Final Report PERFORMANCE-RELATED SPECIFICATIONS FOR CONCRETE BRIDGE SUPERSTRUCTURES FHWA/IN/JTRP-2001/8. Volume 3 Nonmetallic Reinforcement

Transcription

1 Final Report PERFORMANCE-RELATED SPECIFICATIONS FOR CONCRETE BRIDGE SUPERSTRUCTURES FHWA/IN/JTRP-2001/8 Volume 3 Nonmetalli Reinforement Robert J. Frosh Professor of Civil Engineering Prinipal Investigator and Christopher P. Mosley, Graduate Researh Assistant A. Koray Tureyen, Graduate Researh Assistant Joint Transportation Researh Program Projet Number: C-36-56WW File Number: SPR-2325 Conduted in Cooperation with the Indiana Department of Transportation and the Federal Highway Administration U.S. Department of Transportation The ontents of this report reflet the views of the authors, who are responsible for the fats and the auray of the data presented herein. The ontents do not neessarily reflet the offiial views or poliies of the Indiana Department of Transportation or the Federal Highway Administration at the time of publiation. This report does not onstitute a standard, speifiation, or regulation. Purdue University West Lafayette, IN Otober 2002

2 ACKNOWLEDGMENTS The work desribed in this report was funded by the Joint Transportation Researh Program at Purdue University through SPR ontrat The support of the Indiana Department of Transportation (INDOT) and the Federal Highway Administration (FHWA) are gratefully aknowledged. The fiber reinfored plasti (FRP) bars used in the experimental program were donated by Tehnora, Marshall Industries Composites, and Hughes Bros., In. Their partiipation in this researh program is also gratefully aknowledged. The authors would like to extend thanks to Dr. Tommy Nantung from the Division of Researh of INDOT for his support throughout this projet. In addition, thanks are extended to the members of the Study Advisory Committee for their partiipation and thoughtful omments. This researh is a testament to their ommitment to build improved bridge superstrutures with a longer servie life. ii

3 INDOT Researh TECHNICAL Summary Tehnology Transfer and Projet Implementation Information TRB Subjet Code: Bridge Design and Performane Otober 31, 2002 Publiation No.: FHWA/IN/JTRP-2001/8, SPR-2325 Final Report Performane Related Speifiations (PRS) for Conrete Bridge Superstrutures- A Four Volume Report Introdution The development of Performane Related Speifiations (PRS) requires the identifiation of key performane levels for a given strutural system. The first attempt to develop a methodology for PRS an be traed to 1980 when the Federal Highway administration (FHWA) instituted a new researh program ategory. The main two objetives of the program were: 1) To provide a more rational basis for payment redution plans. 2) To develop additional speifiations related to the performane of flexible and rigid pavement strutures. In the early and mid-1980s, the FHWA, the National Cooperative Highway Researh Program (NCHRP), and the Amerian Assoiation of State Highway and Transportation Offiials (AASHTO) began a ooperative effort searhing for supporting data needed for the development of PRS. The idea was to develop performane models that would allow relating the material and onstrution testing parameters olleted at the time of onstrution to the future performane of the omplete projet. However, it was onluded that the existing databases were inadequate to derive the needed performane models. A known example of a PRS is the one developed for Portland Cement Conrete (PCC) pavements by Eres Consultants, In. and the FHWA (Darter et. al., 1998) in a ooperative effort. In this study, the overall objetives of a methodology for PRS were not ompletely fulfilled due to the lak of adequate supporting information in the existent databases to onstrut aurate performane preditive models. As a result, the proposed PRS was presented only as a methodology providing a more rational basis for payment plans. The objetive of the researh study was to develop the essential omponents of a PRS for onrete bridge superstrutures for appliation in the state of Indiana. The work onduted in this researh projet is presented in four volumes. Volume 1 summarizes the work onduted on the identifiation of performane levels and key parameters, and the development of aeptane riteria are addressed in Volume 1. The main objetive of this volume is to present a proposed methodology for a PRS for onrete bridge superstrutures. Volume 2 presents the researh findings dealing with development of High- Performane Conrete (HPC) for appliations in the bridge strutures in the state of Indiana. The objetive of the study presented in Volume 2 was to identify and develop onrete mixtures with adequate performane harateristis in terms of durability for the purpose of using these harateristis in performane-related speifiations. Volume 3 summarizes the work onduted to investigate the behavior of fiber reinfored polymer (FRP) reinfored onrete strutures with an emphasis on bond and shear. The main objetive of this volume is to provide design guidelines for the use of FRP reinforement in bridge superstrutures. Volume 4 summarizes the results of an evaluation of the bond performane of epoxyoated bars with a oating thikness up to 18 mils. Findings /02 JTRP-01/08 INDOT Division of Researh West Lafayette, IN 47906

4 In this study emphasis has been plaed on the development of a methodology for a Performane Related Speifiation, PRS, for onrete bridge superstrutures. The implementation of the methodology, presented in the form of a user-friendly omputer program in Volume 1 of this report, is projet speifi. It requires the mean and standard deviation (or definition of a probability distribution) of the input parameters for the performane preditive models. This is done for both the as-designed ondition and the as-built ondition of the struture. The ontrator is expeted to ahieve ertain level of ompliane during the onstrution as ditated by the as-designed ondition (whih is defined based on the submitted design in ompliane with ageny speifiations). Based on performane preditive models, ost models, and statistial simulation, the methodology reports a relative as-built/asdesigned Life-Cyle Cost (LCC). This relative LCC measures the level of ompliane of the asbuilt struture with the design. The ageny (INDOT) implementing the methodology ould then onsider the relative LCC in the form of a pay fator modifying the ontrator s bid prie. Statistial simulation is onduted to evaluate the effets of the variations in the input parameters for the performane preditive models. The differenes in the LCC for the as-designed and as-built elements ome from the differenes in the input parameters that are under the ontrol of the ontrator (referred to as quality harateristis). The framework of the proposed methodology has been fully developed and illustrated with four numerial examples in an initial ase study of a simply supported reinfored bridge dek or slab. The researh effort desribed in Volume 2 of this report was divided in two phases. Phase I was foused on development of onrete mixtures optimized with respet to seleted performane-related parameters. During this phase, ten optimum onrete mixes have been identified from 45 mixes in terms of ompressive strength, Young s modulus of elastiity, rapid hloride penetration and hloride ondutivity using a statistial design proedure. Through surfae response methodology, 27 statistial models were developed for eah of four parameters. Based on the models developed, 81 ontour maps were generated, whih indiated how performane of onrete varied in response to the hange of dosages of binders at onstant water-binder ratio. Based on the overlaid ontour maps and the threshold values hosen for the properties of onrete, optimum onrete mixtures inluding Portland ement and the ombinations with fly ash, silia fume and slag were identified. In Phase II of the HPC study, the ten optimum mixtures were further evaluated with respet to mehanial properties and durability harateristis. Several different tests related to the evaluation of the resistane of onrete to hloride permeability were used: rapid hloride permeability test, hloride ondutivity test, test for the resistane of onrete under DC eletrial field, ponding test for the determination of the resistane of onrete to hloride penetration, and rapid test for the determination of diffusion oeffiient from hloride migration. Tests related to the resistane of onrete to freezing & thawing, and saling were also investigated. Other tests suh as, the determination of drying shrinkage, and test for uring effets on the properties of high performane onrete were also evaluated in this researh. Speial emphasis was plaed on determining and quantifying these parameters that ontrol the ingress of the hloride ions. Based on the results generated during this researh, models have been developed that allow for predition of ertain mehanial and durability-related parameters related to the mixture omposition. The parameters that an be predited inlude strength, rapid hloride permeability (RCP) values, and hloride diffusion oeffiient. Limited validation of these models was performed using field data provided by INDOT. The strength and hloride diffusion oeffiient values generated by these models an serve as an input for the life-yle osting (LCC) model desribed in Vol. 1 of this report As summarized in Volume 3, experimental investigations were performed to speifially investigate the behavior of FRP reinfored onrete strutures in both bond and shear. For the bond investigation, three series of beam splie tests were performed on speimens reinfored with steel, glass FRP, and aramid FRP to determine the effet of the different types of reinforement on bond, raking, and defletions. The test results indiate that the use of FRP reinforement leads to lower bond strengths and, therefore, require longer development lengths. The speimen rak widths and defletions were substantially larger for FRP speimens than steel speimens due to /02 JTRP-01/08 INDOT Division of Researh West Lafayette, IN 47906

5 the signifiantly lower modulus of elastiity. Analysis of the test results resulted in reommendations for modifying the empirial development length equation of ACI design ode for use with FRP reinforement. For the shear investigation, two series of beam tests were onduted on speimens reinfored with steel, glass FRP, and aramid FRP to determine the effet of the different types of reinforement on the onrete shear strength. All speimens did not ontain transverse reinforement. The test results show that the use of FRP reinforement leads to lower onrete shear strengths than steel reinforement for equal reinforement ross-setional areas (longitudinal reinforement perentages). In addition, the test Implementation Based on the results from the researh onduted on the framework for a PRS, it was onluded that the most pratial implementation of the methodology had to onsider the orrosion deterioration problem as the only distress determining/affeting the LCC of the struture. It was onluded that other distress indiators applied at a setion level should be inluded in the framework of a PRS to give more integrity to the proess of quality ontrol. The needed software for the implementation of the proposed PRS has been provided to INDOT as part of this report. It must be noted that orrosion deterioration represents almost 50% of the problems of the urrent bridge infrastruture in Indiana. As part of the implementation efforts for the part of the researh dealing with HPC, a series of mathematial models were onstruted that allow for the predition of strength, rapid hloride permeability and hloride diffusion oeffiient values based on the binder omposition of the mixture. The data generated using these models have been arranged in an Exel sheet, whih allows the user to input desired minimum and maximum values of strength (at 28 days) and/or RCP values (at 56 days) and obtain binder ombinations whih yield/satisfy the desired input values. Binder system 1 refers to mixtures, whih ontain PC, SF and GGBS. Binder system 2 refers to mixtures, whih ontain PC, SF and FA. Binder system 3 refers to mixtures, whih ontain PC, GGBS and FA. The perentage inrements of SF represented in the Exel worksheet are 0, 5 and 7.5 %. The results point that the shear strength is a diret funtion of the longitudinal reinforement stiffness. The test results further substantiated the findings that larger rak widths and defletions are ahieved by FRP speimens relative to steel speimens due to the lower modulus of elastiity. Analysis of the test results resulted in reommendations for the alulation of onrete shear strength. The experimental work on the bond performane of epoxy-oated bars with thikness up to 18 mils summarized in Volume 4 of the final report indiates that the urrent AASHTO requirements for development length of epoxyoated bars ould be extended to oating thikness of up to 18 mils. perentage inrements of FA and GGBS represented are 0, 20, 25 and 30 %. The strength and hloride diffusion oeffiient values determined for the 10 onrete mixtures tested in Phase II of the study were also used as input values for the LCC model desribed in Vol. 1 of this report. The LCC model was run for a single, simply supported span. The same type of data was also obtained form three existing Indiana bridges and the LCC model was re-run for these strutures. The results indiate that LCC for all laboratory mixtures was lower than the LCC for standard INDOT lass C onrete mixture. Furthermore, the LCC of the atual field mixtures was slightly higher than the LCC of standard lass C mixture. Currently, the ability of the models developed as a part of the HPC study to predit the atual properties of a field onrete is being validated on several QC/QA bridge jobs and a supplementary report summarizing the results of these evaluations is expeted by June Based on the researh onduted on the use of FRP reinforement, design and onstrution reommendations are provided that an be used in the design and onstrution of FRP reinfored bridge deks. These reommendations will be implemented in a JTRP study Implementation of a Non-Metalli Reinfored Bridge Dek. This study will evaluate the design and onstrution reommendations in a prototype laboratory dek speimen as well as through a pilot field study that inorporates nonmetalli reinforement in a bridge dek /02 JTRP-01/08 INDOT Division of Researh West Lafayette, IN 47906

6 No hange of the bond speifiations is required to implement the use of up to #8 diameter deformed bars with epoxy-oating thikness up to 18 mils. Contat For more information: Prof. Julio A. Ramirez Prinipal Investigator Shool of Civil Engineering Purdue University West Lafayette IN Phone: (765) Fax: (765) Prof. Jan Olek Co-Prinipal Investigator Shool of Civil Engineering Purdue University West Lafayette IN Phone: (765) Division of Researh 1205 Montgomery Street P.O. Box 2279 West Lafayette, IN Phone: (765) Fax: (765) Purdue University Joint Transportation Researh Program Shool of Civil Engineering West Lafayette, IN Phone: (765) Fax: (765) Prof. Robert J. Frosh Co-Prinipal Investigator Shool of Civil Engineering Purdue University West Lafayette IN Phone: (765) Indiana Department of Transportation /02 JTRP-01/08 INDOT Division of Researh West Lafayette, IN 47906

7 1. Report No. 2. Government Aession No. 3. Reipient's Catalog No. FHWA/IN/JTRP-2001/8 TECHNICAL REPORT STANDARD TITLE PAGE 4. Title and Subtitle Performane Related Speifiations for Conrete Bridge Superstrutures- Volume 3: Nonmetalli Reinforement 7. Author(s) Robert Frosh, Christopher Mosley, A. Koray Tureyen 9. Performing Organization Name and Address Joint Transportation Researh Program 1284 Civil Engineering Building Purdue University West Lafayette, Indiana Report Date Otober 31, Performing Organization Code 8. Performing Organization Report No. FHWA/IN/JTRP-2001/8 10. Work Unit No. 11. Contrat or Grant No. SPR Sponsoring Ageny Name and Address Indiana Department of Transportation State Offie Building 100 North Senate Avenue Indianapolis, IN Type of Report and Period Covered Final Report 14. Sponsoring Ageny Code 15. Supplementary Notes Prepared in ooperation with the Indiana Department of Transportation and Federal Highway Administration. 16. Abstrat In Volume 3 of the final report, researh work onduted to investigate the behavior of fiber reinfored polymer (FRP) reinforement is summarized. This study foused on the behavior of FRP reinfored onrete strutures with an emphasis on bond and shear. For the bond investigation, three series of beam splie tests were performed on speimens reinfored with steel, glass FRP, and aramid FRP to determine the effet of the different types of reinforement on bond, raking, and defletions. The test results indiate that the use of FRP reinforement leads to lower bond strengths and, therefore, require longer development lengths. The speimen rak widths and defletions were substantially larger for FRP speimens than steel speimens due to the signifiantly lower modulus of elastiity. Analysis of the test results resulted in reommendations for modifying the empirial development length equation of ACI design ode for use with FRP reinforement. For the shear investigation, two series of beam tests were onduted on speimens reinfored with steel, glass FRP, and aramid FRP to determine the effet of the different types of reinforement on the onrete shear strength. All speimens did not ontain transverse reinforement. The test results indiate that the use of FRP reinforement leads to lower onrete shear strengths than steel reinforement for equal reinforement ross-setional areas (longitudinal reinforement perentages). Analysis of the test results resulted in reommendations for the alulation of onrete shear strength. Based on the findings of this researh, design and onstrution reommendations are provided that an be used for the design and onstrution of FRP reinfored bridge deks. 17. Key Words Corrosion, durability, fiber reinfored polymer (FRP) reinforement, onrete bridge deks, onrete bond strength, onrete shear strength. 18. Distribution Statement No restritions. This doument is available to the publi through the National Tehnial Information Servie, Springfield, VA Seurity Classif. (of this report) 20. Seurity Classif. (of this page) 21. No. of Pages 22. Prie Unlassified Unlassified 73 Form DOT F (8-69)

8 TABLE OF CONTENTS LIST OF TABLES...v LIST OF FIGURES...vi 1. INTRODUCTION Reinforement Corrosion Fiber Reinfored Plasti (FRP) Reinforement Bond Bakground Failure Modes ACI Design Provisions Orangun Equation ACI Design Provisions AASHTO Design Speifiations Researh on Bond of FRP Reinforement ACI Committee 440 Proposed Reommendations Shear Bakground Mehanisms of Shear Transfer Types of Shear Failure Fators Influening Shear Strength Shear Strength of FRP Reinfored Conrete Beams Without Stirrups OBJECTIVES AND SCOPE BOND INVESTIGATION Introdution Design of Speimens Materials Steel Glass FRP Aramid FRP Conrete Test Setup and Proedure Experimental Results General Behavior Flexural Craking Failure Appearane After Failure Test Results Beam Stiffness Crak Widths Bond Strength Data Analysis ACI Design Provisions ACI Committee 440 Proposed Reommendations AASHTO Design Provisions Crak Widths Defletions SHEAR INVESTIGATION Introdution Design of Speimens Materials...44 iii

9 4.3.1 Reinforement High Yield Strength Steel (Dywidag) Conrete Test Setup and Proedure Experimental Results General Behavior Loading Failure Craking Load Shear Strength Load-Defletion Curves Data Analysis ACI Building Code ACI Committee 440 Proposed Design Reommendations AASHTO LRFD Bridge Design Speifiations Alternative Analysis CONSTRUCTION RECOMMENDATIONS Introdution Handling And Storage Plaing & Assembling of Reinforement And Pouring of Conrete Quality Control DESIGN RECOMMENDATIONS Development Length Crak Width Defletions Shear REFERENCES...71 APPENDIX A: CONVERSTION FACTORS...74 iv

10 LIST OF TABLES Page Table 1.1: Proposed Values of C E...10 Table 1.2: Proposed K Table 3.1: Details of Bond Series...22 Table 3.2: Conrete Bath Weights Per Cubi Yard...24 Table 3.3: Average Conrete Strengths...24 Table 3.4: Bond Test Results...30 Table 3.5: Lap-Splie Bond Strengths...34 Table 3.6: X Fators Based on Experimental Results...36 Table 3.7: Proposed Values of C E...37 Table 3.8: K 2 Fators Based on Experimental Results...37 Table 3.9: Calulated Stress by Equation Table 4.1: Shear Speimen Nominal Design Values...43 Table 4.2: Mehanial properties of reinforing materials...44 Table 4.3: Conrete Bath Weights Per Cubi Yard...45 Table 4.4: Average Conrete Strengths...45 Table 4.5: Shear Test Results...51 Table 4.6 Analysis Results...55 v

