fire behaviour of fibre-reinforced polymer (frp

FIRE BEHAVIOUR OF FIBRE-REINFORCED POLYMER (FRP) REINFORCED OR CONFINED CONCRETE

by

Luke Alexander Bisby

A thesis submitted to the Department of Civil Engineering in conformity with the requirements for the degree of Doctor of Philosophy

Queen’s University Kingston, Ontario, Canada April, 2003

Copyright © Luke A. Bisby

“It is the mark of an educated mind to rest satisfied with the degree of precision which the nature of the subject admits, and not to seek exactness where only an approximation is possible.” - Aristotle, 3rd Century B.C.

ABSTRACT Fibre-reinforced polymer (FRP) materials are rapidly gaining acceptance as structural materials for a range of civil engineering applications, particularly as internal tensile reinforcement for concrete beams and slabs, or as confining reinforcement for concrete columns. Until now, the majority of applications of these materials have been for bridges, where performance in fire is not a primary design consideration. There is a potentially much larger market for the use of these materials in buildings, parking garages, and infrastructure, where firesafety is a key design criterion. However, there is currently very little information available on the behaviour of FRP materials and FRP-reinforced or wrapped concrete members in fire. This thesis presents the initial results of an ongoing study into the effects of fire on FRP-reinforced or wrapped concrete members, in an attempt to fill some of the numerous gaps in knowledge. A detailed and comprehensive literature review is presented that provides background information on the high-temperature behaviour of FRP materials and FRP-reinforced concrete members, as well as information on fire testing procedures and objectives. The results of two (2) full-scale fire endurance tests on FRP-wrapped and insulated reinforced concrete columns are presented and discussed, and the test data are used to validate numerical models (also developed and presented herein) which can predict the behaviour of FRP-wrapped columns during fire. The numerical models are subsequently used to conduct parametric studies to investigate the effects of varying a range of parameters on the fire performance of FRP-wrapped columns, and simple fire design recommendations are presented. Numerical models are also developed herein to predict the behaviour in fire of FRP barreinforced concrete slabs. The slab models are validated against fire test data available in the literature, and are subsequently used to conduct parametric studies and suggest fire design guidelines for FRP bar-reinforced concrete slabs. It is demonstrated that FRP materials are extremely sensitive to high temperatures, although the results presented herein indicate that it is possible, through careful consideration of the numerous factors involved, and by providing supplemental fire insulation in some cases, to achieve excellent fire ratings ( in excess of 5 hours) for FRP-reinforced or confined concrete members.

i

ACKNOWLEDGEMENTS Throughout my time as a graduate student at Queen’s University I have been surrounded by a community of tremendously supportive individuals. I am indebted to so many for their kindness, wisdom, and friendship, and I dedicate this thesis to all of those people who have helped out along the way. First and foremost I would like to thank my supervisor and colleague, Dr. Mark Green, for his patience, friendship, generosity, and mentorship over the past six years, and for allowing me the freedom to pursue unique opportunities, outside my core research, that have enabled me to grow and mature as a researcher and instructor. I would also like to thank my co-supervisor Dr. Venkatesh Kodur at NRC for lending his unique insight and experience to the experimental and numerical work presented herein. The full-scale fire endurance tests conducted during the course of this thesis would not have been possible without the visionary support of Fyfe Co. LLC, and I am tremendously grateful to Mr. Ed Fyfe for having the vision to support and participate in this important work. I would also like to thank Mr. Bob Lundborg, of RD Installations Inc., whose practical know-how, calm demeanor, and incomparable skill contributed greatly to the success of the experimental program. I would like to thank the Faculty, Staff, and Graduate Students in the Department of Civil Engineering at Queen’s University for their daily assistance and guidance. In particular, but in no particular order, thanks are due to Dr. T. Ivan Campbell, Dr. Dave Turcke, Joyce Row, Darlene Gaffney, Cathy Wagar, Fiona Froats, Karen McIntyre, Lloyd Rhymer, Neil Porter, Dave Tryon, Jim Roettger, Paul Thrasher, Dorian Tung, Patrick Robitaille, Natalie Rizkalla, Dr. Raafat ElHacha, Dr. Ahmed Debaiky, Dr. Abass Braimah, John Ford, Brea Williams, Scott Shillinglaw, Aaron Dent, Marie Gauthier, Chester Shi and Dr. Laurent Bizindavyii.

ii

For their assistance in conducting some of the ancillary testing presented herein, I would like to thank Dr. Gord Wight and Mr. Jaime Valeria at the Royal Military College of Canada. The full-scale fire endurance tests reported on in this thesis were conducted at the Fire Research Management testing facilities at the National Research Council of Canada. Thanks are due to the NRC Technical Officers and Contractors who worked with me on the project, particularly Patrice Leroux, Joe Hum, and John Latour, whose dedication to their work, competence, and willingness to lend a hand or give advice was one of my most invaluable resources. I would like to thank my friends and family for their support and encouragement over the past four years. I’d like to thank my parents for giving me a rich and happy childhood, for allowing me the freedom to make my own choices, and for helping me at every step along the way.

Finally, I’d like to thank my wife, Lindsay, for her support and encouragement on

occasions far too numerous to mention. This work was financially supported by the Natural Sciences and Engineering Research Council of Canada, The Canadian Networks of Centres of Excellence on Intelligent Sensing for Innovative Structures, Queen’s University, The National Research Council of Canada, and Fyfe Co. LLC.

iii

TABLE OF CONTENTS Abstract………………………………………………………………………………..

i

Acknowledgements………………………………………………………………..

ii

Table of Contents……………………………………………………………………

iv

List of Tables…………………………………………………………………………..

viii

List of Figures………………………………………………………………………….

ix

Notation………………………………………………………………………………..

xvii

Chapter 1 INTRODUCTION 1.1

General………………………………………………………………………...

1

1.2

Motivation: The Global Infrastructure Crisis………………………………….

2

1.3

New and Innovative Infrastructure Solutions………………………………….

4

1.4

Statement of Problem………………………………………………………….

5

1.5

Research Objectives…………………………………………………………...

6

1.6

Scope of the Project……………………………………………………………

7

1.7

Outline of Thesis………………………………………………………………

8

Chapter 2 LITERATURE REVIEW 2.1

General………………………………………………………………………...

10

2.2

Fibre-Reinforced Polymers (FRP)…………………………………………….

10

2.2.1 2.2.2 2.3

General…………………………………………………………………. Composition and Properties of Fibre-Reinforced Polymers……………

10 11

Fibre-Reinforced Polymers in Civil Engineering Applications……………….

15

2.3.1 2.3.2 2.3.3 2.3.4 2.4

Overview……………………………………………………………….. FRP Bars for Beam and Slab Reinforcement…………………………... Column Strengthening with FRP Wraps……………………………….. Economic Considerations……………………………………………….

15 18 21 27

Material Properties at Elevated Temperatures………………………………...

27

2.4.1 2.4.2

General…………………………………………………………………. Concrete………………………………………………………………… 2.4.2.1 Thermal Properties………………………………………………. 2.4.2.2 Mechanical Properties………………...………………………… 2.4.3 Reinforcing Steel……………………………………………………….. 2.4.3.1 Thermal Properties………………………………………………. 2.4.3.2 Mechanical Properties……………………………………………

iv

27 28 28 30 32 32 33

2.4.4

2.5

Fibre Reinforced Polymers....................................................................... 2.4.4.1 Matrix Behaviour............................................................................ 2.4.4.2 Fibre Behaviour.............................................................................. 2.4.4.3 Thermal Properties......................................................................... 2.4.4.4 Mechanical Properties.................................................................... 2.4.4.5 Bond Properties at Elevated Temperatures.................................... 2.4.4.6 Smoke Generation and Toxicity..................................................... 2.4.4.7 Ignition and Flame Spread.............................................................. 2.4.4.8 Barrier Treatments..........................................................................

34 36 38 40 44 49 50 51 52

Fire Endurance...................................................................................................

53

2.5.1 2.5.2

2.6

Philosophy................................................................................................ Procedures to Evaluate Fire Endurance.................................................... 2.5.2.1 Experimental Procedures to Evaluate Fire Endurance................... 2.5.2.2 Numerical Procedures to Evaluate Fire Endurance........................ 2.5.3 Fire Endurance Tests on Reinforced Concrete Members......................... 2.5.4 Fire Endurance Tests on FRP-Reinforced Concrete Members................ 2.5.4.1 FRP Bar-Reinforced Concrete........................................................ 2.5.4.2 FRP-Plated or Wrapped Reinforced Concrete..............................

53 56 57 58 59 59 59 64

Summary............................................................................................................

67

Chapter 3 EXPERIMENTAL PROCEDURE 3.1

General...............................................................................................................

82

3.2

Column Test Program........................................................................................

82

3.2.1

Column Fabrication and Design............................................................... 3.2.1.1 Concrete Mix............................................................................... 3.2.1.2 FRP Wrapping Scheme............................................................... 3.2.1.3 Fire Protection Scheme............................................................... Wrapping and Insulation Procedures........................................................ Column Instrumentation........................................................................... Fire Test Procedures.................................................................................

82 84 84 85 88 93 95

Ancillary Test Program......................................................................................

99

3.2.2 3.2.3 3.2.4 3.3

3.3.1 3.3.2 3.3.3 3.3.4

Concrete Tests.......................................................................................... Reinforcing Steel Tests............................................................................ Fibre-Reinforced Polymer Tests............................................................... Thermogravimetric Analysis....................................................................

99 100 100 101

Chapter 4 EXPERIMENTAL RESULTS AND DISCUSSION 4.1

General...............................................................................................................

118

4.2

Ancillary Tests...................................................................................................

118

4.2.1 4.2.2 4.2.3

Concrete.................................................................................................... Reinforcing Steel...................................................................................... Fibre-Reinforced Polymer........................................................................

v

118 119 119

4.2.4 4.3

4.4

Thermogravimetric Analysis....................................................................

119

Column Tests......................................................................................................

123

4.3.1 General..................................................................................................... 4.3.2 Preload Phase........................................................................................... 4.3.3 Fire Endurance Tests................................................................................ 4.3.3.1 Timeline and Observations................................................................ 4.3.3.2 Temperatures...................................................................................... 4.3.3.3 Axial Deformation and Reinforcement Strains.................................. 4.3.3.4 Load Capacity.................................................................................... 4.3.4 Comparison..............................................................................................

123 123 125 125 127 135 137 138

Discussion and Summary...................................................................................

139

Chapter 5 NUMERICAL MODELLING 1 – COLUMNS 5.1

General...............................................................................................................

158

5.2

Column Modelling.............................................................................................

159

5.2.1 Heat Transfer Model................................................................................. 5.2.2 Overall Load Capacity Model.................................................................. 5.2.3 Pure Axial Load Capacity........................................................................ 5.2.4 Axial Deflection....................................................................................... 5.2.5 Confinement Modelling at High Temperature......................................... 5.2.6 Modelling Intumescent Coatings.............................................................. 5.2.7 Thermomechanical Subroutines............................................................... 5.2.8 Fire Endurance Ratings............................................................................ 5.2.9 Validation................................................................................................. 5.2.9.1 Pre-Test Validation............................................................................ 5.2.9.2 Post-Test Validation...........................................................................

160 169 173 174 174 178 180 180 181 181 183

5.3 Parametric Studies................................................................................................

192

5.3.1 5.3.2 5.3.3 5.3.4

Numerical Comparison of Member Types............................................... Unprotected FRP-Wrapped Reinforced Concrete Columns..................... FRP-Wrapped and Insulated Reinforced Concrete Columns................... Unwrapped but Insulated Reinforced Concrete Columns........................

194 195 195 204

5.4

Consequences for Design...................................................................................

204

5.5

Summary............................................................................................................

206

Chapter 6 NUMERICAL MODELLING 2 – SLABS 6.1 Slab Modelling..................................................................................................... 6.1.1 6.1.2 6.1.3 6.1.4 6.1.5

Heat Transfer Model................................................................................. Flexural Capacity Model.......................................................................... Fire Endurance Requirements.................................................................. Validation................................................................................................. Critical Temperatures for FRP Reinforcing Bars and Grids....................

vi

243 243 246 248 249 251

6.2

Results and Parametric Studies.......................................................................... 6.2.1 6.2.2

252

Fire Endurance Based on Critical Temperature....................................... Structural Fire Endurance.........................................................................

252 255

6.3 Consequences for Design.....................................................................................

258

6.4 Summary..............................................................................................................

260

Chapter 7 CONCLUSIONS AND RECOMMENDATIONS 7.1 Summary................................................................................................................

273

7.2 Conclusions...........................................................................................................

273

7.3 Summary of Design Recommendations................................................................

280

7.4 Recommendations for Further Research...............................................................

281

References....................................................................................................................

285

Appendix A: Additional Heat Transfer Equations........................................

305

Appendix B: Design Charts for FRP Bar-Reinforced Concrete Slabs...

311

Appendix C: Models for FRP-Confined Concrete Columns...................

315

Appendix D: Mechanical Properties of FRP at High Temperature.......

341

Appendix E: Load Calculations for FRP-Wrapped Columns...................

352

Appendix F: Material Properties at Elevated Temperature.....................

363

Vita....................................................................................................................................

370

vii

LIST OF TABLES Table 2.1:

Selected FRP materials currently available for civil engineering applications...............................................................................................

69

Table 2.2:

Qualitative comparison of different FRP types........................................

70

Table 2.3:

Thermal properties of Dexter Hysol® epoxy...........................................

70

Table 2.4:

CTEs of various unidirectional FRP and building materials....................

70

Table 2.5:

Thermal conductivities of various unidirectional FRPs and building materials...................................................................................................

71

Table 2.6:

Gases released during combustion of FRPs.............................................

71

Table 2.7:

UBC classifications for interior finishes..................................................

71

Table 2.8:

Summary of results from fire endurance tests on FRP-plated reinforced concrete beams.........................................................................................

71

Table 3.1:

Mix design proportions............................................................................

102

Table 3.2:

Typical dry fibre properties of Tyfo® SCH-30T FRP.............................

102

Table 3.3:

Typical laminate properties for Tyfo® SCH-30T FRP............................

102

Table 3.4:

Typical properties for Tyfo® S epoxy.....................................................

102

Table 3.5:

Manufacturer specified properties of Tyfo® VG.....................................

103

Table 3.6:

Details of the column test program..........................................................

103

Table 3.7:

Instrumentation for the column tests........................................................

104

Table 4.1:

Results of ancillary tests on concrete.......................................................

142

Table 4.2:

Results of ancillary tests on reinforcing steel...........................................

142

Table 4.3:

Results of ancillary tension tests on SCH FRP coupons..........................

143

Table 4.4:

Visual observations for Fire Test #1: Column ISIS-2..............................

143

Table 4.5:

Visual observations for Fire Test #2: Column ISIS-1..............................

144

Table 5.1:

Summary of various column, wrap, and insulation configurations that were analyzed in conducting parametric studies using QCFIRE......

208

Table 6.1:

Details of analyses conducted during parametric studies.........................

262

Table 6.2:

Details of the 4 slabs analyzed to produce Figure 6.11............................

262

Table 6.3:

Details of the 2 slabs analyzed to produce Figure 6.13............................

263

Table C.1:

Modified empirical confinement models..................................................

334

Table D.1:

Semi-empirically derived coefficients for various mechanical properties of FRP at high temperature......................................................

347

Summary of design load calculations.......................................................

360

Table E.1:

viii

LIST OF FIGURES CHAPTER 2 Figure 2.1:

Variation in shear strength of a typical epoxy resin with temperature.....

72

Figure 2.2:

Manufacturer specified stress-strain behaviour of various currently available FRP reinforcing products..........................................................

72

Differences in behaviour for plain concrete, spirally reinforced, and FRP wrapped concrete cylinders under axial compression......................

72

Figure 2.4:

Variation in density of concrete with temperature...................................

73

Figure 2.5:

Variation in thermal conductivity of concrete with temperature..............

73

Figure 2.6:

Variation in the specific heat of concrete with temperature.....................

73

Figure 2.7:

Idealized variation in thermal capacity of concrete with temperature, for use in numerical modelling.................................................................

74

Variation in (a) the ultimate compressive strength and (b) the modulus of elasticity of concrete with temperature.................................

74

Temperature dependency of concrete and steel strength and modulus of elasticity as idealized for numerical modeling by Lie.........................

74

Figure 2.10:

Stress-strain curves for 35 MPa concrete at different temperatures.........

75

Figure 2.11:

(a) Thermal strain of concretes with different aggregates and (b) effect of load level on thermal strains of concrete..............................

75

Variation in (a) the thermal conductivity and (b) the specific heat of steel with temperature...............................................................................

75

Stress-strain curves for 300 MPa yield-strength steel at different temperatures.............................................................................................

76

Figure 2.14:

Variation in elastic modulus of mild steel with temperature....................

76

Figure 2.15:

Variation in the thermal expansion of mild steel with temperature.........

76

Figure 2.16:

Variation of thermal conductivity, density, and specific heat with temperature for carbon/epoxy FRP..........................................................

77

Variation of strength with temperature for GFRP, timber, concrete, and steel....................................................................................................

77

Figure 2.18:

Variation in tensile strength of various fibres with temperature..............

78

Figure 2.19:

Variation of strength of various carbon FRPs with temperature..............

78

Figure 2.20:

Variation of strength of various glass FRPs with temperature.................

79

Figure 2.21:

Variation of strength various aramid FRPs with temperature..................

79

Figure 2.22:

Variation of elastic modulus of (a) carbon (b) glass and aramid FRPs with temperature.......................................................................................

80

Results of smoke generation tests on various FRPs.................................

81

Figure 2.3:

Figure 2.8: Figure 2.9:


Figure 2.17:

Figure 2.23:

ix

Figure 2.24:

Details of selected FRP plated beams fire tested by Blontrock et al........

81

Figure 3.1:

Column dimensions and reinforcement details........................................

105

Figure 3.2:

Typical column base plate with welded longitudinal and spiral steel......

106

Figure 3.3:

Top of column reinforcement cage including the aluminum collar used to maintain the longitudinal bars in the correct location curing fabrication and pouring.............................................................................

106

CHAPTER 3

Figure 3.4:

Casting of the concrete columns............................................................... (a) Top of forms with aluminum collar and instrumentation wires (b) Forms inside scaffold, crane, and hopper (c) Pouring concrete into elephant trunk chute with hopper

107

Figure 3.5:

Details of the FRP wrapping scheme for columns ISIS-1 and ISIS-2.....

108

Figure 3.6:

Application of the epoxy prime coat........................................................

108

Figure 3.7:

Saturation of SCH sheet with epoxy........................................................

108

Figure 3.8:

Application of a single lift of SCH sheet..................................................

109

Figure 3.9:

Removal of air voids at the FRP/concrete interface.................................

109

Figure 3.10:

Columns ISIS-1 and ISIS-2 with SCH sheets installed............................

109

Figure 3.11:

Sanding the surface of the SCH sheets prior to application of the VG primer.......................................................................................................

110

Photo showing connection detail of steel angle bracket to concrete column......................................................................................................

110

Schematic showing connection detail for insulation brackets and diamond lath.............................................................................................

110

Figure 3.14:

Column ISIS-1 with diamond lath installed.............................................

111

Figure 3.15:

Photo of ring spacer on column ISIS-2....................................................

111

Figure 3.16:

Application of VG primer........................................................................

111

Figure 3.17:

Application of VG....................................................................................

111


st

nd

Figure 3.18:

Columns between 1 and 2 lifts of sprayed VG....................................

111

Figure 3.19:

Trowel application of VG to obtain desired thickness and surface finish.........................................................................................................

111

VG insulation immediately after removal of spacer ring on column ISIS-1........................................................................................................

112

Figure 3.21:

Columns ISIS-1 and ISIS-2 after application of VG................................

112

Figure 3.22:

Column ISIS-1 after application of EI......................................................

112

Figure 3.23:

Location and number of various sensors for the column tests................. (a) Thermocouples in the concrete, FRP, and insulation (b) Thermocouples on the reinforcing steel (c) Strain gauge coupons welded to the vertical reinforcing steel

113

Figure 3.20:

x

Figure 3.24:

Thermocouple frame for sensors TC3 to TC7 before being installed on the reinforcement cage........................................................................

113

Figure 3.25:

Thermocouples TC10 and TC11 installed on reinforcing steel................

114

Figure 3.26:

Placement of TC1 and TC9 installed on outside of FRP.........................

114

Figure 3.27:

Thermocouple TC18 before application of VG........................................

114

Figure 3.28:

Strain coupon installed on the longitudinal reinforcing steel...................

115

Figure 3.29:

Locations of thermocouples inside the test furnace.................................

115

Figure 3.30:

Photo of the FRM/IRC/NRC column testing facility with column ISIS-2 just prior to testing........................................................................

116

Figure 3.31:

ULC S101 time-temperature curve for fire endurance tests.....................

116

Figure 3.32:

SCH coupon used for tensile testing of FRP............................................

117

Complete experimental stress-strain curves for 20M reinforcing bars...........................................................................................................

145

Figure 4.2:

Elastic range stress-strain curves for 20M reinforcing bars.....................

145

Figure 4.3:

Complete experimental stress-strain curves for 10M spiral steel.............

145

Figure 4.4:

Elastic range stress-strain curves for 10M spiral steel.............................

146

Figure 4.5:

Stress-strain data for Tyfo®SCH-30T coupons tested in tension............

146

Figure 4.6:

Tyfo®SCH-30T coupon immediately after tensile failure.......................

146

Figure 4.7:

TGA data for S-Epoxy.............................................................................

147

Figure 4.8:

TGA data for SCH and SEH fibres..........................................................

147

Figure 4.9:

Variation in mass from TGA data for SCH FRP system and for a carbon/epoxy FRP tested by Dimitrienko................................................

147

Figure 4.10:

TGA data for VG insulation.....................................................................

148

Figure 4.11:

TGA data for EI paint...............................................................................

148

Figure 4.12:

Photo-documentation of EI expansion and contraction with temperature...............................................................................................

149

Load versus stroke data during the preload phase for columns ISIS-1 and ISIS-2.................................................................................................

150

Figure 4.14:

Load versus strain plot during the preload phase for column ISIS-1.......

150

Figure 4.15:

Load versus strain plot during the preload phase for column ISIS-2.......

150

Figure 4.16:

Column ISIS-2 in the test furnace just before commencing test #1.........

151

Figure 4.17:

View of column ISIS-2 through a view-port during the intumescent reaction.....................................................................................................

151

Figure 4.18:

Column ISIS-2: EI in the process in falling off........................................

151

Figure 4.19:

Column ISIS-2: VG completely exposed to fire......................................

151

CHAPTER 4 Figure 4.1:

Figure 4.13:

xi

Figure 4.20:

Column ISIS-2: Immediately after failure................................................

152

Figure 4.21:

Column ISIS-1: Close-up of vertical crack formation in the VG.............

152

Figure 4.22:

Column ISIS-1: Close-up of flaming at a vertical crack in the VG.........

152

Figure 4.23:

Column ISIS-1: Immediately after failure................................................

152

Figure 4.24:

Time versus temperature plots for column ISIS-2................................... (a) Furnace, EI, VG, and FRP temperatures (b) Concrete temperatures (c) Reinforcing steel temperatures

153

Figure 4.25:

Time versus temperature plots for column ISIS-1................................... (a) Furnace, EI, VG, and FRP temperatures (b) Concrete temperatures (c) Reinforcing steel temperatures

154

Figure 4.26:

Column ISIS-2: Initial EI/VG interface time-temperature curves...........

155

Figure 4.27:

Column ISIS-1: Initial EI/VG interface time-temperature curves...........

155

Figure 4.28:

Axial elongation and applied load during both fire tests..........................

155

Figure 4.29:

Strains measured on the longitudinal reinforcing steel during Fire Test #1......................................................................................................

156

Figure 4.30:

Complete load versus axial elongation plots for both columns................

156

Figure 4.31:

Design, predicted, and tested strength of columns ISIS-1 and ISIS-2.....

156

Figure 4.32:

Comparison of temperature histories at various locations within columns ISIS-1 and ISIS-2 during fire tests............................................. (a) Temperatures at the EI/VG and VG/FRP interfaces (b) Temperatures at the FRP/concrete interface, rebar-outside, and centerline locations

157

CHAPTER 5 Figure 5.1a:

Main program schematic for QCFIRE.....................................................

209

Figure 5.1b:

Program schematic for heat transfer in an unwrapped column................

210

Figure 5.1c:

Program schematic for heat transfer in an FRP-wrapped column............

212

Figure 5.1d:

Program schematic for heat transfer in an FRP-wrapped and insulated column......................................................................................................

215

Figure 5.1e:

Program schematic for the load capacity analysis portion of QCFIRE....

218

Figure 5.1f:

Program schematic for the algorithm to account for moisture in the concrete.....................................................................................................

220

Figure 5.1g:

Program schematic for the confining effect of an FRP wrap...................

221

Figure 5.2:

Heat transfer discretization for an uninsulated FRP-wrapped concrete column......................................................................................................

222

Figure 5.3:

Load capacity discretization used in QCFIRE.........................................

222

Figure 5.4:

Assumed variation in column curvature and deflection used in QCFIRE....................................................................................................

223

xii

Figure 5.5:

Subprogram schematic for QCFIRE-Axial...............................................

224

Figure 5.6:

Subprogram schematic for QCFIRE-Deflection......................................

226

Figure 5.7:

Schematic of the intumescent process......................................................

228

Figure 5.8:

Details of the columns tested by Lie and Celikkol...................................

228

Figure 5.9:

Predicted and observed temperatures in the concrete for columns tested by Lie and Celikkol........................................................................

229

Predicted and observed overall axial elongation for columns tested by Lie and Celikkol.......................................................................................

229

Predicted and observed fire endurance (load capacity) for the columns tested by Lie and Celikkol........................................................................

229

Figure 5.10: Figure 5.11: Figure 5.12:

Predicted and observed temperatures for column ISIS-2......................... a) Temperatures in the concrete b) Temperatures in the reinforcing steel c) Temperatures in the EI, VG, and FRP

230

Figure 5.13:

Predicted and observed temperatures for column ISIS-1......................... a) Temperatures in the concrete b) Temperatures in the reinforcing steel c) Temperatures in the EI, VG, and FRP

231

Figure 5.14:

Forced validation plots for column ISIS-2............................................... a) Temperatures in the concrete (FRP/Concrete Forced) b) Temperatures in the reinforcing steel (FRP/Concrete Forced) c) Heat transfer through the FRP (VG/FRP Forced) d) Heat transfer through the VG (EI/VG Forced)

232

Figure 5.15:

Forced validation plots for column ISIS-1............................................... a) Temperatures in the concrete (FRP/Concrete Forced) b) Temperatures in the reinforcing steel (FRP/Concrete Forced) c) Heat transfer through the FRP (VG/FRP Forced) d) Heat transfer through the VG (EI/VG Forced)

233

Figure 5.16:

Measured and predicted axial elongation for columns ISIS-1 and ISIS-2........................................................................................................

234

Predicted and observed axial load capacities for columns ISIS-1 and ISIS-2........................................................................................................

234

Fire endurance curves for unwrapped, wrapped, and wrapped and insulated reinforced concrete columns.....................................................

234

Fire endurance curves for the initial 2 hours of fire exposure for unwrapped, wrapped, and wrapped and insulated reinforced concrete columns....................................................................................................

235

Effect of FRP fibre type on the fire behaviour of FRP-wrapped and insulated concrete columns.......................................................................

235

Figure 5.17: Figure 5.18: Figure 5.19:

Figure 5.20:

xiii

Figure 5.21:

Figure 5.22:

Figure 5.23:

Figure 5.24:

Figure 5.25: Figure 5.26: Figure 5.27: Figure 5.28: Figure 5.29: Figure 5.30: Figure 5.31:

Effect of insulation thickness on the fire endurance of FRP-wrapped and insulated concrete columns................................................................ a) Fire endurance based on matrix GTT b) Fire endurance based on matrix ignition temperature c) Structural fire endurance Effect of insulation thermal conductivity on the fire behaviour of FRP-wrapped and insulated concrete columns......................................... a) Fire endurance based on matrix GTT b) Fire endurance based on matrix ignition temperature c) Structural fire endurance Effect of insulation specific heat on the fire behaviour of FRPwrapped and insulated concrete columns................................................. a) Fire endurance based on matrix GTT b) Fire endurance based on matrix ignition temperature c) Structural fire endurance Effect of insulation density on the fire behaviour of FRP-wrapped and insulated concrete columns................................................................ a) Fire endurance based on matrix GTT b) Fire endurance based on matrix ignition temperature c) Structural fire endurance

236

237

238

239

Effect of FRP matrix GTT on the criterion 1 fire endurance of FRPwrapped and insulated concrete columns.................................................

240

Effect of FRP ignition temperature on the fire behaviour of FRPwrapped and insulated concrete columns.................................................

240

Effect of concrete aggregate type on the fire behaviour of FRPwrapped and insulated concrete columns.................................................

240

Effect of concrete compressive strength on the fire behaviour of FRP-wrapped and insulated columns.......................................................

241

Effect of steel reinforcement ratio on the fire behaviour of FRPwrapped and insulated concrete columns.................................................

241

Effect of confinement ratio on the fire behaviour of FRP-wrapped and insulated concrete columns................................................................

241

Maximum allowable strength increases for FRP-wrapped members according to the recommendations presented herein and those suggested by ACI 440.2 R-02..................................................................

242

Discretization of a concrete slab for the purposes of heat transfer analysis .....................................................................................................

264

Schematic showing the program logic for the heat transfer portion of QSFIRE....................................................................................................

266

Schematic showing the program logic for the flexural capacity portion of QSFIRE....................................................................................

268

CHAPTER 6 Figure 6.1: Figure 6.2: Figure 6.3:

xiv

Figure 6.4:

QSFIRE temperature validation plot using carbonate aggregate slab test data from PCA...................................................................................

268

QSFIRE temperature validation plot using siliceous aggregate slab test data from PCA...................................................................................

269

Figure 6.6:

QSFIRE temperature validation plot using data from NEFCOM............

269

Figure 6.7:

Effect of overall slab thickness on the fire endurance of FRPreinforced concrete slabs..........................................................................

269

Effect of concrete cover thickness and aggregate type on the fire resistance of FRP-reinforced concrete slabs.............................................

270

Effect of critical temperature of the reinforcement on the fire resistance of FRP-reinforced concrete slabs.............................................

270

Effect of concrete moisture content on the fire endurance of FRPreinforced concrete slabs..........................................................................

270

Effect of FRP type on the Load capacity of FRP reinforced concrete slabs during fire........................................................................................

271

Effect of FRP type on the capacity of FRP reinforced slabs designed based on permissible crack width.............................................................

271

Sample design chart for a 120 mm thick carbonate aggregate reinforced concrete slab............................................................................

271

Comparison of flexural capacity during fire for a steel reinforced concrete slab and an equivalent slab reinforced with CFRP and designed using Figure 6.12.......................................................................

272

Discretization of an unwrapped circular concrete column for heat transfer purposes.......................................................................................

309

Discretization of a wrapped and insulated circular concrete column for heat transfer purposes.........................................................................

309

Design chart for 120 to 200 mm thick carbonate aggregate concrete slabs..........................................................................................................

311

Design chart for 120 to 200 mm thick siliceous aggregate concrete slabs..........................................................................................................

311

Design chart for 120 to 200 mm thick expanded shale aggregate concrete slabs............................................................................................

312

Design chart for 200 mm thick or larger carbonate aggregate concrete slabs............................................................................................

312

Design chart for 200 mm thick or larger siliceous aggregate concrete slabs..........................................................................................................

313

Design chart for 200 mm thick or larger expanded shale aggregate concrete slabs............................................................................................

313

Various proposed stress-strain curves for FRP-confined concrete...........

335

Figure 6.5:

Figure 6.8: Figure 6.9: Figure 6.10: Figure 6.11: Figure 6.12: Figure 6.13: Figure 6.14:

APPENDICES Figure A.1: Figure A.2: Figure B.1: Figure B.2: Figure B.3: Figure B.4: Figure B.5: Figure B.6: Figure C.1:

xv

Figure C.2:

Figure C.3: Figure C.4:

Figure C.5: Figure C.6: Figure C.7: Figure C.8:

Figure D.1: Figure D.2: Figure D.3: Figure D.4: Figure D.5: Figure D.6: Figure D.7: Figure D.8: Figure D.9: Figure D.10:

A comparison of mean absolute errors and mean errors for ultimate strength predictions of the various confinement models as compared with the database of experimental data....................................................

336

Comparison of model prediction and experimental data for ultimate strength prediction by the Saafi confinement model................................

336

A comparison of mean absolute errors and mean errors for failure strain predictions of the various confinement models as compared with the database of experimental data....................................................

337

Comparison of model prediction and experimental data for ultimate strain prediction by the Miyauchi II confinement model.........................

337

A comparison predicted and experimental stress-strain behaviour of a CFRP wrapped concrete cylinder..........................................................

338

A comparison predicted and experimentally observed ultimate stress for the ACI 440 confinement model with reduction factors included......

338

A comparison predicted and experimentally observed ultimate stress for the ISIS Canada confinement model with reduction factors included....................................................................................................

339

Variation of tensile strength properties coefficient for CFRP with temperature...............................................................................................

348

Assumed form for the analytical expression describing deterioration of strength, stiffness, and bond of FRP reinforcements...........................

348

Database points and analytical expressions for the variation of tensile strength of pure carbon, glass, and aramid fibres with temperature.........

349

Database points and analytical expression for the variation of elastic modulus of CFRP with temperature.........................................................

349

Database points and analytical expressions for the variation of elastic modulus of GFRP and AFRP with temperature.......................................

350

Database points and analytical expression for the variation of tensile strength of CFRP with temperature..........................................................

350

Database points and analytical expression for the variation of tensile strength of GFRP with temperature..........................................................

351

Database points and analytical expression for the variation of tensile strength of AFRP with temperature..........................................................

351

Database points and analytical expression for the variation of bond strength of glass/vinyl-ester FRP with temperature.................................

352

Comparison of analytical expressions for the variation of strength, elastic modulus, and bond strength of FRP temperature..........................

352

xvi

NOTATION Ac

cross-sectional area of the concrete column

Acom

cross-sectional area of the FRP wrap

Ag

gross cross-sectional area of concrete

Am ,n

cross-sectional area of element m, n

Ast

cross-sectional area of reinforcing steel

Asurf

surface area of a ring element

aE

residual FRP elastic modulus constant

aσ

residual FRP ultimate strength constant

bE

curve severity constant for FRP elastic modulus

bσ

curve severity constant for FRP ultimate strength

Cc

specific heat of concrete

Cc, T

specific heat of concrete at temperature T

Celem

specific heat of the element

C H 2O

specific heat of water

Ci

specific heat of insulation

Cj

confinement modulus

CVG ,T

specific heat of VG insulation at temperature T

Cw

specific heat of the FRP wrap

Cw,T

specific heat of the FRP at temperature T

cE

central value constant for FRP elastic modulus

cσ

central value constant for FRP ultimate strength

d

concrete column diameter

de

distance from element centroid to column centreline

d m ,n

distance from the centroid of element m, n to the centerline of the column

dw

average diameter of the FRP wrap

eo

assumed initial eccentricity of the compressive load

E1

initial modulus of confined concrete stress-strain curve

E2

final modulus of confined concrete stress-strain curve

xvii

Ecom

elastic modulus of FRP wrap

Ecom,T

elastic modulus of FRP wrap at temperature T

Ec

initial elastic modulus of unconfined concrete

Eeff

effective axial elastic modulus for an FRP wrapped concrete column

Esec

secant modulus of concrete

Esec,µ

secant modulus of confined concrete at failure

Ew

modulus of the FRP wrap at current average wrap temperature

fa

concrete stress at the point of damage initiation

fc

current concrete axial stress

f c',T

current concrete axial stress at temperature T

f’cc

confined concrete ultimate strength

f’co

unconfined concrete ultimate compressive strength

fcom

FRP ultimate tensile strength

fcom,T

FRP ultimate tensile strength at temperature T

fl

lateral confining pressure at ultimate

flat

lateral confining pressure at any point less than ultimate

fl(max)

maximum allowable lateral confining pressure at ultimate for design

fo

stress axis intercept of linear portion of stress-strain curve

f s ,T

(hrad )max

current stress in the steel at temperature T maximum value of the equivalent heat transfer coefficient due to radiant heating at the column-fire interface

i

time step counter

k

thermal conductivity

k1

confinement effectiveness coefficient for confined ultimate strength

k2

confinement effectiveness coefficient for confined ultimate strain

kc, T

thermal conductivity of concrete at temperature T

ke

effectiveness coefficient for FRP-confined concrete

k mi

thermal conductivity of element m at time step i

(k c )max

maximum thermal conductivity to be expected in the concrete at any point during exposure to fire

xviii

(ki )max kVG ,T

(k w )max

maximum thermal conductivity to be expected in the insulation at any point during exposure to fire thermal conductivity of VG insulation at temperature T maximum thermal conductivity to be expected in the wrap material at any point during exposure to fire

kw,T

thermal conductivity of the FRP wrap at temperature T

K

effective length factor

L

distance over which conduction occurs

m

radial (ring) element counter

melem

mass of an element

m H 2O

mass of water evaporated from and elemental layer during time step ∆t

M1

ring element counter at the FRP-concrete interface

M2

ring element counter at the columns centreline

i M ext

external moment at the column mid-height at time step i

i M int

internal moment at the column mid-height at time step i

Mins

ring element counter at the insulation-FRP interface

n

curve-shape parameter, circumferential (annular) element counter, number of layers of FRP wrap

nwrap

number of elemental layers in the FRP wrap material

Pr(max)

CSA A23.3 design axial load capacity

q&

heat generation term

Rc

outer radius of reinforced concrete column

Ri

outer radius of insulation

Rw

outer radius of FRP wrap

r

variable as defined in Eq. 2.15a

SDL

service live load

SLL

service dead load

T

temperature

Tair

temperature of ambient air above concrete slab

Tc

temperature of the concrete

Tcrit

critical temperature of FRP reinforcement

Tfi

fire temperature at time step i

xix

Tmi

temperature of element m at time step i

Ts

temperature of the reinforcing steel

Tsurf

surface temperature of the column or slab

Tsurr

temperature of the surroundings

TVG

temperature of VG insulation

Tw

temperature of the FRP wrap

i Twrap

average temperature of the FRP wrap at time step i

t, t w

thickness of FRP wrap (time in some cases)

t slab

slab thickness

Vlay

volume of a ring element layer

VmH 2O

initial volume of water in concrete element m

x

variable as defined in Eq. 2.15b

y

column mid-height lateral deflection

α1

stress block factor for concrete

αpc

performance coefficient for FRP wraps

αc,T

coefficient of thermal expansion of concrete at temperature T

αs,T

coefficient of thermal expansion of steel at temperature T

αw

coefficient of thermal expansion of FRP wrap

β

constant as defined in Eq. 2.17

β1

stress block factor for concrete

∆Qin

heat supplied to an elemental layer during time step ∆t

Qinrad

heat into an element due to radiation during time interval ∆t

cond Qout

heat out of an element due to conduction during time interval ∆t

∆Qst

change in the energy stored in an element during time interval ∆t

∆T

change in temperature

∆t

time step used in numerical analysis

∆VmH 2O

change in volume of moisture in element m during a time interval ∆t

∆x

distance increment in the concrete in slab analysis

∆xc

distance increment in the concrete in column analysis

xx

∆xi

distance increment in the insulation in column analysis

∆xw

distance increment in the FRP wrap

ε

strain, emmissivity

ε axial

axial strain in the concrete

εA

strain in the concrete at the point of damage initiation

εc

concrete axial strain, emmissivity of concrete

εcc

strain in confined concrete at ultimate

εco

strain in unconfined concrete at ultimate

εcom

failure strain of FRP wrap

εf

emmissivity of the fire

(εf )m

flexural strain in slab element m

ε1

lateral strain of concrete at ultimate

ε1at

lateral dilation (strain) of concrete element

εmax

crushing strain of concrete

εp

yield strain in reinforcing steel

εs

current strain in reinforcing steel

εsurf

emmissivity of the outer column surface

εsurr, εs

emmissivity of the surroundings (strain in steel sometimes)

εT

thermal strain

εtan

strain the concrete at the point of tangency

ε total

overall lateral stain of the concrete column

εw

emmissivity of the FRP wrap

λ

constant

λH O

heat of vaporization of water

νc

initial Poisson’s ratio for unconfined concrete

ν’c

Poisson’s ratio for confined concrete at ultimate

π

Pi

ρc

density of concrete

ρ c,T

density of concrete at temperature T

ρH O

density of water

2

2

xxi

ρi

density of insulation

ρVG ,T

density of VG insulation at temperature T

ρw

density of the FRP wrap

ρw,T

density of the FRP at temperature T

(ρ c Cc )min (ρ i Ci )min (ρ wC w )min

minimum value of thermal capacity of the concrete to be expected during exposure to fire minimum value of thermal capacity of the insulation to be expected during exposure to fire minimum value of thermal capacity of the wrap material to be expected during exposure to fire

σ

stress, Stephan-Boltzman constant

σA

stress in concrete at the point of damage initiation

i σ m, n

stress in element m, n at time step i

φc

resistance factor for concrete

φfrp

resistance factor for FRP

φi

initial moisture content in the concrete

(φRn )existing

existing strength of a column member to be strengthened

φs

resistance factor for reinforcing steel

φ mi

moisture content of element m at time step I

χ

curvature

ωw

volumetric confinement ratio

xxii

CHAPTER 1: Introduction

CHAPTER 1 INTRODUCTION 1.1 General Fibre-reinforced polymer (FRP) materials have, in the last 10 years or so, received a great deal of interest in the civil engineering research community.

Although developed for the

aerospace, automotive, and chemical processing industries, FRPs display many characteristics, such as high strength-to-weight ratios and resistance to corrosion, which are advantageous in a variety of civil engineering applications. Recent advances in knowledge and FRP technology, as well as an increase in the use of these materials, have led to reduced material costs and increased confidence among structural designers, such that there are now hundreds of field applications of FRPs in civil engineering applications around the world. Possibilities for the use of FRPs in structural applications are numerous. There are however, several applications for which these materials have shown an almost immediate promise. The first of these is internal reinforcement of concrete. In this application, FRP rods, grids, or tendons are used to provide internal tensile reinforcement in place of conventional reinforcing steel. The motivation for the substitution is primarily that FRPs are non-corrosive (electrochemically) and will not deteriorate by oxidation.

Corrosion of conventional steel

reinforcement in many existing reinforced concrete structures has caused widespread deterioration in the last 50 years or so, and has led to what many have referred to as the current global infrastructure crisis. A second, and equally promising, use of FRPs in structural engineering applications involves repair and rehabilitation of existing reinforced concrete structures by bonding FRP plates, sheets, or wraps, to the exterior of reinforced concrete members. This is done to provide tensile or confining reinforcement that supplements the reinforcement provided by the internal

1

L.A. Bisby, Ph.D. Thesis, 2003


reinforcing steel. Repair and strengthening of reinforced concrete structures with externally bonded FRP has proven to be a very promising application, and the technique is now widely recognized for its effectiveness and ease of application. Research in the area of FRP-reinforced concrete, with internal or external FRP reinforcement, has now advanced to a point where design guidelines are being produced and widespread acceptance of these materials seems inevitable. Many structures have been built incorporating internal FRP reinforcement, and many more have been rehabilitated with external FRP plating. However, until now, applications of FRPs in structures have been limited mainly to bridges and exterior applications, where fire endurance is not a primary concern in design. There is an enormous potential for applications of FRP-reinforced and/or -repaired concrete in multistorey buildings, parking garages, and industrial structures. However, before FRP reinforcement can be used with confidence in buildings, the performance of these materials during fire, and their ability to meet the fire endurance criteria set out in building codes, must be evaluated. To date, information in this area is extremely scarce, and a great deal of further work is required to fill all the gaps in knowledge. The purpose of this thesis is to start to fill some of the gaps in the understanding of the fire endurance of structures built with FRPs.

1.2 Motivation: The Global Infrastructure Crisis The population of the modern civilized world depends on a complex and extensive system of infrastructure for economic prosperity. The existing public infrastructure of Canada, the United States, Europe, and other countries has suffered from decades of neglect and overuse, leading to the accelerated deterioration of bridges, buildings, municipal and transportation systems, and resulting in a situation that many in the civil engineering industry are calling a global infrastructure crisis. Much infrastructure is unsatisfactory in some respect, and public funds are not generally available for the required replacement of existing structures or construction of new ones.

This has led to a building boom in the upgrading, repair, and

2



rehabilitation industry, where the focus is now on extending the service lives and/or increasing the live load capacity of existing structures. Many factors have contributed to the unsatisfactory condition of our infrastructure. For the purposes of this thesis, which is concerned primarily with reinforced concrete infrastructure, these factors include: design errors in the original construction, deterioration due to corrosion of reinforcing steel or vehicle collision, increases in required load carrying capacity, changes in building codes (particularly with regard to seismic requirements), or changes in loads and load paths due to building renovation and/or addition. As an example of the infrastructure problems facing North America, in 2001 the American Society of Civil Engineers (ASCE) released a report card on America’s infrastructure (ASCE, 2001), that presented data on the deterioration of infrastructure within the United States. The report states, in part, that 29% of bridges in America are structurally deficient and/or functionally obsolete in 1998. According to the Federal Highway Administration (FHWA), expenditures in the order of US$10.6 billion are required annually for the next twenty years in order to eliminate current bridge deficiencies. In addition, the maintenance of existing bridges will require an expenditure of approximately US$5.8 billion per year. Not included in the ASCE report is any information on the current condition of hospitals, parking garages, and industrial buildings, many of which are deficient in some respect and require repair, rehabilitation, or upgrading. Nonetheless, it is evident that the expenditures required to bring public and private infrastructure to acceptable levels are extremely daunting. The Canadian Society for Civil Engineering recently released a report (CSCE, 2002) that outlines the magnitude of the infrastructure crisis in Canada. The report states, in part, that the Canadian infrastructure – a $1.6 trillion asset – is not sustainable, that we are accumulating an infrastructure debt that could grow to $112 billion in 2027, and that on average, our civil infrastructure system has used over 80% of its service life. Obviously, new and innovative infrastructure solutions are required.

3



1.3 New and Innovative Infrastructure Solutions Soaring costs associated with structural repair, renovation and replacement, combined with severe financial constraints placed on public and private organizations, have spawned many new and innovative solution techniques for the repair and rehabilitation of reinforced concrete infrastructure. New construction has also witnessed the introduction of many new materials in an attempt to prolong the service lives of reinforced concrete structures. In cases where repair or rehabilitation of existing reinforced concrete is required, or where additional live load capacity is needed, some initial options that were studied and implemented included: bonding steel plates to the tension and side faces of beams and slabs (Ali, 1997; Swamy et al., 1987), which has been shown to increase the flexural and shear capacities of these members; steel jacketing of columns (Priestly et al., 1994), resulting in increased axial, flexural, and shear capacity, and increased ductility; and external post-tensioning of beams with steel tendons (Bruggeling, 1992). These techniques have all been implemented with varying degrees of success, although all suffer from the disadvantage that steel corrodes electrochemically, often resulting in loss of bond between the concrete or steel, loss of crosssectional area of the steel, anchorage failure, etc. More recently, many of the above approaches have been implemented using fibre-reinforced polymers in place of steel. FRPs have numerous advantages over steel, such as very high strength-to-weight ratios and resistance to electrochemical corrosion. These advantages have led to their widespread use for external plating of reinforced concrete beams and for wrapping of reinforced concrete columns (Bakis et al., 2002). In new construction, FRPs have also started to see increased use as reinforcement for concrete (Bakis et al., 2002). The impetus toward the use of FRPs in new construction is provided by the non-corrosive nature of these materials, which prevents the cracking and spalling

4



invariably observed when conventional reinforcing steel corrodes and expands inside reinforced concrete structures. By all indications, it would appear that FRPs are set to revolutionize the way in which reinforced concrete infrastructure is constructed, repaired, or replaced.

1.4 Statement of Problem As their name suggests, FRPs are composed of slender fibres embedded in a polymer matrix. In Civil Engineering applications the fibres are generally carbon (graphite), glass, or aramid (KevlarTM), and the matrices are generally epoxies, polyesters, or vinyl esters.

As

discussed in greater detail in Chapter 2, the thermal and mechanical properties of FRPs are highly dependent on temperature, and severe degradation of mechanical properties is observed for relatively mild (100ºC to 200ºC) increases in temperature. Most reinforced concrete structures are designed subject to fire endurance requirements that are set out in building codes. There is a concern that FRPs may not be able to survive exposure to fire. The temperature increases that are likely in the event of fire (over 1000°C) would be sufficient to cause loss of mechanical strength and stiffness of the FRP, or perhaps loss of bond between the FRP and the concrete. Loss of strength, stiffness, or bond could potentially lead to failure of the structure, resulting in less than adequate fire endurance. When FRPs are used in lieu of conventional reinforcing or prestressing steel for reinforcing concrete, there are several fundamental differences in material properties that must be considered. Some of the potential concerns associated with internal FRP reinforcement in the event of fire include: internal stresses developed through differential thermal expansion, strength and stiffness deterioration of the reinforcement, thermally induced creep of the reinforcement, loss of bond, smoke and toxic gas evolution, and flammability (Tanano et al., 1999). When FRP is used for external plating of reinforced concrete members, the concerns are amplified by the fact that no concrete cover is present to provide fire protection for the FRP.

5



Many of the concerns associated with FRPs in external plating applications are the same as for internal FRP reinforcing applications. These include: differential thermal expansion, strength and stiffness deterioration, and thermally induced creep. Of particular concern in the case of external plating are: smoke and toxic gas evolution, loss of bond, and flammability. While research into the use of FRP reinforcement for concrete has advanced to a point where design codes for the use of FRPs in structures are currently being produced, the behaviour of FRP-reinforced, -plated, or -wrapped concrete members in fire remains largely unknown.

1.5 Research Objectives The experimental and numerical study presented in this thesis seeks to examine the implications of the high temperature thermal susceptibility of FRPs currently used in civil engineering applications on the structural behaviour of FRP-reinforced concrete slabs and FRPwrapped reinforced concrete columns. To achieve this objective, exhaustive literature reviews and meta-analyses have been conducted, numerical models have been developed, and full-scale fire endurance tests on FRP-wrapped and insulated reinforced concrete columns have been conducted at the Fire Risk Management Centre (FRM) of the Institute for Research in Construction (IRC) at the National Research Council of Canada (NRC). The primary objectives of this research project were: To experimentally investigate the behaviour in fire (as defined by ASTM E119 or ULC S101) of circular FRP-wrapped and insulated reinforced concrete columns. To develop, and validate against experimental data, numerical models to simulate the behaviour in fire of circular FRP-wrapped and insulated reinforced concrete columns. To investigate, both numerically and experimentally, techniques to improve the behaviour in fire of FRP-wrapped reinforced concrete columns. To use experimental data and numerical models to provide guidance to designers regarding the fire behaviour of FRP-wrapped reinforced concrete columns.

6



The secondary objectives of this research were: To develop a numerical model to study the behaviour in fire of FRP bar- or gridreinforced concrete slabs. To validate the numerical model for FRP bar-reinforced concrete slabs against data available in the literature from previous studies on the fire endurance of steel and/or FRP bar- or grid-reinforced concrete slabs. To use the numerical slab model to provide guidance on fire endurance to designers of concrete slabs incorporating FRP bar or grid reinforcement.

1.6 Scope of the Project The work presented in this thesis involved both experimental tests and numerical analyses. The experimental program consisted of the fabrication and fire endurance testing of full-scale FRP-wrapped and insulated circular reinforced concrete columns.

Six (6) fully

instrumented full-scale reinforced concrete columns were fabricated, and 2 of these were tested for the purposes of this thesis. The remaining 4 columns will be tested at a later date. The columns were 3810 mm tall with a diameter of 400 mm, and were reinforced internally with longitudinal (vertical) and spiral (transverse) reinforcement in the form of conventional reinforcing steel. Numerical work involved the development, and subsequent validation, of computer programs to simulate the fire behaviour of circular FRP-wrapped and insulated reinforced concrete columns and FRP bar-reinforced concrete slabs.

These computer programs were

developed using a finite-difference (FD) heat transfer methodology, coupled with a strainequilibrium approach for structural behaviour at high temperature. The numerical models were validated against the experimental results presented in this thesis (where possible) and against experimental data available in the literature. Once validated, the numerical computer models were used to develop design guidelines and charts for FRP-

7



wrapped reinforced concrete columns and FRP bar-reinforced concrete slabs. The numerical models also assisted in pre-defining the column insulation schemes that were selected for the fullscale fire endurance tests on FRP-wrapped reinforced concrete columns.

1.7 Outline of Thesis Chapter 2 presents an extensive review of the literature pertinent to the study of the fire behaviour of FRPs as reinforcement for concrete members. The chapter begins with a discussion of FRP materials’ composition and properties, followed by a description of the various uses of FRP in structural engineering applications and a discussion of some of the economic considerations involved with their use. The material properties of concrete, steel, and FRPs at high temperature are subsequently presented in detail.

The discussion then turns to issues

associated with fire endurance, including concerns associated with FRPs at high temperature, methods to determine fire endurance, and results of fire endurance tests on both conventionally reinforced concrete and FRP-reinforced concrete members. Chapter 3 provides details of the experimental program conducted during the course of this thesis.

Ancillary tests to determine thermal and mechanical properties of the various

materials involved are first presented and discussed, followed by details of the fire endurance tests on full-scale circular FRP-wrapped and insulated reinforced concrete columns. The results of the experimental program are presented and discussed in Chapter 4. Details of the numerical models developed to simulate the fire behaviour of FRPwrapped reinforced concrete columns and FRP bar-reinforced concrete slabs are given in Chapters 5 and 6, respectively, as are results of validation and parametric studies. Additional information pertinent to the development of the numerical models is presented in the appendices, particularly Appendix A. Chapter 7 gives conclusions from this study, as well as a series of recommendations for further work in this area.

8



A considerable amount of supplementary information is included in the appendices of this thesis. Appendix B provides a series of design charts that can be used to evaluate the fire endurance of FRP bar or grid-reinforced concrete slabs. Appendix C contains a detailed metaanalysis and comparison of various currently available models for the stress-strain behaviour of FRP-confined concrete columns, and presents new confinement models based on a least squares regression analysis of the test data assembled from the literature. Appendix D takes a detailed look at the thermomechanical behaviour of carbon, glass, and aramid FRPs at high temperature based on data available in the literature, and provides a series of semi-empirically derived analytical equations that can be used for preliminary modelling of FRP-reinforced concrete members at high temperature.

9


CHAPTER 2: Literature Review

CHAPTER 2 LITERATURE REVIEW 2.1 General As discussed in Chapter 1, the deteriorated and/or under strength state of the world’s reinforced concrete buildings and infrastructure has forced the development and implementation of new and innovative repair and rehabilitation techniques for reinforced concrete structures. Additionally, the civil engineering industry is being forced to examine new materials and methods of construction to ensure that new structures are capable of living long and healthy service lives. Fibre-reinforced polymers (FRPs), a relatively new class of materials in Civil Engineering applications, have received widespread attention in recent years, both for repair and rehabilitation and for new construction of reinforced concrete structures.

As mentioned

previously, there is a concern that structures strengthened and/or reinforced with FRPs may be severely damaged by exposure to high temperatures such as would be expected in the case of fire. This chapter presents a brief review of the literature with regard to FRPs as a class of materials, their applications in civil engineering, and some of the concerns associated with their behaviour in fire when used as reinforcement for concrete. Also included is a discussion of the current philosophy and methodologies used to evaluate the fire endurance of reinforced concrete members.

2.2 Fibre-Reinforced Polymers 2.2.1 General FRPs are not new materials; they have been used in the automotive and aerospace industries for over 40 years, where their high-strength and light-weight can be used to great advantage. In the last fifteen years or so, the civil engineering community, spurred by soaring

10



infrastructure repair and replacement costs, has begun to use FRPs in a variety of structural applications. FRPs are a subgroup of the class of materials referred to as composites. Composites are defined as materials created by the combination of two or more materials on a macroscopic scale, to form a new and useful material with enhanced properties that are superior to those of its constituents (Neale and Labossière, 1991). An FRP is essentially a two-component material, consisting of high strength fibres embedded in a polymer matrix. The study of FRPs is invariably complicated by the innumerable combinations of materials that can be combined to create an FRP composite. This is both an advantage and a disadvantage for FRP as an engineering material. FRPs can be tailored to suit virtually any application. However, this versatility leads to an enormous range in properties, making it extremely difficult to arrive at generalizations with regard to a number of important issues; behaviour at elevated temperatures is just one example.

2.2.2 Composition and Properties of Fibre-Reinforced Polymers Because FRPs are generally composed of two distinct materials, material properties depend primarily on those of the constituents. It is instructive to look at the role and properties of each component separately before examining the properties of the composite material as a whole. Fibres The fibres provide the strength and stiffness of an FRP. Because fibres used in most structural applications are continuous and are oriented in specified directions, FRPs are orthotropic, and much stronger and stiffer in the fibre direction(s). Fibres are selected to have high stiffness, high ultimate strength, low variation of strength between individual fibres, stability during handling, and uniform diameter (Hollaway, 1990). For structural applications, fibres are also characterized by extremely large length-to-diameter ratios (they are generally considered continuous) and by near crystal size diameters (Neale and Labossière, 1991). The small diameter

11



of the fibres is significant in that crystals in the material are aligned along the length of the fibre, giving them high tensile strength, and in that the probability of a sample of material containing a flaw large enough to cause brittle failure decreases with its volume. Hence, microscopic fibres have fewer defects than the bulk fibre material (Jones, 1975). In the event of a single fibre break, force transfer to adjacent fibres through shear stresses that develop in the polymer matrix prevents failure of the overall FRP composite. It is interesting to consider that this force transfer (required to prevent overall failure) depends primarily on the shear strength of the matrix material, which is severely degraded at high temperature. In civil engineering applications, the three most commonly used fibre types are carbon, glass, and to a lesser extent, aramid (Rostasy, 1993). While the various fibre types all have advantages and disadvantages, carbon fibres are rapidly becoming the material of choice for structural composites for rehabilitation, due mainly to their high stiffness, low relaxation, and superior fatigue and durability characteristics. A complete review of fibre types and properties is avoided here, since this information is available in most composite materials textbooks (e.g. Schwartz, 1997). A more complete discussion of fibre properties as they relate to FRP behaviour at high temperature is presented in Section 2.4.4.2. Matrix The matrix is the binding material for an FRP and generally has poor mechanical characteristics. Matrix materials are required to support and protect the fibres, to transfer and distribute forces to the fibres, and to disperse and separate the fibres (Hollaway, 1990). Matrix materials should also be as thermally compatible as possible with the fibres to reduce the magnitude of stresses resulting from differential thermal expansion, although in most cases the coefficients of thermal expansion of the fibres and matrices actually differ substantially. A major selection criterion for matrix materials is that they have a low density, usually considerably less than the fibres, such that the overall weight of the composite is minimized (Neale and Labossière, 1991).

12



Matrix materials for FRPs can be grouped into two broad categories: thermoplastics and thermosetting resins. Thermoplastics include such polymer compounds as polyethylene, nylon, and polyamides, while thermosetting materials include epoxies, polyesters, and vinyl esters. Thermosets are used almost exclusively in structural applications (Blontrock et al. 1999). They are actually low molecular weight liquids with very low viscosities. They have good thermal stability (at service temperatures) and chemical resistance, and low creep and relaxation in comparison with thermoplastics. However, thermosets generally have a short shelf life after mixing, lower strains at failure, and lower impact strengths than thermoplastics (El-Hacha, 2000). Both thermoplastics and thermosets are characterized by low thermal conductivities (Cengel, 1998). In terms of the fire behaviour of FRPs, it is the polymer matrices that are the more problematic component. Polymers are extremely sensitive to temperature (refer to Figure 2.1) and rate of loading. Thus, at elevated temperatures deterioration of matrix mechanical properties and creep, neither of which are fully understood (Neale and Labossière, 1991), can be a serious concern.

Again, a detailed discussion of matrix properties is avoided here, but additional

information on matrices relevant to the behaviour of FRP in fire is presented in Section 2.4.4.1. Fibre-Reinforced Polymers Although the strength and stiffness of an FRP are governed by the fibres, the overall material properties depend also on the mechanical properties of the matrix, the fibre volume fraction, the fibre cross-sectional area, the fibre orientation within the matrix, and the method of manufacturing (Jones, 1975). The current work focuses exclusively on unidirectional FRPs. It is the interaction between the fibres and the matrix that gives FRPs their unique physical and mechanical characteristics. A wide variety of FRP formulations are available for use with reinforced concrete structures. Glass and carbon are the two most commonly used fibres in North America, and matrices are generally epoxies or vinyl esters. Aramid fibres and polyester resins are used very

13



occasionally (ISIS, 2001a, b). Glass is widely used because of its comparatively low cost, and because there is historically much more experience with it (El-Hacha, 2000). However, glass fibres have demonstrated certain significant disadvantages, such as a relatively low elastic modulus and durability concerns in alkaline environments (Uomoto, 2001). These disadvantages have made carbon FRPs, with elastic moduli that compare more closely with steel, more attractive, even given their considerably higher cost. The primary concerns with aramid FRPs are that they are sensitive to creep (Hollaway, 1990; Uomoto, 2001) and have displayed poor durability characteristics resulting from their propensity for moisture absorption (Braimah, 2000; Uomoto, 2001). Aramid fibres also perform very poorly at high temperature. Figure 2.2 shows the stress-strain behaviour of several FRP materials. Also included in the figure is a stress-strain curve for mild steel. Some commonly available FRPs and their properties are listed in Table 2.1. It is evident that both glass and aramid FRPs have moduli that are considerably less than steel in the pre-yield zone, but that carbon FRPs have moduli that are comparable to, or even higher than, steel in some cases. Also evident is that FRPs have ultimate strengths that can be many times greater than steel. The most significant characteristic of the stress-strain behaviour of FRPs as compared with steel is their linear elastic nature. FRPs do not display the post-yield behaviour observed with steel, and generally have much lower strains at failure. Consequently, the safe use of FRPs as reinforcement for concrete is complicated by the lack of ductility inherent in the materials. It has been shown however, that adequate ductility can be obtained for FRP-reinforced concrete members provided they are properly detailed using the concept of deformability (ACI, 2002; ISIS 2001a,b). Table 2.2 gives a qualitative comparison of the three main types of FRPs based on a number of important criteria.

14



2.3 Fibre-Reinforced Polymers in Civil Engineering Applications 2.3.1 Overview Reinforced concrete continues to be one of the most popular construction systems for buildings, bridges, and other infrastructure worldwide. The most pressing problem reinforced concrete structures face is that electrochemical corrosion of conventional reinforcing steel is difficult to prevent, and can cause severe deterioration of concrete structures through expansion and cracking, which invariably leads to spalling of the concrete cover and/or loss of tensile reinforcement. FRPs are non-corrosive and thus present a potential alternative to steel as tensile reinforcement for concrete structures.

FRPs have also demonstrated enormous potential as

materials for repair and rehabilitation of existing corrosion damaged and/or under strength concrete structures (Meier, 2000; Munley and Dolan, 2001; Neale, 2000; Taly and GangaRao, 1999). Various techniques for using FRPs in lieu of (or in combination with) steel for reinforced concrete construction have been gaining acceptance in the civil engineering community. Some examples of FRP applications in reinforced concrete repair, rehabilitation, and construction include: external post-tensioning of damaged or under strength concrete girders (Burgoyne, 1993; El-Hacha, 1997), plating for flexural or shear strengthening of reinforced and/or prestressed concrete beams (ACI, 2002; Deniaud and Cheng, 2000; El-Hacha et al., 2001; Hazen et al., 1998; Ichimasu et al., 1993; ISIS, 2001a,b; Labossière et al., 2000; Labossière et al., 1997; Meier et al., 1992; Nanni, 1997; Rostasy et al., 1992; Steiner, 1996; Walser and Steiner, 1997), wrapping for confinement and/or ductility enhancement of reinforced concrete columns (ACI, 2002; Challal and Shahawy, 2000; Fam and Rizkalla, 2001; ISIS, 2001a, b; Karbhari and Gao, 1997; Lavergne and Labossière, 1997; Lee et al., 2000; Liu and Foster, 1998; Mirmiran and Shahawy, 1997; Mirmiran et al, 1998, 1999, 2000; Monti et al., 2001; Parent and Labossière, 2000; Purba and Mufti, 1999; Saaman et al., 1998; Sheikh and Yau, 2002; Spolestra and Monti, 1999; Theriault

15



and Neale, 2000, 2001; Xiao and Wu, 2000), and internal reinforcement for shear, nonprestressed, and prestressed reinforcement for new concrete structures (ACI, 2002; Benmokrane et al, 2000; CSA, 2001a, 2002; Hassan et al, 2000; ISIS, 2001a, b; Shehata et al., 2000; Tadros, 2000). In some cases, FRPs can be used in combination with fibre-optic sensors (FOS) to create smart structures and allow for remote monitoring, which has recently received widespread attention (Benmokrane et al., 2000; Meier, 2000). The advantages of FRPs over conventional reinforcing steel are cited often in the literature and include: FRPs are non-corrosive, although they may be susceptible to other forms of equally damaging environmental distress, usually caused by elevated temperatures and/or moisture high strength-to-weight ratios, as much as 10 to 15 times greater than steel excellent fatigue characteristics, particularly carbon FRPs electromagnetic neutrality, which can be extremely useful in some special structures high tensile strength as compared with steel rapid and easy installation, significantly lowering construction costs and downtime. There are however, a number of disadvantages to using FRPs. Some of the most pressing concerns include: high material cost, although prices have dropped drastically in the past 10 years as use has increased (El-Hacha, 2000) low strain at failure, requiring new design approaches and raising concerns over insufficient member ductility extremely low lateral load capacity due to the relatively poor mechanical properties of the matrix excessive creep and relaxation in some cases, particularly for aramid FRPs

16



the potential for ultra-violet (UV) degradation of polymer matrices in external applications expansion and deterioration due to moisture-absorption, particularly for aramid FRPs lack of clear guidelines for design with and use of these new materials, although this problem has been addressed recently by several organizations (ACI, 2002; CSA, 2001a, 2002; ISIS, 2001a, b). rapid and severe loss of bond, strength, and stiffness at elevated temperatures, as would be expected in the case of fire. The final disadvantage listed is extremely significant and is at the core of the work presented in this thesis.

Sorathia et al. (2001) presented a gap analysis for durability of FRP

composites for civil infrastructure, in which they included a chapter on the effects of fire. They state that an enormous amount of additional information is required in terms of: flame spread, fire endurance, smoke generation and toxicity, and heat release, before many infrastructure applications using FRPs can be implemented safely and with confidence. Munley and Dolan (2001) point out that fire endurance uncertainties associated with the use of FRPs may be addressed, for the time being, through a rational set of design limitations. This approach has indeed been taken by ACI Committee 440 (ACI, 2002), whose design guidelines state that the load factors for a structure that is retrofitted with bonded FRP reinforcement be set to 1.2 and 0.85 for dead and live loads respectively. It is intended that these load factors prevent collapse by assuming that the FRP is lost completely during a fire. The ACI 440 documents further states that the existing strength of a structural member with a specified fire-resistance rating should satisfy fire endurance requirements under the contemplated increased loads if it is to be strengthened with an FRP system. The service load effects in this case should be determined using the current load requirements for the structure. If the FRP system is meant to allow greater load carrying capacity, such as an increase in live load, the load effects should be computed using

17



these greater loads (ACI, 2002). However, while such an approach is certainly conservative, it makes FRP materials less cost-effective as they otherwise might be. Munley and Dolan also state that the professional community would be far more comfortable with a set of full-scale fire endurance tests, which is precisely what is undertaken in this thesis. Although FRPs have been used in a wide variety of applications for reinforced concrete construction, repair, and rehabilitation, the focus in this thesis is on two specific, yet fundamentally different applications: FRP bars for internal reinforcement of concrete slabs (in lieu of conventional reinforcing steel), and FRP-wrapping for confinement (a rehabilitation and retrofit technique) of reinforced concrete columns. Both applications are discussed in detail in the following sections.

2.3.2 FRP Bars for Beam and Slab Reinforcement Because of the widespread and costly problems associated with corrosion of reinforcing steel in concrete structures, a great deal of research in recent years has focused on the use of FRP bars as a substitute for conventional tensile and temperature and shrinkage reinforcement for concrete beams and slabs (Benmokrane and Rahman, 1998; Benmokrane et al, 2000; Neale and Labossière, 1992; Saadatmanesh and Ehsani, 1996; Taerwe, 1995). Recently, there have been several field applications using FRPs for non-prestressed slab reinforcement in bridge decks (ACI, 2002; Benmokrane et al., 2000; Hassan et al., 2000). Some of the initial factors that were considered in investigating the suitability of FRP bars, rods, and grids, for slab and beam reinforcement have included: creep (Ando et al., 1997; Currier and Dolan, 1995; Matthys and Taerwe, 1998; Plevris and Triantafillou, 1994; Rostasy, 1988; Saadatmanesh and Tannous, 1999a, 1999b; Seki et al., 1997; Sen et al., 1999; Uomoto, 2001), fatigue (Adimi et al., 1998; Hayes et al., 1998; Rahman and Kingsley, 1996; Rahman et al., 1997; Rostasy, 1997; Saadatmanesh and Tannous, 1999a; Uomoto, 2001; Yagi et al., 1997), durability (Benmokrane and Rahman, 1998; Hamilton and Dolan, 2000; Uomoto, 2001), bond

18



and development (Bakis et al., 1998a, b; Bank et al., 1998; Freimanis et al., 1998; Katz, 1998, 1999, 2000; Nanni et al., 1997), and failure modes and ductility (Jaeger et al., 1995; Newhook et al., 2000), to name only a few. An exhaustive review of the literature in this area is not presented here, because it is now relatively widely recognized that internal FRP reinforcement can be used for concrete slabs and beams, provided that certain design guidelines are adhered to. FRP barreinforced concrete structures can be designed with the aid of draft design guidelines such as the Guide for the Design and Construction of Concrete Reinforced With FRP Bars (ACI, 2002), Reinforcing Concrete Structures with Fibre-Reinforced Polymers (ISIS, 2001a, b), The Canadian Highway Bridge Design Code (CSA, 2001a), or CSA-S806: Design and Construction of Building Components with Fibre-Reinforced Polymers (CSA, 2002). Assuming that durability, fatigue, bond and development, and creep rupture problems can be avoided, generally through using appropriate FRP fibre/matrix formulations and by limiting service stress levels in the FRP reinforcement – sometimes severely – the major concerns during a fire when reinforcing for flexure with FRP bars as opposed to steel are: Reduction in Strength and Stiffness:

The strength and stiffness of FRP materials

decreases with increasing temperature. When the matrix is called upon to transfer loads, the strength will be drastically reduced at temperatures of about 300 C for most conventional FRP materials, with decreased stiffness well below that. Loss of Bond: Because of the susceptibility of matrix materials to high temperature, loss of bond will occur for fire exposed composite bars and wraps within minutes. In cases where bond is relied upon to transmit shear stresses, the FRP will be rendered ineffective at temperatures below 300 C (Blontrock et al., 2000). In cases where the bond is not critical to the performance of the FRP, as would be the case in a filament wound concrete column or a multi-bay concrete slab, loss of bond is not as critical (assuming anchorage is provided, see below). Loss of bond in this case would essentially create an unbonded

19



FRP that would still be able to provide significant confinement or tensile reinforcement (Fardis and Khalili, 1982). Supplemental insulation: Tests on FRP-plated reinforced concrete beams (Blontrock et al., 2000) have shown that, when the anchorage of the FRP material can be maintained through fire protection in the form of insulation, the FRP reinforcement will perform in a similar manner to the case of an FRP insulated along its full length.

Thus, it is

conceivable that a concrete member reinforced or wrapped with continuous fibres could maintain some increased level of strength up to temperatures well above 1000 C (for carbon FRPs), as long as loss of bond is prevented (possibly by insulating only the anchorage regions against fire). Insulating Effect: One factor that should be considered is the potential insulating effect of an externally applied composite wrap on the reinforced concrete member. FRPs generally have very low transverse thermal conductivities (Mallick, 1988) and would thus tend to insulate the reinforced concrete member, possibly increasing its fire endurance. Potential for Increased Spalling: Tests on the effect of FRP wrapping on corrosion rates in reinforced concrete columns (Debaiky, 2000) have indicated that the FRP wraps form an impermeable membrane around the column, preventing the transport of moisture and oxygen into (or out of) the concrete.

While an advantage in terms of corrosion

prevention, this is a concern in fire. As a concrete member is heated in fire it will expel moisture when its temperature reaches approximately 100 C. If moisture is prevented from exiting the column, pressure will be developed internally which could potentially lead to explosive failures when the wrap eventually loses its confining capabilities. Development of thermal stresses resulting from differential thermal expansion between the concrete and the FRP can be a concern. Minimum concrete covers are specified to ensure that cracking will not occur as a result of differential thermal expansion.

20



However, the current draft codes consider thermal expansion due to temperature differences that could be expected under normal service conditions, and no mention is made of the extreme temperature differentials that could be expected in a fire. Fire endurance concerns are commented on in several of the draft codes. The approach for fire design in conventional (steel-reinforced) slab design is to specify required concrete covers for particular fire endurances based on thermal protection for the tensile reinforcement. In the case of steel-reinforced concrete the design guidelines are based on the observation that steel loses about 50% of its room temperature yield strength at 593ºC – its critical temperature. For FRPs it is difficult to establish the acceptable critical temperatures given the wide variety of FRP materials currently available. Further testing will be required in this area before simple design charts can be used with confidence.

