ASSESSMENT OF STRENGTH, STIFFNESS, AND DEFORMATION CAPACITY OF CONCRETE SQUAT WALLS REINFORCED WITH GFRP BARS

Transcription

1 Faulté de génie Département de génie ivil ASSESSMENT OF STRENGTH, STIFFNESS, AND DEFORMATION CAPACITY OF CONCRETE SQUAT WALLS REINFORCED WITH GFRP BARS Évaluation de la résistane, la rigidité et la apaité en déformation des voiles ourts en béton armé d armature en PRFV Thèse de dotorat Spéialité: génie ivil Ahmed Noureldean Mohamed Arafa A dissertation submitted in partial fulfillment of the requirements for the degree of Dotor of Philosophy (Civil Engineering) Jury: Prof. Brahim Benmokrane (direteur de reherhe) Prof. Murat Saatioglu (Examinateur) Prof. Ehab El-Salakawy (Examinateur) Prof. Sébastien Langlois (Examinateur) Sherbrooke (Québe) Canada August 2017

2

3 ABSTRACT The present study addressed the feasibility of reinfored-onrete squat walls totally reinfored with GFRP bars to attain reasonable strength and drift requirements as speified in different odes. Nine large-sale squat walls with aspet ratio (height to length ratio) of 1. one reinfored with steel bars (as referene speimen) and eight totally reinfored with GFRP bars were onstruted and tested to failure under quasi-stati reversed yli lateral loading. The key studied parameters were: (1) use of bidiagonal web reinforement; (2) use of bidiagonal sliding reinforement; and () web reinforement onfiguration (horizontal and/or vertial) and ratio. The reported test results learly revealed that GFRP-reinfored onrete (RC) squat walls have a satisfatory strength and stable yli behavior as well as selfentering ability that assisted in avoiding sliding shear that ourred in the ompanion steelreinfored wall following steel yielding. The results are promising regarding using GFRPreinfored squat walls in areas prone to seismi risk where environmental onditions are adverse to steel reinforement. Bidiagonal web reinforement was shown to be more effetive than onventional web reinforement in ontrolling shear-raks width. Using bidiagonal sliding reinforement was demonstrated to be not neessary to prevent sliding shear. The horizontal web reinforement ratio was found to have a signifiant effet in enhaning the ultimate strength and deformation apaity as long as the failure is dominant by diagonal tension. Existene of both horizontal and vertial web reinforement was shown to be essential for raks reovery. Assessment of the ultimate strengths using the available FRP-reinfored elements ode and guidelines (CSA S and ACI 440.1R-15) was onduted and some reommendations were proposed to attain a reasonable estimation of ultimate strengths. Given their importane in estimating the walls later displaement, the effetive flexural and shear stiffness of the investigated walls were evaluated. It was found that the raked shear stiffness ould be estimated based on the truss model; while the flexural stiffness an be estimated based on the available expressions in FRP-reinfored elements odes and guidelines. Based on a regression analysis, a simple model that diretly orrelates the flexural and shear stiffness degradation of the test walls to their top lateral drift was also proposed. Keywords: Squat walls, onrete, GFRP bars, seismi, hystereti response, sliding shear, residual displaement, web reinforement, flexural and shear deformations, stiffness. i

4 RÉSUMÉ La présente étude traite de la faisabilité de voiles ourts en béton armé totalement renforés ave des barres de polymères renforés de fibres de verre (PRFV), obtenant une résistane et un déplaement latéral raisonnable par rapport aux exigenes spéifiées dans divers odes. Neuf voiles à grande éhelle ont été onstruits: un renforé ave des barres d'aier (omme spéimen de référene) et huit renforés totalement ave des barres de PRFV. Les voiles ont été testés jusqu à la rupture sous une harge quasi-statique latérale ylique inversée. Les voiles ont une hauteur de 2000 mm, une largeur de 1500 mm (élanement 2000 mm/1500 mm = 1,) et une épaisseur de 200 mm. Les paramètres testés sont : 1) armature bi-diagonale dans l âme; 2) armature bi-diagonale dans l enastrement du mur à la fondation (zone de glissement); ) onfiguration d armature vertiale et horizontale réparties dans l âme et taux d armature. Les résultats des essais ont lairement montré que les voiles ourts en béton armé de PRFV ont une résistane satisfaisante et un omportement ylique stable ainsi qu'une apaité d'auto-entrage qui ont aidé à éviter la rupture par glissement à l enastrement (sliding shear). Ce mode de rupture (sliding shear) s est produit pour le voile de référene armé d aier après la plastifiation de l armature. Les résultats sont prometteurs onernant l'utilisation de voiles en béton armé de PRFV dans les régions sismiques dans lesquelles les onditions environnementales sont défavorables à l armature d aier (orrosion). L armature bi-diagonale en PRFV dans l âme s est avérée plus effiae pour le ontrôle des largeurs de fissures de isaillement omparativement à l armature répartie dans l âme. L'utilisation d'un renforement de isaillement bi-diagonal a été démontrée omme n'étant pas néessaire dans les voiles ourts en béton armé de PRFV pour prévenir la rupture par glissement à l enastrement (shear sliding). Par ailleurs, les résultats d essais ont montré que le taux d armature horizontale répartie dans l âme a un effet signifiatif sur l augmentation de la résistane et la apaité en déformation des voiles dont la rupture par effort tranhant se fait par des fissures diagonales (tension failure). L'existene d armature vertiale et horizontale répartie dans l âme du voile en béton armé de PRFV s'est révélée essentielle pour l ouverture et la fermeture des fissures au ours des hargements yliques. Les normes alul CSA S et ACI 440.1R-15 ont été utilisées pour évaluer la résistane au isaillement des voiles ourts en béton armé de PRFV. Certaines reommandations ont été proposées pour obtenir une ii

5 iii Résumé estimation raisonnable des fores ultimes. Compte tenu de leur importane dans l'estimation du déplaement latérale des voiles, la rigidité effetive en flexion et en isaillement des voiles étudiés a été évaluée. On a onstaté que la raideur de isaillement du béton fissuré pourrait être estimée en utilisant le modèle de treillis. La rigidité à la flexion peut être, quant à elle, estimée en fontion des expressions disponibles dans les normes et les guides de oneption de membrures en béton armé ave des barres en PRFV. Sur la base d'une analyse de régression, un modèle simple qui orrèle diretement la dégradation de la rigidité en flexion et en isaillement des voiles ourts en béton armé de PRFV testés ave le déplaement latérale dans la partie supérieure des voiles a également été proposé. Mots-Clés : Voiles ourts, béton, barres d armature en polymère renforé de fibre de verre (PRFV), sismique, réponse hystérétique, rupture par glissement, déplaement résiduel, armature de l âme, déformation due à la flexion et à l effort tranhant, rigidité.

6 ACKNOWLEDGEMENT Thanks to Almighty ALLAH for the graious kindness in all the endeavors I have taken up in my life. The author would like to express his gratefulness to the valuable advies and patiene of his supervisor, Prof. Brahim Benmokrane, and for giving him the opportunity to ondut suh researh in Sherbrooke University and providing him support at times when it was most needed. To Dr. Ahmed Sabry Farghaly, your passion to strutural engineering has been an inspiration to me. I annot thank you enough for the ountless enouragement, disussion, and support. This is beside the hand-by-hand work with dediation and devotion in every single step during the whole projet. Gratitude must be extended to Prof. Sebastien Langlois for his hospitality to ondut a large sale test at his faility. Sinere words of thanks must also go to our group tehnial staff; Mr. Martin Bernard, Mr. Mar Demers. I value the assistane of my olleagues Khaled Mohamed, Ahmed Hassanein, and Mohamed Gaber for their invaluable help during the experimental program. I am grateful for the sholarship granted to me by the Canada Researh Chair in Advaned Composite Materials for Civil Strutures and Natural Sienes and Engineering Researh Counil of Canada (NSERC-Industry Researh Chair program). To my parents, thank you for your ommitment to my eduation and for making me the person that I am today. To my brothers and sisters, thank you for your unonditional love and enourage, your patiene and supporting made this possible. iv Ahmed Noureldean Mohamed Arafa August 2017