11 LIST OF FIGURES Page Figure 1.1: Prodution of a GFRP Bar...2 Figure 1.2: Bearing Fores Applied to Reinforement...3 Figure 1.3: Splitting Craks...3 Figure 1.4: Types of Bond Tests...9 Figure 1.5: Shear Transfer Mehanisms in A Beam without Transverse Reinforement...12 Figure 1.6: The Influene of Arh Ation on Shear Strength...13 Figure 1.7: D and B-Regions in a Beam...13 Figure 1.8: Examples of No Arh Ation...14 Figure 1.9: Examples of Arh Ation...14 Figure 1.10: Diagonal Tension Failure...15 Figure 1.11: Influene of ρ on u v...16 Figure 3.1: Test Setup...20 Figure 3.2: Speimen Details...21 Figure 3.3: Stress-Strain Curve for GFRP...23 Figure 3.4: Stress-Strain Curve for AFRP...23 Figure 3.5: Loading Detail...25 Figure 3.6: Comparison of Crak Propagation...26 Figure 3.7: Comparison of Craking...27 Figure 3.8: Brittle Splitting Failure...28 Figure 3.9: Splitting Failure Modes...29 Figure 3.10: Reassembled Conrete Cover...29 Figure 3.11: Series I Load vs. Defletion Plot...31 Figure 3.12: Series II Load vs. Defletion Plot...31 Figure 3.13: Series III Load vs. Defletion Plot...32 Figure 3.14: Series I Bar Stress vs. Avg. Crak Width...33 Figure 3.15: Series II Bar Stress vs. Avg. Crak Width...33 Figure 3.16: Series III Bar Stress vs. Avg. Crak Width...34 Figure 3.17: Maximum Crak Width Comparison...40 Figure 3.18: Series I Load vs. Defletion Plot...41 Figure 4.1: Test Setup...43 Figure 4.2: Cross Setion Details...43 Figure 4.3: Stress-Strain Curve for Dywidag Steel Bars...44 Figure 4.4: Supports...46 Figure 4.5: Series 1 Test Setup...46 Figure 4.6: Series 2 Test Setup...47 Figure 4.7: Instrumentation Plan for Series Figure 4.8: Speimen V-G2-2 During Two Different Load Stages...49 Figure 4.9: Speimen V-D-2 Prior to Failure...49 Figure 4.10: Speimens after Failure...50 Figure 4.11: Tension Crak on the Top Surfae After Failure (V-S-2)...50 Figure 4.12: Shear strength of speimens...52 Figure 4.13: Load defletion urves for Series 1 speimens...53 Figure 4.14: Load defletion urves for Series 2 speimens...53 Figure 4.15: Comparison of Strength Calulations by ACI Code...55 Figure 4.16: ACI Committee 326 design urve development...56 Figure 4.17: Comparison of Strength Calulations by ACI 440 Proposed Reommendations...57 Figure 4.18: Comparison of Strength Calulations by 1999 AASHTO LRFD Code...59 Figure 4.19: Influene of ρ on u v...61 Figure 4.20: Comparison of Strength Calulations by The Proposed Method...61 Figure 5.1: Epoxy Coated Steel Chairs...64 Figure 5.2: FRP Bars Seurely Tied Down to Prevent Floating...65 vi

12 Figure 5.3: Plasti Coated Steel Tie-Wires...65 Figure 5.4: Casting Operations...66 Figure 5.5: Dek Finishing...66 Figure 5.6: Internal Vibrator Compation...67 Figure 5.7: Damage to glass FRP bars due to poor onstrution planning...67 vii

13 1. INTRODUCTION 1.1 Reinforement Corrosion Corrosion of steel reinforement is a serious problem in reinfored onrete strutures. As steel orrodes, it undergoes large volume expansion that auses exessive tensile stresses in the onrete that an eventually ause spalling. Not only does reinforement orrosion lead to aestheti problems in strutures, but it an also affet strutural performane and ultimately redue servie life. Corrosion of steel reinforement an lead to a loss of flexural tensile strength due to a loss of steel ross-setion. At the same time, orrosion of steel an weaken the surrounding onrete, destroy bond, and lead to a loss of flexural ompressive strength. Reinfored onrete exposed to water is highly suseptible to orrosion, and the problem worsens when it is exposed to salts, whih is the ase in marine environments, parking garages, and bridge deks. There are many methods for reduing the risk of reinforement orrosion in reinfored onrete suh as dereasing the permeability of the onrete, inreasing the onrete over, waterproofing the onrete, and oating the reinforement. In many instanes, more than one of the methods mentioned above are employed. Providing a barrier to eletrially isolate the reinforement, suh as applying an epoxy oating on steel reinforement, is a ommon appliation. The use of epoxy-oated reinforement does have its drawbaks suh as affeting bond performane, whih is addressed in many design odes. In addition, this barrier is not foolproof and an often be ompromised by onstrution praties (Samples, 1998). The use of non-metalli reinforement is gaining a signifiant amount of attention by the engineering and onstrution ommunities. Fiber reinfored plasti or polymer (FRP) reinforement is by nature eletrially isolated, whih is of tremendous benefit both eonomially and struturally. FRP reinforing bars have a muh higher strength to weight ratio than steel bars. FRP is generally one-fourth the weight of steel and has ultimate tensile strength muh higher than the yield of onventional ASTM Grade 60 steel reinforement. It should be noted that the behavior of FRP bars is very different than that of steel bars and the behavior is highly dependent on fiber type. FRP reinforement is linear elasti up to failure, whih poses a problem as far as strutural dutility is onerned. Another onern is that the modulus of elastiity of FRP bars is signifiantly lower than that of steel. Bars reinfored with glass fibers (GFRP) typially have a modulus of elastiity of 20-25% of steel, whereas arbon fiber reinfored bars (CFRP) have a modulus muh loser to steel (75%). Aramid FRP (AFRP) bars have a modulus of elastiity anywhere from that of GFRP to slightly above that of steel, depending on the matrix. At present there are no manufaturing standards for FRP reinforement; therefore, their harateristis are highly variable from produer to produer. It is due to these differenes that the design of FRP reinfored onrete must be approahed with a great deal of aution. Most design equations used for reinfored onrete are based on laboratory tests using steel reinforement. Naturally, design of FRP reinfored onrete should utilize equations that reflet test data for FRP reinfored speimens. For that reason there has been a onsiderable amount of researh involving FRP reinforement. 1.2 Fiber Reinfored Plasti (FRP) Reinforement There are many benefits to the use of FRP reinforement in strutural appliations. Corrosion resistane and high ultimate tensile strength are two of the most important harateristis enouraging its use. The low density of the material is of importane as well. FRP bars do not fatigue when stressed to less than half their ultimate apaity. Finally, FRP bars have a oeffiient of thermal expansion that is lose to that of onrete, whih is important in situations where a large thermal gradient is expeted (Ehsani, 1992). There are several obstales, however, to overome in using FRP reinforement in onrete. The most ritial of these is the fat that the material is linear elasti up to failure. Properly designed steel reinfored onrete is under-reinfored so the reinforement will yield (dutile behavior) before the onrete is rushed in ompression. In effet, under-reinforing a struture with FRP will result in a brittle failure of the reinforement. The resulting lak of dutility in the FRP reinfored struture provides no warning signs of distress and ompromises the safety of the publi. Over-reinforing a struture with FRP will result in onrete failure in ompression, whih is a brittle failure mode as well. As stated earlier, the modulus of elastiity (Young s Modulus, E) of most FRP reinforement is lower than that of steel. The result is larger defletions and rak widths for a given stress. While neither result jeopardizes the safety of the publi, they are servieability issues that must be onsidered. These servieability onerns ultimately limit the stress that an be developed in the reinforement; therefore, in many ases the ultimate strength of the bar will not affet design. Due to these differenes, there is adequate evidene that the redued modulus of elastiity ould also affet the bond and shear performane of FRP reinfored onrete. 1

14 The ost of reinforement is another issue that may restrit wide spread use. Based on stiffness, CFRP reinforement is 10 to 25 times the ost of steel reinforement (Jerrett, 1995). The same is true of GFRP and AFRP bars. Though there is ertainly life-yle ost benefits assoiated with FRP reinforement, many projets may not be able to afford the apital expenditure required to inorporate this type of reinforement. Fiber reinfored plasti reinforement is not subjet to standards governing its prodution. This has led to a plethora of different fiber types, resin matries, and deformations or oatings. Although there are a multitude of different FRP bars on the market, most ommerially available bars are produed by a pultrusion proess. An example of this proess is shown in Figure 1.1 for the prodution of a GFRP bar. Glass Fibers Forming and Curing Die FRP Reinforing Bar Resin Bath Glass Fibers Figure 1.1: Prodution of a GFRP Bar The prodution of most FRP bars involves pulling a multitude of fibers through a resin bath as shown on the left side of the figure. The fiber-resin matrix ontinues through a series of forms, giving it a irular rosssetion, while a uring ompound is applied. The FRP bar then passes through rollers that produe the desired diameter and help prevent die wiking (formation of voids in the bar). Finally, surfae indentations an be applied to the bars before the matrix hardens. The proess in Figure 1.1 shows a helial wrapping of glass fibers around the exterior of the bar. This is one of many methods that FRP reinforing bar produers have developed to help improve bond harateristis. There are three main methods for improving the bond harateristis: forming deformations or indentations onto the bars, wrapping fibers around the exterior of the bar, and oating the surfae of the bar with sand. The first two methods aim to improve bond by inreasing bearing, whereas the latter improves bond by inreasing frition. 1.3 Bond Bakground As Ersoy (Ersoy, 1994) states, reinfored onrete is the happy marriage of reinforing bar and onrete. A reinfored onrete member generally relies on onrete to resist flexural ompressive stress. When the onrete an no longer do so, reinforement resists flexural tensile stress. Any strutural system that is based on reinfored onrete, whether the onrete is reinfored with steel or fiber reinfored polymers (FRP), relies on the bond between the onrete and the reinforing material for the suessful transferal of stress from onrete to reinforement. A system devoid of adequate bond will fail, likely in a very brittle manner. In order to design satisfatory reinfored onrete strutures and make for a happy marriage, appropriate development lengths must be provided. Bond stresses are present in reinforement whenever there is a hange in stress along the length of the reinforement. There are several mehanisms that ontribute to bonding ation and allow the reinforement to resist the fores or stresses imparted on it. First, there is adhesion between the onrete and the reinforing material. The amount of adhesion is highly dependent on the harateristis of the onrete and the reinforing material in question. The seond mehanism is frition between the onrete and reinforement. This mehanism is dependent on the onrete quality, surfae texture of the reinforement, and the presene of onfining (radial) pressure. If a smooth, ylindrial length of reinforement embedded in onrete was pulled to a stress that exeeded the bond stress ontributed by those two mehanisms alone, the system would experiene slip and would fail immediately. Adhesion and frition generally do not generate a signifiant amount of bond stress. Therefore, deformations are manufatured on reinforement, thus generating a third bond mehanism, whih is referred to as bearing or mehanial interlok (MaGregor, 1997). The fores due to bearing on reinforement deformations are shown in Figure 1.2. From statis it an be onluded that equal and opposite fores bear on the onrete. 2

15 The bearing of onrete on reinforement deformations auses two fore omponents, a longitudinal and radial omponent. If the longitudinal omponent is large enough, the onrete between suessive deformations shear, resulting in a pullout failure. Alternatively, when reinforement with low shear strength is used, the deformations may shear off the reinforing bar. If the radial omponent of bearing is large enough, the onrete surrounding the reinforement an split along the length of the bar. The raks aused by this phenomenon are alled splitting raks, whih an propagate to the surfae of the onrete and are potential failure planes. The resultant of the longitudinal and radial omponents is inlined at an angle, β, whih is dependent on the materials involved. β Figure 1.2: Bearing Fores Applied to Reinforement Failure Modes There are generally two modes of bond failure. The first is often termed a pullout failure, whih is aused by high longitudinal stress. The seond is termed a splitting failure, whih is aused by high radial stress. A splitting failure is onsidered a lower bound failure mode as it an our at a lower reinforement stress than a pullout failure. The failure planes assoiated with a splitting failure are shown in the ross-setions illustrated in Figure 1.3. The reinforement stress at whih splitting failure ours is primarily a funtion of the minimum distane between reinforing bars in a plane or the onrete over provided, the tensile strength of the onrete, and the average bond stress (MaGregor, 1997). (a.) Side and top splitting (b.) Side splitting (.) Top splitting Figure 1.3: Splitting Craks It is important to note that the amount of over affets the stress that a reinforing bar an develop. For a given development length, providing smaller overs results in lower stress levels being developed by the reinforement at failure. In under-reinfored members, additional onrete over does not provide additional flexural strength beause the steel yields before the onrete fails in ompression. Consequently, designers typially provide the minimum amount of over allowed by the governing design ode. Therefore, the minimum over requirement often ontrols the maximum bond stress that an be developed. The minimum over is generally speified by building odes and is based on exposure onditions. Sine the onrete often serves as the only soure of orrosion protetion, onrete exposed to weather requires more over than onrete that is not. However, sine FRP reinforement does not need a orrosion barrier, it is likely that the amount of over required would be similar to that required of steel reinfored onrete that is not exposed to weather. 3

16 1.3.2 ACI Design Provisions The development length design provisions used presently in ACI (ACI 318, 1999) are remarkably different from those used 25 years previously. The ontemporary design equation at that time was the equation in ACI (ACI 318, 1971), whih was in the following form: where: l A f d b y = development length of = ross = speified l d - setional ab f y = 0.04 (Eq. 1-1) f ' area of yield stress of f' = speified ompressive strength of onrete,(psi) reinforement,(in.) reinforing bar,(in. reinforement,(psi) 2 ) The equation was designed so that the alulated development length develops 5/4 of the yield stress of the reinforement, thus helping insure dutility in the design. In addition, the derivation was based on a maximum bond stress of 800 psi. The 5/4 term is hidden in the 0.04 fator on the right side of the equation. No φ fator is inluded in the equation beause a φ fator of 0.9 is already inluded in the flexural design (ACI 318, 1971). It is important to note that Equation 1-1 does not onsider the effet of onrete over or presene of transverse reinforement, whih both play an important role in splitting behavior Orangun Equation In a paper published in the Marh 1977 ACI (Amerian Conrete Institute) Journal, Orangun, Jirsa, and Breen (Orangun, 1977) proposed a development length design equation based on a non-linear regression analysis of available test data. All test data inluded speimens that were reinfored with steel. The design equation suggested by Orangun et. al. aounts for the effet of onrete over and the presene of transverse reinforement and is shown below: where: K l d 10,200 db ld = 12 in. C f ' ( K tr ) φ d d b = diameter of reinforing bar,(in.) b = development length of reinforment,(in.) f' = speified ompressive strength of onrete,(psi) C = lesser of lear over or 1/2 lear spaing between bars,(in.) tr Atr f yt = 600 s d b 2.5 φ = apaity redution fator = A = area of transverse reinforement normal to C,(in ) f tr yt = speified yield stress of transverse reinforement,(psi) s = spaing of transverse reinforement,(in.) 2. 5 (Eq. 1-2) Equation 1-2 was developed for Grade 60 reinforement (f y = 60 ksi), therefore, f y does not appear in the equation. The authors reommend the development length alulated in Equation 1-2 be multiplied by the appliable fators given below: 4

17 Grade 40 Reinforement Grade 75 Reinforement Top Reinforement (12 in. of onrete below bar) Wide Spaing, C s /(C b d b ) > Wide Spaing, C s /(C b d b ) > Reinforement in a flexural member in exess of that required... (A s,req /A s,pro ) where: C s = half lear spaing between bars or splies or half available onrete width per bar or splie resisting splitting in the failure plane, in. C b = lear bottom over to main reinforement, in. d b = diameter of main reinforement, in. A s,req = area of steel required, in 2 A s,pro = area of steel provided, in 2 While the equation aounts for over and transverse reinforement, it beomes an abstrat and umbersome design provision. Comparison of Equation 1-1 and Equation 1-2 indiates an inrease in development length of 10-25% for minimum over and spaing over that required by Equation 1-2, whereas development lengths of large diameter bars in the presene of inreased over and transverse reinforement derease by as muh as 60% (Orangun, 1977) ACI Design Provisions Signifiant hanges to the development length equation in the ACI 318 Building Requirements have been made reently, speifially in 1989 and While a historial bakground of the urrent equation is not neessary, it should be said that the urrent equation is based on the reommendations proposed by Orangun et. al. (Orangun, 1977), but is not quite as abstrat or umbersome to the designer as the original equation. The development length equation for ACI (ACI 318, 1999) is the following: where: l d d b = 3 40 f y f ' αβγλ + K d b tr (Eq. 1-3) d f' = speified ompressive strength of onrete,(psi) α = reinforement loation fator K β = oating fator A f l f d b y = development length of reinforment,(in.) = diameter of reinforing bar,(in.) = speified yield strength of reinforement,(psi) = lesser γ = reinforement size fator λ = lightweight aggregate fator tr tr yt Atr f yt = transverse reinforement index = 1500 s n = area of = speified s = spaing of transverse reinforement,(in.) n = number of lear over or 1/2 lear spaing between bars,(in.) transverse reinforement within s rossing the 2 failure plane,(in ) yield stress of transverse reinforement,(psi) of bars being developed along the plane of splitting

18 The values of the modifiation fators (Greek letters) are the following: α = Bar Loation Fator Horizontal reinforement with more than 12 in. of fresh onrete ast below the development length Other reinforement β = Coating Fator Epoxy-oated bars with over less than 3d b, or lear spaing less than 6d b All other epoxy-oated bars Unoated reinforement γ = Bar Size Fator No. 6 and smaller bars No. 7 and larger bars λ = Lightweight Aggregate Conrete Fator When lightweight aggregate is used However, when f t is speified, λ shall be permitted to be taken as 6.7 f ' f t but not less than When normal-weight onrete is used The value of f t is speified as the average splitting tensile strength of lightweight aggregate onrete in psi. The produt of αβ need not be greater than 1.7. The minimum development length of a straight bar is speified as 12 in. The ode limits the value of ( + K tr ) / d b to a maximum of 2.5, whih as the ommentary indiates, guards against a pullout failure. Sine the ode aknowledges the effet of over on splitting behavior, it is important to report the minimum over and spaing requirements speified in the ode. For any onrete exposed to earth or weather, the minimum over shall be: No. 6 through No. 18 bars 2 in. No. 5 bar or smaller 1 ½ in. For onrete not exposed to weather: Beams and Columns 1 ½ in. The minimum spaing between parallel bars in a layer is speified as d b, but not less than 1 in AASHTO Design Speifiations The Amerian Assoiation of State Highway and Transportation Offiials (AASHTO) produes Standard Speifiations For Highway Bridges and AASHTO LRFD Bridge Design Speifiations, both of whih inlude pertinent information required for the design of highway bridges. Sine FRP reinforement learly has appliations benefiial to the servie life of highway superstrutures, the speifiations of AASHTO should be taken into onsideration. The development length provisions set forth in the AASHTO Standard Speifiations, Sixteenth Edition (AASHTO, 1996) (AASHTO LRFD Bridge Design Speifiations (AASHTO, 1998) development length provisions are the same) are found in Chapter 8 as follows: No. 11 bars and smaller: but not less than: l d Ab f y = 0.04 (Eq. 1-4) f ' d b f y 6