2.3.3 Column Strengthening with FRP Wraps While the above section discussed FRP bars for new construction applications, bonded repair systems using FRPs, such as bonded plates for external strengthening or column wrapping for confinement, currently show the greatest potential for the use of FRPs in the construction industry. Indeed, such systems are currently competing economically with conventional repair and retrofit solutions (Munley and Dolan, 2001). This section examines applications of FRP sheets for strengthening reinforced concrete columns. In this technique, FRP sheets are applied to the exterior of reinforced concrete columns in either the longitudinal direction (to provide additional flexural capacity) or in the circumferential direction (to provide additional confining reinforcement which increases both the ductility and the compressive strength). The discussion in this section focuses on column wrapping. Although steel casings have been widely used for confinement of concrete columns, they are difficult to install, heavy, and prone to corrosion. Steel encasements are also isotropic (the casing ends up taking a significant portion of the axial load) and have a high Poisson’s ratio

21



(resulting in separation of the tube from the concrete under axial load (Karbhari and Gao, 1997)). FRPs offer an ideal alternative material, since they are non-corrosive, lightweight, easy to install, and orthotropic, with negligible strength in the direction perpendicular to the fibres. Over the past ten years, numerous research studies and field applications have demonstrated that FRP-wrapping can significantly increase both the strength and ductility of reinforced concrete columns (ACI, 2002; Callery, 2000; Challal and Shahawy, 2000; Demers and Neale, 1994; Fam and Rizkalla, 2001; Hosotani et al., 1997; ISIS, 2001a, b; Karbhari and Gao, 1997; Lavergne and Labossière, 1997; Lee et al., 2000; Liu and Foster, 1998; Mirmiran and Shahawy, 1997; Mirmiran et al, 1998, 1999, 2000; Monti et al., 2001; Parent and Labossière, 2000; Pilakoutas and Mortazavi, 1997; Saadatmanesh et al., 1994; Saaman et al., 1998; Santarosa et al., 2001a, b; Sheikh and Yau, 2002; Soudki and Green, 1997; Spolestra and Monti, 1999; Suter and Pinzelli, 2001; Theriault and Neale, 2000; Theriault et al., 2001; Watanabe et al., 1997; Xiao and Wu, 2000). The observed increases in strength and ductility can be attributed to the fact that the FRP wrap generates a confining pressure on the dilating concrete core, which places the concrete in a triaxial state of stress and increases both the ultimate strength and strain of the concrete in compression by reducing shear stresses and controlling crack initiation. Numerous analytical models for the stress-strain behavior of FRP-confined concrete are available in the literature (Cusson and Paultre, 1995; Fam and Rizkalla, 2001; Fardis and Khalili, 1981, 1982; Hoppel at al. 1997; Karabinis and Rousakis, 2001; Karbhari and Gao, 1997; Lam and Teng, 2001; Manfredi and Realfonzo, 2001; Mirmiran and Shahawy, 1997; Miyauchi et al., 1997; Parent and Labossière, 2000; Saafi et al., 1999; Saaman et al., 1998; Spolestra and Monti, 1999; Toutanji, 1999; Vintzileou, 2001; Xiao and Wu, 2000), and design guidelines for confinement of reinforced concrete columns have recently been published by both the American Concrete Institute (ACI, 2002) and ISIS Canada (ISIS, 2001a). The following section outlines the key concepts to which one must remain cognizant when modelling confined concrete. One recently developed model, that is particularly useful for

22



modelling FRP-confined concrete at stress levels less than ultimate, is discussed in more complete detail, because it has been modified and used in the computer programs developed later in this thesis. The reader is encouraged to consult Appendix C for a more complete discussion of confinement models. Early attempts at describing the behaviour of concrete under a state of hydrostatic (triaxial) stress demonstrated that the following empirically derived equations could be used to describe the peak confined stress, f cc' , and strain, ε cc :

f cc'

ε cc

f co'

k1 f 1

ε co 1 k 2

f1 f co'

[2.1]

[2.2]

where f co' is the peak stress of the unconfined concrete, f1 is the lateral (confining) pressure, ε co is the unconfined strain at ultimate, and k1 and k2 are empirically determined coefficients (Karbhari and Gao, 1997). These equations have been used extensively to estimate the ultimate stress and strain of concrete columns confined with steel (Ahmad and Shah, 1982; Chan, 1955; Cusson and Paultre, 1995; Frangou et al., 1995; Furlong, 1967; Iyengar et al. 1970; Knowles and Park, 1969; Kotsovos and Perry, 1986; Morales et al., 1982; Richart et al., 1929). Richart et al. (1929) initially found that k1 = 4.1 and k2 = 5k1 for concrete under hydrostatic fluid pressure. In the years since, studies have been conducted on a variety of concrete mixes and lateral reinforcement ratios, and values of k1 from 2.8 to 7.0 have been suggested. A variety of modifications to the above formulas to account for the arrangement and effectiveness of spiral or tied lateral steel reinforcement have also been suggested in the literature (Karbhari and Gao, 1997). In any case, the mechanism of confinement when using FRP-wraps is fundamentally different to that arising from transverse reinforcing steel, and a somewhat different modelling approach is required.

23



In the development of confinement models for steel, it has traditionally been assumed that the confining steel yields, and that the confinement pressure is thus constant throughout the load history of the member. An FRP wrap, which displays a linear elastic stress-strain response, will apply a varying confinement pressure. This varying confinement pressure depends on the strain in the FRP, which in turn depends on the dilation of the concrete core, which again depends on the confinement pressure. It is evident that the active nature of FRP confinement complicates the analysis significantly, and forces an iterative approach to determine the stress-strain response of confined concrete in compression.

Figure 2.3 shows the enhanced load-deflection

characteristics of FRP-confined concrete under uniaxial compression. Appendix C gives a comprehensive review and comparison of the various existing models for FRP-confined concrete.

These models are largely empirical and are generally

applicable only to hand calculation. The primary shortcoming of the FRP confinement models described in Appendix C is that many assume a constant confining pressure, and can predict, at best, a bilinear approximation to the stress-strain response of an FRP-wrapped concrete column. Indeed, several give only the failure stress and strain for FRP-confined concrete and make no attempt to describe behaviour at service load levels. Often, as in the case of the load capacity analysis for fire-exposed columns (developed in Chapter 5 of this thesis), the complete stressstrain response of FRP-confined concrete is required. To address this problem, Spolestra and Monti (1999) developed a rational iterative procedure to determine the complete stress-strain response of a circular concrete column confined by an active FRP wrap. Based on a simple constitutive model for unconfined concrete under uniaxial loading presented by Pantazopoulou and Mills (1995), the Spolestra and Monti procedure takes into account the interaction between the confining pressure and the dilation of the concrete core. The model is implemented as follows: An overall axial compressive strain, ε c , is assumed in the concrete.

24



The lateral confining pressure, fl , is assumed. The peak confined concrete strength, f cc' , is calculated based on a confined concrete relationship presented previously by Mander et al. (1988):

f cc'

f co' 2.254 1 7.94

fl f co'

2

fl f co'

1.254

[2.3]

The current concrete stress, f c' , is determined using the Popovics (1973) stress-strain relationship for concrete:

f cc' xr r 1 xr

f c'

[2.4]

where,

εc

x

ε co

f 1 5 f

' cc ' co

and

Ec

r Ec

1

f cc'

[2.5a, b]

ε co

and in the above expressions, ε co is the strain corresponding to the peak stress of the unconfined concrete, generally taken as 0.002, and E c is the initial tangent elastic ' modulus of the unconfined concrete, taken as 5700 f c (Spolestra and Monti ,1999).

The lateral strain in the concrete is calculated according to the recommendations of Pantazopoulou and Mills (1995), with some manipulation, to give:

εl

Ec ε c f c' 2 βf c'

[2.6]

where

β

5700 f c'

500

[2.7]

The lateral confinement pressure is updated using compatibility of strains between the laterally expanding concrete and the circumferentially strained FRP wrap:

25



fl

2tEcom ε l

[2.8]

If this value for the lateral pressure is significantly different from the value assumed, then the lateral confining pressure is updated with the new value and the procedure is repeated. The Spolestra and Monti model is particularly useful in the current study, in that it provides confinement pressures for any assumed axial compressive strain in the concrete, which becomes important in the fire endurance analysis presented later in this thesis. It has also been used in moment curvature analyses of FRP-wrapped bridge piers with great success (Monti and Spolestra, 2000; Monti et al., 2001). Because many FRP confinement models have been presented in the literature, it was decided to perform a meta-analysis to evaluate the validity of the various models based on a database of experimental data compiled from the literature over the past 30 years. The results of the study are discussed in detail in Appendix C. Essentially, it was determined that, while the Monti and Spolestra (2000) model does not give the best least-squares fit with experimental database, it is certainly an adequate predictor of ultimate stress and strain for confined concrete, and it is useful in that gives the compete stress-strain response while accounting for the confinement provided by an FRP wrap. It should be noted that a number of finite element studies on FRP-confined concrete have also been reported on in the literature (Liu and Foster, 1998; Parent and Labossière, 2000). These studies have focused on the development of constitutive relationships for FRP-confined concrete, in a manner similar to the non-iterative methods discussed in Appendix C. While the results of finite element studies have generally agreed well with experimental results from tests conducted by the authors, these models are highly specialized and are not readily applicable to many different cases nor to hand calculation.

26



2.3.4 Economic Considerations In many cases, especially in repair and rehabilitation applications, FRPs are the only viable material. In some cases, however, as in new construction or in certain external plating applications, a choice must be made between FRPs and more conventional techniques. FRP materials are more costly on a per-weight basis than steel. However, the costs of FRPs have dropped drastically in the past few years, and this trend should continue as use of these materials becomes more widespread. For example, Buyukozturk et al. (1999) reported that the price of raw carbon fibres dropped from US$ 20/lb to US$ 6.50/lb in the years between 1990 and 1999, with projections for a cost of US$ 5/lb by the end of 2000. It is also important, in today’s economic climate, to consider factors other than capital cost when selecting materials for construction. If a comparison is made on a cost per force basis, then the cost discrepancy diminishes due to the extremely high strength-to-weight ratios of FRP materials (El-Hacha, 2000). From a construction point of view, costs can also be reduced when using FRPs due to the lightweight nature and ease of application, which can significantly reduce both construction time and costs associated with downtime of the structure in repair applications. Finally, life cycle costs should be considered wherever possible. Structures built using FRPs are presumably non-corrosive and should have service lives that exceed those of conventionally reinforced concrete structures, with repair and maintenance costs that are also considerably less. When FRPs can be combined with fibre optic sensor (FOS) technologies, intelligent sensing, and remote monitoring, inspection and servicing costs can also be reduced.

2.4 Material Properties at Elevated Temperature 2.4.1 General An understanding of material behaviour at high temperature is essential to experimentally or analytically investigate the fire endurance of structural members. The properties that are of interest for structural materials can be divided into two broad categories: thermal and mechanical.

27



Thermal properties include: thermal conductivity, specific heat, emmissivity, and density; while mechanical properties include: thermal expansion, creep, and stress-strain behaviour. The above mentioned properties are discussed in the following sections as they apply to concrete, steel, and FRPs. It should be noted that both steel and concrete do not combust, and hence will not contribute fuel to a fire, evolve toxic gases, or generate smoke. This is not generally true in the case of FRPs, most of which are combustible.

2.4.2 Concrete The behaviour of concrete at elevated temperature has been extensively studied in the literature. Comprehensive reviews are presented by Lie (1992) and Khoury (2000). The most important information with regard to fire endurance is summarized below.

2.4.2.1

Thermal Properties

The thermal properties of concrete vary widely with temperature, and are dependent to a great extent on the type of aggregate used, since different rock types with different mineralogical composition react differently to heat. In order to perform heat transfer calculations for concrete members, approximate information on the variability of density, thermal conductivity, and specific heat of concrete with temperature is required.

Density The density of concrete shows only a slight dependence on temperature (Schneider, 1988), as shown in Figure 2.4. This is due primarily to the effects of moisture loss during heating. Concretes with carbonate aggregates experience a significant decrease in density at temperatures of about 800 C due to thermal degradation of carbonate minerals in the aggregate.

Thermal Conductivity Thermal conductivity refers to the ability of a material to transfer thermal energy by conduction from a region of high temperature to a region of lower temperature. High thermal

28



conductivity will result in rapid heat transfer, causing internal temperatures in a member to rise quickly, and low thermal conductivity will have the reverse effect. In general, the thermal conductivity of concrete decreases with increasing temperature. The initial value of the thermal conductivity and the magnitude of the decrease with increasing temperature depend on the degree of crystalinity of the aggregate used, and on the moisture content. Figure 2.5 shows the thermal conductivity of normal and lightweight concretes as reported by Lie (1972). These data agree well with those given in other sources (Lie, 1992; Schneider, 1986). Figure 2.5 also shows idealized thermal conductivity curves for concrete (as used in Chapter 5 for numerical modelling) based on equations presented by Lie (1992).

Specific Heat The variation in the specific heat of concrete with increasing temperature is shown in Figure 2.6. The specific heat is influenced both by the character of the cement used and the aggregate type, and may also be influenced by the amount of moisture in the concrete. Schneider (1986) states that the specific heat of concrete is the property of concrete that is least understood, because heating of concrete to high temperature is accompanied by numerous chemical reactions. Differences in the specific heats for different concretes may be caused by the latent heat of the different reactions that occur during heating. These may include: water release, dehydration, decarbonization, and quartz inversion. Below temperatures of 600 C, there is no clear effect of aggregate type on the specific heat of concrete. Above 600 C however, concrete with carbonate aggregate experiences a rapid rise in specific heat due to decarbonization. The amount of cement paste in the mix will influence the specific heat in that rich mixes experience a higher latent heat caused by dehydration effects. The water content in the concrete is extremely important at temperatures below 200 C, because evaporation of water in concrete at 100 C causes an apparent specific heat that is about twice that of oven dried concrete. This can be taken into account when modelling by accounting for the latent heat of evaporation of water (the approach that has been

29



taken in this thesis), or by artificially increasing the specific heat of concrete in the temperature range of 100ºC to 200 C. For the purposes of numerical modelling, approximate equations have been developed to describe the variation in thermal capacity of concrete – defined as the product of the specific heat and density – with temperature for concretes incorporating different aggregate types. These are shown in Figure 2.7.

2.4.2.2

Mechanical Properties

Strength The strength of concrete at elevated temperatures depends primarily on the type and amount of aggregate and the level of stress. Greater proportions of aggregate result in smaller strength loss at high temperature, and the presence of applied load also tends to decrease the strength loss. It has been found (Schneider, 1988) that the stressed strength at high temperature is higher than the unstressed strength at the same temperature, but that the magnitude of applied stress has little effect. Strength loss occurs at higher temperatures for concretes with carbonate or lightweight aggregates. Figure 2.8a shows the results of strength loss tests at high temperature conducted by a number of authors. The temperature dependency of concrete strength has been idealized for numerical modelling by Lie (1992) and is shown in Figure 2.9. Some additional comments on the compressive strength of concrete at high temperature are provided by Schneider (1986): the original strength and water-cement ratio do not significantly affect strength loss at high temperature; the type of cement has little effect on strength loss at high temperature; the maximum aggregate size appears to have a minor effect on strength loss at high temperature; the rate of heating has little effect on strength loss so long as temperature gradients are maintained below 10 C/mm; residual strengths obtained after heating and cooling are less than values for concretes tested at elevated temperature.

30



Modulus of Elasticity There is a significant decrease in the modulus of elasticity of concrete with increased temperature, as shown in Figure 2.8b. Modulus decreases with increasing temperature, and the rate of decrease is slightly dependent on aggregate type. It has been observed in tests (Schneider, 1988) that: the original strength of the concrete and the water-cement ratio do not significantly affect the changes in modulus at high temperature; the type of cement has little effect; and that stressed elasticities are always higher than unstressed elasticities.

The variation in elastic

modulus of concrete with temperature has been idealized for numerical modelling and is presented in Figure 2.9.

Stress-Strain Behaviour The overall changes in stress-strain (σ-ε) behaviour of concrete at high temperature are shown in Figure 2.10. It has been observed that (Schneider, 1986, 1988): The original concrete strength and water-to-cement ratio do not significantly influence the shape of the σ-ε curve. The aggregate-cement ratio has a significant effect on the initial slope of the σ-ε curve. Rich mixes display a lower initial slope and less curvature than normal concretes. The type of aggregate is the main factor influencing the shape of the σ-ε curve. Concretes with harder aggregates generally have a more pronounced decrease of the initial slope with increasing temperatures. The ultimate strain is essentially independent of the aggregate type. Curing conditions influence the shape of the σ-ε curve at temperatures below 300 C. The initial slopes of σ-ε curves and the ultimate stresses are lower for specimens cured underwater than for those cured in air. If the concrete is heated under a sustained load, specimens indicate a significant relative increase of compressive strength and modulus of elasticity, and a significant decrease of

31



ultimate strain, as compared to unstressed specimens under the same conditions. The magnitude of the stress has little influence.

Thermal Expansion Figure 2.11a shows the variation in thermal expansion of concretes made with different aggregates. It is evident that the aggregate type can significantly affect both the magnitude and direction of thermal expansion, and that most concretes expand when heated. Thermal expansion is non-linear and highly influenced by the level of stress in the concrete during heating. It is essentially eliminated for concrete subjected to more than 35% of ultimate stress, as shown in Figure 2.11b.

Creep The creep behaviour of concrete is extremely complex and depends on a variety of factors including: age, moisture content, cement type, load condition, and strength. Creep of concrete is also highly influenced by both temperature and stress level. It has been shown that much larger creep strains are observed at higher temperatures, but that the role of creep is insignificant, and can essentially be ignored, in the overall behaviour of concrete at temperatures below 400 C (Khoury, 2000; Lie, 1992).

2.4.3 Reinforcing Steel The behaviour of steel at elevated temperatures has also been studied extensively in the literature. Again, comprehensive reviews are given by Lie (1992) and Khoury (2000).

2.4.3.1 Thermal Properties Thermal Conductivity Although the thermal conductivity of steel at room temperature can vary slightly based on chemical composition, these differences are not considered significant at high temperatures and all structural steels are assumed to behave the same in this regard (Lie, 1992). The thermal conductivity decreases linearly up to a temperature of about 850ºC, at which point it remains

32



constant for increasing temperature. Figure 2.12a shows the variation in the thermal conductivity (both observed and idealized for numerical modelling) of steel with temperature.

Specific Heat The specific heat of steel displays a great deal of variability with temperature up to 1000ºC.

The variation in specific heat with temperature, both observed and idealized for

numerical modelling, is shown in Figure 2.12b.

2.4.3.2 Mechanical Properties Strength As is the case for concrete, the strength of steel decreases with increasing temperature (Cooke, 1988). Figure 2.13 shows stress-strain curves for mild steel at different temperatures. It is evident that the yield strength of steel decreases with increasing temperature and that the welldefined yield plateau disappears. The yield strength decreases by about 50% at 600 C. Figure 2.9 shows the yield strength reduction of steel at high temperature as idealized for numerical modelling.

Modulus of Elasticity Increased temperature decreases the modulus of elasticity of steel (Cooke, 1988). Figure 2.14 shows the variation in elastic modulus of steel with temperature. It is evident that more than 50% of the original modulus is lost at temperatures of about 700 C. The changes in the stressstrain behaviour of steel have been idealized for use in numerical analyses and are presented in Figure 2.9.

Thermal Expansion The variation in thermal expansion of steel with temperature is given in Figure 2.15. It is evident that the coefficient of thermal expansion increases slightly up to a temperature of 600 C. Beyond 650 C, the coefficient of thermal expansion decreases to zero, at about 850 C, and then

33



begins to increase again. This variation in thermal expansion is due to thermally induced changes in the molecular structure of the steel.

Creep Creep of steel can become significant at temperatures above 450 C (lie, 1992). While a comprehensive discussion of creep in steel is avoided here, it should be pointed out that the influence of creep on fire performance may be evaluated by the use of strength values determined at a strain (or heating) rate equivalent to that achieved during fire. It may alternatively be evaluated through the use of creep equations (Harmathy, 1967). Previous modelling of the overall fire behaviour of reinforced concrete members has indicated that the effects of creep in reinforcing steel can be ignored for practical purposes (Lie, 1992).

2.4.4 Fibre-Reinforced Polymers As early as 1982 it was recognized that fire posed a significant risk to FRP-reinforced concrete members. In their pioneering work on FRP-wrapped concrete columns, Fardis and Khalili (1982) included a section that discussed various concerns associated with the flammability of the polymer matrix and the consequences for reinforced concrete structures. At that time, they suggested the use of flame retardant additives and fillers to improve the fire performance of polymer matrices, but did not attempt to improve or test fire performance themselves. It is interesting, and somewhat alarming, to note that relatively few studies have been conducted to investigate the fire resistance of FRPs for structural applications in the twenty-one years since. Two types of performance against fire are extremely important (Tanano et al., 1999): performance against initial fire and performance in the post-flashover stage. Performance against initial fire includes: flammability, which affects the spread of fire (non-combustibility and flame retardency), and smoke and gas generating properties, which affect the ability to safely evacuate a building. The performance against fire in the post-flashover stage includes: heat-insulating, flame

34



resisting, and smoke barrier properties of separating members, such as floors or walls, and structural safety (or load bearing capability) of framing members, such as columns and beams. Fibre-reinforced polymers display high temperature performance that is drastically different than steel. All polymer matrix composites will burn if subjected to a sufficiently high heat flux. In addition, commonly used matrix materials such as polyester, vinyl ester, and epoxy not only support combustion, but evolve large quantities of dense black smoke (Sorathia et al., 1992). Compared to non-filled plastics however, composites have two advantages with regard to their involvement in fire. First, the non-combustible fibres (with contents as high as 70% by weight) displace polymer resin, making less fuel available for the fire.

Second, when the

outermost layers of a composite lose their resin due to combustion, the remaining fibres act as an insulating layer for the underlying composite, significantly reducing heat penetration to the interior (Sorathia et al., 2001). The properties of FRP elements at high temperature have not been extensively studied in the literature, and what little work has been done does not give results that are readily applicable to the wide range of FRP composites used in civil engineering applications. Before detailed numerical studies on the structural fire endurance of FRP-reinforced concrete elements can be conducted with confidence, a more complete understanding of the thermomechanical properties of these materials would be beneficial. Appendix D contains a summary and analysis of data available in the literature on the high temperature mechanical behaviour of various FRP materials with temperature.

Approximate analytical equations to describe the reductions in strength,

stiffness and bond with temperature are presented and discussed. The following sections outline the results of studies performed to date to investigate the high temperature behaviour of FRP materials.

35



2.4.4.1 Matrix Behaviour The wide variety of available matrix materials and additives makes it extremely difficult to provide generalizations with respect to their high temperature behaviour.

However, this

section gives a brief synopsis of the limited work that has been done in this area. For a detailed discussion of matrix materials and their properties, the reader should consult Bakis (1993). As far as the fire endurance of FRP-reinforced concrete is concerned, some of the more important matrix properties are the thermal conductivity, the upper use temperature – or glass transition temperature (GTT), the coefficient of thermal expansion (CTE), and the flame resistance. The burning characteristics of thermoplastics and thermosets differ significantly. Sorathia at al. (1992) offer a review of the fire behaviour of different resin types used for FRPs in marine applications. They state that thermosets will degrade, thermally decompose, or char when exposed to fire, but will not soften or melt like thermoplastics. In general, thermosets burn for a shorter duration than thermoplastics, and have much higher heat release rates. Thermoplastics, on the other hand, tend to soften when exposed to high temperature due to their primarily linear chain molecular structure. Thermoplastics burn longer and release less heat than thermosets. Currently, thermosets are most often used in civil engineering applications. With respect to thermosets, Bakis (1993) states that polyesters can be made quite resistant to fire, and that their upper use temperatures are about 100 C to 140 C. Vinyl esters have advantages over polyesters in terms of high temperature resistance, with upper use temperatures in the range of 220 to 320 C. Epoxy resins, the most versatile FRP resins and subsequently the most widely used in structural applications, can have upper use temperatures anywhere from 50 C to 260 C depending on the particular formulation and resin additives. Polyamide resins, which can be either thermoplastic or thermosetting, have maximum use temperatures as high as

36



316 C. Thermoplastics can have upper use temperatures anywhere from 85 C to 277 C, but have rarely been used in structural applications to date. Probably the most important property of the matrix material, as far as fire behaviour in reinforced concrete applications is concerned, is the glass transition temperature (GTT). Drastic changes in the strength and stiffness properties of matrix materials occur at temperatures close to the GTT (Bakis, 1993). Reductions in elastic and shear modulus of a factor of 10 to 100 have been observed in a temperature interval of 10 C to 20 C around the GTT (Blontrock et al., 1999). The magnitude of the reduction in mechanical properties of a matrix at temperatures near the GTT depends very heavily on the degree of cross-linking of the polymer, and a detailed discussion of this topic is beyond the scope of this thesis. The GTT for a particular FRP is the temperature at which the amorphous polymeric regions of a material undergo a reversible change from hard and brittle to viscous and rubbery (Bank, 1993). The changes are due to changes in the molecular structure of the material. The GTT for resins used in commonly available FRPs are relatively low, generally less than 200 C, while the fibres can withstand comparatively high temperatures (more than 1000 C for carbon). Because the GTT of a polymer is specific to that material, it is virtually impossible to make generalizations with regard to safe temperature ranges for the enormous variety of FRPs currently available for structural applications. Table 2.3 provides maximum upper use temperatures for a variety of epoxy resins available in industry (Gluguru, 2000). It is evident that there is an enormous amount of variability in the data. For instance, while a general purpose clear epoxy adhesive has an upper use temperature of about 71 C, a specially designed high temperature (aluminum filled) epoxy has an upper use temperature of about 204 C. Also, we note that while the general purpose clear epoxy has lost about 75% of its shear strength at 80 C, the high temperature epoxy has lost only about 14% at the same temperature.

37



Plecnik et al. (1986) investigated the fire behaviour of epoxy resins as repair materials for concrete members. A series of tests were performed (tensile, compressive, and shear) to evaluate the high temperature mechanical properties of the resins. Their results indicated that the strength of these materials dropped off very rapidly at temperatures near the glass transition temperature, and that the strength was negligible at temperatures of 100 C greater than the glass transition temperature. Dimitrienko (1999) conducted a series of tests to determine thermomechanical properties of a variety of matrix materials. His tests, on a specific epoxy material that is not described in detail, indicate a reduction in the elastic modulus of about 50% at 150 C and about 95% at 300 C. Dimitrienko also found that the rate of reduction in strength depended on the heating rate and was greater for higher rates of heating. Numerous resin additives are available for enhancing the resistance of matrix materials to flames, smoke generation, oxidation, and heat. For example, phosphorus-based flame-retardant additives function by developing a protective char that insulates the unburned polymer from the flame.

Hydrate-based flame-retardants, such as aluminum trihydrate, undergo exothermic

reactions and release water upon heating, thereby quenching the combustion reactions. These specific additives each function differently and it is again difficult to make generalizations with respect to their behaviour. In most cases however, resin additives result in a reduction in mechanical properties of the resin material, making many of them unsuitable for FRP applications. A complete discussion of resin additives can be found in Mallick (1988).

2.4.4.2 Fibre Behaviour The three commonly used fibre types have drastically different thermomechanical properties at high temperature. Aside from a tendency to oxidize at temperatures above 400 C, some carbon fibres have shown negligible strength loss up to temperatures of 2000 C. Aramid fibres have a high thermal stability, but oxidation limits their use above 150 C. Glass fibres will not oxidize but tend to soften at temperatures in the range of 800 C to 1000 C (Bakis 1993).

38



Rehm and Franke (1979) tested the tensile strength as a function of temperature of different kinds of glass fibres. It was found that at 550 C the tensile strength was reduced to half of its value at room temperature. Rostasy (1992) conducted a series of tests to examine the effect of temperature on the tensile strength of carbon, glass, and aramid fibres. The tests indicated that the tensile strength of aramid fibres was more dependent on temperature than glass fibres, but that the tensile strength of carbon fibres seemed to be affected only slightly by temperatures up to 1000 C.

Sen et al. (1993) performed tests on a variety of different glass fibres at high

temperature. They concluded that the strength of glass fibres was reduced to about half of the room temperature value at about 550 C, and that the strength reduction was independent of the type of glass fibre being used. Sumida et al. (2001) tested the tensile strength of both carbon and aramid fibres at high temperature and determined that, while carbon fibres are unaffected by temperatures up to 300 C, aramid fibres experience an almost linear decrease in strength at temperatures above 50 C with a strength reduction of 50% at 300 C.

Dimitrienko (1999)

provides experimental data from tests on a variety of fibres at temperatures up to 1400 C. Tests were performed on carbon, glass, and aramid fibres in pure tension under exposure to elevated temperature. It was determined that carbon fibres were relatively insensitive to high temperature, with strength and stiffness actually increasing at temperatures above 600 C and up to 1400 C. Glass fibres were found to weaken and soften at temperatures above 400 C, with a reduction of 20% in both strength and stiffness at 600 C and of 70% at 800 C. Glass fibres showed negligible strength and stiffness at temperatures above 1200 C. Aramid fibres performed very poorly, with significant reductions in strength and stiffness at temperatures above 100 C. Aramid fibres demonstrated a 20% decrease in strength and stiffness at 250 C, and a 70% decrease at 450 C. Figure 2.18 shows the temperature dependence of the tensile strength of a number of different fibre materials based on tests conducted by a variety of authors. It is evident that, while all fibres seem to be affected by elevated temperatures, aramid is the most severely affected with

39



reductions of over 50% at 500 C, and carbon is the least with reductions of less than 5% at the same temperature. Sorathia et al. (1992) state that the type and quantity of the fibre will significantly influence the fire performance of an FRP composite. Glass and carbon FRPs generally smoke less, and give off less heat than those with organic fibres such as aramid. The fibre type also significantly influences the thermal conductivity of FRP, with carbon FRPs having higher thermal conductivities than glass (particularly in the fibre direction).

2.4.4.3 Thermal Properties Thermal Expansion Thermal expansion is potentially a very important factor in the fire behaviour of FRPreinforced concrete members since the coefficient of thermal expansion (CTE) of concrete may differ substantially from that of a particular FRP. The potential consequences of differential thermal expansion between FRP and concrete include spalling of concrete cover (due to the development of internal pressure when FRPs are used as internal reinforcement) or the development of shear stresses within the adhesive layer (possibly contributing to bond damage or failure) when FRPs are used in external applications. Although differential thermal expansion of concrete and FRP may not be a primary concern for the temperature ranges commonly encountered in service, the temperature variation experienced during fire could be in the order of hundreds of degrees, causing significant differential thermal expansion. Because of the concerns associated with thermal expansion of FRP, it has been studied quite extensively in the literature. The CTE of FRP materials are highly directionally dependent, and can also vary a great deal depending on the type and proportion of constituent materials present.

The CTE of

unreinforced polymers is generally higher than that of reinforcing steel (Mallick, 1988). However, the addition of fibres to a polymer matrix reduces the CTE. Depending on the fibre type, orientation, and volume fraction, the CTE of an FRP can vary widely, as shown in Table

40



2.4, which provides typical CTEs for various FRP materials. It is evident that the transverse CTEs are generally much higher than the longitudinal. This is because the longitudinal properties of a unidirectional FRP are dominated by the fibre properties, whereas the transverse properties are dominated by the matrix (the CTEs of fibres are generally much less than for commonly used polymer matrices). Hence, the concerns associated with spalling of concrete cover when FRP bars are used internally are highlighted. Rahman et al. (1993) determined the CTE for a carbon/glass/vinyl ester hybrid grid FRP reinforcement in the range of -30 C to +50 C and found it to be 8.39e-6/ C, or about 84% that of concrete. Silverman (1983) conducted tests on a series of glass FRP reinforcements with a variety of different resins, and found that the transverse (matrix dominated) CTE for the GFRP material was not constant, but rather increased with increasing temperature. Drastic increases in the CTE were observed in the 150 to 200 C temperature range, and were thought to be associated with the attainment of the glass transition temperature of the resin materials, established as 185 C by the author. Gentry and Hudak (1996) experimentally determined the CTEs of two different glass/vinyl ester composite rods and found them to be 4.8 µ / C and 8.2 µ / C in the longitudinal direction in the temperature range 0 C to 60 C and 38 µ / C and 32 µ / C in the transverse direction. This highlights both the high directional dependence of the CTE and the inconsistency in CTE values for FRPs of similar composition. Gentry and Hussain (1999) studied the thermal compatibility of concrete and composite reinforcements using both a thermoelastic solution and a numerical finite element approach. They concluded that it is likely that some cracking due to differential thermal expansion of the bars and concrete could result if small covers and bar spacings were used. Because of the potential for cracking and spalling associated with thermal expansion of FRP reinforcement, several researchers have attempted to determine the critical concrete cover required to prevent cracking. Matthys et al. (1996) performed a finite element analysis to study

41



the effects of transverse thermal expansion of FRP reinforcement. They considered aramid FRP prestressing bars and strips in a temperature range of 20 to 80 C, and determined that the critical concrete cover was about 3.5 to 5 times the bar diameter, depending on the concrete strength and the shape of the FRP element. Aiello et al. (1999) performed an evaluation of the effects of temperature variation on glass and aramid FRP-reinforced concrete elements.

Their study

confirmed the influence of temperature variations on the state of thermal strain and stress within FRP-reinforced concrete, and the necessity of a minimum concrete cover to prevent cracking. The above mentioned studies demonstrate that thermal expansion of FRP reinforcement in concrete can cause tensile stresses to develop. If these stresses are large enough and the concrete cover is small enough, cracking or failure of the cover will occur. Experimental work in this area has been limited at present to consider only temperature variations that are likely to be experienced under normal service conditions. No attempt has apparently been made to consider the effects of extreme temperature variation, as would occur in a fire, on the potential for spalling of the concrete cover. Indeed, the potential for damage under high temperatures is increased by the fact that the CTE of some FRP materials has been shown to increase rapidly at temperatures above the GTT of the matrix (Gentry and Hudak, 1996). This is a potentially very important consideration that must be examined with regard to the fire endurance of internally FRPreinforced concrete, although it has not been investigated for the purposes of this thesis.

Thermal Conductivity In the study of the high temperature behaviour of FRP bar or grid-reinforced concrete, thermal conductivity of FRPs is not a primary consideration since the amount of FRP in a concrete member will be small in comparison to the amount of concrete. Hence, its contribution to the overall heat transfer within the member will be negligible.

When used as external

reinforcement, however, the effect of wrapping or plating may be an important consideration

42



given that the low transverse thermal conductivity of FRPs may act to insulate the substrate concrete from fire. In general, polymers have comparatively low thermal conductivities (Mallick, 1988), which is one reason that polymers are used as insulating materials for wires and cables. The thermal conductivity of an FRP depends on the resin type, the fibre type and orientation, and the fibre volume fraction. For unidirectional composites used in civil engineering applications, the fibres control the longitudinal thermal conductivity and the matrix controls the transverse thermal conductivity. Some typical values of thermal conductivities for various FRP materials at room temperature are given in Table 2.5, although again, generalizations are difficult to make. Thermal conductivities of FRPs are generally quite low, with the exception being CFRPs in the fibre direction due to the high thermal conductivity of carbon fibres. Little work on the variation of thermal conductivity and other thermally important properties such as density and specific heat are available for FRPs in the literature. Griffis et al. (1984) conducted tests on a specific carbon/epoxy FRP used in the aerospace industry by subjecting specimens to radiant heat by laser irradiation up to temperatures of 3000 C. The results of these studies are shown in Figure 2.16, and indicate that the thermal and physical properties of the FRP vary a great deal with temperature. Scott and Beck (1992) also conducted tests on a carbon/epoxy FRP used in the aerospace industry. They found that the thermal conductivity of the FRP varied almost linearly from 0.77 W/m C to 0.85 W/m C between the temperatures of 30 C and 135 C.

Specific Heat The rate of heat transfer through an FRP depends to a great deal on its specific heat. Because of the complex chemical reactions that take place in an FRP at high temperature, it is extremely difficult to determine the variability of specific heat with temperature. Griffis et al. (1984) suggested the use of specific heats that varied as shown in Figure 2.16 for heat transfer

43



calculations in a carbon/epoxy FRP. In the development of this curve, the specific heat was artificially increased in the temperature range 343 C to 510 C to simulate the thermal degradation effect of the epoxy. Scott and Beck (1992) provide specific heat data that agree well with Griffis et al. (1984), although over a much smaller range of temperatures.

2.4.4.4 Mechanical Properties It is well established that there is a general deterioration in the mechanical properties of most engineering materials with increasing temperature. The limited work that has been done in the area of FRP materials at elevated temperatures suggests that the same holds true for FRP composites. Deterioration in the mechanical properties in the case of FRP reinforcement for concrete structures is extremely important, since decreases in elastic modulus and strength during a fire could lead to unserviceable deflections, loss of tensile or confining reinforcement, and eventually to collapse. Bank (1993) states that all of the mechanical properties of an FRP are functions of temperature.

The critical temperature will usually be the GTT of the matrix; although

degradation in both strength and stiffness may be observed well before the GTT is reached. Because of to the anisotropy of FRP materials, the transverse (matrix dominated) properties are more affected by elevated temperatures than the longitudinal (fibre dominated) properties. Thus, for a unidirectional FRP, the transverse strength and stiffness decrease rapidly as the temperature approaches the GTT of the polymer matrix. Gates (1991) investigated the variation in the longitudinal, transverse, and shear moduli of a carbon/thermoplastic FRP and a carbon/bismaleimide thermoset FRP in a temperature range of 23 C to 200 C. No significant trend was observed in the longitudinal modulus up to 200 C for either FRP, although the transverse and shear moduli appeared to be affected at elevated temperatures. This result agrees well with the notion that matrix-dominated properties are more severely affected by elevated temperature. It is important to note that the GTT for the resin used

44



in these tests was unusually high, quoted as 220 C. Gates also studied the stress-strain response, in both tension and compression, of the two CFRPs at elevated temperatures, and observed a significant reduction in strength with increasing temperature. The reduction was about 40 to 50% at 125 C and 80 to 90% at 200 C, which indicates extreme strength degradation at temperatures well below the matrix GTT. The anchorage zones were protected from high temperature during testing, so anchorage effects (loss of bond) should not be reflected in the data. Sorathia et al. (1992) conducted residual flexural strength tests on a variety of FRP panels after 20 minutes of exposure to flame. The tests indicated residual flexural strengths of only about 21% of the original for thermoset matrix FRP. The reader will note that the flexural strength of an FRP is dependent largely on the shear strength of the matrix. Kumahara et al. (1993) conducted a detailed study on the tensile strength and longitudinal modulus of a variety of continuous glass, carbon, and aramid FRP bars at high temperature. They found that aramid fibre bars showed the most pronounced changes in properties due to heating, with tensile strength values that dropped by 20% at 100 C, and by about 80% at 400 C. The Young’s modulus also dropped by 15% at 100 C, and decreased to about 30% at 250 C. Glass fibre bars demonstrated somewhat different behaviour depending on the type of matrix used. The tensile strength of the glass/vinyl ester FRP decreased by 20% at temperatures of 100 C. At 250 C, the tensile strength of the GFRP had decreased by 40%, and at 400 C, the reduction was 60%. In contrast, glass/polyphenylene sulfide (PPS) bars demonstrated little reduction in strength at temperatures up to 250 C. The authors stated that the binder used in the glass/vinyl ester FRP rods was less resistant to heat than that used in the glass/PPS material. No significant reduction in the elastic modulus was observed for the GFRP materials up to temperatures of 250 C. Tests on carbon/epoxy FRP showed a tensile strength reduction of 0 to 25% by 100 C, and 0 to 50% at 250 C. A strength loss of at least 40% was observed at 400 C. The Young’s modulus of the carbon FRP did not appear to change significantly up to temperatures of 250 C. The anchorage

45



zones of the FRPs tested in this study were isolated from high temperature.

The authors

concluded that the tensile strength and modulus of FRP bars was significantly affected by exposure to high temperature, and that the type of fibre and resin used was a key factor in high temperature performance of FRPs. In particular, it was deemed important to use a matrix with as high a GTT as possible. Fujisaki et al. (1993) conducted heat resistance tests on grid-shaped FRP reinforcement composed of carbon, glass, or carbon/glass hybrid fibres with a vinyl-ester resin. Two types of heating tension tests were conducted on carbon/glass hybrid FRP grid reinforcement: a series of tension tests at high temperature, and a series of residual strength tests performed after heating and cooling. Experimental results indicated that the residual strength of the specimens after heating did not seem to be affected by temperatures up to 250 C. However, the strength during heating was reduced by about 40% at 100 C, and remained about 60% of the unheated strength up to 250 C. Again, the anchorage zones of the FRPs were insulated in this study. This finding could be important for the fire behaviour of FRP-reinforced structures in that the residual strength of the members might not decrease significantly, allowing members to retain much of their strength after exposure to high temperature. The residual strength of FRP-reinforced concrete members and FRP reinforcing materials warrants further study. Uematsu et al. (1995) used uniaxial tension tests on carbon/poly-ether-ether-keytone (PEEK) FRP.

The longitudinal elastic modulus was found to be relatively insensitive to

temperatures up to 200 C, demonstrating a loss of less than 10% at that temperature. However, the matrix-dominated properties showed a significant deterioration at temperatures above 100 C. For example, the shear modulus decreased by 5% at 100 C, 55% at 150 C, and 85% at 200 C. Dimitrienko (1999), in an attempt to develop analytical models to describe the high temperature thermomechanical properties of FRP materials, conducted tests on a number of epoxy matrix FRP products. His tests indicated severe reductions in the elastic and strength

46



properties of carbon and glass FRPs at temperatures below 300 C. Both FRPs displayed strength and stiffness reductions of 20% at 200 C and about 40% at 250 C. No mention is made in this study of a provision for thermal protection of the FRP anchorages during these tests, and hence it is not clear if thermal degradation of the shear properties of the polymer matrix at the anchorage were responsible for the apparent deterioration in tensile properties. Alsayed et al. (2000) found that, at 350 C, carbon FRP retained about 35% of its room temperature tensile strength and 40% of its tensile modulus. The corresponding percentages for aramid FRP were 15% and 40%. Sumida et al. (2001) conducted a test program to investigate the heat resistance of FRP rod reinforcements available in Japan. They conducted tensile tests on both carbon/epoxy and aramid/epoxy FRP rods at elevated temperature and after exposure to high temperature. Tests at high temperature, with the anchor zones insulated from heat, indicated tensile strength reductions of approximately 40% and 60% respectively, for carbon and aramid FRP rods at 260 C, with essentially linear trends of strength loss with increasing temperature. Residual strength tests after heating indicated less than 10% reduction in tensile strength at heating temperatures up to 300 C for both carbon and aramid FRP rods. Kodur and Baingo (1998) conducted a detailed literature survey on the properties of FRP reinforcement at high temperature, and subsequently presented a numerical model to describe the heat transfer behaviour of FRP bar-reinforced concrete slabs which has been used to provide design guidance in CSA S806: Design and Construction of Building components with Fibre Reinforced Polymers (CSA, 2002). Based on their work, they suggested a strength versus temperature curve for glass FRP reinforcements subjected to fire. Figure 2.17 shows the variation of strength with temperature for GFRP, timber, concrete, and steel based on their survey of the literature, which predicts a strength loss for FRP of about 15% at 100 C and 75% at 250 C.

47



Based on the experimental data discussed above, a database of test results on the strength and stiffness properties of various fibre and FRP types was assembled. This database has been used in Appendix D to develop semi-empirical analytical expressions that describe the reduction of strength and stiffness of FRP materials at elevated temperatures – information which is essential for the numerical modelling presented in Chapter 5 of this thesis.

Based on the

database, the temperature dependant behaviour of carbon, glass, and aramid FRP elements compiled from a number of different studies is shown in Figures 2.19, 2.20, and 2.21 respectively. Variation in the elastic modulus of different FRP materials is shown in Figure 2.22. It is evident that all types of FRPs show diminished strength and stiffness properties with increasing temperature, although there is a great deal of scatter in the results – as should be expected given the wide range of possible matrix formulations, fibre orientations, and fibre volume fractions represented in the data. The data also indicate that a significant reduction in strength should be expected for any FRP material at temperatures well below 500 C. This would indicate that the critical temperature for FRP reinforcement is significantly less than that for steel (593°C).

Creep Creep can become a critical factor in the design of FRP-reinforced concrete in that excessive long-term deformations can lead to unserviceable structures or to creep-rupture of FRP reinforcement. One major concern when FRPs are exposed to elevated temperatures, as in the case of fire, is that thermally accelerated creep may be severe and could lead to large deflections and/or failure. Creep, defined as the increase in strain with time at a constant load level, is caused by a combination of elastic deformation and plastic flow. It is widely recognized that creep strain in polymers and polymer composites is dependent on temperature (Mallick 1988). In general, creep strain of FRPs increases with increasing temperature and is largely dependent on the matrix

48



material. Also, highly cross-linked thermoset matrices with a higher GTT exhibit less creep than thermoplastics. Fibre orientation greatly influences the temperature dependence of the creep characteristics of FRPs. If the fibres are in the loading direction, creep in the fibres governs creep in the composite, and it has been observed that commercially available fibres do not creep significantly, with the exception of some aramid fibres (Mallick 1988). Rahman et al. (1993) conducted tensile creep tests on a uniaxial carbon/glass hybrid FRP loaded to 40% of its ultimate strength for 175 days at room temperature, and determined that the creep in the fiber direction was only 1.8% of the initial strain. Thus, with the exception of aramid fibres, little temperature dependence of creep is expected in the fibre direction, and thermally accelerated creep during fire should not be a significant problem in civil engineering applications, where uniaxial composites are the FRPs of choice.

2.4.4.5 Bond Properties at Elevated Temperature The bond between FRP and concrete is essential to transfer loads, through shear stresses that develop in the polymer matrix or adhesive layer, from the FRP to the concrete and vise-versa. In the event of fire, changes in the mechanical properties of the matrix material have the potential to cause loss of bond at modestly increased temperatures, and result in loss of interaction between FRP and concrete. The result could be catastrophic, both for internally FRP-reinforced concrete and for externally FRP-wrapped reinforced concrete, since loss of interaction could very rapidly lead to loss of tensile or confining reinforcement, and subsequent failure of the concrete member. Katz et al. (1998, 1999) and Katz and Berman (2000) studied the effect of elevated temperature on the bond properties of FRP bars in concrete. They investigated the pullout strength of glass FRP reinforcement, with six different types of surface textures, subjected to temperatures up to 250 C and found that the bond strength of FRP bars decreased as the surrounding temperature increased. Up to 100 C, the loss of bond was found to be similar to that observed in steel-reinforced concrete, but at temperatures of 200 C to 220 C, the bond strength

49



decreased dramatically to a value of about 10% of the original. The authors commented that the reduction in bond strength was likely due to changes in the properties of the polymer matrix at the surface of the rod. Sumida et al. (2001) conducted bond strength tests at high temperature on carbon and aramid/epoxy FRP rods, and found large bond strength reductions at rod temperatures above 100 C. They concluded that the surface temperature of FRP rods should be kept below 100 C when subjected to high levels of permanent stress, and that advanced resins with superior high temperature properties are required to improve the fire resistance of FRP reinforcing materials. More information on the deterioration of the bond between FRP reinforcements and concrete is available in Appendix D. In the case of externally bonded FRP reinforcements, no work has been done on bond at high temperature. Temperature effects are more critical in externally bonded applications as loss of bond would result in complete loss of FRP reinforcement, and further work is required in this area.

2.4.4.6 Smoke Generation and Toxicity The temperature at which a polymer matrix will ignite, the flame spread characteristics, and the amount and toxicity of smoke produced, are all dependent on its particular formulation (Nelson, 1995). Smoke toxicity is potentially a major concern for the fire resistance of FRP reinforcements for concrete, and there is an alarming absence of published research in this area, likely because most of the research to date has been performed by the defense and aerospace industries, and the results are not available to the general public. Combustion gases from burning FRPs create a toxicity hazard to humans, and can be highly corrosive to equipment and electronics. Although smoke from burning resins may be toxic and corrosive, it has been stated in the past that the level of hazard is low because FRP repairs are normally in the open or the amount of material is small (Ballinger et al., 1993). This is not the

50



case when concrete elements are reinforced or strengthened with FRPs in buildings. Neale and Labossière (1991) comment on the problem, stating that polymers will generally produce large quantities of very dense, sooty, black smoke. Some components of the smoke such as carbon monoxide may be toxic, and other toxic gases such as hydrogen cyanide may also be produced, depending on the component materials used in the manufacture of the FRP. Sorathia et al. (1992) studied the smoke generation and toxicity characteristics of a variety of FRP materials for use in marine and offshore applications. They conducted optical smoke density tests according to ASTM E662-97 (ASTM, 1997a). The results, shown in Figure 2.23, demonstrate that the thermoset resins commonly used in structural FRPs (vinyl ester in particular) generate unacceptable quantities of smoke. The limits quoted by the authors for smoke density in this study were 100 within the first 300 seconds and 200 at any point during the test. Tests were also conducted to determine the nature of combustion gases for a variety of FRPs. The results are shown in Table 2.6 and indicate that: thermoplastics generate significantly lower concentrations of carbon monoxide during combustion, burning fluorocarbons produce hydrogen fluoride, chlorinated resins produce hydrogen chloride, sulfur-containing compounds produce hydrogen sulfide, and nitrogen containing resins produce hydrogen cyanide, all of which are potentially harmful compounds. Smoke toxicity is obviously most critical in cases where the FRP is installed on the exterior of a concrete member, although internal reinforcement may generate smoke that could escape through cracks in the concrete at high temperature.

Much more work is required to

adequately characterize the gases released and the temperatures at which smoke generation will occur for FRPs in civil engineering applications.

2.4.4.7 Ignition and Flame Spread The impact of a severe fire on an unprotected polymer matrix FRP may be substantial and obvious. Widespread charring, melting, delamination, bucking, and ignition may occur

51



depending on the severity of the fire. Auto-ignition has been observed for FRP composites exposed to high levels of irradiation. Typically, pyrolysis – the process by which solid organic materials are converted into gases, liquids, and a solid char – is initiated between 200 C and 300 C for organic matrix materials (Milke and Vizzini, 1990). During a fire, the FRP matrix will be the most at risk of ignition, due to its high content of carbon, hydrogen, and nitrogen, all of which are flammable. The higher the hydrogen-to-carbon ratio in the polymer matrix the greater is its tendency to burn, and so different matrices may have entirely different susceptibilities to ignition. Despite this disadvantage, the reader should recognize that polymer matrix materials tend to have higher ignition temperatures than wood or other cellulose materials (Neale and Labossière, 1991). Sorathia et al. (1992) conducted a series of flame spread tests on a variety of different FRP materials according to ASTM E162-98 (ASTM, 1998b), which covers the measurement of surface flammability of materials and is not intended for use as a basis of ratings for building code purposes. The results indicated relatively poor flame spread characteristics for vinyl ester and epoxy as compared with other matrix materials. It should be noted that there exist a number of resin additives that can significantly increase the ability of polymer materials to resist ignition, although these additives generally cause reductions in the mechanical properties which discourages their use in structural applications. Additives may also be incorporated to increase the resistance of FRPs to flames, heat, smoke generation, moisture absorption, oxidation, various chemical actions, and shrinkage. Because of the enormous variety of possible additives, and their effects on the various properties of FRPs, no further discussion of this topic is included here.

2.4.4.8 Barrier Treatments The performance of fire exposed FRP systems can be greatly improved by the use of barrier treatments or coatings. These treatments function either by reflecting radiant heat back

52



towards the heat source, or by delaying heat penetration to the FRP through their insulative and/or ablative properties (Sorathia et al., 1992). Sorathia et al. (1992) studied the effectiveness of a ceramic fabric, a ceramic coating, and several intumescent coatings (silicon foam, ablative materials, and phenolic resin as a sacrificial fire barrier), in protecting vinyl ester/glass and epoxy/carbon FRPs exposed to a radiant heat source. For the vinyl ester/glass FRP, ignition times were increased by a factor of 20 using a water-based intumescent coating. For epoxy/carbon FRP, use of an ablative layer with a phenolic skin increased the ignition time by a factor of 10 and reduced the heat release rates significantly. Apicella and Imbrogno (1999) conducted flame spread and smoke generation tests according to ASTM E84-95 (ASTM, 1995) on a carbon/epoxy FRP wrap. They also investigated the ability of three different fire coatings: an intumescent latex coating, a flame-retardant latex, and an exterior grade acrylic latex, to delay or suppress flame spread. Tests demonstrated that the FRP system without protection achieved only a UBC class III rating, with an ASTM E-84 flame spread index of 155 and a smoke development index of 405. However, with the intumescent latex protective layer, the same system was able to achieve an UBC class I rating, with a zero flame spread index and a smoke development index of 20. The UBC Classifications for interior finishes are shown in Table 2.7. Class III materials are approved for use in factories, storage areas, and industrial rooms, Class II materials are approved for use in institutional rooms and spaces, and Class I materials are approved for use in critical areas like unsprinkled stairwells and exit ways (Apicella and Imbrogno, 1999). No results were reported with regard to ignition temperature or smoke toxicity.

2.5 Fire Endurance 2.5.1 Philosophy The discipline of fire engineering is primarily concerned with the protection of life and property from fire (Lie, 1992). The high temperatures experienced during a fire in a building can

53



cause dramatic changes in member behaviour and cause safety or ultimate limit states to be exceeded. Provision for adequate fire endurance of structural members is required to ensure that, when all methods of fire containment fail, structural integrity is maintained for an adequate period of time. Design for fire safety is an extremely complex and rapidly evolving discipline, which requires consideration of a wide range of factors, from materials, to sprinkler systems, to exit signage, to structural integrity. In what follows, we are concerned with structural fire endurance requirements as they apply to buildings. Fire endurance requirements for buildings are specified in building codes such as the National Building Code of Canada (NRC, 1995). They are generally expressed in terms of minimum allowable times to reach specified failure criteria. The prescribed time to failure is chosen based on a number of factors, such as the building size and occupancy, and is a function of: applied load, member type and dimensions, fire intensity, and the materials involved. For traditional structural materials and methods of construction (structural steel, timber, and reinforced concrete), behaviour in fire has been extensively studied and is relatively well understood. However, full-scale fire endurance tests are often extremely time-consuming and costly, and for new materials and construction methods, fire endurance requirements are not well established in most cases. When examining the fire endurance of reinforced concrete members, it is essential to consider the three basic types of members: beams, slabs, and columns. Each member type has specific requirements in terms of the applicable failure criteria and the required time to reach said criteria. For instance, columns, which are the primary load bearing members in a structure, tend to have the highest required fire ratings (up to 4 hours) whereas slabs, the failure of which would likely be more localized, have comparatively low fire ratings (as low as 1 or 2 hours) (Lie, 1992). There are essentially three distinct failure criteria for different types of reinforced concrete members: loss of load bearing capacity, loss of insulating capacity, and loss of integrity or separating capacity (Kodur, 1999).

54



Loss of load bearing capacity refers to structural failure (or collapse) such that a member no longer performs the structural task for which it was constructed. Loss of insulating capacity refers to floors or walls that are required to provide fire separation in a building. The particular requirements in Canada (NRC, 1995) are such that the maximum temperature rise at the unexposed face of the specimen shall not exceed 181 C, and that the average temperature rise at the exposed face shall not exceed 139 C. Loss of integrity refers to walls, floors, and roofs and states that no openings or holes should form that would allow fire to move through the assembly. All three criteria do not apply to all types of members. Obviously, a column must only satisfy load-bearing requirements. The behaviour of reinforced concrete in fire has been studied quite extensively in the literature (Abrams, 1977; Khoury, 2000; Lie, 1992).

As discussed earlier, concrete has a

relatively low thermal conductivity, and it thus experiences elevated internal temperatures far more slowly than steel. Concrete members can consequently achieve very high fire ratings with no supplemental fire protection, a factor which has contributed to the success of concrete as a structural material during the last century. Adequate fire endurance for reinforced concrete members is usually ensured by providing a sufficient concrete cover to the steel reinforcement. The low-conductivity concrete cover acts as insulation for the internal steel reinforcement, and ensures that the temperature in the reinforcement does not exceed its critical temperature (defined as the temperature at which the reinforcement loses about 50% of its room temperature strength). However, it is difficult to evaluate the fire performance of load-bearing members by heating tests when the permissible temperatures for the reinforcing materials are unknown (Tanano et al., 1999). The critical temperatures for reinforcing and prestressing steel have been established as 593 C and 426 C respectively (Kodur, 1999). No such critical temperatures have yet been established for the wide variety of available FRP reinforcements, although Kodur and Baingo (1998) suggest a temperature of 250 C for internal GFRP reinforcement based on a review of the

55



literature. The lower critical temperature suggested for FRP reinforcement is highly significant to structural designers, since much larger concrete covers (or supplemental insulation) would be required to achieve similar fire endurance ratings as would be achieved with steel-reinforced concrete. This point is demonstrated further through numerical modelling in Chapter 5. In FRP plating applications there is no concrete cover, and hence there is no insulation for the FRP reinforcement (unless supplementary insulation is installed). Additionally, in steelreinforced concrete design there is a provision for concrete cover for corrosion protection which is usually sufficient to satisfy fire endurance requirements. For FRP-reinforced concrete there is no such corrosion provision (Kodur and Baingo, 1998), although most design guidelines suggest minimum concrete covers to FRP reinforcement based on the prevention of concrete cracking as a consequence of thermal incompatibility. Thus, the thickness of concrete cover might be less in FRP-reinforced concrete than in steel-reinforced concrete. In cases where the cover cannot be relied on to provide adequate insulation, additional fire insulation will be required. There are a variety of options as far as supplemental fire insulation is concerned; gypsum board, mineral wool, shotcrete, spray-up cellulose, and vermiculite plaster, are all possible insulative materials, as are a wide variety of currently available intumescent coatings. Because of the wide variety of insulating materials available, and because they are for the most part proprietary products, a detailed discussion of supplemental insulation methods is avoided here.

2.5.2 Procedures to Evaluate Fire Endurance In the past, the fire endurance of a particular structural assembly was determined entirely through testing (Lie, 1992). However, this method is extremely time-consuming and costly, so with the advent of powerful digital computers, numerical procedures have been developed that can accurately predict the behaviour of structural members during a fire. These numerical procedures use detailed information about the thermal and mechanical properties of the members’ constituent materials, and have the potential to substantially reduce both the time and expense of

56



fire endurance evaluation. Brief discussions of both the experimental and numerical methods of fire endurance evaluation follow.

2.5.2.1

Experimental Procedures to Evaluate Fire Endurance

The experimental procedure traditionally consists of subjecting a full-scale structural member, under full and sustained unfactored service load, to a standard fire in a specialized testing furnace. The specialized testing furnace is designed to reproduce as accurately as possible the Canadian standard fire, which is represented by a time-temperature curve developed specifically for fire endurance tests and meant to be a conservative reproduction of a severe building fire. The Canadian standard fire (which is essentially the same as the American version) is shown in Figure 3.31. There has been substantial debate in recent years as to the validity of the various standard fire curves used in different countries (Khoury, 2000), with many fire researchers expressing the view that the standard fires are sometimes highly conservative and other times dangerously unconservative. This debate is beyond the scope of this thesis and is not discussed further, although it is interesting to note that significant changes in the criteria and procedures for fire testing of building elements can be expected in the near future, as many countries shift their building code philosophies from prescriptive to objective or performance based. Although different standard fires and procedures are used in different countries, most are essentially equivalent. In Canada, the test method for fire endurance is outlined in CAN/ULCS101: Standard test methods of fire endurance tests on building construction and materials (CAN/ULC, 1989). The fire endurance of the member is defined as the time to reach one of the three specified failure criteria with respect to load bearing capacity, insulation, and integrity. With regard to the fire endurance of reinforced concrete members, results obtained from fire endurance tests have been used to develop guidelines that give minimum dimensions and concrete cover to reinforcement to achieve specific fire endurance ratings (Kodur, 1999). For standard building elements, these tables can be used to determine the fire endurance instead of

57



having to conduct expensive and time-consuming laboratory tests. These data are of little use however, in the determination of the fire endurance of FRP-reinforced concrete members, since they were developed from numerous tests on conventionally reinforced concrete.

2.5.2.2

Numerical Procedures to Evaluate Fire Endurance

Rather than performing laboratory tests to determine fire endurance, it is now possible to conduct detailed numerical analyses to obtain the fire endurance of a variety of structural member types. Models have been developed and used in Canada to obtain fire endurance ratings and to perform parametric studies on the factors that influence fire endurance for many types of concrete members, including: beams, slabs, and columns (Lie, 1992, 1994; Lie and Denham, 1993; Lie and Irwin, 1993; Lie and Kodur, 1996; Lie and Stringer, 1994; Lie et al., 1984, 1992). Although there are an enormous variety of numerical approaches to take, most studies in Canada, conducted at FRM/IRC/NRC, have employed an explicit finite-difference heat transfer procedure, coupled in some cases with a stress-strain or equilibrium analysis. In this approach, the fire temperature is calculated based on a standard fire time-temperature curve outlined in CAN/ULC-S101. The member temperatures as a function of time are obtained using an explicit finite-difference heat transfer procedure derived from an elemental heat balance formulation. Finally, a stress-strain analysis, which accounts for the variation in mechanical properties of the constituent materials with temperature, is used to determine the structural behaviour of the member as its internal temperatures change. Over the past two decades, a variety of increasingly complex thermohydromechanical finite element models have also been developed to describe the behaviour of concrete structures in fire (Khoury, 2000). These models, developed primarily in Europe, have evolved to the point that such complex factors as moisture movement, spalling, chemical decomposition, load-induced thermal strain, and phase changes can be accounted for. However, these models are proprietary

58



in nature and are not well suited to nor available for the manipulation required for the purposes of this thesis. The reader will note that, to accurately use any numerical fire analysis procedure, a detailed knowledge of the thermal and mechanical properties of the constituent materials, at a variety of different temperatures, is required.

Although thermomechanical properties are

relatively well established for concrete, timber, and steel (Lie, 1992), information on the thermomechanical properties of FRPs at high temperature is scarce.

2.5.3 Fire Endurance Tests on Reinforced Concrete Members Fire endurance tests on a wide range of reinforced concrete members and assemblies have been performed over the last thirty years. These tests have examined a variety of factors and have led to the development of design codes for reinforced concrete members. Some of the more important factors that influence the fire behaviour of reinforced concrete are the cover to the steel reinforcement, the size of the member, the aggregate type, and the concrete compressive strength. In recent years, the primary purpose in conducting full-scale fire endurance tests has been to validate numerical models, such that parametric studies can subsequently be performed with little additional cost. For a comprehensive review of test methods and results, the reader is referred to Lie (1992) and Khoury (2000).

2.5.4 Fire Endurance Tests on FRP-Reinforced Concrete Members Studies investigating the thermal and structural behaviour of FRP-reinforced concrete elements are extremely scarce. The few tests results that have been presented in the literature represent tests on specific FRP reinforcing systems and materials, and are not generally applicable to many different FRP-reinforced concrete elements.

2.5.4.1 Fire Endurance Tests on FRP Bar-Reinforced Concrete NEFCOM Corporation (1998) conducted fire endurance tests on concrete slabs that were internally reinforced with either glass or carbon FRP grids produced under the trade name

59



NEFMACTM.

A total of ten 3500mm by 500mm, 120mm thick, slabs were exposed to fire on

one side for a maximum duration of 2 hours according to the Japanese Industrial Standard (which is essentially equivalent to the CAN/ULC S101 standard fire up to 2 hours). Parameters that were varied in the experimental program included the load intensity, the type of reinforcement (GFRP, CFRP, GFRP/CFRP in combination, or conventional reinforcing steel), the type of polymer matrix used (vinyl ester or unsaturated polyester), the bar size of the grids, the thickness of concrete cover, the presence of a construction joint, and the presence of supplemental insulation in the form of a 25 mm thick rock wool board. Deflections, cross-sectional temperatures, and reinforcement temperatures were all monitored during the tests. It was observed that the deflection of all slabs increased dramatically when the temperature at the bottom of the reinforcement reached 600 C. This was due to a severe drop in the stiffness of the FRP grid at these elevated temperatures. The performance in fire of the FRPreinforced slabs did not appear to be affected by the type of resin used in the fabrication of the FRP grid. The rise in temperature in the FRP grid, for the same concrete cover thickness, did not appear to be affected by the type of fibre used. However, the temperature rise at the level of the reinforcement for the steel-reinforced slab was slower than for the slabs reinforced with NEFMACTM. It is not clear in the NEFCOM study exactly where thermocouples were located in the slabs. If the thermocouples measuring reinforcement temperature were placed at the bottom of the reinforcement (the side closest to the fire), then the slower temperature rise in the reinforcement observed for the steel reinforced slab was likely due to the higher thermal conductivity and heat capacity of steel, such that it acted as a thermal sink to draw heat further into the slab, and thus reducing the observed temperature at that location. This idea is discussed further in Chapter 6. Slabs with construction joints failed before the bottom surface of the reinforcement reached 600 C because of rapid thermal degradation of the epoxy joint filling agent, resulting in very high temperatures at the location of the joint. The insulated slabs showed

60



substantially higher fire endurance than those without insulation. After two hours of the test, the temperature in the reinforcement in the insulated slab was only 170 C and the deflection only 25mm, as opposed to 600 C and 73mm in the uninsulated slabs. Specimens with higher applied loads showed lower fire endurance based on the time to reach a limiting deflection of 73 mm. The authors concluded that there was no recognizable difference in the fire deflection behaviour of slabs reinforced with NEFMACTM or with steel. The most interesting information presented in the above paper is that the NEFMACTM grid reinforcement was apparently able to maintain strength and stiffness until it reached a temperature of 600 C. Most FRP materials should have lost a significant portion of their strength and stiffness at temperatures well below 600 C. While these results seem contradictory, it is possible that special chemical additives were incorporated in the FRP matrix to improve the fire behaviour, although the authors do not comment in this regard. Tanano et al. (1995) performed a study on the fire behaviour of FRP-reinforced concrete beams in Japan. Their study focused on the residual strength of beams after exposure to fire and did not investigate the structural behaviour of the elements during exposure while under load. In this study, 3m long beams with a 200 mm by 300 mm cross-section, and reinforced with either carbon, glass, or aramid FRPs, were heated in a furnace according to a modified version of the Japan Industrial Standard heating curve, such that their temperature reached some specified value in one hour, and was then maintained at a constant level for one and a half hours until the temperature at the level of the internal tensile reinforcement reached 250 C, 350 C, or 450 C. The authors observed several explosive failures during the heating. It was noted that, because these failures were only observed in beams with an epoxy matrix FRP, the explosive failures were not thought to be associated with generation of steam within the specimens, but with the use of epoxy matrix FRP with a spiral configuration. The specific cause of the explosive failures remains unknown.

61



After heating, the beams were returned to room temperature and tested in four-point bending. It was observed that bond strength and stiffness decreased for epoxy matrix FRPreinforced concrete beams as the heating temperature increased, but that the rate of decrease was different depending on the type of FRP used. The rates of decrease in both strength and stiffness were greater for epoxy matrix FRP-reinforced beams than for those reinforced with conventional reinforcing steel. Beams reinforced with an inorganic matrix FRP showed only a small reduction in residual strength after exposure to temperatures of 250 C and above. The residual tensile strength of the FRP reinforcement decreased as exposure temperature increased for all materials, as evidenced by a change in failure mode of the beams from compression failure in the concrete to tensile failure of the internal reinforcement. Fujisaki et al. (1993) conducted fire endurance tests on FRP-reinforced concrete panels in an attempt to justify the use of FRP-reinforced precast concrete curtain walls. The grid-shaped FRP reinforcement was composed of carbon, glass, or carbon/glass hybrid fibres with a vinyl ester resin. Thirty-minute fire endurance tests were conducted on precast concrete panels with the grid shaped FRP reinforcement in accordance with the Japanese Industrial Standard.

All three

types of FRP material (glass, carbon, and carbon/glass hybrid) were examined in a single test. The test results indicated that the maximum temperature on the exposed face of the 150mm thick lightweight concrete slab remained below 260 C, and the maximum temperature observed in the internal reinforcement was 300 C, at which point its strength was assumed to be equal to 60% of the unheated tensile strength based on preliminary testing. The maximum out of plane deflection at the center of the slab was 14 mm, and the maximum crack width was 0.55mm. Based on the results of the tests, the authors concluded that it was not necessary to test the structural capacity as well as the fire-resistance for the members, and judged that the panels displayed sufficient fireresistance characteristics to be used for curtain wall elements.

62



Okamoto et al. (1993) conducted a series of full-scale fire endurance tests on partially prestressed concrete beams reinforced and prestressed with braided FRP bars.

Braided

aramid/epoxy or carbon/epoxy FRP bars were used as the main tendons and transverse reinforcement. Two specimens, one with aramid FRP reinforcement and the other with carbon, were tested. The beams were heated according to the Japanese industrial Standard heating curve and subjected to a constant load in four-point bending. It was evident in the tests that heating significantly increased the deflection of the FRPreinforced concrete beams. The time to failure for both beams tested was 114 minutes, and no explosive failure of the concrete cover was observed. Temperature profiles at different locations were measured throughout the tests, and it was noted that the temperature of the FRP tendons did not exceed 120 C. The authors concluded that the FRP-prestressed beams should be considered able to resist fire for about two hours, although they stated that the residual strength of these beams warranted study. Kodur and Baingo (1998) conducted a literature review and a numerical parametric study in order to examine the fire endurance of FRP-reinforced concrete slabs. In their study, a numerical finite difference procedure was used to study the time-temperature response of an internally FRP-reinforced concrete slab under exposure fire. The numerical procedure essentially gave fire endurance ratings based on the achievement of a critical temperature of 250 C in the internal FRP reinforcement. The choice of 250 C as the critical temperature was based on what the authors referred to as a worst-case scenario for GFRP chosen after a review of the scarce literature in this area. Some of the more important conclusions of their study were that: FRPreinforced concrete slabs have a lower fire endurance than slabs reinforced with conventional reinforcing steel; higher fire endurance for FRP-reinforced concrete slabs can be obtained by using a thicker concrete cover and by using carbonate aggregate concrete; data on the material

63



properties of FRPs at elevated temperatures are required; and structural fire endurance tests are required to validate the numerical models. Sakashita (1997) investigated the effect of fire on concrete beams reinforced with carbon, glass, and aramid FRP rods with different surface textures and fibre orientations (braided, spiral, or straight). The behaviour of these beams was compared to that of a conventionally reinforced concrete beam in a fire test. All specimens were heated to 100ºC for three hours prior to testing and then heated to 1000ºC under load in 180 minutes. It was found that, at a furnace temperature of 350ºC, specimens containing aramid FRP experienced a sudden increase in vertical deflection. These beams failed at a furnace temperature of 500ºC. However, specimens containing glass or carbon FRPs, or conventional steel, completed the 180 minute test without failure. At the end of the tests, it was observed that the average midspan deflections and temperature at the bottom face of the beams were 160 mm and 680ºC for GFRP, 30 mm and 700ºC for CFRP, and 100 mm and 680ºC for conventional reinforcing steel. Nakagawa et al. (1993) performed a fire test on an FRP-reinforced concrete curtain wall section. The curtain wall was a 75 mm thick concrete slab and was reinforced with a 3-D carbon FRP grid.

Fire endurance tests of 60 minutes duration were conducted on two specimens

according to the Japanese Industrial Standard. A maximum temperature of 516 C was recorded in the FRP reinforcement and of about 110 C on the unexposed face of the slab. Based on the results of heating tests on the specific FRP reinforcement, which indicated loss of about 50% of the room temperature tensile strength at 400 C, it was concluded that the reinforcement was damaged by the extremely high temperatures recorded during the test, although the panels (which were tested under load) did not fail. No cracks or fractures due to differential thermal expansion were observed. The authors concluded that the wall panels displayed an adequate fire endurance of more than 60 minutes.

64



2.5.4.2 FRP-Plated or -Wrapped Reinforced Concrete In concrete members externally reinforced with FRP, unless an insulating or intumescent protective layer (or both) is applied, the FRP will be immediately exposed to the heat of the fire, likely resulting in rapid loss of composite action. In these cases, it is required that the reserve strength of the member, which would revert to a conventional steel-reinforced concrete member, would be relied on to carry the necessary loads for the duration of the fire. Few tests on externally FRP-reinforced concrete have been reported in the literature. Only three studies to date have examined the behaviour in fire of externally FRP-reinforced concrete beams and slabs, and two preliminary studies have examined the fire behaviour of FRPwrapped columns. In terms of tests on beams and slabs, Deuring (1994) studied flexural strengthening with externally bonded FRP materials on six concrete beams during exposure to fire. One beam was unstrengthened, one was strengthened with an adhesive bonded steel plate, and four were strengthened with CFRP plates. Two of the FRP plated beams were tested without insulation and two were protected with insulating plates of different thickness. The results of this initial test program demonstrated the need for thermal insulation of the FRP plates. Bond between the FRP and concrete was lost very rapidly (within minutes) for the unprotected specimens but occurred after about an hour for those with supplemental insulation. In an effort to gain further insight into the behaviour of FRP-plated reinforced concrete beams during fire, a second study was conducted by Blontrock et al. (2000). The focus of this test program was to investigate a number of different thermal protection materials and layouts. The program included tests on a total of ten beams. An unstrengthened reference beam and a strengthened reference beam were statically tested to failure in four point bending, 2 unprotected and unstrengthened beams were loaded to full service load and tested under fire exposure, and 6 strengthened and protected beams were loaded to full service load and tested under fire exposure.