7 TABLE OF CONTENTS ABSTRACT RÉSUMÉ ACKNOWLEDGEMENT TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES i ii iv v viii ix CHAPTER 1: INTRODUCTION General Bakground Objetives and Sopes 1.. Methodology 1.4. Thesis Outlines 4 CHAPTER 2: LITERATURE REVIEW Introdution Steel RC Squat Walls General Bakground Mode of Failures Diagonal Tension Failure Diagonal Compression Failure Sliding Shear Failure Fators Affeting Squat Walls Behavior Walls Aspet Ratio Presene of Boundary Elements Horizontal and Vertial Web Reinforement Use Diagonal Web Reinforement Constrution Joint Axial Load Shear Strength Predition FRP Composite Materials FRP Constituents Manufaturing Proess Mehanial Properties Tensile Strength Compression Strength Flexural Strength Shear Strength Fatigue Strength Bent Portion Strength FRP RC Strutural System for Seismi Fores Strutural Frame Systems v

8 vi Table of Contents Shear Walls 7 CHAPTER : EXPERIMENTAL PROGRAM 9.1 Introdution 9.2 Testing Matrix 9. Material Properties 44.4 Speimens Constrution 45.5 Preliminary Speimens Design 47.6 Test-setup 50.7 Loading Proedure 5.8 Instrumentations 5 CHAPTER 4: EXPERIMENTAL BEHAVIOR OF GFRP-REINFORCED CONCRETE SQUAT WALLS SUBJECTED TO SIMULATED EARTHQUAKE Introdution Experimental Program Test Matrix of Speimens Material Properties Speimens Design Test Setup and Proedure Instrumentations Test Results and Disussion General Behavior and Mode of Failures Hystereti Response Steel-Versus GFRP-reinfored Walls Effet of Different Configurations Predition of Ultimate Strength Energy Dissipation Conlusions 84 CHAPTER 5: EFFECT OF WEB REINFORCEMENT ON THE SEISMIC RESPONSE OF CONCRETE SQUAT WALLS REINFORCED WITH GLASS-FRP 86 BARS 5.1 Introdution Experimental Program Desription of Test Speimens Material Properties Test Setup and Proedure Instrumentations Experimental Results and Disussion Failure Progression and Hystereti Response Load-Top Displaement Envelop Curves Strain in Vertial Reinforement Strain in Horizontal Reinforement Shear Crak Width 107

9 vii 5..6 Influene of Web Reinforement on Conrete Shear Resistane Predition of Speimens Ultimate Strength Flexural Strength Shear Strength Comparison of Predited Ultimate Strength to Test Results Confinement Influene on Wall Response Conlusions 119 CHAPTER 6: EVALUATION OF FLEXURAL AND SHEAR STIFFNESS OF CONCRETE SQUAT WALLS REINFORCED WITH GLASS-FIBER- 122 REINFORCED-POLYMER (GFRP) BARS 6.1 Introdution Researh Signifiant Summary of Experimental Program and Results Deoupling of Flexural and Shear Deformations Original and Correted Flexural and Shear Deformations Shear Stiffness Design Codes and Guides for Shear-Strength Preditions Evaluation of Shear Crak Angle Evaluation of Conrete Shear Strength Proposed Shear-Stiffness Models Flexural Stiffness Conlusions 148 CHAPTER 7: CONCLUSIONS AND RECOMMENDATIONS General Conlusions Reommendations for Future Work Conlusion Reommandation pour des Travaux Futurs 158 REFERENCES 160

10 LIST OF TABLES Table.1 Reinforement details 42 Table.2 Reinforement mehanial properties 44 Table 4.1 Reinforement details and alulated apaities of the walls 62 Table 4.2 Reinforement mehanial properties 6 Table 4. Wall failure progression 7 Table 5.1 Conrete strength and reinforement details 91 Table 5.2 Summary of the test results 95 Table 5. Ultimate strength predition 117 Table 5.4 Evaluation of the onfinement effet 119 Table 6.1 Tensile properties of the reinforement 126 Table 6.2 Summary of the test results 128 viii

11 LIST OF FIGURES Figure 1.1 Corrosion in squat walls 2 Figure Strutural walls ategories based on height to length ratio 8 Figure Squat wall appliations 9 Figure 2. Diagonal tension failure; (a) shemati details; (b) photo (Woods et al. 2015) 10 Figure 2.4 Diagonal ompression failure 11 Figure 2.5 Sliding shear failure; (a) shemati detail; (b) Photo (Paulay et al. 1982) 12 Figure 2.6 Load transfer in squat walls (Barda et al. 1977) 1 Figure 2.7 Typial Shapes of Squat Walls 14 Figure 2.8 Shear resistane mehanisms in squat walls (Paulay 1972) 18 Figure 2.9 Sliding shear resistane omponents 22 Figure 2.10 Pultrusion proess 28 Figure 2.11 Stress strain for steel and FRP bars 29 Figure 2.12 Bends in GFRP bars reinforement 2 Figure 2.1 Effet of the lengthening of period on design fore levels 4 Figure 2.14 Measured shear strain at ultimate load (Mohamed et al. 2014b) 8 Figure.1 Overall dimensions of test walls 40 Figure.2 Reinforement details 42 Figure.2 Reinforement details (ontinue) 4 Figure. GFRP reinforement 45 Figure.4 Prepared formwork and age of the base 45 Figure.5 Assembly of the wall age to the base age 46 Figure.6 Assembly and alignment of wall formwork 46 Figure.7 Casting the base 46 Figure.8 Casting the wall 46 Figure.9 Speimens after uring 46 Figure.10 Test-setup 52 Figure.11 Loading history of testing program 5 Figure.12 Stain gauges instrumentation 54 Figure.1 LVDTs instrumentation 55 Figure 4.1 Conrete dimensions and reinforement details 60 Figure 4.2 GFRP reinforement and wall age 6 Figure 4. Test setup 68 Figure 4.4 Loading history 68 Figure 4.5 Instrumentation 69 Figure 4.6 Crak pattern 70 Figure 4.7 Failure progression of speimen S Figure 4.8 Failure progression of speimens G4-80, G6-80, and G4 72 Figure 4.9 Failure of speimen GD 72 Figure 4.10 Hystereti response 76

12 x List of Figures Figure 4.11 Envelope urves: Steel vs. GFRP 78 Figure 4.12 Envelope urves: Top displaement 80 Figure 4.1 Conrete strain 82 Figure 4.14 Energy dissipation 84 Figure 5.1 Conrete dimensions and details of reinforement 90 Figure 5.2 Test Setup 9 Figure 5. Displaement history 9 Figure 5.4 Instrumentation: (a) LVDTs instrumentation; (b) strain-gauge instrumentation 94 Figure 5.5 Crak pattern 96 Figure 5.6 Lateral load versus top displaement 97 Figure 5.7 Failure modes: (a) G4-250; (b) G4-160; () G4-80; (d) G6-80; (e) G-V; (f) G-H 99 Figure 5.8 Damage aspets 99 Figure 5.9 Load-top displaement envelope urves 102 Figure 5.10 Vertial strains distribution along the wall length 10 Figure 5.11 Maximum measured vertial strain 105 Figure 5.12 Horizontal strain distribution along the wall height 106 Figure 5.1 Load-average horizontal strain envelope urves 107 Figure 5.14 Load-shear rak width envelope urves 108 Figure 5.15 Vertial strain versus horizontal strain at the same loation (G4-250) 108 Figure 5.16 Maximum measured horizontal strain versus shear rak width 109 Figure 5.17 (a) Conrete shear strength versus top displaement envelope urves; (b) shear resistane omponents 111 Figure 6.1 Conrete dimensions, reinforement details, test setup, load history, and instrumentation 126 Figure 6.2 Failure modes of the test speimens 127 Figure 6. Load top-displaement envelope urves 128 Figure 6.4 Deoupling of flexural and shear deformations 129 Figure 6.5 Method for estimating α (rotation profile over the wall height) 10 Figure 6.6 Calulated urvature and rotation profiles for the test speimens 10 Figure 6.7 Comparison between the measured and alulated displaement at a height equal to the wall length 12 Figure 6.8 Displaement omponents at a height equal to the wall length 1 Figure 6.9 Comparison between flexural and shear deformations in test speimens 14 Figure 6.10 Truss model for shear deformation estimation 16 Figure 6.11 Lateral load versus experimental and predited shear deformations using different shear rak angles 19 Figure 6.12 Lateral load versus experimental and predited shear deformations using different onrete shear strength (θ = 60 o ) 140 Figure 6.1 Conrete shear strength versus top displaement 141 Figure 6.14 Shear stiffness degradation normalized with gross shear stiffness versus drift ratio 141 Figure 6.15 Validation of the proposed model for shear stiffness degradation 14 Figure 6.16 Comparison of predited normalized shear stiffness with experimental data 144 Figure 6.17 Comparison of Predited (a) flexural displaement (b) normalized flexural stiffness 147 Figure 6.18 Normalized flexural stiffness versus drift ratio 148