19 where: l d A f f' b y = development length of reinforement,(in.) = ross - setional area of reinforing bar,(in = speified yield stress of reinforement,(psi) = speified ompressive strength of onrete,(psi) 2 ) Clearly, Equation 1-4 is the same as Equation 1-1. However, the ode inludes multipliation fators for a number of different onditions. The following fators shown are to be multiplied by the development length alulated in Equation 1-4: Top reinforement with more than 12 in. of fresh onrete ast below the reinforement Lightweight aggregate onrete 6.7 f ' when f t is speified... but not less than 1.0 ft When f t is not speified all lightweight onrete sand lightweight onrete Linear interpolation may be used when partial sand replaement is used. Bars oated with epoxy with over less than 3d b or lear spaing between bars less than 6d b All other ases The produt obtained when ombining the fator for top reinforement with the appliable fator for epoxy oated reinforement need not be taken greater than 1.7. Reinforement being developed in the length under onsideration is spaed laterally at least 6 in. on enter with at least 3 in. of lear over measured in the diretion of spaing Anhorage or development for reinforement strength is not speifially required or reinforement in flexural member is in exess of that required by analysis (A s required)/(a s provided) Reinforement is enlosed within a spiral of not less than ¼ in. in diameter and not more than 4 in. pith The development length alulated inluding multipliation fators shall not be less than 12 in. The provisions state that the minimum distane between bars shall be no less than 1.5 bar diameters, 1.5 times the maximum aggregate size, or 1 ½ in. The over requirements are summarized below: Conrete exposed to earth or weather: Primary reinforement Stirrups, ties, and spirals Top reinforement exposed to deiing salts Conrete not exposed to earth or weather: Primary reinforement Stirrups, ties, and spirals 2 in. 1 ½ in. 2 ½ in. 1 ½ in. 1 in. 7

20 1.3.6 Researh on Bond of FRP Reinforement A literature review of researh performed on bond in reinfored onrete identified two widely used bond test methods, pullout and beam tests. The majority of researh performed on the bond of FRP reinforement to onrete has been in the form of pullout tests. Pullout tests, while relatively inexpensive and easy to perform, subjet onrete around the bonded rebar to onfining pressure, whih an provide an overestimate of bond strength. Beam tests are onsidered to provide a more realisti representation of bond behavior; and therefore, bond strengths that are more aurate. Examples of eah type of test setup an be seen in Figure 1.4. Pullout tests (Figure 1.4(a.)) do not generate fores that are an aurate refletion of the onditions that a reinforing bar in reinfored onrete is subjeted to in normal servie onditions. In reinfored onrete, bending is generally the mehanism that introdues tension into the reinforement, not an axial load. Between raks in the onrete, the tensile stress is shared between the reinforement and the onrete, and the stress is transferred by what is alled in-and-out bond stresses. Due to the fat that pullout tests generate onfining pressure on the onrete surrounding the reinforement, the onrete is not free to rak; and onsequently, no in-and-out bond stresses an develop (MaGregor, 1997). Tests of 48 beam and 18 pullout speimens by Ehsani et al. (Ehsani, 1992) indiate lower bond strengths for beam speimens than pull-out speimens. Zenon et al. (Zenon, 1998) observed the same phenomenon. Therefore, test results from pullout tests represent upper-bound bond strength. Despite this drawbak, pullout tests are valuable in investigating the possible bond benefits of partiular aspets of different FRP reinforement. Comparing different deformation heights and spaing as well as different surfae oatings are appropriate uses of pullout tests. The onlusions of several of the testing programs involving pullout tests are desribed below, and although not speifially addressed, other literature (Boothby, Cosenz, Jerrett) has indiated similar findings. Nanni et al. (Nanni, 1998, Nanni and Al-Zahrani, 1995, Nanni and Bakis, 1995, Nanni and Boothby, 1995) and Larralde et al. (Larralde and Silva-Rodriguez, 1993, Larralde and Mueller-Rokholz, 1998, Larralde and Silva- Rodriguez, 1994) have onduted pullout tests on a variety of FRP bars. Both have reported that the failure was governed by shearing of the bar deformations. In pullout tests with steel reinfored speimens, the failure is generally governed by rushing of the onrete in front of the deformations. Nanni onluded that the failure mode in FRP reinfored speimens indiated that the resin was the ontrolling parameter and found that epoxy based resins performed better than vinyl-ester resins. Failure of the deformations before reahing the ultimate bar apaity is a serious onern in utilizing FRP reinforement. Karlsson (Karlsson, 1997) and Tepfers et al. (Tepfers, 1998) reported three modes of failure in pull-out tests performed with FRP bars. Crushing of the onrete, shearing of the reinforement deformations, and a ombination of the two modes was observed for onrete strengths of approximately 4,200 psi, 8,500 psi, and 6,200 psi, respetively. Apparently, the benefit of high onrete strengths on bond strength an be undermined by the lak of reinforement deformation shear strength. Tests by Malvar (Malvar, 1994, Malvar, 1995) indiate that deformations formed simultaneously with the bar performed better than those that were applied afterward. A beam test in whih the reinforement is splied in a region of onstant moment is shown in Figure 1.4(b.). It should be noted that this is only an example of a beam test setup, as many different setups have been reported in the literature. The major findings of beam tests by Zenon et al. (Zenon, 1998), Ehsani et al. (Ehsani, 1992), Kanakubo et al. (Kanakubo, 1992, Kanakubo, 1993), Makitani et al. (Makitani, 1992), and Tighiouart et al. (Tighuart, 1998) are summarized below: Bar stresses inrease as development length inreases, but the relationship between development length and bar stress developed is not linear. When the mode of failure is splitting, an inrease in onrete strength results in a pronouned inrease in bond strength. Development lengths between 1.3 and 1.9 times the development length required of a omparable steel speimen are required for FRP bars. Bond strengths are higher for smaller diameter bars. A top bar effet was observed indiating a redution in bond strength ranging from 1.1 to 1.3 for bars ast above more than 12 in. of onrete. There is no onsensus regarding the effet of the modulus of elastiity of the fibers in the reinforement on bond performane. Limited investigations have been onduted regarding the deterioration of bond in aggressive environments. Al-Dulaijan et al. (Al-Dulaijan, 1996) found signifiant bond degradation after subjeting speimens 8

21 to high ph and temperature. This effet should be investigated further sine onrete is an alkali environment (high ph), and in many instanes is subjeted to high temperatures. Mashima and Iwanmoto (Mashima, 1992) onduted pullout tests on speimens made with several types of FRP after performing up to 600 yles of freezing and thawing. Their findings indiate that Aramid fiber bars do not perform well after repeated freeze-thaw yles, a ondition that an be experiened in bridge deks. Bond strength redution was as muh as 50% in 500 yles. Conrete Speimen Bonded Length FRP Reinforement Reation Reation Pull-out Load (a.) Pullout Test Setup (Adapted from Referene 15) P P Reinforement d Splie Elevation Cross Setion (b.) Beam Test Setup Figure 1.4: Types of Bond Tests ACI Committee 440 Proposed Reommendations ACI Committee 440 has proposed the Guide for the Design and Constrution of Conrete Reinfored with FRP Bars (ACI 440, 2000) based on researh and experiene generated to this point. While the guide has not been 9

22 endorsed outside of the ommittee or distributed by ACI for publi use, it is likely that many of the reommendations will be eventually published and inorporated into a design doument. Chapter 11 of the proposed guide involves the development of FRP reinforement in onrete. For splitting ontrolled failure, the ommittee has suggested the following equation: 2 d b f fu lbf = K 2 (Eq. 1-5) f ' where: l = development length required,(in.) K d bf f C f f' b 2 fu E fu = onstant determined experimentally = reinforing bar diameter,(in.) = design tensile strength of reinforement onsidering environmental fators,(psi) = C = environmental redution fator, dependent on fiber type and exposure onditions * = guaranteed tensile strength,(psi) = speified onrete strength,(psi) The value of f fu is defined as the mean tensile strength of a sample of test oupons minus three standard deviations. The ommittee proposes values of C E based on fiber type and exposure onditions, whih are shown in Table 1.1. E f fu * Exposure Condition Conrete not exposed to earth and weather Conrete exposed to earth and weather Table 1.1: Proposed Values of C E Fiber Type Environmental Redution Fator, C E Carbon 1.0 Glass 0.8 Aramid 0.9 Carbon 0.9 Glass 0.7 Aramid 0.8 The proposed doument does not reommend a speifi value for K 2, but the K 2 fators suggested by several investigators are shown in Table 1.2. The equation is in the same form as the development length equation in the AASHTO speifiation and ACI The π/4 term that is inluded in the bar area of Equation 1-1 is absorbed by the K 2 term in Equation 1-5. Table 1.2: Proposed K 2 Investigator K 2 Pleimann (Pleimann) 1/19.4 (GFRP) 1/18.0 (AFRP) Gao (Gao, 1998) 1/16.7 Ehsani (Ehsani, 1996) 1/21.3 Tighiouart (Tighiouart, 1998) 1/

23 For pullout ontrolled failure, the ommittee has suggested the following equation: d b f fu l bf = (Eq. 1-6) 2700 The form of this development length equation (Equation 1-6) is unique in omparison to the others disussed thus far. Aording to this reommendation, the development length is not dependent on f '. The design reommendations state, if the onrete over exeeds two bar diameters, a pull-out failure is more likely (ACI 440, 2000). The ommittee does not speify whih equation should be used in design, but does reommend a modifiation fator be used with Equation 1-6. This modifiation fator, alled a onrete over modifiation fator, is reommended in instanes in whih the over is between d b and 2d b, and varies linearly from 1.5 to 1.2. For over greater than 2d b the modifiation fator beomes 1.0. The ommittee reommends that onrete over should be no less than d b. The development length alulated in either Equation 1-5 or 1-6 should be multiplied by a bar loation modifiation fator of 1.3 if the reinforement layer is above more than 12 in. of onrete. 1.4 Shear Bakground Mehanisms of Shear Transfer A entury of researh has been onduted on shear in reinfored onrete beams without transverse reinforement. However, an understanding of shear behavior is still limited. Therefore design methods are based on empirial formulations based on laboratory tests and observations. A 1973 ACI-ASCE Committee 426 (Shear Strength of Reinfored Conrete Members) report (ACI 426, 1973) identified four mehanisms of shear transfer; namely, shear stresses in the unraked onrete, interfae shear transfer, dowel ation of the longitudinal reinforing bars, and arh ation. Sine that report was published, the 1998 ACI-ASCE Committee 445 (Shear And Torsion) report (ACI 445, 1998) has identified a fifth mehanism of shear transfer, residual tensile stresses transmitted diretly aross raks. These five mehanisms are illustrated in Figure 1.5 and will be disussed in more detail in the following. The labels in the figures oinide with the headings in the disussion that follows. a) Shear Stresses in Unraked Conrete: This shear transfer mehanism is the simplest and ours in unraked members and unraked portions of raked reinfored onrete members (Figure 1.5). The interation of shear stresses with ompressive and/or tensile stresses on the unraked onrete produes a omplex state of stresses. Depending on whether the prinipal tensile or ompressive stresses produed from this stress field reah the orresponding strength of onrete, failure may our either by inlined raking (tensile strength reahed) or rushing of the onrete (ACI 426, 1973) (ompressive strength reahed). MaGregor (MaGregor, 1997) states that this omponent may be apable of arrying approximately 30% of the total shear fore. b) Interfae Shear Transfer: This shear transfer mehanism, labeled as (b) in Figure 1.5, relies on frition along the inlined rak interfae whih develops as the two rak surfaes slide relative to eah other. Physially the aggregates protruding from the rak surfae provide resistane against this slip. The mehanism was widely aepted as a shear transfer mehanism after it was desribed and reognized in the 1973 ACI-ASCE Committee 426 report (ACI 426, 1973), based on researh by Fenwik and Paulay (Fenwik, 1968), Mattok and Hawkins (Mattok, 1972), and Taylor (Taylor, 1970). The ability of this mehanism to transfer shear depends on three fators: rak normal stresses, rak widths, and the amount of rak slip (ACI 426, 1973). As the rak width and slip inrease the ability of the rak interfae to transfer shear dereases. In addition, ompressive stresses normal to the rak inrease the shear transfer along the rak while tensile stresses derease the shear transfer. Ersoy (Ersoy, 1994) reports that measurements on test beams without web reinforement by indiated that 50 to 70% of the total shear might be resisted by interfae shear transfer. 11

24 (e) (d) (a) (b) () Figure 1.5: Shear Transfer Mehanisms in A Beam without Transverse Reinforement ) Dowel Ation: Where longitudinal reinforement rosses an inlined rak, it serves as a dowel transmitting shear fores aross the rak (mehanism () in Figure 1.5). When these shear fores, ombined with radial fores developed by bond fores exeeds the tensile strength of the area of onrete supporting the bars, splitting along the longitudinal reinforement ours. In a beam without transverse reinforement, the strength of this ation depends on the area and tensile strength of the onrete supporting the dowel bars and the rak width at the level of the dowel (Vintzeleou, 1986). As the rak width inreases and/or the strength of onrete support under the bars dereases the amount of dowel shear that an be transferred dereases. It is very diffiult to quantify the amount of dowel fore that an be ativated in any given situation; therefore, there are only semi-empirial methods treating very simple ases involving one bar are available (Vintzeleou, 1986). Due to the diffiulty of quantifying dowel ation separately, in general, design odes treat the dowel ontribution to shear strength impliitly. d) Residual Tensile Stresses aross Craks: When onrete is loaded in diret tension with a displaement ontrolled atuator, a signifiant softening branh is obtained after the peak tensile stress is reahed (Gopalaratnam, 1985, Reinhardt, 1986). This softening branh is attributed to residual tensile stresses aross the rak after onrete raks. Residual tensile stresses an be explained as follows. When onrete raks, a lean break does not our. Small piees of onrete bridge the rak and ontinue to transmit tensile fore up to rak widths in the range of in. (ACI 445, 1998). Reinek (Reinek, 1991) has found that residual tensile stresses aross inlined raks an provide a signifiant perentage of shear resistane for very shallow members (for depths less than about 4-in.), where the width of flexural and diagonal tension raks is small. e) Arh Ation: Saint Venant s priniple suggests that a loal disturbane suh as a onentrated load or reation will dissipate in approximately one beam depth from the point at whih it is applied. Shlaigh et al (Shlaigh, 1987) alled these regions in a beam as D-regions, where the D stands for disontinuity or disturbed. Therefore, shear transfer by arh ation (labeled (e) in Figure 1.5) is most pronouned in reinfored onrete beams with shear span to effetive depth (a/d) ratios less than 2.5, where there is overlapping of the D-regions. In Figure 1.6, nominal shear strength of reinfored onrete beams without transverse reinforement vs. a/d ratio was plotted to illustrate the effet of arh ation. Figure 1.6 indiates that there is a signifiant inrease in the nominal shear strength for a/d ratios less than approximately 2.5 to 3. This inrease is attributed to inreased arh ation due to overlapping of D-regions in the beams. 12

25 Vult f b w d a/d Figure 1.6: The Influene of Arh Ation on Shear Strength Any region, whih is not a D-region, is alled a B-region, where the B stands for beam or Bernoulli. A very simple illustration of these regions is provided in Figure 1.7. Arh ation is different from the other transfer mehanisms beause it does not transfer shear along the rak interfae. As illustrated in Figure 1.5, arh ation transfers the shear to the supports through a ompression strut extending between the onentrated load and the supports. a 1 a 2 d D D D B D Figure 1.7: D and B-Regions in a Beam For arh ation to our, it is neessary for load to be applied at the top of the beam while it is supported at the bottom by non-yielding supports, Figure 1.7. Arh ation annot be mobilized with support or loading arrangements shown in Figure 1.8. In addition, a horizontal tensile fore to balane the ompression strut must be developed at the base of the arh. This horizontal fore an be developed by well-anhored longitudinal reinforement loated at the tension side of the beam. Shear transfer in B-regions is primarily attributed to the other four mehanisms disussed previously. Therefore, the shear strength and behavior of longer beams (a/d > 2.5), where B-regions exist, is determined by beam ation mehanisms and their strengths as previously disussed. 13

26 (a) Loading (b) Support Figure 1.8: Examples of No Arh Ation Types of Shear Failure Modes of shear failures for reinfored onrete beams of retangular ross setions without transverse reinforement, with a/d ratios greater than 1, and with properly anhored longitudinal reinforement will be disussed in this setion. A beam with these qualities will fail either by rushing or raking of onrete in the ompression zone (ACI 445, 1998). Craking in the ompression zone ours on a horizontal plane due to the ombined effet of transverse tensile stresses and ompressive stresses exeeding the tensile strength of onrete in this region. Conrete over splitting along the longitudinal bars in tension may or may not aompany failure in the ompression zone. Shear failures, whih involve raking of onrete in the ompression zone, are termed diagonal tension failures (DT), and those, whih involve rushing in the ompression zone, are termed shear ompression failures (SC). Experiments reveal that the ontrolling mode of failure depends primarily on the a/d ratio and longitudinal reinforement ratio, ρ. Kani (Kani, 1979) found that the onrete strength did not signifiantly influene the mode of failure ( f ranged from 2500 to 5000-psi.). These failure modes are desribed in detail as follows: a) Shear Compression Failure (SC): Shear ompression failures our in reinfored onrete beams ontaining typial amounts of reinforement (ρ = 1%-2%) for a/d ratios ranging from 1.0 to 2.5. The inlined rak is a ontinuation of a flexural rak for the upper end (a/d = 2.5) of the given a/d ratios. For a/d ratios loser to 1.0, the inlined rak may form by itself independent of the flexural raks. As the applied load is inreased, the inlined rak propagates deeper into the beam and may penetrate the ompression zone. A reinfored onrete beam without transverse reinforement, whih is loaded with a onentrated load in the enter of a simple span, with fully developed inlined raks, is illustrated in Figure 1.9(a). One the inlined raks are fully developed as illustrated, the bond transfer annot our between the reinforement and onrete between these raks. Therefore the onrete between these two inlined raks an be assumed to be non-existent. The beam essentially transforms into an equivalent tiedarh, whih an be visualized as shown in Figure 1.9(b). The formation of inlined raks, however, does not result in failure. Considerably additional loads an be arried through redistribution of internal stresses in the beam. However, sine the inlined rak generally extends higher into the beam than a flexural rak, failure ours at a apaity less than that of the development of the flexural moment apaity. The rotations in suh a beam are onentrated in the viinity of the onrete ompression zone under the load point. With further loading these rotations result in ompressive stresses higher than the onrete an resist, resulting in failure through rushing of onrete in the viinity of the onentrated load. (a) Idealized Fully Developed Inlined Craks 2d (b) Equivalent Tied-Arh Model of Beam Figure 1.9: Examples of Arh Ation 14