65



The protection schemes were different for all six protected beams and consisted of gypsum board/rock wool combinations.

All strengthened beams were strengthened using the Sika

CarboDurTM carbon/epoxy FRP strengthening system. The fire endurance tests were conducted in accordance with the International Standards Organization (ISO) test method 834 for fire testing of concrete members, which is essentially the same as the Canadian CAN/ULC S101 fire testing procedure. A summary of the test results is presented in Table 2.8. The “U”-shaped protection scheme shown in Figure 2.24 was most effective at prolonging the time before loss of interaction between the plate and the concrete. This scheme had the additional advantage of lowering the temperature of the internal reinforcing steel, thus contributing to lower deflections throughout the tests. Blontrock et al. (2001) conducted a series of fire endurance tests on externally CFRPreinforced concrete slabs in an effort to evaluate their fire endurance. As was the case in the beam study discussed above, various fire insulation schemes (consisting of rock wool and/or gypsum board layers) were implemented to prevent debonding of the carbon FRP plating material. A total of 10 slabs were tested including: unstrengthened and strengthened reference slabs tested at room temperature, unstrengthened and unprotected slabs tested under exposure to fire, and strengthened and protected slabs tested under exposure to fire. Some of the more important conclusions reached in these studies were that: thermal protection is required in order to maintain the interaction between the FRP plates and the concrete; without protection it is impossible to achieve the same fire endurance as for the unprotected and unstrengthened beams; interaction between the externally glued composite and the concrete was lost when the temperature in the epoxy adhesive reached temperatures of 66 C to 81 C for the SikaTM CFRP product, and 47 C to 69 C for an S&P LaminatesTM CFRP product; partial protection of the external strengthening system (applied to the anchorage zones only) was able to maintain interaction between the FRP and the concrete; and the fire endurance for the

66



strengthened and protected beams was at least the same as for the unstrengthened unprotected beams. To the knowledge of the author, no fire endurance tests on full-scale on FRP-wrapped reinforced concrete columns have ever been reported in the literature. One full-scale fire test was conducted at the National Research Council of Canada (NRC) in conjunction with R.J.Watson Engineering in 1997 but has not been commented on in the literature, likely because of its proprietary nature. The only available information from this test was the conclusion that the fullscale, rectangular, carbon FRP-wrapped concrete column performed satisfactorily during the test and was able to achieve a satisfactory fire performance rating (Watson, 2000). Saafi and Romine (2002) conducted a series of residual strength tests on FRP-wrapped reinforced concrete cylinders after exposure to elevated temperatures. A total of 40 cylinders, wrapped with two layers of a unidirectional glass/epoxy FRP, was tested in axial compression after exposures of up to 3 hours at 90 C, 180 C, and 360 C. The results of these tests indicated significant reductions in the overall strength and ductility of the wrapped cylinders at exposure temperatures at or above the 180 C (the GTT for this system). However, this study is not particularly useful for a number of reasons. First, the authors observed several explosive failures of their cylinders, which would indicate that the cylinders were not allowed sufficient time to dry before testing.

In addition, the temperature regimes were not nearly severe enough to be

representative of a building fire, and the tests were conducted after the cylinders had returned to room temperature, a condition which is not representative of a severe building fire.

2.6 Summary It is evident from the material presented in this chapter that information on the fire and high temperature behaviour of FRPs and FRP-reinforced concrete members is extremely scarce. What little information is available in the literature is largely case specific and cannot be applied to a wide range of FRP materials or applications.

67



At elevated temperatures, all FRP materials currently available for civil structural applications will experience a reduction of both strength and stiffness. They may experience significant transverse thermal expansion leading to cracking or spalling of the concrete cover or to the development of shear stresses in their adhesive layer. They may ignite. Upon ignition they may emit dense smoke and toxic gases.

They may lose their bond with the substrate or

surrounding concrete. All of these concerns have not, at present, been adequately studied or addressed by current design guidelines. The development of standard tests for FRP materials are required both at room and high temperatures, with both static and dynamic loading and temperature regimes, such that a database of test results can be formulated and expanded on an international scale. The mechanical and thermal behaviour of FRP materials currently available in industry must be accurately ascertained, such that experimental and parametric numerical studies can be executed with accuracy. Detailed models must be developed and continually updated in to study the effect of varying a wide range of parameters on the fire behaviour of FRP-reinforced and FRP-prestressed concrete members.

Finally, full-scale fire endurance tests are required in order to validate

numerical procedures, and to raise awareness of and confidence in FRP reinforcing materials in the construction industry. Eventually, it is hoped that research will lead to the development of complete design guidelines for the use of FRP-reinforced concrete in buildings and structures. Only when such a design code is produced and sanctioned with confidence by the engineering research community will the use of FRPs for reinforcement gain widespread acceptance and implementation.

68


Glass Glass Glass Carbon Glass Carbon Aramid Hybrid Glass Carbon Carbon Aramid Aramid Aramid Carbon Glass Carbon Carbon Carbon Glass Aramid Glass Carbon Carbon Glass Carbon

C-Bar B1 (#12)1 ASLAN 100 (#12) 1 ISOROD 1 Vinyl ester (VE) VE or polyester Vinyl ester Epoxy Epoxy Epoxy Epoxy Epoxy Vinyl ester Epoxy Epoxy Epoxy Epoxy Epoxy Epoxy Epoxy Epoxy Epoxy Epoxy Epoxy Epoxy

Matrix

ROTAFLEX1 LEADLINE1 CFCC2 Arapree2 FiBRA2 Technora2 SIKA CarboDur S3 SikaWrap Hex 100G3 Mitsuishi Replark 203 M-Brace CF 1303 M-Brace CF 5303 M-Brace AK 603 M-Brace EG 9003 Tyfo SEH-51 Tyfo SCH-41S Tyfo SCH-35 Tyfo SEH-51A Tyfo SCH-30T (1) ISIS, 2001a (2) Braimah, 2000 (3) ISIS, 2001b (4) http://www.fyfeco.com/tyfosys.html

NEFMAC1

Fibre

Name 60 by weight 70 by weight 80 by weight 65 by vol. 53.2 by vol. 45 by vol. 60 by vol. 65 by vol. 68 by vol. -

Fibre Content (%)

69

42 40.8 37.43 111.1 30 100 54 37 56 147 137 125 69 54 > 155 26.1 235 227 373 120 72.4 26.1 72.4 78.6 26.2 90.2

Modulus (GPa) 770 690 635-747 1596 600 1200 1300 600 1800 2550 1750 2860 1400 2150 > 2400 600 > 3000 3800 3500 2000 1520 575 876 991 575 1351

Ultimate Stress (MPa)

Table 2.1: Selected FRP materials currently available for civil engineering applications

8 6-10 8.9-9.1 0.7 .6 -2 -5.2 -3 -

Longitudinal 32 21-23 16.9-21.2 -

Transverse

CTE (µε/ C)


0.27 0.27-0.29 0.38 0.6 -

Poisson’ s Ratio

Rod Rod Rod Rod Grid Grid Grid Grid Rod Rod Rod Strip Rod Rod Strip Sheet Sheet Sheet Sheet Sheet Sheet Sheet Sheet Sheet Sheet Sheet

Shape



Table 2.2: Qualitative comparison of different FRP types (after Meier, 1994) Criterion

Carbon Very Good Very Good Very Good Very Good Excellent Good Very Good Adequate

Tensile Strength Compressive Strength Modulus of Elasticity Long Term Behaviour Fatigue Behaviour Bulk Density Alkaline Resistance Price

Fibre Type Aramid Very Good Inadequate Good Good Good Excellent Good Adequate

Glass Very Good Good Adequate Adequate Adequate Adequate Inadequate Very Good

Table 2.3: Thermal properties of Dexter Hysol® epoxy (after gluguru.com, 2000) Maximum Suggested Service Temp. ( C)

Lap Shear Strength @ 20 C (GPa)

Lap Shear Strength @ 80 C (GPa)

71

12.7

2.8

107

24.1

6.9

Fast Cure Epoxy

54

13.8

1.4

High Temp. Epoxy

204

26.2

20.7

Description General Purpose Clear Epoxy General Purpose High Performance

Table 2.4: CTEs of various unidirectional FRPs and building materials (Mallick, 1988) Material Glass/Epoxy Aramid/Epoxy High Modulus Carbon/Epoxy Ultra-High Modulus Carbon/Epoxy Boron/Epoxy Aluminum Steel Epoxy

Coefficient of Thermal Expansion (10-6 / C) Longitudinal Transverse 6.3 19.8 -3.6 54 -.09 27 -1.44 30.6 4.5 14.4 21.6 – 25.2 10.8 – 18 54 – 90

70



Table 2.5: Thermal conductivities of various unidirectional FRPs and building materials (after Mallick, 1988) Material Glass/Epoxy Aramid/Epoxy High Modulus Carbon/Epoxy Ultra-High Modulus Carbon/Epoxy Boron/Epoxy Aluminum Steel Epoxy

Thermal Conductivity (W/m· C) Longitudinal Transverse 3.46 0.35 1.73 0.73 48.44 – 60.55 0.87 121.1 – 129.8 0.04 1.73 1.04 138.4 – 216.3 15.57 – 46.71 0.346

Table 2.6: Gases released during combustion of FRP (after Sorathia et al., 1992) Fibre/Matrix

Glass/Vinyl Ester Glass/Epoxy Glass/Phenolic Glass/BMI Glass/PEEK

Carbon Monoxide (ppm) 230 283 300 300 TRACE

Carbon Dioxide (% volume) 0.3 1.5 1.0 0.1 TRACE

Hydrogen Cyanide (ppm) NONE 5 1 7 NONE

Hydrogen Chloride (ppm) NONE NONE 1 TRACE NONE

Table 2.7: UBC classifications for interior finishes (Apicella and Imbrogno, 1999) Interior Finish Classifications Class I Class II Class III

ASTM E-84 Flame Spread Index 0-25 26-75 76-2000

ASTM E-84 Smoke Developed Index 0-450 0-450 0-450

Table 2.8: Summary of results from fire tests on FRP-plated reinforced concrete beams (after Blontrock et al., 2000) Beam Number 3 4 5 6 7 8 9 10

Time to Loss of CA* (min) 7 38 26 39 18 22

Temp. in FRP at Loss of CA ( C) 61.2 65.2 52.1 57.4 54.9 56.0

Midspan Defl. @ 90 min (mm) 56.0 49.4 15.3 26.1 47.7 40.7

Temp. in Internal Steel @ 90 min ( C) 536.2 547.5 137.3 212.8 466.4 419.5

* CA – Composite Action

71



Shear Strength (MPa)

40 35 30 25 20 15 10 5 0

0

20

40

60

80

100

Temperature

120

140

160

(oC)

Figure 2.1: Variation in shear strength of a typical epoxy resin with temperature (reproduced after Mays and Hutchinson, 1992) 2500

C-Bar (GFRP) Hughes Bros. No. 6 (GFRP) Hughes Bros. No.32 (GFRP) ISOROD (GFRP) NEFMAC (GFRP) ROTAFLEX (GFRP) LEADLINE (CFRP) ISOROD (CFRP) NEFMAC (CFRP) S&P Laminates (CFRP) NEFMAC (AFRP) NEFMAC (hybrid) Reinforcing Steel

Stress (MPa)

2000 1500 1000 500 0

0

1

2

3

Strain (%)

Figure 2.2: Manufacturer specified stress-strain behaviour of various currently available FRP reinforcing products (reproduced after ISIS, 2001a)

Normalized Axial Stress

2.5 2.0 1.5 1.0

Plain Concrete Steel Spiral GFRP-Wrapped CFRP-Wrapped

0.5 0.0

0

5

10

15

20

Normalized Axial Strain

Figure 2.3: Differences in behaviour for plain concrete, spirally reinforced, and FRP wrapped concrete cylinders under axial compression (reproduced after Spolestra and Monti, 1999)

72



2.6

3

Density (kg/m )

2.4 2.2 2.0 1.8

Basalt Aggregate Carbonate Aggregate Siliceous Aggregate

1.6 1.4

0

200

400

600

800

1000

o

Temperature ( C)

Thermal Conductivity (W/m.k)

Figure 2.4: Variation in density of concrete with temperature (reproduced after Schneider, 1988) 2.5

Siliceous Aggragate (model) Pure Quartz Aggregate (model) Carbonate Aggregate (model) Lightweight Aggragate (model)

2.0

Quartz Aggregate (exp.)

1.5 1.0

Anorthosite Aggregate (exp.) Expanded Shale Aggregate (ρ

1450 kg/m3) (exp.)

0.5 0.0

Expanded Shale Aggregate (ρ

0

200

400

600

3

1150 kg/m )

800

1000

o

Temperature ( C)

Figure 2.5: Variation in thermal conductivity of concrete with temperature (reproduced after Lie, 1992) Heat Capacity (MJ/m3.K)

5 4

Normal Weight

3 2 Lightweight

1 0

0

200

400

600

800

1000

o

Temperature ( C)

Figure 2.6: Variation in the heat capacity of concrete with temperature (reproduced after Lie, 1992)

73


-6

3

Thermal Capacity (×10 J/m .K)


20 18

Siliceous Carbonate Expanded Shale

16 14 12 10 8 6 4 2 0

0

200

400

600

800

1000

o

Temperature ( C)

Figure 2.7: Idealized variation in thermal capacity of concrete with temperature, for use in numerical modelling (reproduced after Lie, 1992)

Practical Design Curve

100

80

Practical Design Curves: Normal (model) Lightweight (model)

80

o

E(T) / E(20 C)

fc'(T) / fc'(20oC)

100

60 40 Abrams Schneider Malhotra Waubke

20 0

60 40

0

0

200

Lightweight Carbonate Basalt Siliceous

20

400

600

Temperature

800

1000

0

200

400

600

800

1000

o

(oC)

Temperature ( C)

(a)

(b)

% of Room Temperature Value Retained

Figure 2.8: Variation in (a) the ultimate compressive strength and (b) the modulus of elasticity of concrete with temperature (reproduced after Schneider, 1986)

100

Concrete Strength Concrete Modulus Steel Yield Strength or Modulus

80 60 40 20 0

0

200

400

600

800

1000

1200

o

Temperature ( C)

Figure 2.9: Temperature dependency of concrete and steel strength and modulus of elasticity as idealized for numerical modeling by Lie (reproduced after Lie, 1992)

74



40 0oC 100oC 200oC 300oC 400oC 500oC 600oC 700oC 800oC

Stress (MPa)

30

20

10

0 0.00

0.01

0.02

0.03

0.04

0.05

Strain

Figure 2.10: Stress-strain curves for 35 MPa concrete at different temperatures (reproduced after Lie, 1992) 1.5

1.2 Carbonate Aggregate Siliceous Aggregate Lightweight Aggregate

0.8 0.6 0.4

0.5 0.0 -0.5

0.2 0.0

0% 22.5 % 35 % 45 % 67.5 % 90 %

1.0

Strain (%)

Strain (%)

1.0

0

200

400

-1.0

600

0

200

o

400

600

800

o

Temperature ( C)

Temperature ( C)

(a)

(b)

12

50 Experimental Idealized for Modelling

3

45

Thermal Capacity (J/m .K)

Thermal Conductivity (W/m.K)

Figure 2.11: (a) Thermal strain of concretes with different aggregates and (b) effect of load level on thermal strains of concrete (reproduced after Lie, 1992)

40 35 30 25 20

10 8 6 4 2 0

0

200

400

600

800

1000

1200

Experimental Idealized for Modelling

0

200

400

600

800

1000

1200

o

o

Temperature ( C)

Temperature ( C)

(a)

(b)

Figure 2.12: Variation in (a) the thermal conductivity and (b) the specific heat of steel with temperature (reproduced after Lie, 1992)

75



Stress (MPa)

500 400 20oC 200oC 400oC 600oC

300 200 100 0 0.00

0.02

0.04

0.06

0.08

0.10

0.12

Strain

o

% of Elastic Modulus at 20 C

Figure 2.13: Stress-strain curves for 300 MPa yield-strength steel at different temperatures (reproduced after Lie, 1992)

100 80 60 40 8 mm Diameter Bar 12 mm Diameter Bar 25 mm Diameter Bar

20 0

0

200

400

600 o

Temperature ( C)

Figure 2.14: Variation in elastic modulus of mild steel with temperature (reproduced after Cooke, 1988)

Expansion (% of Original)

1.2 1.0 0.8 0.6

Transformation into Austenite

0.4 0.2 0.0

0

200

400

600

800

1000

o

Temperature ( C)

Figure 2.15: Variation in the thermal expansion of mild steel with temperature (reproduced after Lie, 1972)

76



7 Specific Heat (kJ/kg.K) Density (g/cm3) Thermal Conductivity (W/m.K)

6 5 4 3 2 1 0

0

200

400

600

800

1000

o

Temperature ( C) Figure 2.16: Variation of thermal conductivity, density, and specific heat with temperature for carbon/epoxy FRP (after Griffis et al., 1984)

Concrete Steel Wood GFRP

100

o

% of Strength at 20 C

120

80 60 40 20 0

0

200

400

600

800

1000

o

Temperature ( C)

Figure 2.17: Variation of strength with temperature for GFRP, timber, concrete, and steel (reproduced after Kodur and Baingo, 1998)

77



120

S-Glass (1) E-Glass (1) HTS Roving (1) E-Glass Roving (1) E-Glass (2) S-2 Glass (2) S-Glass (3) E-Glass (3) Glass (5) Aramid Fibre (3) Aramid (4) Aramid (5) HM-Carbon (3) Carbon (4) Carbon (5) Carbon Model (see App. D) Glass Model (see App. D) Aramid Model (see App. D)

% of Strength at 20°C

100

80

60

40

20

0 0

200

400

600

800

1000

(1) Rehm and Franke, 1979 (2) Sen et al., 1993 (3) Rostasy, 1992 (4) Sumida et al., 2001 (5) Dimitrienko, 1999

Temperature (°C) Figure 2.18: Variation in tensile strength of various fibres with temperature 120

PAN Carbon / Epoxy 1 (1) PAN Carbon / Epoxy 2 (1) PAN Carbon / Epoxy 3 (1) Pitch Carbon / Epoxy 1 (1) Pitch Carbon / Epoxy 2 (1) Braided Carbon / Epoxy (4) Stranded Carbon / Epoxy (4) Braided Carbon / Epoxy (5) Spiral Carbon /Epoxy (5) Carbon / Epoxy (6) Carbon / Epoxy (7) Carbon / Epoxy-Phenolic (7) Pitch Carbon / Cement (1) Carbon / Inorganic (5) Carbon / Glass / Vinyl Ester (2) Carbon / Glass / Vinyl Ester (3) Carbon / Polyimide (7) Model (see Appendix D)


100

80

60

40

20

0 0

100

200

300

400

500

(1) Kumahara et al., 1993 (2) Rahman et al., 1993 (3) Fujisaka et al., 1993 (4) Tanano et al., 1997 (5) Tanano et al., 1995 (6) Sumida et al., 2001 (7) Dimitrienko, 1999

Temperature (°C) Figure 2.19: Variation of strength of various carbon FRPs with temperature

78



120

Carbon / Glass / Vinyl Ester (1) Carbon / Glass / Vinyl Ester (2) Glass / PPS (3) Glass / Vinyl Ester (3) Polystal (4) Spiral Glass / Epoxy (5) Glass / Epoxy (6) Glass / Epoxy (7) Model (see Appendix D)


100

80

60

(1) Rahman et al., 1993 (2) Fujisaki et al., 1993 (3) Kumahara et al., 1993 (4) Clarke, 1993 (5) Tanano et al., 1995 (6) Dimitrienko, 1999 (7) Dimitrienko, 1997

40

20

0 0

100

200

300

400

Temperature (°C) Figure 2.20: Variation of strength of various glass FRPs with temperature 120

Aramid / Vinyl Ester (1) Braided Aramid / Epoxy (2) Arapree (3)

100

Aramid / Epoxy (1)


Braided Aramid / Epoxy(4) Aramid / Vinyl Ester (5)

80

Model (see Appendix D) (1) Kumahara et al., 1993 (2) Tanano et al., 1997 (3) Clarke, 1993 (4) Tanano et al., 1995 (5) Sumida et al., 2001

60

40

20

0 0

100

200

300

400

500

Temperature (°C) Figure 2.21: Variation of strength various aramid FRPs with temperature

79



120

PAN Carbon / Epoxy 1 (1) Pitch Carbon / Epoxy 1 (1) Pitch Carbon / Cement (1) Pitch Carbon / Epoxy 2 (1) Carbon / Glass / Vinyl Ester (2) Braided Carbon / Epoxy (3) Stranded Carbon / Epoxy (3) Carbon / Inorganic (3) Carbon / Inorganic (4) Braided Carbon / Epoxy (4) Spiral Carbon / Epoxy (4) Carbon / PEEK (5) Carbon / Bismalemide (6) Carbon / Thermoplastic (6) Carbon / Epoxy (7) Idealized Carbon / Epoxy

% Retained

100 80 60 2

R =0.60

40 20 0 -50

50

150

250

350

450

Temperature (deg. C)

(1) Kumahara et al., 1993 (2) Clarke, 1993 (3) Tanano et al., 1997 (4) Tanano et al., 1995 (5) Uematsu et al., 1995 (6) Gates, 1993 (7) Dimitrienko, 1999

(a) 120

Glass / PPS (1) Spiral Glass / Epoxy (3) Glass / Epoxy (4) Aramid / Epoxy (1) Braided Aramid / Epoxy 1 (2) Braided Aramid / Epoxy 2 (2) Stranded Aramid / Epoxy (2) Braided Aramid / Epoxy (3) Glass Model (see App. D) Aramid Model (see App. D)

% Modulus at 20°C

100

80

60

(1) Kumahara et al., 1993 (2) Tanano et al., 1997 (3) Tanano et al., 1995

40

20

0 0

100

200

300

400

500

Temperature (°C)

(b) Figure 2.22: Variation of elastic modulus of (a) carbon (b) glass and aramid FRPs with temperature

80


FRP Type (MATRIX/FIBRE)


300 Second Limit

PEEK/CA

Max. Value Limit

PH/CA

At 300 Seconds Maximum Value VE = vinyl ester EP = epoxy PH = phenolic PEEK = poly ether ether keytone GL = glass CA = carbon AR = aramid BMI = bismaleimide

PH/AR PH/GL BMI/CA EP/GL VE/GL 0

100

200

300

400

500

600

ASTM E662 Smoke Density

Figure 2.23: Results of smoke generation tests on various FRPs (after Sorathia et al., 1992)

Beam Dimensions and Reinforcement Details

Beam 6: Insulation Scheme

Insulation

Insulation Beam 7: Insulation Scheme

Figure 2.24: Details of selected FRP plated beams fire tested by Blontrock et al. (1999)

81


CHAPTER 3: Experimental Procedures

CHAPTER 3 EXPERIMENTAL PROCEDURES 3.1 General The main experimental study presented in this thesis involved 2 full-scale fire endurance tests to investigate the behaviour of circular FRP-wrapped and insulated reinforced concrete columns under exposure to a standard fire. A variety of different tests were conducted to this end, and all are described in this chapter. The experimental program can be divided into two main thrusts: full-scale fire endurance tests on circular FRP-wrapped and insulated reinforced concrete columns, and ancillary tests to determine the mechanical (and in some cases thermal) properties of the columns’ constituent materials (concrete, reinforcing steel, FRP, and insulating materials). Results of both ancillary and full-scale fire endurance tests are reported and discussed in Chapter 4.

3.2 Column Test Program The column test program involved the fabrication of 6 full-scale spirally-reinforced concrete columns, and the testing of 2 of these columns under exposure to a standard fire. All columns were fabricated and instrumented in the Structures Testing Laboratory at Queen’s University before being transported by truck to NRC, in Ottawa, where two were wrapped with CFRP, insulated, and tested under sustained axial compressive load during exposure to fire.

3.2.1 Column Fabrication and Design The column test specimens fabricated in this study were designed to be representative of typical circular columns that are common to reinforced concrete buildings and bridges in North America. The dimensions of the test furnace at NRC dictated an overall column height of 3810 mm.

The column diameter was subsequently chosen as 400 mm, a longitudinal steel

reinforcement ratio of approximately 2% was assumed, and the remaining details of the

82



reinforcement were determined in accordance with CSA A23.3: Design of Concrete Structures (CSA, 1994).

The resulting overall dimensions and reinforcement details are shown in

Figure 3.1. To test the columns in the column furnace at IRC/NRC, steel end plates 38 mm thick were required at both ends of the columns. A typical column base-plate with welded longitudinal and spiral steel is shown in Figure 3.2. These mild-steel plates measured 864 mm by 864 mm square, and were fillet-welded to the longitudinal reinforcing bars. The longitudinal steel for the columns consisted of eight 19.5 mm diameter deformed steel bars, distributed evenly around the column, and the spiral steel consisted of 11.3 mm diameter deformed steel bar with a centre-tocentre pitch of 50 mm. Both longitudinal and spiral reinforcement conformed to CSA Standard G30.12-M77, with a minimum yield strength of 400 MPa. An aluminum collar was used to maintain the tops of the reinforcing bars in position during construction (refer to Figure 3.3). Cement grout, consisting of 1 part Type 10 cement and 3 parts siliceous sand (with a maximum diameter of 0.5 mm) by mass with a water-to-cement ratio of 0.4, was used to fill the space below the top end plates to ensure a tight fit. The formwork for the columns consisted of 400 mm inside-diameter SonotubeTM formwork, which was anchored to the base plates using welded steel angles and tap-screws, and held in place at the top of the columns with plastic ties (again refer to Figure 3.3). The columns were all poured vertically in a single lift, with concrete supplied by a local ready-mix plant. Details of the mix proportions for the concrete are provided in Section 3.2.2.1. The concrete was placed using a large hopper attached to an overhead crane, and a series of rubber “Elephant Trunk” chutes of various lengths (fabricated specifically for this project to ensure that the concrete never free-fell more than 0.6 m while being placed, and so that the concrete placing operations could be conducted without damaging the instrumentation at the column mid height) were used to place the concrete within the formwork. The concrete was

83



hand-vibrated during each lift with a 3.65 m long, 32 mm diameter, wand vibrator to ensure excellent consolidation. Figure 3.4 shows various aspects of the pouring operations. Once cast, the columns were cured in a humidified plastic enclosure at 21 C to 24 C and 100% relative humidity for seven days, after which point the formwork was removed. The columns were allowed to cure in the Structures Testing Laboratory at Queen’s University, at ambient temperature and relative humidity, until they were ready for transport to NRC.

3.2.1.1

Concrete Mix

Two slightly different concrete mixes were used to fabricate the columns in this study, such that the influence of aggregate type on the fire behaviour of the columns could be examined. Two of the 6 columns were cast using concrete containing pure siliceous aggregate, which was supplied as 14 mm maximum aggregate size (MAS) crushed granite from Green’s Quarry near Kingston, Ontario. The remaining 4 columns were cast using pure carbonate aggregate, supplied as 14 mm MAS crushed limestone from Lafarge Canada’s limestone quarry in Napanee, Ontario. The proportions of the various constituents for both mixes, as supplied by the concrete ready-mix supplier, are given in Table 3.1. The mixes were designed specifically for this study based on the particular requirements for 35 MPa non-air-entrained concrete, with pure siliceous or pure carbonate aggregate. Through discussions with the ready-mix supplier it was decided that, given the requirements for an average strength of 35 MPa at 28-days, and given the method of placement of the concrete, the specified 28-day strength for the concrete mixes should be 28 MPa.

3.2.1.2

FRP Wrapping Scheme

The FRP-wrapping scheme was chosen based on discussions with Fyfe Co. LLC, of San Diego, California, who were an industrial partner in this project.

Because of the superior

thermomechanical behaviour of carbon fibres at high temperature (see Section 2.4.4.2 or Appendix D), it was decided that the Fyfe Co. Tyfo® SCH-30T unidirectional carbon/epoxy FRP

84



system (referred to herein as SCH) would be used for wrapping the columns, and that the adhesive matrix would consist of Fyfe Co. Tyfo® S Epoxy (referred to as S-Epoxy herein). Manufacturer specified properties of the SCH and S-Epoxy materials are given in Tables 3.2, 3.3, and 3.4. Because significantly different design procedures are currently used by Fyfe Co. Engineers as compared to either the American Concrete Institute (ACI, 2002) or ISIS Canada (ISIS, 2001a), there was initially some negotiation as to what the details of the FRP wrapping scheme should be. The design suggested by Fyfe Co. Engineers, which called for 2 layers of SCH sheet, was thought to be over-conservative and to provide excessive confinement. As such, the final wrap design for the columns consisted of a single layer of the SCH system, with a 300 mm overlap in the circumferential direction and a 25 mm overlap in the vertical direction. The overlap lengths chosen were based on best-practice guidelines and their effects on the confined behaviour of FRP-wrapped columns are not yet known (although they have not been studied in detail, they are not considered significant factors). The final wrap design, using a single layer of SCH wrap, was selected because it resulted in a theoretical ultimate load capacity increase of about 53% based on the ACI (2002) design guidelines, and of about 26% based on the ISIS (2001a) guidelines. The larger increases in design axial load capacity that would result from higher confinement ratios (with two layers of SCH) are thought to be unrealistic and would not currently be practical for design. Details of the final FRP wrapping scheme are provided in Figure 3.5, and a complete discussion of load calculations, based on various design guidelines, for the columns is included in Appendix E along with a discussion of current fire design recommendations as they apply to the columns tested herein.

3.2.1.3

Fire Protection Scheme

The literature review presented in Chapter 2 demonstrated that rapid and severe deterioration of the mechanical properties of FRPs can be expected at high temperature and

85



highlighted the need for thermal insulation of FRP wraps. Up to now, the approach used in industry to provide thermal insulation has been to apply an intumescent paint to the exterior of FRP wraps. In the event of a fire, the intumescent paint is expected to expand to many times its original thickness and form an insulating char that protects the FRP from the thermal insult of the fire. Because intumescent fire protection systems experience changes in physical and thermal properties under exposure to a heat flux, they are often referred to as active insulation. The primary shortcoming of these systems, however, is that the activation temperature – the temperature at which expansion and charring of the intumescent coating begins – for commonly available intumescent coatings is generally in the range of 200°C to 300°C, substantially higher than the GTT of most currently available polymer matrices.

As a result, the mechanical

properties of the FRP wrap (or the bond between the FRP and the substrate concrete) are likely to be lost, or at least seriously deteriorated, before the intumescent reaction occurs. Anecdotal evidence of this problem comes from a fire test on a GFRP-wrapped concrete column conducted in 1999 (not reported on in the literature), where an unidentified intumescent coating failed to protect the FRP wrap from fire, and the wrap burned and debonded from the column within 30 minutes of exposure to a standard fire. Indeed, analysis using the numerical computer model QCFIRE (developed and discussed in Chapter 5 of this thesis) also indicated that the mechanical strength of FRP wraps should deteriorate completely within 15 to 30 minutes of exposure, even with an intumescent protective coating applied. Because the insulation scheme used in the test program of this thesis incorporates an intumescent coating as part of the fire protection system for FRP-wrapped concrete columns, a more complete discussion of intumescent materials is provided in Section 5.2.6. To improve the behaviour of intumescent fire protection systems for use with FRP wraps, it was decided that a secondary passive layer of insulation was required between the FRP and the intumescent coating. The rationale for this addition was that the intermediate layer, which would

86



ideally consist of a passive layer of a highly insulative (low thermal conductivity and high heat capacity) and thermally inert material, would maintain the temperature of the FRP well below its GTT up until the onset of intumescence.

After activation of the intumescent coating, the

significant thermal insulation provided by the intermediate insulation in combination with the multi-cellular intumescent char would provide significant fire protection for the wrap for a substantial period of time. With the above rationale in mind, it was decided that a combination of fire protection materials now sold under trade names: Fyfe Co. Tyfo® VG (referred to herein as VG) and Fyfe Co. Tyfo® EI (referred to as EI), would be used to protect the 2 columns wrapped and tested during this study. VG represents the intermediate passive insulation, and EI was specified as the exterior intumescent coating. VG is a cementitious fireproofing material consisting of a gypsum binder with an expanded and pulverized vermiculite filler. Materials similar to VG were originally developed by the Vermiculite Association as early as 1958 (Type5, 1999) and have traditionally been used as spray or trowel applied fireproofing materials for structural steel. To the knowledge of the author, VG has not previously been applied to an FRP wrap. Although supplied by Fyfe Co. LLC, VG is produced by Southwest Vermiculite Co. under the trade name “Southwest Type 5GP”. Manufacturer specified properties for this material are presented in Table 3.5. VG was selected as the passive fire protection material herein because of its proven performance record as fire insulation for steel structures. It has an extremely low thermal conductivity and is thermally inert (retains its shape and bond to the substrate) up to temperatures in excess of 1000°C. When exposed to flames, VG plaster releases chemically combined water in the form of water vapor, which helps to maintain the plaster’s temperature below 100°C until all of the water has been driven off as steam. Meanwhile, the insulating action of the vermiculite filler (which gives VG its low thermal conductivity) delays the release of steam and retards the transmission of heat, thus improving overall fire-proofing characteristics. Unlike ordinary sand

87



plaster which can expand or explode during fires or when subjected to high temperatures, vermiculite plasters have low linear expansion characteristics, which greatly reduce the number and size of cracks. By reducing the number and size of cracks through the plaster, the heat of the fire cannot penetrate the plaster surface as quickly (Schundler, 2002). Details on the thermal properties of VG as a function of temperature are proprietary and difficult to obtain. Approximate thermal data is presented in Appendix F. EI is an intumescent coating developed for fire protection of structural steel and manufactured by Fire & Thermal Protection Engineers Inc., of Petersburg, Indiana. This material is a two-part coating comprised of bisphenol-A epoxy resin, polyphosphates, and an aliphatic polyamine compound. Because the material is proprietary, little additional information on its behaviour or composition is available. Some useful information on the activation temperature and thermophysical properties of EI was obtained using thermogravimetric analysis (TGA), as described in Section 3.3.4. Both of the FRP-wrapped columns tested herein were insulated with the VG/EI composite insulation scheme. Column ISIS-1 was protected with approximately 32 mm of VG and approximately 0.56 mm of EI, and Column ISIS-2 had approximately 57 mm of VG and 0.25 mm of EI. Installation procedures for the FRP wrap and the fire protection materials are described in Section 3.2.2. Table 3.6 gives details of the experimental program, including descriptions of the two wrapping and insulation schemes, and Figure 3.5 gives a visual representation of the insulation schemes.

3.2.2 Wrapping and Insulating Procedures Because this research project represents the first attempt at providing a specialized passive/active fire protection system for FRP-wrapped concrete columns, it is perhaps instructive to describe the procedures used, both to wrap and to insulate the columns. Much of the following information will be useful to future researchers, and so it is included here in significant detail.

88



FRP Wrapping During the installation of both the FRP wrap and the fireproofing materials, which generated copious amounts of dust and debris, the columns were placed inside a large plastic enclosure at IRC/NRC. The step-by-step procedure for application of the FRP wrap was as follows: 1. The surface of the columns was checked for any major defects or protrusions that could affect the ability of the FRP to bond to the substrate concrete. None were found. 2. The surface of the columns was lightly brushed with a heavy-duty scrub brush to remove any dust or loose debris. 3. A thin primer coat, consisting of S-Epoxy mixed with Cabot Cab-O-SilTM silicafume, was applied to the entire surface of the column by hand using a trowel. The objective was to fill any surface voids or minor imperfections and to ensure a strong bond between the SCH system and the concrete (refer to Figure 3.6). 4. The SCH carbon sheets were cut to the appropriate length and placed on a clean plastic drop sheet where they were saturated with epoxy using a standard paint roller (refer to Figure 3.7). 5. The SCH sheets were wrapped around the columns in 610 mm wide lifts, starting at the top and working toward the base of the column. The 300 mm overlapping vertical seam was staggered 90 degrees around the column for each subsequent lift of SCH sheet (refer to Figure 3.8). 6. Pressure was applied to the SCH sheets by hand to ensure that all air voids were removed from the FRP-concrete interface (refer to Figure 3.9). 7. The FRP was allowed to cure for 16 hours at ambient indoor temperature (approximately 18°C). The FRP-wrapped columns are shown in Figure 3.10.

89



VG Insulation Application To ensure a strong bond between the SCH system and the VG insulation, a single layer of galvanized steel diamond lath was secured to the column over its full height using steel concreteanchors and brackets. The VG was then spray-applied to the surface of the column in 20 mm lifts until the desired thickness was achieved. The procedure to apply the VG was as follows: 1. The surface of the FRP wrap was lightly sanded with a coarse-grit sand paper to ensure as strong a bond as possible with the VG (refer to Figure 3.11). 2. Steel shelving channel sections were attached vertically to the surface of the columns by placing 6 mm diameter Red HeadTM Trubolt threaded concrete anchors through the FRP wrap and into the substrate concrete. The steel channels were then bolted to the anchors at two locations each. Steel finishing washers (4 mm thick) were used to create a gap between the steel channel sections and the surface of the SCH (to prevent heat transfer directly from the steel channels to the SCH system). Figures 3.12 and 3.13 should assist the reader with visualization of the system used to attach the channels to the column. 3. Galvanized steel diamond lath (with 6 mm openings), traditionally used for plastering work, was attached to the steel shelving channels using conventional metal screws. The diamond lath was applied so as to completely cover the surface of the column such that it would provide reinforcement for the sprayed VG Insulation. Care was taken to ensure that the distance between the surface of the column and the diamond lath was constant (since it was critical to ensure a uniform overall thickness of VG). Refer to Figures 3.13 and 3.14 for clarification. 4. On each column, 4 circular depth guides were installed at various heights around the column. The depth guides consisted of PVC tubing of known outside diameter, wrapped circumferentially around the diamond lath. The intent was to use these guides to obtain a consistent depth of sprayed VG by leveling between the guides

90



with a straight section of aluminum angle. The guides were essential to level and screed the surface of the VG after spraying and to obtain a uniform surface. Refer to Figure 3.15 for clarification. 5. A thin layer of Fyfe Co. Tyfo® VG Primer (VGP) was applied to the surface of the columns using a general-purpose spray bottle. VGP is a bonding agent which helps to promote a strong bond between the water-based VG insulation and the hydrophobic S-Epoxy resin at the surface of the SCH wrap. The VG Primer was allowed to set for about 30 minutes before commencing the VG spray application (see Figure 3.16). 6. The VG insulation was spray applied in lifts approximately 20 mm thick using an EZ-TEXTM DX Electric Texture Sprayer until the desired overall thickness of VG was achieved (refer to Figures 3.17 and 3.18). The depth of the VG was verified periodically by running a straight aluminum section between adjacent depth guide rings. It was necessary to wait approximately 4 hours between lifts to ensure that the VG had hardened sufficiently for the next lift to be applied without localized sloughing-off of the entire thickness of VG material (which occurred twice during the VG spraying operation). 7. Once the desired thickness of VG had been applied, its surface was hand-trowelled to as smooth a finish as possible using drywall trowels (Figure 3.19). Any areas where additional VG material was required were filled in by hand using a trowel. Trowelling the surface of the VG was difficult however, and it is suggested that skilled labour (plastering contractors) be utilized for this portion of the work in the future. Column ISIS-2 was particularly difficult to trowel finish, and hence the final surface of this column was quite rough, adversely affecting the EI application (as discussed below). The VG was allowed to cure at room temperature overnight.

91



8. Once the VG had hardened sufficiently to remain in place while being manipulated, the circumferential depth guides were removed (Figure 3.20) and the resulting grooves were hand-filled with VG using a trowel. The columns were left to dry for 4 days before application of the first coat of EI. The finished VG coating is shown in Figure 3.21. EI Coating Application According to the manufacturer, EI can be applied by roller or by industrial paint sprayer onto most surfaces. Since spraying EI requires specialized equipment that was unavailable at the time of application, it was initially intended that the EI would be roller applied to the surface of the VG. Initial attempts at roller application failed however, because the EI, which is extremely viscous and sticky, tended to pull a layer of VG (approximately 5 mm thick) away from the surface of the column. It was subsequently determined that the EI was most easily applied to the surface of the VG by trowel, and this technique was eventually used. The design specifications for the EI material, as outlined by the material supplier, called for 0.51 mm and 0.25 mm thicknesses on columns ISIS-1 and ISIS-2 respectively. To achieve this approximate coverage of material, the required volumes of EI were determined based on the surface areas of each of the columns, and the amounts of material actually applied to the columns were monitored throughout the application process. It was observed during application of the first coat of EI that virtually all of the material was absorbed into surface voids in the VG (due to its rough, porous surface). The result was that little-to-no EI remained as a solid film on the exterior of the column. Thus, the first coat of EI was left to set and “seal” the surface of the VG for 5 days before a second coat was applied. The second coat of EI was also applied by trowel, and it was observed that the EI formed a thin rubbery film over the surface of the VG. Based on the volumes of EI applied to the columns, enough material was applied to column ISIS-1 (not considering the first coat of absorbed EI) for an average coating thickness of 0.56 mm, slightly more than the 0.51 mm

92



specified. For column ISIS-2 the approximate average thickness of EI was 0.25 mm, as specified by the initial design. Figure 3.22 shows column ISIS-1 after application of the EI. It is important to remain cognizant of the fact that the surface of the VG was porous and rough, so there was considerable uncertainty associated with the actual effective thicknesses of EI that were present on each of the columns at the time of testing. However, EI thickness does not turn out to be a significant factor in the fire behaviour of the columns in any case.

3.2.3 Column Instrumentation All columns were instrumented with thermocouples and electrical resistance strain gauges at the column mid height. Figure 3.23 shows the location and number of various sensors, and Table 3.7 provides details of sensor designations and locations. Thermocouples The thermocouples were fabricated by the author using chromel-alumel (Type-K) thermocouple wire and braised with an acetylene torch. Thermocouple frames were fabricated for the thermocouples in the concrete (thermocouples 4 to 7) by fastening individual thermocouples to 3.1 mm diameter steel drill-rod (Figure 3.24) and securely fastening the drill rod to the steel reinforcement cage with wire ties prior to pouring the concrete. Thermocouples 10 to 15 were attached directly to the reinforcing cage using wire ties (refer to Figure 3.25). Thermocouple wires for sensors placed inside the columns were threaded up the sides of the vertical reinforcing bars inside the reinforcing cage and exited the side of the columns just below the top end-plate. Thermocouples 2 and 8, at the FRP/concrete interface, caused special problems in installation, since it was desired that the surface of the column be completely smooth prior to wrapping. Shallow vertical grooves were ground into the side of the columns from the top-plate to the column mid-height using a hand grinder. The wires for thermocouples 2 and 8 were placed inside the grooves and run down the sides of the columns to the appropriate locations. The

93



grooves were then filled in with an epoxy patching compound developed specifically for concrete surface repair in FRP-wrapping applications. Thermocouples 1, 9, 16 and 17 were attached to the exterior of the FRP wrap using 5-minute epoxy (Figure 3.26). Thermocouples 18 and 19 were attached to the steel diamond lath with wire ties before commencing spray application of the VG insulation (Figure 3.27). These thermocouples were manipulated by hand during the finishing of the VG to ensure that they were barely visible, located just below the surface of the VG. Strain Gauges The number and location of strain coupons in the columns are depicted in Figure 3.23. Strain gauges were installed inside the column at its mid-height in an attempt to gain insight into the effectiveness (or lack thereof) of the FRP wrap and to detect any bending in the column during the tests. The assumption was that if the FRP wrap lost effectiveness during the fire test, this would cause an increased strain in the concrete and hence an observable increased strain in the longitudinal reinforcing bars. In hindsight, because the strains in the FRP wrap were so small at the load level applied to the columns during the fire endurance tests, loss of the wrap would likely not cause an observable change in reinforcement strains. The strain gauges malfunctioned during the fire testing in any case, as discussed in Chapter 4. High temperature strain coupons were fabricated by the author and installed on the longitudinal reinforcing steel at 4 locations, distributed evenly around the column cross-section at mid-height. Figure 3.28 shows a single strain coupon installed in the reinforcing cage just prior to casting the columns.

The high-temperature strain coupons were fabricated by installing

KyowaTM KFU-5-120-C1-11 high-temperature foil gauges on 150 mm long coupons of 3 mm by 13 mm cold-flat-finished mild steel plate.

The gauge lead-wires were soldered to high-

temperature polyamide terminal pads using high-temperature silver solder, and high-temperature KyowaTM Type L-4 strain gauge wire was soldered to the terminal pads. The strain gauge wires were sheathed with protective PVC tubing, which served to provide mechanical protection for the wires while the columns were cast, as well as waterproofing protection inside the hardened

94



concrete.

A surface protection and waterproofing layer was applied to the gauge/lead-

wire/terminal pad assembly using SuperflexTM Red High Temp RTV Silicone Adhesive Sealant, manufactured by Loctite Corporation. Once fabricated, the metal coupons were tack-welded to the vertical reinforcing bars, and the lead wires ran up the sides of the vertical reinforcing bars and exited the forms just below the top end-plate.

3.2.4 Fire Test Procedures Test Furnace Both columns were tested by exposing them to heat in a specially designed furnace at IRC/NRC that was built specifically for testing loaded full-scale columns under fire exposure (Figure 3.29). The test furnace, which was designed to reproduce the temperature, load, and heat transfer conditions to which a member might be exposed during a building fire, consists of a steel framework supported by 4 steel columns. The furnace chamber sits inside the steel framework, and 3 hydraulic jacks can be used to produce load in a column along the three principal axes. A jack below the floor of the test chamber applies axial load to the column with a maximum capacity of 9790 kN. The heating chamber of the test furnace is 2642 mm square and 3048 mm high, and is lined with high-emissivity insulating materials which effectively transfer heat to the specimen being tested. Heat is supplied by 32 propane burners arranged in 8 vertical lines, with a total thermal output of 4700 kW. Uniformity of heat throughout the furnace is achieved by automatically and individually adjusting the propane burners using information provided by 8 thermocouples located at various heights throughout the heating chamber at a distance of 305 mm from the test specimen. The location of the thermocouples within the furnace chamber is depicted in Figure 3.30. The applied load is measured with load cells that have an accuracy of ±20 kN at low load levels, and considerably better accuracy at higher load levels (Lie and Woolerton, 1988). The

95



axial deformation of the column is measured using transducers with an accuracy of ±0.002 mm, mounted on the jack that supports the column. A more detailed description of the test furnace is given by Lie (1980). End Conditions The column tests in this study were conducted with both ends restrained against rotation and horizontal translation (a fixed-end condition). This was accomplished by bolting the bottom end-plate to the hydraulic jack and the top end-plate to the loading head using eight 19 mm diameter bolts each.

Fixed-end end-conditions are the standard for full-scale column fire

endurance tests, since they most accurately simulate the end-conditions to be expected for concrete columns in an actual structure. Loading Before the fire endurance tests could be conducted, it was required to determine the concentric axial sustained load that would be applied to the columns during the tests. Both ULCS101 (CAN/ULC, 1989) and ASTM E119 (ASTM, 2001) state that, throughout a fire endurance test, columns shall be exposed to fire on all sides and shall be loaded in such a manner as to develop, as nearly as practicable, the working (service) stress contemplated by the design. In conventional cases (i.e. for steel reinforced concrete or structural steel columns) this load is determined by calculating the factored ultimate strength of the column in question and, assuming some ratio of dead-to-live load (between 0.25:1 and 3:1), back-calculating the unfactored service loads. A complication that arises for the case of an FRP-wrapped column is that there remains considerable uncertainty as to what the design load capacity of the column actually is, since no legally binding design codes are currently available for FRP wrap design. The Canadian (ISIS, 2001a) and American (ACI, 2002) design procedures yield substantially different design strengths for FRP-wrapped columns, and hence a judgment call must be made to determine which set of design recommendations to follow. In the present case, it was decided that the Canadian design

96



strength, calculated using ISIS (2001a), should be used to estimate the design load capacity, and the service loads were back-calculated from this value. The actual tested properties have been used in the load calculations for all materials involved. Tested material properties are discussed in detail in Chapter 4 of this thesis and were as follows: Concrete Compressive Strength: f c,

39.5 MPa

Yield Strength of Longitudinal Reinforcing Steel: f y Ultimate Tensile Strength of SCH FRP System: f frp ,ult Ultimate Strain of SCH FRP System: ε frp ,ult Elastic Modulus of SCH FRP System E frp

456 MPa 1510 MPa

1.64 % 90.2 GPa

According to the ISIS Canada design guidelines, the confining pressure at ultimate, fl, is calculated as follows:

fl

2 n φ frp f com t w d

2 1 0.75 1510 0.76 400

4.30 MPa

[3.1]

where φfrp is equal to 0.75 for CFRP with an interior conditioned exposure, n is the number of layers of FRP wrap, fcom is the ultimate tensile strength of the FRP wrap, tw is the wrap thickness, and d is the column diameter.

The ISIS Canada guidelines specify a maximum effective

confinement pressure, fl(max), as follows:

f l (max) where

pc

f c, 2

pc

1 ke

φc

39.5 1 0.6 2 1.0 1.0

7.8 MPa

[3.2]

= 1.0 and k e = 1.0 for a round column, and f’c is the unconfined concrete compressive

strength. The volumetric confinement ratio, ωw, is calculated as:

2 fl φc f c,

w

2 4.30 0.6 39.5

0.3697

[3.3]

The confined ultimate strength of the concrete, f’cc, is determined from:

f cc,

f c, (1

pc

w

)

39.5 (1 1.0 0.3697 ) 53.1 MPa

[3.4]

Finally, the design strength of the FRP-wrapped column is determined from:

97



Pr max

0.85

φ f cc' (Ag

1 c

Ast ) φ s f y Ast

0.85 0.79 0.6 (53.1) 3430 kN

(

(200)2

)

8 (300)

0.85 456 (8 300)

[3.5]

where Ag is the gross cross-sectional area of concrete, Ast is the area of longitudinal reinforcing steel in compression, φc is equal to 0.6, φs is equal to 0.85, fy is the yield strength of the reinforcing steel, and α1 is equal to 0.85-0.0015·f’c. Finally, if a dead-to-live load ratio of 1:1 is assumed, the service load, S, on the column can be back-calculated using the CSA A23.3 load factors of 1.25 for dead and 1.5 for live. Thus:

3430 1.5 S LL 1.25 S DL S LL

1143 kN and

and 1.5 S LL

1.25 S DL

S DL 1372 kN

which results in a service load of 1143 + 1372 = 2515 kN. This load was applied to both columns during the fire endurance tests described herein. A more detailed discussion of the load calculations for the columns tested herein, including loads resulting from currently available American design guidelines (ACI, 2002), is presented in Appendix E. Fire Exposure The fire exposure for the columns was simulated by controlling the average temperature in the furnace such that it followed as closely as possible the standard fire time-temperature curve of ULC-S101 (CAN/ULC, 1989). This time-temperature history is shown in Figure 3.31 and is equivalent to the American (ASTM E119) standard fire. Past tests performed at the IRC/NRC facility have indicated that deviations in the furnace temperature from the standard timetemperature curve are generally less than 2% (Lie, 1980). Actual furnace time-temperature curves recorded during both tests are presented in Chapter 4. Failure Criterion The columns were considered to have failed, and the tests stopped, if the hydraulic jack, which has a maximum speed of 76 mm/minute, could no longer maintain the sustained load. If

98



this failure criterion was not reached during the first 5 hours of the test, the load was increased until the column failed and the test was stopped after 5½ hours, beyond which time damage to the test furnace could occur.

3.3 Ancillary Test Program The ancillary test program consisted of a variety of standard tests, primarily to determine the room temperature mechanical properties of the columns’ constituent materials. In some cases, tests were also performed to examine thermal properties of the various constituents. Results from the ancillary test program are presented and discussed in Chapter 4.

3.3.1 Concrete Tests A total of 36 concrete cylinders (15 with siliceous aggregate and 21 with carbonate aggregate) with radius 75 mm and height 200 mm was cast in conjunction with the column fabrication such that a minimum of 3 cylinders would be available for compressive testing at twenty-eight days and at the time of the fire test (for each of the six columns). Compressive strength tests on concrete were conducted according to test method CSA A23.2-9C: Compressive Strength of Cylindrical Concrete Specimens (CSA, 1990) using a Reihle 1350 kN testing machine. In addition to the concrete cylinders, 6 concrete bricks 100 mm × 200 mm × 50 mm were fabricated for each of the 2 concrete mixes. These blocks were cast such that the thermal conductivity of the concrete mixes could be determined if necessary. The relative humidity (RH) of the concrete in the columns was measured just prior to testing to determine the approximate moisture content of the concrete; this information is important for numerical modelling since water vapor evaporation from concrete has a significant effect on the heat transfer within concrete in the temperature range 100°C to 200°C. Concrete with very high moisture content is also prone to spalling during fire endurance tests, and so it was desired to verify that spalling was not likely to be a problem. The RH was measured by drilling

99



into the concrete at three locations (top, mid-height, and bottom) of the columns, inserting a VaisalaTM handheld humidity probe, and plugging the hole's opening with plasticine. The RH data obtained from the humidity meter was used to estimate the moisture content of the concrete using the procedure outlined in Appendix A2 of ULC-S101 (CAN/ULC, 1989). The thermomechanical properties (variation in thermal conductivity, density, specific heat, strength, and stiffness with temperature) of the concrete were not tested for the purposes of this thesis. It was decided that such tests were not warranted given the wealth of information available in the literature (refer to Chapter 2) and the time and effort required to perform the tests.

3.3.2 Reinforcing Steel Tests Tensile tests at room temperature were conducted on reinforcing steel used for both the spiral and longitudinal reinforcement of the full-scale reinforced concrete columns. Tests were conducted according to ASTM A370-97A: Standard Test Methods and Definitions for Mechanical Testing of Steel Products (ASTM, 1997b). Bars were tested in groups of three on specimens cut from at least two different sections of reinforcement (to verify uniformity of the results). The stress-strain characteristics of the steel bars were obtained using a standard flat load cell in conjunction with a long-gauge MTS extensometer. Again, a Reihle 1350 kN testing machine was used. Thermal properties of the steel were not tested and have been assumed based on information provided in Chapter 2.

These properties are relatively well documented and

accepted.

3.3.3 Fibre-Reinforced Polymer Tests Tensile tests on FRP coupons at room temperature were conducted according to ASTM D3039 (ASTM, 1976). A photo of an SCH tensile test coupon and its instrumentation prior to testing are shown in Figure 3.32. Five coupons were fabricated from SCH sheets using S-Epoxy as a matrix material. The SCH coupons consisted of a single ply of SCH sheet, with 51 mm wide

100



cross-plies of SCH sheet installed on both faces in the anchorage zones. A large flat sheet of cured SCH material was fabricated and allowed to cure, after which it was saw cut into 25 mm wide, 150 mm long test coupons. A single 5 mm electrical resistance foil strain gauge was installed on each of the test coupons at the mid-length prior to testing. Load and longitudinal strain were monitored throughout the tests, which were conducted using an MTS Materials Testing System in the Structures Laboratory of the Royal Military College of Canada (RMC).

3.3.4 Thermogravimetric Analysis Thermogravimetric analysis (TGA) refers to a testing procedure in which the change in mass of a sample with temperature is monitored over a range of temperatures.

For many

materials, TGA can be instructive in providing information on the thermal behaviour (dehydration, decomposition, combustion, etc.) at elevated temperatures. TGA was conducted on at least 3 samples each of SCH fibres, Tyfo® SEH-51A (glass) fibres (SEH), S-Epoxy, VG insulation, and EI paint. These tests were performed according to ASTM E1131: Test Method for Compositional Thermal Analysis by Thermogravimetry (ASTM, 2000) with a heating rate of 10°C per minute. The results of the TGA analyses are presented in Chapter 4.

101



Table 3.1: Concrete mix design proportions (supplied by Lafarge Kingston) Mix Designation Specified 28-Day Strength (MPa) Aggregate Type Maximum Aggregate Size (mm) Type 10 Cement (kg/m3) Coarse Aggregate (kg/m3) Fine Aggregate (kg/m3) Water (kg/m3) Mid-Range Water Reducer Slump (mm) Water-Cement Ratio

#1 28 Crushed Granite 14 280 1,020 980 152 As per MBT* specs Max 150 0.54

#2 28 Crushed Limestone 14 280 1,070 980 152 As per MBT* specs Max 150 0.54

*MBT = Master Builders Technologies Inc.

Table 3.2: Typical dry fibre properties of Tyfo® SCH-30T FRP (supplied by Fyfe Co.) Property Tensile Strength (MPa) Tensile Modulus (MPa) Ultimate Elongation (%) Density (kg/m3) Weight/m2 (kg) Nominal Fibre Thickness (mm)

Value 4,900 230,000 2.1 1,800 0.457 0.254

Table 3.3: Typical laminate properties for Tyfo® SCH-30T FRP (supplied by Fyfe Co.) Property Ultimate Strength (MPa) Elongation at Break (%) Tensile Modulus (GPa) Nominal Thickness (mm)

ASTM Method D-3039 D-3039 D-3039 D-3039

Test Value 1,351 1.4 95.2 0.76

Design Value 1,149 1.4 80.7 0.76

Table 3.4: Typical properties for Tyfo® S epoxy (supplied by Fyfe Co.) Property Glass Transition Temperature (°C)* Tensile Strength (MPa) Tensile Modulus (MPa) Elongation Percent (%) Flexural Strength (MPa) Flexural Modulus (MPa)

ASTM Method D-4065 D-638 Type 1 D-638 Type 1 D-638 Type 1 D-790 D-790

Test Value 93 72.4 3,180 5.0 123.4 3,120

* Curing schedule 72 hours post cure at 60°C

102



Table 3.5: Manufacturer specified properties of Tyfo® VG (supplied by Fyfe Co. and Type5.com) Property Density (kg/m3) Compressive Strength (kPa) Bond Strength (kPa) Combustibility Surface Flame Spread, Smoke Developed Erosion by Air (kg/m3) Effect of Impact Effect of Deflection Corrosion of Steel Thermal Conductivity @ -14°C (W/m·°C) * ρ = 269 kg/m3 ρ = 291 kg/m3 ρ = 372 kg/m3 Thermal Conductivity @ ρ = 762 kg/m3 (W/m·°C) * Temp. = 38°C Temp. = 260°C Specific Heat* (J/kg·°C) * Temp. = -18°C Temp. =149°C *

ASTM Method -E-761 E-736 E-136 E-84 E-859 E-760 E-759 E-937

Test Value 240-272 112 18.6 Non-Combustible 0, 0 0.00 Passed Passed Passed

----

0.0815 0.0919 0.1142

---

0.1673 0.1716

---

1004 1047

Proprietary Information, Patents Pending

Table 3.6: Details of the column test program

* ŧ

Column

Date Poured

Date Tested

Agg. Type

FRP Wrap

Insulation

Test Purpose

ISIS-3

27/09/01

--

CAŧ

None

None

CA-Baseline

ISIS-5

27/09/01

--

SAŧ

None

None

SA-Baseline

ISIS-4

27/09/01

--

CA

ISIS-1*

27/09/01

17/12/02

CA

ISIS-2*

27/09/01

26/11/02

CA

ISIS-6

27/09/01

--

SA

Tyfo® SCH-30T 1-layer Tyfo® SCH-30T 1-layer Tyfo® SCH-30T 1-layer Tyfo® SCH-30T 1-layer

None 32mm VG 0.56mm EI 57mm VG 0.25mm EI TBA

Uninsulated Wrap Insulation Scheme 1 Insulation Scheme 2 SA-Insulation Scheme 3

Column tests reported in this thesis CA – Carbonate Aggregate, SA – Siliceous Aggregate

103



Table 3.7: Instrumentation for the column tests£ Name

Location*

Type

TC1 TC2 TC3 TC4 TC5 TC6 TC7 TC8 TC9 TC10 TC11 TC12 TC13 TC14 TC15 TC16 TC17 TC18 TC19 Strain1 Strain2 Strain3 Strain4 FT1 FT2 FT3 FT4 FT5 FT6 FT7 FT8 Load Disp.

VG / FRP Interface FRP / Concrete Interface Inside Concrete (D/8) Inside Concrete (D/4) Inside Concrete (3D/8) Concrete Centreline Inside Concrete (D/4) FRP / Concrete Interface VG / FRP Interface Top of Spiral Reinforcement Outside Edge of Longitudinal Rebar Inside Edge of Longitudinal Rebar Inside Edge of Longitudinal Rebar Outside Edge of Longitudinal Rebar Top of Spiral Reinforcement VG / FRP Interface VG / FRP Interface EI / VG Interface EI / VG Interface On Longitudinal Rebar 1 On Longitudinal Rebar 2 On Longitudinal Rebar 3 On Longitudinal Rebar 4 Furnace Interior (2440 mm height)ŧ Furnace Interior (1220 mm height)ŧ Furnace Interior (1830 mm height)ŧ Furnace Interior (610 mm height)ŧ Furnace Interior (1830 mm height)ŧ Furnace Interior (610 mm height)ŧ Furnace Interior (2440 mm height)ŧ Furnace Interior (1220 mm height)ŧ Total Applied Load (@ base) Overall Column Elongation (@ base)

Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Axial Strain (µε) Axial Strain (µε) Axial Strain (µε) Axial Strain (µε) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Temp. (°C) Axial Load (kN) Axial Stroke (mm)

Concrete Cover (mm) -0.76 0 50 100 150 200 100 0 -0.76 45 50 70 70 50 45 -0.76 -0.76 -32.5 or -57.9¥ -32.5 or -57.9 60 60 60 60 -505ŧ -505 -505 -505 -505 -505 -505 -505 N/A N/A

£

Refer to Figure 3.22 * All sensors were located at the column mid-height unless otherwise noted ¥ Cover to TCs 18 and 19 depended on the thickness of VG installed ŧ Refer to Figure 3.30

104



864 400

864

No. SPECIMENS: 6 COLUMNS CONCRETE: 35 MPa 28-DAY STRENGTH REINFORCEMENT: 8 19.5 mm DIAMETER BARS LONGITUDINAL 11.3 mm DIAMETER SPIRAL w/ 50 mm PITCH c/c 40 mm COVER TO SPIRAL 50 mm COVER TO PRINCIPAL REINFORCEMENT

SECTION A-A N.T.S.

*all dimensions in mm A

A

3810 3734

50 mm pitch c/c

10M Spiral

Concrete Rebar

38 mm Thick Steel Plate

13 mm Diameter Machined Bar Ends

ELEVATION N.T.S.

Figure 3.1: Column dimensions and reinforcement details

105



Spiral Steel

Longitudinal Reinforcing Steel

Base Plate

Figure 3.2: Typical column base plate with welded longitudinal and spiral reinforcement

Machined Ends of Rebars

Aluminum Collar

SonotubeTM Formwork

Plastic Ties

Figure 3.3: Top of column reinforcement cage including the aluminum collar used to maintain the longitudinal bars in the correct location curing fabrication and pouring

106



Thermocouple Lead-wires

(a)

(b)

(c) Figure 3.4: Casting of the concrete columns (a) Top of forms with aluminum collar and exiting instrumentation wires (b) Forms inside scaffold, crane, and hopper (c) Pouring concrete into “elephant trunk” chute with hopper

107



11.3 mm Diameter Spiral

25 mm overlap at all butt splices

50 mm pitch c/c

Tyfo Tyfo Tyfo Tyfo Tyfo

S Prime coat SCH-30T (See Table 1) VG Primer VG (See Table 1) EI (See Table 1) Table 1 ISIS-1 Single ply of Tyfo SCH-30T 610 mm (24") wide bands to full height of column.

ISIS-2 Single ply of Tyfo SCH-30T 610 mm (24") wide bands to full height of column.

Tyfo VG

Applied in 2 lifts to the maximum thickness possible; approx 38 mm/1.5 in. or greater

Applied in 3 lifts to the maximum thickness possible; approx. 64 mm/2.5 in. or greater

Tyfo EI

20 mils

10 mils

Tyfo SCH-30T

Concrete Rebar * 300 mm overlap at all circumferential splices for Tyfo SCH-30T sheets

All dimension are in millimiters U.O.S.

Figure 3.5: Details of the FRP wrapping scheme for Columns ISIS-1 and ISIS-2

Figure 3.6: Application of the epoxy prime coat

Figure 3.7: Saturation of SCH sheet with epoxy

108



Figure 3.8: Application of a single lift of SCH sheet

Figure 3.9: Removal of air voids at the FRP/concrete interface

ISIS-2

ISIS-1

Figure 3.10: Columns ISIS-1 and ISIS-2 with SCH sheets installed

109



Steel Shelving Channel

Finishing Washer

Hex-Nut

Concrete Anchor Head

SCH Sheet

Figure 3.11: Sanding the surface of the SCH sheets prior to application of the VG primer

Figure 3.12: Photo showing connection detail of steel angle bracket to concrete column

A

¼” Diamond Lath

Concrete Steel Shelving Channel

Depth Guide Tubing

Tyfo® SCH-30T Wrap

Finishing Washer

Hex Nut Flat Washers

Red Head Anchor Drilled Hole

Metal Screw

A 3.7

10.3

1.8

19.1 34.9

Section A-A

Elevation

NOTE: the depth guide tubing for column ISIS-2 had an outside diameter of 42.3 mm, resulting in an overall depth of Tyfo® VG of 58.1 mm

Figure 3.13: Schematic showing connection detail for insulation brackets and diamond lath

110



Figure 3.14: Column ISIS-1 with diamond lath installed

Figure 3.15: Photo of ring spacer on column ISIS-2

Figure 3.16: Application of VG primer

Figure 3.17: Application of VG

Figure 3.18: Columns between 1st and 2nd lifts of sprayed VG

Figure 3.19: Trowel application of VG to obtain desired thickness and surface finish

111



Figure 3.20: VG insulation immediately after removal of spacer ring on column ISIS-1

ISIS-2

ISIS-1

Figure 3.21: Columns ISIS-1 and ISIS-2 after application of VG

Figure 3.22: Column ISIS-1 after application of EI

112



Tyfo

S Prime coat

Tyfo

VG Primer

SCH-30T

Tyfo

VG

Tyfo

EI

Tyfo D

9

16 18

2 3 4 5 6

7

8

1

SG1 10 12

19

D/4

18

SG3

D/4

9, 17

1, 16

2 3 4 5 6

SG2 SG4

14

11

17

5 TCs @ D/8

13 15

7

8

13

10 11

19

14 15

SG1 SG4

12

(a)

SG2 SG3

(c)

(b)

Figure 3.23: Location and number of various sensors for the column tests (a) Thermocouples in the concrete, FRP, and insulation (b) Thermocouples on the reinforcing steel (c) Strain gauge coupons welded to the vertical reinforcing steel

Thermocouple Lead Wires

TC4 TC5

Thermocouple Frame

TC6

TC7

Figure 3.24: Thermocouple frame for sensors TC3 to TC7 before being installed on the reinforcement cage

113



TC9 TC1

TC10

TC11

Figure 3.25: Thermocouples TC10 and TC11 installed on reinforcing steel

Figure 3.26: Placement of TC1 and TC9 installed on outside of FRP

TC18

Figure 3.27: Thermocouple TC18 before application of VG

114



Strain gauge beneath silicone coating

Lead-wire inside PVC sheathing

Metal coupon

Tack-welded

Spiral steel

Tack-welded

Figure 3.28: Strain coupon installed on the longitudinal reinforcing steel (rotated 90°)

Figure 3.29: Photo of the FRM/IRC/NRC column testing facility with column ISIS-2 inside ready for testing

115



2642 816 505 505 1, 2 5, 6

2642

3, 4

7, 8 Furnace Door PLAN

Furnace Chamber

1,7 610 5

3

610

2,8

3048

610

6

4

610 * all dimensions are in mm U.O.S.

ELEVATION

Figure 3.30: Locations of thermocouples inside the column furnace

1200

o

Temperature ( C)

1000 800 600 400 200 0

0

50

100

150

200

250

300

Time (min) Figure 3.31: ULC S101 time-temperature curve for fire endurance tests (reproduced after CAN/ULC, 1989)

116



Strain Gauge

50 mm

150 mm

50 mm

Figure 3.32: SCH coupon used for tensile testing of FRP

117


CHAPTER 4: Experimental Results and Discussion

CHAPTER 4 EXPERIMENTAL RESULTS AND DISCUSSION 4.1

General Experimental results for both ancillary tests and full-scale fire endurance tests are

presented and discussed in this chapter. Ancillary test results are presented first, since they provide insight into some of the observations from the full-scale fire endurance tests. Further discussion of the test results as they apply to validation of the QCFIRE numerical models is included in Chapter 5.

4.2 Ancillary Tests 4.2.1 Concrete Results of compression tests on concrete cylinders are presented in Table 4.1. Data are presented for each of the two concrete mixes (siliceous or carbonate aggregate) at 28-days and at the time of testing. It is evident that the desired 28-day compressive strength of approximately 35MPa was achieved for both concrete mixes, even though the mix design was actually performed for a 28-day strength of 28 MPa. Both columns tested herein (ISIS-1 and ISIS-2) were fabricated from the same carbonate aggregate concrete mix. The average concrete strengths at the time of testing were 40 MPa for column ISIS-1 and 39 MPa for ISIS-2. These concrete strengths were used to determine the appropriate applied loads for the full-scale fire endurance tests as described in Appendix E. The relative humidity of the concrete at the time of testing and the resulting calculated approximate volumetric moisture content, both included in Table 4.1, were relatively high for the columns, which had been stored in a conditioned environment for more than 1 year. However, because both columns were provided with a significant amount of supplemental fire insulation, it was not felt that spalling was likely to be a significant issue during the fire tests unless the

118



insulation fell off of the columns. Indeed, during the full-scale fire endurance tests, spalling was observed in the concrete only after prolonged exposure to fire and was coincident with failure of the VG insulation.

4.2.2 Reinforcing Steel The results of tensile tests on the longitudinal and spiral reinforcing steel used in the fabrication of the fire test specimens are given in Table 4.2. Figures 4.1 to 4.4 show complete and elastic-region stress-strain plots for both types of reinforcement. In Figure 4.1, it is evident that the 20M bar displayed classical stress-strain behaviour for mild steel rebar, with well-defined elastic, yielding, and strain hardening phases. Conversely, the 10M bar did not display a welldefined yield plateau, likely because the tensile specimens used in the tensile tests were cut from a section of straightened spiral that had experienced considerable inelastic deformation prior to the tension tests. The yield point for the 10M bar was consequently defined in terms of the 0.2% offset stress (see Figure 4.4).

4.2.3 Fibre-Reinforced Polymer Results of tensile tests on SCH FRP coupons conducted at room temperature are presented in Figure 4.5 and Table 4.3. All 5 tensile coupons displayed similar stress-strain behaviour, with similar ultimate stress, failure strain, and elastic moduli. The data obtained in tests also compare favourably with manufacturer specified properties for this material (see Table 3.3). Failure of all SCH specimens was sudden and violent, but was preceded by numerous cracking sounds.

Figure 4.6 shows an SCH tensile coupon in the MTS testing machine

immediately after tensile failure.

4.2.4 Thermogravimetric Analysis As outlined in Chapter 3, TGA was conducted on a number of the materials used in this study. Results and discussion of all TGA tests are included in this section.

119



S-Epoxy The results of TGA on S-Epoxy are shown in Figure 4.7. The test data demonstrate that at least 95% of the room temperature mass of the S-Epoxy samples was retained at temperatures up to 390°C. Between 390°C and 510°C, approximately 90% of the mass of the S-Epoxy samples was lost in an essentially linear fashion. This mass loss indicates severe degradation of material properties and was accompanied by the generation of large amounts of dense black smoke. Auto-ignition and free-burning of the S-Epoxy samples was observed at about 450°C. It is interesting to note that the relationship between specific heat and temperature for a carbon FRP – as quoted by Griffis et al. (1984), included in Figure 4.7, and used in the numerical modelling of Chapter 5 – shows a marked increase in the region of 343 C to 510 C to simulate the effect of epoxy degradation. This increased region in the Griffis et al. specific heat data coincides well with the mass loss region observed in the S-Epoxy TGA data. This gives credence to the specific heat versus temperature relationships used for the thermal properties of FRP in the numerical modelling of this thesis (refer to Chapter 5 and Appendix F). It is also interesting that the GTT for S-Epoxy, the temperature beyond which the strength and stiffness of the polymer are severely degraded, is quoted as 93°C by the manufacturer, well below the temperature at the onset of mass loss recorded in TGA data. Thus, two potential critical temperatures are possible for the SCH system incorporating S-Epoxy: first, above 93°C severe reductions in strength, stiffness, and bond properties are expected; and second, above about 390°C decomposition of the S-Epoxy is expected. Free combustion may occur near 450°C. The implications of choosing one critical temperature versus another are discussed later in this chapter, and further in the parametric discussions of Chapter 5. SCH Fibres Results of TGA on raw SCH fibres are shown in Figure 4.8. Also shown in Figure 4.8, for the purposes of comparison, is TGA data for raw Tyfo SEH-51A glass fibres. Both the carbon and glass fibres retained at least 90% of their mass up to 750ºC. In fact, the SEH fibres

120



retained 92% of their mass up to 1000ºC, while the carbon fibres showed a marked decrease in mass above 750ºC, likely as a result of oxidation. Figure 4.9 shows a comparison of the variation in density of an SCH/S-Epoxy composite with a fibre volume fraction of 0.5 with temperature based on the rule of mixtures. Also shown in this figure is the variation in density of a carbon/epoxy FRP with the same fibre volume fraction as tested by Dimitrienko (1999). There is relatively good agreement between the two data sets within the range of temperatures considered, particularly considering that numerous carbon/epoxy combinations are possible and that the specific materials tested by Dimitrienko are unknown. VG Insulation Figure 4.10 provides the results of TGA on samples of VG insulation and demonstrates that 50 mm cube samples experienced a mass loss of about 15% between 100°C and 500°C. Beyond 500°C, virtually no change in mass was observed, and the VG appeared to be thermally inert up to temperatures of at least at least 1100°C. As stated earlier, VG is a cementitious mixture of exfoliated vermiculite in a gypsum binder. When exposed to heat, the gypsum binder releases chemically combined water in the form of water vapor. It is likely that the mass loss experienced by the VG TGA samples can be attributed to dehydration of the gypsum binder in combination with loss of free water from pores within the material. The fact that the mass loss was observed over a range of temperatures up to 500°C, and not in a smaller range close to 100°C as would be expected for moisture evaporation, can be attributed to the fact that it took time for the 50 mm cube specimens to heat completely to the centre (because of the extremely low thermal conductivity of VG).