13

14 CHAPTER 1 INTRODUCTION 1.1. General Bakground Squat walls are defined as strutural walls with a height to length ratio (aspet ratio) less than 2.0. This type of wall is widely used as the primary seismi-fore resisting omponent in lowrise strutures suh as nulear failities, industrial buildings, and parking strutures. Moreover, suh walls also frequently serve as bridge piers and abutments. Squat walls may also be used in high-rise buildings where a substantial part of the lateral load may be assigned to them when extending over the first few stories above foundation level (Paulay et al. 1982). Beause of their aspet ratio, squat walls unlike slender walls generate high shear fores at their bases to develop strutural flexural strength, whih makes shear apaity a major issue in their design (Paulay et al. 1982; Kuang and Ho 2008; Whyte and Stojadinovi 2014). Investigations revealed that the flexural and shear deformations are intimately orrelated in squat walls. By the onset of flexural reinforement yielding, shear deformations either shear distortion and/or sliding have been shown to be ativated and loalized along the yielding zone then begins to dominate the behavior ausing rapid degradation in strength and stiffness with subsequent premature shear failure (Paulay et al. 1982; Saatioglu 1991; Sittipunt et al. 2001). In North Ameria, many bridges and other types of buildings in whih squat walls are used are defiient due to the orrosion of steel reinforement and onsequent failure in onrete (Figure 1.1). Some onditions, suh as, signifiant temperature flutuations and environmental aggression aggravate this phenomenon and make the hazard more severe. The high eonomi onsequenes of orrosion problems led engineers all over the world to searh for new and affordable onstrution materials as well as innovative approahes and systems to problem solving. In reent years, the use of fiber-reinfored polymer (FRP) as an alternative reinforing material in onrete strutures has emerged as an innovative solution to overome the orrosion problem. In addition to its non-orrodible nature, FRP reinforement 1

15 2 Chapter 1: Introdution presents many advantages suh as high strength-to-weight ratio, ease of handling, and immunity against the eletrohemial orrosion (Rizkalla et al. 200, Benmokrane et al. 2006, 2007). These advantages paved the way for their appliations into numerous onstrution elements suh as slabs, beams, olumns (Arafa et al. 2016a, El-Salakawy et al. 2005, Kassem et al. 2011, Tobbi et al. 2012). However, sine the investigations mainly foused on the behavior under stati-loading onditions omitting the seismi design; the feasibility of using FRP as internal reinforement for a omplete struture immune to orrosion while having strength, stiffness, and deformation apaity to resist seismi loads, has beome questionable. Figure 1.1 Corrosion in squat walls To address this issue, an experimental study was onduted by Mohamed et al. (2014a) to investigate the behavior of mid-rise shear walls totally reinfored with glass (G) FRP bars. Four large-sale shear walls one reinfored with steel bars (as referene speimen) and three totally reinfored with GFRP bars were onstruted and tested to failure under quasi-stati reversed yli lateral loading. The reported test results learly showed that properly designed and detailed GFRP-reinfored onrete (RC) walls ould reah their flexural apaities with no strength degradation while ahieving high level of deformability. The test results also revealed the potential of GFRP bars in ontrolling shear distortion ompared to steel reinforement (Mohamed et al. 2014b). It was explained that using elasti material (GFRP bars) gave uniform distribution of shear strain along the shear region of GFRP RC shear walls, resulting in a better ontrol of shear distortion ompared to steel RC shear wall, in whih the yielding of the flexural reinforement redistributed shear strains ausing their loalization at the yielding zone.

16 The onduted results of using FRP bars in mid-rise shear walls in term of its ability in ontrolling shear distortion and overriding the orrosion problem alled for a new investigation to study the feasibility of using this material in squat walls in whih these problems are frequently enountered. Currently; however, no experimental data are available on the seismi behavior of squat walls reinfored with FRP reinforing bars. This has been the main impetus to arry out this study to fill the gap of knowledge and provide information about this behavior Objetives and Sope Experimental program on large-sale squat walls reinfored with GFRP bars under reversed yli lateral loading was onduted. The ability of suh strutural element to ahieve the strength and drift requirements, speified in various odes, is the main sope of this study. Basially, the objetives of the urrent study an be summarized as follows: 1. Investigate the differene in behavior between GFRP and steel RC walls; 2. Assess the influene of bidiagonal web reinforement in lieu of horizontal and vertial web reinforement on the shear strength of squat walls;. Study the effiieny of bidiagonal sliding reinforement to suppress sliding failure; 4. Examine the effet of web reinforement onfiguration (horizontal and/or vertial) and ratio; 5. Assess methods to reasonably predit the ultimate flexural and shear strength; 6. Assess methods to alulate realisti flexural and shear deformations in squat walls; and 7. Evaluate the walls flexural and shear stiffness. 1.. Methodology To ahieve the foregoing objetives, a series of test speimens that omprised nine large-sale squat walls were onstruted and tested laterally under quasi-stati reversed yli loading up

17 4 Chapter 1: Introdution to failure. One wall was reinfored totally with steel bars and served as a ontrol speimen while the others were reinfored with GFRP bars in different onfigurations and reinforement ratios aording to the studied parameters. The preliminary design and reinforement details of the wall speimens were onduted aording to the CSA A2. (2014) for steel RC wall and the CSA S806 (2012) for GFRP RC walls. Given the absene of seismi provisions in the CSA S806 (2012); however, similar philosophies that are being used in the ompanion ode, CSA A2. (2014), were adopted. Analysis of the experimental results in term of deformability, energy dissipation, ultimate strength, and failure mode was onduted showing the main aspets of differene between the behavior of steel and GFRP RC squat walls. Evaluation of the influene of either bidiagonal web reinforement or sliding reinforement was disussed. The experimental results were also analyzed, identifying the effet of different web reinforement onfigurations and ratio on the rak pattern and failure mode, drift apaity and ultimate strength as well as the shear rak widths. Doumentation of the strain distribution in either horizontal or vertial diretion was also presented. Evaluation of the ultimate apaity aording to the ACI and CSA odes provisions was also introdued. As a result, some reommendations that assist in a reasonable estimation of the ultimate strength were given. The experimental measurements permitted to deouple flexural and shear deformations and examine their ontribution to the lateral deformations. The results showed that shear deformation signifiantly inrease by the onset of the first shear rak initiation and represented a signifiant portion of the total deformations. Consequently, predition of shear stiffness of squat walls beside its flexural stiffness was found to be neessary. The available expressions in guidelines and odes as well as some tehnial papers were used in lateral shear and flexural stiffness predition. Additionally, in the ontext of displaement based design, a simple model that diretly orrelates the flexural and shear stiffness degradation of the tested walls to their top lateral drift was proposed Thesis Organization The thesis onsists of seven hapters. The ontents of eah hapter an be summarized as follows:

18 5 Chapter 1 of this thesis presents bakground information on the researh topi, the work objetives and the adopted methodology. Chapter 2 introdues a literature review reporting the past known harateristis of steel RC squat walls. Some aspets suh as failure modes and the most deisive parameters that affet the walls behavior are presented. Additionally, the available methods in the ACI 18 (2014) and CSA A2. (2014) for ultimate shear strength s predition are introdued. This is followed by a bakground about some mehanial harateristis of FRP reinforement. Finally, the available knowledge of seismi behavior of FRP RC systems is disussed. Chapter : gives the details of the experimental program and the testing proedure. The geometry and reinforement details of the test speimens, web reinforement onfiguration, test setup and proedure, and the instrumentation details are presented. In addition, detailed harateristis of the used materials are provided. The subsequent three hapters respetively orrespond to three tehnial papers and one tehnial note that have submitted for publiation in sientifi journals: Chapter 4: (Paper 1) Arafa, A., Farghaly, A. S., and Benmokrane, B., 2016 (submitted) Experimental Behavior of GFRP-Reinfored Conrete Squat Walls subjeted to Simulated Earthquake Load, Journal of Composites for Constrution, ASCE. Chapter 5: (Paper 2) Arafa, A., Farghaly, A. S., and Benmokrane, B., 2017 (submitted) Effet of Web Reinforement on the Seismi Response of Conrete Squat Walls Reinfored with Glass-FRP Bars, Engineering Strutures. (Paper ) Arafa, A., Farghaly, A. S., and Benmokrane, B., 2017 (submitted tehnial note) Predition of Flexure and Shear Strength of Conrete Squat Walls Reinfored with Glass-FRP Bars, Journal of Composites for Constrution, ASCE. Chapter 6: (Paper 4) Arafa, A., Farghaly, A. S., and Benmokrane, B., 2017 (submitted) Evaluation of Flexural and Shear Stiffness of Conrete Squat Walls Reinfored with Glass-Fiber-Reinfored-Polymer (GFRP) Bars, ACI Strutural Journal.