27 b) Diagonal Tension Failure (DT): Diagonal tension failures typially our in relatively slender beams, where a/d is approximately in the range between 2.5 and 7, with ommon amounts of longitudinal reinforement (ρ = 1%-2%). Sine there is no overlapping of D-regions in this range of a/d ratios, shear is arried through beam ation mehanisms previously disussed. One an inlined rak forms in a reinfored onrete beam without transverse reinforement for a/d ratios greater than 2.5, redistribution of stresses does not our. Therefore, there is little reserve apaity after the onset of inlined raks. Diagonal tension failure follows shortly afterwards. A typial rak pattern on a beam, whih fails in diagonal tension, is shown in Figure The bolded rak reates the diagonal tension failure. The inlined portion of the rak penetrating into the ompression zone is usually a ontinuation of a flexural rak. With a small inrease of load, the rak penetrates into the ompression zone and propagates towards the onentrated load. At the same time, a splitting rak initiates from the tension side of the inlined rak and travels toward the support. It is ommon for a vertial tension rak to form in the ompression zone above the diagonal tension rak. It develops from the top of the beam down to the inlined rak. The sequene of events, however, is disputed among researhers. In partiular, there is disagreement as to whih of the two raks form first; the splitting rak on the reinforement that propagates towards the support or the diagonal rak that penetrates into the ompression zone towards the load point. Tension rak Splitting rak Figure 1.10: Diagonal Tension Failure The ranges of a/d ratios provided to separate diagonal tension and shear ompression failures are only approximate values as observed in tests. In a typial reinfored onrete beam (1% < ρ < 2%), the reinforement yields prior to shear failure beyond an a/d ratio of 7.0. The longitudinal reinforement ratio is known to hange these ranges, and for heavily reinfored speimens (ρ > 3.0) the upper end of a/d ratios for shear ompression failures may shift to 3.0 (Kani, 1979). The same is true for the upper end of a/d ratios for diagonal tension failures; the value an shift up to a value of 9.0. The modes of failures and the range of a/d ratios provided are only valid for simply supported, retangular beams. Both the mode of failure and the a/d ratio may hange for ontinuous beams and/or T-beams (Kani, 1979) Fators Influening Shear Strength The shear strength attributed to onrete, whih is named V by the ACI Building Code, is affeted by 5 primary variables (MaGregor, 1997). These variables will be disussed in more detail. a) Tensile Strength of Conrete: The stress at whih inlined raking ours is diretly related to the tensile strength of onrete. In a beam without external axial fores, tensile stresses arise from the interation of flexural and shear stresses. When the resulting priniple tensile stress from this interation exeeds the tensile strength of onrete, an inlined rak forms. Sine the formation of inlined rak is related to tensile strength of onrete, the inlined raking load an be related to the tensile strength from split ylinder tests. Sine split ylinder tests are not routinely performed in pratie, design odes generally use a relationship for the onrete tensile strength that is related to the 15

28 speified ompressive strength. Tensile strength is often related to inlined raking load of a member is ommonly a funtion of this quantity. f or 3 f. Therefore, the b) Longitudinal Reinforement Ratio: Figure 1.11 shows the relationship between the longitudinal reinforement ratio, ρ, and V for beams without transverse reinforement. 6 5 = u bd f V u v f ρ (%) Figure 1.11: Influene of ρ on v u The data is taken from Tompos (Tompos, 2000) and filtered to only inlude beams with a/d ratios greater than 2.75 and onrete strengths less than 8500-psi. The vertial axis indiates the nominal Vu shear stress in the setion at failure ( vu = bd ) normalized with f. Historially, the nominal shear stress has been used as a measure of shear strength for reinfored onrete beams without transverse reinforement. Figure 1.11 reveals that the shear strength of beams inrease as ρ inreases. The ACI building ode uses 2 f as the shear stress that produes shear failure. However, as illustrated, for reinforement ratios less than 1%, the shear stress at failure may be lower than 2 f. In a beam reinfored with low ρ, flexural raks penetrate higher into the setion and wider raks are experiened as ompared to a beam reinfored with higher amounts of longitudinal reinforement. Deeper flexural raks derease the depth of ompression zone thereby reduing the ontribution of unraked onrete to shear strength. Wider raks, on the other hand, result in a redution in the shear strength ontributions of interfae shear transfer as well as residual tensile stresses. Furthermore, the derease in the amount of longitudinal reinforement oupled with the inrease in rak widths redues dowel ation. Therefore, the longitudinal reinforement ratio is a very important fator that influenes all beam ation shear transfer mehanisms. Speial attention should be given to the range of ρ less than 1%, where the shear strength is not linearly related to ρ and signifiant redutions in shear strength an be noted. ) Shear Span-to-Depth Ratio (a/d): The shear span-to-depth ratio, a/d, has a signifiant effet on the shear strength of a beam without transverse reinforement. The a/d ratio provides a measure of the arh ation ontribution to shear strength. Deep beams with a/d ratios less than 2.5 an develop arh ation and, therefore, experiene an inrease in shear strength. In fat, beams with a/d ratios less than 16

29 2.5 an arry shear stresses signifiantly in exess of the ACI shear stress value of 2 f at failure. For a/d ratios beyond 2.5 to 3, the effet of a/d ratio on inlined raking shear and shear strength an be negleted (ACI 445, 1998). d) Size of beam: As the overall depth of a beam inreases the shear stress at inlined raking tends to derease for a given f, ρ, and a/d (ACI 445, 1998). The rak width along the depth of a beam without skin reinforement is a funtion of the distane from the main reinforement to the rak loation. The rak widths at points above the main reinforement tend to be wider for deeper beams. Wider rak widths derease the ontribution of residual tensile stress and the interfae shear transfer mehanisms to shear strength; thereby reduing ν u. Taylor (Taylor, 1972) showed that when the aggregate and the speimen are saled appropriately, the dereasing trend of shear strength with inreasing beam depth was not observed in tests of suh speimens. e) Axial Fore: In general, externally applied axial tensile fores tend to redue the shear strength, while axial ompressive fores inrease it (ACI 445, 1998). Axial ompressive stress delays the onset of flexural raking and limits the depth of rak penetration into the beam. Axial tensile stresses, on the other hand, produe the opposite effet. However, approximately the same additional inrement of shear load is required between flexural raking in the shear span and inlined raking regardless of whether there is axial fores (tension or ompression) on the speimen or not (ACI 445, 1998). Sine slender beams with a/d ratios greater than approximately 2.5 fail shortly after the onset of inlined raking, the shear strength of suh members under axial fores may hange signifiantly due to the hange in the onset of flexural raking. Axial tension or ompression, therefore, results in a derease or inrease in the shear strength (V ) of beams with a/d ratios greater than 2.5. However, The inlination of inlined raks on beams tested with the existene of axial fores is not signifiantly affeted aording to ACI-ASCE Committee 445 (Mattok, 1969) Shear Strength of FRP Reinfored Conrete Beams Without Stirrups The Modulus of elastiity (E) of some FRP bars may be lower than that of steel bars. As previously disussed, due to lower modulus of elastiity, there may be a drop in the shear strength of onrete members reinfored with FRP bars (ACI 440, 2000). The shear arrying mehanism for beam ation was illustrated in Figure 1.5. This model onsiders four shear transfer mehanisms: (1) unraked onrete ontribution, (2) aggregate interlok, (3) dowel ation, and (4) residual tensile stresses aross inlined raks. Using bars with a lower modulus of elastiity, while keeping the amount of longitudinal reinforement onstant will result in deeper and wider raks. Deeper raks translate into a smaller depth of onrete ompression zone, whih results in a redution in the shear strength ontribution from unraked onrete mehanism. Naturally, larger rak widths result in a redution in the shear strength ontributions of aggregate interlok, residual tensile stresses, and doweling ation. Doweling ation is also redued due to the lower modulus of elastiity. Therefore, there is a redution in every shear mehanism, whih should result in a redution in the overall shear strength of onrete members reinfored with the FRP bars relative to that of members reinfored with steel. Experimental researh on the onrete ontribution to shear strength of FRP bar reinfored onrete members is limited. Most test data on the onrete ontribution has been derived from tests that were not designed to study the shear strength of these members. Only three studies desribing shear failures of FRP bar reinfored onrete beams without stirrups are available in the literature. These investigations and their onlusions will be briefly summarized. Nawy et al (Nawy, 1971) studied the flexural behavior of glass FRP reinfored onrete beams. In the experiments, 20 simply supported retangular beams, 10 ontaining FRP bars as tensile reinforement were loaded with onentrated loads at the third spans. All beams were 7 in. deep by 3-1/2 in. wide with effetive depths varying from 6.25 in. to 6.5 in. The total speimen length was 78 in. and the distane between supports was 72 in. These dimensions orrespond to a/d ratios varying from 3.7 to The reinforement ratio, ρ, was varied from 0.19 % to 0.41 % to obtain flexural failures. The FRP bars were 0.12 in. diameter, smooth bars. The tensile strength was 155- ksi while the modulus of elastiity was 7,300-ksi. The onrete strength from 6x12 in. test ylinders registered strength values ranging from 4000-psi to 5000-psi. All speimens were reported to fail in diagonal tension. Sine the investigation was onduted on the flexural behavior of FRP reinfored onrete beams, no attention was given to the shear strength. 17

30 Deitz (Deitz, 1998) tested 5 full sale, two-way bridge deks reinfored with #5 glass FRP bars in tension, loaded at the third points with two onentrated loads to study the flexural behavior of FRP reinfored onrete bridge deks. The ross setion of the dek panels was 7.5 in. deep by 12 in. wide and the total length of the speimens was 9 ft-9 in. A lear over of 1 in. was used in both the top and bottom mats. Conrete strengths ranging from 4000 psi to 4500 psi were registered. 3 speimens ontained glass FRP bars as reinforement in both the top and bottom mats (FRP speimens). 2 speimens were reinfored with #5 epoxy oated steel bars in the ompression zone and #5 glass FRP bars in the tension zone (hybrid speimens). The a/d ratio was 5.8 for 2 of the FRP speimens and both of the hybrid speimens. Sine shear failures were observed in the first two FRP speimens, the a/d ratio of the third FRP speimen was lowered to 4.5 to study the effet of a/d ratio on the behavior. The nominal shear strength v u of the speimens registered values ranging from 1.25 f to 1.41 f. Mihailuk et al (Mihaluk, 1998) tested 8 simply supported one-way onrete slabs reinfored with glass FRP, arbon FRP, and steel. The slabs were loaded with two onentrated loads. No data was provided on the loation of these loads. The slabs were 39.4-in. (1-m) wide by 11.5-ft (3.5-m) long with a lear span of 9.84-ft (3- m). Two depths were investigated, 5.9-in. (150-mm) and 7.8-in. (200-mm). The bars were plaed in the tension side of the speimens with a lear over of 1.5-in. (38-mm). Bar sizes were varied (#4, #5 and #6) to ahieve reinforement ratios ranging from 0.23% to 0.96%. Glass FRP bars had a tensile strength of 100 ksi and modulus of elastiity of 6,000 ksi. Similarly, arbon FRP bars had tensile strength of 326 ksi and a modulus of elastiity of 21,300 ksi. Conrete strength varied between 8600 psi and 9600 psi at the time of testing. One glass FRP reinfored speimens, with 150-mm deep and 0.95% reinforement ratio and one with 200-mm deep and 0.77% reinforement ratio, failed due to shear failure. The glass bars, rossed by a rak extending vertially almost along the entire depth of the slab in the shear span, were ruptured at failure. It is interesting to note that the measured rak widths on this partiular rak in both beams were 9/16 in. (15 mm) wide and the ompression zone depth was measured as 9/16 in. (15 mm) prior to failure. It was onluded that the large rak widths eliminated shear transfer by aggregate interlok and the small onrete ompression zone ontributed little to the shear transfer. Therefore shear was transferred by dowel ation beyond the dowel ation apaity of the glass FRP bars, whih eventually ruptured the bars. From an analytial study, it was found that the dowel ation of the bars in these two beams varied from 7.5% to 13.8% of the tensile strength of glass FRP reinforement. It was reommended that the alulated shear strength of FRP bar reinfored onrete members be multiplied by the ratio of modulus of elastiity of FRP bars to that of steel to determine onservative shear strength estimates. This reommendation was inorporated into Provisional Reommendations for Design of Conrete Members Reinfored with FRP Reinforement (ACI 440, 2000), prepared by ACI Committee 440 (Committee on FRP Reinfored Conrete) JSCE Reommendations for Design and Constrution of Conrete Strutures Using Continuous Fiber Reinforing Materials (JSCE, 1997) disusses the shear strength of FRP bar reinfored onrete members without stirrups. This doument reports that: Previous studies indiate that the shear apaity of beam members with CFRM (FRP reinforement) used for tensile reinforement but without shear reinforement an generally be evaluated by taking into aount the axial rigidity of the tensile reinforement. V d is therefore alulated aording to the equation used for steel, allowing for the ratio of the Young s modulus of CFRM to that of steel. The term V d is the equivalent of V in the ACI Building Code. The JSCE doument indiates that the reinforement ratio term (ρ) used in alulations is multiplied by the ratio of modulus of elastiity of FRP bar to that of steel to onsider the effet of the axial rigidity of tensile reinforement. FRP and steel reinforement may differ from eah other in their physial and mehanial harateristis. These differenes are likely to hange the behavior of onrete members reinfored with FRP bars from those reinfored with steel. As indiated, the modulus of elastiity of FRP bars has the potential to signifiantly affet the shear strength of beams without stirrups. This hypothesis is supported by the limited experimental work available in the literature. Experimental work onduted on the shear strength of FRP bar reinfored onrete members without stirrups, however, is extremely limited. Therefore, further researh is neessary to develop safe but eonomial design methods for FRP reinfored onrete members under shear loads. 18

31 2. OBJECTIVES AND SCOPE The objetive of this study was to investigate bond and shear performane of onrete reinfored with FRP (non-metalli) reinforement and develop design reommendations for the development of suh reinforement in bridge deks. To evaluate the bond strength of glass and aramid FRP bars, three series of beam splie tests were ondued. In addition, the raking and defletion behavior of the test speimens were investigated. To evaluate the shear strength of beams longitudinally reinfored with FRP reinforement, two series of shear beam tests were onduted. In partiular, the effet of the axial stiffness of the tensile reinforement and the longitudinal reinforement perentage on the onrete ontribution to shear strength for beams with a/d ratios greater than 2.5 was investigated. The following report presents a summary of the experimental studies and the resulting design reommendations. Additional information regarding these studies is available in Mosley, 2000 and Tureyen,

32 3. BOND INVESTIGATION 3.1 Introdution Three FRP reinforing bars were investigated in the experimental program. Two types of Glass FRP and Aramid FRP reinforement were supplied by manufaturers for the study. Three series of beams were tested to ompare the bond strength of FRP reinforing bars of various fiber types with the bond strength of onventional Grade 60 steel reinforement. Eah series onsisted of four beams, one speimen for eah of the three types of FRP reinforement and one speimen reinfored with steel. Craking behavior and stiffness of eah speimen was observed and reorded. 3.2 Design of Speimens Eah series was designed suh that the same test setup ould be used throughout the experimental program. The speimen design was similar to that of other tests of bond strength, ommonly referred to as beam tests. The onfiguration of load points and support points were arranged to reate a onstant moment region in the entral span of the speimen (Figure 3.1). This entral span is the most ritial loation for a splie as the onstant moment region is devoid of shear fores and the entire splie length is experiening maximum tensile stress. To provide ease and safety while reording rak widths and taking photographs, the reinforement was plaed in the top of the speimens, whih were loaded at the ends and supported on the reation floor at the quarter-points. The length of the speimens was ditated by the spaing of the threaded inserts loated in the reation floor of the Kettlehut Strutural Laboratory (the inserts are loated at 6 ft intervals). Therefore, it was deided that the threaded-rods used to support the loading beam would be loated 12 ft apart. The plaement of the supports on the floor was important in two respets. First, if the distane between the support and the load point is too short, very high load is needed to produe the required failure moment. Therefore, the threaded inserts ould yield or pullout of the floor, or the steel-loading beam ould fail. Seond, a short distane between the supports will not allow a random distribution of raks in the onstant moment region, thus influening the raking behavior of the speimens. Based on alulations, it was onluded that plaing the supports 3 ft from the load points would provide adequate moment generating apaity for the range of splie lengths and bar spaings that would be tested in the experimental program. In addition, the 6 ft onstant moment region would allow for random rak generation. To aommodate the loading head, an additional 8 ¼ in. was added to eah end making the beams a total of 13 ft- 4 ½ in. The speimen dimensions are shown in Figure 3.1. P L s P 3-0 Shear Span 6-0 Constant Moment Region 3-0 Shear Span 13-4 ½ Figure 3.1: Test Setup To allow bond strength omparisons between the individual tests, it is important that the reinforement remains in its elasti deformation range. The speimens of eah series were designed aordingly. The speimens were reinfored for negative moment by three longitudinal #5 bars in the top of the speimen and were lap splied in the enter of the onstant moment region. Series I speimens had a 18 in. splie length, while Series II and III had a 20