As a result, by the time the centre of the cubes was experiencing

dehydration reactions, the furnace temperature had increased to 500°C. It is also plausible that thermally induced moisture migration (which tends to move water away from warmer regions of a porous material) caused a retardation of the dehydration reaction near the surface of the VG specimens.

121



In an attempt to test these hypotheses, a thin sample of VG, 50 mm × 50 mm × 10 mm, was subjected to TGA at the same rate of heating as the cubic VG specimens. Data from this test have been included in Figure 4.10. As expected, this sample experienced a similar total mass loss, but the loss occurred over a smaller range in temperature, providing some credibility to the above stated hypotheses. The VG TGA data yielded two main observations of significance for the full-scale fire endurance tests: first, VG experiences a dehydration reaction in the range of 100°C to 200°C which consumes thermal energy and maintains the temperature of the insulation within this range; and second, aside from the dehydration reaction, VG appears to be thermally inert up to temperatures in excess of 1000°C. Both of these factors play pivotal roles in the ability of the VG to protect a substrate material from fire. EI Paint Before examining the results of TGA on EI paint, readers who are unfamiliar with intumescent materials will benefit from reviewing section 5.2.6 which provides a short discussion of intumescent materials and the reactions that occur on heating. TGA data for EI paint is provided in Figure 4.11. While the thermophysical behaviour of the EI material is extremely complex, good repeatability is observed in the TGA data.

The test data indicate that the

activation temperature for EI is approximately 235°C, where the onset of rapid mass-loss coinciding with the release of the intumescent blowing agent was observed. Figure 4.12 shows a sequence of photos which document the expansion of EI paint at various temperatures.

From

this series of photos, it is evident that a change in color of the paint, from white, to light yellow, to brown, is observed between 225°C and 250°C.

This corresponds with the onset of

intumescence. Rapid expansion and charring does not begin until about 325°C however, after which point expansion is rapid to a maximum thickness of at least 50 times the original thickness at about 525°C. Beyond 525°C the EI maintains its shape well until about 600°C, before contracting into a thin residual char.

122



While the TGA results discussed in the above paragraph are instructive in several respects, the tests were conducted in an oven with a heating rate (and thermal flux) substantially lower than would be expected in a standard fire test. It is extremely difficult to predict exactly how a particular intumescent coating will react in a specific fire test situation based solely on these data.

4.3 Column Tests 4.3.1 General The following sections present various data obtained during the 2 full-scale fire endurance tests on FRP-wrapped and insulated circular reinforced concrete columns. The tests consisted of two distinct phases: the preload phase and the fire test phase.

Qualitative

descriptions and visual observations from the fire endurance tests, including an approximate timeline of significant events, are provided first to give the reader a mental framework with which to approach the test data. The quantitative fire test data are then presented and discussed with respect to observed temperatures, strains, load carrying capacity, and overall axial elongation of the columns. The columns tested have been designated as ISIS-1, which was tested in fire test #2, and ISIS-2, which was tested in fire test #1.

4.3.2 Preload Phase Preload refers to the phase of testing wherein the columns were placed in the test furnace and the applied load was gradually increased up to the full service load. The magnitude of the test load was determined using the procedures outlined in Section 3.2.4 and Appendix E. During this phase, no heat was applied to the columns, and once the target load was reached it was maintained at the full value for the 5-hour duration of the test.

Load, temperature, axial

deflection, and longitudinal strain in the reinforcement were all monitored during the preload phase, which began approximately one hour before the start of the fire tests.

123



Load was applied to the columns at a rate of approximately 52 kN/minute up to a total sustained load of 2515 kN. Figure 4.13 gives load versus stroke plots for both columns during the preload phase. Aside from an initial non-linearity – likely the result of seating effects – both columns displayed essentially linear load-deflection behaviour, as should be expected at loads less than 50% of ultimate. Both columns behaved in a similar manner, with slightly larger deflections observed for column ISIS-2. Strains measured in the vertical reinforcing bars at mid-height were monitored throughout the preload phase. The results of strain monitoring during the preload phase are presented graphically in Figures 4.14 and 4.15, where total axial load has been plotted as a function of axial strain in the reinforcement. Also included in these figures are lines giving the axial strain averaged from the four gauges in each column, as well as the axial strain in the reinforcing bars as predicted by the iterative confinement model used in the computer program QCFIRE (presented in Chapter 5). There is generally very good agreement between the theoretical (QCFIRE) load-strain behaviour and the average strain observed during preload for the columns, aside from the initial non-linearities in the experimental data. Column ISIS-2 showed considerably more variability between the individual strain readings than did ISIS-1.

This is likely an indication of an

eccentricity of the column loading for ISIS-2, resulting in an unintentional curvature in the column at mid-height. However, if the column curvature is approximately calculated based on the maximum difference between observed strains on opposite sides of the column and the known distance between the reinforcing bars, the column would have a radius of curvature of about 800 m (at the maximum applied load of 2515 kN). A curvature of this magnitude is not likely to be significant at loads less than 50% of ultimate, and the average strain value agrees well with both numerical predictions and test data from column ISIS-1 in any case.

124



4.3.3 Fire Endurance Tests Data and observations from both full-scale fire endurance tests are included in this section. Both qualitative and quantitative observations are presented. In each subsection, column ISIS-2 is discussed first, since it was the first tested. 4.3.3.1

Timeline and Observations Both columns were exposed to fire for more than 5 hours without failing. During the fire

exposure, periodic visual observations were made through small view-ports in the furnace walls. Tables 4.4 and 4.5 outline timelines and observations for each of the two tests respectively. Where possible, photographs of the columns were taken through small view-ports in the furnace chamber walls. Several of these photographs have been included in this chapter to assist the reader with visualization. The most significant visual observations from both tests relate to the performance of the EI coating within the first 20 to 30 minutes of the tests, and to cracking of the VG insulation later in the fire exposure. For both columns, the EI intumescent coating activated within the first 3 to 4 minutes of the test, which is as expected given the activation temperature of 235°C determined from TGA (Section 4.2.4). Expansion of the EI coating was complete within 10 minutes of fire exposure, and the expanded EI char began to debond from the columns’ surface in both tests within 15 minutes of exposure. Given the relatively short-lived effectiveness of the EI coating, it is not clear whether it provided significant fire protection for the columns. The reader should remain cognizant of the fact that the EI coating was not originally developed for application on a porous substrate such as the VG insulation used here, so the relatively poor performance of the EI during fire is not necessarily indicative of a poor fire performance of intumescent coatings in general, which are more commonly used for fire protection of steel structures. In addition, the EI material was trowel-applied to the surface of the columns tested herein, whereas the manufacturer recommends it be applied by roller or industrial

125



paint sprayer. Further tests are required to investigate the effect of the EI paint on the fire performance of the insulating system and should include a test with VG insulation only. Once the EI material had fallen off of the columns, the VG insulation was fully exposed to fire. The VG insulation performed very well under fire exposure, and remained intact until late in the fire tests when the load was increased dramatically and explosive concrete spalling caused it to debond from the diamond lath, exposing the SCH FRP to fire. The only change that was observed in the appearance of the VG insulation during fire was the formation of cracks, generally less than 5 mm wide, which gradually appeared and widened as the test progressed. The location of the cracks appeared to be associated with the thickness of the VG and the location of installation joints in the material. For example, the initial VG cracks formed at the locations of the ring spacers, where VG had been in-filled by trowel after spray application the material. Later in the fire tests, larger vertical cracks formed in the VG and appeared to coincide with the location of the steel channels. This could potentially have been caused by differential thermal expansion between the metal channels and the VG, since VG is known to experience mild shrinkage on heating as a consequence of dehydration. In addition, cracks were larger and formed earlier for column ISIS-1, which had a smaller overall thickness of VG. Indeed, at later stages of the fire test #2 (column ISIS-1), flame was observed emanating from cracks in the VG (Figure 4.24). This flaming was thought to be associated with burning of the S-Epoxy matrix beneath the VG, a theory which is consistent with the observed temperatures at the VG-FRP interface and the known (from TGA) ignition temperature for S-Epoxy. The formation of cracks in the VG insulation is of utmost importance to the fire performance of insulated FRP-wrapped columns, since burning of the FRP, which is accompanied by flaming and the release of toxic gases, should be avoided at all cost in interior exposures. Cracking of the VG also creates regions of locally increased temperature in the FRP wrap and the substrate concrete, which are likely to damage the mechanical properties of the wrap and potentially cause concrete spalling. Further tests will be required to determine installation

126



procedures and VG thicknesses that will minimize and/or prevent cracking. As a final comment in this regard, cracking of the VG, and the subsequent regions of local increased temperature that result, are not accounted for in the numerical models of Chapter 5. As will become evident later in this thesis, VG cracks are hypothesized to be a primary factor in the relatively poor agreement between numerical and experimental results for column ISIS-1. 4.3.3.2

Temperatures As described in Chapter 3, temperatures were monitored at various locations within both

columns for the duration of the fire tests. Temperature data are presented and discussed in detail in this section. Throughout the presentation of temperature data, it is important to remain cognizant of the fact that temperatures were monitored at specific locations within the column. Because of the variability inherent in the constituent materials (particularly with respect to cracking of the VG), localized variations in temperature may or may not have been accurately captured by the thermocouple instrumentation, although the temperature data appear to be consistent with visual observations in most cases. Temperature data from test #1 on column ISIS2 are presented first (Figure 4.24), followed by data for test #2 on column ISIS-1 (Figure 4.25). Fire Test #1: Column ISIS-2 Figure 4.24 provides time versus temperature plots for various locations within column ISIS-2. Figure 4.24a shows: the actual furnace temperature (calculated as the average of all 8 furnace thermocouples), the furnace temperature as specified by ULC-S101, and the temperatures at the EI/VG interface (TCs 18 and 19), the VG-FRP interface (TCs 1, 9, 16, and 17), and the FRP/concrete interface (TCs 2 and 8). Some of the more important information gleaned from the data is discussed below. Furnace Temperatures A comparison of the Furnace Ave. and ULC-S101 curves in Figure 4.24a demonstrates that the furnace temperature was well controlled by the automatic burner control system, and that the ULC-S101 time-temperature curve was accurately reproduced by the column test furnace.

127



EI/VG Interface Temperatures: Effect of EI Coating The effect of the intumescent EI coating is clearly evident in the test data by examining the shape of the EI/VG Interface time-temperature curves within the first 40 minutes of fire exposure (Figure 4.26). The temperature at the EI/VG interface increased rapidly within the first 3 to 4 minutes of fire exposure, at which time the intumescent reaction started to occur, consuming heat in the expansion process and protecting the underlying VG from the heat of the fire as evidenced by a reduced slope in the time-temperature curves. For TC18 the reduced slope lasted until about 14 minutes of exposure, at which point the EI fell off of the column, exposing the VG to the full effects of the fire. The time-temperature history then “caught up” with the fire temperature curve over a period of about 5 minutes. For TC19, the reduced slope intumescent phase of the time-temperature curve lasted slightly longer than for TC18 (about 16 minutes total). This delay is thought to be due to the EI coating falling off of the column in sections, with the result that the insulating EI char remained intact over TC19 longer than over TC18. All of the above conclusions are consistent, in terms of both temperatures and times, with TGA on the EI coating and with visual observations taken during the fire tests. It is not clear from the test data whether the time-temperature curve for the EI/VG interface beyond 25 minutes of fire exposure was significantly affected by the presence of the EI coating. Hence, it is difficult to state conclusively whether EI is a necessary component of the fire protection system used herein. Additional tests on columns insulated with VG only would be required to investigate this point. VG/FRP Interface Temperatures: Effect of VG Insulation While the effect of the EI coating appears to be relatively short lived under exposure to fire, the VG insulation remained essentially intact for the full duration of the test, and provided substantial protection for both the FRP wrap and the substrate concrete. Examining Figure 4.24a, the temperatures at the VG/FRP interface increased quite rapidly during the first 60 minutes of

128



fire exposure to a temperature between 95ºC and 100ºC.

However, the temperature then

remained constant until at least 190 minutes of exposure, after which time it began to rise rapidly again until the end of the test. All of the VG/FRP Interface thermocouples did not remain at the 95-100ºC plateau for the same length of time. Rather, one thermocouple experienced an increase to temperatures above 100ºC significantly earlier than the other three. It is possible that the earlier increase in temperature for the lone thermocouple could be the result of a crack in the VG close to the sensor, resulting in a localized increase in temperature. The 95ºC-100ºC plateau observed in the VG/FRP temperature data occurs as a result of the unique thermal properties of the VG material and its tendency to release both free and chemically-combined water at temperatures close to 100ºC. In combination with the very low thermal conductivity of the VG, the moisture loss results in a coating that is extremely effective at maintaining low temperatures in the underlying material.

In fact, aside from the lone

thermocouple that experienced a jump in temperature at 190 minutes, the FRP wrap temperature was maintained below 100ºC for about 240 minutes (4 hours).

The reader will note that

maintenance of the FRP temperature below 100ºC for 4 hours would likely be sufficient experimental evidence to meet even the most stringent fire design guidelines for a 4-hour fire rating. Even after 5 hours of fire exposure, the temperature at the outside edge of the wrap was low enough to prevent burning of the S-Epoxy matrix. FRP/Concrete Interface Temperatures: Heat Transfer through the Wrap The thermal insulating ability of the FRP wrap can be approximately ascertained by examining the temperatures at the FRP/concrete interface. Although the SCH FRP material has a low thermal conductivity in the radial (transverse) direction, it was also very thin in comparison with the VG or the underlying concrete. Hence, only a small drop in temperature should be expected across the FRP wrap.

Indeed, the data presented in Figure 4.24a show that the

FRP/concrete interface temperature closely follows, with a slight lag, the VG/FRP interface temperature.

The magnitude of the temperature differential across the FRP is observed to

129



increase late in the fire endurance test, and could be due to changing thermal properties of the wrap at elevated temperature or to the fact that the thermocouples on each side of the wrap were not necessarily adjacent to each other (so that the effects of VG cracking and local thermal spikes could have influenced the data). A particularly interesting observation gleaned from the FRP/concrete interface data is that the temperature at the FRP/concrete bond-line exceeded the GTT of the S-Epoxy, quoted as 93ºC by the manufacturer, within the first 60 minutes of fire exposure. Hence, the bond between the SCH FRP wrap and the substrate concrete may have been severely degraded after less than 1 hour of fire exposure. Concrete Temperatures Figure 4.24b shows the temperatures within the concrete at depths of 50 mm (TC3), 100 mm (TCs 4 and 7), 150 mm (TC5) and 200 mm (TC6, which was at the centerline of the column). As expected, given the outstanding insulation provided by the VG fire protection, the temperatures in the concrete remained very low (below 100ºC) for the full duration of the test. Given the information on the variation in strength and stiffness of concrete with temperature, presented in Chapter 2, which suggested that the strength of concrete is virtually unaffected by such moderately increased temperatures, we can conclude that the concrete remained at fullstrength for the duration of the test. The reader will note that this is the full unconfined strength of the concrete, since we have already stated that the effectiveness of the wrap was likely lost (or at least severely degraded) within the first hour of fire exposure. The observed temperature increases in the concrete were largest closer to the outside of the column, with temperatures at greater depths lagging behind, although with similar trends. The 2 thermocouples at a depth of 100 mm, installed on opposite sides of the column, exhibit a slight and unexpected difference in temperature (of no more than about 7ºC). This is likely due in part to dislocation of the thermocouples during the concrete pouring operations.

130



Reinforcement Temperatures Figure 4.24c shows the time-temperature history of the reinforcing bars within column ISIS-2 during fire test #1. Presented in the plot are data for opposite sides of the column on the top of the spiral steel (TCs 10 and 15), the outside surface of the longitudinal reinforcing bars (TCs 11 and 14), and the inside face of the longitudinal reinforcing bars (TCs 12 and 13). As was the case for temperatures measured in the concrete, the temperatures in both the spiral and longitudinal reinforcement remained low (below 120ºC) for the duration of the test. Although this data suggests that the steel reinforcement heats up faster than the surrounding concrete, it is important to recall that steel has a comparatively high thermal conductivity. Hence, the steel spiral, which has a clear concrete cover of 40 mm and is connected to the vertical steel bars, draws heat into the reinforcing steel and raises its overall temperature.

Referring to the

information on reinforcing steel presented in Chapter 2, such modest increases in temperature can be assumed to have had a negligible effect on the strength of the reinforcement, and it can thus be assumed that the reinforcement maintained its room temperature strength for the full 5 hours of fire exposure. It is also interesting to note that the “spiral top” temperature is generally greater than the “outside rebar” temperature, which is generally greater than the “inside rebar” temperature, as should be expected. Fire Test #2: Column ISIS-1 Figure 4.25 shows time versus temperature plots for column ISIS-1 during fire test #2. Figure 4.25a shows the furnace temperature, along with the temperatures at the EI/VG interface, the VG/FRP interface, and the FRP/concrete interface. Furnace Temperatures Comparison of the Furnace Ave. and ULC-S101 curves in Figure 4.25a shows that the furnace temperature was well controlled by the automatic control system and that the ULC-S101 time-temperature curve was accurately reproduced by the column test furnace.

131



During this test, a small thermal spike was observed in the average furnace temperature curve at about 13 minutes into the fire exposure. The temperature spike appeared to correspond with the end of the intumescent phase of the EI coating, and is thought to have been caused by free-burning of the EI during the intumescent reaction. The EI, which exhibited substantial flaming while intumescing, contributed energy to the fire and increased the temperature in the column furnace too rapidly for the automatic control systems to sufficiently decrease the rate of heat gain. Indeed, it was observed during the free-burning of the EI that the furnace control system had effectively closed off the supply of propane to the burners. The behaviour of the EI in this regard is worrisome from a fire engineering standpoint in that it suggests that, during a building fire, activation of the EI coating could contribute fuel to the fire, and actually increase the fire’s severity. EI/VG Interface Temperatures: Effect of EI Coating The effect of the intumescent reaction of the EI coating is clearly evident in the test data by examining the shape of the EI/VG interface time-temperature curves within the first 40 minutes of fire exposure (Figure 4.27). The behaviour for column ISIS-1 is similar to that observed for ISIS-2, although there are two aspects of the plot in Figure 4.27 that are worthy of comment: the shorter length of the intumescent phase for ISIS-1, and the fact that both EI/VG interface TCs experienced an intumescent phase of the same length. Comparison of Figure 4.27 with Figure 4.26, which has been plotted with identical scaling, shows that the intumescent phase for column ISIS-1 was shorter than for column ISIS-2 (about 8 minutes long as compared with 10 to 17 minutes). This result is somewhat counterintuitive given that ISIS-1 had about twice the amount of EI paint applied to its surface. A potential explanation is revealed if we consider that the EI coating, once expanded and charred, was observed to suddenly and completely fall off of column ISIS-1 under its own weight at about 13 minutes of fire exposure, whereas column ISIS-2 lost its EI coating gradually over a period of about 10 minutes. The sudden and complete loss of the EI coating for ISIS-1 occurred earlier and

132



more completely than for column ISIS-2 because the thicker layer of EI resulted in excessive expansion, causing the intumescent char to fall off suddenly under its own weight.

This

hypothesis also provides an explanation for the identical length of the intumescent phase observed for TCs 18 and 19 in test #2. Hence, it appears that using a greater thickness of EI actually results in a poorer fire performance for the intumescent char, since it appears to swell excessively and fall off suddenly and completely under its own weight. If EI is to be used in future fire tests, it is necessary to determine the optimal thickness of raw material to prolong its fire performance. As was the case for column ISIS-2, it is not clear whether the time-temperature curve for the EI/VG interface beyond 25 minutes of exposure was significantly affected by the presence of the EI coating. One aspect of the EI/VG interface temperatures recorded during test #2 that is difficult to explain, is the apparent decrease in temperature observed at that location beyond about 230 minutes of fire exposure (refer to Figure 4.25a). It is unlikely that such a large decease in temperature actually occurred as the instrumentation suggests. It is felt that the decreasing temperature readings were caused by sensor errors due to thermocouple damage at high temperature. VG/FRP Interface Temperatures: Effect of VG Insulation Although the thickness of the VG insulation for column ISIS-1 was considerably less than for ISIS-2, the VG insulation remained intact for the full duration of the test. Major cracks were observed in the VG as early as 90 minutes, although it provided significant protection for both the SCH FRP wrap and the substrate concrete until late in the fire test. In Figure 4.25a, it is evident that the temperatures at the VG/FRP interface increased quickly during the first 15 minutes of fire exposure to a temperature between 95ºC and 100ºC (the equivalent increase took about 60 minutes for ISIS-2). As for column ISIS-2, the temperature at this level then remained essentially constant, although only until about 55 minutes in this case, at which point it began to rise rapidly, and continued to rise until the end of the test. Again, not all of the thermocouples

133



remained at the 95-100ºC plateau for the same length of time. Rather, some began to increase significantly before the others, likely due to the localized effects of cracking in the VG. For ISIS-1, the FRP surface reached a temperature of 750ºC after 5 hours of fire exposure. The data show that the surface temperature of the FRP was as high as 450ºC after about 150 to 180 minutes, which is consistent with the observation that flames were emanating from the VG cracks at 183 minutes. Recall that, during TGA tests on S-Epoxy (Section 4.2.4), ignition of the samples was observed at a temperature of approximately 450ºC. FRP/Concrete Interface Temperatures: Heat Transfer through the Wrap As was the case for ISIS-2, the FRP/concrete interface temperature for ISIS-1 closely followed that of the VG/FRP interface. Again, the magnitude of the temperature differential across the FRP increased later in the fire exposure. The temperature at the FRP/concrete interface exceeded the glass transition temperature of the S-Epoxy within the first 15 minutes of fire. Concrete Temperatures Figure 4.25b shows the temperatures within the concrete at depths of 50 mm, 100 mm, and 200 mm. Although less effective than for ISIS-2, the VG insulation provided outstanding fire protection for column ISIS-1. Again, the temperatures in the concrete remained very low (below 200ºC in this case) for the duration of the test. Thus, assuming that there were not any local temperature spikes in the concrete that were not captured by the thermocouples, it is safe to assume that the concrete retained much of its original strength for the duration of the test. As expected, the temperature increases that were observed in the concrete were most severe close to the outside of the column, with temperatures at greater depths lagging behind but with similar trends. For test #2, the thermocouples at a depth of 100 mm again exhibited a slight difference in temperature, although the difference remained less than 10ºC. Reinforcement Temperatures Figure 4.25c shows the temperatures of the reinforcing bars within column ISIS-1. Again, presented in the plot are data for the top of the spiral steel, the outside surface of the

134



longitudinal reinforcing bars, and the inside face of the longitudinal reinforcing bars. As was the case for the temperatures observed in the concrete, the temperatures in both the spiral and longitudinal reinforcement remained low (below 220ºC) for the duration of the test. Again, such modest increases in temperature can be assumed to have had a negligible effect on the strength of the reinforcement. Again, the spiral-top temperature was generally greater than the outside-rebar temperature, which was generally greater than the inside-rebar temperature. 4.3.3.3

Axial Deformation and Reinforcement Strains Axial deformation and reinforcement strains measured during the preload phase were

discussed in section 4.3.2. Once the sustained load had been reached in the preload phase, the load was maintained at a constant value of 2515 kN for the full 5 hour duration of the fire test. After 5 hours of exposure to fire, with both columns showing no signs of impending failure, the load was increased gradually until failure occurred. Because the columns were loaded only to their full service load for the tests, the strains in the FRP wrap theoretically remained very low (less than 10% of ultimate) and hence the confining effect at service load levels was negligible. Thus, when (or if) the confining effect of the wrap was lost, as a result of increased temperature at some point in time during fire exposure, there was no noticeable change in overall axial deflection. Also, because the temperature of both the concrete and the reinforcing steel remained low during both tests, very little overall thermal expansion was observed for either column tested. Axial Elongation Figure 4.28 shows axial elongation versus time and total applied load versus time curves for columns ISIS-2 and ISIS-1 during their respective fire endurance tests. The reader should note the small magnitude of the observed axial elongation for both columns. Column ISIS-2, in which the concrete temperatures remained less than 100ºC for the entire duration of the test, displayed a small (0.1 mm) contraction during the first 15 minutes of fire exposure. This contraction could have been due either to seating effects within the load frame, or to short-term

135



creep of the column itself. The deflection is extremely small in any case. After the initial contraction, the column very slowly expanded, due to slight thermal expansion of the concrete upon heating, until the load was increased at the 5-hour mark, resulting in contraction of the column at an increasing rate up until failure. Column ISIS-1 behaved in a similar manner, although it did not display the small initial contraction experienced by ISIS-2. The magnitude of the expansion during exposure to fire was larger for ISIS-1, as should be expected given the higher internal temperatures it experienced. The overall magnitude of the column’s expansion was small in comparison with the overall length of the column, with a maximum elongation of approximately 0.02% of the column length. Strains in the Longitudinal Reinforcement The strains in the longitudinal reinforcing steel were monitored throughout the test. During the preload phase, all four strain gauges in both columns were shown to yield data that agreed well with theoretical predictions. However, during the fire test phases the strain gauges, which based on their specified upper use temperatures, should have performed throughout the tests, behaved in an erratic and erroneous fashion. reinforcement strains measured during test #1.

Figure 4.29 shows data for all four

It is evident that the measured strains are

nonsensical, since the column experienced virtually no axial elongation for the entire duration of the test, and hence the reinforcement strains should have remained essentially constant. Similar behaviour was observed for the strain gauges installed in column ISIS-1 during fire test #2. The strain data was hence deemed to be useless beyond the preload phase for both columns. It is thought that the erratic performance of the strain gauges during the fire tests was due to the operation of the high voltage ignition and control systems employed by the column furnace. This could have resulted in electrical interference with the strain signals, although it was not possible to pinpoint the source of the problem. It is unlikely that any conventional foil strain gauges can be used in fire endurance tests.

136



4.3.3.4

Load Capacity The room temperature axial load capacity of columns ISIS-1 and ISIS-2 was assumed to

be the same, since the VG and EI were assumed to have negligible strength and were not likely to significantly increase the axial load capacity of the columns. The nominal room temperature strength of both columns was calculated to be 5094 kN based on the ISIS Canada (2001a) design guidelines without reduction factors (or 7054 kN using the ACI 440 Guidelines (ACI, 2002)). As stated earlier, examination of the temperature data obtained during the fire endurance tests leads to the conclusion that, after 5 hours of fire exposure, the wrap had likely been rendered ineffective due to increased temperatures in excess of its GTT but that the concrete and reinforcing steel should have retained virtually all of their room-temperature strength. Thus, when the load was gradually increased after 5 hours of exposure, we should have expected the columns to fail at, or only slightly below, the nominal load for an equivalent unwrapped column at room temperature. The nominal (unfactored) compressive strength of an equivalent unwrapped column was calculated to be 4149 kN using the CSA A23.3 (CSA, 1994) code or 4386 kN using ACI 318 (ACI, 1995) (refer to Appendix E). During the fire endurance tests, data was sampled at a rate of 0.0167 Hz (once per minute), so the failure load of the columns was not captured precisely. Examination of Figure 4.30, which shows load deflection data for both columns (including the preload and fire test phases), shows that columns ISIS-2 and ISIS-1 had failure loads of approximately 4680 kN and 4437 kN respectively. Figure 4.31 gives a graphical representation of the load capacities of the two columns with respect to their design and predicted values. In both cases, the tested strength of the columns was similar to the unfactored room temperature predicted strength, calculated according to either the Canadian or American concrete design guidelines. In both fire endurance tests, the columns actually failed at loads slightly higher than those predicted by either code, even after 5 hours of exposure to the ULC S101 standard fire.

137



In terms of the fire endurance of the tested columns, it can be stated with confidence that, for the full 5-plus hour duration of the tests, both columns were able to carry their unfactored FRP-confined service load. This would result in a 5-hour ULC-S101 fire endurance rating with design loads calculated in accordance with the ISIS Canada design guidelines. It is interesting to note also that the unfactored FRP-confined service load calculated according to the ACI 440.2R02 (ACI, 2002) design guidelines is 2942 kN for the columns tested herein, which is also considerably less than the observed failure load for either column after 5 hours of exposure to the standard fire.

4.3.4 Comparison Figure 4.32 provides a comparison of recorded temperatures at various locations in columns ISIS-1 and ISIS-2 for the purposes of comparison. Shown in the figure are average temperature histories at the EI/VG interface, the VG/FRP interface, the FRP/concrete interface, the outside-rebar location, and the column centreline. Keeping in mind that temperatures of less than 200ºC are not structurally significant in terms of deterioration of mechanical properties for either concrete or steel, and realizing that the confining effectiveness of the SCH FRP will be lost regardless of the amount of insulation applied to the exterior of the wrap (within reason), it is evident from Figure 5.32 that an increase in the thickness of VG insulation from 32 mm (for ISIS-1) to 57 mm (for ISIS-2) is not likely to provide a significant increase in the structural fire endurance of the columns. Although the latestage temperatures in the column are reduced by the presence of the thicker VG insulation for ISIS-2, fire endurances longer than 4 hours are rarely specified in design, and so the extra thickness of VG represents a somewhat unnecessary use of material. Indeed, it was observed that the column with the greater thickness of VG insulation failed at only a slightly higher load after 5.5 hours of fire exposure.

138



Thus, only small thicknesses of VG insulation are required for the structural fire protection of the FRP-wrapped concrete columns tested herein up to 4 hours of fire exposure, assuming that the wraps are effectively lost regardless of the type and amount of fire protection. However, it is possible that using a greater thickness of VG insulation would allow FRP wraps to remain at a low enough temperature during a severe fire to retain a significant portion of their original strength. Test data on the residual strength properties of FRP materials will be required to support this hypothesis.

4.4

Discussion and Summary The preceding sections presented results of ancillary tests on the various materials used in

the fabrication, strengthening and insulation of two full-scale reinforced concrete columns, as well as the results of full-scale fire endurance tests on said columns. The ancillary tests demonstrated that the mechanical behaviour of the concrete, the reinforcing steel, and the carbon FRP used herein was as expected. Thermogravimetric analysis of the SCH FRP, S-Epoxy, EI, and VG provided information on the thermal behaviour of the various materials, with the most important observations being: 90% of the mass of the S-Epoxy samples was lost in the temperature range of 390°C to 510°C. Auto-ignition of the S-Epoxy occurred at approximately 450°C. Hence it is deemed important to maintain the temperature of S-Epoxy below 450°C in all fire resistant applications to avoid increasing fuel load for potential fires. SCH carbon fibres are relatively thermally inert (in terms of mass loss) up to 750°C. Mass loss at higher temperatures occurs without flame or smoke generation. Samples of VG insulation experienced a mass loss of approximately 15% near 100°C. This mass loss is due to release of free and chemically combined water, primarily from the gypsum binder. Otherwise, VG appears to be thermally inert up to 1100°C.

139



The activation temperature for EI is about 235°C, at which point it expands to form a thick multi-cellular char. However, this char disintegrates into a thin residual layer at temperatures above 600°C. Experimental results from the full-scale fire endurance tests revealed the following important information: During the preload phase, strain gauge data from gauges installed on the reinforcing steel at mid-height were found to agree well with theoretical predictions.

Strains

measured in column ISIS-2 suggested that it was experiencing an inadvertent curvature, although the magnitude of the curvature was small and not deemed to be significant. Visual observations collected during the fire tests were helpful in analyzing timetemperature data at a number of locations within the column and vise-versa. This was particularly true in terms of evaluating the performance and effectiveness of the EI intumescent paint. Temperatures measured at a number of different locations within the columns provided a wealth of important information. Thermal data were used to evaluate the performance of the various insulating materials and to shed light on the fire endurance of both the FRP wrapping scheme and the overall concrete members. Axial Deflection of the columns was monitored and found to be small, as expected, up until the columns failed under increased load after more than 5 hours of fire exposure. The columns behaved in a similar fashion to each other throughout the tests, with slightly greater thermal expansion observed for ISIS-1. Strain data collected during the fire tests were highly erratic and were not examined in detail.

140



The failure loads of the columns were similar, and indicated that both columns were able to maintain their full nominal unwrapped design strengths for the full duration of the fire endurance tests according to both Canadian and American design guidelines. The test program described in this chapter sought to examine the behaviour in fire of fullscale FRP-wrapped and insulated reinforced concrete columns. It was demonstrated through testing that it is possible to maintain the temperature of the FRP wrap below 100°C for up to 4 hours by providing the requisite thicknesses of EI and VG. In addition, both wrapped and protected columns achieved a 5-plus hour fire endurance rating based on the requirements of ULC-S101 (CAN/ULC, 1989), with service loads calculated according to the ISIS Canada design guidelines (ISIS, 2001a). The following chapters (5 and 6) discuss the development of a series of computer programs written to numerically determine temperature distributions and load carrying capacities for both FRP-wrapped reinforced concrete columns and FRP bar-reinforced concrete slabs during fire. The results of these models are compared to the experimental results presented in this chapter.

141



Table 4.1: Results of ancillary tests on concrete Test Series

Age (days)

Test Number

fc’ (MPa)

Ave. fc’ (MPa)

Std. Dev. (MPa)

Internal R.H. (%)

1 33.5 32.7 0.74 N/A 2 32.0 3 32.6 1 38.7 28-day 28 38.5 0.75 N/A 2 37.7 (carbonate) 3 39.1 1 39.9 ISIS-1 440 40.1 0.21 92.5 2 40.3 Test Day 3 40.0 1 38.4 ISIS-2 418 38.8 1.00 92.5 2 38.0 Test Day 3 39.9 ¥ Determined in accordance with Appendix A2 of ULC-S101 (CAN/ULC, 1989) 28-day (siliceous)

28

Moisture Content¥ (%) N/A

N/A

7.5

7.5

Table 4.2: Results of ancillary tests on reinforcing steel Specimen

11.3-1-A 11.3-1-B 11.3-1-C 11.3-2-A 11.3-2-B 11.3-2-C Average Std. Dev. 19.5-1-A 19.5-1-B 19.5-1-C 19.5-2-A 19.5-2-B 19.5-2-C Average Std. Dev.

Yield Stress (MPa)

Yield Strain (%)

Elastic Modulus (GPa)

Peak Stress (MPa)

Spiral Steel (11.3 mm diameter) 414 0.425 237 657 405 0.427 212 657 402 0.441 176 663 401 0.441 155 659 391 0.434 149 657 365 0.445 128 659 396 0.436 176 659 17 0.008 41 3 Longitudinal Reinforcing Bars (19.5 mm diameter) 454 0.214 211 594 462 0.235 222 599 458 0.223 201 594 454 0.212 220 583 455 0.212 230 585 453 0.224 211 585 456 0.220 216 590 3 0.009 10 6.3

142

Ultimate Strain (%)

17.8 17.1 19.2 17.9 15.4 18.9 17.7 1.4 18.8 20.3 19.1 20.5 19.5 19.5 19.6 0.7



Table 4.3: Results of ancillary tension tests on SCH FRP coupons Specimen Coupon #1 Coupon #2 Coupon #3 Coupon #4 Coupon #5 Average Standard Deviation Coeff. of Variation (%)

Ultimate Strength (MPa) 1640 1500 1540 1540 1320 1510 120 8

Strain at Failure (%) 1.61 1.66 1.76 1.86 1.33 1.64 0.20 12

Elastic Modulus (GPa) 95.4 82.9 86.6 88.5 97.6 90.2 6.1 7

Table 4.4: Visual observations for Fire Test #1: Column ISIS-2 Time (min) 0 3-4 5 11 13 17 17-300 300 330 334

Comment Test started at 10:01 AM Intumescent reaction begins; irregular expansion of coating accompanied by flaming Flames observed on surface of EI; intumescent reaction continues EI falling off in various places; EI surface extremely irregular EI almost completely fallen off; VG exposed to fire Few small cracks (< 5 mm width) observed in VG No significant new observations; slight and gradual (1 to 2 mm) widening of cracks in VG Applied load increased at a rate of approximately 87 kN/min Explosive sounds from inside furnace; spalling of concrete cover; VG blown off of column; failure of column; fire test halted Furnace doors opened; SCH burning; VG debonded from column over entire height; diamond lath exposed; widespread spalling of concrete cover

143

Figure 4.16 4.17 -4.18 4.19 ----4.20



Table 4.5: Visual observations for Fire Test #2: Column ISIS-1 Time (min) 0 3-4 5 12 28 28-90 90 183 231 237-300 300 330 336

Comment Test started at 09:47 AM Intumescent reaction begins; irregular expansion of coating accompanied by flaming Flames observed on surface of EI; intumescent reaction continues EI falls off suddenly and completely; VG exposed to fire Few small cracks (< 5 mm width) observed in VG at the ringspacer locations No significant new observations; slight and gradual (1 to 2 mm) widening of cracks in VG Large vertical cracks ( 5 mm width) observed in VG at location of steel channels Flame observed emanating from vertical cracks in VG; assumed to be burning of FRP matrix resin Flames emanating from all cracks in VG Hissing sound of moisture evaporation from top of column Applied load increased at a rate of approximately 72 kN/min Explosive sounds from inside furnace; spalling of concrete cover; some VG blown off of column; failure of column; fire test halted Furnace doors opened; SCH burning; VG debonded from column in some areas; diamond lath exposed in some areas; some spalling of concrete cover

144

Figure ------4.21 -4.22 ---4.23



700 Ave. Ult. Stress = 590 MPa

600

Stress (MPa)

500 400 300 200 20M Bar #1

100

Ave. Failure Strain = 19.6%

20M Bar #2

0 0

5

10

15

20

Strain (%)

Figure 4.1: Full experimental stress-strain curves for 19.5 mm diameter reinforcing bars 500

Yield Stress = 456 MPa 20M Bar #1

400

Stress (MPa)

20M Bar #2

300

200

200 000 MPa Yield Strain = 0.225% 1

100

0 0

0.1

0.2

0.3

0.4

0.5

Strain (%)

Figure 4.2: Elastic range stress-strain curves for 19.5 mm diameter reinforcing bars 800 Ave. Ult. Stress = 659 MPa

700

Stress (MPa)

600 500 400 300 200

Spiral 1

100

Ave. Failure Strain = 17.7%

Spiral 2

0 0

5

10

15

20

Strain (%)

Figure 4.3: Full experimental stress-strain curves for 11.3 mm diameter spiral steel

145



500

Stress (MPa)

400 413.7 405.1 401.8 400.9 391.1 365.3

300

396.3 MPa

200 0.2% Offset

100

Spiral 1 Spiral 2

0 0

0.1

0.2

0.3

0.4

0.5

0.6

Strain (%)

Figure 4.4: Elastic range stress-strain curves for 11.3 mm diameter spiral steel 2000 fult,ave = 1510 ± 117 MPa εult,ave = 1.64± 0.20 % Eave = 90200 MPa

Stress (MPa)

1500

1000

500 Test Data Manufacturer Data

0 0

0.5

1

1.5

2

Strain (%)

Figure 4.5: Stress-strain data for Tyfo®SCH-30T coupons tested in tension

Figure 4.6: Tyfo®SCH-30T tensile coupon immediately after failure (rotated 90°)

146



1.2

6

Mormalized Mass

1.0

5

0.8 4

0.6 0.4

3

Specific Heat for CFRP (Griffis et al., 1984)

0.2

2

0.0 0

100

200

300

400

Specific Heat (kJ/kg.K)

Tyfo S Epoxy TGA Data

1 600

500

o

Temperature ( C)

Figure 4.7: TGA data for S-Epoxy

Normalized Mass

1.0 0.8 0.6 0.4 0.2 0.0

Tyfo SCH-30T Fibres Tyfo SEH-51A Fibres

0

200

400

600

800

1000

o

Temperature ( C)

Figure 4.8: TGA data for SCH and SEH fibres

Normalized Mass

1.0

TGA Data Dimitrienko (1999)

0.8

Matrix Mass Loss

0.6 0.4 Fibre Mass Loss

0.2 0.0

0

200

400

600

800

1000

o

Temperature ( C)

Figure 4.9: Variation in mass from TGA data for SCH FRP system and for a carbon/epoxy FRP tested by Dimitrienko (1999)

147



Normalized Mass

1.0 0.8 Dehydration

0.6 0.4 0.2 0.0

50 mm Cubes 50 × 50 × 10 mm Block

0

200

400

600

800

1000

Temperature (oC) Figure 4.10: TGA data for VG insulation

Activation Temp. = 235oC

Normalized Mass

1.0 0.8 0.6

Inert

0.4

Expansion and Charring

0.2 0.0

0

200

400

Contraction

600

Residue

800

1000

Temperature (oC) Figure 4.11: TGA data for EI paint

148



150°C

175°C

200°C

225°C

250°C

275°C

300°C

325°C

350°C

375°C

400°C

425°C

450°C

475°C

500°C

550°C

575°C

600°C

625°C

675°C

700°C

725°C

750°C

525°C

650°C

Figure 4.12: Photo-documentation of EI expansion and contraction with temperature

149



3000 Column ISIS-2 Column ISIS-1

Axial Load (kN)

2500 2000 1500 1000 500 0

} 0

Initial seating effects

1

2

3

4

5

6

Axial Stroke (mm)

Figure 4.13: Load versus stroke data during the preload phase for columns ISIS-1 and ISIS-2

Axial Load (kN)

2500 2000 1500 Strain 1 Strain 2 Strain 3 Strain 4 Ave. Strain Model

1000 500 0

}


0

200

400

600

Strain (µε)

Figure 4.14: Load versus axial strain on the reinforcing steel during the preload phase for column ISIS-1

Axial Load (kN)

2500 2000 1500 Strain 1 Strain 2 Strain 3 Strain 4 Ave. Strain Model

1000 500 0

} 0


200

400

600

Strain (µε)

Figure 4.15: Load versus axial strain on the reinforcing steel during the preload phase for column ISIS-2

150



Figure 4.16: ISIS-2 in the test furnace just before commencing fire test #1

Figure 4.17: View of column ISIS-2 through a view-port during the intumescent reaction

Figure 4.18: Column ISIS-2: EI in the process of falling off

Figure 4.19: Column ISIS-2: VG completely exposed to fire

151



Figure 4.20: Column ISIS-2: Immediately after failure

Figure 4.21: Column ISIS-1: Close-up of vertical crack formation in the VG

Figure 4.22: Column ISIS-1: Close-up of flaming at a vertical crack in the VG

Figure 4.23: Column ISIS-1: Immediately after failure

152



(a)

1200 TC18

Temperature (oC)

1000 800

Furnace Ave. (FT1-FT8) EI/VG Interface (TCs 18 & 19) VG/FRP Interface (TCs 1, 9, 16, & 17) FRP/Concrete Interface (TCs 2 & 8) ULC-S101

600 400 TC19

200 0

0

60

120

180

240

300

240

300

240

300

Time (min) 100 50 mm Depth (TC3) 100 mm Depth (TCs 4 & 7) 150 mm Depth (TC5) Centreline (TC6)

80

o

Temperature ( C)

(b)

60 40 20 0

0

60

120

180

Time (min)

(c)

120 Spiral Top (TCs 10 & 15) Rebar Outside (TCs 11 & 14) Rebar Inside (TCs 12 & 13)

o

Temperature ( C)

100 80 60 40 20 0

0

60

120

180

Time (min)

Figure 4.24: Time versus temperature plots for column ISIS-2 (a) Furnace, EI, VG, and FRP temperatures (b) Concrete temperatures (c) Reinforcing steel temperatures

153



1200

(a) Temperature (oC)

1000 800

Furnace Ave. (FT 1-FT8) EI/VG Interface (TCs 18 & 19) VG/FRP (TCs 1, 9, 16, & 17) FRP/Concrete Interface (TC 2 & 8) ULC-S101

600 400 200 0

0

60

120

180

240

300

240

300

240

300

Time (min) 200 50 mm Depth (TC3) 100 mm Depth (TCs 4 & 7) 150 mm Depth (TC5) Centreline (TC6)

150

o

Temperature ( C)

(b)

100

50

0

0

60

120

180

Time (min) 200 Spiral Top (TCs 10 & 15) Rebar Outside (TCs 11 & 14) Rebar Inside (TCs 12 & 13)

150

o

Temperature ( C)

(c)

100

50

0

0

60

120

180

Time (min)

Figure 4.25: Time versus temperature plots for column ISIS-1 (a) Furnace, EI, VG, and FRP temperatures (b) Concrete temperatures (c) Reinforcing steel temperatures

154



800 TC18: Intumescent Phase

TC18: Post-Intumescent Phase

Pre-Intumescent Phase

600

o

Temperature ( C)

700

500 400 300 200

TC18 TC19

Loss of EI above TC 19

100 0

Loss of EI above TC18

0

5

10

15

20

25

30

35

40

Time (min)

Figure 4.26: Column ISIS-2: Initial EI/VG interface time-temperature curves 800 Intumescent Phase

600

Post-Intumescent Phase

Pre-Intumescent Phase

Temperature (oC)

700

500 400 300 200 100 0

TC18 TC19

Loss of EI

0

5

10

15

20

25

30

35

40

Time (min)

Figure 4.27: Column ISIS-1: Initial EI/VG interface time-temperature curves 4000 ISIS-2 Axial Elongation ISIS-1 Axial Elongation ISIS-2 Applied Load ISIS-1 Applied Load

1.0

3600

0.5

3200

0.0

2800

-0.5

0

50

100

150

200

250

300

Load (kN)

Axial Elongation (mm)

1.5

2400

Time (min)

Figure 4.28: Axial elongation and applied load during both fire endurance tests

155



4000 Strain 1 Strain 2 Strain 3 Strain 4

Strain (µε)

2000 0 -2000 -4000 0

100

200

300

Time (min)

Figure 4.29: Strains measured on the longitudinal reinforcing steel during fire test #1

Total Applied Load (kN)

5000 4000

* *

ISIS-2 ISIS-1

Failing Phase

3000 2000 1000

Load Constant During Fire Test: 2515 kN

Preload Phase

0 -15

-10

-5

0


8000 6000 4000 2000

ISIS-1 Test

ISIS-2 Test

ACI 318 Pred.

CSA Pred.

ACI 318 Design

CSA Design

ACI 440 Pred.

ISIS Pred.

ACI 440 Design

0 ISIS Design

Axial Load Capacity (kN)

Figure 4.30: Complete load versus axial elongation plots for both columns

Figure 4.31: Design, predicted, and tested strength of columns ISIS-1 and ISIS-2

156



(a) o

Temperature ( C)

1000 800 SENSOR ERROR

600 ISIS-2: EI/VG Interface ISIS-2: VG/FRP Interface ISIS-1: EI/VG Interface ISIS-1: VG/FRP Interface

400 200 0

0

60

120

180

240

300

240

300

Time (min)

200 ISIS-2: FRP/Concrete Interface ISIS-2: Rebar-Outside ISIS-2: Centreline ISIS-1: FRP/Concrete Interface ISIS-1: Rebar-Outside ISIS-1: Centreline

150

o

Temperature ( C)

(b)

100

50

0

0

60

120

180

Time (min)

Figure 4.32: Comparison of temperature histories at various locations within columns ISIS-1 and ISIS-2 during fire endurance tests (a) Temperatures at the EI/VG and VG/FRP interfaces (b) Temperatures at the FRP/concrete interface, rebar-outside, and centerline locations

157


CHAPTER 5: Numerical Modelling 1 – Columns

CHAPTER 5 NUMERICAL MODELLING 1 – COLUMNS 5.1 General This and the following chapter present and discuss a series of numerical models developed to simulate fire endurance tests on FRP-wrapped reinforced concrete columns and FRP bar-reinforced concrete slabs. Because full-scale fire endurance tests are time-consuming and expensive, the numerical models were developed such that, once validated by relatively few laboratory tests, they could be used to conduct parametric studies and investigate the effects of varying a number of variables on the fire endurance behaviour of FRP-reinforced concrete members. The models described herein have been incorporated into computer programs which have been designated as QCFIRE (Queen’s Column Fire Analysis Software) and QSFIRE (Queen’s Slab Fire Analysis Software). This chapter focuses on models developed for columns, and Chapter 6 presents details of the numerical model for FRP bar-reinforced concrete slabs. The column models were developed using procedures and format similar to those used previously, with great success, by researchers at the National Research Council of Canada (Kodur and Baingo, 1998; Lie, 1992; Lie and Celikkol, 1991; Lie and Irwin, 1993; Lie and Stringer, 1994). The approach consists of a coupled heat-transfer and load-capacity analysis programmed for computer in DIGITALTM Visual Fortran. The initial portions of both the column and slab analyses use an explicit finite-difference formulation based on an elemental energy balance to calculate the temperatures inside concrete members when subjected to a standard timetemperature curve (ASTM, 2001; CAN/ULC, 1989).

The second portion of the analysis

calculates the load capacity of the member in question during fire, based on strain compatibility and equilibrium of forces, and on the known distribution of internal temperatures determined from the heat-transfer analysis.

158



While various models have been developed previously to describe the behaviour in fire of a number of different concrete member types, none were readily available to the author. It is important to recognize that the numerical models presented in this thesis were developed entirely independently by the author for both the QCFIRE and QSFIRE analyses, although a similar methodology (for much of the analysis) has been used by other researchers in the past. In the QCFIRE analysis, the completely original aspects of the numerical model are that the thermal properties and heat transfer equations for the FRP wrap and insulation have not been previously presented in the literature. In addition, the model described herein is the first to account for the confining effect of an FRP wrap in calculating the load capacity (by both buckling and crushing) and elongation of a reinforced concrete column at elevated temperature. In the QSFIRE analysis, the model developed herein is original in that it accounts for the evaporation of moisture in the concrete at temperatures close to 100°C (not accounted for by previous models). In addition, the flexural capacity portion of the QSFIRE analysis (described in detail in Chapter 6) represents the first attempt to model the structural behaviour of FRP barreinforced concrete slabs at elevated temperature. This chapter presents numerical procedures for FRP-confined reinforced concrete column fire models, and discusses their validation based on test data available in the literature and on that obtained during the experimental program discussed in Chapters 3 and 4 of this thesis. The results of parametric studies for both the column and slab models are also presented in their respective chapters, as are discussions examining the consequences of fire for the design of FRPwrapped or reinforced concrete members.

5.2 Column Modelling As stated previously, the fire endurance of concrete columns has traditionally been defined in terms of their load carrying capacity since columns do not generally perform fireseparation or fire-barrier functions.

When an FRP-wrapped reinforced concrete column is

159



exposed to fire, the strength will decrease as its internal temperatures and the mechanical properties of its constituent materials deteriorate. The fire endurance is reached at the point in time after the onset of fire exposure that the load carrying capacity of the column falls below its full service load. Thus, any model of the fire behaviour of FRP-wrapped reinforced concrete columns must first determine the distribution of temperatures within the column during fire exposure, and subsequently use this distribution of temperatures, in conjunction with the known thermomechanical properties of the constituent materials involved, to estimate the load carrying capacity of the member. A variety of numerical models to predict the fire behaviour of different types of structural members have been developed during the last 35 years at the National Research Council of Canada (Kodur and Baingo, 1998; Lie, 1992; Lie and Celikkol, 1991; Lie and Irwin, 1993; Lie and Stringer, 1994). Models are available at NRC for the fire behaviour of reinforced concrete columns, concrete-filled steel hollow structural sections, and reinforced concrete slabs, and all have been found to agree reasonably well with results from fire endurance tests conducted at NRC and elsewhere (Lie, 1992). The overall rationale of the NRC numerical procedures has been used herein as a basis for the development of the numerical models. To assist the reader with comprehension during the discussions below, a complete schematic of the program logic for the QCFIRE model is shown in Figures 5.1a to 5.1g.

5.2.1 Heat Transfer Model In this section, the derivation of the finite difference equations for an FRP-wrapped reinforced concrete column is presented. A similar methodology can be used to develop heat transfer equations for FRP-wrapped and insulated reinforced concrete columns, unwrapped concrete columns, and FRP bar-reinforced concrete slabs. The equations in this section were derived using basic heat transfer principles based on a conservation of energy approach, and the

160



final heat transfer equations are similar to those used previously by Lie et al. (1992) for fire modelling of concrete-filled steel hollow structural sections. Heat transfer in a solid can be described by the heat diffusion differential equation, which, for a differential volume in Cartesian coordinates, can be expressed as (Incropera and DeWitt, 2002):

x

k

T x

y

k

T y

z

k

T z

q&

ρC p

T t

[5.1]

where k is the thermal conductivity of the solid, T is the temperature, t is time, ρ is the material’s density, Cp is the specific heat, and q& is a term representing heat generation within the material. The physical significance of the various terms in Equation 5.1 demonstrates that it is simply a mathematical description of the principle of conservation of energy. Each of the first three terms represents heat transfer into or out of the differential volume due to heat conduction, and the term on the right hand side represents the heat (energy) stored in the differential volume per unit time. The above differential equation can be used, with appropriate simplifications and boundary conditions, to arrive at the heat transfer equations used in the current analysis. Alternatively, the finite difference equations for heat transfer in a solid can be derived on the basis of an energy balance. In the discussions below, an energy balance for a 1 m long section of column is used to arrive at finite difference equations for an FRP-wrapped concrete column. For heat transfer purposes, the column is assumed to be infinitely long. While the assumption of infinite length is not strictly correct, and there are likely some end-effects due to the true finite size of the column, this assumption is reasonable and has been used with success in the past (Lie, 1994; Kodur and Lie, 1997; Lie et al., 1984, 1992). It is also assumed that the contribution of the internal reinforcing steel to heat transfer is minimal, and can be neglected, due to the relatively small volume of steel as compared with concrete. This assumption has also been used with success in the past, and simplifies the heat transfer analysis dramatically.

161



Fire Temperature For a 1 m long section of an FRP-wrapped reinforced concrete column, exposed to fire around its entire circumference, the temperature of the surroundings can be determined in accordance with a standard fire curve as described by ULC-S101 (CAN/ULC, 1989). This timetemperature curve is shown in Figure 3.31 and can be approximated mathematically as:

T fi

(

20 750 1 e

3.79553 i∆t

)

170.41 i∆t

[5.2]

where ∆t is the time step used in the analysis (in hours), and Tfi is the fire temperature (in C) at time step i. The column cross-section is divided into a series of circular layers, and for each layer an energy balance is developed. Figure 5.2 shows the discretization of an FRP-wrapped reinforced concrete column with concrete outer radius Rc, and wrap outer radius Rw. The wrap and concrete are divided into M1 and M2-M1+1 layers respectively. Also introduced in Figure 5.2 are distance increments, ∆xw in the wrap and ∆xc in the concrete. FRP Surface Element For an element at the surface of the column, the change in energy stored in the element, rad

∆Qst, must be equal to the difference between the heat into the element due to radiation, Qin , cond

and the heat out of the element due to conduction, Qout , thus:

∆Qst

cond Qinrad Qout

[5.3]

In the above equation, convection is ignored as a mechanism for heat transfer from the surroundings to the column. Previous work in this area (Lie, 1992) has led to the conclusion that convection is responsible for less than 10% of the heat transfer at the surface of the column in standard fire endurance tests and can thus be neglected for the purposes of numerical fire modelling. The heat stored in an elemental layer of the column can be described using (Incropera and DeWitt, 2002):

162



∆Qst

T dxdydz t

ρC p

[5.4]

If the volume of the layer is taken as Vlay, and the time interval is taken as ∆t, then for some change in temperature, ∆T, Equation 5.4 can be expressed approximately as :

ρ wC wVlay

∆Qst

∆T ∆t

[5.5]

where ρ w C w is the heat capacity of the wrap material (the product of specific heat and density). For a column discretized as shown in Figure 5.2, Equation 5.5 can be rewritten as:

ρ wC w 2

∆Qst

∆x w ∆xw T1i T1i 4 2 ∆t

Rw

1

[5.6]

where T1i is the element temperature at the current time step and T1i 1 is the element temperature at the previous time step. The heat transferred into a surface element with surface area As, due to radiation can be approximated after Dusinberre (1962) as:

(

Asσε surf Tsurr

Qinrad

273

) (T 4

273

surf

)

4

[5.7]

where σ is the Stephan-Boltzman constant, εsurf is the emmissivity of the column surface, Tsurr is the temperature of the surroundings, and Tsurf is the surface temperature of the object being heated. For the circular elemental layer considered here, Equation 5.7 becomes:

(

2 Rwσε w T fi

Qinrad

1

273

) (T 4

i 1 1

where εw is the approximate emmissivity of the FRP wrap, T fi the previous time step, and T1i

1

273 1

)

4

[5.8]

is the temperature of the fire at

is the surface temperature of the wrap at the previous time step.

For a differential volume element, the heat conduction equation in one of the three orthogonal directions (take the x direction for the purposes of illustration) can be described using:

163



Qxcond

k

T dydz x

[5.9]

Thus, because the present heat transfer problem is one-dimensional due to radial symmetry, the heat out of an elemental layer of material per unit time, with thermal conductivity k, surface area Asurf, and thickness L, by conduction can be approximated after Cengel (1998) as: cond Qout

kAsurf

∆T L

[5.10]

which, in the case of a circular elemental layer with variable thermal conductivity, gives: cond Qout

where T2i

1

k1i

1

k 2i 1 2

2

Rw

∆x w T1i 1 T2i ∆x w 2

1

[5.11]

is the temperature of the element adjacent to the surface element at the previous time

step, and k1i

1

and k 2i 1 are the respective thermal conductivities of the surface and adjacent

elements at their previous time step temperatures. Substituting Equations 5.6, 5.8, and 5.11 into Equation 5.3, and rearranging to isolate the temperature at the current time step, yields the finite-difference heat transfer equation for the element at the FRP-fire interface:

T1i

T1i

1

ρ wC w

2 Rw ∆t σε w T fi ∆x w ∆xw Rw 4

(

∆x w ∆t 2 k1i ∆xw Rw ∆x w2 4

1

) (T

273

4

i 1 1

4

[5.12]

Rw

ρ wC w

)

273

1

k 2i 1 T1i

1

T2i

1

The temperature at the FRP-fire interface is thus expressed in terms of thermal properties and temperatures at the previous time step at elements directly adjacent to the interface, which for a given instant in time are all known.

164



Internal FRP Element For any elemental layer, denoted layer m, inside the wrap, only conductive heat transfer occurs. Using a similar approach as used above for the FRP-fire interface element, the following equation for the temperature, Tmi, of element m at time step i can be derived:

Tmi

Tmi 1

2 ρ w C w Rw Rw

∆t (m 1)∆xw ∆xw2

Rw

(

)(

(

1 ∆x w k mi 1 k mi 11 Tmi 1 Tmi 11 m 2

)(

3 ∆x w k mi 11 k mi 1 Tmi 11 Tmi 1 2

m

)

) [5.13]

FRP/Concrete Interface Element At the FRP/concrete interface only conduction is present. However, at this location there are two distinct materials involved – FRP and concrete – and so a modified version of Equation 5.13 must be derived. A further complication arises from the fact that moisture evaporates from concrete at temperatures close to 100°C and consumes heat as discussed in Chapter 2, subsequently retarding the rise of temperatures within the column. The effect of moisture is taken into account in the present analysis by assuming that the moisture in any layer of the concrete starts to evaporate when the temperature in that layer reaches 100 C. All heat supplied to the layer is assumed to be consumed in the evaporation reactions, so that the temperature in the element is assumed to remain constant until all of the moisture evaporates. The initial volume of water, VMH12O , in the element at the FRP/concrete interface is equal to the volume of concrete times the initial volumetric moisture content, φi, and can be expressed as:

VMH12O

VM 1 φ i

2

Rc

∆xc ∆xc φi 4 2

[5.14]

If the temperature remains constant at 100ºC until all of the moisture has evaporated, then all of the heat supplied to the layer must be used for evaporation of moisture. This can be expressed as:

165



∆Qin

m H 2O λ H 2 O

ρ H O ∆VH O λ H O

∆t

∆t

2

2

[5.15]

2

where ∆Qin is the heat supplied to the layer during time step ∆t, mH 2O is the mass of water evaporated, and λH 2O is the heat of vaporization of water. For an element at the wrap/concrete interface, heat both enters and exits the element due to conduction, and Equation 5.15 can be combined with versions of Equation 5.10 that have been modified to represent conduction into or out of the surface element and rearranged to give the volume of moisture evaporated from the element, ∆VMH12O , during a time interval, ∆t, as:

∆VMH12O

∆t

Rc

∆x w 2 i 1 k M1 1 ∆x w

k Mi 11 TMi 11 1 TMi 11

Rc

∆xc 2 i 1 k M1 ∆x c

i 1 M1 1

ρ H O λH O 2

2

[5.16]

k

T

i 1 M1

T

i 1 M1 1

This treatment of water leads to the finite difference equation for heat transfer at the wrap/concrete interface, where a term has been added to account for the heat capacity of unevaporated moisture in the concrete:

TMi 1

TMi 11

ρ w C w Rw

(M 1

5 4)∆xw ∆xw

∆t ρ c Cc

Rc

∆xw 2 i 1 k M1 1 k Mi 11 TMi 11 1 TMi 11 ∆xw

Rc

∆xc 2 i 1 k M1 k Mi 11 1 TMi 11 TMi 11 1 ∆xc

ρ H O C H Oφ Mi 1 Rc 2

2

1

(1 4)∆xc ∆xc

[5.17] In the above expression, ρ c Cc is the heat capacity of the concrete, ρ H 2O C H 2O is the heat capacity of water, and φ Mi 11 is the volumetric moisture content of the FRP/concrete interface element at the previous time step, all of which are known for a given point in time.

166



Internal Concrete Element For any layer inside the concrete the effect of moisture is treated as above and the following two equations can be developed for the initial volume of water in an element and the amount of water evaporated during a time interval, ∆t:

VmH 2O ∆VmH 2O

∆t

ρ H O λH O 2

(m

2 Rc

M 1 )∆xc ∆xc φi

[5.18]

Rc

(m

M 1 1 2 )∆xc k mi 11 k mi 1 Tmi 11 Tmi 1

Rc

(m

M 1 1 2 )∆xc k mi 1 k mi 11 Tmi 1 Tmi 11

2

[5.19]

The heat transfer equation obtained is similar to that derived for heat transfer inside the wrap, with the addition of a term to account for moisture:

Tmi

Tmi 1

2 ρ c Cc Rc Rc

ρ H OC

(m (m

2

∆t φ Rc

i 1 H 2O m

(m

M 1 )∆xc ∆xc

2

M 1 1 2)∆xc k mi 11 k mi 1 Tmi 11 Tmi 1

[5.20]

M 1 1 2)∆xc k mi 1 k mi 11 Tmi 1 Tmi 11

Central Concrete Element At the centre of the concrete, symmetry dictates that conduction into the element is the only heat transfer that occurs. Using the same approach as above, equations can be developed for the initial volume of water in the element: 2

VMH22O

∆xc φi 4

[5.21]

and the volume of water evaporated during a time interval, ∆t:

∆VMH22O

∆t 2 ρ H 2O λ H 2O

k Mi 12 1 k Mi 12 TMi 21 1 TMi 21

[5.22]

The change in temperature of the centerline element during a time interval, ∆t, is:

TMi 2

TMi 21

ρ wCw

2∆t k Mi 1 k Mi 12 1 TMi 21 1 TMi 21 ρ H 2O C H 2Oφ mi 1 ∆xc 2 2

167

[5.23]



Repeated application of Equations 5.12 through 5.23 using a numerical algorithm programmed for computer enables the calculation of the temperatures at any point in the column at any time during fire exposure. Stability Criteria The finite-difference formulation presented above is explicit in nature, and so it must be ensured that the time step chosen for the analysis is sufficiently small to guarantee stability of the numerical solution. Using a criterion for stability of explicit finite difference formulations taken from Dusinberre (1962), the maximum allowable time step for stability at a particular elemental layer can be stated as:

∆t

melem C elem K

[5.24]

where melem is the mass of the element in question, Celem is the specific heat of the element, and K is a term representing the sum of the heat transfer coefficients into the element. Equation 5.24 leads to the following four criteria for stability of the current analysis, two of which have been stated previously by Lie et al. (1992). At the wrap/fire interface:

∆t

(ρ wC w )min ∆xw 2 2 (hrad )max ∆x w (k w )max

[5.25]

where (ρ w C w )min is the minimum value of the heat capacity of the wrap material to be expected during exposure to fire,

(hrad )max

is the maximum value of the equivalent heat transfer

coefficient due to radiant heating at the wrap/fire interface, and (k w )max is the maximum thermal conductivity to be expected in the wrap material at any point during exposure to fire. Inside the wrap, Equation 5.24 becomes:

∆t

(ρ wC w )min ∆x w 2 4(k w )max

[5.26]

At the wrap/concrete interface:

168



∆t

(ρ wC w )min 2

Rw Rw

(M 1 (M 1

5 4 )∆x w ∆x w

3 2 )∆x w

∆x w

(ρ c Cc )min Rc (M 1 1 4)∆xc (k w )max Rc (1 2)∆xc (k c )max

∆xc

[5.27]

∆xc

where (ρ c C c )min is the minimum value of the heat capacity of the concrete to be expected during exposure to fire, and (k c )max is the maximum thermal conductivity to be expected in the concrete during exposure. Inside the concrete:

∆t

(ρ c Cc )min ∆xc 2 4(k c )max

[5.28]

The smallest time step determined from the above three equations governs and is chosen as the time step used during application of the finite difference heat transfer algorithm. As stated earlier, models have also been developed by the author for unwrapped reinforced concrete columns and insulated and FRP-wrapped reinforced concrete columns. Derivation of the heat transfer equations for these additional member cases has not been included here, since all of the equations were derived using the same rationale as those presented above. Appendix A presents the additional heat transfer equations used in the other types of analyses for those readers who are interested.

5.2.2 Overall Load Capacity Model Once the distribution of temperatures throughout the column during exposure to fire is known, the axial load carrying capacity of the column during fire can be approximated. The procedure used herein is essentially a strain-equilibrium analysis that approximates the buckling strength of the column by discretizing the column cross-section into a series of elements with corresponding strain and temperature values. A similar approach has been used by previous authors to estimate the load carrying capacity of circular (Lie and Celikkol, 1991) and rectangular (Lie and Irwin, 1993) reinforced concrete columns, and concrete-filled steel hollow structural sections (Kodur and Lie, 1997; Lie et al., 1992) in fire. The load capacity of a column during

169



exposure to fire depends largely on the compressive stress-strain behaviour of the concrete in the cross-section, which in turn depends on the temperature of the concrete at that particular location and time, and on the confining pressure applied by the FRP wrap – which itself depends on the temperature of the wrap material and the dilation of the concrete. The details of the numerical procedure used to approximate axial load capacity of the columns are described in the following sections. The load capacity analysis relies on the following assumptions: 1. Plane sections before bending remain plane after bending. 2. Concrete has no strength in tension. 3. There is no slip between the internal reinforcing steel and the concrete. 4. The deterioration of mechanical properties for confined concrete with temperature can be treated in the same manner as the deterioration of mechanical properties for unconfined concrete. The column is divided into a series of annular elements as shown in Figure 5.3. Neither the FRP wrap nor the insulation (in cases where insulation is included in the heat transfer analysis) are included in the discretization of the column for the load capacity analysis. The FRP wrap is assumed to have fibres in the circumferential direction only, and its direct contribution to the axial strength of the column is assumed to be negligible. The effect of confinement provided by the FRP is included in the analysis as described later. Insulation is assumed to provide thermal protection only and has no strength in tension or compression. If it is assumed that the curvature of the column varies linearly from inflection points to midheight, as shown in Figure 5.4, then for any assumed curvature, χ , the mid-height deflection of the column, y, can be calculated using:

y

(KL)2

χ

12

[5.29]

where KL is the effective length of the column.

170



For each element in the column cross-section, the temperature, stress, and strain are assumed to be represented by those at the centroid of the element. The axial (vertical) strain that causes stress in any element of the column cross-section, which is subjected to both axial and flexural loads, is equal to the sum of the thermal strain, (ε T )e , bending strain, d e χ , and overall (average) axial strain, ε axial .

Thus, for any element on the left-hand side of the column

centreline, the axial (vertical) strain to cause stress is calculated from:

(ε e )l

(ε T )e ε axial

de χ

[5.30]

where d e is the distance from the element centroid to the column centerline (refer to Figure 5.3). The thermal strain is simply calculated as expansion,

c

c

(Tmi ,n

Tinit ) with the coefficient of thermal

, obtained from the thermomechanical subroutines at the current temperature. On

the right-hand side of the column centerline, the sign of the bending strain is reversed and is determined from:

(ε e )l

(ε T )e ε axial

de χ

[5.31]

Once the strain in each element is determined, the stress-strain characteristics at its particular temperature, which are determined using the confinement routine described below in combination with the concrete thermomechanical subroutines, can be used to determine the stress, and hence the elemental force due to each element can be obtained. The strain and temperature in any of the longitudinal reinforcing bars is assumed to be the same as the concrete element that the bar lies within. The stress in the reinforcement is then determined using thermomechanical subroutines for steel, and the force due to each reinforcing bar is calculated based on its cross-sectional area. Using the above procedure, the overall axial strain in the column is varied until the internal moment at mid-height, due to the contributions of each of the annular elements and reinforcing bars about the centreline of the column at mid-height, is equal to the external moment

171



at mid-height due to the applied loads. The internal moment at mid-height is calculated using (refer to Figures 5.2 and 5.3): i M int

2

M2

N

(σ ) (σ ) i m ,n l

m M1 n 1

i m ,n r

Am,n d m,n

N bars

σ qi Aq d q

[5.32]

q 1

i where σ m, n , is the stress in any single element of the cross section at time step I (the subscripts l

and r refer to the elements on the left and right-hand sides of the column centerline respectively),

Am,n is the area of the element, and d m ,n is the distance from the centroid of the element to the centerline of the column. σ qi , is the stress in any longitudinal reinforcing bar at time step i, Aq , is the area of the reinforcing bar, and d q is the distance from the centroid of the bar to the centerline of the column. The external moment is calculated as the product of the total vertical i

force at midheight, Pm , times the horizontal deflection of the column at that location: i M ext

Pmi y eo

[5.33]

i

where Pm is calculated as a summation of the force contributions from the individual elements. Thus:

Pmi

2

M2

N

m M1 n 1

(σ ) (σ ) i m ,n l

i m ,n r

Am,n

N bars

σ qi Aq

[5.34]

q 1

where y is the midheight deflection of the column, and eo is the assumed initial eccentricity of the compressive load (assumed as some small value, 2 mm in the present analysis). By repeating the above procedure for increasing curvatures – until the maximum axial load is reached – load versus mid-height deflection plots for the column can be obtained for a range of times during exposure to fire. From these plots the maximum load capacity of the column is determined, and a plot of load capacity versus fire exposure time can be obtained. The reader is encouraged to consult Figures 5.1a to 5.1f for improved comprehension.

172



5.2.3 Pure Axial Load Capacity In addition to the numerical model described above for the overall load capacity of the column, two sub-programs, called QCFIRE-Axial and QCFIRE-Deflection, were also developed. QCFIRE-Axial calculates the variation of the pure axial compressive strength of the cross-section with fire exposure, and QCFIRE-Deflection approximates the column’s axial deflection under fire exposure. QCFIRE-Axial is essentially a simplified version of the QCFIRE load capacity analysis that ignores flexural effects in calculating the strength of the column. The pure axial strength of the column is always greater than the overall (buckling) strength, and the effects of loss of the effectiveness of the FRP wrap are more pronounced. The QCFIRE-Axial analysis is applicable only to very short columns where buckling is highly unlikely and flexural effects can safely be ignored. The analysis is conducted in a manner similar to that described above for QCFIRE, with an identical heat transfer procedure. The difference lies in the fact that the load in the column is calculated for any overall axial strain simply by adding up the force contributions, Pi, from the various concrete elements and reinforcing bars using (refer to Figure 5.2):

P

i

M2

σ Am i m

m M1

N bars

σ qi Aq

[5.35]

q 1

The overall axial strain in the column is incrementally increased so long as the total axial load in the column is increasing. Once the maximum load is reached, the load value is written to the output file and the time step is incremented. In this manner, a series of axial compressive loadelongation plots for the column can be obtained for various fire exposure times, and these plots can then be used to generate a plot of axial load capacity versus fire exposure time. A schematic of the QCFIRE-Axial subprogram is shown in Figure 5.5.