19 6 Chapter 1: Introdution Chapter 7 presents a general onlusion of the results obtained from the experiments and analyses with respet to the problems and observations disussed throughout the thesis in addition to reommendations for future work.

20

21 CHAPTER 2 LITERATURE REVIEW 2.1. Introdution In this hapter, a survey of relevant previous works related to this researh study is presented. A review for the behavior of steel RC squat walls under seismi loading is first presented. Emphasis is given on the identifiations of failure modes and main fators that affet the behavior. In addition, a summary of the available models in the ACI 18 (2014) and CSA A2. (2014) that are being used in prediting the ultimate shear strength of steel RC walls is presented. This is followed by a brief summary about FRP material onstituents, manufaturing, and properties as well as the available knowledge of seismi behavior of FRP RC systems Steel RC Squat Walls General Bakground Reinfored onrete walls are ommonly used as the primary omponent for lateral loadresisting system in buildings prone to seismi risk and/or wind pressure. Compared to frametype strutures, the main advantages that an be rendered by strutural walls are the signifiant inrease in building s lateral stiffness whih leads to a redution of seond-order effets and subsequent inrease of safety against ollapse (Fintel 1995, Paulay 1972). Even after extensive raking, the strutural walls are able to maintain most of their vertial load-bearing apaity whih is not always the ase for frame-type systems. These advantages were reognized and evidened from the observation during the Marh 1985 Chilean earthquake (Wyllie et al. 1986); where buildings equipped with well-designed RC walls showed overwhelming suess in ontrolling damages. Strutural walls are generally lassified based on its height to length ratio into three ategories; slender walls, mid-rise walls, and squat walls. Slender walls are strutural walls with a height 7

22 8 Chapter 2: Literature Review to length ratio larger than 4.0. Suh type of walls is used in high rise buildings and it an be treated as ordinary reinfored onrete antilever beam. In slender walls, it is relatively easy to ensure developing adequate dutility as flexural behavior is dominated. Strutural walls with a height to length ratio between 2.0 and 4.0 are lassified as mid-rise walls and widely used in mid-rise buildings (buildings with 4 to 10 stories). Both nonlinear flexural and shear deformations signifiantly ontribute to the lateral response of suh shear walls. The last ategory is squat walls with a low height to length ratio that is typially less than 2.0. V V hw hw lw (b) Mid-rise walls (4.0>h w /l w > 2.0) V hw lw lw (a) Slender walls (h w /l w > 4.0) () Low-rise walls (h w /l w < 2.0) Figure Strutural walls ategories based on height to length ratio

23 9 Squat walls are widely used in low-rise buildings suh as nulear plants, industrial buildings; parking strutures, highway overpasses, and bridge abutments (Figure 2.2 shows some appliations of using squat walls). Squat walls may also be used in high-rise buildings where a signifiant portion of lateral load an be assigned to them (Paulay et al. 1982). (a) (b) () (d) (e) Figure Squat wall appliations; (a) wall in nulear reator (Whyte and Stojadinovi. 201), (b) bridge piers, () overpass piers, (d) bridge abutment, (e) low-rise housing made of walls and slabs (Sánhez-Alejandre and Aloer 2010) In squat walls, relatively large shearing fores are generated at the wall base, that are suffiient to destroy the struture in brittle shear manner before ahieving its flexural strength. This is mainly due to the ombination of a squat (low height to length ratio) and the uniform distribution of vertial reinforement aross the wall setion. Therefore, squat wall behavior is generally dominated by shear deformations (shear distortion and/or sliding shear deformations) and its shear apaity onstitutes a major onern in their design. Numerous experimental and analytial investigations have been devoted to study the behavior of steel RC squat walls under quasi-stati reversed yli lateral loading as a simulation for seismi loading. Based on these studies; failure modes, hystereti behavior harateristis as well as the main parameters affeting the behavior and ultimate shear strength were identified. These

24 10 Chapter 2: Literature Review issues will be disussed in the following subsetions in addition to the odes methods for ultimate shear strength estimation Mode of Failures Diagonal Tension Failure Failure due to diagonal tension ours when horizontal web reinforement is inadequate (Figure 2.a). In suh ase, horizontal reinforement yields and widely spaed diagonal raks appear. The failure then ours suddenly through a sliding along diagonal plane assoiated with the rupture of the horizontal reinforement rossing this plane as shown in Figure 2.b. Test results demonstrated that the inlination of failure plane an be affeted by the wall s aspet ratio (height to length ratio) and the existene of a beam at the top of the wall (Paulay et al. 1982). The results also showed that avoiding suh types of failure an be ahieved by providing horizontal shear reinforement apable of transferring a larger shear fore than the one whih produes flexural yielding. Horizontal bars rupture (b) (a) Figure 2. Diagonal tension failure; (a) shemati details; (b) photo (Woods et al. 2015) Diagonal Compression Failure If adequate horizontal shear reinforement is provided and the average shear stress in the wall setion is large, onrete may rush under diagonal ompression. Resistane of the onrete ompression struts in the web of the wall deteriorates as the inlined raks in two opposite diretions open and lose suessively under yli loading. Ultimately, the onrete struts are rushed as shown in Figure 2.4a. This type of failure mode ours for walls with a very high shear stress suh as walls with flanges or barbells and/or walls with a high axial load. Flanged and barbell walls an potentially aommodate more reinforement at the wall ends, whih

25 11 provides substantial flexural strength and inrease the shear demands in the wall web. In spite of the preferable effet of axial load in term of ontrolling shear rak width and inreasing the shear strength; their existene with large value inrease the ompressive stresses in the web of the wall and aelerate the ourrene of diagonal ompression failure. Diagonal ompression failure is usually assoiated with dramati and irreoverable loss of strength. Therefore, diagonal ompression failure is highly undesirable (Paulay et al. 1982). This mode of failure an only be avoided if the average shear stress in the wall ritial setion is limited between 0.5 f 1/2 and 0.9 f 1/2 ; as a funtion of the dutility requirements imposed on the wall (Park and Paulay 1975, Oesterle et al. 1980). An example of a diagonal ompression failure from Maier and Thürlimann (1985) is shown in Figure 2.4b. (a) (b) Figure 2.4 Diagonal ompression failure; (a) shemati detail; photo (Maier and Thürlimann 1985) Sliding Shear Failure Sliding shear failure ours when the wall has suffiient horizontal reinforement to prevent a diagonal tension failure, and relatively small amount of vertial reinforement in the wall web with low axial loading. Inelasti deformation required for energy dissipation would be expeted to be reated mainly from post yielding strains originated in the vertial flexural reinforement. However, after a few yles of reversed loading that auses signifiant yielding in the flexural reinforement, sliding displaement an our along flexural raks that interonnet and form a ontinuous horizontal shear path, as depited in Figure 2.5a. Suh sliding displaements are responsible for a signifiant redution of stiffness, partiularly at

26 12 Chapter 2: Literature Review low load intensities, onsequently, a redution of energy dissipation (Paulay et al. 1982). This mode of failure is also responsible for signifiant degradation of stiffness. Typial sliding shear failure reported by Paulay et al. (1982) is shown in Figure 2.5b. (a) (b) Figure 2.5 Sliding shear failure; (a) shemati detail; (b) Photo (Paulay et al. 1982) Fators Affeting Squat Walls Behavior The experimental database ompiled from previous experimental and analytial researhes revealed that the shear behavior of squat wall was governed by many fators suh as wall s aspet ratios, presene of boundary elements (barbells or flanges), horizontal and vertial web reinforement, diagonal web reinforement, onstrution joint as well as the magnitude of axial loading. Summary about the effet of eah parameter is briefly presented in the following subsetions Wall s Aspet Ratio There is an agreement that aspet ratio (height to length ratio) is one of the most ruial parameters that affets the shear strength, deformability and mode of failure of squat walls. The work onduted by Barda et al. (1977) is one of the early investigations that studied this parameter. Eight flanged squat walls with aspet ratios of 0.25, 0.5, and 1.0 were onstruted and tested under lateral loading. It was observed that walls with lower aspet ratio possess high shear strength; the shear strength of speimen with aspet ratio of 0.5 was found to be 20% higher than the shear strength of the ompanion speimen with aspet ratio of 1.0. This