33 12 in. splie length. There were two bar spaing onfigurations seleted for the experimental program. Series I and II speimens were designed with lear spaing between the bars in the splie zone equal to 1 in. and the side over equal to 1-½ in. This represents the minimum lear spaing of bars and the minimum over for reinfored onrete exposed to weather as allowed by ACI (ACI 318, 1999). Series III speimens had a enter to enter bar spaing of 6 in. and side over of 2-11/16 in. This over orresponds to half the lear bar spaing and was seleted to represent a onrete dek with a onstant bar spaing of 6 in. All three series had a top over of 1-½ in., whih represents the requirements of ACI and AASHTO (AASHTO, 1996) for onrete ontaining primary reinforement that is not exposed to weather. This ondition was seleted in light of the fat that the FRP reinforement will not be subjet to orrosion; therefore, the minimum over for interior exposure should be appropriate. The spaing of the top longitudinal reinforement for eah series is shown in Figure #5 bars #5 bars (a.) Series I & II (b.) Series III Figure 3.2: Speimen Details In Series I and II, splitting of the side over was onsidered as the likely failure mode sine the average distane between bars and the side faes of the beam (1 ¼ in.) are less than the top over (1 ½ in.). The failure mode of Series III swithes from side over splitting to top over splitting sine the top over is the smaller dimension. The speimens of all three series had a depth of 16 in and more than 12 in. of onrete ast below the reinforement. Therefore, the reinforement was designed to be in the top ast position. The shear spans of the beams were reinfored with losed stirrups, fabriated from #3 steel bars, at approximately 6 in. on enter to prevent a shear failure in that region. The shear reinforement provided enough shear apaity to insure yielding of the steel speimen in the event a splie failure did not our. No shear reinforement was provided in the onstant moment region sine it was desired to test the splie without the presene of transverse reinforement as transverse reinforement will inrease the bond strength ahieved by the splie (Orangun, 1977). Therefore, the tests were designed to produe a lower bound strength and are onsistent with splies in a bridge dek where transverse reinforement is not typial. Two steel #3 bars were provided in the bottom of the beam to seure the stirrups during age onstrution and to provide tensile reinforement for the beam for handling purposes after failure. A summary of the details of eah series is shown in Table 3.1. The speimens are identified first by the type of test performed (Bond Test), then by the type of reinforing material used in the speimen, and finally by the series number. The reinforing material abbreviations are as follows: S for onventional Grade 60 Steel rebar; G1 for Hughes Bros. Glass FRP rebar; G2 for Marshall Industries Glass FRP rebar; and A for Tehnora Aramid FRP rebar. 21

34 Series I II III Speimen Name B-S-1 B-G1-1 B-G2-1 B-A-1 B-S-2 B-G1-2 B-G2-2 B-A-2 B-S-3 B-G1-3 B-G2-3 B-A-3 Table 3.1: Details of Bond Series Bar Type Steel Glass FRP Glass FRP Aramid FRP Steel Glass FRP Glass FRP Aramid FRP Steel Glass FRP Glass FRP Aramid FRP Splie Length (in.) Top Cover (in.) Side Cover (in.) Clear Spaing (in.) Materials Steel All reinforing steel used was Grade 60 onforming to ASTM A 615. The #5 bars were ordered from the same heat of steel to insure onsistent yield strength for the bars. Three tension tests resulted in yield strengths of 74,500 psi, 74,500 psi, and 78,000 psi, indiating average yield strength of 76,000 psi. The reinforement used for the ompression (bottom) reinforement and stirrups was #3 Grade 60 rebar. The stirrups were prefabriated with standard bends and details required by ACI (ACI 318, 1999) Glass FRP The glass FRP bars were reeived from their respetive produers. In general, the reinforement was in good ondition with only minor surfae defets. Hughes Brothers, In. provided #5 bars for the experimental program. Careful inspetion of the reinforement upon arrival led to the disovery of resin puddles that formed during prodution on the bottom side of the bars while being extruded through the resin bath and then hardened. The manufaturer indiated that it is a ommon phenomenon during prodution. The puddles were surfae treated with a sand oating just as the rest of the bar, and therefore were not viewed as defets. Marshall Industries Composites provided #5 bars, ommerially referred to as C-Bar, for the experimental program. Slight saling of the thin protetive oating of some of the bars was observed. While this may be a durability onern, it is highly unlikely that bond strengths were affeted. Tensile tests on the Hughes Brothers bar (Glass1) indiated failure at 88,000 psi with a modulus of elastiity of approximately 5,900 ksi, while the C-Bar speimens (Glass2) failed at 82,000 psi with a modulus of elastiity of approximately 5,400 ksi. The modulus of elastiity was of primary interest, but it should be noted that both bar types failed at a lower ultimate stress than speified by the manufaturer. Other researhers (Ehsani, 1996) have reported similar findings with respet to tensile tests on FRP bars. The range of stresses experiened by the speimens did not approah the ultimate strength of any of the FRP bars, therefore, this was not of onern in the experimental program. The stress-strain urves for eah type of glass FRP bar are shown in Figure

35 Glass1 Stress (ksi) Glass Strain Figure 3.3: Stress-Strain Curve for GFRP Aramid FRP A Japanese produer, Tehnora, provided the aramid FRP reinforement. The reinforement was in good ondition upon arrival to the laboratory. Tensile tests were performed up to 117,000 psi, whih is signifiantly lower than the ultimate strength speified by the produer. Unlike the tensile tests on GFRP speimens, the ultimate load was governed by a grip failure. As stated earlier, the ultimate tensile strength of the reinforement was not a onern for the experimental program. The tension tests indiated a modulus of elastiity of 6,800 ksi. The stressstrain urve for the aramid FRP bar is shown in Figure Stress (ksi) Strain Figure 3.4: Stress-Strain Curve for AFRP Conrete A loal ready-mix onrete supplier delivered all onrete used in the experimental program. The same bath of onrete was used for all speimens in a series. The bath weights for eah series are shown below in Table 23

36 3.2. Adjustments based on the water ontent of the aggregates were made by the supplier. Water added to the mix before plaement to adjust slump is inluded in the total. The oarse aggregate had a maximum size of ¾ in. in all series. Table 3.2: Conrete Bath Weights Per Cubi Yard Material Series I Series II Series III Cement Type 1 (lbs) Fly Ash Type C (lbs.) Fine Aggregate (lbs.) Coarse Aggregate (lbs.) Water (lbs.) Air (oz.) Water Reduer/Retarder (oz.) Series I, II, and III speimens had a slump of 5 to 6 in. The ompressive strength of the speimens was monitored by testing 6 x 12 ylinders at 7, 14, 21, and 28 days. In addition to testing ylinders for ompressive strength, split ylinder (6 x 12 ) tests and flexure beam (6 x 6 x 18 ) tests were performed on days in whih speimens were tested. Flexure beams were not tested on the day of testing for Speimens B-G1-2 and B-G2-3 due to the proximity of previous or subsequent tests. Table 3.3 shows the average ompressive strength (f ) based on 6 x 12 ompression ylinders, average tensile strength (f t ) based on split ylinders, and average rupture strength (f r ) based on flexure beams for eah speimen. Table 3.3: Average Conrete Strengths Series Speimen Age (days) f (psi) f t (psi) f r (psi) Steel (B-S-1) 104 5, I Glass1 (B-G1-1) 111 5, Glass2 (B-G2-1) 120 5, Aramid (B-A-1) 113 5, Steel (B-S-2) 45 4, II Glass1 (B-G1-2) 50 4, Glass2 (B-G2-2) 108 3, Aramid (B-A-2) 76 4, Steel (B-S-3) 38 5, III Glass1 (B-G1-3) 63 5, Glass2 (B-G2-3) 59 5, Aramid (B-A-3) 68 5, Test Setup and Proedure The test setup was designed to load the ends of the speimens with hydrauli rams. The rams were seured to the loading beams using threaded rods and bolts. As shown in Figure 3.5, rams were loated at either end of the beam. The rams reated against steel load transfer beams that were seured to the reation floor using threaded pipe. The reation supports were attahed to the floor using hydrostone. A plate was ast on the top of eah support; one flat plate and one grooved plate. Steel rods were loated between the plates on the support and the bearing plates that were attahed to the underside of the beams (with hydrostone) at the support points. This onfiguration was seleted to simulate a pin-roller support system. Bearing plates were also used at the loading points on the end of the speimens. Hydrostone was used for the attahment of all plates due to its high rate of strength gain and high ultimate ompressive strength. Load ells were used to monitor load appliation during the tests. The load ells used for the first two series were rated for 20 kips, while a 100 kip and 150 kip load ell were used for the third series. All load ells were alibrated before testing using a 120 kip universal testing mahine. The alibration of the load ells was performed in the range in whih they were used during testing. 24

37 Defletions were measured at several points along the speimen using linear voltage displaement transduers (LVDT s) that were mounted on the beam. All tests inluded defletion readings at eah end of the speimen (load points), at the support points, and at the middle of the beam. The LVDT s loated at the support points were used to detet support settlement, whih were used to orret the values obtained at the ends and middle of the beam. Load, defletion, and strain readings were reorded ontinuously throughout the tests. Load was applied at 1 kip intervals. After the first flexural raks appeared, raks were marked and rak widths were reorded at eah load stage until it was deemed unsafe due to eminent failure of the beam. In some instanes, loading was momentarily stopped and rak widths were reorded between load stages. Figure 3.5: Loading Detail 3.5 Experimental Results General Behavior Load Cell Flexural Craking For a given series, eah speimen raked at approximately the same load while exhibiting approximately the same stiffness up to the raking load. The first raks ourred in the onstant moment region of the speimens, generally either over a support or at the end of the splie region. As the load inreased, additional flexural raks ourred in the onstant moment region, and flexural raks began to form in the shear span. Overall, the rak pattern losely resembled the bending moment diagram for the speimens. It was observed that the raks in the FRP reinfored speimens propagated deeper into the setion than that of the ompanion steel speimen at a given load (stress level), as shown in Figure 3.6. With the reinforement at a stress of approximately 36 ksi, the flexural raks of the steel speimen (Figure 3.6a.) have propagated roughly two-thirds through the 16 in. deep setion, while the raks of a omparable aramid speimen have propagated muh further. The rak widths observed in the FRP reinfored speimens were several times larger than those observed in the ompanion steel speimen. 25

38 The FRP reinfored speimens exhibited extensive branhing of the priniple flexural raks near the level of reinforement as ompared to the steel speimens. Figure 3.7 shows the onstant moment region of Speimen B- S-1 (Figure 3.7a) and Speimen B-G2-1 (Figure 3.7b). Although the steel speimen (B-S-1) is at a higher longitudinal stress, the raking has not branhed from the priniple flexural raks as it has for the glass speimen (B-G2-2). Depth of Craks (a.) Steel Speimen at 36 ksi (B-S-2) Depth of Craks (b.) Aramid Speimen at 36 ksi (B-A-2) Figure 3.6: Comparison of Crak Propagation 26

39 (a.) Top View of Speimen B-S-1 at 40 ksi Branhing of Craks Priniple Flexural Craks (b.) Splie Region of Speimen B-G2-1 at 25 ksi Figure 3.7: Comparison of Craking Failure All speimens in the experimental program failed by a splitting mode in the splie region. The speimens failed abruptly; the first and seond series speimens failing by side over splitting, while the third series with a wider bar spaing, failed due to top splitting. Beause hydrauli rams were used for loading, the load was partially removed at the initiation of failure. Failure of the speimens ommened when the over in the splie region blew 27

40 apart due to exessive radial stress imparted on the onrete by the reinforement. The nature of the brittle splitting failure is aptured in Figure 3.8. (a.) Wide Angle View of Failure of B-S-2 (b.) Failure of B-S-2 Figure 3.8: Brittle Splitting Failure Appearane After Failure The splie regions of the speimens were investigated immediately after failure. The side splitting failure mode of the first and seond series is shown in Figure 3.9(a.). Figure 3.9(b) shows the top splitting failure mode of the third series. The over for eah series was removed as muh as possible by hand (if the fore of the failure had not removed it), and the onrete fragments were reassembled as shown in Figure The reassembled over was used to determine the exat bar loations. Both the reassembled over and the reinforement in the splie zone were arefully examined to determine whether any loalized rushing had ourred in the onrete around the reinforement deformations, or in the ase of the FRP reinforement, whether there was damage to the deformations. A visual determination of bond quality was also made. 28

41 Top Splitting Craks (a.) Side Cover Splitting (b.) Top Cover Splitting Figure 3.9: Splitting Failure Modes Figure 3.10: Reassembled Conrete Cover 29

42 3.5.2 Test Results The maximum load and orresponding reinforement stress ahieved for eah test speimen is provided in Table 3.4. Speimen harateristis suh as onrete strength, splie length, top over, and lear bar spaing, are also provided for ease of omparison. The reinforement stress values in Table 3.4 were alulated based on the load ahieved and a raked setion analysis. The Hognestad onrete stress blok (MaGregor, 1997) was used in the alulation of reinforement stress to aount for the non-linear behavior of onrete at high stress levels. All alulations were based on the design ross-setional dimensions. In general, alulation of the depth of the neutral axis from the raked setion analysis agreed well with the observed depth of rak propagation. In addition, the stresses measured from the strain gages at the end of the splie zone were generally similar to the alulated values. Series I II III Speimen f (psi) Table 3.4: Bond Test Results l s (in.) Cover (in.) Clear Spaing (in.) P u (k) f bu (ksi) B-S-1 5, B-G1-1 5, B-G2-1 5, B-A-1 5, B-S-2 4, B-G1-2 4, B-G2-2 4, B-A-2 4, B-S-3 5, B-G1-3 5, B-G2-3 5, B-A-3 5, It is signifiant to note that the reinforing stress in Speimen B-S-1 and B-S-3 did not yield aording to the alulations (average f y = 76 ksi). These alulations were supported by the load-defletion urves (Figures ) Beam Stiffness As stated earlier, all speimens within eah series raked at approximately the same load. Analysis of the load vs. defletion urves (Figures ) of eah series shows that the speimens within eah series had approximately the same stiffness up to raking. The average load applied to eah end of the beam as determined by the alibrated load ells is presented on the vertial axis, while the average defletion of the speimens at the loading points as determined by LVDT s, inluding orretions for support settlement is presented on the horizontal. After raking, the stiffness of the speimens hange, evident by the hange in slope of the load vs. defletion urve, as the reinforement must now resist tensile stress that was resisted primarily by the onrete. The modulus of elastiity (stiffness) of the reinforement beomes the governing parameter determining the stiffness of the beam speimens, as the raked moment of inertia of the FRP reinfored speimens is approximately 20-30% that of the steel speimens. The saw-tooth pattern of the load vs. defletion urves after initial raking reflets the fat that the tests were essentially defletion ontrolled. In instanes in whih the urves seem disontinuous, new raks formed in the speimens or old raks were propagating deeper into the setion. 30

43 30,000 25,000 Avg. End Load (lbs.) 20,000 15,000 10,000 5,000 Steel Glass2 Aramid Glass Avg. End Defletion (in.) Figure 3.11: Series I Load vs. Defletion Plot 30,000 25,000 Avg. End Load (lbs.) 20,000 15,000 10,000 5,000 Steel Aramid Glass1 Glass Avg. End Defletion (in.) Figure 3.12: Series II Load vs. Defletion Plot 31

44 Avg. End Load (lbs.) 30,000 25,000 20,000 15,000 10,000 Steel Aramid Glass2 Glass1 5, Avg. End Defletion (in.) Figure 3.13: Series III Load vs. Defletion Plot Crak Widths Crak widths assoiated with eah type of reinforement were ompared by analysis of the reinforement stress vs. average rak width urves. These urves for eah series are shown in Figures The alulation of reinforement stress was based on the load ell reading at the time the rak widths were measured using the Hognestad onrete stress blok. Calulations were based on the design dimensions of the ross-setion. Craks that ourred over the support, in the splie region, and at the ends of the splie zone were not inluded in the average. Craks over the support and at the end of the splie zone ourred as a result of the onfiguration of the test setup, and therefore do not represent a random distribution of raking. The raks that ourred in the splie zone were restrained by double the reinforement as the other raks in the beam, leading to signifiantly redued rak widths. The rak width urves end before speimen failure, when it was deemed unsafe to take readings due to eminent failure of the speimens. As shown in Figures , the urves for the steel speimens as ompared to the FRP speimens are remarkably different. At a given stress level, the average rak width in the steel speimens were several times less than (as muh as 10 times) that of FRP reinfored speimens within the same series. The rate at whih the average rak widths inreased as the reinforement stress inreased (slope of the urve) in the steel speimen is several times less than the FRP speimens as well. The FRP speimens in eah series ahieved average rak widths that exeeded in. at relatively low stress levels (approximately 20 ksi). 32

45 Bar Stress (ksi) Steel Aramid Glass2 Glass Average Crak Width (in.) Figure 3.14: Series I Bar Stress vs. Avg. Crak Width Bar Stress (ksi) Steel Glass2 Glass1 Aramid Average Crak Width (in.) Figure 3.15: Series II Bar Stress vs. Avg. Crak Width 33

46 Steel Bar Stress (ksi) Glass1 Glass2 Aramid Average Crak Width (in.) Figure 3.16: Series III Bar Stress vs. Avg. Crak Width Based on this data, it is diffiult to make any generalizations about the performane of one type of FRP reinforement ompared to another. In the first series, the aramid FRP speimen ahieved a smaller average rak width at a given stress when ompared to the two glass FRP speimens. However, in the third series, the Glass1 speimen ahieved smaller rak widths than the Aramid speimen. One trend that is evident in the urves is the similarity of the slope of the FRP speimens. In eah series, the FRP speimens average rak width inreases with stress level at approximately the same rate. Although the deformation pattern on eah type of FRP reinforement is different, the raking behavior is quite similar for all bar types Bond Strength The bond strength for the speimens of eah series of speimens is presented in Table 35. The strengths were determined by alulating the average bond stress ating along the splie length. The nominal bar diameter and design dimensions were used in all alulations. The bond ratio shown in the table represents the ratio of bond stress (or bar stress) for the speimen to the bond stress of the steel speimen within that series. The Series I and II FRP speimens attained a bond stress of approximately one half that of the steel speimens. With wider bar spaing (Series I), the FRP speimens ahieved roughly two-thirds of the bond stress of the steel speimen. It is evident that the bond stress ahieved appears to be related to the modulus of elastiity of the reinforing materials, as the steel speimens had the highest bond strength and modulus of elastiity, followed by the aramid FRP speimens, and finally the glass FRP speimens. Table 3.5: Lap-Splie Bond Strengths Bond Strength, u Series Speimen P u (k) f bu (ksi) avg Bond Ratio (psi) (u avg / u steel ) B-S B-G I B-G B-A