173



5.2.4 Axial Deflection QCFIRE-Deflection is a subprogram that was developed to approximate the overall axial deflection of a fire-exposed concrete column (unwrapped, wrapped, or wrapped and insulated). Again, the heat transfer portion of the analysis is identical to that in QCFIRE. During the QCFIRE analysis, the load capacity of the column was determined by generating load-deflection plots for the column at various fire exposure times.

QCFIRE

Deflection uses the load versus axial strain data generated in the QCFIRE analysis to determine the overall axial strain value at a predefined user-specified applied load level. The overall axial strain value is then used in conjunction with the exposed length of the column to approximate its overall elongation. The reader will note that this procedure does not account for the effects of bending on axial deflection, although the contributions from bending are small (until just prior to failure) in comparison to those caused by the applied load or by thermal expansion. A similar approach has been used in the past for reinforced concrete columns (Lie and Celikkol, 1991). A schematic for the QCFIRE-Deflection subprogram is shown in Figure 5.6.

5.2.5 Confinement Modelling at High Temperature A unique aspect of the numerical models described herein is that they account for the beneficial effects of confinement on the strength of an FRP-wrapped concrete column exposed to fire. The confinement effect has been incorporated into the models using a modified version of the iterative confinement procedure developed by Spolestra and Monti (1999) and discussed more completely in Chapter 2 and Appendix C. The Spolestra and Monti model was chosen for use in the models because it is on of the few currently available model that can rationally calculate the confining pressure at any value of the axial compressive strain in the concrete (Bisby et al., 2003). As discussed further in Appendix C, most currently available confinement models provide only the ultimate stress and strain for confined concrete, and others suggest bilinear approximations of the complete stress-strain curve.

174



The use of the Spolestra and Monti confinement model in the analysis presented herein is complicated by the fact that the mechanical properties of the concrete in the column (required as inputs for the confinement model) are non-uniform over the column cross-section due to material deterioration as a consequence of increased temperature. Hence, an extension of the confinement model was required to account for the effect of fire on the stress-strain behaviour of both the concrete and the FRP wrap. With the overall axial strain in the concrete assumed, and the temperatures throughout the cross-section known, the maximum unconfined concrete stress and strain for each ring of the column, and the modulus and ultimate tensile strength of the wrap (based on its average temperature) are obtained from thermomechanical subroutines. These mechanical properties are used as inputs for the Spolestra and Monti model, and the confining pressure is determined. The procedure for calculation of the confinement pressure at some point in time is as follows (refer to Figure 5.1g): 1. The column is discretized into a series of ring elements (the same discretization as was used in the main program’s thermal analysis). See Figure 5.2. 2. The overall axial strain in the column, ε axial , is assumed (it is taken as the value that is passed to the confinement routine from the main program). 3. The confining pressure applied by the wrap, f lat , is assumed. i , at time step, i, is calculated using 4. The average temperature of the FRP wrap, Twrap

(refer to Figure 5.2): M1 i Twrap

m

i

T 1 m

M1

[5.36]

and the stress-strain characteristics of the wrap at the current temperature are obtained from thermomechanical subroutines.

175



'

5. For each ring of the column, the confined concrete strength, f cc , is obtained, using the Mander et al. (1988) equation as follows: f cc'

f co' 2.25 1 7.9

f lat f co'

2

f lat f co'

[5.37]

1.25

'

where f co is the compressive strength of the unconfined concrete at the current temperature of the element in question. 6. The confined compressive strength of the element is used, along with the concrete stress-strain curve of Popovics (1973), to determine the current element stress, f c , at the current strain, ε axial , as follows:

fc

f cc' x r where x r 1 xr and r

ε axial , ε cc ε cc

f cc' 1 f co'

ε co 1 5

Ec , Esec Ec Esec

f cc'

[5.38a,b,c]

[5.39a,b]

ε cc

In the above equations Ec is the initial tangent modulus for the unconfined concrete and has been taken as 5700

f co' (Spolestra and Monti, 1999).

7. The dilation or lateral strain, ε lat , for each element is determined using an equation developed by Spolestra and Monti (1999), which gives dilation as a function of axial strain and concrete stress:

ε lat

Ec ε axial f c where β 2 βf c

5700 f co'

500

[5.40]

8. The lateral strain in each ring due to thermal expansion, ε T , is determined using c

(Tmi ,n

Tinit ) assuming that all elements are free to expand laterally, and the total

176



lateral strain in each concrete ring is calculated by summing the dilation and thermal strains over the cross-section. 9. The overall lateral stain of the column, ε total , is approximated by averaging the lateral strain contributions from each elemental concrete ring over the cross-section: M2

ε total

(ε

lat

ε T )m

m M1

M2

M1

[5.41]

10. The total lateral strain is used to determine the strain in the FRP wrap, and hence to update the lateral confinement pressure, which is calculated using an equation that represents a modification to the Spolestra and Monti procedure suggested by Manfredi and Realfonzo (2001):

f lat

0.685

2t w E wε total dw

[5.42]

In the above equation, t w is the overall thickness of the FRP wrap, Ew is the modulus of the FRP wrap at the current average wrap temperature, and d w is the average diameter of the wrap. The 0.685 coefficient is inserted to account for the general over-prediction that is observed when the Spolestra and Monti model is compared against a large database of test data on FRP-confined concrete columns (Manfredi and Realfonzo, 2001). 11. The calculated confinement pressure is used as the new assumed value in step 3, and steps 3 to 10 are repeated until convergence of the confining pressure within a satisfactory tolerance is achieved. The confining pressure is passed back to the main program. The confinement model as implemented here essentially assumes that the wrap is unbonded, since it results in a constant confining pressure at all points in the column cross

177



section. Tests have indicated that for columns subjected to both axial and flexural loads, a bonded wrap will actually provide a higher level of confinement in the regions of the cross section subjected to compressive flexural strains (Monti and Spolestra, 2000). The assumption of a constant confining pressure is thus conservative.

5.2.6 Modelling Intumescent Coatings Intumescent coatings, while extremely useful for fire protection of structural members in many applications, are notoriously difficult to model or analyze. These materials respond to heat by swelling into an insulating char of thickness between 5 and 100 times that of the original material. They provide thermal protection to their underlying surface through absorption of heat during the endothermal reactions of the intumescent process, and through the insulating properties of the multicellular char formed during those reactions (Butler, 1997).

The basic

physical, thermal, and chemical reactions remain poorly understood, and so most intumescent materials have been developed using a trial-and-error approach and have not been extensively modelled (Butler et al., 1995). When subjected to a sufficiently high heat flux, the following sequence of events generally occurs for modern intumescent materials: an inorganic acid stored in the form of a salt is released; the acid dehydrates a carbon-rich polyhydric compound in preparation for the formation of the final char (a reaction catalyzed by an organic amine or amide); the intumescent mixture melts; an endothermal chemical reaction generates gases; the gases diffuse into small (10 to 60 micron diameter) bubbles, resulting in the formation of a foam; and the material solidifies through cross-linking into a thick multicellular char (Butler et al., 1995).

To assist with

visualization of the intumescent process, Shih et al. (1998) provide a detailed explanation of intumescence when a coating is subjected to an external heat flux (refer to Figure 5.7). The initial heat transfer mode is dominated by conduction through the virgin material. The temperature of the coating increases continuously under the influence of the heat flux, and as the temperature

178



reaches the activation temperature of the material, a thin intumescent zone is formed on the surface. The intumescent reaction may include thermochemical decomposition, thermochemical expansion, and solid-liquid, liquid-vapor, and solid-vapor phase changes. The temperature in the intumescent zone remains essentially constant.

Gases are given off in the intumescent zone to

blow and expand the softened but viscous solid into a foam, thus completing the intumescent reaction. Thus, there are three distinct (but constantly varying) regions that develop in sequence within an intumescent material under heating: virgin coating, intumescent zone, and final char. Each of these regions has its own thermal properties, none of which are well understood at present. The foaming process and its effects on the heat transfer process are extremely complex and could easily be the topic for a doctoral thesis in their own right. Furthermore, the EI system used to provide protection for the columns tested herein is proprietary, and very little is known about the coating’s chemical constituents and properties. Thus, no significant attempt has been made here to model the intumescent reaction. In the numerical analyses conducted herein, the beneficial effects of the intumescent coating are ignored. In Chapter 4 it was noted that, during fire tests, the intumescent coating activated and fell off of the columns within the first 20 minutes of exposure to fire. We shall see that relatively good overall agreement between the model and the test data of Chapter 4 is obtained by ignoring the short-lived presence of the EI coating entirely. In addition, the heat transfer models can alternatively be validated by forcing the temperature at the EI/VG interface to follow the experimentally observed behaviour, thus eliminating uncertainties associated with the intumescent layer. For additional information on modelling intumescent materials the reader is referred to: Anderson et al. (1988), Bhargava and Griffin (1999), Buckmaster et al. (1986), Butler (1997), Butler et al. (1995), Di Blasi and Branca (2001), and Shih et al. (1998).

179



5.2.7 Thermomechanical Subroutines To accurately model the effects of fire on structural members, a detailed knowledge of the thermal and physical properties of the constituent materials is required. In the numerical models presented in this and the following chapter, the variation in thermal and physical properties of the various constituent materials with temperature is included in the analysis by passing temperatures to thermomechanical subroutines. These subroutines pass thermal and mechanical information back to the main program. The thermomechanical subroutines for steel and concrete have been developed based on mathematical relationships (Lie, 1992) that are relatively well established for numerical fire modelling. For FRP, the density, thermal conductivity, and specific heat were modelled using data presented by Griffis et al. (1984), as shown in Figure 2.17, and the mechanical properties (strength and stiffness) were modelled using the semi-empirical relationships derived and summarized in Appendix D.

Only thermal property subroutines were required for the VG

insulation. These subroutines were derived based on a review of the relatively sparse research literature on vermiculite-gypsum fire protection materials, and on approximate data provided by the material’s manufacturer.

A summary of the equations used in the thermomechanical

subroutines is presented in Appendix F.

5.2.8 Fire Endurance Ratings In Canada, fire endurance ratings for various member types and structures are given in Part 3 of the National Building Code of Canada (NBCC) (NRC, 1995). This document outlines fire ratings, which are essentially required fire endurance times, based on a number of important factors. Some of the factors considered in assigning fire ratings to structural members are: use and occupancy, construction type (combustible versus non-combustible), and whether or not the

180



structure is outfitted with sprinklers.

Fire endurance guidelines in other countries may be

substantially different than those required in Canada. To determine the required fire rating for a particular structural member, the use and occupancy must first be classified according to one of the major occupancies outlined in the NBCC. These major occupancies are divided into groups and divisions. Within each group and division, there are a number of sub-categories that are related to the specifics of the structure in question, such as the provision of sprinklers, the number of storeys, and the potential fire load (the amount and nature of materials available for combustion). It is not relevant to outline all of the specific fire rating requirements here. The reader is referred to the NBCC for additional information. However, for slabs, fire ratings are typically required to be anywhere from 45 minutes (for a group A, division 1, sprinklered, limited area, single storey structure, for example) to 2 hours (for a group A, division 1, any height, any area, sprinklered structure). Columns are generally required to have fire ratings at least as high as the members which they support, and in some cases ratings as high as 4 hours are required. Indeed, in the wake of the September 11th, 2001 terrorist attacks on the World Trade Center in New York City, the Materials, Equipment and Acceptance (MEA) Division of the New York Department of Buildings requires 4 hour fire ratings for columns in structural upgrading applications. Thus, a wide range of fire rating requirements exist for reinforced concrete structures, although the likely ranges are about 1 to 2 hours for slabs and 2 to 4 hours for columns.

5.2.9 Validation 5.2.9.1 Pre-Test Validation When the QCFIRE program was initially developed, no test data were available on the fire behaviour of FRP-wrapped reinforced concrete members.

It was desired however, to

compare the predictions of the model with experimental results from fire endurance tests on reinforced concrete members such that the model could be used with relative confidence to assist

181



in pre-selecting the fire insulation schemes that would be used in the full-scale fire endurance tests. Few test results from fire endurance tests on circular reinforced concrete columns are available in the literature, and to the knowledge of the author, only one study on these members has ever been performed. Lie and Celikkol (1991) reported the results of full-scale fire endurance tests on 2 circular spirally-reinforced concrete columns at NRC. The columns tested were essentially identical to each other, and had dimensions and reinforcement details as shown in Figure 5.8. Temperatures QCFIRE was used to evaluate the fire behaviour of the Lie and Celikkol (1991) columns, and a comparison was made between the experimental results and the predictions of the model. Figure 5.9 shows experimental and predicted temperatures in the concrete at various depths as a function of fire exposure time. Significant variability is observed in the experimental data, likely due to uncertainty as to the exact location of the thermocouples in the concrete. At a depth of 25 mm, there is generally good agreement between the experimental and predicted temperatures, although the model tends to under-predict temperatures within the first hour of fire exposure. At depths of 64 and 178 mm, the experimental curves are characterized by rapid increases in temperature followed by regions of relatively constant temperature. This behaviour is more pronounced at greater depths, and has been attributed by previous authors (Kodur and Lie, 1997; Lie and Celikkol, 1991; Lie and Irwin, 1993; Lie et al., 1992) to thermally induced moisture migration in the concrete that is not accounted for in the model. While the model accounts for evaporation of moisture, it does not account for its migration toward the centre of the column (away from the heat source). The migration appears to account for the deviation between the predicted and observed temperatures in the early stages of fire. Hence, the numerical model tends to under-predict concrete temperatures early in the exposure, with closer agreement demonstrated at later stages in the fire exposure. It is important to recognize that the late-stage temperatures are

182



those which are more critical in estimating the fire endurance of the columns, and hence the initial discrepancy between measured and predicted temperatures is not a major concern.

Axial Deformation Figure 5.10 shows the measured and predicted axial deformation of the Lie and Celikkol (1991) columns as a function of fire exposure time using an applied axial load of 1431 kN (as was applied during the fire endurance tests).

Within the first two hours of fire exposure, the

calculated deformation is generally greater than that observed in tests.

However, both the

maximum deformation and the point of failure are predicted by the model with reasonable accuracy. The largest difference between predicted and measured axial deformation is in the order of 1.5 mm, which is small in comparison to the overall length of the columns. Considering the various unaccounted for factors that could contribute to axial deformation, the agreement is good. Load Capacity Finally, Figure 5.11 shows the predicted axial strength of the column as a function of fire exposure time. Also shown in Figure 5.11 is a horizontal line showing the sustained axial load that was applied to the columns during the fire endurance tests, and the corresponding observed and predicted fire endurances (the points where load capacity drops below applied load). The predicted and measured fire endurances for the two columns are similar, with the numerical model under-predicting the fire endurance by 11 minutes for column 1, and 35 minutes for column 2, both of which are conservative results. With the above discussion in mind, QCFIRE, and subprogram QCFIRE-Deflection, were assumed to satisfactorily predict the behaviour in fire of circular reinforced concrete columns, although the above comparisons provide no validation of the confinement effect of an FRP wrap during fire. 5.2.9.2 Post-Test Validation

183



Once the column tests described in this thesis were conducted, the resulting experimental data were used to investigate the validity of the numerical analysis accounting for both the FRP wrap and the supplemental (VG) insulation. Figures 5.12 and 5.13 show time versus temperature histories for various locations in the columns during the two fire endurance tests along with predictions made using QCFIRE. Test #1: Temperatures Figure 5.12a shows the experimental and predicted temperatures within the concrete during fire test #1. At a depth of 50 mm the agreement is generally good with a maximum prediction error of about 10°C, although the model appears to under-predict temperatures within the first 200 minutes of fire and over-predict temperatures thereafter. At depths greater than 50 mm, the model under-predicts the temperatures for the full fire-exposure. The under-prediction by the model is likely due to the effects of moisture migration within the concrete toward the centre of the column as discussed previously. It is important to note that the temperatures predicted within the concrete depend to a great extent on the model’s ability to accurately capture the heat transfer behaviour in the EI, VG, and FRP, since the temperatures in the concrete depend on the model predictions for the outermost layers of the column. Later in this section, an attempt is made to determine which portions of the analysis are most accurate using an approach wherein the temperatures at various locations are forced to follow the experimental data and the model is used to predict temperatures based on these forced curves. In this manner, each material in the heat-transfer analysis can be examined independently to determine which materials require better modelling. The overall ability of the complete heat transfer model to predict temperatures within the concrete is satisfactory, particularly given the small vertical scale of the plot (temperatures remain less than 100°C for the full fire exposure) and the fact that the temperatures remain sufficiently low to prevent degradation of material properties for the full fire exposure.

184



Figure 5.12b shows experimental and predicted temperatures in the reinforcement for test #1.

Again the agreement is generally good.

The predicted temperature drop across the

reinforcing bar (the difference between the Rebar-Outside and Rebar-Inside curves) is generally greater than that observed experimentally. This can be explained by recalling that the heat transfer model does not account for the presence of the reinforcing steel. Rather, it assumes the entire cross-section to be concrete. In reality, heat transfer across the rebar is accelerated by the comparatively high thermal conductivity of steel, and so the observed temperature drop is less than predicted. Again, the vertical scale of this plot is small and the temperatures in the steel reinforcement remain sufficiently low to prevent degradation of material properties for the full fire exposure. Figure 5.12c shows time-temperature curves for the EI/VG, VG/FRP, and FRP/concrete interfaces along with the ULC-S101 curve. As discussed in Chapter 4, the effect of the EI paint on the heat-transfer behaviour at the EI/VG interface is evident in the experimental data. The model tends to over-predict the EI/VG temperature, as should be expected, but the agreement becomes closer later in the fire exposure as the effect of the EI coating is gradually “forgotten”. The ability of the model to predict the temperature at the VG/FRP interface is an important factor in the success of the numerical model. While the model is capable of predicting the approximate temperature differential across the VG insulation, the model does not perform as well at capturing the 100°C plateau observed in the test data. The discrepancy between model predictions and test data could be due to a number of factors, the most likely of which is moisture migration in the insulation toward the surface of the FRP wrap. The SCH FRP is impervious to water, so any moisture in the insulation that migrates away from the fire will collect at the VG/FRP interface until it has completely evaporated.

This could explain both the under-

prediction of VG/FRP temperatures early in the fire exposure as well as the apparent inability of the model to capture the 100°C plateau as the fire progresses. In addition, the model does not predict the relatively rapid increases in temperature that are observed at the VG/FRP interface

185



after about 240 minutes of fire exposure. However, these sudden increases are thought to be due to cracking of the VG (which is not accounted for in the model). Clearly, a more detailed understanding of the thermohydromechanical properties of VG insulation would be beneficial in accurately modelling its behaviour in fire, particularly since thermal properties are a key factor in the overall predictive ability of the model. Figure 5.12c provides a comparison of the observed and predicted temperature differential across the SCH FRP sheet. The numerical model predicts that the differential should be very small at all times. Examination of the test data reveals that the temperature differential, while small for the majority of the fire exposure, increases to become comparatively large after 4 hours. As discussed in Chapter 4, this could be due to changing thermal properties of the wrap material at temperatures above 100°C, or to the fact that the VG/FRP and FRP/concrete interface thermocouples were not necessarily installed directly adjacent to each other. Test #2: Temperatures Figure 5.13a shows the experimental and predicted temperatures within the concrete during fire exposure for test #2. As for fire test #1, the agreement between the model and observations is best at a depth of 50 mm, with a maximum prediction error of about 30°C. Also as before, at depths greater than 50 mm, the model under-predicts the temperature for the full fireexposure, likely due to moisture migration with in the concrete. Again, the overall ability of the complete heat transfer model to predict temperatures within the concrete is satisfactory given the small vertical scale of the plot (in this case the concrete temperatures are maintained below 200°C). Figure 5.13b shows experimental and predicted temperatures in the reinforcement for test #2.

Again, the agreement is generally quite good, although in this case the model tends to

slightly under-predict spiral and rebar temperatures beyond 2 hours of fire exposure.

As

discussed in Chapter 4, cracks developed in the VG insulation within the first hour of test #2 and it is hence likely that localized thermal spikes near these cracks are responsible for the under-

186



prediction of temperatures by the model beyond 2 hours of exposure. Again, the predicted temperature drop across the reinforcing bars is greater than the observed temperature drop. Although higher internal temperatures were observed for test #2, the temperatures in the reinforcement remained sufficiently low to prevent degradation of material properties for the full fire exposure. Figure 5.13c shows time versus temperature curves for the EI/VG, VG/FRP, and FRP/concrete interfaces. The model tends to over-predict the EI/VG temperature early in the exposure, as should be expected given that it does not account for the effect of the EI coating. The agreement improves later in the fire exposure, although for this test, the model under-predicts the EI/VG interface temperature late in the tests.

This could be due to dislocation of the

thermocouples during the VG placement. Figure 5.13c further demonstrates that the model does not accurately capture the 100°C plateau observed in the test data, and tends to under-predict the temperature at the VG/FRP interface beyond 70 minutes of fire exposure and the FRP/concrete interface after 110 minutes. The under-prediction of these temperatures is almost certainly due to the VG cracks that formed within the first hour of fire exposure. It is evident that both the overall fire endurance of the column and the ability of the model to predict internal temperatures depend to a large extent on the prevention or minimization of crack formation in the VG insulation as the fire progresses. Since it was observed during tests that the largest VG cracks formed above the steel channel sections used to attach the diamond lath to the column, new installation approaches should be developed that promote the formation of fewer and smaller cracks. Again, further work is required, both analytical and experimental, to accurately model the behaviour in fire of the VG insulation and to determine installation procedures that discourage cracking. Figure 5.13c also provides data on the observed and predicted temperature differential across the FRP sheet. As for test #1, the test data reveal that the temperature differential increases

187



later in the fire exposure (in this case after 70 minutes), and strengthens the hypothesis that the thermal properties of the wrap material experience significant changes at temperatures above 100°C. It is interesting to note that the GTT for the S-Epoxy used in the tests was 93°C, so it is plausible that achieving the GTT causes changes in FRP’s thermal properties. To the knowledge of the author, no studies have been conducted that comment on thermal properties of polymers beyond their GTT. Forced Validation: Temperatures Forced validation refers to a process by which the ability of the model to predict temperatures in various materials was examined for each material in the column cross-section. Essentially, by forcing the temperature at a given location to follow the experimental time versus temperature behaviour, the performance of the model can be evaluated for specific materials. Figure 5.14 shows validation plots created for test #1 by forcing, (a, b) the temperature at the FRP/concrete interface, (c) the temperature at the VG/FRP interface, and (d) the temperature at the EI/VG interface. Similar plots for test #2 are shown in Figure 5.15. From Figures 5.14 and 5.15 many of the same observations result as made previously regarding the ability of the numerical model to predict temperatures in an FRP-wrapped and insulated reinforced concrete column. These can be summarized as follows: 1. There is generally good agreement between the experimental data and model predictions for the temperature in the concrete at a depth of 50 mm, and for the spiral and reinforcing steel temperatures. At greater concrete depths the model tends to under-predict temperatures, likely due to moisture migration in the concrete which is not accounted for in the model. 2. The predicted temperature drop across the reinforcing bars is greater than that observed in tests due to the fact that the heat transfer model does not account for the presence of the reinforcing steel, but rather assumes that the entire cross-section is concrete.

188



3. The numerical model predicts that the temperature differential across the FRP sheet should be small at all times. Examination of the test data reveals that the temperature differential becomes greater than that predicted beyond a wrap temperature of 100°C. This is most likely due to changing thermal properties of the wrap material at temperatures above the matrix GTT. 4. The model is capable of predicting the approximate temperature differential across the VG insulation, although it does not accurately capture the 100°C plateau observed in the test data and it tends to under-predict the temperatures at the VG/FRP interface within the first hour of fire exposure. Both of these observations can be at least partially explained by moisture migration in the insulation toward the surface of the FRP wrap. 5. The model does not predict the relatively rapid increases in temperature that were observed at the VG/FRP interface after 4 hours of fire exposure for test #1 and after 70 minutes for test #2. These increases are likely due to cracking of the VG which is not accounted for in the model. The results of the forced validation studies agree well with conclusions reached previously in the full validation discussion. The key parameters that require improved modelling are the effects of moisture migration in the concrete, the thermal properties and effects of moisture migration in the VG, and the effects of VG cracking at late-stage fire exposures. Axial Deformation Figure 5.16 shows the measured and predicted axial deformation of columns ISIS-1 and ISIS-2 as a function of fire exposure time, using an applied axial load of 2515 kN as was applied during the fire tests. The model tends to over-predict the expansion of both columns for the full length of fire exposure. However, the largest difference between predicted and measured axial deformation is in the order of 1.5 mm, which is small in comparison to the overall length of the column (although the total deflection predicted is only about 1 to 2 mm). In terms of the design

189



of structures for fire safety the model is conservative, since larger deflections in fire would result in the development of larger induced strains in a structure. It is suspected that the discrepancy between the predicted and observed column deformations can be attributed primarily to shortterm creep of the concrete which acts to counteract axial elongation due to mild thermal expansion. Given the small magnitude of the deflections, both predicted and observed, it is difficult to state conclusively whether the model is an adequate predictor of axial elongation for the FRP-wrapped and insulated case. Load Capacity Finally, Figure 5.17 shows the predicted axial load capacity of columns ISIS-1 and ISIS2 as a function of fire exposure time, using both the QCFIRE and QCFIRE-Axial numerical procedures. Also shown in Figure 5.17 are applied axial load capacity versus time curves for both fire endurance tests (2515 kN up to 5 hours of exposure, and then increasing until failure). As expected, the strength of the columns is predicted to decrease only slightly with fire exposure because of the relatively mild temperature increases experienced as a consequence of the outstanding protection provided by the VG insulation during both tests. Both columns are predicted to remain essentially at full strength for the entire duration of the fire, although it should be pointed out that the model does not account for the effects of bond degradation, which would likely reduce the confining effect of the wrap at temperatures close to the matrix GTT. The model predicts slightly superior fire behaviour for column ISIS-2, which was protected with the greater thickness of VG insulation. QCFIRE-Axial predicts similar behaviour, although the pure axial load capacity of the model is greater than the buckling strength calculated by QCFIRE, which can be attributed to the removal of flexural effects from the analysis. QCFIRE-Axial also predicts a slight increase in the strength of these columns under fire exposure, which can be attributed to the effects of differential thermal expansion between the wrap and the concrete, resulting in activation of the confining mechanism and higher confining pressures.

190



QCFIRE over-predicts the failure strength of the columns by 21% for ISIS-2 and 24% for ISIS-1. This result is unconservative and can be partially explained by recalling that the model tends to under-predict the temperatures in the concrete late in the fire exposure. In addition, localized regions of increased temperatures in the concrete, adjacent to cracks in the VG, could potentially have caused spalling of the concrete cover and contributed to failure. This hypothesis is supported by the failure mode of the columns, which was sudden and accompanied by explosive spalling. If QCFIRE is used to calculate the CAN/ULC-S101 (or ASTM E119) fire endurance of the wrapped and insulated columns under a sustained load of 2515 kN, then the fire endurance is predicted to be in excess of 7 hours. Another potential factor in the over-prediction of the load capacity by the QCFIRE analysis is the initial eccentricity used. In the analysis presented herein, the initial eccentricity was taken as 2 mm based on previous similar column fire analyses on conventionally reinforced concrete columns (Lie and Celikkol, 1991). However, Clause 10.15.3 of CSA A23.3 (CSA, 1994), in a discussion of slenderness effects in columns, suggests that members should be designed for a minimum initial eccentricity, eo, in millimeters, of:

eo

15 0.03h

[5.43]

where h is the diameter of the column. For the columns tested herein the above equation results in an initial eccentricity of 27 mm. Figure 5.17 shows the fire endurance curves predicted for columns ISIS-1 and ISIS-2 if the initial eccentricity in the QCFIRE analysis is taken as 27 mm. It is evident that a better agreement between the model and the tested failure loads is obtained for this case, and the model prediction very slightly conservative. Thus, it is likely that the true initial eccentricity of load for both columns was somewhere in the range of 2 mm to 27 mm, although it is not possible to state the eccentricity with certainty. Certainly, if the model is to be used in the future for specific design guidance, it would be wise to assume an eccentricity in accordance with Equation 5.43, which results in a more conservative prediction.

191



QCFIRE-Axial also over-predicts the strength of the columns after 5 hours of fire exposure, even if the QCFIRE Axial strength is multiplied by 0.85 (as suggested for concrete columns by CSA A23.3 to account for accidental flexural effects in axially loaded members). Because only 2 fire endurance tests were conducted, it is difficult to state conclusively whether the model is capable of predicting the axial strength of an FRP-wrapped and insulated reinforced concrete column during fire. However, both the test data and the numerical models demonstrate that even a small thickness of VG insulation (32 mm for ISIS-1) enabled the columns to retain at least 94% of their QCFIRE-calculated room temperature strength for a full 5 hours of fire exposure. With the above points in mind, the numerical models were assumed to satisfactorily predict overall trends in the fire behaviour of FRP-wrapped and insulated circular reinforced concrete columns, such that they could be used to conduct qualitative parametric studies. It is premature at this time to use the models as quantitative design tools for FRP-wrapped members.

5.3 Parametric Studies As is hopefully evident, the behaviour of FRP-wrapped and insulated columns in fire is extremely complex. The numerical models and experimental data demonstrate that it is possible to approximately predict the thermal and structural behaviour of both plain reinforced concrete and FRP-wrapped and reinforced concrete columns during fire using the QCFIRE numerical models. However, information on the thermal properties of both VG insulation and SCH FRP are extremely scarce, and the properties used in the analysis have been assumed based on approximate information available in the literature and from the material manufacturers. A more complete understanding of the materials involved (particularly with respect to the thermal properties of the VG insulation and the thermomechanical properties of the SCH FRP) is required before the numerical models can be used with confidence for design of specific FRP-wrapped members and insulation schemes.

192



In the following section, QCFIRE is used to conduct parametric studies to investigate the effects of varying a range of parameters, within realistic ranges, on the fire performance of FRPwrapped and insulated reinforced concrete columns.

The information obtained from the

parametric analysis is qualitative in nature, and the goal in the discussion is to provide guidance as to which factors are likely to be important in the design and testing of insulation schemes for FRP-wrapped columns. Three distinct failure criteria have been suggested for FRP-wrapped reinforced concrete columns in fire and are used in the discussion below. These are: Criterion 1: The average temperature of the FRP wrap shall not exceed the glasstransition temperature of the polymer matrix. This criterion is thought to protect against loss of the confinement effect. Criterion 2: The temperature at the outside face of the FRP wrap shall not exceed the ignition temperature of the polymer matrix, which prevents toxic gas and smoke generation. Criterion 3: The load-bearing capacity of the column shall not fall below the full design service load for the strengthened member. This is equivalent to the traditional ULC-S101 or ASTM E119 failure criterion for reinforced concrete columns. Only criterion 3 is applicable to current North-American structural fire design guidelines, although flame spread and smoke generation characteristics are also important from an environmental fire protection standpoint. In conducting the parametric studies presented herein, the following assumptions have been made unless otherwise stated: Insulation consists of Tyfo® VG, with thermal properties assumed as outlined in Appendix F. The VG insulation is assumed not to crack. The FRP matrix consists of Tyfo® S Epoxy, with a GTT of 93°C (as quoted by the manufacturer) and an ignition temperature of 450°C (as observed in TGA).

193



The degradation of mechanical properties for FRP at high temperature is assumed to be described by the semi-empirical relationships of Appendix D.

The thermal

properties of the FRP depend primarily on the matrix material, and so they are assumed to be those for SCH FRP as presented in Appendix F. For FRP-wrapped columns, service loads are determined by back-calculating from the ISIS Canada design strength (ISIS, 2001a) with a live-to-dead load ratio of 1:1. The concrete column is assumed to have the dimensions and internal reinforcement details of those fabricated during the experimental program of this thesis (refer to Figure 3.1). Table 5.1 presents a summary of the various column, wrap, and insulation configurations that were analyzed in conducting the parametric study.

5.3.1 Numerical Comparison of Member Types Before discussing parametric studies, it is instructive to examine the results obtained from the model for various types of reinforced concrete columns. Figure 5.18 shows structural fire endurance curves (load capacity versus time of fire exposure) calculated using QCFIRE for full-scale concrete columns equivalent to those described in Chapter 3 in the unwrapped, wrapped, and wrapped and insulated configurations. The strength of the unwrapped column is predicted to decrease steadily during exposure to fire as should be expected. The reduction in strength is such that the unwrapped column is predicted to fail under the 2515kN sustained axial load after approximately 5 hours of fire exposure. Adding a single layer of SCH FRP wrap is predicted to increase the initial strength of the column by about 5%. However, the increase in strength due to wrapping is essentially lost in less than 20 minutes of fire exposure (refer to Figure 5.19), and the strength of the column is subsequently reduced to slightly more than that of the unwrapped column. The slightly higher strength is due primarily to the residual strength of the FRP wrap. As exposure time increases,

194



the strength of the wrapped column remains slightly higher than that of the unwrapped column. Hence, the column with an FRP wrap displays a predicted fire endurance approximately 12 minutes longer than the unwrapped column. By insulating the column with VG insulation, the fire endurance of the column is increased substantially. The most significant observation that can be gleaned from Figures 5.18 and 5.19 is that, for the FRP-wrapped and insulated column the fire endurance of the overall concrete member is significantly improved as compared to the unwrapped column. The model predicts that the confining effect of the wrap is maintained for a significant period of time during fire exposure, although the reader should recall that the model does not account for the effects of bond degradation, which is likely severe beyond 1 hour of fire exposure. Thus, although the structural benefits of the FRP wrap may be difficult to maintain in fire, the overall fire behaviour of the member is vastly superior to that of the unwrapped member, and an increase in the fire endurance for the wrapped and insulated member certainly seems warranted.

5.3.2 Unprotected FRP-Wrapped Reinforced Concrete Columns The above discussion demonstrates that unprotected FRP wraps will be rendered ineffective within minutes of exposure to fire. Thus, during fire, unprotected wraps should be considered completely ineffective at providing confining reinforcement to concrete columns. In addition, there are significant flame spread and smoke generation concerns, not discussed in detail here, associated with the use of fire-exposed FRPs in buildings. Parametric studies on unprotected FRP-wrapped reinforced concrete members did not provide any useful information and are not included in this thesis.

5.3.3 FRP-wrapped and Insulated Reinforced Concrete Columns For protected (insulated) FRP-wrapped concrete columns, various factors have been investigated using QCFIRE. These include: fibre type, insulation thickness, thermal conductivity,

195



density, and specific heat, FRP matrix GTT and ignition temperature, concrete aggregate type, compressive strength, steel reinforcement ratio, and FRP confinement ratio.

Effect of Fibre type Figure 5.20 shows fire endurance curves for three identical reinforced concrete columns wrapped with CFRP, GFRP, or AFRP, and protected with 25 mm of VG insulation. To generate the plot, it was assumed that the GFRP and AFRP wraps were applied in such a manner as to provide the same confinement ratio (fl/f’c) in pure compression as a single layer of SCH FRP sheet by adjusting the sheet thickness. The GFRP wrap was assumed to consist of the Tyfo® SEH 51A glass FRP system manufactured by Fyfe Co. (www.fyfeco.com). This system has a tensile strength of 575 MPa and an elastic modulus of 26.1 GPa. The AFRP wrap was assumed to consist of the Wabo® MBrace AK 60 system (www.wbacorp.com), which has a tensile strength of 2000 MPa and an elastic modulus of 120 GPa. Hence, the assumed thicknesses of GFRP and AFRP were 2.0 mm and 0.57 mm respectively, to give a lateral confinement pressure of 5.7 MPa at ultimate, as was the case for the single layer of Tyfo® SCH 30T FRP. The thermal properties of the FRP wraps were assumed to be the same regardless of fibre type, since the transverse thermal conductivity of FRP depends primarily on the thermal characteristics of the polymer matrix (which was assumed to be S-Epoxy for all 3 systems). As such, there is no significant difference in the times to reach failure criteria 1 and 2 for columns wrapped with different fibre types. In terms of criterion 3, which relates to the load carrying capacity of the column during fire, there are some minor differences in the predicted behaviour. Referring to Figure 5.20, the predicted load capacity of the glass FRP-wrapped column decreases comparatively rapidly under exposure to fire, whereas the load capacity of the aramid FRPwrapped column actually increases briefly before decreasing to match that of the carbon FRPwrapped column. This behaviour can be explained by considering the longitudinal coefficients of thermal expansion of the respective FRP materials. Glass FRP has a positive longitudinal CTE,

196



and hence the wrap becomes less effective (relaxes) as a confining mechanism due to heating. Conversely, aramid FRP has a negative longitudinal CTE (it contracts on heating), so that aramid wraps briefly become more effective under heating. This benefit is lost later in the fire exposure though, because thermal degradation of the wrap’s mechanical properties occurs at modestly increased temperatures. Carbon FRP displays virtually no thermal expansion on heating, so the CFRP-wrapped column displays behaviour that is intermediate between glass and aramid. The above discussion assumes that the bonded overlap of the FRP wrap remains intact during fire, an assumption which may or may not hold true in reality. Bond tests at high temperature are required to shed light in this area. Effect of Insulation thickness Figure 5.21 shows the relationship between the VG insulation thickness and the column’s fire endurance based on the 3 failure criteria outlined above (i.e. matrix GTT, matrix ignition temperature, and overall load capacity of the column). In Figure 5.21a, the criterion 1 fire endurance of the member is plotted as a function of VG thickness.

As expected, the fire

endurance increases for increasing insulation thickness. In addition, the figure suggests that keeping the wrap temperature below its GTT (as required by criterion 1) for longer than 120 minutes is unlikely for insulation thicknesses that could be used in practice. The criterion 2 fire endurance, plotted as a function of insulation thickness, is shown in Figure 5.21b, and displays a similar trend to that observed for criterion 1. In this case however, the analysis predicts that fire endurances in excess of 4 hours could be obtained with as little as 15mm of VG insulation, although the model does not account for the VG cracking that was observed in tests. The design service load for the FRP-wrapped column used in the development of Figure 5.21 is 2515 kN (refer to Appendix E), so the criterion 3 fire endurance is not reached during the first 5 hours of fire exposure for any thickness of VG insulation. Figure 5.21c, which plots fire endurance curves for various insulation thicknesses, demonstrates that increasing the insulation

197



thickness even slightly can drastically improve the load carrying capacity of an FRP-wrapped reinforced column during fire. Increasing the thickness of the insulation can also prolong the effectiveness of the FRP confinement, although the effect of the confinement is not significant to the overall structural fire behaviour of the member in any case. It is primarily the insulating effect of the wrap and insulation on the reinforced concrete column that contributes to increased fire endurance. Effect of Insulation Thermal Conductivity To investigate the effect of the insulation’s thermal conductivity, it was assumed that the thermal conductivity of the VG material remained constant with increasing temperature. While this assumption is certainly false, it is useful for the purposes of illustration. The fact remains that the variation of thermal conductivity with temperature for insulation materials commonly available in the construction industry remains poorly understood because of the proprietary nature of the materials and the cost and difficulty associated with obtaining accurate thermal data. Figure 5.22 shows the effect of varying the thermal conductivity of the insulation, assuming it to be 25 mm thick, on the fire endurance of an FRP-wrapped reinforced concrete column with respect to the three previously defined failure criteria. The thermal conductivity of the insulation is observed to play a crucial role in the effectiveness of the VG insulation. A reduction in the insulation’s thermal conductivity from 0.5 W/m·K to 0.1 W/m·K results in an increase of the criterion 1 fire endurance from 9 minutes to 41 minutes (refer to Figure 5.22a), and from 85 minutes to over 8 hours for criterion 2 (refer to Figure 5.22b). Within the range of thermal conductivities of the best fire insulation materials (0.1 to 0.5 W/m·K), even slight reductions in thermal conductivity result in large increases in fire endurance. Figure 5.22c shows fire endurance curves for columns with various insulation thermal conductivities. Again, the service load on these columns is 2515 kN, so failure by criterion 3 will not occur for at least 5 hours of exposure. The benefits of low thermal conductivity insulation are immediately evident, particularly in the 0.1 to 0.5 W/m·K range. For example, an increase in the

198



VG thermal conductivity from 0.1 to 0.5 W/m·K results in a 27% load capacity increase after 4 hours of fire exposure. It is clear that thermal conductivity is a key consideration in the selection of insulating materials for FRP wraps. Effect of Insulation Specific Heat As was the case in the discussion of thermal conductivity, the effect of the insulation’s specific heat was investigated by assuming that the specific heat of the VG insulation was constant with temperature. Again, while this assumption is false it is useful in this context for the purposes of illustration. Figure 5.23 shows the predicted effect of varying the specific heat for a 25 mm thick layer of insulation on the fire endurance of an FRP-wrapped column, again with respect to the three previously mentioned criteria. Also included in Figures 5.23a and 5.23b are similar plots for a 10 mm thick layer of insulation. A linear trend is observed between specific heat and fire endurance for Criteria 1 and 2, and materials with a higher specific heat are preferable. However, the effect of the insulation’s specific heat on fire endurance is not nearly as dramatic as for the thermal conductivity. For the 25 mm thick insulation, an increase in the specific heat from 500 to 2500 J/kg·K results in criteria 1 and two fire endurance increases of 15 minutes and 22 minutes respectively. The effect of increased specific heat is even less for thinner insulation thicknesses, with criteria 1 and 2 fire endurance increases of only 3 minutes and 5 minutes, respectively, for the 10 mm thick insulation for the same increase in insulation specific heat. In addition, low thermal conductivity and high specific heat are contradictory thermal properties for most common materials. Figure 5.23c demonstrates that the effect of the insulation’s specific heat on the load carrying capacity of the column during fire exposure is minimal, with only extremely mild improvements in behaviour observed for substantial increases in specific heat. For example, with all other parameters constant, an increase in the specific heat of the VG from 500 to 2500 J/kg·K, a 500% increase, results in a load capacity increase of only 0.8% after 5 hours of fire exposure.

199



Effect of Insulation Density To investigate the effect of the insulation’s density, it was assumed that the density of the VG material was constant with temperature. Figure 5.24 shows the predicted effect of the insulation density of a 25 mm thick layer of insulation on fire endurance. There is an essentially linear trend between density and fire endurance for criteria 1 and 2, and it is evident that higher density materials are preferable for insulation, although low thermal conductivity and high density are also contradictory thermal properties for common insulating materials. For the 25 mm thick insulation, an increase in the density from 250 to 2500 kg/m3 results in criteria 1 and two fire endurance increases of 50 minutes and 83 minutes respectively. The effect of increased density is less for thinner insulation thicknesses, with criteria 1 and 2 fire endurance increases of only 10 minutes and 17 minutes, respectively, for the aforementioned increase in insulation density for the 10 mm thick layer of VG insulation. Figure 5.24c demonstrates that the effect of the insulation’s density on the load carrying capacity of the column is also minimal, with only very mild improvements in behaviour observed for substantial increases in density. In this case, an increase in the VG density from 250 kg/m3 to 2250 kg/m3 results in only a 2.5% increase in the load capacity after 5 hours of fire exposure. Thus, while low density and high specific heat are beneficial properties for insulation materials, thermal conductivity is by far the most important. Effect of Matrix GTT One potential method to increase the fire endurance of FRP-wrapped columns is with the use of supplementary insulating materials, as discussed above. Another possible technique would be to use FRP materials with superior behaviour at high temperature. While the GTT of S-Epoxy is approximately 93°C and its ignition temperature was determined by the author to be approximately 450°C, various polymer materials are available that have GTTs in excess of 300°C and that are extremely resistant to ignition and flaming. Indeed, a new generation of highly thermally resistant composite materials called Geocomposites are currently being investigated by

200



several researchers (Balaguru et al., 1997; Lyon et al., 1996; Kurtz and Balaguru, 2001). These materials are composed of carbon fibres, identical to those used in FRP composites, in conjunction with a geopolymer matrix (which is an alumino-silicate powder that reacts with water to form a solid material). Research on Geocomposites has indicated that they demonstrate similar room-temperature physical and mechanical properties to FRPs, but that they display superior resistance to high temperatures. Kurtz and Balaguru (2001) reported retention of 63% of the room temperature tensile strength for a carbon/geopolymer composite after exposure to a temperature of 800°C for one hour. Additionally, geocomposites are non-combustible up to at least 800°C. These materials can be considered to be an emerging technology, since they are not readily available for study or use, and are not discussed further. Many high-temperature composites currently exist or are under development, and it is instructive to examine the effect of improving the thermomechanical properties of FRPs on the overall fire behaviour of FRP-wrapped columns. Figure 5.25 shows the predicted fire endurance (based solely on the criterion 1 fire endurance) as a function of matrix GTT, for a number of VG insulation thicknesses. This figure confirms that an increase in the matrix GTT will increase the criterion 1 fire endurance of the members, and that the effect of increasing the matrix GTT on the fire endurance is more pronounced at greater insulation thicknesses. Hence, a combination of slightly increased matrix GTT in combination with some kind of supplementary fire protection has the potential to significantly increase the criterion 1 fire endurance of FRP-wrapped columns. Effect of Matrix Ignition Temperature As mentioned above, composites have been developed that are highly resistant to ignition and flaming. Figure 5.26 shows the predicted fire endurance of an FRP-wrapped concrete column based on criterion 2 for various matrix ignition temperatures. This figure is similar to Figure 5.25 in that an increase in the matrix GTT will increase the criterion 2 fire endurance of the member. Again, the effect of increasing the matrix ignition temperature on the fire endurance is more pronounced at greater insulation thicknesses.

201



Effect of Aggregate Type In design of conventionally reinforced concrete structures, aggregate type is sometimes a key consideration in assigning fire endurance values to different members. Results of parametric studies on columns with carbonate or siliceous aggregates indicated that the criterion 1 and 2 fire endurances were within 5 minutes of each other. Figure 5.27 shows the predicted fire endurance curves for columns fabricated from concretes with different aggregate types. Siliceous and Carbonate aggregate concrete columns are predicted to perform similarly during fire, and the effect of aggregate type is minimal. Effect of Concrete Strength A comparison of the load carrying capacity of columns with different unconfined concrete compressive strengths during fire exposure is shown in Figure 5.28, where the load capacities of columns with various concrete strengths (normalized to their room temperature strength) are plotted as a function of fire exposure time. All the columns were analyzed with identical FRP wrap details. This figure demonstrates that columns with lower concrete strengths can be expected to lose a greater proportion of their axial strength under exposure to fire. This is expected, since a lower unconfined concrete strength results in a higher proportional confinement effect for the column, and a greater proportional strength loss when the wrap’s mechanical properties degrade at high temperature. Effect of Steel Reinforcement Ratio The effect of the steel reinforcement ratio, which is the ratio of the cross-sectional area of reinforcing steel in the column to the gross cross-sectional area, on the fire behaviour of FRPwrapped reinforced concrete columns was examined by varying the reinforcement ratio between 1.3% and 7.2%. These percentages are close to the minimum and maximum steel reinforcing ratios of 1% and 8% suggested for reinforced concrete columns according to CSA A23.3-94 (CSA, 1994), and were achieved using 6 or 8 identical reinforcing bars. The column diameter

202



was maintained constant, as were the details of the FRP wrap, spiral steel, and fire insulation provided, and in all cases the cover to the principal reinforcement was assumed to be 50 mm. Figure 5.29 shows the resulting fire endurance curves for the various columns, normalized to their room temperature strength. As the steel reinforcement ratio increases, the proportional fire behaviour of the columns worsens, although only slightly. This can be attributed to the fact that columns with higher steel reinforcement ratios are proportionally more dependent on the reinforcing steel for strength. The reinforcing steel is close to the outside of the column, and so it is affected by elevated temperatures before the bulk of the concrete in the column. In any case, after 5 hours of fire exposure the difference in retention of room-temperature strength for columns with reinforcement ratios of 1.27 and 7.16 is observed to be only 3.8%, so the effect is not highly significant within the range of steel reinforcement ratios to be expected in practice. Effect of Confinement Ratio The effect of the confinement ratio on the fire behavior of FRP-wrapped columns was examined by varying the thickness of SCH FRP applied to the columns. Five FRP thicknesses were selected so as to provide room temperature confinement ratios of 0%, 10%, 20%, 30% and 40%, which resulted in FRP thicknesses of 0 mm, 0.51 mm, 1.02 mm, 1.54 mm, and 2.06 mm. Because the thicknesses of FRP are small in comparison to the other materials in the column, the heat transfer behaviour of columns with different confinement ratios was essentially identical.

Thus, the criterion 1 and 2 fire endurances were unaffected by increasing the

confinement ratio within reasonable limits. The effect of increased confinement ratio on the overall load carrying capacity of the columns is presented in Figure 5.30, where the load capacity of the columns (normalized to the room-temperature capacity of an unwrapped column) is plotted as a function of fire exposure time. As expected, higher confinement ratios result in higher initial strengths for the column. During fire, the columns (with a 25 mm thick layer of VG protection in this case) are predicted to retain much of their room-temperature strength during exposure to fire, and in some cases the strength is actually predicted to increase, as a consequence of activation of

203



the FRP wrap resulting from differential thermal expansion. At later stages of fire exposure, columns with larger confinement ratios lose a greater proportion of their strength, as the mechanical properties of the wrap are degraded at high temperature. The reader will note that this degradation of strength would occur earlier for smaller VG insulation thicknesses.

5.3.4 Unwrapped but Insulated Reinforced Concrete Columns It is interesting to consider that the primary factor contributing to the enhanced fire endurance of an FRP-wrapped and insulated concrete member is the presence of supplemental insulation, not the confining effect of the FRP wrap. As such, an unwrapped but insulated member could be expected to achieve a similar level of fire endurance as a wrapped and insulated column subjected to the same applied load. This is because both columns are assumed to fail by buckling at elevated temperature, which depends largely on the modulus of the concrete and reinforcing steel in the column. Wrapping with FRP can be expected to increase the modulus of the concrete only very slightly, and thus the FRP wrap does not significantly increase the buckling strength of the column (refer to Figure 5.18). Consequently, loss of the wrap during fire has only minor consequences for the column, which is assumed to be loaded only to service load levels. However, columns are not wrapped with FRP to increase their fire resistance, and in most cases the columns cannot be insulated sufficiently to prevent loss of the wrap. Columns are wrapped for increased axial load capacity (or to improve seismic performance), and fire insulation is generally required in these cases to ensure that the column can carry the increased (wrapped) service load for an adequate period of fire exposure. It is important to examine the fire endurance of FRP-wrapped concrete columns using a holistic approach, and to keep the goals (primarily prevention of collapse) clearly in mind throughout the design process.

5.4 Consequences for Design

204



The above discussion and parametric study indicates that it is unlikely that any of the FRP-wrapping materials currently available in the Civil Engineering industry will be able to perform satisfactorily for any significant period of time during fire unless they are provided with some form of supplemental fire insulation. Depending on the failure criterion selected, the critical factors in the fire design of FRP-wrapped reinforced concrete columns for fire appear to be: the supplementary insulation’s thickness, the insulation’s thermal conductivity, and the glass transition and ignition temperatures of the polymer matrix. Given the wide variety of column sizes, lengths, reinforcing details, concrete strengths, aggregate types, fibre and matrix types, and insulation types, it is neither possible nor practical at this juncture to provide detailed design guidelines for the fire design of FRP-wrapped reinforced concrete columns. It is more appropriate, particularly given the uncertainties that remain with regard to the predictive abilities of the numerical models, to examine these members on a caseby-case basis. However, both the experimental results and the numerical modelling presented in this thesis indicate that outstanding fire endurances can be obtained for FRP-confined concrete members provided that supplemental insulation is applied to the exterior of the FRP wrap and that the insulation stays in place during exposure to fire. The Tyfo® VG material tested in this thesis is a good example of an outstanding insulation material. The following two simple guidelines for FRP-wrapped concrete members can be suggested to guard against failure by collapse during fire (which is the current ULC-S101 or ASTM E119 fire endurance criterion): 1. Under no circumstances should the strengthened (increased) service load on the upgraded column exceed the ultimate design strength of the unstrengthened column. Thus, maximum allowable strength upgrades for FRP-wrapped columns can be suggested for various dead-to-live load ratios using the following expression:

(φRn )existing (S DL

205

S LL )new

[5.44]



where (φRn )existing is the existing strength of the member to be strengthened with FRP, S DL is the strengthened service dead load, and S LL is the strengthened service live load. 2. Supplemental fire protection insulation is essential in most cases. Depending on the degree of upgrading, analysis should be conducted with QCFIRE to determine the approximate thickness of fire insulation required to achieve the desired criterion 3 fire endurance rating. Tests should be conducted to ensure that the fire insulation will remain in place during exposure to fire. The design recommendation (1.) above is similar to the approach suggested in ACI 440.2R-02 (ACI, 2002) and discussed in Appendix E, and will provide a measure of protection against vandalism or poor workmanship in addition to fire. Figure 5.31 shows the resulting allowable strength increases for members with various live-to-dead load ratios using the above recommendation in conjunction with either the Canadian (NRC, 1995) or American (ACI, 1995) load factors. Also included in Figure 5.31 are curves showing the allowable strength increases that result from using the ACI 440.2R-02 (ACI, 2002) strengthening limit equation.

The

strengthening limits indicate allowable strength increases that range from 50% to 100% for predominantly live-loaded members and from 5% to 40% for predominantly dead-loaded members. Obviously, this is because live load factors are greater than dead load factors, and so structures with a greater proportion of live load will have a greater allowable strength increase. The design recommendation (2.) above is essentially a statement of the ULC S101 (or ASTM E119) fire endurance requirement. If it can be reliably shown that the column can carry the increased service load for the required fire duration without the wrap and without supplemental insulation then fire insulation may not be required on the basis of structural fire endurance. However, in these cases some form of fire-barrier coating would be required on the outside surface of the FRP wrap to control flame spread and smoke generation.

206



5.5 Summary In this chapter a numerical model was developed to predict the heat transfer and structural behaviour of FRP-wrapped and insulated reinforced concrete columns exposed to a standard fire. The heat transfer and load capacity models were validated where possible against the results of fire endurance tests on reinforced concrete columns available in the literature, and against the results of the fire endurance tests discussed in Chapters 3 and 4 of this thesis. Parametric studies using the model indicated that insulation is essential to provide fire protection for FRP-wrapped members. While it appears that it will be virtually impossible in practice to prevent loss of mechanical or bond properties of the FRP, the overall structural fire endurance of FRP-wrapped members was predicted to be superior to unwrapped members with even small amounts of supplemental insulation. Further parametric studies were conducted with respect to 3 distinct failure criteria, and indicated that the primary factors to consider in the selection and design of insulation schemes for FRP-wrapped reinforced concrete columns were the insulation thickness and thermal conductivity. In addition, fire performance of FRP-wrapped concrete members can be improved by improving the high temperature retention of mechanical properties of the FRP materials themselves, particularly by increasing the matrix glass transition temperature.

Finally, two simple design recommendations were suggested based on the

assumption that FRP-wraps will be effectively lost during fire.

207


CHAPTER 5: Numerical Models 1 – Columns

Table 5.1: Summary of the various column, wrap, and insulation configurations that were analyzed in conducting parametric studies using QCFIRE Section

Factor Varied

5.3.1: Member Type

Wrap and insulation configuration

5.3.3: Fibre Type

Type of FRP wrap*

5.3.3: Insulation Thickness 5.3.3: Insulation Thermal Cond. 5.3.3: Insulation Specific Heat 5.3.3: Insulation Density

Range 1. Unwrapped 2. Wrapped 3. ISIS-1, ISIS-2 1. Carbon 2. Glass 3. Aramid

Failure Criterion

Relevant Figures

3

5.18, 5.19

1, 2, 3

5.20

VG thickness (mm)

tins = 0-60

1, 2, 3

5.21

VG thermal conductivity (W/m·K)

kins = 0.1-5.0

1, 2, 3

5.22

VG specific heat (J/kg·K)

Cpi = 500-2500

1, 2, 3

5.23

VG density (kg/m3)

ρi = 250-2500

1, 2, 3

5.24

5.3.3: Matrix GTT

Matrix GTT (°C), VG thickness (mm)

Tg = 50-500, ti = 0-60

1

5.25

5.3.3: Matrix Ignition Temp. 5.3.3: Aggregate Type 5.3.3: Concrete Strength 5.3.3: Reinforcement Ratio 5.3.3: Confinement Ratio

Matrix ignition temp. (°C), VG thickness (mm)

Tignition = 200-800, ti = 0-30

2

5.26

Type of aggregate

1. Carbonate 2. Siliceous 3. Expanded shale

3

5.27

Concrete compressive strength (MPa)

f’c = 20-50

3

5.28

Steel reinforcement ratio (%)

ρs = 1.27-7.16

3

5.29

FRP confinement ratio (%)

fl/f’c = 0-40

3

5.30

* Carbon FRP = Fyfe Co. Tyfo® SCH-30T sheets, 0.76 mm thick Glass FRP = Fyfe Co. Tyfo® SEH-51A sheets, 2.00 mm thick Aramid FRP = Wabo® MBrace AK 60 system, 0.57 mm thick

1 – Exceeding the matrix GTT 2 – Exceeding the matrix ignition temperature 3 – Loss of load bearing capacity

208



START

INPUT Type of Analysis Unwrapped, Wrapped, Wrapped-Insulated

Unwrapped

Wrapped

Wrapped-Insulated

Calculate: Temperatures over the column cross-section using the procedure of Figure 5.1b

Calculate: Temperatures over the column cross-section using the procedure of Figure 5.1c

Calculate: Temperatures over the column cross-section using the procedure of Figure 5.1d

OUTPUT Data Files for temperature related output (used primarily for validation of the program)

Calculate: The Load carrying capacity of the column using the procedure of Figure 5.1e

OUTPUT Data files relating to LoadCapacity and deformations with time

END

Figure 5.1a: Main program schematic for QCFIRE

209



IN FROM MAIN PROGRAM

READ FROM INPUT FILE Concrete column radius, Rc Number of rings in the concrete Number of radial slices in the concrete Duration of test Type of aggregate (siliceous, carbonate, expanded shale) Concrete moisture content, φ Initial concrete compressive strength, f’co Initial steel yield strength, fyo Initial steel modulus, Eso Radius to reinforcing bar centroid, rbars Reinforcing bar diameter, dbars Number of longitudinal reinforcing bars Axial load on column

Discretize the column crosssection into a number of ring elements (see Figure A.1) Stability Criteria Calculate minimum time step for stability using Equations A.6 and 5.28 Smallest time step value governs, ∆t

Allocate and Initialize Fire temperature and member temperatures initialized to 20 C

Allocate and initialize Element moisture volumes initialized using Equations 5.14, A.4, and 5.21

Set the time-step number, i=1

Calculate: Fire temperature Equation 5.2

1

2

Figure 5.1b (part 1 of 2)

210



1

2

Calculate: Fire-concrete interface temperature Figure 5.1f Equations A.1 and A.3 Set element counter, m=2

Calculate: Temperature of concrete element Figure 5.1f Equations A.2 and A.5

Increase element counter

NO

Increase time-step counter

Check if all concrete element temperatures have been calculated

YES Calculate: Temperature of the centerline concrete element Figure 5.1f Equations 5.22 and 5.23

Check if the temperature is known at all time steps

NO

YES BACK TO MAIN PROGRAM

Figure 5.1b (part 2 of 2) Figure 5.1b: Program schematic for heat transfer in an unwrapped column

211




READ FROM INPUT FILE Concrete column radius, Rc FRP outer radius, Rw Number of rings in the concrete Number of radial slices in the concrete Number of rings in the FRP Duration of test Type of aggregate (siliceous, carbonate, expanded shale) Concrete moisture content, φ Initial concrete compressive strength, f’co Initial steel yield strength, fyo Initial steel modulus, Eso Radius to reinforcing bar centroid, rbars Reinforcing bar diameter, dbars Number of longitudinal reinforcing bars Design thickness of FRP Sheet, tw Does the wrap fall off? (Y/N) Temperature at which the wrap falls off, Tloss Applied axial Load Ultimate stresngth of FRP, ffrpu Elastic modulus of FRP, Efrp Fibre type (carbon, glass, aramid)

Discretize the column crosssection into a number of ring elements (see Figure 5.2) Stability Criteria Calculate minimum time step for stability using Equations 5.25-5.28 Smallest time step value governs, ∆t


Allocate and initialize Element moisture volumes initialized using Equations 5.14, 5.18, and 5.21

1 Figure 5.1c (part 1 of 3)

212



1 Set the time-step number, i = 1

YES

NO

Check if the wrap has fallen off



Calculate: Fire-concrete interface temperature Figure 5.1f Equations A.1, A3

Calculate: Fire-FRP interface temperature Equation 5.12 Set the element counter, m = 2

Calculate: Temperature of the element, m, in the FRP Equation 5.13


NO

Check if all FRP element temperatures have been calculated

YES Calculate: FRP-concrete interface temperature Figure 5.1f Equations 5.16, 5.17

Set element counter to m = M1+1

2

3

Figure 5.1c (part 2 of 3)

213



2

3

Calculate: Temperature of the element inside the concrete Figure 5.1f Equations 5.19, 5.20


NO



YES Calculate: Temperature of the centerline concrete element Figure 5.1f Equations 5.22, 5.23

Check if the Temperature is known at all time steps

NO

YES BACK TOMAIN 6 PROGRAM

Figure 5.1c (part 3 of 3) Figure 5.1c: Program schematic for heat transfer in an FRP-wrapped column

214




READ FROM INPUT FILE Concrete column radius, Rc FRP outer radius, Rw FRP outer radius, Ri Number of rings in the concrete Number of radial slices in the concrete Number of rings in the FRP Number of rings in the insulation Duration of test Type of aggregate (siliceous, carbonate, expanded shale) Concrete moisture content, φ Initial concrete compressive strength, f’co Initial steel yield strength, fyo Initial steel modulus, Eso Radius to reinforcing bar centroid, rbars Reinforcing bar diameter, dbars Number of longitudinal reinforcing bars Design thickness of FRP Sheet, tw Does the wrap fall off? (Y/N) Temperature at which the wrap falls off, Tloss Applied axial Load Ultimate stresngth of FRP, ffrpu Elastic modulus of FRP, Efrp Fibre type (carbon, glass, aramid)

Discretize the column crosssection into a number of ring elements (see Figure A.2)

Stability Criteria Calculate minimum time step for stability using Equations A.12 - A.14, 5.26 - 5.28 Smallest time step value governs, ∆t


Allocate and initialize Element moisture volumes initialized using Equations 5.14, 5.18, and 5.21

Set the time-step number, i = 1

1 Figure 5.1d (part 1 of 3)

215



1

YES


NO



Calculate: Fire-concrete interface temperature Figure 5.1f Equations A.1, A.3

Calculate: Fire-insulation interface temperature Equation A.7 Set the element counter, m = 2

Calculate: Temperature of the element, m, in the insulation Equation A.8


NO

Check if all insulation element temperatures have been calculated

YES Calculate: Insulation-FRP interface temperature Equation A.9 Set the element counter, m = Mins+1

2

3

4

Figure 5.1d (part 2 of 3)

216



2

3

4

Calculate: Temperature of the element, m, in the FRP Equation A.10


Check if all FRP element temperatures have been calculated

NO

YES Calculate: FRP-concrete interface temperature Figure 5.1f Equations 5.16, A.11

Set element counter to m = M1+1


Calculate: Temperature of the element inside the concrete Figure 5.1f Equation 5.19, 5.20


NO


YES Calculate: Temperature of the centerline concrete element Figure 5.1f Equations 5.22, 5.23

Check if the temperature is known at all time steps

NO

YES BACK TOMAIN 6 PROGRAM

Figure 5.1d (part 3 of 3) Figure 5.1d: Program schematic for heat transfer in an FRP-wrapped and insulated column

217




Discretize the total fire duration into specific points where load capacity will be calculated Discretize the column cross-section into a number of annular elements (see Figures 5.3, A.1, A.2) Set the time to, t = 0 hours

Set the initial curvature to φ=0

Assume an axial strain value, εaxial

Check what type of analysis is being performed

Unwrapped

Wrapped

YES

Wrapped-Insulated


NO Set the confinement pressure,

Calculate: The confinement pressure,

f lat = 0 MPa

1

2

f lat , using

the procedure outlined in Figure 5.1g

3

4

Figure 5.1e (part 1 of 2)

218



1

2

3

4 Reinforcing Steel Strains For each bar calculate: Total strain of bar using Equation 5.30, 5.31 Stress in bar using mechanical properties subroutine for steel Axial force and internal moment at mid-height due to bar

Increase the time, t

Increase the curvature, φ

Modify the assumed total axial strain,

εaxial

(using a false position algorithm)

Concrete Strains For each annular concrete element calculate: Strain of element using Equation 5.30, 5.31 Stress in element using mechanical properties subroutine for confined concrete Axial force and internal moment at mid-height due to element Summation of Forces and Internal Moments, Mint Add up the contributions of all elements to the total axial load and the total internal moment at mid-height using Equation 5.32 External Moment, Mext Calculate: External moment at mid-height due to applied loads using Equation 5.33, 5.34

NO

MINT = MEXT? YES Check if the total axial load is decreasing

NO

YES Check if the load capacity is known at all time steps

NO

YES BACK TO MAIN PROGRAM

Figure 5.1e (part 2 of 2) Figure 5.1e: Program schematic for the load capacity analysis portion of QCFIRE

219



IN FROM HEAT TRANSFER ROUTINE

YES

Check if the temperature of the concrete is < 100 C

NO YES

Check if the element moisture content = 0%

NO Calculate: The temperature of the element at the current time step using the appropriate Equation

Calculate: The amount of moisture evaporated from the element during the current time step using the appropriate Equation

Update the volume of moisture in the concrete element

Set the temperature of the concrete element to 100 C for the current time step

BACK TO HEAT TRANSFER ROUTINE

Figure 5.1f: Program schematic for the algorithm to account for moisture in the concrete

220



IN FROM LOAD CAPACITY ROUTINE Calculate: Average wrap temperature, Twrap, from Equation 5.36

Assume confinement pressure,

f lat

Set the element counter to the outermost concrete ring number

Calculate: Confined concrete strength, f’cc, using Equation 5.37 Calculate: Current axial stress in the concrete element using Equations 5.38a, b, c, and 5.39a, b

Increment element number

Calculate: Current lateral strain, εlat, using Equation 5.40 and thermal strain, εT.

Check if done for all concrete ring elements?

NO

YES Calculate: Total lateral strain of column using Equation 5.41

Update confining pressure

Calculate: Confining pressure using Equation 5.42

Check if calculated confining pressure is equal to assumed

NO

YES BACK TO LOAD CAPACITY ROUTINE

Figure 5.1g: Program schematic for the confining effect of an FRP wrap

221



½ ∆xw 1

∆xw

2 M1

Fire

Wrap

Rc

Rw

½ ∆xc

M1+1

∆xc

m Concrete M2 Figure 5.2: Heat transfer discretization for an uninsulated FRP-wrapped concrete column Column Centreline

FRP-Concrete Boundary

M1, 1

∆C

M1+1, 1

M1+2, 1

Line of Symmetry

Rc

M1+2, n

M1+2, N

M2, 1

Figure 5.3: Load capacity discretization used in QCFIRE

222 222

L.A. Bisby, Bisby, Ph.D. Ph.D. Thesis, Thesis, 2003 2003 L.A.


y

χ

(KL)2 12

y

KL

Deflection

Curvature

Figure 5.4: Assumed variation in column curvature and deflection used in QCFIRE

223



IN FROM HEAT TRANSFER ANALYSIS

Discretize the total fire duration into specific points where load capacity will be calculated Discretize the column cross-section into a number of ring elements

Set the time to, t = 0 hours

Set axial load capacity to zero

Set the overall axial strain,

ε axial = 0 MPa


Unwrapped

Wrapped

YES


NO Calculate: The confinement pressure,

f lat , using


Set the overall confining pressure,

YES

f l = 0 MPa

Check if the wrap has ruptured

NO 1

2

3 Figure 5.5 (part 1 of 2)

224



1

2

3 Reinforcing Steel Strains For each bar calculate: Total strain of bar using Equation 5.30, 5.31 Stress in bar using mechanical properties subroutine for steel Axial force due to bar

Increase the time, t

Increase the assumed total axial strain, εaxial

Concrete Strains For each ring concrete element calculate: Total strain of element using Eq. 5.30, 5.31 Stress in element using mechanical properties subroutine for confined concrete Axial force and internal moment at mid-height due to element Summation of Forces Add up the contributions of all elements to the total axial load using Equation 5.35

Check if the total axial load is decreasing

NO

YES Write data to output file

Check if the Load Capacity is known at all time steps

NO

YES End Program

Figure 5.5 (part 2 of 2) Figure 5.5: Subprogram schematic for QCFIRE-Axial

225



IN FROM HEAT TRANSFER ANALYSIS

Discretize the total fire duration into specific points where load capacity will be calculated Discretize the column cross-section into a number of annular elements (see Figures 5.3, A.1, A.2)


Set the initial curvature to φ=0

Assume an axial strain value, εaxial


Unwrapped

Wrapped

YES

Wrapped-Insulated


NO Set the confinement pressure,

Calculate: The confinement pressure,

f lat = 0 MPa

f lat , using


Reinforcing Steel Strains For each bar calculate: Total strain of bar using Equation 5.31, 5.31 Stress in bar using mechanical properties subroutine for steel Axial force and internal moment at mid-height due to bar

1

2

3

4

Figure 5.6 (part 1 of 2)

226



1

2

3

4 Concrete Strains For each annular concrete element calculate: Total strain of element using Eq. 5.30, 5.31 Stress in element using mechanical properties subroutine for confined concrete Axial force and internal moment at mid-height due to element

Increase the time, t Increase the curvature, φ Modify the assumed total axial strain,

Summation of Forces and Internal Moments, Mint Add up the contributions of all elements to the total axial load and the total internal moment at mid-height using Equation 5.32, 5.33

εaxial

External Moment, Mext Calculate: External moment at mid-height due to applied axial loads using Equation 5.34

(using a false position algorithm)

NO

MINT = MEXT? YES Check if: force @ mid-height > applied load

NO

YES Interpolate between the previous and current total axial strains and the resulting axial loads to determine the overall axial strain at the applied load Calculate the approximate axial elongation of the column based on the overall axial strain and the effective length Write to output file

Check if the axial elongation is known at all time steps

NO

YES END PROGRAM

Figure 5.6 (part 2 of 2) Figure 5.6: Subprogram schematic for QCFIRE-Deflection

227



Substrate

Substrate

Virgin Coating

Substrate

Virgin Coating

Char

Decomposition Gases Heat Flux

Heat Flux

Char

Conduction

Intumescent Zone Intumescent Front

Initial Stage

Intermediate Stage

Final Stage

Figure 5.7: Schematic of the intumescent process (reproduced after Shih et al., 1998) 533 356

533

No. SPECIMENS: 2 COLUMNS CONCRETE: 42 MPa 28-DAY STRENGTH REINFORCEMENT: 6 20M BARS LONGITUDINAL 10M SPIRAL w/ 54 mm PITCH c/c 38 mm COVER TO SPIRAL 48 mm COVER TO PRINCIPAL REINFORCEMENT

SECTION A-A N.T.S.

*all dimensions in mm A

A

3810

54 mm pitch c/c

10M Spiral

Concrete Rebar

25 mm Thick Steel Plate

13 mm Diameter Machined Bar Ends

ELEVATION N.T.S.