27 1 inrease an be explained based on load transfer mehanism whih was reported by the authors. Barda et al. (1977) reported that a signifiant amount of shear transmitted in squat walls from the top slab to the foundation by the so-alled lattie ation (arh ation). This mehanism onsists of the vertial wall reinforement ating in tension (tie) and onrete struts in the wall between inlined raks ating in ompression (Figure 2.6). One of the most ruial parameter that influenes the strut effiieny is the angle between strut and tie; the strut will lose its apaity as it approahes the diretion of the tie. Sine this angle is higher in low aspet ratio; it is therefore the walls with low aspet ratios that have higher peak shear strength ompared to taller walls with similar properties. The inrease of shear strength with lower aspet ratio was also doumented by Gule and Whittaker (2009) based on statistial study inluding the results of 44 test results of squat walls subjeted to lateral load. The effet of aspet ratio on the shear strength has also been onsidered by most of the available models for shear strength estimations of squat walls; for example, Barda et al. (1977); Hwang et al. (2001); Gule and Whittaker (2009); and Kassem (2015). Nevertheless; ontradition regarding the effet of aspet ratio is still found between the ACI 18 (2014) and CSA A2. (2014) methods for shear strength predition. Whereas; the aspet ratio effet is introdued within the onrete shear ontribution term in the ACI 18 (2014), the CSA A2. (2004) does not aount for this parameter in shear strength estimations. V u Conrete Strut Conrete Strut h w Vertial Reinforement l w Figure 2.6 Load transfer in squat walls (Barda et al. 1977) In addition to their influene on shear strength, lower aspet ratio was also found to result in stiffer strutures with lower deformability. These remarks were evidened from the behavior

28 14 Chapter 2: Literature Review of 26 squat walls with aspet ratios ranged from 0.7 to 2.0 (Hidalgo et al. 2002). Other test results also indiated that walls with a low aspet ratio may be more vulnerable to sliding shear than those with a higher aspet ratio as reported by Salonikios et al. (1999 and 2000) Presene of Boundary Elements Squat walls are generally grouped by plan geometry into retangular walls or walls framed by either end boundary olumns (barbell walls) or boundary flanges (flanged walls) (Figure 2.7). Boundary elements are often presented to allow effetive anhorage of transverse beams. Even without beams, they are often provided to aommodate the prinipal flexural reinforement, to provide stability against lateral bukling of a thin-walled setion and to provide more effetive onfinement of the ompressed onrete in the potential plasti hinge. Walls meeting eah other at right angles will give rise to flanged setions. Suh walls are normally required to resist earthquake fores in both prinipal diretions of the building. a) Retangular wall b) Barbell wall ) Flanged wall Figure 2.7 Typial Shapes of Squat Walls A review of the literature revealed that most previous studies on squat walls mainly foused on a single shape; hene the effet of this parameter on squat walls behavior ould only be estimated by omparing results for similar walls tested in different programs. More speifially, only the works onduted by Paulay et al. (1982) and Maier and Thurlimann (1985) have two types of wall shapes; retangular, and flanged. Paulay et al. (1982) reported that flanged walls are more seriously affeted by sliding shear along interonneting flexural raks. Maier and Thurlimann (1985), on the other hand, reported that the horizontal

29 15 resistane is a funtion of the ross-setional geometry; adding boundary elements improve the strength of squat walls. The onept of omparing results for similar walls tested in different programs with the objetive of estimating the ruial parameters and validating some proposed analytial models in squat walls was followed by some researhers (Hwang et al. 2001, Gule and Whittaker 2009; Kassem 2015). Hwang et al. (2001) reviewed and atalogued the results of 62 reinfored onrete squat walls with different shapes with the aim of alibrating a proposed softened strut and tie model for the determination of shear strength of squat walls. The investigation showed that walls with barbell or flanged ross-setion resist signifiantly higher shear fores than a retangular wall with the same amount and arrangement of web reinforement. Some flanged walls ahieved 20% higher shear strength than other idential retangular walls. The authors explained that the higher apaity of flanged walls an be attributed to the improved end onditions of its diagonal strut provided by the ompression boundary element. The authors added that the depth of ompression strut of a wall inreases with the presene of boundary elements whih leads to an inrease in the shear strength. The same onlusions were drawn by Gule and Whittaker (2009) and Kassem (2015) based on a database of information from tests of 44 and 645 squat walls, respetively. Kassem (2015); however, added that the existene of boundary elements ause additional onfining effet whih redues the onrete softening effet and limits the raking and extension of the walls; subsequently, inreasing in the shear apaity. Even though the review emphasized on the effetiveness of boundary elements in enhaning squat wall s shear strength; urrent ACI 18 (2014) and CSA A2. (2014) ignore the rosssetion effet on shear strength preditions; resulting in less-than-satisfatory estimates with a disagreement with the experimentally observed strutural behavior. This remark with other defiienies has been reported by many researhers (for example, Hwang et al. 2001, Gule and Whittaker 2009, Kassem 2015) who pointed to the importane of developing more rational predition methods apable of fully haraterizing the real response of RC squat walls under various loading onditions while onsidering those parameters that signifiantly influene the behavior of suh walls.

30 16 Chapter 2: Literature Review Horizontal and Vertial Web Reinforement Previous investigations on steel RC squat walls have onfirmed that minimum horizontal and vertial web reinforement are essential for rak ontrol. Nevertheless, there is no onsensus between researhers about the effet of horizontal and vertial web reinforement on the shear strength. Some researhers reported that using proper amount of horizontal reinforement suppresses the diagonal tension, hene inreasing in the shear and deformation apaity of squat walls (Paulay et al. 1982; Pilakoutas and Elnashai 1995; Hidalgo et al. 2001). In ontrast to what adopted by those researhers, other experiments showed that horizontal web reinforement has no impat on the shear strength of squat walls, whereas, it signifiantly inreases as a funtion of vertial web reinforement (Maier and Thurlimann 1985; Wood 1990; Lefas et al. 1990; Gupta and Rangan 1996; Emamy Farvashany et al. 2008). In between the first and seond opinion, other investigations; however, indiated that the presene of both horizontal and vertial web reinforement ativate more ontribution of onrete for shear resistane mehanism by providing additional load paths beside the diagonal strut, and hene inrease the shear apaity (Cardenas et al. 1980, Hwang et al. 2001, Kassem 2015). Current ACI 18 (2014) and CSA A2. (2014) methods for shear strength estimation of squat walls only aount for the amount of horizontal web reinforement. Nevertheless, both odes reognized that vertial web reinforement is essential to maintain equilibrium of internal fores. Considering the Barada et al. (1972) onlusion that for squat walls having aspet ratio less than 1.0, the shear fores are transmitted to the base by inlined struts developed within the web (Figure 2.8a), the inlined struts fores must be vertially in equilibrium by vertial web reinforement (Figure 2.8b), while both horizontal and vertial web reinforement ontribute in equilibrium at the wall edges (Figure 2.8) (refer to Paulay 1972). The vertial web reinforement is therefore reommended by both the ACI 18 (2014) and CSA A2. (2014) to be a ratio of the horizontal web reinforement. Apparently from Figure 2.8, the ratio that satisfies equilibrium is typially equal to ot 2 θ; where θ is the struts inlination angle to the longitudinal axis. The ACI 18 (2014) onservatively assume that θ = 45 ; hene, it is required that the vertial web reinforement should be equal to the horizontal web reinforement for all squat walls; regardless their aspet ratio. The CSA A2. (2014); however, provide Eq. 2.1, to alulate θ that is based on the modified ompression field theory

31 17 and is a funtion of the wall depth, reinforement axial rigidity, and internal fores applied at the setion of interest l (2.1) where εl = average longitudinal strain at mid-height of the setion of interest and an be alulated as follows: l M f V dv 2E A F F f (2.2) where Mf, Vf, are the applied moment and shear at the setion of interest, dv is the effetive shear depth, EF and AF are the modulus of elastiity and the total ross-setional area of the longitudinal reinforement at the same setion. Based on the alulated θ, the CSA A2. (2014) gives the following equation to alulate vertial web reinforement ratio: v P 2 s h ot (2.) s f y Ag where ρv, ρh are the vertial and horizontal web reinforement, Ps is the axial load; φs is the material resistane fator fy is the speified yield strength of web reinforement; and Ag gross area of the wall setion. It an be notied from Eq. 2. that the CSA A2. (2014) aounts for another omponent in fores equilibrium that is the ompression stresses applied on the wall. This effet; however, is onservatively ignored in the ACI 18 (2014). Furthermore, whereas the ACI required that vertial web reinforement must equal horizontal web reinforement regardless their aspet ratio, the CSA A2. (2014) limits Eq. 2. appliation for squat walls having an aspet ratio less than 1.0. In spite of the aforementioned disrepany between the ACI 18 (2014) and CSA A2. (2014), both odes agree that minimum horizontal and vertial web reinforement (equal to 0.25%) should be provided to ontrol raks propagation and width.