47 II III B-S B-G B-G B-A B-S B-G B-G B-A Data Analysis ACI Design Provisions The design provisions of ACI (ACI 318, 1999) have their roots in the reommendations presented by Orangun et. al. (Orangun, 1977). It should be noted that these reommendations were based on available data of steel reinfored speimens; therefore, this equation is not meant to be used with other reinforing materials. It is possible, however, that slight modifiations ould result in its appliability for different types of reinforing materials. The development length equation of ACI (Equation 3-1) is shown for the purpose of this disussion: ld 3 f y αβγλ = (Eq. 3-1) db 40 f ' + tr K d b The 3/40 (0.075) oeffiient on the right side of the equation is a value that provides onservative development lengths for the majority of available steel data. Ultimately, an equation of the same form ould be utilized for other types of reinforement, but a multiplier (X), based on test data for that type of reinforement, would have to be used. By dividing Equation 1-3 by the 5/4 overstrength fator ontained within the equation, the 3/40 oeffiient beomes 3/50. The modified equation is shown below: where: l d d b = 3 50 f b f ' M f + K d b tr X (Eq. 3-2) M l d f f' = speified ompressive strength of onrete,(psi) K d b b f = development length of reinforement,(in.) = diameter of reinforing bar,(in) = speified produt of = lesser tr = = transverse reinforement index X = modifiation stress in reinforement,(psi) modifiation fators α, β, γ,and λ of lear over or 1/2 lear fator for reinforement type spaing between bars,(in.) Values for X for eah of the speimens (inluding steel) tested in the experimental program are shown in Table 3.6. Values were determined for the steel speimens for the purpose of omparison with the values obtained for FRP speimens. The values alulated were based on solving Equation 4-1 for X. The values of l d and were based on design dimensions, while the values of f b and f were taken as the alulated bar stress at failure and the 35

48 average ompressive strength at the time of testing. The modifiation fator for bar loation was not onsidered sine its inlusion would likely provide unonservative values for the X fator. Table 3.6: X Fators Based on Experimental Results Series Speimen X Fator Steel (B-S-1) 1.03 I Glass1 (B-G1-1) 1.95 Glass2 (B-G2-1) 2.30 Aramid (B-A-1) 1.85 Steel (B-S-2) 0.78 II Glass1 (B-G1-2) 1.50 Glass2 (B-G2-2) 1.47 Aramid (B-A-2) 1.39 Steel (B-S-3) 1.01 III Glass1 (B-G1-3) 1.56 Glass2 (B-G Aramid (B-A-3) 1.48 The values of X in the last olumn of Table 3.6 indiate that the urrent form of the ACI development length equation is onsiderably unonservative (X is muh greater than 1.0) for all of the FRP reinfored speimens. In the ase of the steel reinfored speimens, the equation provides results in X fators very lose to 1.0, even though the 5/4 overstrength fator and the bar loation fator have been removed from Equation 4-1. Comparing results of all the speimens in Series I and II indiates that Equation 4-1 is more onservative for Series II, with a shorter splie length. While the ACI development length equation may onsistently alulate onservative results if appropriate multipliation fators are used, the inonsistent level of onservatism indiates that the equation does not reflet the atual bond behavior. Clearly, if Equation 4-1 is to be used with FRP reinforement, an appropriate X fator should be seleted for eah type of reinforement. For the FRP reinforement used in this experimental program, the X fator seleted for eah type of FRP reinforement is 2. It should be noted that this reommendation is only for the types of FRP reinforement used in this experimental program. No overstrength fator has been inluded sine providing suh a fator does not insure dutility in design with FRP reinforement ACI Committee 440 Proposed Reommendations ACI Committee 440 (ACI 440, 2000) has proposed reommendations for the design of members reinfored with FRP. The development length equation proposed by ACI Committee 440 for splitting failure is: 2 d b f fu lbf = K 2 (Eq. 3-3) f ' where: l d bf K f C f b 2 fu E fu = development length required,(in.) = onstant determined experimentally = reinforing bar diameter,(in.) = design tensile environmental fators,(psi) = C = environmental and exposure onditions guaranteed tensile strength,(psi) f' = speified onrete strength,(psi) * = strength of reinforement onsidering redution fator, dependent on fiber type E f fu * 36

49 The value of f fu is defined as the mean tensile strength of a sample of test oupons minus three standard deviations. The ommittee proposes values of C E based on fiber type and exposure onditions, whih are shown in Table 3.7. Exposure Condition Conrete not exposed to earth and weather Conrete exposed to earth and weather Table 3.7: Proposed Values of C E Fiber Type Environmental Redution Fator, C E Carbon 1.0 Glass 0.8 Aramid 0.9 Carbon 0.9 Glass 0.7 Aramid 0.8 The alulated development length is dependent on the bar diameter (bar area), longitudinal stress to be generated, and the onrete strength. ACI Committee 440 has proposed inluding in the K 2 fator the π/4 term from the area of the bar. ACI Committee 440 has not speifially endorsed a proposed K 2 fator, but has reported the fators determined by many investigators. The average of all K 2 values reported was 1/18.2 with the largest being reported as 1/15.6 by Tighiouart (Ehsani, 1996). The K 2 fators determined for the all speimens tested in this experimental program are shown in Table 3.8. In the determination of K 2, the alulated stress and the average onrete strength at the time of testing were used. An overstrength fator similar to that used in ACI (ACI 318, 1999) was not used in the determination of the K 2 values presented in Table 3.8 sine the Committee s reommendations do not inlude suh a fator. This fator is not reommended sine the ultimate bar stress is used in design. Providing an overstrength fator does not provide dutility. The modifiation fator of 1.3 for top bar loation was not used in the alulations, whih is a onservative assumption. Table 3.8: K 2 Fators Based on Experimental Results Series Speimen Required K 2 Perent of 1/18.2 Perent of 1/15.6 Steel (B-S-1) 1/ % 96% I Glass1 (B-G1-1) 1/ % 140% Glass2 (B-G2-1) 1/ % 166% Aramid (B-A-1) 1/ % 134% Steel (B-S-2) 1/ % 73% II Glass1 (B-G1-2) 1/ % 140% Glass2 (B-G2-2) 1/ % 137% Aramid (B-A-2) 1/ % 130% Steel (B-S-3) 1/ % 64% III Glass1 (B-G1-3) 1/ % 98% Glass2 (B-G2-3) 1/ % 103% Aramid (B-A-3) 1/ % 92% The perentages shown in the fouth olumn of Table 3.8 are the ratio of the required K 2 determined for eah test to the average of the K 2 values presented by ACI Committee 440. The perentages in the last olumn are the ratio of the required K 2 to the largest value determined by the investigators (Tighiouart, 1998). The values obtained for Series III FRP speimens indiate that the average of the K 2 fators proposed by the investigators is slightly unonservative (perentages above 100%) for the partiular over and spaing. Series III had development length of 12 in., while the lear spaing between bars was 4 ¾ in. For the FRP speimens of Series I and II, whih had minimum over and lear spaing, the proposed K 2 fators beome very unonservative. It appears that hanging the development length, as is the ase with Series I and II, does not affet the required K 2. However, 37

50 omparison of the value of K 2 in Series I and II with the value in Series III indiates that the value hanges as the over or lear spaing is hanged Aording to the proposed reommendations, pullout failures are more likely when onrete over is greater than 2d b. All three series had a onrete over greater than 2d b, thus the alulation of development length using Equation 3-4 should be appliable aording to the reommendations. d b f fu l bf = (Eq. 3-4) 2700 The development length and bar diameter used in eah series of tests was substituted into Equation 3-4, and the equation was subsequently solved for the bar stress. The results of this alulation are shown in Table 3.9 and ompared to the atual bar stress ahieved in eah speimen. Table 3.9: Calulated Stress by Equation 3-4 Series Speimen Stress Ahieved (ksi) Calulated Stress by Equation 3-4 (ksi) Glass1 (B-G1-1) I Glass2 (B-G2-1) Aramid (B-A-1) Glass1 (B-G1-2) II Glass2 (B-G2-2) Aramid (B-A-2) Glass1 (B-G1-3) III Glass2 (B-G2-3) Aramid (B-A-3) Comparison of the atual and alulated stress in Table 3.9 indiate that Equation 3-4 is extremely unonservative in ases with minimum over and spaing, whih is the ase with Series I and II. Although the equation is meant for use in instanes where pullout failure governs, this omparison is valid sine the proposed reommendations indiate that a pullout failure was likely for all speimens in the experimental program. Clearly, the reommendations should be modified to orret this problem. Based on the tests in this experimental program, it is lear that the proposed reommendations of ACI Committee 440 should be modified. It has been found that the K 2 fator (Equation 3-3) proposed by other investigators is modestly unonservative for a top over of 1.5 in. and a lear spaing of 4.75 in. (Series III), but beomes seriously unonservative for the minimum bar spaing of 1 in. (Series I and II). While a larger K 2 fator or other modifiation fators ould be inorporated into Equation 3-3 to provide more onservatism, there is an inherent problem with the form of the equation; Equation 3-3 does not aount for the influene of over on the required development length. Therefore, the use development length provisions of the proposed ACI Committee 440 doument are not reommended AASHTO Design Provisions The development length equation set forth in AASHTO Standard Speifiations, Sixteenth Edition (AAHSTO, 1996) is of the same form as Equation 3-3 of the proposed reommendations of ACI Committee 440 (ACI 440, 2000). Therefore, the results presented in Setion are equally appliable. Due to the problems indiated with the equations presented, it is not reommended to use a modified form of the AASHTO design equations. 38

51 3.7 Crak Widths While it is unlikely that raking in FRP reinfored onrete will ause serious durability onerns; it is, however, an aestheti onern. Due to the lower modulus of elastiity of many types of FRP reinforing bars, defletions and rak widths will be larger than with a omparable reinfored steel struture as evidened by the experimental results. Several expressions have been proposed for alulating rak widths of steel reinfored members. ACI (ACI 318, 1999) has adopted a rak width alulation proedure based on an expression developed by Frosh (Frosh, 1999). This expression is based on a physial model, and therefore, should be appliable to other types of reinforement. The equation is reviewed below: w = β ε S (Eq. 3-5) s where: w = rak width (h - ) β = = amplifiation fator ( d ) h = overall depth of the member = depth from ompression fae to the neutral axis d = effetive depth f s ε s = reinforing steel strain = Es S = rak spaing = Ψs d * f s = reinforing steel stress Es = reinforing bar modulus of elastiity Ψs = rak spaing fator :1.0 for minimum spaing; 1.5 for average; and 2.0 for maximum In the form shown, Equation 3-5 is only appliable to steel reinfored members, sine it uses the modulus of elastiity of steel, E s. However, this model an be expressed for any type of reinforement by simply using the appropriate modulus for the reinforement used. The maximum rak widths alulated using Equation 3-5 for all speimens in this experimental program were ompared with the atual maximum rak widths measured during testing. The results of this omparison are shown in Figure The results were grouped into reinforement types to determine the auray of the model in eah ase. The reinforement stress used in the alulation for rak widths was the alulated stress based on the load determined by the load ells. Figure 3.17 indiates that Equation 3-5 results in alulated rak widths for FRP reinfored speimens with the same order of preision as for steel speimens, but is slightly less onservative. The average ratio of alulated rak width to measured rak width was 1.2 for the steel reinfored speimens; 1.0 for the Aramid speimens; and 0.9 for the Glass1 and Glass2 speimens. Sine the primary onern about rak widths in FRP reinfored onrete is aesthetis, modifiation of Equation 3-5 to make it more onservative is unneessary. 39

52 Number of Observations Calulated Value/Measured Value Number of Observations Calulated Value/Measured Value (a.) Steel (b.) Glass1 Number of Observations Calulated Value/Measured Value Number of Observations Calulated Value/Measured Value (.) Glass2 (d.) Aramid Figure 3.17: Maximum Crak Width Comparison 3.8 Defletions The low modulus of elastiity of FRP reinforement indiates that in many instanes the design of onrete reinfored with FRP will be ontrolled by defletion. This requires that the alulation of defletions be as aurate as possible. As disussed in Setion 3.5.3, the speimens of eah series had approximately the same stiffness up to the raking load (or raking moment). After raking the stiffness is highly dependent on the modulus of elastiity. These stiffnesses are ommonly referred to as the unraked stiffness (EI g ) and the raked stiffness (EI r ), respetively. The load vs. defletion plots for Series I were ompared to the alulated raked and unraked stiffness in Figure The unraked stiffness is approximately the same for eah speimen, and therefore EI g plotted in Figure 3.18 represents the average value for the entire series. The raked stiffness for Speimen B-S-1 and the average raked stiffness for the FRP speimens are shown in the figure as well. The values of EI were alulated using the transformed setion for both the unraked and raked ase. The modulus of elastiity of the onrete was determined using the expression in ACI (ACI 318, 1999): 40

53 E = 57,000 f ' (Eq. 3-6) where: E = alulated modulus of elastiity of f' = ompressive strength of onrete,(psi) onrete,(psi) In alulating the modular ratio, n (E b /E ), the modulus of elastiity of steel was taken as 29,000 ksi, while the modulus of the FRP bars was taken as 6,000 ksi, whih is the average of the three types. Figure 3.18 indiates that the use of the transformed setion to determine the unraked and raked stiffness is an effetive tool in determining the load-defletion relationship in a onrete member reinfored with FRP. In Series I, the average raked stiffness alulated for the FRP speimens fell loser to the load-defletion urves than did the raked stiffness alulated for the steel speimen. 30,000 Avg. End Load (lbs.) 25,000 20,000 15,000 10,000 EI g EI r (steel) B-S-1 B-A-1 B-G2-1 B-G1-1 EI r (FRP) 5, Avg. End Defletion (in.) Figure 3.18: Series I Load vs. Defletion Plot 41

54 4. SHEAR INVESTIGATION 4.1 Introdution The experimental program investigated the shear strength and behavior of beams reinfored only in the longitudinal diretion with FRP bars. Two series of tests, Series 1 and Series 2, were onduted in whih three different FRP bars (two types of Glass FRP and one type of Aramid FRP) were evaluated. For omparison, a steel reinfored speimen was inluded in eah series in addition to three FRP reinfored speimens. The seond series inluded a fifth speimen reinfored with a high-yield-strength steel bar to investigate the effet of axial rigidity of longitudinal bars on shear behavior. 4.2 Design of Speimens Speimens were designed to be representative of a bridge dek. Therefore, only longitudinal reinforement was provided to obtain the shear strength provided by the onrete. The speimen height was seleted as 16-in. to be representative of typial slab span bridges used in Indiana. The nominal onrete strength was seleted as psi, whih is typial bridge dek design strength. The shear strength of reinfored onrete beams is influened signifiantly by the shear-span to effetive depth ratio (a/d). The aim was to size and test the speimens so that beam behavior rather than arh behavior would be dominant. In addition, it was desirable to set the a/d ratio lose to 2.5 where the shear strength due to onrete ontribution is lowest. Considerations of the a/d ratio together with spae and manageability onstraints in the laboratory resulted in an 8-ft. span length, (a/d = 3.4). The longitudinal reinforement ratio is also known to affet the shear strength of reinfored onrete members. Minimum and maximum longitudinal reinforement amounts for steel reinforement are typially speified by design odes. These limits are a funtion of strength and dutility requirements and vary with onrete strength. Aording to the AASHTO Bridge Design Speifiations (AASHTO, 1996) and the ACI Building Code (ACI 318, 1999), the reinforement ratio to satisfy these requirements ranges from 0.35 to 2.5% for 5000-psi onrete. In Shear Series 1, 1% longitudinal reinforement was used, whih is loser to the lower end of the limits and a typial amount used in pratie. A lower perentage of reinforement was not tested due to onerns regarding servieability requirements. Sine the modulus of elastiity of FRP bars is low, raking and defletions may often ontrol design and a minimum amount of longitudinal reinforement is required for their ontrol. The width of speimens was ontrolled by the perentage of longitudinal reinforement and the allowable bar spaing and over requirements, whih were all satisfied by providing an 18-in width. Shear Series 2 speimens were designed to investigate the pratial upper limit of longitudinal reinforement, whih is allowed by the AASHTO Bridge Design Speifiations and the ACI Building Code. Twie the reinforement was used to study the effet of varying longitudinal reinforement ratio on shear behavior. By using two layers of reinforement and the same beam width, the pratial limit, whih is ontrolled by minimum spaing requirements, is alulated as 2.15%. However, ease in laying out the bars was preferred to stritly omplying with the alulated ratio, whih resulted in a design reinforement ratio of 2%. Series 2 speimens provide a diret omparison with those in Series 1. An additional speimen reinfored with high yield-strength steel bars was inluded in this series. The reinforement ratio provided for this speimen was seleted as 0.36% so that the axial stiffness of the steel reinforement used in this speimen would be approximately equal to the axial stiffness of reinforement in the beams reinfored with 2% glass FRP bars. Therefore, Shear Series 2 also allowed a omparison between the shear strengths of beams that were provided with the same axial stiffness of reinforement but different reinforing bars. Anhorage length required beyond supports to prevent premature bond failures prior to shear failure was alulated as 16-in. However, 2.5-ft anhorage beyond supports was provided to eliminate pull out of bars due to splitting of onrete over, a typial failure mode observed in shear tests when the longitudinal bars are not properly anhored. These anhorage requirements resulted in an overall beam length of 13-ft. A summary of design properties for eah speimen is provided in Table 4.1. The test setup is shown in Figure 4.1 while ross-setion details used in eah series are illustrated in Figures 4.2. The speimens are designated in the following manner. The first letter, V, stands for a shear series speimen. The seond letter after the hyphen desribes the type of reinforement used in the speimen; S for steel, D for high strength steel, G for glass, and A for aramid bars. The number following the seond letter in the glass FRP reinfored speimens designates the type of glass bar used; 1 for ribbed, 2 for indented and sand oated. The last number the test series, 1 for Series 1 and 2 for Series 2. 42

55 Table 4.1: Shear Speimen Nominal Design Values Series Speimen f (psi) E r (ksi) a/d ρ (%) V-S V-G V-G V-A V-S V-D V-G V-G V-A P Anhorage Shear span Anhorage zone zone 13-0 Figure 4.1: Test Setup 16.81" 16.8" 1" d=14.19" 16" 16" d=14.19" 2" 2" = 14" 2" 18" (a) (a) Series 1 Speimens. 2" 7 2" = 14" 2" 3" 7 2" = 14" 3" 18" 18" (b) (b) Series 2 Speimens with () () Series 2 Speimen with two two layers reinforement of reinforement layer. one one layer reinforement of reinforement layer. Figure 4.2: Cross Setion Details 43