Figure 5.8: Details of the columns tested by Lie and Celikkol (1991)

228



Model Predictions Test Data

o

Temperature ( C)

800

25 mm Depth

600 64 mm Depth

400 178 mm Depth

200 0

0

60

120

180

Time (min)

Figure 5.9: Predicted and observed temperatures in the concrete for columns tested by Lie and Celikkol (1991)


8 Model Prediction Test Data

6

4

2

0

0

60

120

180

240

Time (min)

Figure 5.10: Predicted and observed overall axial elongation for columns tested by Lie and Celikkol (1991)

Load Capacity (kN)

5000 Model Prediciton Lie and Celikkol Service Load

4000 3000 2000 1000 0

Predicted Fire Resistance = 209 minutes

0

60

120

180

240

300

Time (min)

Figure 5.11: Predicted and observed fire endurance (load capacity) for the columns tested by Lie and Celikkol (1991)

229



100 50 mm Depth 100 mm Depth 150 mm Depth 200 mm Depth

80

o

Temperature ( C)

(a)

Test Data Model Predictions

60 40 20 0

0

60

120

180

240

300

240

300

Time (min)

120

(b)

Spiral-Top Rebar-Outside Rebar-Inside

o

Temperature ( C)

100


80 60 40 20 0

0

60

120

180

Time (min)

1000

o

Temperature ( C)

(c)

800

ULC-S101 EI/VG VG/FRP FRP/Concrete

600


400 200 0

0

60

120

180

240

300

Time (min)

Figure 5.12: Predicted and observed temperatures for column ISIS-2 a) Temperatures in the concrete b) Temperatures in the reinforcing steel c) Temperatures in the EI, VG, and FRP

230



200 50 mm Depth 100 mm Depth 150 mm Depth 200 mm Depth

150

o

Temperature ( C)

(a)


100

50

0

0

60

120

180

240

300

240

300

Time (min)

(b) o

Temperature ( C)

200

Spiral-Top Rebar-Outside Rebar-Inside

150


100 50 0

0

60

120

180

Time (min)

1000

o

Temperature ( C)

(c)

800

ULC-S101 EI/VG VG/FRP FRP/Concrete


600 400 200 0

0

60

120

180

240

300

Time (min)

Figure 5.13: Predicted and observed temperatures for column ISIS-1 a) Temperatures in the concrete b) Temperatures in the reinforcing steel c) Temperatures in the EI, VG, and FRP

231


(d)

(c)

Figure 5.14: Forced validation plots for column ISIS-2 a) Temperatures in the concrete (FRP/Concrete Forced) b) Temperatures in the reinforcing steel (FRP/Concrete Forced) c) Heat transfer through the FRP (VG/FRP Forced) d) Heat transfer through the VG (EI/VG Forced)

(b)

(a)


232


(d)

(c)

Figure 5.15: Forced validation plots for column ISIS-1 a) Temperatures in the concrete (FRP/Concrete Forced) b) Temperatures in the reinforcing steel (FRP/Concrete Forced) c) Heat transfer through the FRP (VG/FRP Forced) d) Heat transfer through the VG (EI/VG Forced)

(b)

(a)


233



Axial Deflection (mm)

2.5 Test #1: ISIS-2 Test #2: ISIS-1

2.0

Model Predictions Test Data

1.5 1.0 0.5 0.0 -0.5

0

60

120

180

240

300

Time (min)

Figure 5.16: Measured and predicted axial elongation for columns ISIS-1 and ISIS-2 8000

Solid Lines - ISIS-2 Dotted Lines - ISIS-1

Load (kN)

6000

4000

2000

0

Applied Load Predicted Capacity (QCFIRE with eo=2 mm) 0.85 x Predicted Capacity (QCFIRE Axial) Predicted Capacity (QCFIRE with eo=27 mm)

0

60

120

180

240

300

360

Time (min)

Figure 5.17: Predicted and observed axial load capacities for columns ISIS-1 and ISIS-2

Load Capacity (kN)

6000 5000 4000 3000 2000

Unwrapped Uninsulated Wrap Column ISIS-1 Column ISIS-2 Applied Load = 2515 kN

1000 0

0

60

120

180

240

300

Time (min)

Figure 5.18: Fire endurance curves for unwrapped, wrapped, and wrapped and insulated reinforced concrete columns

234



Load Capacity (kN)

5800 5700 5600 5500 5400

Unwrapped Uninsulated Wrap Column ISIS-1 Column ISIS-2

5300 5200

0

20

40

60

80

100

120

Time (min) Figure 5.19: Fire endurance curves for the initial 2 hours of fire exposure for unwrapped, wrapped, and wrapped and insulated reinforced concrete columns

Carbon FRP Glass FRP Aramid FRP

Load Capacity (kN)

5800

5600

5400

5200

0

60

120

180

240

300

Time (min)

Figure 5.20: Effect of FRP fibre type on the fire behaviour of FRP-wrapped and insulated concrete columns

235


(a)

Criterion 1 Fire Endurance (min)


120 100 80 60 40 20 0

0

10

20

30

40

50

60

(b)


Insulation Thickness (mm) 300 250 200 150 100 50 0

0

5

10

15

20

Insulation Thickness (mm)

(c)

Load Capacity (kN)

6000 5000 4000 VG Thickness:

3000

0 mm 5 mm 10 mm 15 mm 20 mm 25 mm 30 mm

2000 1000 0

0

60

Predicted Service Load

120

180

240

300

Time (min)

Figure 5.21: Effect of insulation thickness on the fire endurance of FRP-wrapped and insulated concrete columns a) Fire endurance based on matrix GTT b) Fire endurance based on matrix ignition temperature c) Structural fire endurance

236


(a)



40

30

20

10

0

0

1

2

3

4

5

(b)


Insulation Thermal Conductivity (W/m.K) 300 240 180 120 60 0

0

1

2

3

4

5

Insulation Thermal Conductivity (W/m.K)

(c) Load Capacity (kN)

6000 5000 4000 Thermal Conductivity:

3000

k = 0.1 W/m.K k = 0.2 W/m.K k = 0.3 W/m.K k = 0.4 W/m.K k = 0.5 W/m.K k = 1.0 W/m.K k = 2.0 W/m.K k = 5.0 W/m.K

2000 1000 0

0

60


120

180

240

300

Time (min)

Figure 5.22: Effect of insulation thermal conductivity on the fire behaviour of FRPwrapped and insulated concrete columns a) Fire endurance based on matrix GTT b) Fire endurance based on matrix ignition temperature c) Structural fire endurance

237




(a)

100 25 mm Thickness 10 mm Thickness

80 60 40 20 0 500

1000

1500

2000

2500


Specific Heat (J/kg.K)

(b)

500 400 300 200 100

25 mm Thickness 10 mm Thickness

0 500

1000

1500

2000

2500

Specific Heat (J/kg.K) 6000 Predicted Service Load

5000

Load Capacity (kN)

Load Capacity (kN)

(c)

5800

4000 3000

Specific Heat (J/kg.K): Cp = 500 Cp = 1000 Cp = 1500 Cp = 2000 Cp = 2500

2000 1000 0

0

60

120

180

240

5700

5600

5500

5400

300

Time (min)

0

60

120

180

240

300

Time (min)

Figure 5.23: Effect of insulation specific heat on the fire behaviour of FRP-wrapped and insulated concrete columns a) Fire endurance based on matrix GTT b) Fire endurance based on matrix ignition temperature c) Structural fire endurance (full and partial y-axes)

238




(a)

100 25 mm Thickness 10 mm Thickness

80 60 40 20 0

500

1000

1500

2000

2500

3


Insulation Density (kg/m )

(b)

500 400 300 200 100

25 mm Thickness 10 mm Thickness

0

500

1000

1500

2000

2500

3

Insulation Density (kg/m ) 7000 6000

Load Capacity (kN)

5800


Load Capacity (kN)

(c)

5000 4000 Insulation Density (kg/m3):

3000

ρ ρ ρ ρ ρ

2000 1000 0

0

60

120

180

240

5700

5600

5500

5400

300

Time (min)

0

60

120

180

240

300

Time (min)

Figure 5.24: Effect of insulation density on the fire behaviour of FRP-wrapped and insulated concrete columns a) Fire endurance based on matrix GTT b) Fire endurance based on matrix ignition temperature c) Structural fire endurance (full and partial y-axes)

239




300 240 Insulation Thickness (mm): 0 10 20 30 40 50 60

180 120 60 0

100

200

300

400

500 o

Matrix Glass Transition Temperature ( C)


Figure 5.25: Effect of FRP matrix GTT on the criterion 1 fire endurance of FRP-wrapped and insulated concrete columns 300 240 Insulation Thickness (mm): 0 5 10 15 20 25 30

180 120 60 0 200

300

400

500

600

700

800

o

Matrix Ignition Temperature ( C)

Figure 5.26: Effect of FRP ignition temperature on the fire behaviour of FRP-wrapped and insulated concrete columns 5800 Predicted Service Load

Load Capacity (kN)

Load Capacity (kN)

5000 4000 3000 2000 1000 0

5700

5600

5500

Carbonate Aggregate Siliceous Aggregate

0

60

120

180

240

5400

300

Time (min)

0

60

120

180

240

300

Time (min)

Figure 5.27: Effect of concrete aggregate type on the fire behaviour of FRP-wrapped and insulated concrete columns (full and partial y-axes)

240



1.02


1.0

Normalized Load Capacity


1.2

0.8 0.6 0.4 20 MPa Concrete 30 MPa Concrete 40 MPa Concrete 50 MPa Concrete

0.2 0.0

0

60

120

180

240

1.00 0.98 0.96 0.94 0.92

300

0

60

Time (min)

120

180

240

300

Time (min)

Figure 5.28: Effect of concrete compressive strength on the fire endurance of wrapped and insulated columns, normalized to room temp. strength (full and partial y-axes) 1.02


1.0



1.2

0.8 0.6 ρ = 1.27% ρ = 2.39% ρ = 3.34% ρ = 4.78%

0.4 0.2

ρ = 7.16%

0.0

0

60

120

180

240

1.00 0.98 0.96 0.94 0.92 0.90

300

0

60

Time (min)

120

180

240

300

Time (min)

Figure 5.29: Effect of steel reinforcement ratio on the fire endurance of wrapped and insulated concrete columns, normalized to room temp. strength (full and partial y-axes) 1.20



1.2 1.0 Predicted Service Load

0.8 0.6 0.4

fl/f'c = 0% fl/f'c = 10% fl/f'c = 20% fl/f'c = 30% fl/f'c = 40%

0.2 0.0

0

60

120

180

240

1.15 1.10 1.05 1.00 0.95 0.90

300

Time (min)

0

60

120

180

240

300

Time (min)

Figure 5.30: Effect of confinement ratio on the fire endurance of wrapped and insulated concrete columns, normalized to unconfined room temp. strength (full and partial y-axes)

241


Allowable Strength Increase (%)


100

(φRn )existing (0.85S DL (φRn )existing (S DL

80

1.2 S LL )new

S LL )new

60 40 Eq. 5.44 and ACI Load Factors Eq. E.14 and ACI Load Factors Eq. 5.44 and NRC Load Factors Eq. E.14 and NRC Load Factors

20 0

0

1

2

3

4

5

Live-to-Dead Load Ratio

Figure 5.31: Maximum allowable strength increases for FRP-wrapped members according to the recommendations presented herein and those suggested by ACI 440.2 R-02 (ACI, 2002)

242


CHAPTER 6: Numerical Modelling 2 – Slabs

CHAPTER 6 NUMERICAL MODELLING 2 – SLABS 6.1 Slab Modelling Fire endurance criteria for reinforced concrete slabs have traditionally been defined in terms of allowable temperatures at various locations in the slab, as opposed to load carrying capacity. If a similar approach is used for FRP bar-reinforced concrete slabs, as is currently done in CSA S806: Design and Construction of Building components with Fibre Reinforced Polymers (CSA, 2002), then numerical modelling to study the fire endurance of these members is considerably less complicated than that required for FRP-wrapped reinforced concrete columns presented in the preceding chapter. For conventionally reinforced concrete slabs that incorporate steel reinforcement, the fire endurance is defined by limiting the allowable temperature at the level of the reinforcement or at the unexposed face of the slab. Conventional fire endurance requirements, as they pertain to both steel and FRP bar-reinforced concrete slabs, are discussed more completely in Section 6.1.3. The reader should note that the slab model makes no attempt to address concerns associated with transverse thermal expansion of FRP bars in concrete, which could potentially lead to the development of internal stresses and concrete cover spalling (as discussed in Chapter 2).

6.1.1 Heat Transfer Model The heat transfer equations required for thermal analysis of reinforced concrete slabs were derived in a similar manner to those presented in Chapter 5 for FRP-wrapped reinforced concrete columns, so the details of the derivation are not included here. The final equations used in the analysis are similar to those presented by Kodur and Baingo (1998), with the exception that the beneficial effects of moisture have been included in the current analysis. Slabs are assumed,

243



for the purposes of analysis, to be of infinite size horizontally (edge effects are ignored), and moisture is accounted for in the same manner as in the column models. Again, the timetemperature behaviour of the surroundings is described by the ULC-S101 standard fire. The concrete slab is divided into a series of elemental layers as shown in Figure 6.3 and the heat transfer behaviour is described by the equations below, all of which were derived using an energy balance approach (refer to Section 5.3). At the fire-slab interface:

T1i

T1i

1

ρ c Cc 1 k1i 2∆x

2∆t σε c T fi i 1 ρ H 2O C H 2Oφ1 ∆x

(

1

k 2i 1 T1i

1

T2i

1

273

) (T 4

i 1 1

273

)

4

[6.1]

1

At any point inside the slab:

Tmi

Tmi 1

∆t ρ H 2O C H 2Oφ1i 1 ∆x 2

2 ρ c Cc

k mi 11 k mi 1 Tmi 11 Tmi 1

[6.2]

k mi 1 k mi 11 Tmi 1 Tmi 11 And at the slab-air interface (the unexposed face of the slab):

TMi

TMi 1

ρ c Cc 2σε c

(T

∆t ρ H 2O C H 2Oφ1i 1 ∆x

i 1 M

273

) (T 4

i 1 air

1 i1 k M 1 k Mi 1 TMi 11 TMi 1 ∆x 273

)

4

[6.3]

i 1 1.5 air

2627 TMi 1 T

In the above expressions, Tair is the temperature of the ambient air above the slab (it is assumed that fire exposure occurs from below), and εs is the emissivity of the ambient surroundings. The final term in Equation 6.3 accounts for convective heat loss from the top surface of the slab and has been adapted from empirical equations for convective heat transfer above a horizontal surface as suggested by Spiers (1962). The initial volume of moisture in the slab-fire interface and slab-ambient interface elements is given by:

V1H 2O

VMH 2O

244

1 ∆xφ i 2

[6.4]



The volume of moisture evaporated from the fire-concrete interface element during a time interval, t, is:

∆V

H 2O 1

∆t

σε c (T

ρ H O λH O 2

i 1 f

273

) (T 4

i 1 1

273

)

k1i

4

1

k 2i 1 i T1 2 ∆x

2

1

T2i

1

[6.5]

and, from the concrete-ambient interface, the volume of moisture evaporated is:

∆VMH 2O

∆t

ρ H O λH O 2

σε c

2

k Mi 1 1 k Mi 1 i 1 TM 1 TMi 1 2∆x

(T

i 1 M

273

) (T 4

i 1 a

273

)

4

[6.6]

The volume of moisture in the internal slab elements is given by:

VmH 2O

∆xφ i

[6.7]

and the volume of moisture evaporated during a single time interval is:

∆VmH 2O

∆t

k mi 11

2∆xρ H 2O λ H 2O

k mi 1 Tmi 11 Tmi 1

k mi 1

k mi 11 Tmi 1 Tmi 11

[6.8]

Three criteria are required for stability of the QSFIRE heat transfer analysis. These are, at the slab-fire interface:

∆t

(ρ c Cc )min ∆xc 2 2 (hrad )max ∆x (k c )max

[6.9]

at any point inside the slab:

(ρ c Cc )min ∆x 2(k c )max

[6.10]

(ρ c Cc )min ∆x 2 2 (hrad ,a )max ∆x (k c )max

[6.11]

∆t and at the slab-ambient interface:

∆t

Repeated application of Equations 6.1 to 6.11, in conjunction with the standard fire curve of Equation 5.2, gives the distribution of temperatures in the slab at any point in time during its

245



exposure to fire. A schematic showing the program logic for QSFIRE is given in Figures 6.1 and 6.2.

6.1.2 Flexural Capacity Model Various factors require careful consideration in the design of reinforced concrete slabs. For FRP-reinforced concrete slabs, serviceability criteria (deflection and crack control) often govern the amount and detailing of tensile reinforcement, since FRP reinforcing products generally have elastic moduli that are considerably less than that of steel (refer to Table 2.1). During fire exposure however, serviceability criteria become less important and the focus shifts toward prevention of collapse and life-safety objectives. In this section, a relatively simple straincompatibility approach is taken to approximate the flexural strength in positive bending of FRPreinforced concrete slabs exposed to fire from below. The following assumptions are required for analysis: 1. Concrete has no strength in tension. 2. Plane sections before bending remain plane after bending. 3. There is perfect bond between the reinforcement and the concrete. 4. The thermomechanical properties of concrete, reinforcing steel, and FRPs are described by the mathematical relationships presented in Appendix F. 5. Serviceability criteria are satisfied by the slab designs at room temperature. Using the above assumptions and the known distribution of temperatures in the concrete at any time during fire exposure (obtained from the heat transfer analysis), the flexural capacity of the slab can be approximated for a 1 metre wide strip of the cross-section using a straincompatibility/force-equilibrium approach. The procedure for analysis at each time step is shown in Figure 6.2 and is implemented as described below. At any instant in time, the neutral axis depth is assumed just above the tensile reinforcement. The temperatures of the extreme compression fibre and the tensile reinforcement

246



are obtained from the heat transfer analysis.

The failure strain of the extreme concrete

compression fibre is obtained from the thermomechanical subroutines, and the strain in the tensile reinforcement is subsequently obtained using strain compatibility (the initial assumption being that failure of the slab occurs by concrete crushing). The slab is divided into a number of elemental layers (refer to Figure 6.3), and for each layer: 1. The temperature, area, and distance from the assumed neutral axis are determined.

( )

2. The flexural strain, ε f

m

, in the element is determined from strain compatibility.

3. The thermal strain in the element is determined based on its temperature at the current time step:

(ε )

T m

c

(T

i m

Tinit

)

[6.12]

4. The strain causing stress is determined by summing the flexural and thermal components:

(ε T )m (ε f )m

εm

[6.14]

5. The stress in each element is obtained from thermomechanical subroutines, the elemental force is calculated, and the contribution to the total moment about the neutral axis is obtained. For the tensile reinforcement, the procedure is slightly different depending whether steel or FRP reinforcement is being considered.

For FRP reinforcement, the thermomechanical

properties are obtained from the FRP subroutines. If the strain in the FRP is greater than the failure strain at the current temperature then the tensile force is determined from the FRP ultimate failure stress. If the strain in the FRP is less than ultimate, the bar stress is determined based on its strain and current elastic modulus.

For steel reinforcement, the stress at the current

temperature is obtained from the thermomechanical subroutines (the steel reinforcement is assumed never to rupture). Once the stress in the bar is known, the overall tensile force can be calculated and its contribution to the moment about the neutral axis can be obtained.

247



Once the compressive and tensile forces are known, QSFIRE checks if the tensile and compressive forces in the slab cross-section are equal. If not, the neutral axis depth is decreased slightly and the above procedure is repeated. If yes, the time is incremented forward and the entire procedure is repeated. Successive implementation of the above procedure results in a curve giving the predicted flexural capacity of the slab at any time during exposure to fire. Refer to Figure 6.2 for clarification.

6.1.3 Fire Endurance Requirements As mentioned above, the fire endurance of a reinforced concrete slab has traditionally been defined in terms of temperature requirements. Two temperature limits are specified for reinforced concrete slabs. The first of these is that the average temperature rise at the unexposed face of the slab (usually the top face of the slab) must be less than 140 C, and that the temperature rise at any specific point must be less than 180 C. In the analysis presented herein, the initial temperature of the slab is assumed to be 20 C, so if the temperature at the centroid of element M (in Figure 6.3) reaches 160 C, the analysis is terminated and the slab is assumed to have failed. This failure criterion depends primarily on the overall slab thickness and aggregate type, and is not likely to be influenced by the type or amount of reinforcement used. Essentially, using thicker slabs will result in greater fire endurances. It is thus evident that the current code requirements for overall slab thickness should be adhered to when designing FRP bar or gridreinforced concrete slabs for fire. The second failure criterion dictates that the temperature at the level of the internal tensile reinforcement must not exceed the critical temperature of the reinforcement. For conventional steel reinforcement the critical temperature is 593 C. The critical temperature for currently available FRP reinforcements, based on the discussion of FRP properties at high temperature presented in Chapter 2 and Appendix D, should be considerably less than 593 C and in the range of 100 C to 300 C, depending on the particular formulation of reinforcing fibre and polymer

248



matrix used in the fabrication of the FRP. Given that different FRP reinforcements could have dramatically different critical temperatures, the slab model has been used herein to generate design charts that give the required concrete cover to achieve a desired fire rating for a prescribed overall slab thickness and aggregate type. The design charts are discussed further in Section 6.3. Required fire ratings for reinforced concrete slabs were discussed previously in Section 5.2.8.

6.1.4 Validation Because of funding and time constraints it was not possible to conduct validation tests on FRP bar-reinforced slabs for this thesis – although the author did participate as an independent research consultant during a joint research study into the fire behaviour of FRP bar-reinforced concrete slabs conducted by NRC in partnership with Public Works and Government Services Canada (PWGSC). The results of the NRC/PWGSC study will be reported elsewhere, but it should be noted that QSFIRE has been used in the analysis of the slabs tested in that study. For the work presented here, it was required that the slab model be validated against results from slab test data available in the literature. The load capacity portion of the analysis has yet to be validated, and tests on FRP-reinforced concrete slabs under load will be required to accomplish this goal. Two sets of data were used in an attempt to validate the heat transfer portion of the QSFIRE analysis. The first set of validation data was obtained from an extensive study into the fire endurance of reinforced concrete slabs conducted by the Portland Cement Association (PCA) in the 1960s (PCA, 1968). This early study was conducted to examine the effect of a variety of factors on fire behaviour of slabs, including: slab thickness, aggregate type, and moisture content. The second set of validation data was taken from a more recent study into the fire behaviour of NEFMACTM FRP grid-reinforced concrete slabs conducted by NEFCOM Corporation (NEFCOM Corporation, 1998). This second study provided validation data for the time-temperature history at various locations within the slab for both steel and FRP reinforcement.

249



To validate QSFIRE, the results of the model for two different aggregate types were compared with experimental data from the PCA (1968) and NEFCOM Corporation (1998) studies. Figure 6.4 shows a comparison of the analytical predictions and experimental results for the time-temperature behaviour at different depths in a carbonate aggregate concrete slab based on the PCA study, and Figure 6.5 shows a similar plot for siliceous aggregate concrete from the same study. It is evident that there is generally good agreement between the analytical and experimental data, particularly in the critical 400 C to 600 C temperature range where changes in thermal and mechanical properties of both steel and concrete show their greatest variation. The agreement is not as good near the exposed face of the slab, likely due to a combination of errors induced by surface effects, moisture migration, and inaccuracies in measuring temperatures experimentally. The model appears to over-predict temperatures near the exposed face, but this deficiency is conservative when the model is used to formulate design recommendations. Figure 6.6 shows a comparison of the analytical predictions and experimental results for the time-temperature behaviour at the level of the reinforcement for a 120 mm thick carbonate aggregate concrete slab based on the NEFCOM study. Again, the agreement between the model and experimental data is excellent, at both 15 mm and 30 mm of concrete cover. Figure 6.6 also demonstrates that the type of FRP reinforcement does not appear to significantly affect the temperatures in the slab, which is consistent with the assumptions made by the heat transfer model. It appears from Figures 6.4 to 6.6 that QSFIRE is capable of adequately predicting the temperatures in both steel and FRP reinforced concrete slabs during fire, and that the type of reinforcement does not significantly affect the temperatures in the slab for exposures up to 4 hours.

250



6.1.5 Critical Temperatures for FRP Reinforcing Bars and Grids For conventional reinforcing steel the critical temperature is 593 C, the temperature at which it loses about 50% of its room temperature yield strength. This temperature arose from observations of axial compression tests on steel structural sections at elevated temperatures, where it was noted that steel columns could no longer support their design service load at 593 C. Given that the critical temperature for steel is defined in terms of about a 50% yield-strength reduction, a similar criterion could be used to define the critical temperature for FRP reinforcing materials. However, it is difficult to make recommendations for FRP reinforcements because of the number that are currently available. A standard test method to determine critical temperatures for FRP reinforcements will be required before they can be defined with confidence and uniformity. Another complicating factor in the fire behaviour of FRP reinforcements arises from the severe bond degradation that has been observed for some FRP bars at modestly increased temperatures. While current fire design guidelines do not appear to be concerned with loss of bond strength for conventional steel reinforcement, loss of bond could, in the case of FRP reinforcement, be a more critical factor than loss of tensile strength. If for instance, bond is relied upon to anchor reinforcing bars in tension (as in most non-prestressed reinforcing applications) then loss of bond could lead to premature failure during a fire. If however, FRP reinforcing bars are continuous over many bays of a structure and the fire is confined to a single bay (a scenario which is certainly plausible), then bond would be maintained in unexposed areas, and localized loss of bond would simply cause the structures to behave as unbonded members, with significant tensile strength remaining in the FRP reinforcement. Such detailed considerations of the potential behaviour of structures in fire are not currently permitted in North-American design codes. However, there has recently been a shift in the engineering design community, and a push is underway to develop performance or objective based design codes (NRC, 1996) where such considerations might eventually be permitted. Until

251



performance based codes are formulated and sanctioned by the engineering community, it seems inevitable that critical temperatures for FRP reinforcements will be required in order for design guidelines to be suggested. It is neither possible nor within the scope of this thesis to define critical temperatures for the wide variety of FRP reinforcements currently available in industry. Thus, the design aids developed and presented herein have been formulated such that the user can choose the critical temperature of the specific reinforcement that they are employing, ideally based on tensile and bond tests at high temperature. At present, the critical temperature of FRP reinforcements should be taken as the temperature at which 50% of the original tensile strength is lost or the temperature at which 50% of the bond strength is lost (in bond-critical applications). Based on the semi-empirical thermomechanical models for FRP presented in Appendix D, and employing a 50% strength loss criterion to define critical temperature, the approximate critical temperatures for carbon, glass, and aramid FRPs are 360 C, 305 C, and 300 C, respectively. The above quoted temperatures represent best-fit critical temperatures based on a sparse database of experimental data, and for a particular FRP the critical temperature might be substantially less. In the following discussions and parametric studies, a critical temperature of 250 C has been selected for the purposes of illustration, unless otherwise stated.

6.2 Results and Parametric Studies 6.2.1 Fire Endurance Based on Critical Temperature The model presented in Section 6.1.1 was used to conduct parametric studies into the effect of a number of factors on the fire endurance of FRP-reinforced concrete slabs. The fire endurance has been defined in this section in terms of the critical temperature of the reinforcing material (taken as 250 C). At the end of this section, a brief discussion examining the results of the QSFIRE flexural capacity analysis is included. The goal in the following parametric study is simply to identify those factors that are most important to consider in the development of fire

252



design tools for FRP-reinforced concrete slabs. Details of the various analyses conducted during the parametric studies are presented in Table 6.1. Slab Thickness The effect of the overall slab thickness on the fire endurance of an FRP-reinforced concrete slab with an assumed critical temperature of 250 C is given in Figure 6.7 for three concrete cover thicknesses. It is evident that only at smaller slab thicknesses (below 120 mm) is the overall depth of slab a significant factor in determining its fire endurance. For larger slab thicknesses (i.e. those that would be used in practice), the fire endurance remains essentially constant. The Canadian reinforced concrete design code, CSA A23.2-94 (CSA, 1994), specifies a minimum slab thickness of 120 mm for 2-way slabs. Hence, only for very large concrete covers (in excess of 60 mm) is the overall slab thickness likely to influence the fire endurance when it is based on the temperature at the level of the reinforcement. Design charts were developed only for 120 mm and 200 mm thick slabs. It is suggested that the 120 mm design charts be used for all slabs less than 200 mm thick, an approach that will yield conservative results. Slabs greater than 200 mm thick could be designed using the 200 mm design chart. Concrete Cover Thickness and Aggregate Type The effect of concrete cover thickness on the fire endurance of a 150 mm thick FRPreinforced concrete slab, again with an assumed FRP critical temperature of 250 C, at the level of the reinforcement is shown in Figure 6.8. Also shown in this figure is the effect of the aggregate type on the fire behaviour of the slab. The thickness of the concrete cover to the reinforcement has a pronounced effect on the fire endurance of the slab. Hence, concrete cover thickness is a primary variable to consider in the development of design charts for FRP-reinforced concrete slabs. It is also evident that use of siliceous aggregate generally results in slightly lower fire endurances, while use of expanded shale aggregate generally results in slightly higher fire endurances.

Carbonate aggregate displays an intermediate fire endurance, with values

253



intermediate between siliceous and expanded shale. Aggregate type has also been incorporated as a variable in the development of design charts for slabs. Reinforcement Critical Temperature Figure 6.9 shows the effect of the critical temperature of the reinforcement on the fire endurance of a 150 mm thick FRP bar-reinforced concrete slab for three different concrete cover depths. The critical temperature of the reinforcement is a primary factor in the fire endurance of a reinforced concrete slab, as expected, and so FRP critical temperature was a key variable in the development of the design charts presented in Section 6.3. Moisture Content Finally, Figure 6.10 shows the relationship between the initial moisture content of the concrete and the fire endurance for a 150 mm thick FRP-reinforced concrete slab for three different cover thicknesses. Although it appears that moisture content plays a significant role in determining the fire endurance of a reinforced concrete slab, it is important to consider that the moisture content of a concrete slab in service would typically be in the range of 5-10%, and would rarely fall below this value. Hence, all design charts were developed for concrete with a moisture content of 5%. This is at the conservative end of the likely moisture content spectrum. The reader will note that the potential for spalling of the concrete cover is increased for slabs with high moisture contents. Spalling of concrete cover has not been investigated in the current analysis, although it is worth pointing out that, for concretes with moisture contents in excess of 10%, spalling of the concrete cover could result in direct exposure of the reinforcement to fire. This would result in a very poor fire endurance rating for the slab. It is assumed in the discussions herein that spalling does not occur. Summary Examination of Figures 6.7 through 6.10 leads to the conclusion that aggregate type, overall slab thickness, critical temperature of the reinforcement, and concrete cover, all have the

254



potential to significantly influence the fire endurance of FRP bar or grid-reinforced concrete slabs.

6.2.2 Structural Fire Endurance The flexural capacity of reinforced concrete slabs during fire is largely dependent on the temperature of the tensile reinforcement, since it is thermal degradation of the reinforcement’s mechanical properties that leads to failure of the slab. As such, the fire endurance based on flexural strength is affected in a similar fashion by many of the factors examined above as is the fire endurance based on the reinforcement’s critical temperature. In addition, reinforcement with superior high temperature mechanical properties would obviously result in higher structural fire endurance ratings for FRP-reinforced concrete slabs. The flexural capacity component of the QSFIRE model was used to conduct a preliminary examination of the structural fire endurances of FRP-reinforced concrete slabs incorporating different reinforcement types. For the purposes of comparison, a 200 mm thick, carbonate aggregate slab was analyzed assuming 30 mm of concrete cover to the reinforcement and a room-temperature concrete compressive strength of 35 MPa. Type of Reinforcement Figure 6.11 shows the variation in flexural strength of the assumed slab during fire exposure for a variety of reinforcement types. To generate this plot, four different slabs were designed using steel, CFRP, GFRP, and AFRP reinforcement respectively. The design strength of all slabs was calculated based on a 1 m wide strip of slab reinforced with 15M steel reinforcing bars. A steel reinforcement area of 2000 mm2/m width was chosen, and the design ultimate strength for the slab was calculated according to CSA A23.3 (CSA, 1994) as 96.7 kN·m. FRP reinforcing products were chosen for various types of FRP (carbon, glass, and aramid), and the areas of FRP reinforcement per metre width of slab were selected to obtain the same ultimate design strength as the steel-reinforced slab (according to the ISIS Canada design guidelines (ISIS,

255



2001b)). The clear cover to the reinforcement was 30 mm for all slabs. Details of the resulting four slabs are presented in Table 6.2. The service load for the slabs was determined by backcalculating from the ISIS ultimate design load, assuming a 1:1 live-to-dead load ratio and using the Canadian load factors of 1.5 for live load and 1.25 for dead load. The resulting service load moment was determined to be 70.9 kN·m. The flexural strength of all slabs is predicted to decrease with increasing temperature. The decrease occurs earlier and is more pronounced for slabs reinforced with FRP. However, because the predicted room-temperature flexural capacities for the FRP-reinforced concrete slabs are substantially greater than that of the steel-reinforced slab, the FRP-reinforced slabs initially display superior load-carrying capacity.

Beyond about 70 to 90 minutes of exposure, the

predicted flexural strength of the FRP-reinforced slabs degrades and becomes less than the steelreinforced slab. Based on the service load moment of 70.9 kN·m, the fire endurances of the slabs are shown in Figure 6.11 as 185 minutes, 90 minutes, 115 minutes, and 97 minutes, for the steel, CFRP, GFRP, and AFRP reinforced slabs respectively. While it appears from the fire endurances quoted above that the structural fire behaviour of glass FRP-reinforced slabs is superior to that of carbon and aramid, this apparent superiority is due to the relatively low strength reduction factor (0.4 for GFRP) used by the ISIS guidelines in calculating the design ultimate strength from which the service load was determined. For CFRP, the reduction factor is 0.8, for AFRP it is 0.6, and for steel it is 0.85. Strength reduction factors were set to unity for the model analysis, which explains why the FRP-reinforced slabs initially display a higher predicted capacity than the steel-reinforced slab. It is important to recognize that the details of the FRP reinforcement in the slabs used in the comparison of Figure 6.11 were chosen based on an equivalent ultimate strength criterion. Because serviceability requirements often govern the design of FRP-reinforced concrete members, the members used in the analysis are not likely to meet serviceability requirements for crack control or deflection. While it is not possible to check the serviceability of the slabs based

256



on allowable deflections (since no span length is required in the analysis presented herein), allowable service load moments can be approximated based on Clause 7.3.2 of the ISIS (2001b) design guidelines.

This clause limits the strain in FRP reinforcement at service load levels to

2000 µ , to ensure that crack widths remain less than about 0.5 to 0.7 mm. The allowable service load moment can be calculated by assuming the strain in the FRP to be 2000 µ and using strain compatibility to determine the internal moment at that strain level.

This calculation was

performed for all three FRP-reinforced slabs presented in Table 6.2, and it was determined that all three had allowable service load moments (based solely on the ISIS permissible crack width criterion) that were substantially less than the service load obtained by back-calculating from the ultimate capacity (71 kN·m). Because the design of all three FRP-reinforced slabs is actually governed by serviceability (crack width) criteria, they were redesigned such that the crack width criterion was satisfied at the same service load level as calculated for the steel-reinforced slab. Thus, the slabs were designed for a service load moment of 71 kN·m at an FRP strain of 2000 µ . The resulting amounts of FRP-reinforcement required in each slab are shown in Table 6.2, and the predicted structural fire endurance curves for the redesigned slabs are shown in Figure 6.12. As expected, the redesigned slabs have substantially higher room-temperature flexural capacities as a consequence of their increased FRP-reinforcement ratios. Another consequence of the increased reinforcement ratios is that the fire endurance of the redesigned slabs is outstanding, and all three slabs actually perform comparably to, or better than, the steel-reinforced slab, with fire endurance times of 165 minutes and 233 minutes for the GFRP and CFRP-reinforced slabs, respectively. The AFRP-reinforced slab displays a predicted fire endurance in excess of 4 hours. These preliminary observations suggest that FRP-reinforced slabs are capable of achieving similar fire ratings as steel-reinforced slabs (with the same clear cover to the reinforcement). This is because the FRP-reinforced slabs are generally designed under serviceability criteria (and are

257



hence required to contain substantially more reinforcement than required by strength requirements alone), but only ultimate strength criteria are critical in fire. The flexural strength model gives a simplified analysis of the flexural strength behaviour of FRP-reinforced concrete slabs in fire, and while it could be used to examine the effects of various other factors such as concrete strength and reinforcement ratio, it does not account for important factors such as bond-degradation at high temperature (which is likely severe) and localized reinforcement degradation at cracks. These cracks would be increased in size as a consequence of deterioration of reinforcement stiffness and bond properties, allowing flame to directly impinge on the reinforcement. Hence, the strength model is considered preliminary.

6.3 Consequences for Design Once parametric studies were conducted to determine which variables most strongly influence the fire endurance of FRP-reinforced concrete slabs, a series of design charts were developed that provide the required cover to FRP reinforcement for a particular overall slab thickness, critical temperature of reinforcement, aggregate type, and fire endurance. The reader will note that the fire endurance for the design charts has been defined exclusively in terms of the critical temperature of the reinforcement, and no attempt has been made to provide design guidelines based on the flexural capacity of the members. This approach is equivalent to that presently used for conventionally reinforced concrete slabs (CSA, 2002). It is likely that the strength model, once validated and extended to account for factors such as bond degradation and cracking, could be a useful tool in objective-based approaches to fire design. A sample design chart is provided in Figure 6.13, while the full spectrum of charts is provided in Appendix B. Examining the design chart presented in Figure 6.13, the user can determine the required concrete cover to the FRP reinforcement for a desired fire endurance rating and a known critical temperature of FRP in question. For instance, for a 120mm thick carbonate aggregate FRPreinforced concrete slab with a desired 1-hour fire endurance rating, a CFRP bar or grid with a

258



critical temperature of 360 C (as determined from the data of Appendix D) would require a concrete cover of about 40 mm. Conventional steel reinforcement (with a critical temperature of 593 C) would require a concrete cover of only 13 mm. Thus, substantially larger concrete cover thicknesses are required for FRP-reinforced concrete slabs to achieve the same level of fire safety (based on a 50% reinforcement strength-loss criterion). As a final point of interest, Figure 6.14 gives a comparison of the predicted flexural capacity versus fire exposure time for a 120 mm thick, carbonate aggregate, steel reinforced concrete slab (with a room temperature concrete compressive strength of 35 MPa) with an equivalent slab reinforced with CFRP reinforcement.

Using the information presented in

Appendix D, the critical temperature of the CFRP is taken as 360 C, based on a 50% strength loss criterion. Both slabs were designed for a 1 hour fire rating using the design chart of Figure 6.13, which resulted in a required concrete cover of 13 mm for the steel-reinforced slab and 40 mm for the FRP-reinforced slab. The cross-sectional area of CFRP reinforcement in the slab was chosen to obtain the same ultimate design flexural strength, calculated using the ISIS (2001b) guidelines, as the steel reinforced slab (which also satisfied crack-width serviceability requirements in this case). Details of the two slabs being compared are provided in Table 6.3. It is important to recognize that, because of the larger concrete cover required for the CFRP-reinforced slab, the amount of reinforcement needed to obtain the same ultimate flexural capacity as the steelreinforced slab is much larger. While the steel reinforced concrete slab displays a predicted structural fire endurance of approximately 68 minutes, the CFRP-reinforced slab has a predicted fire endurance in excess of 4 hours. The superior structural fire endurance of the CFRP-reinforced slab as compared with the steel-reinforced slab, for this specific case, is due to a combination of factors. The material resistance factor of 0.8 applied to CFRP in the ISIS design guidelines, as opposed to 0.85 for steel, results in a true strength for the CFRP reinforcement that is higher than for the steel. In

259



addition, the steel-reinforced slab is assumed to fail by concrete crushing after yielding of the steel reinforcement. No such yield behaviour occurs for the CFRP reinforcement (and the strains in the reinforcement at failure are far less than ultimate), which results in a greater total tensile force in the reinforcement prior to concrete crushing, a greater neutral axis depth at failure, and a larger ultimate moment. The observations in the above paragraph are significant in that they highlight a fundamental difference between the flexural design of FRP and steel-reinforced concrete slabs. Steel reinforced slabs fail by crushing of the concrete after yielding of the tensile reinforcement, such that a 50% reduction in the yield strength of the steel results in about a 50% reduction in the flexural capacity of the slab (hence the 50% strength-loss criterion used to define the critical temperature of steel reinforcement). FRP-reinforced slabs are generally designed to fail by concrete crushing (because tensile failure of FRP is sudden and catastrophic), and so the strains in FRP at failure can be significantly less than ultimate in many cases. Thus, definition of the critical temperature for FRP reinforcement in terms of a 50% strength reduction may not correspond to a 50% reduction in the flexural capacity of the slab. Obviously, the degree to which loss of FRP mechanical properties at high temperatures will influence the flexural capacity of FRP-reinforced concrete slabs depends on a variety of factors, and, because the model has not been validated against test data, it is premature to investigate this point further here. However, there is enormous potential for future work, both parametric and experimental, in this area.

6.4 Summary In this chapter, a numerical model was developed to predict the heat transfer and structural behaviour of steel or FRP-reinforced concrete slabs exposed to a standard fire. The heat transfer model was validated against results from fire endurance tests on slabs reported in the literature.

The structural (flexural capacity) model represents a simplified analysis for the

purposes of illustration only, and has yet to be validated, although results obtained with the

260



structural model raised some very interesting observations. Parametric studies using the model indicated that the primary factors to consider in the design of FRP-reinforced concrete slabs for fire are the cover to the reinforcement and the critical temperature of the reinforcement. Two secondary factors are the slab thickness and the aggregate type, and since slab thickness is only a factor for slabs greater than 120 mm thick at concrete covers greater than 60 mm, most slabs can be designed on the basis of a 120 mm thick slab. With these variables in mind, the numerical model was used to develop design charts for FRP-reinforced concrete slabs. A complete set of design charts is presented in Appendix B.

261


CHAPTER 6: Numerical Models 2 – Slabs

Table 6.1: Details of the analyses conducted during parametric studies Relevant Figures

Section

Factor Examined

Range

6.2.1: Slab Thickness

Slab thickness (mm), Concrete cover (mm)

tslab = 0-200, tcover = 20-60

6.7

Concrete cover (mm), Aggregate type*

tcover = 0-70

6.8

6.2.1: Cover Thickness and Aggregate Type 6.2.1: Reinforcement Critical Temperature 6.2.1: Moisture Content

Critical Temperature (°C), Concrete cover (mm) Concrete Moisture Content (%), Concrete cover (mm)

tins = 0-60, tcover = 20-60 φ = 0-20 tcover = 20-60 1. Steel 2. CFRP 6.2.2: Type of + Type of reinforcement 3. GFRP Reinforcement 4. AFRP 6.3: Consequences for 1. Steel rebar ¥ Equivalent slab designs Design 2. CFRP rebar * Aggregate type was taken as carbonate, siliceous, or expanded shale

6.9 6.10 6.11 6.13

Table 6.2: Details of the 4 slabs analyzed to produce Figure 6.11 Parameter Slab Thickness (mm) Concrete Cover (mm) Aggregate Type

Steel

Slabs used in the Analysis GFRP* CFRP*

AFRP*

200

200

200

200

30

30

30

30

Carbonate

Carbonate

Carbonate

Carbonate

Concrete RT 35 35 35 35 Strength (MPa) Moisture Content 5 5 5 5 (%) Reinforcement ISOROD #10 ISOROD #10 NEFMAC A6 15M Rebar Name Carbon/Vinylester Glass/Vinylester Aramid RT Reinforcement 1596 689 1300 fy = 400 Strength (MPa) RT Reinforcement 200 120 42 54 Modulus (GPa) Reinforcement C/S 2000 740 4229 2115 Area (mm2) Structural Fire 185 90 115 97 Endurance (min) Reinforcement N/A 1965 5600 4320 Area (mm2)† * FRP properties were taken from ISIS (2001b) † Redesigned on the basis of the ISIS (2001b) permissible crack width criterion

262



Table 6.3: Details of the 2 slabs analyzed to produce Figure 6.13 Slabs used in the Analysis Steel CFRP* Slab Thickness (mm) 120 120 Concrete Cover (mm) 13 40 Aggregate Type Carbonate Carbonate Concrete RT Strength MPa) 35 35 Moisture Content (%) 5 5 Reinforcement Name 10M Rebar ISOROD #10 Carbon/Vinylester RT Reinforcement Strength (MPa) fy = 400 1596 RT Reinforcement Modulus (GPa) 200 120 Reinforcement C/S Area (mm2) 1000 1390 Structural Fire Endurance (min) 68 > 240 * FRP properties were taken from ISIS (2001b) Parameter

263



START

READ FROM INPUT FILE Slab thickness, tslab Number of layers in the slab Type of aggregate (siliceous, carbonate, expanded shale) Duration of test Concrete moisture content, φ Type of reinforcement (steel, CFRP, GFRP, AFRP) Cover to reinforcing bar centroid, cbars Cross-sectional area of steel per metre width, As Concrete strength, f’c Tensile strength of reinforcement (if FRP), fcom Tensile modulus of reinforcement (if FRP), Ecom

Discretize the slab cross-section into a number of layer elements (see Figure 6.1)

Stability Criteria Calculate minimum time step for stability using Equations 6.9-6.11 Smallest time step value governs, ∆t


Allocate and initialize Element moisture volumes initialized using Equations 6.4 and 6.7.

Set the time-step number, i=1

Calculate: Fire temperature Equation 5.1 Calculate: Fire-concrete interface temperature Figure 5.1f Equations 6.1 and 6.5

1

2


264



1

2

Set element counter, m=2

Calculate: Temperature of concrete element Figure 5.1f Equations 6.2 and 6.8


NO



YES Calculate: Temperature of the centerline concrete element Figure 5.1f Equations 6.3 and 6.6

NO

Check if the Temperature @ the unexposed face < limit

YES Check if the temperature is known at all time steps

NO

YES OUTPUT Data files for temperature related output

TO FLEXURAL CAPACITY ANALYSIS

Figure 6.1 (part 2 of 2) Figure 6.1: Schematic showing the program logic for the heat transfer portion of QSFIRE

265



FROM HEAT TRANSFER ANALYSIS

Discretize the total fire duration into specific points where flexural capacity will be calculated Discretize the slab cross-section into a number of layer elements (see Figure 6.1)


Set the neutral axis depth just above the reinforcement

Calculate: The temperature of the reinforcement and the unexposed surface temperature Calculate: The failure strain of the extreme compression fibre (from thermomechanical subroutines) Calculate: The strain in the reinforcement based on strain compatibility (assuming the concrete crushes)

Check what type of reinforcement is being used Steel

FRP

Calculate: The stress in the bars at the current strain and temperature (thermomechanical subroutines)

Calculate: The stress in the bars at the current strain and temperature (thermomechanical subroutines)

1

2

3


266



1

2

3

Check if the bars have ruptured

YES Set: The strain in the FRP to its failure strain

NO

For the Reinforcement Calculate: Temperature and distance from N.A. Strain in layer from strain compatibility Strain in layer from thermal expansion Total strain to cause stress Stress in the reinforcement from thermomechanical subroutines Axial tensile force, T, and internal moment about the N.A. due to the reinforcement

Decrease: The N.A. depth

For all ConcreteElemental Layers Calculate: Temperature, area, and distance from N.A. Strain in layer from strain compatibility Strain in layer from thermal expansion Total strain to cause stress Stress in the layer from thermomechanical subroutines Axial force and internal moment about the N.A. for the element Add up force and moment contributions to get the total compressive force, C, and moment about the N.A.

NO

Check if T=C

YES OUTPUT Flexural capacity

END

Figure 6.2 (part 2 of 2) Figure 6.2: Schematic showing the program logic for the flexural capacity portion of QSFIRE

267



Surroundings (Ambient Air)

M ½ ∆x

Concrete

...

∆x

m+1

∆x

m

∆x

...

∆x

2

∆x

1

½ ∆x

Fire

Figure 6.3: Discretization of a concrete slab for the purposes of heat transfer analysis

Concrete Depth*:

o

Temperature ( C)

800

6.35 12.7 19.0 25.4 38.1 50.8 76.2 6.35 12.7 19.0 25.4 38.1 50.8 76.2

600 400 200 0

Points are Experimental Lines are Model Prediction

0

60

* in millimetres

120

180

240

300

Time (min)

Figure 6.4: QSFIRE temperature validation plot using carbonate aggregate slab test data from PCA (1968)

268



Concrete Depth*:

o

Temperature ( C)

800

6.35 12.7 19.0 24.5 38.1 50.8 76.2 6.35 12.7 19.0 24.5 38.1 50.8 76.2

600 400 200 0

Points are Experimental Lines are Model Prediction

0

60

120

180

240

300

Time (min)

* in millimetres

Figure 6.5: QSFIRE temperature validation plot using siliceous aggregate slab test data from PCA (1968)

600

o

Temperature ( C)

800

400 15 mm Depth - Model Prediction 30 mm Depth - Model Prediction 15 mm Cover - Vinyl Ester/Glass FRP 15 mm Cover - Polyethylene/Carbon/Glass FRP 30 mm Cover - Vinyl Ester/Carbon/Glass FRP

200

0

0

30

60

90

120

150

180

Time (min)

Figure 6.6: QSFIRE temperature validation plot using data from NEFCOM (1998)

Fire Endurance (min)

120 20 mm Cover 40 mm Cover 60 mm Cover

100 80

Lower limit on slab thickness due to CSA A23.3 thickness requirements

Lower limit on slab thickness due to heat transmission criterion

60 40 20 0

0

60

120

180

Slab Thickness (mm)

Figure 6.7: Effect of overall slab thickness on the fire endurance of FRP reinforced concrete slabs (Tcrit = 250°C, carbonate aggregate)

269




100 80 60 40 Carbonate Aggregate Siliceous Aggregate Expanded Shale Aggregate

20 0

0

20

40

60

Cover to Reinforcement (mm)

Figure 6.8: Effect of concrete cover thickness and aggregate type on the fire endurance of FRP reinforced concrete slabs (Tcrit = 250°C, tslab = 150 mm)


250 20 mm Cover 40 mm Cover 60 mm Cover

200 150 100 50 0

0

100

200

300

400

500

600

o

Reinforcement Critical Temperature ( C)

Figure 6.9: Effect of critical temperature of the reinforcement on the fire endurance of FRP reinforced concrete slabs (tslab = 150 mm, carbonate aggregate)


120 100 80 60 40 20 mm Cover 40 mm Cover 60 mm Cover

20 0

0

5

10

15

20

Volumetric Moisture Content (%)

Figure 6.10: Effect of concrete moisture content on the fire endurance of FRP reinforced concrete slabs (Tcrit = 250°C, tslab = 150 mm, carbonate aggregate)

270



Flexural Capacity (kN.m)

250 Steel Reinforcement CFRP Reinforcement GFRP Reinforcement AFRP Reinforcement Service Load

200 150 100 50 90 97

0

0

60

115

185

120

180

240

Time (min)

Figure 6.11: Effect of FRP type on the capacity of FRP-reinforced slabs designed based on ultimate strength (tslab = 200 mm, cover = 30 mm, carbonate aggregate)


250 Steel Reinforcement CFRP Reinforcement GFRP Reinforcement AFRP Reinforcement Service Load

200 150 100 50

165

0

0

60

120

185

233

180

240

Time (min)

Figure 6.12: Effect of FRP type on the capacity of FRP reinforced slabs designed based on permissible crack width (tslab = 200 mm, cover = 30 mm, carbonate aggregate) Critical Temperature = 100oC

Concrete Cover (mm)

120

200oC

100 80

300oC

60

400oC

40

500oC 600oC

20 0 0.0

0.5

1.0

1.5

2.0

Fire Endurance (hours)

Figure 6.13: Sample design chart for a 120 mm thick carbonate aggregate reinforced concrete slab

271




60 CFRP Reinforced (40 mm cover) Steel Reinforced (13 mm cover) Service Load

50 40 30 20 10 Steel Fire Resistance = 68 min

0

0

60

120

180

240

Time (min)

Figure 6.14: Comparison of flexural capacity during fire for a steel reinforced concrete slab and an equivalent slab reinforced with CFRP and designed using Figure 6.12 (tslab = 120 mm, carbonate aggregate)

272


CHAPTER 7: Conclusions and Recommendations

CHAPTER 7 CONCLUSIONS AND RECOMMENDATIONS 7.1 Summary The experimental and numerical studies presented in this thesis sought to examine the behaviour of FRP-reinforced concrete slabs and FRP-wrapped reinforced concrete columns exposed to fire. To achieve this objective, an exhaustive literature review was conducted; numerical models were developed, validated, and used to conduct parametric studies; and fullscale fire endurance tests were conducted on circular reinforced concrete columns that were wrapped with FRP and insulated with a unique two-part fire protection system. Because of time and financial constraints, only two of the six columns fabricated for this study have been tested to date. However, sufficient experimental data were obtained from these two tests to validate the numerical column models such that they could be used to conduct preliminary parametric studies. Slab models were validated against test data available in the literature. On the basis of the experimental and numerical parametric studies presented herein, simple fire-design recommendations for both FRP-wrapped reinforced concrete columns and FRP bar-reinforced concrete slabs were subsequently presented. Thus, both the primary and secondary objectives of this thesis, as outlined in section 1.5, have been achieved.

7.2 Conclusions A number of significant conclusions can be drawn from the experimental and numerical studies presented and discussed in this thesis. The key conclusions are: 1. The fire endurance tests showed that FRP-wrapped columns that are adequately insulated can achieve a 5-hour fire rating. 2. The fire endurance tests also showed that the VG insulation played a major role in protecting the FRP-wrapped column. The benefits of the EI were not as significant.

273



3. Based on comparison with test results, the numerical models can adequately predict temperatures during fire exposure in FRP-wrapped reinforced concrete columns and FRP bar-reinforced concrete slabs.

Temperatures in the concrete and internal

reinforcement were generally predicted within ±50°C for columns, which is adequate for structural fire endurance prediction, and within ±20°C for slabs. 4. Also based on comparison with test results, the QCFIRE and QCFIRE Axial models tend to slightly overestimate the load carrying capacity of the fire exposed columns. If QCFIRE is to be used for specific design guidance, then the initial eccentricity of the axial load should be assumed in accordance with CSA A23.3 (CSA, 1994), which results in a more conservative prediction of load-capacity. 5. Based on parametric studies conducted using QCFIRE, the model suggests that very small thicknesses of supplemental fire insulation can dramatically improve the fire endurance of FRP-wrapped RC columns, particularly if fire resistance is defined in terms of the overall load-carrying capacity of the member. The model also suggests that the insulation thermal conductivity and thickness are key parameters to consider in the selection of insulation schemes. 6. Based on parametric studies conducted using QSFIRE, it appears that the critical temperature of the reinforcement and the concrete cover thickness are the two key parameters that influence the fire endurance of FRP bar-reinforced concrete slabs. Aggregate type and the overall slab thickness are secondary factors which should also be considered. If the fire endurance of FRP-reinforced concrete slabs is defined in terms of the temperature at which the reinforcement loses 50% of its room temperature tensile strength, then simple design charts can be used to give the required concrete cover to achieve particular fire endurances.

Concrete covers

obtained using this procedure are generally substantially greater for FRP-reinforced slabs as compared to those incorporating steel reinforcement.

274



A series of more specific conclusions can be drawn from the individual chapters of this thesis. The following conclusions were drawn on the basis of the literature review presented in Chapter 2 (along with information contained in Appendices C and D): Information on the fire and high temperature behaviour of FRPs and FRP-reinforced or wrapped concrete members is extremely scarce. In addition, the information available in the literature is case specific and cannot be applied to a wide range of FRP materials or applications. At elevated temperatures, all FRP materials currently available for structural engineering applications will experience a reduction of strength and stiffness. They may experience significant transverse thermal expansion leading to cracking or spalling of the concrete cover, or to the development of shear stresses in their adhesive layer. They may ignite. Upon ignition they may emit dense smoke and toxic gases. They may lose their bond with the substrate or surrounding concrete. All of these concerns have not, at present, been adequately studied or addressed by current design guidelines. The following conclusions were drawn on the basis of the experimental program presented in Chapters 3 and 4: Both of the FRP-wrapped and insulated columns tested herein performed extremely well under exposure to a standard fire. While the temperature of the FRP material reached its GTT within 1 hour of fire exposure during both tests, potentially resulting in loss of effectiveness of the FRP material, the overall behaviour of the columns was outstanding, and they were both able to carry their full service load for at least 5 hours of fire exposure. This would result in a 5-hour ULC-S101 (CAN/ULC, 1989) or ASTM E119 (ASTM, 2001) fire endurance rating, with design loads calculated in accordance with the ISIS Canada design guidelines (ISIS, 2001a).

275



During the preload phase of the fire endurance tests, both columns behaved similarly, and the FRP confinement model used in the QCFIRE analysis was able to accurately predict observed axial strains in the columns. For both fire endurance tests, expansion of the intumescent EI coating was complete within about 10 minutes of fire exposure, and the expanded EI char began to debond from the columns’ surface shortly thereafter.

Given the relatively short-lived

effectiveness of the EI coating, it is not clear whether it provided significant fire protection for the columns, nor is it clear if the EI is an essential component of the fire protection system used herein. The VG vermiculite-gypsum insulation remained essentially intact for the full duration of exposure for both fire endurance tests, and provided outstanding protection for both the FRP wrap and the substrate reinforced concrete. However, cracking of the VG during fire exposure is a potential concern, and further tests are required to determine installation procedures and VG thicknesses that will minimize or prevent cracking. During both fire endurance tests, the temperature at the FRP/concrete bond-line exceeded the GTT of the S-Epoxy within the first 60 minutes of fire exposure. Hence, the bond between the SCH FRP wrap and the substrate concrete may have been severely degraded after less than 1 hour of fire exposure. During both fire endurance tests, the temperatures in the concrete remained below 200ºC for the full duration of exposure. In addition, the temperatures in the internal steel reinforcement remained below 220ºC for the duration of the tests. Thus, both the concrete and the reinforcing steel remained essentially at full-strength for the full 5 hour duration of the test, and the columns retained their unwrapped strength for the full duration of exposure.

276



Using a greater thickness of EI resulted in a poorer fire performance for the intumescent char. If EI is to be used in future fire endurance tests, it is necessary to determine the optimal thickness to prolong its fire performance The following conclusions were drawn on the basis of the QCFIRE numerical models, which were developed to describe the fire behaviour of circular FRP-wrapped and insulated reinforced concrete columns, as presented in Chapter 5: The QCFIRE suite of analysis programs appear to be capable of satisfactorily predicting the heat transfer behaviour of unwrapped or FRP-wrapped and insulated reinforced concrete columns during exposure to a standard fire.

However, the

models should be used with extreme caution, since they do not account for several important factors (cracking of the VG insulation, for instance). Only small thicknesses (less than 50 mm) of VG insulation are required for the structural fire protection of FRP-wrapped concrete columns up to 4 hours of fire exposure, assuming that the wraps will be lost regardless of the amount of insulation. However, using a greater thickness of VG insulation could allow FRP wraps to maintain high post-fire residual strengths, which could potentially allow them to retain some effectiveness after a severe fire. It is difficult to state conclusively, because of the small number of tests conducted to date, whether the QCFIRE model is capable of predicting the axial strength of an FRP-wrapped and insulated reinforced concrete column during fire. The models suggest that it is unlikely that any FRP-wrapping materials currently available in the civil engineering industry will be able to perform satisfactorily for a significant period of time during fire unless they are provided with supplemental fire insulation. Depending on the failure criterion selected (FRP GTT, FRP ignition temperature, or load-capacity of the column), critical factors in the fire design of FRP-wrapped reinforced concrete columns for fire appear to be: the magnitude of the

277



applied load, the supplementary insulation’s thickness, the insulation’s thermal conductivity, and the glass transition and ignition temperatures of the polymer matrix. Aggregate type plays a secondary role. Care must be taken to ensure that the insulation will remain intact for the full duration of the desired fire rating, and that cracking of the insulation is minimized. The QCFIRE models suggest that it is possible to achieve outstanding fire behaviour for FRP-wrapped reinforced concrete columns, as long as the fire endurance is defined in terms of the overall load-carrying capacity of the member or the ignition temperature of the polymer matrix. The following conclusions with respect to the QCFIRE models are also of interest: The ability of QCFIRE to predict the temperature at the VG/FRP interface is an important factor in the success of the numerical model. While the model is capable of predicting the approximate temperature differential across the VG insulation, it does not perform as well at capturing the 100°C plateau observed in test data. A more detailed understanding of the thermophysical properties of VG insulation is required. Good agreement was observed between the experimental data and QCFIRE predictions for the temperature in the concrete at a depth of 50 mm, and for the spiral and reinforcing steel temperatures. At greater concrete depths, the model tends to under-predict temperatures, likely due to moisture migration in the concrete which is not accounted for in the model. The numerical model predicts that the temperature differential across the FRP sheet should be small at all times, while the test data suggests that the differential becomes greater than predicted at wrap temperatures above 100°C. This is likely due to changing thermal properties of the wrap material at temperatures above the matrix GTT.

278



The model does not predict the relatively rapid increases in temperature that were observed at the VG/FRP interface after 4 hours of fire exposure for test #1 and after 70 minutes for test #2. These increases are likely due to cracking of the VG which is not accounted for in the model. The model over-predicted the expansion of both columns for the full length of fire exposure.

However, the difference between predicted and measured axial

deformation is small in comparison to the overall length of the column. It is difficult to state conclusively whether the model is an accurate predictor of axial elongation for the FRP-wrapped and insulated case. The following conclusions were drawn on the basis of the QSFIRE numerical model, which was developed to describe the fire behaviour of FRP bar-reinforced concrete slabs, as presented in Chapter 6: QSFIRE is capable of adequately predicting the temperatures in both steel and FRP reinforced concrete slabs during fire. The type of reinforcement (steel versus FRP) does not significantly affect the distribution of temperatures in concrete slabs for fire exposures up to 4 hours. If the fire endurance of FRP bar-reinforced concrete slabs is defined in terms of their flexural capacity, preliminary numerical modelling indicates that FRP-reinforced slabs are capable of achieving similar fire endurance ratings to steel-reinforced slabs (with the same clear cover to the reinforcement).

This occurs because FRP-

reinforced slabs are generally designed using serviceability criteria, and are thus required to contain substantially more reinforcement than needed for strength requirements alone. In addition, material resistance factors for FRPs are generally less than for reinforcing steel. Because of the fundamental difference between the design criteria and failure modes of FRP-reinforced slabs versus steel-reinforced slabs, it is perhaps overly

279



conservative to define the fire endurance of FRP bar-reinforced members in terms of a 50% strength reduction of the reinforcement. Further work is required in this area.

7.3 Summary of Design Recommendations It is instructive to include in this chapter a summary of design recommendations contained in this thesis. For the fire-safety of FRP-wrapped reinforced concrete columns, it is recommended that: The strengthened service load on an upgraded column should not exceed the ultimate design strength of the unstrengthened column. Thus:

(φRn )existing (S DL

S LL )new

[7.1]

The current requirements suggested by ACI Committee 440 (ACI, 2002), and given by Equation E.14, should also be adequate and will give similar results in most cases. Supplemental fire insulation, applied to the exterior of the FRP-wrap, is required for most applications where fire ratings are to be satisfied. Analysis should be conducted to determine the approximate thickness of fire insulation required to achieve the desired fire endurance rating. Fire tests should be conducted to ensure that the fire insulation will remain in place during exposure to fire. For the fire-safety of FRP bar-reinforced concrete slabs, it is recommended that: Until fire endurance tests on loaded FRP-reinforced concrete slabs have been conducted and it has been demonstrated otherwise, FRP-reinforced slabs should be designed using the design charts of Appendix B. To use these charts accurately, it is required to conduct tests on the specific FRP reinforcement being used to determine its critical temperature. At present, the critical temperature for FRP bars should be defined in terms of a 50% room-temperature strength reduction criterion.

7.4 Recommendations for Further Research

280



Although a number of significant conclusions have been drawn from the work presented herein, further research is required before rational and realistic fire design recommendations for FRP-reinforced or wrapped concrete members can be suggested with confidence. Some of the most important recommendations for further research into the fire behaviour of FRP-wrapped reinforced concrete columns include: Further full-scale fire endurance testing of FRP-wrapped reinforced concrete columns is required, both to ensure repeatability of the experimental results from the two tests conducted to date, and to provide additional validation data for the QCFIRE models. The key tests that would contribute to increased confidence in the models’ predictive capabilities would involve: an unwrapped and uninsulated column (to provide a baseline fire endurance for the purposes of comparison), a wrapped and uninsulated column (to demonstrate the need for supplemental fire protection), and a column wrapped and insulated with VG only (to shed light on the specific benefits and effectiveness of the EI intumescent coating). Additionally, further tests should be conducted to investigate a variety of additional issues including: load intensity, combined axial and bending loads, fibre type, matrix type, etc. Both the overall fire endurance of the column and the ability of the model to predict internal temperatures depend on the prevention or minimization of crack formation in the VG insulation as the fire progresses. New installation approaches should be developed that promote the formation of fewer and smaller VG cracks. Further work is required, both analytical and experimental, to accurately model the behaviour in fire of the VG insulation. A more complete understanding of the variation in thermal properties with temperature, and the movement of moisture within the pores of the material and its effect on thermal properties, is essential for accurate modelling of VG-insulated members in fire.

281



While the EI coating was essentially ignored in the modelling of this thesis, it would be beneficial to include the effect of the EI if possible. Accurate modelling of intumescent coatings in complex thermal situations is currently not possible however, and it is not clear at present if the effort required for the analysis is reasonable in any case, given the number of unknowns involved. A comment is made several times in this thesis that certain poor agreement between predicted and observed temperatures can be attributed to the fact that the model does not account for moisture migration in the concrete. It is suggested that an attempt be made in the future to numerically account for moisture migration in the concrete (and in VG when present). This extension of the procedure used herein would also be of benefit to researchers investigating the fire behaviour of conventionally reinforced concrete structures, and steel structures insulated with VG. With regard to the fire behaviour of FRP bar-reinforced concrete slabs, the following are recommendations for further research: Fire endurance tests on FRP bar-reinforced slabs under sustained load are required to investigate the structural behaviour of these members during fire exposure. Test data are also required to validate the flexural capacity portion of the QSFIRE analysis. Further research is required to determine rational methodologies for assigning critical temperatures of FRP reinforcing bars. This work must account for the differences in design approaches and failure modes for FRP-reinforced slabs as compared with steel reinforced slabs, and will require that full-scale fire endurance tests on FRPreinforced slabs under sustained load be conducted. Further research is required to determine both the nature and magnitude of FRPconcrete bond degradation at high temperature (for both internal reinforcing and external plating applications), and the effects of bond degradation on the flexural capacity of FRP bar-reinforced concrete slabs.

282

In addition, tests to determine L.A. Bisby, Ph.D. Thesis, 2003


residual bond characteristics would be beneficial in the evaluation of fire-damaged FRP bar-reinforced concrete structures. The QSFIRE model should be refined and extended to account for both positive and negative bending, the effects of bond degradation, and considerations of span length. Once the model has been extended and validated against the results of fire endurance tests on loaded FRP-reinforced concrete slabs, further parametric studies should be conducted such that a series of straightforward and rational design guidelines (based on load-carrying capacity) can be suggested. While the above recommendations represent extensions of the work presented in this thesis, the following are general recommendations for further research into the fire behaviour of FRPs and FRP-reinforced concrete: A more complete understanding of the thermal and mechanical properties of FRP materials at high temperature is required. The mechanical and thermal behaviour of FRP materials currently available in industry must be accurately ascertained, such that experimental and parametric numerical studies can be executed with confidence and accuracy. In addition, the residual (post-fire) mechanical properties of FRP materials require investigation, such that fire damaged FRP-reinforced or wrapped concrete members can be evaluated and repaired if necessary. The development of standard test methods for FRP materials are required both at room and high temperatures, with both static and dynamic loading and temperature regimes, such that a database of test results on the high-temperature behaviour of FRP materials can be formulated and expanded on an international scale. To date, virtually no consideration has been given to the potential environmental effects of FRP combustion and decomposition during a building fire. The potential for the generation of significant amounts of dense smoke and toxic gas is a significant concern when FRPs are used in buildings. Further research is required to

283



characterize the nature and amount of gaseous FRP decomposition products produced in fire, such that environmental threats to building occupants can be addressed. During the course of this thesis, it was noted by the author that detailed information on the variation of several thermally important properties (thermal conductivity, specific heat, density) with temperature is extremely scarce for most currently available fire insulation materials. In the near future, given the current shift toward objective-based design codes, it is likely that numerical fire modelling will become a much more important research tool for fire protection engineers.

Research is

required to develop test methods to determine these thermally significant properties at a variety of temperatures. Once these test methods are in place, a significant research effort is required to obtain and compile thermal data for the wide variety of insulating materials currently available in industry. Finally, in Section 5.3.3, the reader was introduced to a new class FRP of materials called geocomposites.

These emerging materials, which are highly resistant to

elevated temperatures but which are not readily available for infrastructure applications at this time, may well represent the future of FRPs for fire-safe design. A significant research effort is required to address the suitability of geocomposites for concrete reinforcing applications and for exposure to fire.

284


REFERENCES

REFERENCES Abrams, M.S. 1977. Performance of Concrete Structures Exposed to Fire. Research and Development Bulletin RD060.01D. Portland Cement Association, Skokie, Illinois, 7 pp. Abrams, M.S. and Gustaferro, A.H. 1968. Fire Endurance of Concrete Slabs as Influenced by Thickness, Aggregate Type, and Moisture. Research Bulletin 223, Research and Development Laboratories of the Portland Cement Association, 10(2), pp. 9-24. ACI 2002. ACI 440.2R-02: Guide for the design and construction of externally bonded FRP systems for strengthening concrete structures. American Concrete Institute, Farmington Hills, MI. ACI 2001. ACI 440.1R-01: Guide for the design and construction Concrete Reinforced with FRP Bars. American Concrete Institute, Farmington Hills, MI.

ACI 1995. Building Code Requirements for Structural Concrete (ACI 318-95). American Concrete Institute, Farmington Hills, MI. Adimi, R., Rahman, H., Benmokrane, B., and Kobayashi, K. 1998. Effect of Temperature and Loading Frequency on the Fatigue Life of a CFRP Bar in Concrete. Proceedings of the Second International Conference on Composites in Infrastructure (ICCI-98), Tucson, Arizona, Vol. 2, pp. 203-210. Ahmad, S.H. 1981. Properties of concrete under Static and Dynamic loads. PhD Thesis, University of Illinois at Chicago, Ill. Ahmad, S.H., and Shah, S.P. 1982. Stress-strain curves of concrete confined by spiral reinforcements. Journal of the American Concrete Institute, 79(6), pp. 484-490. Ahmad, S.H., Khaloo, H., and Irshaid, A. 1991. Behaviour of Concrete Spirally Confined by Fiberglass Filaments. Magazine of Concrete Research, 43(156), pp. 143-148. Aiello, M.A., Focacci, F., Huang, P.C., and Nanni, A. 1999. Cracking in Fibre Reinforced Polymer Reinforced Concrete under Thermal Loads. In Fibre Reinforced Polymer Reinforcement for Reinforced Concrete Structures. Edited by A. Nanni, C.W. Dolan, and S.H. Rizkalla. American Concrete Institute, Detroit, Michigan, pp. 233-243. Ali, M.S. Mohamed 1997. Reinforced Concrete Beams with Steel Plates Glued to Their Tension Faces: Improving Shear Peeling Strength by Steel Plates Bonded to Their Sides. Research Report No. R 145, Department of Civil and Environmental Engineering, University of Adelaide, Australia. 29 pp. Alsayed, S.H., Al-Salloum, Y.A., and Almusalfam, T.H. 2000. Fibre-reinforced polymer repair materials – some facts. Proceedings of the Institution of Civil Engineers, Civil Engineering, 138, pp.131-134. Anderson, C.E., Ketchum, D.E., and Mountain, W.P. 1988. Thermal Conductivity of Intumescent Chars. Journal of Fire Sciences, Vol. 6, November/December, pp. 390-410.

285


REFERENCES

Ando, N., Matsukawa, H., Hattori, A., and Mashima, A. 1997. Experimental Studies on the Long-Term Tensile Properties of FRP Tendons. Proceedings of the Third International Symposium on Non-Metallic (FRP) Reinforcement for Concrete Structures (FRPRCS-3), Sapporo, Japan, Vol. 2, pp. 203-210. Apicella, F., & Imbrogno, M. 1999. Fire performance of CFRP-composites used for repairing and strengthening concrete. In Materials and Construction: Exploring the Connection. Proc. 5th ASCE Materials Engineering Congress, Cincinnati, OH, May, pp. 260-266. Arockiasamy, M., Shahawy, M., and Chidambaram, S. 1998. Environmental and long term studies on CFRP cables and CFRP reinforced concrete beams. In Proceedings of the First International Conference on the Durability of Composites for Construction. Edited by B. Benmokrane and H. Rahman, Sherbrooke, Quebec, pp. 599-610. ASCE 2001. Report Card on America’s Infrastructure. http://www.asce.org/reportcard/. ASTM 2001. Test Method E119-01: Standard Methods of Fire Test of Building Construction and Materials. American Society for Testing and Materials, West Conshohocken, PA. ASTM 2000. Test Method E1131-00: Test Method for Compositional Thermal Analysis by Thermogravimetry. American Society for Testing and Materials, West Conshohocken, PA. ASTM 1998b. Test Method E162-98: Standard Test Method for Surface Flammability of Materials Using a Radiant Heat Energy. American Society for Testing and Materials, West Conshohocken, PA. ASTM 1997a. Test Method E662-97: Standard Test Method for Specific Optical Density of Smoke Generated by Solid Materials. American Society for Testing and Materials, West Conshohocken, PA. ASTM 1997b. Test Method A370-97a: Standard Test Methods and Definitions for Mechanical Testing of Steel Products. American Society for Testing and Materials, West Conshohocken, PA. ASTM 1996. Test Method A615-96a: Standard Specification for Deformed and Plain Billet Steel Bars for Concrete Reinforcement. American Society for Testing and Materials, West Conshohocken, PA. ASTM 1995. Test Method E84-95: Standard Test Method for Surface Burning Characteristics of Building Materials. American Society for Testing and Materials, West Conshohocken, PA. ASTM 1990. Test Method C295-90: Standard Guide for Petrographic Analysis of Aggregates for Concrete. American Society for Testing and Materials, West Conshohocken, PA. ASTM 1976. Test Method D3039-76: Standard Test Method for Tensile Properties of FiberResin Composites. American Society for Testing and Materials, West Conshohocken, PA. Bakis, C.E. 1993. FRP Reinforcement: Materials and Manufacturing. In Fibre-ReinforcedPlastic (FRP) Reinforcements for Concrete Structures: Properties and Applications. Edited by A. Nanni. Elsevier Science Publishers B.V., pp. 13-58.