32 18 Chapter 2: Literature Review V u V u Conrete Struts V u 2 1 h w Vertial omponents θ Horizontal and vertial omponents l w (a) Wall under loading (b) Equilibrium at zone 1 ()Equilibrium at zone 2 Figure 2.8 Shear resistane mehanisms in squat walls (Paulay 1972) Use Diagonal Web Reinforement Extensive researhes have been devoted to study the influene of using diagonal reinforement; either onentrated at the base to redue the exessive sliding, or distributed in the web to ontrol shear distortion. Consistent results about this effet have been drawn. Using diagonal reinforement was found to onsiderably improve the seismi response of squat walls ompared to that with onventional web reinforement (horizontal and vertial). The aspets of enhanement inluded preventing sliding shear and diagonal ompression failure, redution in shear distortion and relatively high energy dissipation. In spite of the merits of using diagonal reinforement, it may attribute higher ost than using horizontal and vertial bars. Given the fat that the diagonal reinforement needs more labor time to be ut in different lengths as well as the diffiulty in their plaing and anhorage relatively to using horizontal and vertial reinforement. Historially, the first attempt for applying diagonal reinforement in squat-walls was onduted by Iliya and Bertero (1980). Two walls with aspet ratio of 1. were onstruted; the first speimen was reinfored with equal amount of horizontal and vertial web reinforement, while the web reinforement in the other speimen was arranged diagonally with angle of 45. The speimens were ylially loaded up to the first yield of the longitudinal steel in the boundary elements. The raks in the speimens were then repaired by epoxy grouting. The repaired speimens were subsequently loaded, with a few intermediate yles, up to failure. Finally, the damaged walls were retrofitted and again subjeted to yli loadings until failure. The test results showed that the 45 arrangement of the wall reinforing

33 19 bars is more effetive in resisting the effet of shear reversals; the speimen with diagonal reinforement exhibited less stiffness and strength degradation under the yli load. In addition, the notied failure in onventionally reinfored speimens was mainly due to diagonal raking, whereas in the speimen with diagonal web reinforement the flexural failure was dominant. In order to suppress the detrimental effet of sliding shear failure at the base of squat walls, speial bidiagonal reinforement extending from the base through the web was provided in two squat walls tested by Paulay et al. (1982). These walls dupliated two other onventionally reinfored walls with equal flexural and shear strengths. It was onluded that bidiagonal reinforement onsiderably improve seismi response of squat walls, even when as little as 0 perent of the applied shear was resisted by suh reinforement. Diagonal reinforement used in these tests was insuffiient to prevent slip. However, when the diagonal bars were yielding due to slip displaements, signifiant energy dissipation additional to that due to flexure resulted. Salonikios et al. (1999) arried out a omprehensive experimental program involving eleven wall speimens; six with aspet ratio of 1.5 while the others have aspet ratio of 1.0. The wall speimens were reinfored against shear, either onventionally (orthogonal grids of web reinforement), or onventionally plus bidiagonal bars. Using bi-diagonal reinforement was found to offer an attrative alternative to urrent pratie from an eonomi point of view, sine for a lower quantity of total web reinforement an improved seismi performane was ahieved. The authors also reported that the using bidiagonal bars ontributed to better ontrol of the inlined shear raks width in the web of speimen. The main reason for enhanement is that bidiagonal bars interset the inlined shear raks almost at right angles; hene, they work essentially in diret tension, whereas the bars in the orthogonal grid interset the shear raks at 5 to 45 and tend to work primarily as dowels. Sittipunt and Wood (1995) arried out an analytial study using finite element models to investigate the effet of using diagonal web reinforement. The study omprised analysis of six walls with varying arrangements of web reinforement. Conventional web reinforement (vertial and horizontal web reinforement) was used in the first speimen. The amount of

34 20 Chapter 2: Literature Review horizontal web reinforement in the lower half of the wall is doubled in seond wall, and the amount of vertial web reinforement in the lower half of the wall is tripled in the third wall. Inrease of vertial web reinforement was also tested in the fourth wall but the bars were not anhored in the foundation. The final two arrangements of reinforement inluded equal diagonal web reinforement in the lower portion of the wall (either distributed or onentrated), in addition to vertial and horizontal web reinforement. The main drawn onlusion was that hystereti response of squat walls an be signifiantly improved by ontrolling the shear distortion near the base. However, the mehanism by whih fores are arried in the web of the wall must be hanged if the overall fore-displaement response is to be hanged appreiably. Adding vertial or horizontal web reinforement did not improve the hystereti response signifiantly beause fores were still arried aross the raks in the web by dowel ation. Therefore, the possibility of web rushing at a given level of deformation was not dereased. When diagonal reinforement was added in the web, the load-arrying mehanism was hanged to diret tension in the reinforing bars, and the shear distortion was redued signifiantly. Continuing their efforts in studying the effet of diagonal web reinforement, Sittipunt et al. (2001) tested four squat walls with aspet ratio of 1.6 and inorporating different web reinforement ratios and onfigurations. Two walls ontained different amount of horizontal and vertial web reinforement, while the others ontained diagonal web reinforement (45 orientation) with amount equal to the ompanion speimens whih was reinfored onventionally. All speimens were tested under horizontal quasi-stati yli load up to failure with absene of axial load. The test results onfirmed the outomes obtained from the pre-desribed analytial investigation. It was reported that the brittle mode of failure due to web rushing ould be avoided by using diagonal web reinforement; both walls with onventional web reinforement failed due to web rushing. Pinhed shapes haraterized the hysteresis urves for top displaement and shear distortion near the base. In ontrast, the walls with diagonal reinforement displayed rounded hysteresis urves and failed due to rushing of the boundary elements. It was also onluded that the advantages in performane of speimens reinfored with diagonal reinforement an offset the diffiulties assoiated with plaement of diagonal bars during onstrution. Other investigations have been onduted later (Chiou et al.

35 , Liao et al. 2004, Shainghin et al. 2007, Zhong et al. 2009) and yielded to similar results Constrution Joint Constrution joints in squat walls, espeially for those under low axial load, may dramatially deteriorate under yli load and beome the weakest link in the hain; leading to sliding shear failure (Doostdar 1994). After a few yles of reversed loading that auses signifiant yielding in the flexural reinforement, sliding displaement an our along flexural raks that interonnet and form a ontinuous, approximately horizontal shear path (Paulay et al. 1982). Suh sliding displaements are responsible for a signifiant redution of stiffness, partiularly at low load intensities, and onsequently, a redution of energy dissipation. Therefore, to ensure strutural resistane of squat walls and ensure energy dissipation, sliding shear aross a onstrution joint should be avoided (Paulay et al. 1982; Salonikios et al. 1999, and 2000). This an be ahieved through a rational design of the onstrution joint suh that it has a shear apaity larger than the shear apaity of the wall s web; hene, it will not onstitute the weakest link in the load transfer. As mentioned earlier, one sliding movement ommenes, major horizontal rak spreads through the entire onstrution joint and start to widen. Thus, any approah that an delay or prevent the widening of this rak will signifiantly promote the shear resistane along the onstrution joints. Defining of these approahes; however, requires the knowledge of how shear is transferred in the onstrution joint. In this regard, a brief summary of publiations overing this topi as well as available reommendations that were proposed to prevent sliding shear are presented in the following paragraphs. In their efforts to explain how shear is transferred along an existing or potential rak in onnetions (Constrution joint in our ase), Birkeland and Birkeland (1966) and Mast (1968) proposed that shear ould be transfer by what they termed shear frition between the rough faes of the raks. It is assumed in this theory that as the uneven rak faes slide past one another, the projetions on the rak faes ride over one another and fore the rak faes apart, strething any reinforement rossing the rak suffiiently to ause it to yield. The tensile fore so developed in the reinforement is assumed to ompress the rak faes together, whih results in fritional resistane to sliding along the rak. Mattok et al. (1975)