56 4.3 Materials Reinforement Five types of reinforement were used in this investigation, three types of FRP bars and two types of steel bars. FRP reinforement inluded 2 types of glass FRP bars, Glass 1 and Glass 2, and one aramid FRP bar. Steel reinforement onsisted of a onventional steel bar, Steel, and high yield strength steel bar, Dywidag. Properties of the reinforement are presented in Table 4.2. Details regarding eah bar type exept the Dywidag bars are as presented previously in Setion 3.3. Bar Type Table 4.2: Mehanial properties of reinforing materials E (ksi) σ y (ksi) σ u (ksi) Steel Dywidag Glass Glass Aramid ε y ε u High Yield Strength Steel (Dywidag) High yield #5 bars were obtained from Dywidag Co. These bars were used beause of their high yield strength (156-ksi) so that a beam with very low perentages of longitudinal reinforement (0.36%) ould be tested without yielding the reinforement. Three tension tests were onduted whih resulted in average yield strength of 108-ksi and an ultimate strength of 145-ksi. The mehanial properties of the bar are listed in Table 4.2 and a representative stress strain urve in Figure Stress(ksi) Strain Figure 4.3: Stress-Strain Curve for Dywidag Steel Bars 44

57 4.3.2 Conrete Conrete was obtained from a loal ready-mix onrete supplier. The same mix design was used for all speimens and bath weights for eah series are listed in Table 4.3. The onrete onsisted of ¾ maximum aggregate (river gravel), minimal water reduer and air entrainment. Slump was adjusted after arrival of the truk through the addition of water. Conrete for Series 1 and 2 had 6-½ slumps. Compressive strength of the speimens was monitored by testing three standard 6 x12 ylinders at 7, 14, and 28 days. Table 4.3: Conrete Bath Weights Per Cubi Yard Material Series 1 * Series 2 Cement (lbs) Fine Aggregate (lbs) Coarse Aggregate (lbs) Water (lbs) Air (oz) 1 1 Water Reduer/Retarder (oz) *Design mix-proportion, not atual. Speimen ompressive strength and split ylinder strength on the day of testing were obtained by testing three 6X12 ylinders for both Series 1 and Series 2 speimens. Additionally, flexural strength of Series 2 speimens was obtained by testing two 6 x6 x30 modulus of rupture beams on the day of testing. The average ylinder ompressive strength (f ), split ylinder strength (f t ), and rupture strength (f r ) on the day of testing are tabulated in Table 4.4. Table 4.4: Average Conrete Strengths Series Speimen Age (days) f (psi) f t (psi) f r (psi) V-S V-G V-G V-A V-S V-D V-G V-G V-A Test Setup and Proedure Test setups were designed to apply a onentrated load at the mid-span of the simply supported beams (Figure 4.1). In Series 1 and Series 2, two different test setups, a 600-kip Universal Testing Mahine and a 220-kip MTS hydrauli atuator respetively, were used. In Series 2 speimens, onrete surfae strains were measured with a Wittemore gage, whih required a displaement ontrolled loading system. Sine the setup used in Series 1 was not apable of displaement ontrolled loading, a new test setup allowing for both load and displaement ontrolled loading was prepared for loading Series 2 speimens. The supports, whih were made from welded steel hannels and plates, were seured on the strong floor with hydrostone in Series 1 and seated on top of two load ells at eah end in Series 2. A roller support was obtained by plaing a 1-½ diameter steel rod between two steel flat plates. The pin support was obtained by plaing a steel plate on top of a beveled steel rod and plaing a flat steel plate on top of the rod. The pin and roller 45

58 support assembly are shown in Figure 4.4. The beam was positioned, leveled, and ast on the top plates by hydrostone. To alleviate bearing onditions under the onentrated load a steel plate was hydrostoned to the top of the beam at midspan. Hydrostone was used for attahing the steel plates for its high rate of strength gain and high ompressive strength. (a) Roller support (b) Pin support Figure 4.4: Supports As previously mentioned, Series 1 speimens were loaded with a 600-kip universal testing mahine under load ontrol (Figure 4.5). The load was inreased ontinuously until first raking was observed and inreased in 5-kip inrements afterwards. At the end of eah interval, the load was held onstant while raks were marked and rak widths were measured while photographs were taken. A load ell plaed under the onentrated load measured the load while defletions were measured at 2-ft intervals with linear voltage displaement transduers (LVDT). Figure 4.5: Series 1 Test Setup Shear Series 2 speimens were loaded with a 220-kip MTS hydrauli atuator (Figure 4.6). The load was inreased ontinuously until first raking. Following first raking, the load was inreased in 5-kip inrements under load ontrol. At eah load stage, raks were marked and their widths measured. At every other load stage (10-kip inrements), the displaement of the atuator head was held onstant. During this time, strains on the 46

59 onrete were measured with Wittemore gages and photographs of the beam were taken. The applied load was measured by the internal load ell on the atuator as well as by load ells plaed under the supports. The load measurements reported in this doument are the ones obtained from the load ells. Figure 4.6: Series 2 Test Setup To monitor the strain profile along the reinforement Series 2 beams were instrumented with strain gages attahed to the reinforing bars at 1-ft intervals from support to support. In addition, two surfae strain gages were attahed on the onrete top surfae 4.5-in from midspan. Displaements were measured at 1-ft intervals along the span and at the supports to obtain the defleted profile of the speimens. To determine the strain distribution in the onrete a Wittemore gage ontat point grid was attahed on the side surfae of the beam. Figure 4.7(a) illustrates the instrumentation plan for LVDTs and strain gages while Wittemore gage ontat point grid is given in Figure 4.7(b). LVDTs Strain gages on the reinforing bars 4.5in. 4.5in. Conrete strain gages ft 1-ft = 8-ft ft (a) LVDT and Strain Gages 47

60 2 5/8" 1'-2.19" 4.5" 3.0" 4.5" = 90" 3.0" (b) Wittemore gage ontat points Figure 4.7: Instrumentation Plan for Series Experimental Results General Behavior Loading In all speimens the defletions inreased linearly with inreasing load prior to first raking. First raking during the test was observed in the load defletion urve by a marked redution in the stiffness. The initial rak ourred under the onentrated load or within its lose viinity. New flexural raks developed loser to the supports while the existing raks grew with inreasing load. The raks were vertial during early stages of loading (when the total shear on the speimens was low) but propagated by inlining towards the onentrated load as the load was further inreased (Figure 4.8). In general, the loser the rak to the support, the more inlined it beame through the ourse of loading. At loads very lose to the speimen s shear apaity, the heights of the outermost inlined rak and the raks under or loser to the load were approximately the same (Figure 4.9). The inlination and height of the outermost rak usually indiated the imminene of shear failure. Crak patterns on eah side between the supports and the onentrated load were generally symmetri prior to shear failure. However, in some ases the rak pattern and the inlined rak on one side of the onentrated load were more developed (severe) than the other side. In other words, the outer inlined rak on one side would penetrate high into the setion, whih indiated the side that the diagonal tension rak would form Failure All speimens failed brittlely in diagonal tension. The diagonal tension rak was usually a ontinuation of one of the outermost flexural raks losest to the supports, also known as a flexural-shear rak. The flexural rak, whih initiated the diagonal tension failure, propagated deeper into the beam and beame more inlined with inreasing shear load. In 7 beams out of the 9 the height of inlined raks, whih lead to failure, were equal to or more than the height of those under the load as illustrated in Figure 4.9. The rak might have penetrated into the ompression zone. The inlined rak kiked bak towards the level of reinforing steel prior to failure. With further loading two different kinds of behavior were observed. In the first senario, splitting along the reinforement past the support took plae simultaneously with the diagonal tension rak growing towards the onentrated load. Providing a very onservative anhorage length beyond the supports eliminated the possibility of bond failure due to insuffiient anhorage. In the seond senario, the splitting rak along the reinforement reahed the bottom of the beam immediately before the support without ausing splitting past the support. Examples of failed beams illustrating these differenes in failure pattern are shown in Figures

61 (a) Craking Pattern at 30-kip (b) Craking Pattern at 60-kip Figure 4.8: Speimen V-G2-2 During Two Different Load Stages Figure 4.9: Speimen V-D-2 Prior to Failure 49

62 (a) Shear Crak Reahing Bottom Before Support (b) Splitting along Reinforement Past Support Figure 4.10: Speimens after Failure A tension rak, an example of whih is shown in Figure 4.11, formed on the onrete ompression zone above the diagonal tension rak approximately 15-in. to 30-in. away from the onentrated load. These tension raks extended aross the width of the speimens on the top surfae. The trae of the diagonal tension rak, whih lead to failure, at the bak surfae of the speimens was not idential to its trae on the front surfae in most tests. Tension rak in the ompression zone (a) Side View (b) Top View Figure 4.11: Tension Crak on the Top Surfae After Failure (V-S-2) Craking Load The speimens in eah series raked at approximately the same load with the exeption of steel reinfored speimens, whih raked at a load signifiantly higher than the flexural raking load of the FRP reinfored speimens. Although raking loads may be expeted to differ due to different modulus of elastiity of the bars used, the inrease observed in the experiments ould not be explained onsidering this effet. Even when the differene between onrete strengths were also onsidered, the differene between the flexural raking loads of steel reinfored and FRP bar reinfored speimens in eah series ould not be explained. If shrinkage were to be the ause, it would have worked against what was observed. Shrinkage would have aused more severe raking in the speimens with higher stiffness reinforement. The differene in the flexural raking loads annot be explained by differenes in storage and handling either sine these were idential. The seemingly lower flexural raking load of speimen V-D-2 within its own series was due to its smaller depth. The FRP reinfored speimens in Series 1, whih had the same depth as this speimen, raked at approximately the same load. 50

63 4.5.3 Shear Strength The ultimate shear strength, V, and the shear at formation of the ritial inlined rak, u V, and the mode of failure for eah speimen are given in Table 4.5. Test variables suh as onrete ompressive strength and longitudinal reinforement ratio are also provided for ease of omparison. In this investigation, the ritial inlined rak is defined as the rak whose inlination has beome more than 45 o to the vertial and pointing towards the onentrated load or one whih has kiked bak towards the level of reinforement. In testing Shear Series 1 speimens, no speial effort was made in order to determine when the ritial inlined rak formed. Therefore, the values reported in Table 4.5 are found from rak patterns observed in photographs and skethes of Series 1 speimens. The ritial inlined raking loads found in this way were then ompared with the load defletion urves, where it was observed that the stiffness hanged slightly in the viinity of these loads. In general, it an be seen that the differene between the ritial inlined raking load and the ultimate load is small. In most ases these values were within 15% of eah other. When a/d ratios greater than 2.5 are used (3.4 in this study) and the longitudinal reinforement ratio is not high, redistribution of internal stresses is limited or annot take plae. Therefore, the inlined raking shears are typially around the same value as the ultimate shear. Only diagonal tension failures were observed in the tests. In the failure mode olumn of Table 4.5, DT indiates that the speimen failed in diagonal tension. The high yield strength steel bars in the speimen marked as Y-DT experiened yielding under the onentrated load prior to failing in diagonal tension. Table 4.5: Shear Test Results Series Speimen f (psi) ρ(%) V (kip) V u (kip) Failure Mode V-S DT 1 V-G DT V-G DT V-A DT V-S DT V-D Y-DT 2 V-G DT V-G DT V-A DT It an be seen that an inrease in the shear strength was observed as the longitudinal reinforement perentage inreased. However, the rate of inrease hanged from speimen to speimen. For example, the inrease in strength of Series 1 and Series 2 speimens reinfored with Glass 1 was 24% whereas it was 58% for speimens reinfored with Glass 2 bars. These perentages were adjusted by onsidering the hange in onrete strengths and that the shear strength is proportional to the f. To provide a learer view of the different shear strengths obtained in eah series, a bar hart is shown in Figure It is noted that the perentage inrease in the shear strength of FRP bar reinfored speimens were more than that in the steel reinfored speimens when the longitudinal reinforement ratio was inreased from 1% to 2%. From Series 1 to Series 2, the FRP bar reinfored speimens experiened an average inrease in shear strength of 44% whereas this inrease was limited to only 13% for steel reinfored speimens. If we ompare V-D-2 and V-S-2, the inrease in shear strength is 56% for an approximately 5-fold inrease in the longitudinal reinforement ratio. However, reall that the speimen V-D-2 yielded prior to failure. Yielding of the reinforement ould only have dereased the shear strength of this speimen and it would have arried slightly more shear had yielding not taken plae, thereby, dereasing the perentage inrease in shear strength. This figure also illustrates that the shear strength of the speimen reinfored with high 51

64 yield strength steel bars was approximately the same as that from speimens reinfored with the glass FRP bars in Series 2. Strength(kip) Series 1 Series 2 Glass1 Glass2 Aramid Steel Dywidag Figure 4.12: Shear strength of speimens Load-Defletion Curves Defletions of all the speimens were measured using linear voltage displaement transduers (LVDT) and reorded using a data aquisition system. Load vs. midspan defletions for eah test series were plotted on the same figure to allow for easy omparisons. Figure 4.13 shows load-defletion plots of Series 1 speimens while Figure 4.14 illustrates the same plot for Series 2 speimens. All load-defletion plots exhibit the same harateristis. Speimen behavior an be defined by noting four distint stages in the load defletion plots. Prior to inlined raking the urves are approximately linear. In this range, the stiffness of all speimens in a given series was the same. Sine the defletions prior to raking were unaffeted by the type of reinforing bar used, defletion alulations an be made by using either the gross-setion or transformed setion moment of inertia. Both result in approximately the same value for the reinforement ranges tested in this investigation. The seond stage was a transition stage, during whih the beam transformed into a fully raked state as new raks formed. In this stage, the stiffness gradually dereased. As the third stage is reahed, the stiffness was a funtion of the bar type and the amount of reinforement and remained relatively onstant. The beam stiffness dropped slightly in the fourth stage, whih ourred at a load slightly below the failure load. This final stage ended by failure due to diagonal tension. The transition stage (Stage 2) was longer in Series 1 FRP bar reinfored speimens as ompared to the rest of the speimens tested. In Series 1 speimens, it was observed that the lower the modulus of elastiity of the reinforing material, the lower the stiffness in the third stage. In Shear Series 2, the stiffness of glass FRP reinfored speimens (V-G1-2 and V-G2-2) and Speimen V-D-2, all of whih had similar axial stiffness of longitudinal reinforement, were almost the same. A straight line best fit to the data of eah speimen in their third stage gives approximately the same slope for Speimens V-G1-2, V-G2-2, and V-D-2. The ratio of the stiffness of steel reinfored speimens to that of FRP reinfored speimens alulated from the straight line best fit was approximately 4.0 in both Series 1 and Series 2. It is noteworthy that the beam stiffness ratio was roughly the same as the ratio of the modulus of elastiity of steel bar to that of the average of the modulus of elastiity of FRP bars, whih was measured as

65 P(kips) P (kips) V-S-2 80 V-A V-D-2 V-G2-1 V-G (in) V-S Figure 4.13: Load defletion urves for Series 1 speimens V-A-1 V-G1-1 V-G (in) Figure 4.14: Load defletion urves for Series 2 speimens 53

66 In both test series, the steel reinfored speimen attained the highest load followed by the Aramid FRP reinfored speimen followed by either of the Glass FRP reinfored speimens. However, in both series, the defletion levels obtained at failure by the FRP reinfored speimens were always larger than those of steel reinfored speimens. It was also notied that the defletions obtained by FRP reinfored speimens at ultimate within eah series were approximately the same. Furthermore, the average ultimate displaement at failure of FRP reinfored speimens was approximately 2.5 times the ultimate displaement reahed by steel reinfored speimens in both test series. 4.6 Data Analysis ACI Building Code The onrete ontribution to shear strength of the speimens in this investigation an be alulated by two methods in ACI Building Code. One of these equations is the well-known shear strength equation 2 f bwd, (Eqn. 4.1). Sine the amount of longitudinal reinforement was varied in this investigation, the strengths of the speimens were alulated also using an alternate equation, (Eqn. 4.2), whih inludes ρ as one of the variables. These two formulas are also used by the AASHTO Standard Speifiations, 16 th Edition. In addition the 1999 AASHTO LRFD Bridge Design Speifiations allows for the use of formula, 2 f bwd, when ertain restritions suh as when the member depth is less than 16-in. or a minimum amount of transverse reinforement speified in the ode is provided. where: 2 Vud V = 1.9 f ρ w bwd f b M 3. 5 u Vud 1.0 M u f b w d w d (Eq. 4.1) (Eq. 4.2) V : Nominal shear strength provided by onrete, lbs f : Speified ompressive strength of onrete, psi ρ w : Area of longitudinal tension steel divided by b w d V u : Fatored shear fore at setion, lbs M u : Fatored moment at setion, (in.-lbs) d : Distane from extreme ompression fiber to entroid of longitudinal tension reinforement, in. b w : web width, in. The strength of speimens omputed aording to the ACI proedures, (Eqn. 4.1 and 4.2) are tabulated in Table 4.6 along with the experimental results. A olumn hart presenting the ratio of experimental to alulated values by both methods is also shown in Figure The following observations were made from the omparison of the ACI alulated and experimental results. Equation 4.1 is only a funtion of the tensile strength of onrete ( f ) and, therefore, results in approximately the same shear strength regardless of reinforing material. Although Equation 4.2 inludes ρ as one of its variables, it does not take the effet of varying modulus elastiity of reinforing bars; thereby, resulting in unonservative alulations for FRP bar reinfored speimens. Furthermore, Eqn. 4.2 is relatively insensitive to the variations in the amount of longitudinal reinforement. The shear strengths of all FRP bar reinfored speimens were over-estimated by both proedures. Shear strength alulations for Series 1 and Series 2 speimens differ only slightly aording to the ACI proedures. However, the measured shear strengths of speimens reinfored with the same material but varying ρ were notieably different for both the steel and FRP bar reinfored beams tested. 54

67 Table 4.6 Analysis Results Speimen P u (kip) Eqn. 4.1 (kip) Eqn. 4.2 (kip) Eqn. 4.3 (kip) Eqn.4.4 (kip) Eqn. 4.5 (kip) V-S V-G V-G V-A V-S V-D V-G V-G V-A Eqn 4.1 Eqn Vexp /Val V-S-1 V-G1-1 V-G2-1 V-A-1 V-S-2 V-D-2 V-G1-2 V-G2-2 V-A-2 V-S-1 V-G1-1 V-G2-1 V-A-1 V-S-2 V-D-2 V-G1-2 Figure 4.15: Comparison of Strength Calulations by ACI Code V-G2-2 V-A-2 ACI-ASCE Committee 326 (ACI 326, 1962), whih was a sub-ommittee on shear and diagonal tension under the ACI Building Code main ommittee (Committee 318) first proposed Eqn. 4.2 for the onrete ontribution to shear strength (V ) in Committee 326 proposal was adopted by ACI 318 Committee without any hanges and first implemented in the ACI Building Code. Eqn 4.2 still serves as the bakbone of shear strength equations used in ACI Building Code. The data used in the development is plotted in Figure 4.16 together with Eqns 4.1 and 4.2. Note that Eqn. 4.1 is a lower bound to the data points from steel reinfored speimens. In addition the data obtained from the tests of FRP bar reinfored speimens onduted here are plotted for omparison. The data points for FRP bar reinfored onrete beams are notieably separate from the rest of the data 55