286


REFERENCES

Bakis, C.E., Bank, L.C., Brown, V.L., Cosenza, E., Davalos, J.F., Lesko, J.J., Machida, A., Rizkalla, S.H., and Triantafillou, T.C. 2002. Fiber-Reinforced Polymer Composites for Construction – State-of-the-Art Review. Journal of Composites for Construction, 6(2), pp. 73-87. Bakis, C.E., Al-Dulaijin, S.U., Nanni, A., Boothby, T.E., and Al-Zahrani, M.M. 1998a. Effect of Cyclic Loading on Bond Behaviour of GFRP Rods Embedded in Concrete Beams. Journal of Composites Technology and Research, 20(1), pp. 29-37. Bakis, C.E., Freimanis, A.J., Gremel, D., and Nanni, A. 1998b. Effect of Resin Material on Bond and Tensile Properties of Unconditioned and Conditioned FRP Reinforcement Rods. In Proceedings of the First International Conference on the Durability of Composites for Construction. Edited by B. Benmokrane and H. Rahman, Sherbrooke, Quebec, pp. 525-535. Balaguru, P., Kurtz, S., and Randolph, J. 1997. Geopolymer for Repair and Rehabilitation of Reinforced Concrete Beams. Géopolymère. Geopolymer Institute, Saint-Quentin, France, 5 pp. Ballinger, C., Maeda, T., and Hoshijima, T. 1993. Strengthening of Reinforced Concrete Chimneys, Columns, and Beams with Carbon Fibre Reinforced Plastics. In Fibre-ReinforcedPlastic Reinforcement for Concrete Structures: An International Symposium. Edited by A. Nanni and C.W. Dolan. American Concrete Institute, Detroit, Michigan, pp. 233-247. Balmer, G.G. 1944. Shear strength of concrete under high triaxial stress – computation of Mohr’s envelope as a curve. Rep. SP-23, Structural Research Laboratory, US Bureau of Reclamation. Bank, L.C. 1993. Properties of FRP Reinforcements for Concrete. In Fibre-Reinforced-Plastic (FRP) Reinforcements for Concrete Structures: Properties and Applications. Edited by A. Nanni. Elsevier Science Publishers B.V., pp. 59-86. Bank, L.C., Puterman, M., and Katz, A. 1998. The Effect of Material Degradation on Bond Properties of FRP Reinforcing Bars in Concrete. ACI Materials Journal, 95(3), pp. 232-243. Benmokrane, B. and Rahman, H., Eds. 1998. Durability of Fiber Reinforced Polymer (FRP) Composites for Construction. Proceedings of the First International Conference (CDCC’98), Sherbrooke, Quebec, 692 pp. Benmokrane, B., Rahman, H., Mukhopadhyaya, P., Masmoudi, R., Chekired, M., Nicole, J-F., and El-Safty, A. 2000. Use of fibre reinforced polymer reinforcement integrated with fibre optic sensors for concrete bridge deck slab construction. Canadian Journal of Civil Engineering, 27, pp. 928-940. Bhargava, A., and Griffin, G.J. 1999. A Two-Dimensional Model of Heat Transfer across a Fire Retardant Epoxy Coating Subjected to an Impinging Flame. Journal of Fire Sciences, Vol. 17, May/June, pp. 188-208. Bisby, L.A., Dent, A.J.S, and Green, M.F. 2003. A comparison of models for FRP-confined concrete. To be Submitted to the Journal of Composites for Construction. 9 pp.

287


REFERENCES

Bisby, L.A., Williams, B.K., Green, M.F., and Kodur, V.K.R 2002. Studies on the fire behaviour of FRP reinforced and/or strengthened concrete members. In Benmokrane, B. and E. ElSalakawy (Eds.), Durability of Composites for Construction, Montreal, PQ, May 29th-31st, pp. 405-417. Bisby, L.A., Green, M.F., and Kodur, V.K.R 2001. Fire Behaviour of FRP-Wrapped Reinforced Concrete Columns. In Forde, M.E. (Ed.), Structural Faults and Repair – 2001, London, England, July 4th-6th, CD-ROM. Blontrock, H., Taerwe, L., and Vandevelde, P. 2001. Fire Testing of Concrete Slabs Strengthened with Fibre Composite Laminates. In The Fifth Annual Symposium on FibreReinforced-Plastic Reinforcement for Concrete Structures (FRPRCS-5). Edited by C. Burgoyne, Thomas Telford, London, pp. 547-556. Blontrock, H., Taerwe, L., and Vandevelde, P. 2000. Fire Tests on Concrete Beams Strengthened with Fibre Composite Laminates. Third Ph.D. Symposium, Vienna, Austria, 10 pp. Blontrock, H., Taerwe, L., and Matthys, S. 1999. Properties of Fiber Reinforced Plastics at Elevated Temperatures with Regard to Fire Resistance of Reinforced Concrete Members. In Fibre Reinforced Polymer Reinforcement for Reinforced Concrete Structures. Edited by A. Nanni, C.W. Dolan, and S.H. Rizkalla. American Concrete Institute, Detroit, Michigan, pp. 43-54. Braimah, A. 2000. Long-term and fatigue behaviour of carbon fibre reinforced polymer prestressed concrete beams. Ph.D. Thesis, Department of Civil Engineering, Queen’s University, Kingston, Ontario. Bruggeling, A.S.G. 1992. External Prestressing – A State of the Art. External Prestressing, ACI SP-120, American Concrete Institute, Detroit, Michigan, pp.61-82. Buchanan, A.H. 2001. Structural Design for Fire Safety. John Wiley & Sons, Ltd. 421 pp. Buckmaster, J., Anderson, C., and Nachman, A. 1986. A Model for Intumescent Paints. International Journal of Engineering Sciences, 24(3), pp. 263-276. Burgoyne, C.J. 1999. Advanced Composites in Civil Engineering in Europe. Engineering International, 4, pp. reports 1-6

Structural

Burgoyne, C.J. 1993. Should FRP be bonded to concrete? In Fibre-Reinforced-Plastic Reinforcement for Concrete Structures: An International Symposium. Edited by A. Nanni and C.W. Dolan. American Concrete Institute, Detroit, Michigan, pp. 367-380. Butler, K. M. 1997. Physical Modeling of Intumescent Fire Retardant Polymers. Polymeric Foams: Science and Technology. In Proceedings of the American Chemical Society Symposium Series 669. Chapter 15. Edited by K.C. Khemani. American Chemical Society, Washington, DC, pp. 214-230. Butler, K.M., Baum, H.R., and Kashiwagi, T. 1995. Heat Transfer in an Intumescent Material Using a Three-Dimensional Lagrangian Model. In Proceedings of the International Conference on Fire Research and Engineering. September 10-15. Edited by D.P. Lund and E.A. Angell. Society of Fire Protection Engineers (SFPE), Boston, MA, pp. 261-266.

288


REFERENCES

Buyukozturk, O., Hearing, B., and Gunes, O. 1999. FRP strengthening and repair: where do we go from here? Structural Faults and Repair ’99. Proceedings of the 8th international conference on extending the life of bridges, civil, and building infrastructure, July 13-15th, London, England, pp. 12. Callery, K.M. 2000. Environmental Effects on the Behaviour of Wrapped Concrete Compression Members. M.Sc. Thesis, Department of Civil Engineering, Queen’s University at Kingston, Ontario, Canada, 216 pp. CAN/ULC 1989. Standard Methods of Fire Endurance Tests of Building Construction and Materials. CAN/ULC-S101-M89, Underwriters’ Laboratories of Canada, Scarborough, ON. Cengel, Y.A. 1998. Heat Transfer: A Practical Approach. WCB/McGraw-Hill, 1006 pp. Challal, O. and Shahawy, M. 2000. Performance of Fiber-Reinforced Polymer-Wrapped Reinforced Concrete Column under Combined Axial-Flexural Loading. ACI Structural Journal, 97(4), pp. 659-668. Chan, W.W.L. 1955. The ultimate strength and deformation of plastic hinges in reinforced concrete frameworks. Magazine of Concrete Research, 7(21), pp. 121-132. Clarke, J. 1993. Alternative Materials for Reinforcement and Prestressing of Concrete. Blackie Academic and Professional, Glasgow, U.K. Comeau, S. 2001. In Conversation with Saeed Mirza. McGill News. Summer 2001. McGill University, Montreal, Canada, pp. 10-11. Concrete Society, 2002. Design Guidance for Strengthening Concrete Structures using Fibre Composite Materials (DRAFT). Technical Report No. 55. Considère, A. 1903. Resistance a Compression du Beton Arme et du Beton Frette. Genie Civil. Cooke, G.M.E. 1988. An introduction to the mechanical properties of structural steel at elevated temperatures. Fire Safety Journal, Vol. 13, pp. 45-54. CSA, 2002. CAN/CSA-S806: Design and Construction of Building components with Fibre Reinforced Polymers. Canadian Standards Association, Ottawa, Ontario. CSA, 2001a. CAN/CSA-S6-00: Canadian Highway Bridge Design Code. Canadian Standards Association, Ottawa, Ontario. CSA, 1994. CAN/CSA A23.3-94: Design of Concrete Structures. Canadian Standards Association, Ottawa, Ontario. CSA, 1990. CAN/CSA A23.2-9C: Compressive Strength of Cylindrical Concrete Specimens. Canadian Standards Association, Ottawa, Ontario. CSCE, 2002. Critical Condition: Canada’s Infrastructure at the Crossroads. Canadian Society for Civil Engineering, Montreal, Quebec.

289


REFERENCES

Currier, J.B. and Dolan, C.W. 1995. Deformation of Prestressed Concrete Beams with FRP Tendons. Draft Final Report for U.S. Army Corps of Engineers, Waterways Experiment Station, Vicksburg, Missouri, 114 pp. Cusson, D. and Paultre, P. 1995. Stress-strain model for confined high strength concrete. Journal of Structural Engineering, ASCE, 121(3), pp. 468-477. Debaiky, A.S. 2000. Efficiency of FRP Wraps on Corrosion-Damaged Concrete Cylinders. In Advanced Composite Materials for Bridges and Structures (ACMBS-III). Edited by J. Humar and A.G. Razaqpur, Canadian Society for Civil Engineering, Montreal, PQ, pp. 679-686. Demers, M. and Neale, K.W. 1994. Strengthening of Concrete Columns with Unidirectional Composite Sheets. Developments in Short and Medium Span Bridge Engineering ’94, pp. 895-905. Demers, M., Hebert, D., Gauthier, M., Labossière, P., and Neale, K.W. 1996. The strengthening of structural concrete with an aramid woven fibre/epoxy resin composite. In Advanced Composite Materials for Bridges and Structures. Edited by M. El-Badry. Avantage Inc., Montreal, PQ, pp. 435-442. Deniaud, C. and Cheng, R.R. 2000. Shear rehabilitation of G-girder bridges in Alberta using fibre reinforced polymer sheets. Canadian Journal of Civil Engineering, Vol. 27, pp. 960971. Deuring, M. 1994. Brandversuche an nachtraglich verstarkten Tragern aus Beton. Research Report EMPA No. 148’795, Swiss Federal Laboratories for Materials Testing and Research, Dubendorf, Switzerland. Di Blasi, C., and Branca, C. 2001. Mathematical Model for the Nonsteady Decomposition of Intumescent Coatings. AIChE Journal, 47(10), pp. 2359-2370. Dimitrienko, Y.I. 1999. Thermomechanics of Composites under High Temperatures. Klewer Academic Publishers, London, 347 pp. Dimitrienko, Y.I. 1997a. Thermomechanics of Composite Materials and Structures under High Temperatures: 1. Materials. Composites Part A, 28A, pp. 453-461. Dimitrienko, Y.I. 1997b. Thermomechanics of Composite Materials and Structures under High Temperatures: 2. Structures. Composites Part A, 28A, pp. 463-471. Dusinberre, G.M. 1962. Heat-Transfer Calculations by Finite Differences. Textbook Company, 293 pp.

International

El-Hacha, R. 2000. Prestressed CFRP Sheets for Strengthening Concrete Beams at Room and Low Temperatures. Ph.D. Thesis, Department of Civil Engineering, Queen’s University, Kingston, Ontario. El-Hacha, R. 1997. Strengthening of concrete members with advanced composite materials. M.A.Sc. Thesis, Concordia University, Montreal, Quebec.

290


REFERENCES

El-Hacha, R., Green, M.F., and Wight, G. 2001. Concrete Beams Post-Strengthened with Prestressed Carbon Fibre Reinforced Polymer Sheets. In Structural Faults and Repair 2001. Edited by M.C. Forde, Engineering Technics Press, Edinburgh, CD-ROM. El-Hacha, R., Green, M.F., and Wight, G. 1999. CFRP wrapped concrete cylinders in severe environmental conditions. In Structural Faults and Repair 1999. Edited by M.C. Forde, Engineering Technics Press, Edinburgh, CD-ROM. Erki, M.A., and Rizkalla, S. 1993. FRP Reinforcement for Concrete Structures. Concrete International, 12(6), pp. 48-53. Fam, A. and Rizkalla, S. 2001. Confinement Model for Axially Loaded Concrete Confined by Circular Fiber-Reinforced Polymer Tubes. ACI Structural Journal, 98(4), pp. 451-461. Fam, A. and Rizkalla, S. 1999. Concrete –Filled FRP Tubes for Flexural and Axial Compression Members. In Advance Composite Materials in Bridges and Structures III. Edited by J. Humar and A.G. Razaqupur. Canadian Society for Civil Engineering, Montreal, Canada, pp. 315322. Fardis, M.N. and Khalili, H. 1982. FRP-encased concrete as a structural material. Magazine of Concrete Research, 34(121), pp. 191-202. Fardis, M.N. and Khalili, H. 1981. Concrete Encased in Fiberglass-Reinforced Plastic. ACI Journal, 78(38), pp. 440-446. Frangou, M., Pilakoutas, K., and Dritsos, S. 1995. Structural repair/strengthening of RC columns. Construction and Building Materials, 9(5), pp. 259-266. Freimanis, A.J., Bakis, C.E., Nanni, A., and Gremel, D. 1998. A Comparison of Pullout and Tensile Behaviors of FRP Reinforcement for Concrete. Proceedings of the Second International Conference on Composites in Infrastructure (ICCI-98), Tucson, Arizona, Vol. 2, pp. 52-65. Fujisaki, T., Nakatsuji, T., and Sugita, M. 1993. Research and Development of Grid Shaped FRP Reinforcement. In Fibre-Reinforced-Plastic Reinforcement for Concrete Structures: An International Symposium. Edited by A. Nanni and C.W. Dolan. American Concrete Institute, Detroit, Michigan, pp. 177-192. Furlong, R.W. 1967. Strength of steel-encased concrete beam-columns. Journal of the Structural Division, ASCE, 93(5), pp. 113-124. Gate, T.S. 1993. Effects of Elevated Temperature on the Viscoplastic Modeling of Graphite/Polymeric Composites. In High Temperature Environmental Effects on Polymeric Composites, ASTM STP 1174, C.E. Harris and T.S. Gates, Eds., American Society for Testing and Materials, Philadelphia, pp. 201-221. Gates, T.S. 1991. Effects of Elevated Temperature on the Viscoelastic Modeling of Graphite/Polymeric Composites. NASA Technical Memorandum 104160, National Aeronautics and Space Administration, Langley, U.S.A., 29 pp.

291


REFERENCES

Gentry, T.R. and Hussain, M. 1999. Thermal Compatibility of Concrete and Composite Reinforcements. Journal of Composites for Construction, ASCE, 3(2), pp. 82-86. Gentry, T.R. and Hudak, C.E. 1996. Thermal Compatibility of Plastic Composite Reinforcement and Concrete. In Advanced Composite Materials for Bridges and Structures. Edited by M. ElBadry. Avantage Inc., Montreal, PQ, pp. 149-156. Gergely, P. and Lutz, L.A. 1968. Maximum crack width in reinforced flexural members. Causes, Mechanism, and Control of Cracking in Concrete, SP-20, American Concrete Institute, Detroit, Michigan, pp. 87-117. Gerritse, A. 1996. Specific Properties of AFRP-Bars. In Advanced Composite Materials for Bridges and Structures 2. Edited by M. El-Badry. Avantage Inc., Montreal, PQ, pp. 75-83. Gluguru 2000. http://www.glueguru.com/HyEpSelecGuide.htm. Griffis, C.A., Masmura, R.A., & Chang, C.I. 1984. Thermal response of graphite epoxy composite subjected to rapid heating. In Environmental Effects on Composite Materials, Vol. 2, Technomic Publishing Company, Lancaster, Pennsylvania, 245-260. Hamilton III, H.R., and Dolan C.W. 2000. Durability of FRP reinforcements for concrete. Progress in Structural Engineering and Materials, 2(2), pp. 139-145. Harmathy, T.Z. 1967. A Comprehensive Creep Model. Journal of Basic Engineering, Transactions of the American Society for Mechanical Engineering, Vol. 89. Harmon, T.G., and Slattery, K.T. 1992. Advanced Composite Confinement of Concrete. In Advanced Composite Materials for Bridges and Structures 1. Edited by K.W. Neale and P.J. Labossière. Canadian Society for Civil Engineering, Montreal, PQ, pp. 149-156. Hassan, T., Abdelrahman, A., Tadros, G., and Rizkalla, S. 2000. Fibre reinforced polymer reinforcing bars for bridge decks. Canadian Journal of Civil Engineering, Vol. 27, pp. 839849. Hayes, M.D., Garcia, K., Verghese, N., and Lesko, J. 1998. The Effect of Moisture on the Fatigue Behavior of Glass/Vinyl Ester Composites. Proceedings of the Second International Conference on Composites in Infrastructure (ICCI-98), Tucson, Arizona, Vol. 1, pp. 1-12. Hazen, J.R., Basset, S., McConnell, V.P., Dawson, D., Morales, L., Reque, K., and Comar, G. 1998. Composites for Infrastructure: A Guide for Engineers. Ray Publishing, Wheat Ridge, Colorado, pp. 64-65. Hollaway, L.C. 1990. Polymer, fibre, and composite material properties and manufacturing techniques. In Polymers and Polymer Composites in Construction. Edited by L.C. Hollaway, Thomas Telford Ltd., London, England, pp. 5-29. Hoppel, C.P.R., Bogetti, T.A., and Gillespie, J.W. 1997. Design and Analysis of Composite Wraps for Concrete Columns. Journal of Reinforced Plastics and Composites, 16(7), pp. 588-601.

292


REFERENCES

Hosotani, M., Kawashima, K., and Hoshikuma, J. 1997. A study on confinement effect of concrete cylinders by carbon fiber sheets. In Non-Metallic (FRP) Reinforcement for Concrete Structures, Proceedings of the Third International Symposium, pp. 209-216. Howie, I., and Karbhari, V.M. 1995. Effect of tow-sheet composite wrap architecture on strengthening of concrete due to confinement. I: Experimental studies. Journal of Reinforced Plastics and Composites, 14(9), pp. 1008-1030. Ichimasu, H., Maruyama, M., Watanabe, H., and Hirose, T. 1993. RC slabs strengthened by carbon FRP plates: Part 2-Application. ACI SP-138: Fiber-Reinforced Plastic Reinforcement for Concrete Structures, American Concrete Institute, Detroit, Michigan, pp. 957-970. Incropera, F.P. and DeWitt, D.P. 2002. Introduction to Heat Transfer (4/e). John Wiley & Sons. 892 pp. ISIS 2001a. Strengthening reinforced concrete structures with externally bonded fiber reinforced polymers. Intelligent Sensing for Innovative Structures Canada, Winnipeg, Manitoba. ISIS 2001b. Reinforcing concrete structures with fiber reinforced polymers. Intelligent Sensing for Innovative Structures Canada, Winnipeg, Manitoba. Iyengar, K.T.S.R., Desayi, P., and Reddy, K.N. 1971. Flexure of reinforced concrete beams with confined compression zones. ACI Journal 68(9): 719-725. Iyengar, S.R., Desayi, P., and Reddy, K.N. 1970. Stress-strain characteristics of concrete confined in steel binders. Magazine of Concrete Research, 22(72), pp. 173-187. Jaeger, L.G., Tadros, G., and Mufti, A.A. 1995. Balanced section, ductility, and deformability in concrete with FRP reinforcement. Research Report No. 2-1995, The Nova Scotia CADCAM Centre, 29 pp. Jones, R.M. 1975. Mechanics of Composite Materials. Scripta Book Company, Washington, D.C.. Karabinis, A.I., and Rousakis, T.C. 2001. Carbon FRP Confined Concrete Elements Under Axial Load. In FRP Composites in Civil Engineering. Edited by J.G. Teng, Elsevier Science Ltd., pp. 309-316. Karbhari, V.M. and Gao, Y. 1997. Composite Jacketed Concrete Under Uniaxial Compression – Verification of Simple Design Equations. Journal of Materials in Civil Engineering, ASCE, 9(4), pp. 185-192. Katz, A. 2000. Bond to Concrete of FRP Rebars after Cyclic Loading, Journal of Composites for Construction, ASCE, 4(3), pp. 103-161. Katz, A. 1999. Bond mechanism of FRP rebars to concrete. Materials and Structures, 32(224), pp. 761-768. Katz, A. 1998. Effect of Helical Wrapping on Fatigue Resistance of GFRP. Composites for Construction, ASCE, 2(3), pp. 121-125.

293

Journal of


REFERENCES

Katz, A., and Berman, N. 2000. Modeling the effect of high temperature on the bond of FRP reinforcing bars to concrete. Cement and Concrete Composites, 22, pp. 433-443. Katz, A., Berman, N., and Bank, L.C. 1998. Effect of Cyclic Loading and Elevated Temperature on the Bond Properties of FRP Rebars. In Durability of Fibre Reinforced Polymer (FRP) Composites for Construction. Edited by B. Benmokrane and H. Rahman. Avantage Inc., Sherbrooke, PQ, pp. 403-414. Katz, A., Berman, N., and Bank, L.C. 1999. Effect High Temperature on the Bond Strength of FRP Rebars. Journal of Composites for Construction, ASCE, 3(2), pp. 73-81. Kawashima, K., Hosotani, M., and Hoshikuma, J. 1997. A model for confinement effect for concrete cylinders confined by carbon fiber sheets. NCEER-INCEDE Workshop on Earthquake Engineering Frontiers of Transportation Facilities, NCEER, State University of New York, Buffalo, N.Y. Khoury, G.A. 2000. Effect of Fire on Concrete and Concrete Structures. Progress in Structural Engineering and Materials, 2(4) pp. 429-447. Knowles, R.B. and Park, R. 1969. Strength of Concrete Filled Tubular Columns. Journal of Structural Engineering, ASCE, 95(12), pp. 2565-2587. Kodur, V. R.; Harmathy, T. Z. 2002. Properties of building materials. SFPE Handbook of Fire Protection Engineering, 3rd edition, P.J. DiNenno, National Fire Protection Association, Quincy, MA, pp. 1.155-1.181 Kodur, V.K.R. 1999. Fire Resistance Requirements for FRP Structural Members. Proceedings of the Annual Conference of the Canadian Society for Civil Engineering, Regina, Saskatchewan, pp. 83-95. Kodur, V.K.R., and Baingo, D. 1998. Fire Resistance of FRP Reinforced Concrete Slabs. IRC Internal Report No. 758. National Research Council of Canada, Ottawa, ON, 37 pp. Kodur, V.K.R. and Lie, T.T. 1997. Evaluation of the fire resistance of rectangular steel columns filled with fibre-reinforced concrete. Canadian Journal of Civil Engineering 24(3), pp. 339349. Kotsovos, M.D. and Perry, S.H. 1986. Behaviour of concrete subjected to passive confinement. Materials and Structures, RILEM, 19(12), pp. 259-264. Kumahara, S., Masuda, Y., Tanano, H., and Shimizu, A. 1993. Tensile Strength of Continuous Fibre Bar Under High Temperature. In Fibre-Reinforced-Plastic Reinforcement for Concrete Structures: An International Symposium. Edited by A. Nanni and C.W. Dolan. American Concrete Institute, Detroit, Michigan, pp. 731-742. Kurtz, S., and Balaguru, P. 2001. Comparison of Inorganic and Organic Matrices for Strengthening of RC beams with Carbon Sheets. Journal of Structural Engineering, ASCE, 127(1), pp. 35-42. La Tegola, A., and Manni, O. 1999. Experimental Investigation on Concrete Confined by Fiber Reinforced Polymer and Comparison with Theoretical Model. In Fibre Reinforced Polymer

294


REFERENCES

Reinforcement for Reinforced Concrete Structures. Edited by A. Nanni, C.W. Dolan, and S.H. Rizkalla. American Concrete Institute, Detroit, Michigan, pp. 243-253. Labossière, P., Neale, K.W., Rochette, P., Demers, M., Lamonthe, P., Lapierre, P., and Desgagne, G. 2000. Fibre reinforced polymer strengthening of the Sainte-Emile-de-l’Energie bridge: design, instrumentation, and field testing. Canadian Journal of Civil Engineering, Vol. 27, pp. 916-927. Labossière, P., Neale, K.W., and Martel, S. 1997. Strengthening with composite materials: practical applications in Quebec. US-Canada Workshop on Bridge Engineering, Zurich, Switzerland, pp. 89-96 Lam, L., and Teng, J.G. 2001. Strength Models for Circular Concrete Columns Confined by FRP Composites. In The Fifth Annual Symposium on Fibre-Reinforced-Plastic Reinforcement for Concrete Structures (FRPRCS-5). Edited by C. Burgoyne, Thomas Telford, London, pp. 835844. Lavergne, S. and Labossière, P. 1997. Experimental study of concrete columns confined by a composite jacket under combined axial and flexural loads. Proceedings of the Annual Conference of the Canadian Society for Civil Engineering, May 27-30th, Sherbrooke, Quebec, pp. 11-19. Lee, C., Bonacci, J.F., Thomas, M.D.A., Maalej, M., Khajehpour, S., Hearn, N., Pantazopolou, S., and Sheikh, S. 2000. Accelerated corrosion and repair of reinforced concrete columns using carbon fibre reinforced polymer sheets. Canadian Journal of Civil Engineering, Vol. 27, pp. 941-948. Lie, T.T. 1994. Fire Resistance of Circular Steel Columns Filled with Bar-Reinforced Concrete. Journal of Structural Engineering, 120(4), pp. 1489-1509. Lie, T.T. 1992. Structural Fire Protection. American Society of Civil Engineers Manuals and Reports on Engineering Practice No. 78. ASCE, New York, NY. Lie, T.T. 1980. New Facility to Determine Fire Resistance of Columns. Canadian Journal of Civil Engineering, 7. pp. 551-558. Lie, T.T. 1972. Fire and Buildings. Applied Science Publishers Ltd., Barking, Essex, England. Lie, T.T. and Kodur, V.K.R. 1996. Fire Resistance of Steel Columns Filled with Bar-Reinforced Concrete. Journal of Structural Engineering, ASCE, 122(1), pp. 30-36. Lie, T.T. and Stringer, D.C. 1994. Calculation of the fire resistance of steel hollow structural section columns filled with plain concrete. Canadian Journal of Civil Engineering, 21, pp. 382-385. Lie, T.T. and Denham, E.M.A. 1993. Factors Affecting the Fire Resistance of Circular Hollow Steel Columns Filled with Bar-Reinforced Concrete. IRC Internal Report No. 651, National Research Council of Canada, Ottawa, ON, 9 pp. Lie, T.T. and Irwin, R.J. 1993. Method to Calculate the Fire Resistance of Reinforced Concrete Columns with Rectangular Cross-Section. ACI Structural Journal, 90(1), pp. 52-60.

295


REFERENCES

Lie, T.T. and Celikkol, B. 1991. Method to calculate the fire resistance of circular reinforced concrete columns, ACI Materials Journal, 88(1), 84-91. Lie, T.T. and Woolerton, J.L. 1988. Fire resistance of reinforced concrete columns: Test results. IRC Internal Report No. 569, National Research Council of Canada, Ottawa, ON, 302 pp. Lie, T.T., Chabot, M., and Irwin, R.J. 1992. Fire Resistance of Circular Hollow Steel Sections Filled with Bar-Reinforced Concrete. IRC Internal Report No. 636, National Research Council of Canada, Ottawa, ON, 15 pp. Lie, T.T., Lin, T.D., Allen, D.E., and Abrams, M.S. 1984. Fire Resistance of Reinforced Concrete Columns. Division of Building Research, DBR Report No. 1167, National Research Council of Canada, Ottawa, ON, 19 pp. Liu, J. and Foster, S.J. 1998. Finite-Element Model for Confined Concrete Columns. Journal of Structural Engineering, ASCE, 124(9), pp. 1011-1017. Lyon, R.E., Sorathia, U., Balaguru, P.N., and Davidovits, J. 1996. Fire-Resistant Geopolymer Composites. Géopolymère. Geopolymer Institute, Saint-Quentin, France, 7 pp. Mallick, P.K. 1988. Fibre-Reinforced Composites: Materials, Manufacturing, and Design. Marcel Dekker Inc., New York, NY. Mander, J.B., Priestley, M.J.N., and Park, R. 1988. Theoretical Stress-Strain Model for Confined Concrete. Journal of Structural Engineering, ASCE, 114(8), pp. 1804-1823. Manfredi, G., and Realfonzo, R. 2001. Models for concrete confined by fiber composites. In The Fifth Annual Symposium on Fibre-Reinforced-Plastic Reinforcement for Concrete Structures (FRPRCS-5). Edited by C. Burgoyne, Thomas Telford, London, pp. 865-874. Matthys, S. and Taerwe, L. 1998. Long-term behaviour of concrete slabs pre-tensioned with AFRP or prestressing steel. In Durability of Fibre Reinforced Polymer (FRP) Composites for Construction. Edited by B. Benmokrane and H. Rahman. Avantage Inc., Sherbrooke, PQ, pp. 95-106. Matthys, S., Taerwe, L., and Audenaert, K. 1999. Tests on Axially Loaded Concrete Columns Confined by Fiber Reinforced Polymer Sheet Wrapping. In Fibre Reinforced Polymer Reinforcement for Reinforced Concrete Structures. Edited by A. Nanni, C.W. Dolan, and S.H. Rizkalla. American Concrete Institute, Detroit, Michigan, pp. 217-229. Matthys, S., De Schutter, G., and Taerwe, L. 1996. Influence of Transverse Thermal Expansion of FRP Reinforcement on the Critical Concrete Cover. In Advanced Composite Materials for Bridges and Structures. Edited by M. El-Badry. Avantage Inc., Montreal, PQ, pp. 665-673. Mays, G.C. and Hutchinson, A.R. 1992. Adhesives in Civil Engineering. Cambridge University Press, Cambridge, England. Meier, U. 2000. Composite Materials in Bridge Repair. Applied Composite Materials, Klewer Academic Publishers, Netherlands, 7, pp. 75-94.

296


REFERENCES

Meier, U. 1994. Rehabilitation of Concrete Structures with the CFRP Sheet Bonding Technique. Proceedings of Advancing with Composites ’94: Materials and Technologies, Vol. 1, May 35th, Milan, Italy, pp. 169-181. Meier, U., Deuring, M., Meier, H., and Schwegler, G. 1992. Strengthening of structures with CFRP laminates: Research and applications in Switzerland. In Proceedings of the First International Conference on Advanced Composite Materials in Bridges and Structures. Edited by K.W. Neale, and P.J. Labossière, Canadian Society for Civil Engineering, Montreal, Quebec, pp. 243-251. McKay, K.S. and Erki, M-A. 1993. Flexural behaviour of concrete beams pretensioned with aramid fibre reinforced plastic tendons. Canadian Journal of Civil Engineering, 20(4), pp. 688-695. Milke, J.A. and Vizzini, A.J. 1990. Modeling and Evaluation of the Thermal Response of Fire Exposed Composites. Journal of Composites Technology Research, 13(3): 15-21. Mirmiran, A. and Shahawy, M. 1997. Behavior of Concrete Columns Confined by Fiber Composites. Journal of Structural Engineering, ASCE, 123(5), pp. 583-589. Mirmiran, A., Shahawy, M., El Khoury, C., and Naguib, W. 2000. Large Beam-Column Tests on Concrete-Filled Composite Tubes. ACI Structural Journal, 97(2), pp. 268-276. Mirmiran, A., Shahawy, M., and Saaman, M. 1999. Strength and Ductility of Hybrid FRPConcrete Beam-Columns. Journal of Structural Engineering, 125(10), pp. 1085-1093. Mirmiran, A., Shahawy, M., Saaman, M., El Echary, H., Mastrapa, J.C., and Pico, O. 1998. Effect of Column Parameters on FRP-Confined Concrete. Journal of Composites for Construction, ASCE, 2(4), pp. 175-185. Miyauchi, K., Nishibayashi, S., and Inoue, S. 1997. Estimation of Strengthening Effects with Carbon Fiber Sheet for Concrete Column. In Non-Metallic (FRP) Reinforcement for Concrete Structures, Proceedings of the Third International Symposium, pp. 217-224. Monti, G. and Spolestra, M.R. 2000. Fiber-Section Analysis of RC Bridge Piers Retrofitted with FRP Jackets. Unpublished. Personal communication with the authors. Monti, G., Nisticó, N., and Santini, S. 2001. Design of FRP jackets for upgrade of circular bridge piers. Journal of Composites for Construction, ASCE, 5(2), pp. 94-101. Morales, S.M., Nilson, A.H., and Slate, F.O. 1982. Spiral-reinforced high-strength concrete columns. Report No. 82-10, Department of Structural Engineering, Cornell University, Ithaca, New York. Munley, E. and Dolan, C.W. 2001. What is the Value of it all? In The Fifth Annual Symposium on Fibre-Reinforced-Plastic Reinforcement for Concrete Structures (FRPRCS-5). Edited by C. Burgoyne, Thomas Telford, London, pp. 585-592. Nakagawa, H., Kobayashi, M., Suenaga, T., Ouchi, T., Watanabe, S., and Satoyama, K. 1993. Application of Three-Dimensional Fabric Reinforced Concrete to Building Panels. In FibreReinforced-Plastic Reinforcement for Concrete Structures: An International Symposium.

297


REFERENCES

Edited by A. Nanni and C.W. Dolan. American Concrete Institute, Detroit, Michigan, pp. 211-232. Nanni, A. 1997. CFRP strengthening. Concrete International, 19(6), pp. 19-23. Nanni, A. and Bradford, N.M. 1995. FRP jacketed concrete under uniaxial compression. Construction and Building Materials, 9(2), pp. 115-124. Nanni, A., Nenninger, J., Ash, K., and Liu, J. 1997. Experimental Bond Behavior of Hybrid Rods for Concrete Reinforcement. Structural Engineering and Mechanics, 5(4), pp. 339-354. Neale, K.W. 2000. FRPs for structural rehabilitation: a survey of recent progress. Progress in Structural Engineering and Materials, 2(2), pp. 133-138. Neale, K.W. and Labossière, P. 1992. Proceedings of the First International Conference on Advanced Composite Materials in Bridges and Structures, Sherbrooke, Quebec. Neale, K.W. and Labossière, P. 1991. Material Properties of Fibre-Reinforced Plastics. In Advanced Composite Materials with Application to Bridges. Edited by A.A. Mufti, M-A. Erki, and L.G. Jaeger. Canadian Society of Civil Engineering, Montreal, PQ, pp. 21-69. NEFCOM Corporation 1998. Technical Leaflet 3: Fire resistance of concrete slabs reinforced by NEFMAC, NEFCOM Corp., Tokyo, Japan, 11 pp. Nelson, G.L, (1995), Fire and Polymers: an Overview. In Fire and Polymers II. ACS Symposium Series 599, American Chemical Society, Washington, DC, pp. 1-26. Newhook, J., Ghali, A., and Tadros, G. 2000. Concrete Flexural Members Reinforced with FRP: Design for Serviceability and Deformability. Submitted for Publication, 31 pp. Newman, K., and Newman, J.B. 1972. Failure theories and design criteria for plain concrete. Proceedings, International Civil Engineering Materials Conference on Structure, Solid Mechanics and Engineering Design (Southampton, 1969), Wiley Interscience, New York, Part 2, pp. 963-995. NRC 1996. Objective-Based Codes: A new Approach for Canada. Institute for Research in Construction, National Research Council of Canada, Ottawa, 7 pp. NRC 1995. National Building Code of Canada 1995. National Research Council of Canada, Ottawa. Okamoto, T., Matsubara, S., Tanigaki, M., and Hasuo, K. 1993. Practical Application and Performance of PPC Beams Reinforced with Braided FRP Bars. In Fibre-Reinforced-Plastic Reinforcement for Concrete Structures: An International Symposium. Edited by A. Nanni and C.W. Dolan. American Concrete Institute, Detroit, Michigan, pp. 875-894. Pantazopoulou, S.J. and Mills, R.H. 1995. Microstructural aspects of the mechanical response of plain concrete, ACI Materials Journal, 92(6), pp. 605-616.

298


REFERENCES

Parent, S. and Labossière, P. 2000. Finite element analysis of reinforced concrete columns confined with composite materials. Canadian Journal of Civil Engineering, Vol. 27, pp. 400411. PCA 1968. Fire Endurance of Concrete Slabs as Influenced by Thickness, Aggregate Type, and Moisture. Research Department Bulletin 223. Portland Cement Association, 24 pp. Picher, F. 1995. Confinement de cylindres en beton par des composites carbone-epoxie unidirectionnelles. MS Thesis, University of Sherbrooke, Sherbrooke, Canada. Pilakoutas, K. and Mortazavi, A.A. 1997. Ductility Through External Lateral Confinement of RC Members with FRP. In Non-Metallic (FRP) Reinforcement for Concrete Structures, Proceedings of the Third International Symposium, pp. 225-232. Plecnik, J.M., Fogarty, J.H., and Kurfees, J.R. 1986. Behaviour of Epoxy Repaired Beams under Fire. Journal of Structural Engineering, 112(4): 906-922. Plevris, N., and Triantafillou, T.C. 1994. Time-dependent behaviour of RC members strengthened with FRP laminates. Journal of Structural Engineering, 120(3), pp. 1016-1042. Popovics, S. 1973. Numerical approach to the complete stress-strain relation for concrete. Cement and Concrete Research, 3(5), pp. 583-599. Priestley, M.J.N., Seible, F., Xiao, Y., and Verma, R. 1994. Steel Jacket Retrofit of Squat RC Bridge Columns for Enhanced Shear Strength - Part 1 - Theoretical Considerations and Test Design. ACI Structural Journal, 91(4), pp.394-405. Purba, B.K., and Mufti, A.A. 1999. Investigation of the behavior of circular concrete columns reinforced with carbon fiber reinforced polymer (CFRP) jackets. Canadian Journal of Civil Engineering, 26, pp. 590-596. Rahman, A.H. and Kingsley, C.Y. 1996. Fatigue Behaviour of a Fiber-Reinforced-Plastic Grid as Reinforcement for Concrete. Proceedings of the First International Conference on Composites in Infrastructure (ICCI-96), Tucson, Arizona, pp. 427-439. Rahman, A.H., Adimi, R., and Crimi, J. 1997. Fatigue Behaviour of a Carbon FRP Grid Encased in Concrete. Proceedings of the Third International Symposium on Non-Metallic (FRP) Reinforcement for Concrete Structures (FRPRCS-3), Japan Concrete Institute, Sapporo, Japan, Vol. 2, pp. 219-226. Rahman, A.H., Taylor, D.A., and Kingsley, C.Y. 1993. Evaluation of FRP as Reinforcement for Concrete Bridges. In Fibre-Reinforced-Plastic Reinforcement for Concrete Structures: An International Symposium. Edited by A. Nanni and C.W. Dolan. American Concrete Institute, Detroit, Michigan, pp. 71-86. Rehm, G., and Franke, L. 1979. Kunstharzgebundene Glasfaserstabe als Bewehrung im Betonbau Deutscher Ausschuss fur Stahlbeton. Heft 304. Richart, F.E., Brandtzaeg, A. and Brown, R.L. 1929. The failure of plain and spirally reinforced concrete in compression. Bulletin 190, University of Illinois, Engineering Experimental Station, Champagne, Illinois.

299


REFERENCES

Rochette, P. and Labossière, P.J. 2000. Axial testing of rectangular column models confined with composites. ASCE Journal of Composites for Construction, 4(3), pp. 129-136. Rodruigez, C.C., and Da Silva, M.G. 2001. Experimental Investigation of CFRP Reinforced Concrete Columns Under Uniaxial Cyclic Compression. In The Fifth Annual Symposium on Fibre-Reinforced-Plastic Reinforcement for Concrete Structures (FRPRCS-5). Edited by C. Burgoyne, Thomas Telford, London, pp. 782-792. Rostasy, F. 1997. On Durability of FRP in Aggressive Environments. Proceedings of the Third International Symposium on Non-Metallic (FRP) Reinforcement for Concrete Structures (FRPRCS-3), Japan Concrete Institute, Sapporo, Japan, Vol. 2, pp. 107-114. Rostasy, F. 1993. FRP Tensile Elements for Prestressed Concrete – State of the Art, Potentials and Limits. In International Symposium on Fiber-Reinforced-Plastic Reinforcement for Concrete Structures, ACI Publication SP-138. Edited by A. Nanni and C.W. Dolan, Detroit, Michigan, pp. 347-366. Rostasy, F. 1992. Fiber Composite Elements and Techniques as Non-Metallic Reinforcement of Concrete. Brite Project 4142/BREU – CT 91 0515, Evaluation of Potentials and Production Techniques of FRP, Technical Report Task 1. Rostasy, F.S. 1988. New Materials for Prestressing. Proceedings of the FIP Symposium, Sept. 49th , Jerusalem, Israel, pp. 147-159. Rostasy, F. S., Hankers, C., and Ranisch, E-H. 1992. Strengthening of R/C and P/C structures with bonded FRP plates. In Proceedings of the First International Conference on Advanced Composite Materials in Bridges and Structures. Edited by K.W. Neale and P.J. Labossière, Canadian Society for Civil Engineering, pp. 253-263. Saadatmanesh, H. and Tannous, F.E. 1999a. Relaxation, Creep, and Fatigue Behaviour of Carbon Fiber Reinforced Plastic Tendons. ACI Materials Journal, 96(2), pp. 143-153. Saadatmanesh, H. and Tannous, F.E. 1999b. Long-Term Behavior of Aramid Fiber Reinforced Plastic Tendons. ACI Materials Journal, 96(3), pp. 297-305. Saadatmanesh, H. and Ehsani, M.R. 1996. Proceedings of the First International Conference on Composites in Infrastructure (ICCI-96), Tucson, Arizona. Saadatmanesh, H., Ehsani, M.R., and Li, M.W. 1994. Strength and Ductility of Concrete Columns Externally Reinforced with Composite Straps. ACI Structural Journal, 91(4), pp. 434-447. Saafi, M., and Romine, P. 2002. Effect of Fire on Concrete Cylinders Confined with GFRP. In Durability of Fibre Reinforced Polymer (FRP) Composites for Construction. Edited by B. Benmokrane and E. El-Salakawy. Avantage Inc., Sherbrooke, PQ, pp. 512-521. Saafi, M., Toutanji, H.A., and Li, Z. 1999. Behaviour of Concrete Columns Confined with Fiber Reinforced Polymer Tubes. ACI Materials Journal, 96(4), pp. 500-509.

300


REFERENCES

Sakashita, M. 1997. Deflection of Continuous Fiber Reinforced Concrete Beams Subjected to Loaded Heating. Non-Metallic (FRP) Reinforcement for Concrete Structures, Japan Concrete Institute, Vol. 2, pp. 51-58. Samaan, M., Mirmiran, A., and Shahawy, M. 1998. Model of Concrete Confined by Fiber Composites. Journal of Structural Engineering, ASCE, 124(9), pp. 1025-1031. Santarosa, D., Filho, A.C., Beber, A.J., and Campagnolo, J.L. 2001. Concrete Columns Confined with CFRP Sheets. In FRP Composites in Civil Engineering. Edited by J.G. Teng, Elsevier Science Ltd., Hong Kong, pp. 301-308. Schneider, U. 1988. Concrete at High Temperatures – A General Review. Fire Safety Journal, Vol. 13, pp. 55-68. Schneider, U. 1986. Modelling of Concrete Behaviour at High Temperatures. In Design of Structures Against Fire. Edited by R.D. Anchor, H.L. Malhotra, and J.A. Purkiss, Elsevier Applied Science Publishers, pp. 53-70. Schundler 2002. Perlite and Vermiculite Plaster: Lightweight and Insulating Fire-Proofing. http://www.schundler.com/plaster.htm Schwartz, M.M. 1997. Composite Materials, Volume I: Properties, Non-Destructive Testing, and Repair. Prentice Hall, 432 pp. Scott, E.P. and Beck, V.B. 1992. Estimation of Thermal Properties in Epoxy Matrix/Carbon Fiber Composite Materials. Journal of Composite Materials, 26(1), pp. 132-149. Seki, H., Sekijima, K., and Konno, T. 1997. Test Method on Creep of Continuous Fiber Reinforcing Materials. Proceedings of the Third International Symposium on Non-Metallic (FRP) Reinforcement for Concrete Structures (FRPRCS-3), Japan Concrete Institute, Sapporo, Japan, Vol. 2, pp. 195-202. Sen, R., Shahawy, M., Rosas, J., and Sukumar, S. 1999. Durability of Aramid Fiber Reinforced Plastic Pretensioned Elements Under Tidal/Thermal Cycles. ACI Structural Journal, 96(1), pp. 95-104. Sen, R., Shahawy, M., Rosas, J., and Sukumar, S. 1998. Durability of Aramid Fiber Pretensioned Elements in a Marine Environment. ACI Structural Journal, 95(5), pp. 578-587. Sen, R., Mariscal, D., and Shahawy, M. 1993. Durability of Fiberglass Prestensioned Beams. ACI Structural Journal, 90(5), pp. 525-533. Shehata, E., Morphy, R., and Rizkalla, S. 2000. Fibre reinforced polymer shear reinforcement for concrete members: behaviour and design guidelines. Canadian Journal of Civil Engineering, Vol. 27, pp. 859-872. Sheikh, S.A. and Yao, G. 2002. Seismic Behaviour of Concrete Columns Confined with Steel and Fiber-Reinforced Polymers. ACI Structural Journal, 99(1), pp. 72-80. Shih, Y.C., Cheung, F.B., and Koo, J.H. 1998. Theoretical Modeling of Intumescent FireRetardant Materials. Journal of Fire Sciences, Vol. 16, January/February, pp. 46-71.

301


REFERENCES

Silverman, E.M. 1983. Elevated Temperature Testing for Comparison of Glass/Resin Composites. Polymer Composites, 4(4), pp. 214-218. Sorathia, U., Ohlemiller, T., Lyon, R., Riffle, J., and Schultz, N. 2001. Chapter 9: Effects of Fire. In Gap Analysis for Durability of Fibre Reinforced Polymer Composites in Civil Infrastructure. American Society of Civil Engineers, pp. 100-121. Sorathia, U., Dapp, T., and Beck, C. 1992. Fire Performance of Composites. Engineering, September, pp. 10-12.

Materials

Soudki, K.A. and Green, M.F. 1997. Freeze-Thaw Response of CFRP Wrapped Concrete. Concrete International, 19(7), pp. 64-67. Spiers, M.A. 1962. Technical Data on Fuel. British National Committee. World Power conference. R&R Clark, Ltd., Edinburgh, Scotland. Spoelstra, M. R., and Monti, G. 1999. FRP-confined concrete model. Journal of Composites for Construction, ASCE, 3(3), pp. 143–150. Steiner 1996. Strengthening of structures with CFRP strips. In Proceedings of the Second International Conference on Advanced Composite Materials in Bridges and Structures. Edited by M. El-Badry, Canadian Society for Civil Engineering, Montreal, Quebec, pp. 407-417. Sumida, A., Fujisaki, T., Watanabe, K., and Kato, T. 2001. Heat Resistance of Continuous Fiber Reinforced Plastic Rods. In The Fifth Annual Symposium on Fibre-Reinforced-Plastic Reinforcement for Concrete Structures (FRPRCS-5). Edited by C. Burgoyne, Thomas Telford, London, pp. 557-565. Suter, R. and Pinzelli, R. 2001. Confinement of Concrete Columns with FRP Sheets. In The Fifth Annual Symposium on Fibre-Reinforced-Plastic Reinforcement for Concrete Structures (FRPRCS-5). Edited by C. Burgoyne, Thomas Telford, London, pp. 791-802. Swamy, R.N., Jones, R., and Bloxham, J.W. 1987. Structural Behaviour of Reinforced Concrete Beams Strengthened by Epoxy-Bonded Steel Plates. Structural Engineer, 65(2), pp. 59-68. Tadros, G. 2000. Provisions for Using FRP in the Canadian Highway Bridge Design. Concrete International, 22(7), pp. 42-47. Taerwe, L. 1995. Non-Metallic (FRP) Reinforcement for Concrete Structures. Proceedings of the Second RILEM Symposium (FRPRCS-2), Ghent, Belgium, 714 pp. Taerwe, L., Matthys, S., Pilakoutas, K., and Guadagnini, M. 2001. European Activities on the use of FRP Reinforcement, fib Task Group 9.3 and the ConFibreCrete Network. In The Fifth Annual Symposium on Fibre-Reinforced-Plastic Reinforcement for Concrete Structures (FRPRCS-5). Edited by C. Burgoyne, Thomas Telford, London, pp. 3-15. Taly, N., and GangaRao, H.V.S. 1999. Guidelines for the design of concrete structures reinforced with FRP materials. 44th International SAMPE Symposium, May 23-27, pp. 1689-1696.

302


REFERENCES

Tanano, H., Masuda, Y., and Tomosawa, F. 1999. Characteristics and Performance Evaluation Methods of Continuous Fibre Bars – State of the Art Studies on Fire Properties and Durability of Continuous Fibre Reinforced Concrete in Japan. In Fibre Reinforced Polymer Reinforcement for Reinforced Concrete Structures. Edited by A. Nanni, C.W. Dolan, and S.H. Rizkalla. American Concrete Institute, Detroit, Michigan, pp. 523-531. Tanano H., Masuda Y., Sakashita M., Oono Y., Nonomura K., and Satake K. 1997. Tensile Properties at High Temperatures of Continuous Fiber Bars and Deflections of Continuous Fiber Reinforced Concrete Beams under High-Temperature Loading. Proceedings of the Third International Symposium on Non-Metallic (FRP) Reinforcement for Concrete Structures. Japan Concrete Institute, pp. 43-50. Tanano, H., Masuda, Y., Kage, T., Fukuyama, H., Nishida, I., and Hashimoto, T. 1995. Fire Resistance of Continuous Fibre Reinforced Concrete. In Non-metallic (FRP) Reinforcement for Concrete Structures. Edited by L. Taerwe. E&FN Spon, London, UK, pp. 368-375. Tannous, F.E. and Saadatmanesh, H. 1998. Environmental Effects on the Mechanical Properties of E-Glass FRP Rebars. ACI Materials Journal, 95(2), pp. 87-100. Theriault, M. and Neale, K.W. 2001. Simplified Design Equations for Reinforced Concrete Columns Confined with FRP Wraps. In FRP Composites in Civil Engineering. Edited by J.G. Teng, Elsevier Science Ltd., Hong Kong, pp. 741-748. Theriault, M. and Neale, K.W. 2000. Design equations for axially loaded reinforced concrete columns strengthened with fibre reinforced polymer wraps. Canadian Journal of Civil Engineering, Vol. 27, pp. 1011-1020. Theriault, M., Claude, S., and Neale, K.W. 2001. Effect of size and slenderness ratio on the behaviour of FRP-wrapped columns. In The Fifth Annual Symposium on Fibre-ReinforcedPlastic Reinforcement for Concrete Structures (FRPRCS-5). Edited by C. Burgoyne, Thomas Telford, London, pp. 765-771. Toutanji, H.A. 1999. Stress-Strain Characteristics of Concrete Columns Externally Confined with Advanced Fiber Composite Sheets. ACI Materials Journal, 96(3), pp. 397-404. Toutanji, H.A., and Saafi, M. 2001. Durability Performance of Concrete Columns Encased in PVC-FRP Composite Tubes. In The Fifth Annual Symposium on Fibre-Reinforced-Plastic Reinforcement for Concrete Structures (FRPRCS-5). Edited by C. Burgoyne, Thomas Telford, London, pp. 753-763. Tsai, S.W. and Hahn, H.T. 1980. Introduction to Composite Materials. Technomic Publishing Company, Westport, Connecticut, 457 pp. Type5 1999. Type 5 Fireproofing, http://www.type5.com/type5.html Uematsu, Y., Kitamura, T., and Ohtani, R. 1995. Delamination of a Carbon-Fibre-Reinforced Thermoplastic Polymer at High Temperatures. Composites Science and Technology 53, Elsevier Science, Northern Ireland, pp. 333-341.

303


REFERENCES

Uomoto, T. 2001. Durability considerations for FRP reinforcements. In The Fifth Annual Symposium on Fibre-Reinforced-Plastic Reinforcement for Concrete Structures (FRPRCS-5). Edited by C. Burgoyne, Thomas Telford, London, pp. 17-32. Uomoto, T. and Ohga, H. 1996. Performance of Fibre Reinforced Plastics for Concrete Reinforcement. In Advanced Composite Materials for Bridges and Structures. Edited by M. El-Badry. Avantage Inc., Montreal, PQ, pp. 125-132. Vintzileou, E. 2001. An empirical model for predicting the properties of concrete confined by means of composite materials. In The Fifth Annual Symposium on Fibre-Reinforced-Plastic Reinforcement for Concrete Structures (FRPRCS-5). Edited by C. Burgoyne, Thomas Telford, London, pp. 845-853. Walser, R. and Steiner, W. 1997. Strengthening a Bridge with Advanced Materials. Structural Engineering International, Journal of the International Association for Bridge and Structural Engineering (IABSE), No. 2, pp. 110-112. Wang, Y.C., Wong P.M.H. and Kodur V.K.R. 2003. Mechanical Properties of Fibre Reinforced Polymer Reinforcing Bars at Elevated Temperatures. Submitted to ASCE-SFPE Conference: Designing Structures for Fire, Baltimore, MD, October. Watanabe, K., Nakamura, H., Honda, Y., Toyoshima, M., Iso, M., Fujimaki, T., Kaneto, M., and Shirai, N. 1997. Confinement effect of FRP sheet on strength and ductility of concrete cylinders under uniaxial compression. In Non-Metallic (FRP) Reinforcement for Concrete Structures, Proceedings of the Third International Symposium, pp. 233-240. Watson, R.J. 2000. Verbal Communication. Third Conference on Advanced Composite Materials in Bridges and Structures, Ottawa, Canada. Williams, B.K., Bisby, L.A., Green, M.F., and Kodur, V.K.R. 2003. An Investigation of the Fire Behaviour of FRP-Strengthened Reinforced Concrete Beams. In Forde, M.E. (Ed.), Structural Faults and Repair – 2003, London, England, July 1st-3rd, CD-ROM Xiao, Y. and Wu, H. 2001. Concrete Stub Columns Confined by Various Types of FRP Jackets. In FRP Composites in Civil Engineering. Edited by J.G. Teng, Elsevier Science Ltd., Hong Kong, pp. 293-301. Xiao, Y. and Wu, H. 2000. Compressive Behavior of Concrete Confined by Carbon Fiber Composite Jackets. Journal of Materials in Civil Engineering, ASCE, 12(2), pp. 139-146. Yagi, K., Tanaka, T., and Otaguro, H. 1997. Ductility of Carbon Fibber Sheet for Repair and Retrofitting. Structural Faults and Repair ’99. Proceedings of the 8th international conference on extending the life of bridges, civil, and building infrastructure, July 13-15th, London, England, 9 pp.

304


APPENDIX E: Load Calculations for FRP-Wrapped Columns

APPENDIX E LOAD CALCULATIONS FOR FRP-WRAPPED CONCRETE COLUMNS This appendix presents detailed load calculations for the reinforced concrete columns tested in this thesis. For unwrapped reinforced concrete columns, loads have been calculated using CSA A23.3 (CSA, 1994) and ACI 318 (ACI, 1995). For calculations relating to FRPwrapped reinforced concrete columns, calculations have been performed in accordance with the ISIS Canada Design Guidelines (ISIS, 2001a) and the ACI 440.2R-02 design guide (ACI, 2002). Fire endurance test loads were calculated in accordance with ULC-S101 (CAN/ULC, 1989), which is equivalent to ASTM E119 (ASTM, 1998) All calculations have been performed in SI units. Design equations from American codes (ACI 318 and ACI 440.2R-02) have been performed using Canadian rebar designations and properties. The actual tested properties have been used where available in design calculations for all materials involved. Tested material properties are discussed in detail in Chapter 4 of this thesis and were as follows: Concrete Compressive Strength: f c,

39.5 MPa

Yield Strength of Longitudinal Reinforcing Steel: f y

456 MPa

Ultimate Tensile Strength of SCH FRP System: f frp ,ult Ultimate Strain of SCH FRP System: ε frp ,ult Elastic Modulus of SCH FRP System E frp

1510 MPa

1.64 %

90.2 GPa

E.1 Axial Load Capacity of the RC Column ACI 318-95 (Design Strength) The maximum axial design strength, φPn (max) , for a spirally reinforced concrete column according to ACI 318 is taken as (Cl. 10.3.5.1):

φPn (max)

0.85φ 0.85 f c' (Ag

354

Ast )

f y Ast

[E.1]



Where φ is equal to 0.75, Ag is the gross cross-sectional area of concrete, and Ast is the area of longitudinal reinforcing steel in compression.

For the columns tested herein this equation

becomes:

φPn (max)

(

0.85(0.75) 0.85(39.5)

(200)2

)

8(300)

456(8 300)

3289 kN ACI 318-95 (Predicted Strength) Omitting member reduction factors in the above equations for axial load capacity gives the predicted strength of the columns according to ACI 318. Thus:

φPn (max)

0.85 0.85 f c' (Ag

(

0.85 0.85(39.5) 4386 kN

Ast ) f y Ast

(200)2

)

8(300)

456(8 300)

CSA A23.3-94 (Design Strength) The maximum design axial load strength in the CSA A23.3 code for a spirally reinforced concrete column is taken as (Cl. 10.10.4):

Pr max

0.85 Pro

[E.2]

For the columns considered here we have:

Pro

φ f c' (Ag

1 c

Ast ) φ s f y Ast

[E.3]

Where φc is equal to 0.6 and φs is equal to 0.85. Thus:

Pr max

0.85

φ f c' (Ag

1 c

Ast ) φ s f y Ast

(

0.85 (0.85 0.0015(39.5))(0.6)(39.5) 2722 kN

(200)2

)

8(300)

0.85(456)(8 300)

CSA A23.3-94 (Predicted Strength) Omitting material reduction factors in the above equations for axial load capacity gives the predicted strength of the columns according to CSA A23.3. This becomes:

355



Pr max

0.85

f (Ag

Ast ) f y Ast

' 1 c

(

0.85 (0.85 0.0015(39.5))(39.5) 4149 kN

(200)2

)

8(300)

(456)(8

300)

E.2 Axial Load Capacity of FRP-Wrapped RC Column ACI 440.2R-02 (Design Strength) The FRP-confined design strength for the columns is calculated in accordance with Chapter 11 of ACI 440.2R-02. The wrap consists of a single layer of SCH FRP, which has the following material properties required for the design calculations: Ultimate strength of FRP: fcom = 1510 MPa FRP Modulus: Ecom = 90.2 GPa Thickness of the wrap: t w = 0.76 mm The effective confining pressure in the jacket at ultimate is calculated as follows, with notation modified for consistency within this thesis. The confinement reinforcement ratio, ρ f , is calculated as:

4 n tw d

ρf

4 1 0.76 400

0.0076

[E.4]

The effective ultimate strength of the FRP wrap is taken as the product of the ultimate strength and an environmental reduction coefficient, CE , as follows:

f fe

C E f com

0.85 1510 1283.5 MPa

[E.5]

where CE is equal to 0.85 for CFRP with an interior conditioned exposure. The confining pressure at ultimate can subsequently be determined as (with a

fl

ρ f f fe 2

a

1.0 0.0076 1283.5 2

= 1.0 for a circular column):

4.88 MPa

[E.6]

And the confined ultimate strength of the concrete is calculated using the Mander equation:

f cc,

f c,

2.25

1

7.94 f l f c,

356

2 fl f c,

1.25

65.1 MPa

[E.7]



The ultimate strength of the FRP-wrapped RC column can now be determined as:

0.85φ 0.85 f cc' (Ag

φPn (max)

Ast ) f y Ast

0.85 (0.75) 0.85 (65.1)

(

)

(200)2

8 (300)

456 (8 300 )

[E.8]

5049 kN ACI 440.2R-02 (Predicted Strength) Omitting reduction factors in the above expressions results in the following predicted strength for the FRP-wrapped columns according to ACI 440.2R-02:

ρf

4 n tw d

f fe

f com

fl

a

f cc,

f c,

4 1 0.76 400

1510 MPa

ρ f f fe

1.0 0.0076 1510 2

2

φPn (max)

0.0076

2.25

1

7.94 f l f c,

0.85 0.85 f cc' (Ag 0.85 0.85 (68.8)

2 fl f c,

5.74 MPa

1.25

Ast ) f y Ast

(

(200)2

68.8 MPa

)

8 (300)

456 (8 300 )

7054 kN ISIS Canada (Design Strength) According to the ISIS Canada design guidelines, the confining pressure at ultimate is calculated as follows:

fl

2 n φ frp f com t w d

2 1 0.75 1510 0.76 400

4.30 MPa

[E.9]

where φfrp is equal to 0.75 for CFRP with an interior conditioned exposure. The ISIS Canada guidelines specify a maximum effective confinement pressure as follows:

f l (max)

f c, 2

pc

1 ke

φc

39.5 1 0.6 2 1.0 1.0

357

7.8 MPa

[E.10]



where

pc

= 1.0 and k e = 1.0 for a round column. The volumetric confinement ratio is calculated

as:

2 fl φc f c,

w

2 4.30 0.6 39.5

0.3697

[E.11]

The confined ultimate strength of the concrete is determined from:

f cc,

f c, (1

pc

w

)

39.5 (1 1.0 0.3697 ) 53.1 MPa

[E.12]

Finally, the design strength of the FRP-wrapped column is determined from:

Pr max

0.85

φ f cc' (Ag

Ast ) φ s f y Ast

1 c

0.85 0.79 0.6 (53.1)

(

(200)2

)

8 (300)

0.85 456 (8 300)

[E.13]

3430 kN ISIS Canada (Predicted Strength) Omitting reduction factors in the above expressions results in the following predicted strength for the FRP-wrapped columns according to the ISIS Canada guidelines:

fl

w

f cc,

Pr max

2 n f com t w d 2 fl f c, f c, (1

0.85

2 1 1510 0.76 400

2 5.74 39.5

)

pc

w

' 1 cc

(A

f

g

5.74 MPa

0.2959 39.5 (1 1.0 0.2959 ) 50.3 MPa

Ast ) f y Ast

0.85 0.79 (50.3)

(

(200)2

)

8 (300)

456 (8 300)

5094 kN Summary of Results Table E.1 presents an overview of the various column strength calculations. Examination of the data in Table E.1 reveals that the ACI 440.2R-02 design guidelines predict a 54% increase in the axial load capacity of the column due to FRP-wrapping, while the ISIS Canada guidelines predict an increase of 26%. Although it is possible that a 54% increase in the load capacity

358



would be observed in tests of these columns, it is unlikely that such a large increase would be used in a design situation.

Furthermore, in Appendix C it was shown that the ACI 440

confinement model is relatively inaccurate and unconservative in comparison with other available confinement models for FRP-confined concrete. Hence, it was decided that the ISIS Canada ultimate load capacity for the columns tested herein was more realistic, and that the load for the fire endurance tests should be determined using the ISIS procedure.

E.3 Strengthening Limits The ACI 440.2R-02 design code (Clause 8.2) states that careful consideration should be given to determine reasonable strengthening limits. These limits must be imposed to guard against collapse of the structure should some compete failure of the FRP system occur due to unforeseen circumstances such as fire or vandalism. ACI 440 recommends that the existing strength of the structure should be sufficient to resist a level of load described by:

(φRn )existing (1.2

S DL

0.85 S LL )new

[E.14]

Where (φRn )existing is the existing strength of the member to be strengthened with FRP, S DL is the strengthened service dead load, and S LL is the strengthened service live load. If it is assumed that the factored design load is 50% due to live load and 50% due to dead load, and using the ACI 440 load factors of 1.4 and 1.7 for dead and live loads respectively, the maximum allowable confined strength, Pmax, can be calculated for the FRP-strengthened column as follows:

(φRn )existing (1.2 3289

1.2

S DL

0.5 Pmax 1.4

0.85 S LL )new 0.85

0.5 Pmax 1.7

Pmax

4847 kN

which, with a 1:1 dead-to-live load ratio, gives the following loads:

S LL

0.5 4847 1426 kN 1.7

and

359

S DL

0.5 4847 1731 kN 1.4



The total service load on the column becomes 1426 + 1731 = 3157 kN.

E.4 Serviceability To prevent radial cracking or yielding of internal reinforcing steel in the column under service load conditions, ACI 440.2R-02 (Clause 11.1.3) limits the stress in the concrete at service conditions to 0.65 f c,

25.2 MPa . If it is assumed that the vertical reinforcing steel and FRP-

wrap are ineffective, this concrete stress corresponds to an axial service load capacity of 3169 kN, which is larger than the service load quoted in Section E.3 above.

E.5 Structural Fire Endurance ACI 440.2R-02 (Clause 8.2.1) states that the existing strength of a concrete member with a fire endurance rating should be sufficient to carry the unfactored service loads acting on the structure for the duration of the required fire endurance. In other words, the effect of the FRPwrap should be ignored and the fire endurance of the existing structural member determined under the new (increased) load condition. For reinforced concrete columns, fire endurance is covered in ACI 216R. For circular reinforced concrete columns however, ACI 216R provides only approximate fire endurances based on the thickness of concrete cover to the primary reinforcement, and it is difficult to accurately estimate the fire endurance of a circular column under a specified load condition. Because of this difficulty, the fire endurance of the unwrapped concrete columns was evaluated using the computer program QCFIRE. Developed and validated in Chapter 5 of this thesis, QCFIRE has been shown to accurately predict both temperature distributions and load capacities for circular concrete columns under exposure to the standard fire. Analysis using QCFIRE indicated that for a 4-hour fire rating, the unwrapped column should be able to safely carry a service load of approximately 2942 kN.

360



E.6 Sustained Load for Fire Endurance Tests As discussed previously, it was felt that the ultimate load capacity for the FRP-wrapped columns as predicted by the ACI design procedure was unconservative. As such, the service load for the fire endurance tests was calculated based on the ISIS Canada design procedures. The ultimate load capacity using the ISIS procedure is calculated above as 3430 kN. If a dead-to-live load ratio of 1:1 is assumed, the service load on the column can be back-calculated using the CSA A23.3 load factors of 1.25 for dead and 1.5 for live. Thus:

3430 1.5 S LL 1.25 S DL S LL

1143 kN and

and 1.5 S LL

1.25 S DL

S DL 1372 kN

which results in a service load of 1143 + 1372 = 2515 kN. This load was applied to both columns during the fire endurance tests described herein.

361



Table E.1: Summary of design load calculations

Design Document ACI 318-95 (F)* CAN/CSA A23.3-94 (F) ACI 440.2R-02 (F) ISIS Canada (F) ACI 318-95 (U)ŧ CAN/CSA A23.3-94 (U) ACI 440.2R-02 (U) ISIS Canada (U) * ŧ

Unwrapped Ult. Strength (kN) 3289 2722 3289 2722 4386 4149 4386 4149

FRP-Wrapped Ult. Strength (kN) --5049 3430 --7054 5094

Strength Increase (kN) --1760 708 --2668 945

Increase in Strength (%) --53.5 26.0 --60.8 22.7

F – refers to factored design load calculations (ultimate design capacities) U – refers to unfactored load calculations (predicted load capacities)

362


APPENDIX A: Additional Heat Transfer Equations

APPENDIX A ADDITIONAL HEAT TRANSFER EQUATIONS The Following equations were derived in a manner similar to the heat transfer equations presented in Chapter 5. As such, details of the derivation are omitted here and only the final forms of the equations are given. Presented in this appendix are finite difference heat transfer equations for the specific cases of an unwrapped reinforced concrete column and a wrapped and insulated reinforced concrete column.

A.1 Unwrapped Reinforced Concrete Column For an unwrapped reinforced concrete column, referring to Figure A.1 and using an elemental energy balance approach as discussed in Chapter 5, the following finite difference heat transfer equation can be derived at the concrete-fire interface:

T1i

T1i

2 Rc ∆t

1

ρ c Cc

ρ H O C H Oφmi 1 Rc 2

2

∆xc ∆t 2

Rc

ρ c Cc

ρ H O C H Oφmi 1 Rc 2

∆xc ∆xc 4

2

∆xc ∆xc2 4

σε c (T fi

k1i

1

1

) (T

273

k 2i 1 T1i

1

4

i 1 1

T2i

)

273

4

1

[A.1] In cases when the above equation is used to represent the heat transfer for a wrapped column for which the wrap has fallen off, the subscripts 1 are changed to M1, and the subscripts 2 become M1+1. For any element inside the concrete we have:

305



Tmi

Tmi 1

∆t 2 ρ c Cc Rc

ρ H OC 2

m

Rc

m

φ

i 1 H 2O m

(m 1)∆xc

Rc

∆xc2

(

)(

)

(

)(

)

3 ∆xc k mi 11 k mi 1 Tmi 11 Tmi 1 2 1 ∆xc k mi 1 k mi 11 Tmi 1 Tmi 11 2

[A.2]

At the centreline of the column, Equation 5.23 is used for heat transfer calculations. The initial volume of water in the fire-concrete interface element is given by Equation 5.14 and the volume of water evaporated from that element during a time interval, ∆t , is:

2 ∆t

∆V1H 2O

ρ H O λH O 2

(

Rcσε c T fi

1

273

) (T 4

)

i 1 1

273

4

2

(Rc

k1i

1 2 ∆xc )

1

[A.3]

k 2i 1

i 1 1

i 1 2

T

2

T

For an internal concrete element, the initial volume of water is given by:

VmH 2O

2 Rc

(m 1)∆xc

∆xc φ i

[A.4]

And the amount of water evaporated during a time interval is given by:

∆VmH 2O

∆t

ρ H O λH O 2

Rc

2

Rc

(m (m1

3 2)∆xc k mi 11

k mi 1 Tmi 11 Tmi 1

1 2)∆xc k

i 1 m 1

i 1 m

k

i 1 m

T

[A.5]

i 1 m 1

T

The initial volume of water in the concrete centerline element, and the volume of water evaporated from that element during a time interval, ∆t , are given by Equations 5.21 and 5.22 respectively. There are three criteria for stability of the finite difference heat transfer analysis of an unwrapped reinforced concrete column. At the fire-concrete interface, we have:

∆t

(ρ c Cc )min ∆x w 2 2 (hrad )max ∆xc (k c )max

[A.6]

At any point inside the concrete, we use Equation 5.28.

306



A.2 Wrapped and Insulated Reinforced Concrete Column For an insulated and wrapped reinforced concrete column, there are three materials present in the analysis and hence three interfaces; 7 finite difference equations are required in the analysis. Referring to Figure A.2, at the insulation-fire interface we have:

T1i

T1i

2 Ri ∆t σε i T fi ∆xi ∆xi ρ i Ci Ri 4

(

1

∆xi ∆t 2 k1i ∆xi ∆xi2 ρ i Ci Ri 4

) (T

1

273

4

)

i 1 1

273

4

[A.7]

Ri

1

k 2i 1 T1i

1

T2i

1

where the subscript, i , refers to the insulation. Inside the insulation:

Tmi

Tmi 1

2 ρ i Ci Ri Ri

(m

∆t (m 1)∆xi ∆xi 2

1 2)∆xi k

i 1 m

k

Ri

i 1 m 1

(m

i 1 m

3 2)∆xi k mi 11 k mi 1 Tmi 11 Tmi 1

[A.8]

i 1 m 1

T

T

For the insulation-wrap interface:

TMi ins

TMi ins1

∆t 5 4)∆xi ∆xi

ρ i Ci Ri Ri

(M ins

(M ins 3 2)∆xi

∆xi Rw

(m

1 2)∆x w

∆x w

k Mi ins1 k Mi ins1

1

ρ wC w (Rw ∆xw 4)∆xw

k Mi ins1 TMi ins1

k Mi ins1

1

TMi ins1

1

TMi ins1

TMi ins1

[A.9]

1

where Mins is the element number at the insulation-wrap interface. At any point inside the wrap:

Tmi

Tmi 1

2 ρ wC w Ri Ri Rw

(m (m

∆t (m M ins )∆xw ∆xw 2

M ins 1 2 )∆x w k mi 11 k mi 1 Tmi 11 Tmi 1

[A.10]

M ins 1 2 )∆x w k mi 1 k mi 11 Tmi 1 Tmi 11

And at the wrap-concrete interface:

307



TMi 1

TMi 11

ρ w C w Rw Rw

(n

(n

wrap

wrap

1 4)∆x w ∆x w

1 2)∆x w

k Mi 11 1

∆x w Rc

∆xc 2 i 1 k M1 ∆xc

∆t ρ c Cc

ρ H O C H Oφ Mm 2

2

1

1

Rc

∆xc 4 ∆xc

k Mi 11 TMi 11 1 TMi 11

k Mi 11 1 TMi 11 TMi 11 1 [A.11]

where nwrap is the number of layers of wrap material. At any point inside the concrete Equation 5.20 can be used, and at the centreline Equation 5.23 is applicable. The equations used to describe the effect of moisture within the concrete for the wrapped and insulated case are the same as those used for the wrapped case. The initial volume of water in the FRP-concrete interface element and the volume of water evaporated from that element during a time interval, ∆t , are given by Equations 5.14 and 5.16 respectively. For an internal concrete element, Equation 5.18 gives the initial volume of water, and Equation 5.19 gives the amount of water evaporated from the element during a time interval. The initial volume of water in the concrete centerline element, and the volume of water evaporated during a time interval are given by Equations 5.21 and 5.22 respectively. There are seven criteria for stability of the analysis in the case of a wrapped and insulated reinforced concrete column. At the fire-insulation interface, we have:

(ρ i Ci )min ∆xi 2 2 (hrad )max ∆xi (k i )max

∆t

[A.12]

At any point inside the insulation, we have:

∆t

(ρ i Ci )min ∆xi 2 4(ki )max

[A.13]

At the insulation-FRP interface, we have:

308



∆t

(ρ i Ci )min 2

Ri Ri

(M ins (M ins

5 4)∆xi ∆xi 3 2)∆xi

∆xi

(ρ w C w )min

(k i )max

Rw

Rw

(M ins

(1 2)∆x w

∆x w

1 4)∆x w ∆x w

[A.14]

(k w )max

At any point inside the FRP, at the FRP-concrete interface, and at any point inside the concrete, we use Equations 5.26, 5.27, and 5.28.