36 22 Chapter 2: Literature Review experimentally demonstrated the validity of shear frition hypothesis; adding that if a ompressive stress is applied to wall setion, it should be added to shear resistane omponent by adding its value to the previously mentioned normal stresses originated by bars rossing the sliding plane. Another investigation by Mattok (1974) showed that when the shear-frition reinforement is inlined with respet to the shear plane suh that the omponent of the shear fore parallel to the reinforement tends to produe tension in the reinforement, part of the shear is resisted by the omponent parallel to the shear plane of the tension fore in the reinforement. Hene; based on the shear frition theory that has proposed by Birkeland and Birkeland (1966) and Mast (1968) and later developed by Mattok (1974 and 1975), the sliding shear resistane an be alulated as the summation of two primary omponents. The first is the frition aused by all reinforement rossing the potential sliding shear plane in addition to any normal fore ating aross it [μ(avf fy +μn)]; where μ is frition oeffiient and depends on the surfae roughness; Avf is the area for all reinforement rossing sliding plane; other terms were previously desribed) (Figure 2.9a). The seond is the omponent parallel to the shear plane of the tension fore in the inlined reinforement (Ain fy os α) (Figure 2.9b). Reinforement N V V A in f y A in f y sin α V Tension in reinforement= A vf f y os α (a) Frition stress= μ(a vf f y +N) Compression on onrete surfae =A vf f y +N A in f y os A in f y α (b) Applied shear Assumed rak or shear plane Figure 2.9 Sliding shear resistane omponents Based on the shear transfer mehanism; Paulay (1972) pointed that if sliding shear failure is desired to be prevented in strutural walls, the following aspets should be taken into onsideration:

37 2 1. The onstrution joint surfae should be leaned and intentionally roughened to use artifiially high values of the oeffiient of frition in the shear-frition equations; 2. Any ompression fore ating aross the joint an be utilized in alulation sliding shear resistane;. If the flexural reinforement that passed through onstrution joint plus the normal fore is not suffiient to resist sliding shear, additional sliding shear reinforement an be added vertially or diagonally; however, diagonal reinforement is preferable; and; 4. To preserve the funtionality of sliding shear reinforement, shear-frition reinforement shall be appropriately plaed along the shear plane and shall be anhored to develop fy on both sides by embedment, hooks, or welding to speial devies Axial Load Little experimental works have been onduted to assess the effet of axial load on the behavior of reinfored onrete squat walls (Lefas et al. 1990; Salonikios et al. 1999; Li and Xiang 2014). Three levels of onstant axial loads were investigated in the testing program by Lefas et al. (1990); 0.0, 0.1, and 0.2 of the axial apaity of the walls; while two levels of axial loading were tested by Salonikios et al. (1999) and Li and Xiang (2014); (0.0, and 0.07), (0.0, and 0.05) of the axial apaity, respetively. It was reported that axial load has a signifiant effet on enhaning the shear strength and stiffness of the squat walls. This is due to the fat that axial loads signifiantly ontrol shear raks width and therefore the ability of struture to transfer shear by the aggregate interlok would be substantial. In addition, axial load was shown to signifiantly enhane the wall resistane against sliding shear. In spite of the agreement between the mentioned studies about the benefiial effet of axial loads on squat walls behavior, other researhers (Park and Paulay 1975; Paulay et. 1982) reported that ommon squat walls generally arry small axial loads, and therefore this effet was suggested to be ignored. The ontradition an be also found between the available analytial models and guidelines methods for prediting the shear strength of squat walls. For instane, the effet of normal fore was inorporated in the softened truss model whih was proposed by Hsu and Mo (1985) and the softened strut and tie model whih proposed by

38 24 Chapter 2: Literature Review Hwang et al. (2001) and simplified later by Kassem (2015). This effet is also involved in the empirial equations suggested by Gule and Whittaker (2009). Nevertheless; the ACI 18 (2014) and CSA A2. (2004) methods for shear strength estimation of squat walls still ignore this effet Ultimate Shear Strength Predition To predit the nominal shear strength of squat walls, two proedures are usually followed. The first depends on derivation of empirial equations based on the experimental investigations and test results. In the seond proedure, the researhers assume a shear model based on the struture mehanis and ondut a formula for predition after use equilibrium, ompatibility and material onstitutive relationships. The urrent ode provisions use empirial or semiempirial equations to alulate the nominal shear strength of squat walls. In the following subsetions, a brief summary of the available equations that predit nominal shear strength of squat walls in the ACI and CSA A2.-14 odes are presented. ACI 18 (2014) The urrent ACI 18 (2014) shear strength expression is based on an assumed shear rak at a 45 angle aross an effetive wall length (lw), and is omprised of two superimposed resisting mehanisms: the shear reinforement strength and the onrete shear strength. The ontribution of shear reinforement is estimated by onsidering equilibrium of fores at a typial joint of the 45 о truss model while onrete ontribution has been empirially obtained from experimental results. The ACI 18 (2014) expression for in-plane shear resistane (V) of steel RC squat walls is given as follows (SI units): V A ' ' v( f h f yh) 0.8Av f (2.4) where α is the aspet ratio oeffiient equal to 0.25 for hw/lw 1.5, 0.17 for hw/lw 2.0, and varies linearly between 0.25 and 0.17 for aspet ratios between 1.5 and 2.0, λ is a oeffiient depending on the onrete type; equal to 1.0 for normal weight onrete and 0.75 for lightweight onrete, f is the onrete ompressive strength, ρh is the horizontal web reinforement ratio, fyh is the horizontal reinforement yield stress, and Av is the gross area of the web of the wall (equal to wall length web thikness).

39 25 CSA A.2 (2014) The CSA A2. (2014) provides a shear design method based on the modified ompression field theory (MCFT) (Vehio and Collins 1986). In this theory, the shear resistane of a onrete member an be expressed as the sum of onrete ontribution (V), whih depends on the tensile stresses in onrete, and the shear reinforement ontribution (Vs). However, due to the signifiant degradation of onrete shear resistane aused by rissross shear raks; the CSA A2. (2014) ignores the onrete ontribution in shear strength. Following this onept, the shear strength of squat walls an be predited based on the shear apaity of shear reinforement only whih an be alulated as follows (SI Units): V s safd v y v ot (2.5) s where, s is the material resistane fator (equal 1.0 in ase of omparison with experimental), Av is the area of transverse web reinforement perpendiular to the axis of member with in the distane s, fy is the speified yield strength of web reinforement, dv is the effetive shear depth equal to the greater of 0.9d or 0.72 lw but should not be less than 0.8 lw (Clause ), θ is the angle of inlination of diagonal ompressive stresses to the longitudinal axis of the wall. Unlike the ACI 18 (2014), a rotating rak provision is used in whih the angle of the prinipal ompression strut varies depending on the longitudinal strain ondition and an be alulated based on the following equations with maximum value of 50 о and minimum value of 0 о : x (2.6) M / d V f v f x (2.7) 2Es As 2.. FRP Composite Materials The deterioration of reinfored onrete strutures due to steel orrosion has beome a serious problem in the last deades. In North Ameria, this phenomenon has been aelerated in many

40 26 Chapter 2: Literature Review strutures suh as parking garages and bridges. This is typially due to temperature flutuating and the inreasing use of deiing salts. The added ost of repairing deteriorated strutures with replaement osts ommonly more than twie the original ost of onstrution, led to the adoption of striter speifiations in some building odes and more stringent limits of hloride ions in onstrution materials. It also stimulated the reent major researh efforts in design and onstrution tehniques to improve the durability of reinfored onrete. The use of fiber reinfored polymer (FRP) reinforing bars as alternative for steel reinforing bars has emerged as one of the many tehniques to enhane the orrosion resistane of reinfored onrete strutures. In partiular, FRP reinforing bars offer great potential for use in reinfored onrete onstrution under onditions in whih onventional steel reinfored onrete has yielded unaeptable servie. If orretly applied in the infrastruture area, FRP an result in signifiant benefits related to both overall ost and durability. Other advantages inlude high-strength and stiffness to-weight ratios, resistane to orrosion and hemial attak, ontrollable thermal expansion and damping harateristis as well as eletromagneti neutrality. In addition, fatigue strength and fatigue damage tolerane for many FRPs omposite showed satisfatory results. Given their lower ost ompared to other types of FRP bars, glass-frp (GFRP) bars have been used extensively in different appliations suh as bridges, parking garages, water tanks, tunnels and marine strutures (Erki and Rizkalla 199, El-Salakawy et al. 2005; Kassem et al. 2011, Tobbi et al. 2014). A brief summary about FRP material onstituents, manufaturing, types, and properties is presented in the following subsetions FRP Constituents FRP produts are omposite materials onsist of reinforing fibers impregnated with a matrix (resin). The fibers are responsible for providing the mehanial strength and stiffness to the omposite, while the resins are responsible for transferring stresses between the fibers, proteting the fibers from mehanial abrasion, and prevent the fibers from bukling. In order to provide the reinforing funtion, the fiber-volume fration should be more than 55% for FRP bars and rods (ISIS 2007).