68 points. The ACI shear strength equation is an empirial equation, whih was developed as a reasonable fit to the available experimental data from steel reinfored onrete speimens. Therefore, satisfatory shear strength alulations of FRP bar reinfored onrete beams should not be expeted from this equation. 6 V f b d w Eqn 4.2 Eqn ACI 326 data FRP data ρ Vd M f Figure 4.16: ACI Committee 326 design urve development The ACI equation alulates the shear strength of FRP beams in this investigation poorly. Furthermore, the ACI 318 alulation proedure results in an inonsistent level of safety against shear failure for varying a/d ratios and longitudinal reinforement ratios even for steel reinfored onrete members. Therefore, the ACI equation is not suitable to determine the onrete ontribution to shear strength of FRP reinfored onrete members in this investigation ACI Committee 440 Proposed Design Reommendations ACI Committee 440 has proposed design reommendations for the determination of the shear strength FRP bar reinfored beams. There are two alternative equations proposed by the ACI Committee 440 (Proposals 1 and 2), both of whih are modified forms of ACI Building Code equation. Both will be used to analyze the beams in this investigation. The first method is presented in Eqn. 4.3 while the seond method is given in Eqn Proposal 1: Proposal 2: V = 2 f b w d E E FRP steel (Eqn 4.3) V = 2 f b w ρ FRP E d 90 β f 1 FRP (Eqn. 4.4) 56

69 where: b w : web width, in. d: effetive depth, whih is the distane between the extreme ompression fiber and the entroid of the tensile reinforement, in. E FRP : modulus of elastiity of FRP reinforement in tension, psi E steel : modulus of elastiity of reinforing steel, 29,000,000-psi f : onrete ompressive strength, psi β 1: fator for alulating the depth of Whitney stress blok as defined in ACI (Set ) ρ FRP : reinforement perentage of FRP bars The alulated shear strengths aording to Eqns. 4.3 and 4.4 are shown in Table 4.6. Also, the ratio of experimental to alulated strengths by both Eqn 4.3 and Eqn 4.4 are given in Figure Eqn 4.3 Eqn Vexp /Val V-S-1 V-G1-1 V-G2-1 V-A-1 V-S-2 V-D-2 V-G1-2 V-G2-2 V-A-2 V-S-1 V-G1-1 V-G2-1 V-A-1 V-S-2 V-D-2 V-G1-2 V-G2-2 V-A-2 Figure 4.17: Comparison of Strength Calulations by ACI 440 Proposed Reommendations The aim of the modifiation of the ACI Code alulation method reommended by ACI Committee 440 was to address the effet of the differene in the Young s modulus of FRP bars on the onrete ontribution to shear strength. The modifiation introdued (Eqn 4.3) results in the onrete ontribution to shear strength alulated for the FRP bar reinfored speimens very onservative ompared to the experimental values. The onservatism inreases as the reinforement ratio inreases from 1% to 2% in Series 1 and Series 2 respetively. The experimental results indiate that the shear strength is not a linear funtion of the modulus of elastiity or the axial stiffness of tensile reinforement. Therefore, Eqn 4.3 is not adequately alulating the onrete ontribution to shear strength of FRP reinfored onrete speimens in this investigation. To improve the auray of onrete ontribution to shear strength alulations, ACI Committee 440 proposed an adjustment to Eqn 4.3, whih resulted in Eqn However, shear strength alulations using Eqn. 4.4 results in results that are not more reasonable than those provided by Eqn The seond proposal by ACI 57

70 Committee 440 alulated the shear strengths of the speimens in Series 2 more aurately than those of Series 1 (Figure 4.17). Eqn. 4.3, however, should not be used for steel reinfored onrete beams sine the strength alulations for 1% and 2% steel reinfored beams are not reasonable. Proposal 1 resulted in dereasing alulation auray as the longitudinal reinforement ratio was inreased from 1% to 2% in Series 1 and Series 2 respetively (Figure 4.17). Proposal 2, however, resulted in inreasing alulation auray as the longitudinal reinforement ratio was inreased from 1% to 2% in Series 1 and Series 2 respetively (Figure 4.17). Both equations proposed by ACI Committee 440 for alulating the onrete ontribution to shear strength of FRP beams resulted in very onservative and uneonomial strength alulations. Furthermore, sine the formulas are based on ACI Code alulation method, they inherit the same shortomings of Eqn. 4.1 disussed previously. Therefore, the ACI Committee 440 proposed equations were not able to reasonably alulate the shear strength of the speimens in this investigation AASHTO LRFD Bridge Design Speifiations The shear strengths of speimens were also alulated using the 1999 AASHTO-LRFD Bridge Design Speifiations (AASHTO, 1996). The nominal shear resistane V n is alulated from the lesser value obtained from the two following equations: V = V + V + V n V = f b d + V n s v v p p (Eqn. 4.5) where: V = β f b d v v b v : effetive web width, in. d v : effetive shear depth in inhes taken as the distane, measured perpendiular to the neutral axis, between the resultants of the tensile and ompressive fores due to flexure; it need not be taken less than the greater of 0.9d e or 0.72h d e : effetive depth from the extreme ompression fiber to the entroid of the tensile fore in the tensile reinforement, in. h : overall thikness or depth of a member, in. f :speified ompressive strength of onrete at 28 days, unless another age is speified, ksi V p : omponent in the diretion of the applied shear of the effetive prestressing fore; positive if resisting the applied shear, kip β : fator relating effet of longitudinal strain on the the shear apaity of onrete, as indiated by the ability of diagonally raked onrete to transmit tension In this method, a setional design approah is followed in whih the shear strength of several setions along the length of a member is alulated and ompared to the demand at those setions. The β fator in the V term is alulated using design aids provided in the form of figures and tables. An iterative solution tehnique is neessary to obtain the orret value of β, whih is a funtion of the assumed rak inlination angle, rak spaing, provided reinforing material, amount of reinforement, and ultimate load that the member will experiene at the setion onsidered. The strength of speimens alulated aording to 1999 AASHTO LRFD Bridge Design Speifiations are provided in Table 4.6. The results of the alulation method are graphially presented in Figure Shear strength alulation by this method is relatively insensitive to the type of reinforement as well as to the reinforement ratios (0.36%, 1% and 2%) used in this investigation. For example, the alulated shear strengths for Series 1 and Series 2 speimens with the exeption of speimen V-S-2 are pratially the same regardless of the reinforing material or the reinforement ratio (Table 4.6). Figure 4.18 indiates that for Series 1 speimens, the shear strength alulations of FRP reinfored speimens were unonservative whereas that of steel reinfored speimen was onservative. Calulated shear strengths in Series 2 on the other hand were onservative for all speimens. Note that the fator of safety of the alulated shear strength inreased as the axial stiffness of the reinforing material ( ρ E ) inreased. It is also noted that the alulated shear strength of FRP bar reinfored speimens by this method was not sensitive to hanges in the longitudinal reinforement ratio. The alulated shear 58

71 strengths for ompanion FRP bar reinfored speimens in Series 1 and Series 2 (1% and 2%) were very lose although the reinforement ratio varied onsiderably. The method, however, was more sensitive to a hange in the steel reinforement ratio as the strength alulated for Series 2 steel reinfored speimen was higher than that of the Series 1 speimen Vexp /Val V-S-1 V-G1-1 V-G2-1 V-A-1 V-S-2 V-D-2 V-G1-2 V-G2-2 V-A-2 Figure 4.18: Comparison of Strength Calulations by 1999 AASHTO LRFD Code The 1999 AASHTO LRFD Bridge Design Speifiations equation (Eqn 4.5) is based on Modified Compression Field Theory developed by Collins and Mithell (Collins, 1997). The method used in the speifiations, however, is simplified by the introdution of tables and monograms as design aids. Although the development of the method is rational it was alibrated by using data from steel reinfored onrete members as previously disussed. In addition lower and upper bounds for the alulation of shear strength were added to these tables in the ode version. The lower and upper bounds in the ode regulates the values of β for given values of strain in the tensile reinforement and the rak spaing. Sine the axial stiffness of FRP reinfored speimens in this investigation were very low, the alulated reinforement strain for all FRP bar reinfored speimens fell below the lower bound speified in the ode. Therefore the 1999 AASHTO LRFD Bridge Design Speifiations equation was not apable of handling the effet of the hanges in the longitudinal reinforement ratio on the shear strength of FRP bar reinfored speimens in this investigation. Furthermore, the iterative nature of the method is not very pratial for design appliations. 4.7 Alternative Analysis To illustrate the effet of the lower modulus of elastiity of FRP bars on shear strength learly, the nominal shear strength data from the FRP reinfored beams obtained in this investigation as well as data obtained from the literature for steel reinfored beams was plotted vs. ρ as shown in Figure The shear strength was normalized by f. For the steel data plotted, the a/d ratio was less than 2.75 and the onrete strength, f, was less than 8500 psi. The longitudinal reinforement ratio for FRP data in Figure 4.19(a) indiates that the shear strength of FRP reinfored onrete speimens is lower than those of steel speimens with similar ρ. The FRP data points plot just below the lower limit of the steel data (Figure 4.19(a)). It is also noted from Figure 4.19(a) that the inrease in the nominal shear strength of FRP bar reinfored speimens normalized with respet to f (ρ is used instead of ρ eff ) is approximately the same as that for steel reinfored speimens with inreasing ρ. 59

72 In Figure 4.19(b) the same data used in Figure 4.19(a) is plotted vs. the effetive longitudinal reinforement EFRP ratio ( ρ eff = ρ ). Therefore, in Figure 4.19(b), the reinforement ratio of the FRP speimens was ESteel normalized by the ratio of the modulus of elastiity of FRP to steel (modular ratio). It is observed that the shear strength of the FRP reinfored onrete speimens follows the same trend as that of steel reinfored onrete beams. Therefore, the effet of the modulus of elastiity of the reinforement on shear strength an be taken into onsideration by the onept of effetive longitudinal reinforement ratio. There is onsiderable satter in the data whih ours due to differenes in the onrete strength, a/d ratio, and testing onditions (support onditions, loading equipment, loading shedule, et.). However, the data plotted in Figure 4.19(b) shows a trend the lower extremes of whih may safely be used as the limiting shear strength of reinfored onrete beams without transverse reinforement. One possible lower bound urve is illustrated. Based on the lower bound urve in Figure 4.19(b), the following equation is reommended to design FRP bar reinfored onrete beams without transverse reinforement and with a/d > 2.75: 6 5 ( ρ eff ) bd f ( ρ eff ) bd f Steel Data FRP-Data for ρ eff 2.0 for ρ eff > 2.0 (Eqn. 4.6) bd V u f ρ (%) (a) FRP Reinforement Ratio Not Normalized 60

73 6 5 Steel Data FRP-Data bd V u f ρ eff (%) (b) FRP Reinforement Ratio Normalized with Modulus of Elastiity Figure 4.19: Influene of ρ on v u A bar hart illustrating the results of the proposed method is presented in Figure 4.20 for omparison V-S-1 V-G1-1 V-G2-1 V-A-1 V-S-2 V-D-2 V-G1-2 V-G2-2 V-A-2 Vexp /Val Figure 4.20: Comparison of Strength Calulations by The Proposed Method Figure 4.20 indiates that the proposed method for the omputation of shear strength is onservative for the FRP reinfored speimens tested in this investigation. The method an safely be applied for alulating the shear 61

74 strength of both FRP and steel reinfored onrete beams and is simple in appliation. Therefore, the proposed method (Eqn. 4.6) is a simple and reliable way to alulate the onrete ontribution to shear strength of FRP reinfored and steel reinfored onrete without transverse reinforement. 62

75 5. CONSTRUCTION RECOMMENDATIONS 5.1 Introdution FRP reinforement is gradually beoming a viable solution for highway onrete strutures in orrosive environments. There are several demonstration and researh (D&R) projets utilizing FRP bars as onrete reinforement in Japan, Canada and Europe. In the U.S., there is growing interest in the transportation industry and a number of onrete highway bridge strutures have been built as D&R projets. These projets inlude: Rogers Creek Bridge Bourbon County, Kentuky, 1997 Rouge River Bridge Southfield, Mihigan, 1997 Buffalo Creek Bridge MKinleyville, West Virginia, 1997 Sierrita de la Cruz Creek Bridge Amarillo, Texas, 2000 The onstrution reommendations herein are intended to provide general guidane for the onstrution of FRP bar reinfored onrete strutures. The reommendations are based on field investigations before and during the asting of the Sierrita de la Cruz Creek Bridge as well as a literature survey (JSCE, 1997, ACI 440, 2000). 5.2 Handling And Storage The fibers in FRP bars are overed with a plasti resin matrix, whih protets them against physial damage and damage from environmental fators suh as moisture, ultraviolet light, ertain hemials, and the alkalinity of onrete. Durability of FRP bars is, therefore, strongly affeted by damage to the resin matrix. For example, glass FRP bars are very suseptible to degradation due to exposure of glass fibers to alkalis and aramid fibers in aramid FRP bars are suseptible to damage due to UV light exposure of the aramid fibers. Therefore, FRP bars, like epoxy oated steel bars, should be handled, stored, and plaed arefully to avoid damage. The following handling and storage guidelines are reommended to avoid damage to the bars and bar handlers: FRP bars should be handled with work gloves to avoid injuries to the bar handlers from exposed fibers or sharp edges. If utting is neessary, a dust mask is reommended FRP bars should be handled in suh a manner as to prevent damage to the surfae. If neessary, handling equipment should have padded ontat areas. Sine FRP bars are very flexible, bundles of FRP bars should be lifted with a strong bak, spreader bar, multiple supports, or a platform bridge. FRP bars should not be dropped or dragged. When neessary, utting should be aomplished with a high-speed grinding utter, a haksaw, or a fine blade saw. FRP bars should never be sheared or bent unless allowed by the manufaturer. FRP bars should be stored free from diret ontat with the ground. Proper supports to avoid exessive deformations should be provided for storage. Neessary measures should be taken to avoid exposure to exessive heat, diret sunlight, and hemials that are harmful to FRP bars. 5.3 Plaing & Assembling of Reinforement And Pouring of Conrete FRP bars should be aurately plaed to onform to the requirements as provided in the design drawings, details, and notes. Constrution praties suh as reinforement plaing and pouring of onrete is similar to that of steel reinforement and ommon praties should apply with the following exeptions: All FRP bars should be visually inspeted prior plaement. The bars should be free from defets suh as deep sores and uts and suh bars should be replaed immediately. If the FRP bar surfae is ontaminated with dirt, grease, oil, or other foreign materials, it should be leaned using appropriate methods and materials reommended by the bar manufaturer. The FRP bars should be transported to their plae in the forms in a manner to prevent exessive deformation and damage. All minor surfae damage and the ut ends of the bars should be oated with repair material as speified by the manufaturer or equivalent prior to onrete plaement. 63

76 FRP reinforement should be adequately supported using onrete hairs (preferably epoxy or plasti oated Figure 5.1). FRP bars are very flexible and may signifiantly sag during onreting if they are not supported at lose intervals. Therefore, the number of hairs supporting the FRP bars may have to be inreased to maintain the bars at the orret depth. The reinforement should be adequately seured in plae to prevent displaement due to onrete plaement operations. Sine the speifi gravity of FRP bars is very low, they should be seurely tied down to avoid floating of the bars in fresh onrete (Figure 5.2). Coated tie wire, plasti or nylon ties, and plasti snap ties may be used in tying the reinforement (Figure 5.3). Conrete an be plaed aording to ommon praties (Figures 5.4 and 5.5). Compation an be performed using internal or external vibrators (Figure 5.6). Sine FRP bars may be damaged by diret ontat with an internal vibrator, use of internal vibrators proteted with polyurethane is reommended (JSCE, 1997). The loation of bars should be arefully inspeted during onreting and if neessary onreting should be stopped until adequate measures to keep the bars in the orret loation are taken. The effets of uring temperature should be disussed with the manufaturer, and if neessary uring temperatures should be kept within ertain limits aording to the manufaturer s reommendations. Constrution should be planned to finish any potentially harmful operations to FRP bars prior to plaement of reinforement. Two examples of bar damage is illustrated in Figures 5.7(a) and (b) due to poor onstrution sheduling. Figure 5.7(a) illustrates a damaged bar due to rubbing of FRP bars to steel lifting hooks on the pre-ast onrete panels. Figure 5.7(b) shows a burned bar due to a flame utting operation very lose to the FRP bars. Figure 5.1: Epoxy Coated Steel Chairs (Sierrita de la Cruz Creek Bridge, Amarillo, Texas, 2000) 64

77 Figure 5.2: FRP Bars Seurely Tied Down to Prevent Floating (Sierrita de la Cruz Creek Bridge, Amarillo, Texas, 2000) Figure 5.3: Plasti Coated Steel Tie-Wires (Sierrita de la Cruz Creek Bridge, Amarillo, Texas, 2000) 65

78 Figure 5.4: Casting Operations (Sierrita de la Cruz Creek Bridge, Amarillo, Texas, 2000) Figure 5.5: Dek Finishing (Sierrita de la Cruz Creek Bridge, Amarillo, Texas, 2000) 66

79 Figure 5.6: Internal Vibrator Compation (Sierrita de la Cruz Creek Bridge, Amarillo, Texas, 2000) (a) Damaged FRP bar due to repeated rubbing (b) Burned FRP bar due to the use of a flame utter Figure 5.7: Damage to glass FRP bars due to poor onstrution planning (Sierrita de la Cruz Creek Bridge, Amarillo, Texas, 2000) 5.4 Quality Control In general, eah FRP bar manufaturer has their own resin formulation and manufaturing proess. As a onsequene, bars of the same fiber type (glass, arbon, aramid) but from different manufaturers, are likely to exhibit different mehanial properties, strutural performane, and durability. Sine, there are not well-established quality standards for FRP bars in the U.S., eah type of FRP bar should be tested prior to use to ensure that they meet the required performane riteria. It is reommended that the quality harateristis of the bars for use in onrete highway bridge strutures be determined using the following tests performed on at least 3 speimens. Tensile strength and tensile modulus of elastiity Fatigue Creep Coeffiient of thermal expansion Alkaline immersion 67