309



1

Fire

Rc

½ ∆x

∆x

2

m Concrete M2

Figure A.1: Discretization of an unwrapped circular concrete column for heat transfer purposes

½ ∆xi 1 2 Mins

Fire Insulation Wrap

Rc

∆xi

Mins+1 M1-1 M1

M1+1

½ ∆xw ∆xw

½ ∆xc ∆xc

Rw m

Ri Concrete

M2

Figure A.2: Discretization of a wrapped and insulated circular concrete column for heat transfer purposes

310310

L.A. L.A. Bisby, Bisby, Ph.D. Ph.D. Thesis, Thesis, 2003 2003

APPENDIX B: Design Charts for FRP Bar-Reinforced Concrete Slabs

APPENDIX B DESIGN CHARTS FOR FRP BAR-REINFORCED CONCRETE SLABS In this appendix, design recommendations for FRP bar-reinforced concrete slabs are presented in graphical form. The charts were developed using the program QSFIRE, which is described in detail in Chapter 6 of this thesis. Design charts are presented for two different slab thicknesses (120 mm and 200 mm) and three different aggregate types (carbonate, siliceous, and expanded shale). It is suggested that the 120 mm series of charts be used for all slabs with an overall thickness of less than 200 mm. In those rare cases where the overall slab thickness is less than 120mm, the design charts presented in this thesis are unconservative and should not be used. However, QSFIRE can be used with relative ease to plot new design charts for different slab thicknesses and aggregate types. The use of the design charts presented in this section is complicated by difficulties involved in determining the critical temperatures for FRP reinforcing materials. Considerable further research is required in this area, including full-scale structural fire endurance tests on loaded FRP-reinforced concrete slabs, such that the consequences of heat on the behaviour of FRP bar-reinforced concrete slabs can be rationally accounted for in design.

311


APPENDIX B: Design Charts for FRP Bar Reinforced Concrete Slabs

Critical Temperature = 100oC

Concrete Cover (mm)

120

200oC

100 80

300oC

60

400oC

40

500oC 600oC

20 0 0.0

0.5

1.0

1.5

2.0

Fire Endurance (hours) Figure B.1: Design chart for 120 to 200 mm thick carbonate aggregate concrete slabs


Concrete Cover (mm)

120

200oC

100 80

300oC

60

400oC 500oC

40

600oC

20 0 0.0

0.5

1.0

1.5

2.0

Fire Endurance (hours)

Figure B.2: Design chart for 120 to 200 mm thick siliceous aggregate concrete slabs

312




120

Concrete Cover (mm)

200oC

100 300oC

80

400oC

60

500oC 600oC

40 20 0 0.0

0.5

1.0

1.5

2.0

2.5

3.0

Fire Endurance (hours) Figure B.3: Design chart for 120 to 200 mm thick expanded shale aggregate concrete slabs


Concrete Cover (mm)

200 175 150

200oC

125 300oC

100

400oC

75

500oC

50

600oC

25 0

0

1

2

3

4

5

Fire Endurance (hours) Figure B.4: Design chart for 200 mm thick or larger carbonate aggregate concrete slabs

313




Concrete Cover (mm)

200 175

200oC

150 125

300oC

100

400oC

75

500oC 600oC

50 25 0

0

1

2

3

4

5

Fire Endurance (hours) Figure B.5: Design chart for 200 mm thick or larger siliceous aggregate concrete slabs

Critical Temperature =

Concrete Cover (mm)

200

o

100 C

175 150

o

200 C

125 o

300 C

100

400oC

75

o

500 C o 600 C

50 25 0

0

1

2

3

4

5

Fire Endurance (hours) Figure B.6: Design chart for 200 mm thick or larger expanded shale aggregate concrete slabs

314


APPENDIX C: Models for FRP-Confined Concrete Columns

APPENDIX C MODELS FOR FRP-CONFINED CONCRETE COLUMNS The information presented in this appendix is of interest to the civil engineering community in that it represents, to the knowledge of the author, the most detailed meta-analysis of confinement models conducted to date. The information presented is not essential to the core topic of this thesis however, and hence this appendix has been written as a stand-alone manuscript with the intention that it will be published as a separate research paper.

C.1 Introduction It is now well recognized that that both the compressive strength and ductility of concrete members can be significantly enhanced using circumferential FRP wraps. These non-corrosive, light weight, and easily installed wraps can be used to rehabilitate corrosion damaged columns, increase the load capacity of under-strength members, and retrofit seismically inadequate bridges and buildings. A significant research effort during the past 20 years has focused on the use of FRP materials for confinement of concrete, and a wide variety of issues concerning this application of FRPs have been examined. Numerous field applications have been implemented around the world. With the recent publication of several design guidelines for FRP-confined concrete members, FRPs appear poised to make a significant impact on the repair and retrofit industry. There are however, several key areas in which there appears to be a lack of consensus among the research community. One such area relates to the analytical modelling of FRP-confined concrete. Since the initial studies on FRP-confined concrete (Fardis and Khalili, 1981), several analytical models, with varying degrees of sophistication, have been presented to predict their stress-strain response. Virtually all such models have been derived empirically, have represented modifications of existing models developed for conventionally (steel spiral or hoop) confined

315



concrete, and have essentially utilized equivalent variables. Many of these confinement models were derived based on relatively few tests, and some display gross inadequacies when compared with a more complete database of experimental results. The most basic models provide only the failure stress and strain for FRP-confined concrete columns, while others approximate the full stress-strain behaviour as bilinear. Recently, more sophisticated iterative procedures have been suggested that are capable of approximating the complete stress-strain response of FRP-confined concrete (Fam and Rizkalla, 2001; Spolestra and Monti, 1999). However, accuracy does not necessarily follow from sophistication, and with new confinement models being presented each year, it remains unclear which model should be used in light of the existing experimental data. In this appendix, various available analytical models for the ultimate stress and strain of FRP-confined concrete members, as well as their complete stress-strain behaviour (where applicable), are compared against the results of tests reported in the literature. The various models are subsequently compared, based on their accuracy and precision using statistical indicators, in an attempt to determine which model is most appropriate in light of the database of available test data. The ability of the various models to accurately describe the complete stress-strain response of FRP confined concrete is briefly examined and discussed. Finally, a least-squares regression analysis is used to modify the numerical coefficients of several of the existing analytical equations, resulting in a series of modified best-fit confinement models.

C.2 Significance The study in this Appendix attempts to provide an impartial evaluation of the available models for circular FRP-confined concrete, such that a consensus can be reached and universal design guidelines can be formulated. This is done, to the knowledge of the author, using the largest database of experimental data assembled to date, representing results from around the world with many different FRP materials and various specimen sizes. Widely sanctioned and

316



defensible design guidelines are essential if FRPs are ever to gain widespread acceptance as repair materials for infrastructure.

C.3 Confinement Models In this section, each of the previously developed analytical models considered in this study is presented, and the key assumptions made are discussed where appropriate. The reader should remain cognizant of the fact that the notation and format of the original equations has been modified in most cases to ensure uniformity of notation within this thesis.

The reader is

encouraged to consult the original references for clarification. In the following equations, fl represents the lateral confining pressure at ultimate, calculated by assuming that the concrete will fail when the wrap reaches its ultimate tensile failure stress (Karbhari and Gao, 1997):

fl

2 f com t d

[C.1]

Fardis and Khalili (F&K)

The first attempt at modelling the confining effect of an FRP wrap on circular concrete columns was presented in the pioneering work of Fardis and Khalili (1981), who performed a series of tests with glass/epoxy FRP-wrapped concrete cylinders. They subsequently suggested two equations to predict the failure stress of FRP confined concrete. The first equation was taken from an early paper on steel confined concrete by Richart et al. (1929), who found that the confined concrete strength was a function of the original concrete strength, f’co, and the confining pressure at ultimate: f cc'

f co' 1 k1

fl f co'

[C.2]

Richart et al. suggested a k1 coefficient of 4.1 as an average value from their tests on concrete confined with steel ties. Subsequent researchers using steel confinement proposed values

317



ranging from 2.8 to 4.8 (Considere, 1903; Iyengar et al., 1971) and in one case between 4.5 and 7.0 (Balmer, 1944). The second equation was taken from Newman and Newman (1972), and was also derived empirically from tests on steel confined concrete. This equation is equivalent to Equation C.2 with: k1

f 3.7 l' f co

0.14

[C.3]

Fardis and Khalili (1981) found that the strain at failure for glass FRP-confined concrete was a function of the differential stiffness parameter (the bracketed term in Equation C.4 below). They subsequently suggested an equation for the failure strain of an FRP-wrapped column:

ε cc

ε co

0.001

E com t d f co'

[C.4]

Karbhari and Gao (K&G)

Karbhari and Gao (1997) examined a variety of models for steel confined concrete and formulated two new models for concrete confined with FRP. The first model was based on a simplistic composite analysis and resulted in equations for concrete stress and strain at ultimate. This model assumed that the behaviour of FRP confined concrete could be described with a bilinear relationship as shown in Figure C.1a and given by the following equations: f cc'

f co' 3.1 f co'

1.04 1

ε cc 1 where

c

c

f co' Eeff

2 f com t d

2t Ecom d Ec 4.1 f co'

1 ε com

c

2t E com d Ec E eff

2

[C.5]

[C.6]

is the initial Poisson’s ratio of the concrete and Eeff is given by: Eeff

Ecom Acom

318

Ec Ac

[C.7]



Ac and Acom are the cross-sectional areas of the concrete and FRP wrap, respectively, in the longitudinal direction. The second Karbhari and Gao model was of the same form as Equation C.2 above, but with: f 2.1 l' f co

k1

0.13

[C.8]

This equation was derived using a least-absolute deviations regression analysis on test data available from Howie and Karbhari (1995). Karbhari and Gao (1997) also commented on the failure strain of FRP confined concrete, and noted that the non-linear portion of the stress-strain curve for FRP-confined concrete was unaffected by the modulus of the wrap, as had been suggested by Fardis and Khalili (1981), but that it was controlled by the confining pressure developed in the jacket. Hence, the failure strain is directly dependent on the strength of the composite:

ε cc

ε co

f 0.01 l' f co

1.0

[C.9]

Karbhari and Gao (1997) used their analysis to suggest a simple bilinear stress-strain curve for FRP confined concrete (refer to Fig. C.1a). The curve is defined in terms of three points: the origin, a point of damage initiation that corresponds to the beginning of second linear portion of the response, given by:

σ A f co' 4.1 f co' ν c

2t Ecom and εA d Ec

σA E eff

[C.10]

and the ultimate point described by Equations C.5 and C.6. Samaan

Samaan et al. (1998) also suggested an equation of the Richart et al. (1929) form, based on a database of 22 data points from studies on FRP-wrapped concrete and concrete-filled FRP

319



tubes conducted by Mirmiran and Shahawy (1997). In their version of Equation C.2 it was suggested that: k1

6 .0 f l

0.3

[C.11]

To predict the failure strain of FRP-confined concrete, Samaan et al. (1998) assumed that the initial slope of the bilinear stress-strain curve is given by Equation C.14, that the slope of the second portion is a function of the stiffness of the confining material and, to a lesser extent, the unconfined concrete strength as described by Equation C.15, and that the intercept of the second portion with the stress axis is a function of the unconfined concrete strength and the confining pressure at ultimate as given in Equation C.16. These assumptions result in the stress-strain curve described by Equations C.13 to C.16, which with some manipulation yields an estimate of the failure strain as:

ε cc

f cc'

0.872 f co ' 0.371 f l 6.258 E t 0.2 245.61 f co' 1.3456 com d

[C.12]

Samaan et al. (1998) suggested a simple bilinear approximation for the complete stressstrain response of FRP-confined concrete using a four-parameter relationship as shown in Fig. C.1b. The curve is described using the following equations:

(E1

fc 1

(E1

E 2 )ε c E 2 )ε c fo

n

E 2ε c

1 n

[C.13]

where n = 1.5 and,

E1 E2 fo

3950 f co'

245.61 f co'

0.2

Ecom t d

[C.15]

6.258

[C.16]

1.3456

0.872 f co' 0.371 f l

320

[C.14]



Miyauchi

Miyauchi et al. (1997) also used the Richart model as a starting point for their FRP confinement model, but introduced an effectiveness coefficient, determined from a regression analysis of results from tests on 10 carbon FRP-wrapped cylinders. This coefficient has the effect of reducing k1, in Equation 2, from 4.1 to 3.485. By assuming that the strain confinement ratio, εcc/εco, is a function only of the confinement ratio, fl/f’co, Miyauchi et al. (1997) performed a regression analysis on the data from their tests, and suggested the following predictive equations for ultimate strain of FRP confined concrete:

ε cc

ε co 1 10.6

fl f co'

ε cc

ε co 1 10.5

fl f co'

0.373

for f co' 30 MPa

[C.17]

for f co' 50 MPa

[C.18]

0.525

However, no guidance was given as to how intermediate or higher unconfined concrete strengths should be treated. Based on their equations, Miyauchi et al. (1997) presented a simple two-part stress-strain curve for FRP-confined concrete. The curve (Figure C.1c) consists of an initial parabolic portion (which is essentially the stress-strain curve for unconfined concrete) and a straight-line portion that is tangent to the parabola and passes through the point described by Equation C.2 (with k1 = 3.485) and Equation C.17 or C.18. Hence: fc

fc

f

' co

ε 2 c ε co

f cc

εc ε co

2

λ ε cc ε c

for 0 ε c

for ε tan

εc

ε tan

[C.19]

ε cc

[C.20]

where,

ε tan ε co

321

λε co2 2 f co'

[C.21]



f co' 2

f tan 2 f co' (ε cc

λ

ε tan ε co

2

ε tan ε co

[C.22]

(

ε co )

4 f co' f co' ε cc2

2 f co' ε co ε cc

f cc' ε cc

2

)

ε co2

[C.23]

Saafi

Saafi et al. (1999) suggest equations of empirical form based on a regression analysis of results from tests on 15 concrete-filled FRP tubes. For strength prediction, their equation can be expressed as Equation C.2, with: 2.2

k1

0.16

fl f co'

[C.24]

Saafi et al. (1999) chose to use a variation of the Mander equation (Mander et al., 1988) for the ultimate strain of steel confined concrete to describe the failure strain of concrete-filled FRP tubes. This empirically defined equation gives failure strain as a function of the confined and unconfined concrete strengths and the failure strain of the confining FRP as:

ε cc

ε co 1 k 2

f cc' f co'

[C.25]

1

where k2 = 537(εcom+2.6) for concrete-filled FRP tubes. A bilinear stress-strain curve for FRP confined concrete, based on a regression analysis and the Ahmad (1981) stress-strain curve for concrete, was suggested in this study. Referring to Figure C.1d, the stress-strain equations are as follows: E1ε c

fc 1

E1 fa

2

ε1

[C.26]

E 2 E1ε 1 εc f a2

1

ε 12

E1 E 2 2 εc f a2

where, E1 10200 f co'

322

1 3

[C.27]



E2

fa

0.272

f co' 1 0.0213

f co'

[C.28]

ε co

0.84

Ecom t f co' d

[C.29]

0.84

E t 1 0.0783 com f co' d

ε 1 ε co

[C.30]

Toutanji

Toutanji (1999) followed an approach identical to Saafi et al. (1999) to suggest stress and strain equations for FRP-wrapped concrete cylinders.

These equations were derived using

regression analysis of test data from 6 different studies on FRP-wrapped concrete columns. Toutanji suggested that: k1

3. 5

fl f co'

0.15

[C.31]

310.57ε com

k2

[C.32]

1. 9

Toutanji also presented equations for the complete stress-strain response of FRP-wrapped concrete. These equations are as follows: E1ε c

fc 1

E1 fa

2

ε1

[C.33]

E 2 E1ε 1 εc f a2

1

ε 12

E1 E 2 2 εc f a2

where, E1 10200 f co'

E2

51000 f co'

1 3

[C.34]

1 3

[C.35]

' co

E t 1 0.0178 com f co' d

ε 1 ε co

E t 1 0.0448 com f co' d

fa

f

323

0.85

[C.36]

0.85

[C.37]



Spolestra and Monti (S&M)

Spolestra and Monti (1999) developed a rational iterative procedure to determine the complete stress-strain response of circular concrete columns confined by FRP wraps. Their iterative model for the complete stress-strain behaviour of FRP confined concrete accounts for the continuous interaction between the dilating concrete core and the confining wrap, and does not inherently assume the observed bilinear behaviour of FRP confined concrete. Details of the iterative procedure to determine the full stress-strain response are discussed in detail in Chapter 2 of this thesis, and the accuracy of the resulting stress-strain curves is discussed later in this Appendix. Based on their incremental-iterative approach, Spolestra and Monti (1999) also suggested two sets of equations for hand calculation of the confined ultimate stress and strain. The first set of equations is referred to as exact and is based on the Mander curve (Mander et al., 1988) for confined concrete. The analysis takes the Mander stress-strain curve for the ultimate confining pressure: f cc'

*

f co' 2.254 1 7.94

ε cc*

fl f co'

f cc' f co'

ε co 1 5

2

fl f co'

1

1.254

[C.38]

[C.39]

Next, the secant modulus at ultimate, Esec,µ , is calculated using: Ec

Esec,u 1 2

β

β f com

5700 f co'

[C.40]

E com 500

[C.41]

which gives the ultimate confined stress and strain at the intersection of the Mander curve and the Esec,µ sloped line through the origin:

324



1

f cc'

ε cc

ε cc*

*

ε cc*

(E

c

Esec,u Ec

f cc'

Esec,u ) f

* f cc' ' ε cc Ec

[C.42]

' * cc * cc

ε

E sec,u ε cc

[C.43]

The second set of equations presented by Spolestra and Monti is based on regression analysis of the results obtained from the incremental-iterative model. Based on observations as to which variables were most important in describing the observed experimental behaviour, three independent parameters were identified and varied between their extremes. From the 600 cases that resulted, the following predictive equations for ultimate confined stress and strain were suggested: f cc'

ε cc

ε co 2

f co' 0.2 3

1.25

fl f co'

Ec ε com f co'

[C.44]

fl f co'

[C.45]

Xiao and Wu (X&W)

Xiao and Wu (2000) conducted independent tests on 27 concrete cylinders wrapped with carbon FRP sheets, and subsequently performed a regression analysis to suggest an equation of the form of Equation C.2 for prediction of confined ultimate stress. The resulting equation is: 2

f

' cc

f

' co

1.1

f' d 4.1 0.75 co 2 Ecom t

fl f co'

[C.46]

Xiao and Wu (2000) also suggested a simple bilinear curve for the complete stress-strain response of FRP-wrapped concrete cylinders, also based on a regression analysis of their experimental data. The initial portion of the curve is described by equations based on elastic theory as:

325



Ec ε c

fc

2ν c f r for 0

fc

f co'

[C.47]

where,

νc

εr

Cj

1

Ec

1 νc

2ν

[C.48]

εc 2 c

After achieving the unconfined concrete strength, the second stabilized portion of the stress-strain curve can be described using: fc

1.1 f co'

kf l for f co'

fc

f cc'

[C.49]

where,

εr

ε ro' ν c' ε c

[C.50]

In the above equations, fl

C jε r

[C.51]

Cj

2t Ecom d

[C.52]

ν c'

7

f co' Cj

0.8

[C.53]

The ultimate failure strain of the FRP-confined concrete can be obtained by substituting the failure strain of the FRP into Equation C.49, which with some manipulation yields:

ε cc

ε com 7

0.0005 ' co

f d 2t E com

0.8

[C.54]

Theriault and Neale (T&N) and Lam and Teng (L&T)

Theriault and Neale (2001), and Lam and Teng (2001), apparently independently, followed the rationale of Richart et al. (1929) along with a regression analysis of available data to suggest a simple design equation for the ultimate strength of FRP-confined concrete cylinders.

326



The equations are presented in slightly different forms, but essentially simplify to Equation C.2 with k1 = 2.0. A version of this equation has also been used in the ISIS Canada guidelines for the rehabilitation of concrete members with advanced composite materials (ISIS, 2001a), although modified by the addition of material resistance factors and a performance factor (which is currently taken as 1.0 and which accounts for various factors such as: FRP stiffness, ultimate failure strain of the FRP, concrete compressive strength, and quality of bond). Lam and Teng (2001) also suggested a simple predictive equation for the confined ultimate strain as:

ε cc

ε co 2 k 2

fl f co'

[C.55]

In the above equation, k2 is a strain enhancement coefficient that depends on the type of FRP (carbon, glass, or aramid) with which the column is wrapped. For a carbon FRP-wrapped columns the authors suggested k2 = 15, and for glass FRP concrete-filled tubes they suggested: f 27 l' f co

k2

0.3

[C.56]

Using their predictive stress and strain equations, Lam and Teng (2001) offered a bilinear stress-strain curve for FRP-confined concrete, consisting of an initial parabolic portion followed by a straight line to failure (Fig. C.1e), and described by the following equations: fc

Ec ε c fc

(E c

E2 )

2

4f f co'

' co

ε c 2 for 0 ε c

E2ε c for ε t

ε c ε cc

εt

[C.57] [C.58]

where,

εt

2 f co' (Ec E 2 )

327

[C.59]



E2

f cc'

ε cc

f co'

[C.60]

ACI

Committee 440 of the American Concrete Institute (ACI, 2002) has taken an approach that is essentially equivalent to the Mander equation for the ultimate strength of steel confined concrete. The only modification made is in some minor simplifications to the coefficients of the Mander equation, giving: f cc'

f co' 2.25 1 7.9

fl f co'

2

fl f co'

1.25

[C.61]

The ACI design document suggests a predictive equation for the strain at failure, also based on the work of Mander et al. (1988) as follows:

ε cc

1.71(5 f cc 4 f co ) Ec

[C.62]

C.4 Experimental Database A database of results from uniaxial compression tests on circular FRP-wrapped concrete columns was compiled using data presented by a number of authors (Ahmad et al., 1991; Callery, 2000; Demers and Neale, 1994; Demers et al., 1996; El-Hacha et al., 1999; Fam and Rizkalla, 1999; Fardis and Khalili, 1981; Harmon and Slattery, 1992; Howie and Karbhari, 1995; Karabinis and Rousakis, 2001; Kawashima et al., 1997; La Tegola and Manni, 1999; Matthys et al., 1999; Mirmiran and Shahawy, 1997; Miyauchi et al., 1997; Nanni and Bradford, 1995; Picher, 1995; Purba and Mufti, 1999; Rochette and Labossiere, 2000; Saafi et al., 1999; Samaan et al., 1998; Santarosa et al., 2001; Suter and Pinzelli, 2001; Theriault et al., 2001; Toutanji, 1999; Watanabe et al., 1997; Xiao and Wu, 2000). The resulting database represents results from tests on over 250 circular FRP-confined concrete columns from 27 different sources. For each test in the database, it was required to know: the unconfined concrete strength, f’co, the column diameter, d, the total FRP wrap thickness, t, modulus, Ecom, and tensile strength,

328



fcom, and the unconfined initial concrete compressive modulus, Ec. The unconfined concrete strain at ultimate, εco, was assumed to be 0.2% unless given as otherwise in the source document. The database of test results was then used in conjunction with each of the analytical models presented previously, to determine the theoretical confined ultimate stress and strain for each column in the database (and in some cases the complete stress-strain response). It was thus possible to examine the performance of each model with respect to the database of experimental results.

C.5 Ultimate Strength Figure C.2 shows a comparison of the mean absolute errors (MAEs) and mean errors (MEs) for each of the ultimate strength models described above, in comparison with the complete database of experimental results. The MAE is used here as an indicator of precision whereas the ME is an indicator of accuracy. It is evident that there are significant differences between the abilities of the various models to predict the ultimate strength of FRP confined concrete. There are 6 models that have comparable MAE values of between 14% and 16%, although there is significant variation among these models in the ME values obtained. The reader will note that an ME value of less than zero corresponds to an under-prediction of strength (conservative prediction). If overall model performance is measured in terms of ability to predict ultimate strength, then the best model is the K&G II or the Saafi model, both of which have low MAEs and MEs close to zero. If the goal is conservatism, as would be the case in a design situation, then the F&K or L&T/T&N models are preferable, since they both have low MAEs and underestimate strength by about 7% on average. It is interesting to note that the L&T/T&N model is used in the ISIS Canada design guidelines for FRP confined concrete. Also, the ACI model appears to be relatively inaccurate and unconservative when used without reduction factors. Figure C.3 shows a comparison of the model predictions and experimental values for the Saafi model. There is generally good agreement between model predictions and experimental

329



data, although there is a significant scatter in the plot. This scatter will likely be very difficult to account for with an empirical confinement model, and could be due to any of a wide variety of factors, including: fiber type, adhesive type, FRP-wrapped versus concrete-filled tube applications, quality of bond, overlap length in FRP-wrapping applications, fiber orientation, etc.

C.6 Failure Strain Figure C.4 shows a comparison of the MAE and ME values for each of the ultimate strain models described above, in comparison with the complete database of experimental data. Large differences are observed among the abilities of the various models to predict the failure strain of FRP confined concrete. The ability of these relatively simple models to predict strain at failure is far less developed than for ultimate stress prediction. In the case of failure strain, 5 of the models have comparable MAE values, between 45% and 60%, although there is significant variation among the models in the observed ME values. If overall model performance is measured in terms of ability to predict ultimate strain, then the best model is the Miyauchi II model, which has the lowest MAE and an ME of about 18% on the unconservative side of zero. If the goal is conservatism, as would be the case in a design situation, then the K&G II model is the only practical option, since it has a low MAE and a negative ME. Figure C.5 shows a comparison of the model predictions and experimental failure strain values for the Miyauchi II model, where the scatter in the data is immediately apparent. The Miyauchi II model appears to overestimate the failure strains for low strain values and underestimate the strains for high values. The ability to predict failure strain is complicated by many of the same factors as ultimate strength, and would likely require the complex considerations of fracture mechanics and non-linearity in order to fully describe the behaviour. In any case, the analysis suggests that it is not advisable to rely on the vastly increased failure strains that are often observed when confining concrete with FRP, as experimental data indicates the

330



potential for unpredictable failures at strain levels well below those predicted by all but the K&G II model. More work is required to develop a satisfactory, and conservative, failure strain model which accounts for the complexities of behaviour involved.

C.7 Stress-Strain Behaviour For those models presented above that have been used to predict the complete stressstrain behavior of FRP-confined concrete, it was desired to determine which was the most satisfactory.

It is difficult however, given the variation observed in experimental stress-strain

curves presented by different authors, to arrive at any conclusive decision as to which stressstrain model is most appropriate. For any specific model it would certainly be possible to find at least one experimental stress-strain curve that is predicted almost perfectly. Indeed, many authors in the past have placed far too much confidence in their models by comparing them with a mere handful of data sets and subsequently proclaiming them to be accurate. For illustrative purposes only, Figure C.6 shows an experimental stress-strain curve (randomly selected from the literature) along with various model predictions for the stress strain response of a carbon FRP-confined concrete cylinder. There is significant variation between the stress-strain models, and although all analytical stress-strain curves capture the characteristic bilinear behaviour of FRP-confined concrete, the L&T model appears to describe the behaviour of this particular carbon FRP-confined concrete cylinder the best. It is important to remember that the experimental curve is provided simply as an example, and data could likely be found in the literature that would agree well with any of the stress-strain models. Given that all of the models adequately describe the overall shape of the confined concrete stress-strain curve, it is suggested that those stress-strain curves that are based on the most accurate failure stress and strain predictions should be used.

331



C.8 North American Design Guidelines Recently both the American Concrete Institute (ACI, 2002) and ISIS Canada (ISIS, 2001a) have published design guidelines for the use of FRP as confining materials for concrete. The ACI confinement model, described by Equations C.61 and C.62 above, is based on the Mander equation for steel-confined concrete and is unconservative with respect to the database of test data when reduction factors are omitted (see Figures C.2 and C.4). Introducing the reduction factors suggested in the ACI 440.2R-02 guide for the design of reinforced concrete columns confined with FRP wraps (ACI, 2002) results in the scatter plot shown in Figure C.7, where the database of experimental data is compared against the predictions of the ACI confinement design equations.

Because the ACI equations specify different environmental reduction factors

depending on the type of FRP and exposure condition, Figure C.7 distinguishes between carbon, glass, and aramid wraps, and considers an internal conditioned exposure condition (the least conservative exposure case). Also included in Figure C.7 are diagonal lines showing the mean ratio of experimental-to-theoretical ultimate strength, µ, and the mean minus three standard deviations, µ-3σ. Figure C.7 shows that while the ACI confinement equation appears to be conservative on average, there remain a number of database points for which the model is unconservative. Indeed, the µ-3σ line, which is essentially a line statistically assuring conservative results with a confidence level of 99% (assuming the data to be normally distributed), falls far on the unconservative side of the plot. The ISIS confinement model is based on the T&N model, developed from tests on FRP confined concrete, and is generally conservative (and more accurate than the ACI equation) with respect to test data, even when reduction factors are omitted. When the reduction factors of the ISIS design guidelines (ISIS, 2001a) are included, the predictions of the model are seen to be conservative in all cases (refer to Figure C.8), and the µ-3σ line remains on the conservative side

332



of the plot. Again, the design guidelines specify different environmental reduction factors, depending on the type of FRP wrap and exposure condition, and thus Figure C.8 distinguishes between carbon, glass, and aramid wraps, while assuming an internal exposure condition.

C.9 Best-Fit Model Because of the scatter inherent in the test results of the experimental database that has been used to evaluate the various FRP-confined concrete models in this Appendix, it is unlikely that models of more complex form than those previously suggested would be able to capture the observed variations in behaviour with any increased success. In this section, the empirical coefficients of the existing confinement equations have been modified to provide new confinement models that represent best-fit versions with respect to the database of test data. Following the lead of previous researchers, 3 new strength equations and 5 new strain equations are presented. The new strength equations are all based on the Richart form, where k1 has been evaluated based on least squares regression analysis assuming it to be (a) constant, (b) an exponential function of the confinement stress ratio, or (c) an exponential function of the confining stress at ultimate. The resulting 3 equations and their MAE and ME values are given in Table C.1. The 5 new ultimate strain equations were also developed using a least squares regression analysis following forms suggested by previous authors. The resulting equations and statistical indicators are also provided in Table C.1. Examination of the statistical accuracy indicators of the new models indicates that it is essentially impossible to predict the ultimate strength and strain of an FRP-confined concrete cylinder within about 15% and 40% respectively. Also, it appears that the increased complexity of the more advanced confinement models is unwarranted given the variation observed in tests. Consequently, it is recommended that Equations C.63 (for ultimate stress) and C.68 (for ultimate

333



strain) be used, with appropriate reduction factors, for the design of FRP-confined concrete columns. It should be pointed out that both Lam and Teng (2001) and Theriault and Neale (2001) have previously suggested virtually identical models for strength using smaller databases of experimental data, as have Miyauchi et al. (1997) for strain.

C.10 Summary and Conclusions Numerous empirical models have been developed to predict the ultimate strength and strain of FRP confined concrete. Most of these models have been developed based on relatively small data samples however, and many tend to over predict both ultimate strength and strain. In this Appendix, a variety of previously developed confinement models for FRPconfined concrete have been evaluated in light of a large database of test data. Because of the variability observed in the test data, it is likely impossible to develop an empirical model with less than about 15% absolute error for ultimate strength, and 40% absolute error for ultimate strain. Examination of both current North-American guidelines for design of FRP-confined concrete demonstrates that the ISIS Canada guidelines are accurate and conservative with at least a 99% confidence level (when reduction factors are used). The ACI design approach appears to be unconservative and may require re-evaluation. Simple modified empirical models for ultimate strength and strain have been suggested. These new models, used in conjunction with appropriate reduction factors, can be used for the design of FRP-wrapped concrete compression members.

334



Table C.1: Modified empirical confinement models Model Form

Proposed Equation MAE (%) Models for Ultimate Stress

F&K I, Miyauchi, L&T, T&N F&K II, K&G II, Saafi, Toutanji

f cc'

f co' 1 2.265

f cc'

f co' 1 2.101

Samaan

f cc'

f co'

fl f co'

fl f co'

3.87 f l

ME (%)

Equation #

14.4

-2.6

C.63

14.2

-1.4

C.64

14.0

-2.2

C.65

47.6

1.02

C.66

40.2

13.7

C.67

38.7

9.67

C.68

42.1

18.5

C.69

41.5

15.6

C.70

0.881

0.802

Models for Ultimate Strain F&K

ε cc

ε co 0.000829 ε cc

K&G II

ε co 0.0375

Ecom t d f co' fl f co'

0.0300

fl f co'

ε cc

ε co 1 8.417

f cc' f co'

ε cc

ε co 2 13.47

fl f co'

Miyauchi I&II

ε cc

Saafi, Toutanji L&T

ε co

0.83

1 0.85

335



fc

fc f cc'

f cc'

E2

A

σA

1 fo

n E1

Eeff 1

1

ε cc

εA

εc

εc

ε cc

εc

(b) Samaan et al. (1998)

(a) Karbhari and Gao (1997) fc

fc

f cc' f co' f tan

ε cc

f cc'

B

E2

A

1

fa

E1

ε tan ε co

ε cc

1

εc

(d) Saafi et al. & Toutanji (1999)

(c) Miyauchi et al. (1999) fc f cc'

E2 1 f co'

Ec

ε co ε t

0.0035

ε cc

εc

(e) Lam and Teng (1999)

Figure C.1: Various proposed stress-strain curves for FRP-confined concrete

336



20

10

-20

F&K II

Toutanji

X&W

ACI

K&G I

Miyauchi

S&M I

S&M II

Samaan

L&T / T&N

F&K I

-10

Saafi

0 K&G II

Ultimate Stress Error (%)

30

Mean Absolute Error Mean Error

-30

Figure C.2: A comparison of mean absolute errors and mean errors for ultimate strength predictions of the various confinement models as compared with the database of experimental data

140

Experimental Value (MPa)

120 100 80 60 40 20 0 0

20

40

60

80

100

120

140

Model Prediction (MPa)

Figure C.3: Comparison of model prediction and experimental data for ultimate strength prediction by the Saafi confinement model

337



200 Mean Absolute Error

175

Mean Error

125 100 75 50 25 S&M I

K&G I

Saaman

L&T

X&W

S&M II

Saafi

Toutanji

F&K I

Miyauchi I

-50

K&G II

-25

ACI

0 Miyauchi II

Failure Strain Error (%)

150

-75

Figure C.4: A comparison of mean absolute errors and mean errors for failure strain predictions of the various confinement models as compared with the database of experimental data 5

Experimental Value (% )

4

3

2

1

0 0

1

2

3

4

5

Model Prediction (% )

Figure C.5: Comparison of model prediction and experimental data for ultimate strain prediction by the Miyauchi II confinement Model

338



100 90

Axial Stress (MPa)

80 70 60

K&G Samaan Miyauchi Saafi Toutanji X&W L&T S&M Experiment

50 40 30 20 10 0 0

0.5

1

1.5 2 2.5 Axial Strain (%)

3

3.5

4

Figure C.6: A comparison predicted and experimental stress-strain behaviour of a CFRP wrapped concrete cylinder

140

µ

Experiment (MPa)

120 100 80 60 µ

σ

40 Carbon FRP Glass FRP Aramid FRP

20 0 0

20

40

60

80

100

120

140

Model Prediction (MPa)

Figure C.7: A comparison predicted and experimentally observed ultimate stress for the ACI 440 confinement model with reduction factors included

339



140 µ

Experiment (MPa)

120 µ

100

σ

80 60 40 Carbon FRP Glass FRP Aramid FRP

20 0 0

20

40 60 80 100 Model Prediction (MPa)

120

140

Figure C.8: A comparison predicted and experimentally observed ultimate stress for the ISIS Canada confinement model with reduction factors included

340


APPENDIX D: Modelling the Behaviour of FRP at High Temperature

APPENDIX D MODELLING THE BEHAVIOUR OF FRP AT HIGH-TEMPERATURE Information on the high-temperature behaviour of FRP materials is scarce. It was not possible to perform high-temperature tests on the specific FRP that was used in the experimental program reported in this thesis, and thus an attempt was made to develop analytical equations to describe the degradation in strength, elastic modulus, and bond properties of currently available FRP reinforcing materials with temperature. To accomplish this goal, an exhaustive literature survey was conducted to obtain as much information as possible on the high-temperature behaviour of FRP materials. The resulting database was used, in conjunction with a number of assumptions, to suggest semi-empirical analytical expressions describing FRP behaviour at high temperature. The form of the analytical expressions was assumed based on the limited previous work in this area (Dimitrienko, 1999; Katz et al., 2001), and coefficients were determined using a multi-variable least-squares regression analysis. Visual and graphical summaries of the data and resulting expressions are presented in Figures D.3 through D.10 and in Table D.1.

D.1 Experimental Database The database of experimental results was assembled from tests on the mechanical and bond properties of FRPs at high temperatures presented in the literature (Clarke, 1993; Dimitrienko, 1997a, b, 1999; Fujisaki et al., 1993; Gates et al., 1993; Katz et al., 1998, 1999; Kumahara et al., 1993; Rahman et al., 1993; Rehm and Franke, 1979; Rostasy, 1992; Sen et al., 1993; Sumida et al., 2001; Tanano et al., 1995, 1997; Uematsu et al., 1995). The database contained tests on carbon, glass, and aramid FRP with a number of different matrix materials, although the most common matrix represented in the database is epoxy. The database points are presented graphically in Figures D.3 through D.10. In most cases, authors provided few specific

341



details regarding the fibre type or polymer matrix formulation used in the fabrication of the FRPs for a particular study, likely due to the proprietary nature of much of the research conducted. As a result, and because of the enormous variety of FRP formulations possible, it is inadvisable to consider the information presented here more than a rough approximation of material behaviour at high temperature.

D.2 Assumed Analytical Form Only two studies that attempt to describe the variation in mechanical properties of FRP materials with temperature (using analytical expressions) could be found in the literature. Dimitrienko (1999) provides a detailed discussion of the thermomechanics of composite materials, and presents a number of semi-empirically derived expressions to describe the variation of strength, stiffness, density, thermal conductivity, and specific heat of FRP with temperature. Dimitrienko’s equations are derived from fairly sparse experimental data, but were found to agree well with tests in most cases. The most significant aspect of Dimitrienko’s work in terms of the present discussion is that thermal degradation of both strength and stiffness for carbon, glass, and aramid FRP were observed to follow a decreasing sigmoidal trend (refer to Figure D.1). This behaviour is characterized by insignificant changes in strength or stiffness up to a certain critical temperature, followed by a relatively rapid reduction to some constant residual value. The critical temperature is generally close to the GTT of the polymer matrix involved. The final portion of the sigmoid curve is a constant residual value. Katz et al. (2001) present semi-empirical analytical expressions to describe the thermal deterioration of bond strength for FRP reinforcements embedded in concrete. In this study, a hyperbolic tangential expression, which behaves as a decreasing sigmoid function, is assumed to describe degradation of bond-strength with temperature. The behaviour of the average bond strength at elevated temperatures is observed to be characterized by constant strength up to some transitional temperature, followed by a relatively rapid decrease to some constant residual value.

342



The reader will recall that the bond strength of FRP materials at high temperature is affected primarily by the shear strength and shear modulus of the polymer matrix. Studies on FRPs at high temperature have suggested that the shear strength and stiffness of the polymer matrices are also critical factors in the observed reductions in overall FRP strength and stiffness. Following the lead of both Dimitrienko (1999) and Katz et al. (2001), a sigmoid function was assumed to describe the observed reductions in strength, stiffness, and bond for FRP materials with temperature. The following analytical form was assumed:

f fo

1 a tanh ( b(T c )) 2

1 a 2

[D.1]

In the above expression, f refers to the mechanical property in question (strength, stiffness, or bond) at some temperature T, fo refers to the room-temperature value of the mechanical property in question, a is an assumed constant that describes the residual value for the mechanical property in question, and b and c are empirically derived (using least-squares regression analysis) constants that describe the central temperature (c) and severity (b) of property degradation with temperature. Figure D.2 should assist the reader with visualization of the various constants’ influence on the analytical expression.

D.3 Determination of Coefficients It was necessary to define the residual value, a, for all of the properties to be described before the regression analysis could be performed to obtain the coefficients b and c. This was accomplished using data supplied by Dimitrienko (1999) as a rough guide. Table D.1 provides the approximate residual strength values assumed in the derivation of the analytical expressions for all material properties. The reader should remain cognizant of the fact that these values have been chosen based on test data obtained using a specific epoxy resin, the specific composition of which remains unknown. Strictly speaking, tests would be required to determine the residual properties for any specific FRP, and hence, only experimental results for FRPs having epoxy resin matrices have been included in the regression analysis. The residual strength values are not

343



critical in any case, as they are assumed to be less than 10% of room temperature values in all cases. Once the constant, a, was estimated for each property and FRP type, a multi-variable least-squares regression analysis was conducted using SigmaPlotTM software to determine the coefficients b and c. The only exception to this procedure was in the analysis for the tensile strength of carbon fibres, which were assumed to display negligible thermal degradation up to at least 1000 C (in fact, data available from the literature appears to indicate a mild increase in strength for carbon fibres with elevated temperature). The Pearson correlation coefficient (R-squared value) was calculated for each of the analytical equations with respect to the appropriate database of experimental data. Table D.1 presents the resulting calculated values for the coefficients b and c, along with R-squared values for each of the analytical expressions.

D.4 Results and Discussion The database of experimental results for tensile strength tests on pure fibres is shown in Figure D.3, as are the semi-empirically derived analytical expressions for the tensile strength of raw fibres which have been normalized to the room-temperature strength. Although considerable scatter is present in the experimental data, the analytical curves appear to adequately describe the overall trends in the degradation in strength of the various fibres with increasing temperature. It is interesting to note that, while carbon appears insensitive to temperature, both glass and aramid fibres experience significant (>5%) reductions in tensile strength at temperatures below 200 C and 100 C respectively. It is hence evident that carbon fibres appear to be more suitable for high temperature applications. Also shown in Figure D.3 are R-Squared values for the analytically derived curves. While both the aramid and glass curves show satisfactory R-squared values, the carbon curve has a value of –0.05, which statistically means that a horizontal line at 100% strength describes the

344



data worse than a horizontal line through the mean strength of all data points. This is because a best-fit line for this data set would actually show a slightly increasing tensile strength with temperature. Assuming a constant strength of 100% for carbon fibres is a conservative treatment of the data. Figure D.4 shows the experimental and analytical data for the variation of CFRP tensile modulus with temperature. Substantial scatter is present in the data (hence the relatively low RSquared value of 0.6), although again, the analytical curve appears to capture the overall trends in the experimental data. Figure D.5 gives experimental and analytical data for the elastic modulus of GFRP and AFRP. The analytical model captures the overall trends, although the R-Squared value for the GFRP modulus is extremely low, likely due to the small number of data points used to fit the analytical curve. Curves showing the database points and analytical expressions for the tensile strength of CFRP, GFRP, and AFRP are shown in Figures D.6, D.7, and D.8 respectively. In all three cases there is substantial scatter in the data, resulting in marginal R-Squared values. The reader must remain cognizant however, of the fact that only data points for FRP products incorporating epoxy matrix materials were included in the derivation of the analytical expressions, so the data sets were generally quite small. Figure D.9 shows the experimental data versus the analytical prediction for the average bond stress of Glass/Vinyl-Ester FRP bars embedded in concrete. The analytical model does an excellent job of describing this data set, with an R-Squared value of 0.89. In the case of bond strength, the predicted reduction takes place at a significantly lower temperature than for the other mechanical properties (c

120). This could be for one of three reasons. First, bond depends

almost exclusively on the shear strength of the matrix material at the surface of the rod, which is severely degraded at high temperature. Second, the surface of the rod is the first region of the FRP to experience elevated temperatures, since it takes time for heat transfer to the center of the

345



poorly transversely thermally conductive FRP material, with the result that the bond will be affected earlier than strength or stiffness of the overall FRP composite. Finally, the matrix used in the bond tests was a vinyl ester, as opposed to epoxy which is represented in the majority of the strength and stiffness data. The use of a different matrix could produce major differences in behaviour at high temperature, although insufficient data is available to validate this hypothesis. Also shown in Figure D.9 is experimental data for the reduction in bond strength of conventional reinforcing steel with temperature, which demonstrates that the deterioration of bond strength for steel, while significant, is not nearly as severe as for FRP materials. As a final point of discussion, the reader will likely remark that the R-squared values presented herein for the analytical expressions are in most cases quite low, and little confidence should be placed in the analytical equations’ ability to predict the high-temperature behaviour of FRP materials. As stated earlier, because of the enormous variety of FRP materials available for reinforcement of concrete, it is important to conduct high temperature tests on any specific FRP to be used in a particular application. The analytical models presented here were developed only as approximate tools in the numerical analysis of FRP-reinforced concrete members subjected to fire. A considerable amount of further work, both experimental and analytical, is required in this area.

D.5 Comparison and Conclusion Figure D.10 shows a comparison of the analytical models developed for the various materials and mechanical properties discussed in this appendix. Examination of the analytical expressions in this light is instructive, as there are a number of interesting observations that can be made. All three curves for FRP modulus and FRP strength, respectively, are similar in both location and severity of degradation for different FRP materials, with the curves for CFRP showing slightly superior retention of mechanical properties at elevated temperature. This is as expected, since all of the expressions were derived from epoxy matrix FRPs, and since both

346



modulus and strength depend substantially on matrix properties.

The slightly superior

performance of CFRP can likely be attributed to the thermal insensitivity of the carbon fibres. The analytical expressions presented in this appendix have been used in the numerical modelling of FRP-wrapped reinforced concrete columns and FRP bar-reinforced concrete slabs presented in Chapters 5 of this thesis.

347



Table D.1: Semi-empirically derived coefficients for various mechanical properties of FRP at high temperature Property Carbon Fibre Strength Glass Fibre Strength Aramid Fibre Strength CFRP Modulus GFRP Modulus AFRP Modulus CFRP Strength GFRP Strength AFRP Strength Vinyl-Ester Matrix Bond Strength

b ( 10-3) N/A 3.67 7.52 8.68 7.91 7.93 5.83 8.10 8.48 22.0

a N/A 0.05 0.10 0.05 0.05 0.05 0.10 0.10 0.10 0.10

348

c N/A 642.78 312.03 367.41 320.35 290.49 339.54 289.14 287.65 119.73

R-Squared -0.05 0.78 0.80 0.60 0.11 0.59 0.32 0.34 0.54 0.89



Strength Coefficient (not normalized)

All Curves CFRP with: 1 – Epoxy ( ) 2 – Epoxy-Phenolic ( ) 3 – Phenolic 4 – Polyimide ( ) 5 – Silicon-Organic 6 – Carbon Fibre

Temperature, T (ºC)

Figure D.1: Variation of tensile strength properties coefficient for CFRP with temperature (Dimitrienko, 1999)

1.2

Normalized Property, f / fo

1 0.8 f fo

0.6

1 a tanh ( b(T c )) 2

1 a 2

c = 300

0.4 b = 0.01 a = 0.1

0.2 0 0

100

200

300

400

500

600

Temperature, T (deg. C)

Figure D.2: Assumed form for the analytical expression describing deterioration of strength, stiffness, and bond of FRP reinforcements

349



120 R2=-0.05

100

% Retained

S-Glass (1) E-Glass (1) HTS Roving (1) E-Glass Roving (1) E-Glass (2) S-2 Glass (2) S-Glass (3) E-Glass (3) Glass (5) Aramid Fibre (3) Aramid (4) Aramid (5) HM-Carbon (3) Carbon (4) Carbon (5) Idealized Glass Idealized Aramid Idealized Carbon

80 60 R2=0.78

40 20

2

R =0.80

0 0

200

400

600

800

1000

(1) Rehm and Franke, 1979 (2) Sen et al., 1993 (3) Rostasy, 1992 (4) Sumida et al., 2001 (5) Dimitrienko, 1999

Temperature (deg. C) Figure D.3: Database points and analytical expressions for the variation of tensile strength of pure carbon, glass, and aramid fibres with temperature 120

% Retained

100 80 60 2

R =0.60

40 20 0 -50

50

150

250

350


450

PAN Carbon / Epoxy 1 (1) Pitch Carbon / Epoxy 1 (1) Pitch Carbon / Cement (1) Pitch Carbon / Epoxy 2 (1) Carbon / Glass / Vinyl Ester (2) Braided Carbon / Epoxy (3) Stranded Carbon / Epoxy (3) Carbon / Inorganic (3) Carbon / Inorganic (4) Braided Carbon / Epoxy (4) Spiral Carbon / Epoxy (4) Carbon / PEEK (5) Carbon / Bismalemide (6) Carbon / Thermoplastic (6) Carbon / Epoxy (7) Idealized Carbon / Epoxy (1) Kumahara et al., 1993 (2) Clarke, 1993 (3) Tanano et al., 1997 (4) Tanano et al., 1995 (5) Uematsu et al., 1995 (6) Gates et al., 1993 (7) Dimitrienko, 1999

Figure D.4: Database points and analytical expression for the variation of elastic modulus of CFRP with temperature

350



120

Glass / PPS (1) Spiral Glass / Epoxy (3) Glass / Epoxy (4) Aramid / Epoxy (1) Braided Aramid / Epoxy 1 (2) Braided Aramid / Epoxy 2 (2) Stranded Aramid / Epoxy (2) Braided Aramid / Epoxy (3) Idealized Glass / Epoxy Idealized Aramid / Epoxy

% Retained

100 80 60 40

(1) Kumahara et al., 1993 (2) Tanano et al., 1997 (3) Tanano et al., 1995

2

R =0.11 2

R =0.59

20 0 0

100

200

300

400

500

Temperature (deg. C) Figure D.5: Database points and analytical expressions for the variation of elastic modulus of GFRP and AFRP with temperature 120

% Retained

100 80 60 40 20

2

R =0.32

0 -50

50

150

250

350


450

PAN Carbon / Epoxy 1 (1) PAN Carbon / Epoxy 2 (1) PAN Carbon / Epoxy 3 (1) Pitch Carbon / Epoxy 1 (1) Pitch Carbon / Epoxy 2 (1) Braided Carbon / Epoxy (4) Stranded Carbon / Epoxy (4) Braided Carbon / Epoxy (5) Spiral Carbon /Epoxy (5) Carbon / Epoxy (6) Carbon / Epoxy (7) Carbon / Epoxy-Phenolic (7) Pitch Carbon / Cement (1) Carbon / Inorganic (5) Carbon / Glass / Vinyl Ester (2) Carbon / Glass / Vinyl Ester (3) Carbon / Polyimide (7) Idealized Carbon / Epoxy (1) Kumahara et al., 1993 (2) Rahman et al., 1993 (3) Fujisaka et al., 1993 (4) Tanano et al., 1997 (5) Tanano et al., 1995 (6) Sumida et al., 2001 (7) Dimitrienko, 1999

Figure D.6: Database points and analytical expression for the variation of tensile strength of CFRP with temperature

351



120

Carbon / Glass / Vinyl Ester (1) Carbon / Glass / Vinyl Ester (2) Glass / PPS (3) Glass / Vinyl Ester (3) Polystal (4)

% Retained

100 80

Spiral Glass / Epoxy (5) Glass / Epoxy (6) Glass / Epoxy (7) Idealized Glass / Epoxy

60

(1) Rahman et al., 1993 (2) Fujisaki et al., 1993 (3) Kumahara et al., 1993 (4) Clarke, 1993 (5) Tanano et al., 1995 (6) Dimitrienko, 1999 (7) Dimitrienko, 1997

40 20 R2=0.34

0 -50

50

150

250

350

450

Temperature (deg. C) Figure D.7: Database points and analytical expression for the variation of tensile strength of GFRP with temperature

120

Aramid / Vinyl Ester (1) Braided Aramid / Epoxy (2) Arapree (3)

100

Aramid / Epoxy (1)

% Retained

Braided Aramid / Epoxy(4)

80

Aramid / Vinyl Ester (5) Idealized Aramid / Epoxy (1) Kumahara et al., 1993 (2) Tanano et al., 1997 (3) Clarke, 1993 (4) Tanano et al., 1995 (5) Sumida et al., 2001

60 40

2

R =0.54

20 0 0

100

200

300

400

500

Temperature (deg. C) Figure D.8: Database points and analytical expression for the variation of tensile strength of AFRP with temperature

352



120

% Retained

100 80

R1 (1)

R2 (1)

R3 (1)

R4 (1)

CB (2)

CPH (2)

CPI (2)

NG (2)

Steel (2)

Idealized

(1) Katz et al., 1998 (2) Katz et al., 1999

60 40 R2=0.89

20 0 0

50

100

150

200

250

300


R1: Glass / Vinyl Ester Helical wrap w/ embedded sand R2: Glass / Vinyl Ester Helical wrap w/o coating R3: Glass / Vinyl Ester Helical wrap w/ embedded sand R4: Glass / Vinyl Ester Molded deformations CB: Glass / Urethane modified Vinyl Ester Molded deformations CPH: Glass / Epoxy Vinyl Ester Helical wrap w/ embedded sand CPI: Glass / Epoxy Vinyl Ester Helical wrap w/ resin roughening NG: Glass / Polyester Helical wrap w/ embedded sand and deformations Steel: Reinforcing Steel 12 mm diameter w/ deformations

Figure D.9: Database points and analytical expression for the variation of bond strength of glass/vinyl-ester FRP with temperature

% of Room Temperature Value

120

Carbon Strength Glass Strength

100

Aramid Strength CFRP Modulus GFRP Modulus

80

AFRP Modulus CFRP Strength GFRP Strength

60

AFRP Strength Ave. Bond Strength

40 20 0 0

100

200

300

400

500

600

Temperature (deg. C) Figure D.10: Comparison of analytical expressions for the variation of strength, elastic modulus, and bond strength of FRP temperature

353


APPENDIX F: Material Properties at Elevated Temperature

APPENDIX F MATERIAL PROPERTIES AT ELEVATED TEMPERATURE This appendix provides a summary of material behaviour used in the numerical modelling and analysis of Chapters 5 and 6 of this thesis. Information is presented for concrete, steel, FRP, VG insulation, and water, with respect to both thermal (specific heat, thermal conductivity) and physical (density, strength, stiffness) properties where required by the numerical analyses.

F.1 Concrete The equations presented in this section have been reproduced after Lie (1992). Thermal Capacity, ρc,TCc,T For siliceous aggregate concrete, with Tc in °C and ρc,TCc,T in J/m3·°C: 0

200:

Tc

ρ c ,T C c ,T

(0.005Tc 2.7 10 6

200

Tc

400: ρ c ,T C c ,T

400

Tc

500: ρ c ,T C c ,T

500

Tc

600: ρ c ,T cc ,T

Tc

1.7 ) 10 6

(0.013Tc 2.5) 10 6 ( 0.013Tc 10.5) 10 6

[Eqns. F.1]

ρ c ,T cc ,T 2.7 10 6

3316:

For carbonate aggregate concrete: 0

Tc

400:

ρ c ,T C c ,T 2.566 10 6

400

Tc

410: ρ c ,T C c ,T

410

Tc

445: ρ c ,T C c ,T

(0.1765T 68.034 ) 10 6 ( 0.05043T 25.00671)

445

Tc

500: ρ c ,T C c ,T

2.566 10 6

500

Tc

635: ρ c ,T C c ,T

635

Tc

715: ρ c ,T C c ,T

715

Tc

785: ρ c ,T C c ,T

Tc

785:

10 6

(0.01603T 5.44881) 10 6 (0.16635T 100.90225 ) 10 6 ( 0.22103T 176.07343) 10 6

[Eqns. F.2]

ρ c ,T C c ,T 2.566 10 6

363



For lightweight aggregate concrete: 0

Tc

400:

ρ c ,T C c ,T 1.930 10 6

(0.0772T 28.95) 10 6 ( 0.1029T 46.706 ) 10 6

400

Tc

420: ρ c ,T C c ,T

420

Tc

435: ρ c ,T C c ,T

435

Tc

600: ρ c ,T C c ,T 1.930 10 6

600

Tc

700: ρ c ,T C c ,T

700

Tc

720: ρ c ,T C c ,T

720:

Tc

[Eqns. F.3]

(0.03474T ( 0.1737T

18.9140 ) 10 6 126.994 ) 10 6

ρ c ,T C c ,T 1.930 10 6

Thermal Conductivity, kc,T For siliceous aggregate concrete, with Tc in °C and kc,T in W/m·°C: 0

800:

Tc

Tc

800:

k c ,T

0.000625Tc

1.5

[Eqns. F.4]

k c ,T 1.0

For carbonate aggregate concrete: 0

Tc

293:

293:

Tc

k c ,T 1.355

k c ,T

0.001241Tc

[Eqns. F.5]

1.7162

For lightweight aggregate concrete: 0

Tc

600:

600:

Tc

k c ,T

0.00039583 Tc

0.925

[Eqns. F.6]

k c ,T 0.6875

Stress-Strain Behaviour The following equations are valid for all aggregate types. 0

Tc

Tc

450:

450:

ε max

0.0025

f c',T

f c'

f c',T

f c'

(6.0 T

c

2.011 2.3353 0.04 Tc2

)

10

Tc 20 1000

[Eqns. F.7]

6

Coefficient of Thermal Expansion, αc,T For siliceous and carbonate aggregate concrete, with Tc in °C and αc,T in °C-1: c ,T

(0.008Tc 6) 10

6

[ F.8]

364



For lightweight aggregate concrete:

7.5 10

c ,T

6

[ F.9]

F.2 Steel The following equations have again been reproduced after Lie (1992). Stress-Strain Behaviour In the following equations, ε p is the yield strain, f y is the room-temperature yield strength of the steel in MPa, ε s is the current steel strain in MPa, f s ,T is the current steel stress in MPa, and the steel temperature, Ts, is in °C. For ε s

εp:

f s ,T

f (Ts , 0.001) εs 0.001

For ε s

εp:

f s ,T

f (Ts , 0.001) εp 0.001

Where f (Ts , 0.001) And ε p

4 10

6

[ F.10]

f (Ts , (ε s ε p

(

(50

0.04 Ts ) 1 e (

0.001)) f (Ts , 0.001) [ F.11]

30 0.03 Ts ) 0.001

)

6.9

fy

[ F.12] [ F.13]

Coefficient of Thermal Expansion, αs,T For reinforcing steel, the coefficient of thermal expansion, αs,T, in °C-1, in terms of the steel temperature, Ts, in °C, is: 0 Ts

Ts

1000:

1000:

s ,T

(0.004

Ts 12 ) 10 6

s ,T

16 10

6

[Eqns. F.14]

F.3 FRP Specific Heat, Cw,T In the following equations, Cw,T has units of kJ/kg·°C and Tw is in °C.

365



0

Tw

325:

C w,T

325

Tw

343:

C w,T

343

Tw

510:

C w,T

510

Tw

538:

C w,T

538

Tw

3316:

C w,T

Tw

3316:

C w,T

0.95 (Tw ) 325 2.8 (Tw 325) 2.2 18 0.15 (Tw 343) 5.0 167 3.59 (Tw 510) 4.85 28 1.385 (Tw 538) 1.265 2778 0

1.25

[Eqns. F.15]

Density, ρw,T In the following equations, ρw,T has units of g/cm3 and Tw is in °C. 0

Tw

510:

ρ w,T

1.6 0.35 (Tw 510) 28

510

Tw

538:

ρ w,T

1.6

538

Tw

1200:

ρ w,T

1.25

[Eqns. F.16]

Thermal Conductivity, kw,T In the following equations, kw,T has units of W/m·°C and Tw is in °C. 0

Tw

500:

500

Tw

Tw

650:

650:

k w,T

1.4

k w,T

1.4

k w,T

0.2

1.1 Tw 500 0.1 (Tw 500) 150

[Eqns. F.17]

Strength, fcom,T , and Elastic Modulus, Ecom,T For a CFRP wrap: aσ = 0.1 bσ = 5.83e-3 cσ = 339.54 aE = 0.05 bE = 8.68e-3 cE = 367.41

f com ,T

f com

1 aσ tanh 2

For a GFRP Wrap: aσ = 0.1 bσ = 8.10e-3 cσ = 289.14 aE = 0.05 bE = 7.91e-3 cE = 320.35

bσ (Tw

cσ )

366

1 aσ 2

For an AFRP wrap: aσ = 0.1 bσ = 8.48e-3 cσ = 287.65 aE = 0.05 bE = 7.93e-3 cE = 290.49 [F.18]



Ecom ,T

Ecom

1 aE tanh 2

bE (Tw c E )

1 aE 2

[F.19]

In the above equations both fcom,T and Ecom,T have units of MPa with Tw in °C. Coefficient of Thermal Expansion, αw Coefficients of thermal expansionfor the wrap, αw, have been assumed constant with temperature based on the literature review presented in Chapter 2. For CFRP: αw = 0.0 °C-1 For GFRP: αw = 6.3-6 °C-1 For AFRP: αw = -3.6-6 °C-1

[Eqns. F.20]

F.4 VG Insulation Because VG vermiculite-gypsum fire insulation is a proprietary material, neither detailed thermal properties nor specific information on the composition of the product were available to the author. Thus, the properties of VG had to be estimated using information from a variety of sources. The dry-density of gypsum plaster is approximately 865 kg/m3.

According to the

Schundler Company (www.schundler.com), the dry density of a standard-graded pulverized exfoliated vermiculite aggregate is between 96 and 160 kg/m3. Assuming the density is given by the average value of 128 kg/m3, based on the measured dry density of the VG material (measured by the author as approximately 330 kg/m3) and assuming that there is little or no volume interaction between the two materials on mixing, the mass ratio of vermiculite to gypsum should be approximately 2:1. Specific Heat, CVG,T The specific heat of VG, CVG,T in kJ/kg°C, at some temperature, TVG in °C, has been approximated using the rule of mixtures. The specific heat of vermiculite has been assumed constant with temperature (http://www.vermiculite.org/properties.htm), and the specific heat of gypsum has been assumed to vary with temperature as suggested by Buchanan (2001), where the

367



effect of dehydration is accounted for by artificially increasing the specific heat near 100°C. By assuming a 2:1 mix ratio of vermiculite to gypsum, we arrive at the following equations for specific heat: 0 TVG

20:

CVG ,T

1.1763

20

TVG

78:

CVG ,T

1.1763

78

TVG

125: CVG ,T

1.3058

125

TVG

137: CVG ,T

6.9066

137

TVG

153: CVG ,T

1.3722

153

TVG

610: CVG ,T

1.0136

610

TVG

663: CVG ,T

0.8509

663

TVG

690: CVG ,T

1.6976

690

TVG:

CVG ,T

0.9167

1.3058 78 6.9066 125 6.9066 137 1.3722 153 1.0136 610 1.6976 663 1.6976 690

1.1763 (TVG 20 1.3058 (TVG 78 1.3722 (TVG 125 1.0136 (TVG 137 0.8509 (TVG 153 0.8509 (TVG 610 0.9167 (TVG 663

20) 78) 125) 137 )

[Eqns. F.21]

153) 610 ) 663)

Density, ρVG,T The variation in density of installed VG insulation, ρVG,T in kg/m3, with temperature,

TVG in °C, was determined from TGA and has been idealized for numerical modelling as follows: 0

TVG 100:

100

TVG

200

TVG:

200:

ρVG ,T

351

ρVG ,T

351

ρVG ,T

287

351 287 (TVG 100) 200 100

[Eqns. F.22]

The TGA demonstrated a mass loss of about 16% between 100 and 300°C. Lie, 1992 describes the variation in density of gypsum by stating that it contains about 24% by mass of water (free and chemically combined), which is lost through evaporation in the temperature range of 100°C to 200°C. Exfoliated vermiculite is thermally inert up to temperatures in excess of 1000°C. It is interesting to note that by assuming a vermiculite-to-gypsum ratio of 2:1 by mass,

368



we would estimate a mass loss of about 18% between 100°C and 200°C, a similar result to that actually observed in TGA testing. Thermal Conductivity, kVG,T The thermal conductivity of VG insulation, and indeed of all gypsum-based plasters, is highly variable with temperature and remains poorly understood.

To estimate the thermal

conductivity of VG insulation, the rule of mixtures has been used (as was the case for specific heat). The thermal conductivity of vermiculite has been assumed constant with temperature (http://www.vermiculite.org/properties.htm), and the thermal conductivity of gypsum has been assumed to vary with temperature as suggested by Buchanan (2001). The following equations are used here, with kVG,T in W/m°C: TVG 100:

0

kVG ,T

0.1158

100

TVG

101: kVG ,T

0.1158

101

TVG

400: kVG ,T

0.0726

400 8000

TVG TVG:

800: kVG ,T

kVG ,T

0.1158 0.0726 (TVG 100) 101 100 [Eqns. F.23]

0.1224 0.0726 800 0.2087 0.1224 1000

0.0726 (TVG 400) 400 0.1224 (TVG 800) 800

In all cases, the thermal properties used for numerical modelling of the VG insulation compare favourably with manufacturer specified properties as presented in Table 3.5.

F.5 Water The following data has been reproduced after Lie (1992). Thermal Capacity of Water

ρ H OCH O 2

2

4.2 10 6 J / m 3

o

[F.24]

C

Heat of Vaporization of Water

λH O 2

[F.25]

2.3 10 6 J / kg

369


VITA

VITA Name:

Luke Alexander Bisby

Place/Year of Birth:

Calgary, Alberta, Canada, 1974

Education:

Queen’s University, Kingston, Ontario, 1997-1999 Department of Civil Engineering (Structures) M.Sc. (Eng.) 1999 McGill University, Montreal, Quebec, 1993-1997 Department of Civil Engineering and Applied Mechanics B.Eng. 1997

Experience:

Research Contractor, Fire Risk Management Group, Institute for Research in Construction, National Research Council of Canada, 2003 Teaching Fellow, CIVL 335, Structural Analysis, Department of Civil Engineering, Queen’s University, 2003, 2001 Research Assistant to Dr. M.F. Green, Department of Civil Engineering, Queen's University, 1997-2003 Course Instructor, CIVL223, Mechanics and Materials, Department of Civil Engineering, Queen’s University, 1999, 2000, 2001 Teaching Assistant, Department of Civil Engineering, Queen's University, 1997-2002 Teaching Assistant, Department of Civil Engineering and Applied Mechanics, McGill University, 1997

Awards: 1996-1997 1996-1997 1997-1998 1997-2002 1997-1998 1998 1998-2000 1998-2000 1999-2000 2000-2002 2000-2002 2000 2001-2002 2002 2002 2002 2002-2003

J.W. McConnell Award, McGill University Zeev Vered Award, McGill University Senator Frank Carrel Fellowship, Queen's University Queen's Graduate Award (multiple years), Queen's University Teaching Assistant Award, Civil Engineering, Queen's University First Place, Student Paper Competition, ISIS Canada Annual Conference Ontario Graduate Scholarship (declined), Queen's University NSERC PGS A Scholarship, Queen's University Educational Excellence Teaching Award, Queen’s Engineering Society Ontario Graduate Scholarship (declined), Queen's University NSERC PGS B Scholarship, Queen's University 2nd Place, Student Competition, CSCE Annual Conference, London, ON ISIS Canada Annual Scholarship Robert Mitchell Prize, Queen’s University First Prize, First ISIS Canada Student Design Competition Best of Group and Best of Conference, ISIS Canada Annual Poster Competition Ontario Graduate Scholarship for Science and Technology

370


VITA

Refereed Publications: Green, M.F., Dent, A.J.S., and Bisby, L.A. 2003. Effect of Freeze-Thaw Cycling on the Behaviour of Reinforced Concrete Beams Strengthened in Flexure with FRP Sheets. Submitted to Canadian Journal of Civil Engineering, January, 14 pp. Bisby, L.A. and Green, M.F. 2002. Freeze-Thaw Durability of the FRP-Concrete Bond. ACI Structural Journal, 99(2), pp. 215-226. Green, M.F., Bisby, L.A., Beaudoin, Y. and Labossiere, P.J. 2000. Effect of Freeze-Thaw Cycling on the Bond Durability Between FRP Reinforcements and Concrete. Canadian Journal of Civil Engineering, 27(5), pp. 949-959. Conference Proceedings: Williams, B.K., Bisby, L.A., Green, M.F., and Kodur, V.K.R. 2003. An Investigation of the Fire Behaviour of FRP-Strengthened Reinforced Concrete Beams. In Forde, M.E. (Ed.), Structural Faults and Repair – 2003, London, England, July 1st-3rd, CD-ROM Bisby, L.A., Williams, B.K., Green, M.F., and Kodur, V.K.R 2002. Studies on the fire behaviour of FRP reinforced and/or strengthened concrete members. In Benmokrane, B. and E. El-Salakawy (Eds.), Durability of Composites for Construction, Montreal, PQ, May 29th-31st, pp. 405-417. Bisby, L.A., Green, M.F., and Kodur, V.K.R 2001. Fire Behaviour of FRP-Wrapped Reinforced Concrete Columns. In Forde, M.E. (Ed.), Structural Faults and Repair – 2001, London, England, July 4th-6th, CD-ROM. Bisby, L.A., Green, M.F., Beaudoin, Y., and Labossiere, P.J. 2000. FRP Plates and Sheets Bonded to Concrete: Bond Behaviour, 3rd International Conference on Advanced Composite Materials in Bridges and Structures, Ottawa, August 15-18th, pp. 209-216. Green, M.F., Bisby, L.A., Labossiere, P., and Beaudoin, Y. 1998. Effects of freeze-thaw action on the bond of FRP sheets to concrete. In Benmokrane, B., and Rahman, H. (Eds.), Durability of Fibre Reinforced Polymer (FRP) Composites for Construction, Sherbrooke, Quebec, Aug. 5-7th, pp. 179-189. Conference Presentations: Kodur, V.K.R., Foo, S., and Bisby, L. 2003. Fire Performance of Concrete Slabs Reinforced with FRP bars. Research in Progress Session, Spring Convention of the American Concrete Institute, Vancouver, April 1st-4th. Bisby, L.A. 1998. Effect of Freeze-Thaw Cycling on the Anchorage of FRP Sheets Bonded to Reinforced Concrete Beams. ISIS Canada Annual Conference, Winnipeg, May 5-7th. Other: Bisby, L.A. 1999. Effects of Freeze-Thaw Cycling on Anchorage of Fibre-Reinforced Polymer Sheets Bonded to Reinforced Concrete Beams. M.Sc. Thesis, Department of Civil Engineering, Queens University.

371


fire behaviour of fibre-reinforced polymer (frp

fire behaviour of fibre-reinforced polymer (frp

Suggest Documents

fire behaviour of fibre-reinforced polymer (frp ...

BOND BEHAVIOUR OF FIBRE REINFORCED POLYMER (FRP ...

Fire behaviour of FRP wrapped square reinforced concrete columns

Fire Resistance Performance of FRP Rebar

Experimental Behaviour of Carbon FRP Reinforced ...

durability of fibre reinforced polymer (frp) wraps ... - University of Toronto

Monitoring the production of FRP composites - eXPRESS Polymer ...

Load transfer of fibre-reinforced polymer (FRP ... - NRC Research Press

Durability Evaluation of Florida's Fiber- Reinforced Polymer (FRP ...

Flexural Behaviour of Polymer Modified Reinforced ...

Monitoring the production of FRP composites - eXPRESS Polymer ...

Flexural Behaviour of Full-scale Basalt FRP RC Beams ...

Results of fire resistance experiments on FRP-strengthened reinforced ...

Fire Resistance of FRP Reinforced Concrete Slabs

Thermal and Fire Characteristics of FRP Composites for ... - MDPI

Evaluating the fire performance of FRP-strengthened concrete structures

Fire Endurance of Insulated FRP-Strengthened Square ... - CiteSeerX

fire performance of concrete beams reinforced with frp bars

Behaviour of unconfined and FRP-confined rubberised concrete in ...

behaviour of frp-confined circular concrete-filled steel ...

Lateral Behaviour of Concrete-Filled FRP Tube Piles

behaviour of shear critical frp reinforced concrete one ...

Behaviour of hybrid FRP composite I-girder with concrete deck

Bond behaviour and durability of FRP composites ...