41 27 Fibers The most ommonly used material for FRP reinforement produts are aramid, arbon, glass, and reently basalt fibers. Given their lower ost ompared to other types of FRP types, glass- FRP (GFRP) are more attrative to the onstrution industry. Glass fibers present many advantages suh as high strength-to-weight ratio, low ost, eletromagneti neutrality, and hemial resistane. Notwithstanding, the disadvantages are relatively low tensile modulus, sensitive to abrasion, and relatively low fatigue resistane. Glass fibers are lassified as fiber drawn from an inorgani produt of fusion that has ooled without rystallizing. The types of glass fibers ommonly used are E-glass, S-glass and C-glass. E-glass has the lowest ost among all ommerially available reinforing fibers, whih is the reason for its widespread use in the FRP industry (ISIS 2007). Matrix (Resin) The final mehanial properties of the FRP produt are signifiantly affeted by the seletion of the proper matrix (resin). The matrix should be able to develop a higher ultimate strain than the fibers to exploit the full strength of the fibers (Phillips. 1989). There are two types of polymeri matries widely used for FRP omposites; namely, thermosetting and thermoplasti. However, thermosetting polymers are used more often than thermoplasti in FRP industry. They are low moleular-weight liquids with very low visosity, and their moleules are joined together by hemial ross-links. Hene, they form a rigid threedimensional struture that one set, annot be reshaped by applying heat or pressure (ISIS 2007) Manufaturing Proess There are three ommon manufaturing proesses for FRP materials; pultrusion, braiding and filament winding. Straight FRP bars are produed using the pultrusion tehnique. In this method, the ontinuous strands of the fibers are pulled from a reel of fibers to be impregnated in a resin tank. One they are saturated with resin, they are shaped through a heated die at whih they an be ured as shown in Figure Before the FRP bars are ut to the required

42 28 Chapter 2: Literature Review lengths, the bars surfae must be treated in the form of spirals or with sand oating to reate rough surfae that reates a strong bond with onrete (ISIS 2007). Figure 2.10 Pultrusion proess 2... Mehanial Properties Tensile Strength When loaded in tension, FRP bars do not exhibit any plasti behavior (yielding) before rupture. The tensile behavior of FRP bars onsisting of one type of fiber material is haraterized by a linearly elasti stress-strain relationship until failure with a general lak of dutility, very high tensile strength and relatively low modulus of elastiity (Figure 2.11). Various fators affet the tensile strength of FRP bars. The most signifiant fators are fiber type and fiber-volume fration that is defined as the ratio of the volume of fiber to the overall volume of the bar over the unit length. Bar manufaturing proess, quality ontrol and the rate of thermoset resin uring also affet tensile strength [ACI 440.1R (2015)].

43 Carbon Glass Aramid Steel Figure 2.11 Stress strain for steel and FRP bars Unlike steel, the unit tensile strength of an FRP bar an vary with diameter. Faza and Gangarao (199) reported that GFRP bars from three different manufaturers exhibited tensile strength redutions of up to 40% as the diameter inreases proportionally from 9.5 to 22.2 mm. However, a 7% strength redution in pultruded AFRP bars has been observed when the bar diameter inreased from to 8mm. Due to this disrepany in results, the ACI 440.1R (2015) design guidelines reommend that the bar manufaturers should be requested to provide the strength values of different bar sizes Compression Strength Limited researh has gone into investigating the behavior of FRP bars under ompression load. Tests on FRP bars with a length-diameter ratio from 1:1 to 2:1 have shown that the ompressive strength is lower than the tensile strength (Wu 1990). The ompressive modulus of elastiity of FRP bars depends on length-to-diameter ratio; bar size and type; and other fators, suh as boundary onditions. In the reported results from ompression tests, it is generally agreed that ompressive stiffness ranges from 77 to 100% of the tensile stiffness (Bedard. 1992, Chaallal and Benmokrane 199, Tavassoli et al. 2015), while the ompressive strength is around 50% of the tensile strength. Experimental test results (Alsayed et al. 1999; De Lua et al. 2009; ; Issa et al. 2011, Deiveegan and Kumaran 2009) on the behavior of onrete olumns entirely reinfored with glass fiber RC polymer (GFRP) reinforement have also demonstrated the feasibility of suh strutural element. Nevertheless, the urrent ACI 440.1R (2015) design guidelines still do not reommend using FRP bars as longitudinal

44 0 Chapter 2: Literature Review reinforement in ompression members, while the CSA S806 (2012) states that the ompressive ontribution of FRP longitudinal reinforement is negligible Flexural Strength The behavior of FRP RC setions is different ompared to setions reinfored with traditional steel reinforement. This is due to the different mehanial behavior between the two types of reinforements. FRP bars exhibit linear stress-strain behavior up to failure without any yielding. Therefore, tension failure in FRP RC setion is sudden and atastrophi; hene, it should be avoided (Jaeger et al. 1997, Theriault and Benmokrane 1998). On the other hand, ompression failure of FRP RC setions offers more favorable response, as onrete dutility is utilized in giving ample warnings before failure (Nanni 199). Most of urrent odes and guidelines require FRP RC setion to be design for ompression failure. Aording to ACI 440.1R (2015), a large amount of FRP reinforement to be provided in the tension zone of flexural members in order to obtain ompression failure mode whih is the most dutile failure mode, as well as, for ontrolling rak width and defletion. The CSA S6 (2014) and CSA S806 (2012), on the other hand, reommends that the moment of resistane of flexure member ross setion reinfored with FRP should be at least 50% greater than the raking moment, to avoid brittle failure Shear Strength As identified by Joint ACI-ASCE Committee 445 (ACI-ASCE 1998), raked reinfored onrete members resist the applied shear stresses by the following five mehanisms: (1) shear stresses in unraked onrete; (2) aggregate interlok; () dowel ation of the longitudinal reinforing bars; (4) arh ation; and (5) residual tensile stresses transmitted diretly aross the raks. Aggregate interlok results from the resistane to relative slip between the two rough interloking surfaes of the rak, muh like fritional resistane. As long as the rak is not too wide, this ation an be signifiant (Razaqpur et al. 2001). Dowel fores generated by longitudinal bars rossing the rak partially resist shearing displaements along the rak. Arhing ation ours in deep members or in members having shear span-to-depth ratio less than 2.5 (Razaqpur et al. 2004). The residual tension in raked onrete is reported to be present for raks less than 0.15 mm in width (ACI-ASCE Committee ).

45 1 Due to the relatively low modulus of elastiity of FRP bars, onrete members reinfored with FRP will develop wider and deeper raks than members reinfored with steel. Deeper raks derease the ontribution to shear strength from the unraked onrete due to the lower depth of onrete in ompression. Wider raks, in turn, derease the ontributions from aggregate interlok and residual tensile stresses. Additionally, due to the relatively small transverse strength of FRP bars and the relatively wider raks, the ontribution of dowel ation may be negligible. Eventually, the overall shear apaity of onrete members reinfored with FRP bars as flexural reinforement is lower than that of onrete members reinfored with steel bars Fatigue Strength Fatigue refers to the degradation or failure of a strutural material or element after repeated yles of loading and unloading. Fators suh as the material properties of the onrete and reinforement, reinforement ratio, transverse reinforement, minimum and maximum values of repeated loading, range and rate of loading, all play a major role in estimate fatigue life and fatigue strength of reinfored onrete elements. The environmental fators suh as temperature and humidity also affet the fatigue life and strength. Aording to many researhers, FRP bars exhibit good fatigue resistane (Uomoto and Ohga 1996, Demers 1998, Adimi et al. 2000, El-Ragaby et al. 2007). It was doumented that arbon-epoxy omposites have better fatigue strength than steel, while the fatigue strength of glass omposites is lower than steel at low stress ratio. Even though GFRP is weaker than steel in fatigue, tests on speimens with unbonded GFRP dowel bars have shown fatigue behavior similar to that of steel dowel bars for yli transverse shear loading of up to 10 million yles. The test results and the stiffness alulations have shown that an equivalent performane an be ahieved between FRP and steel bars subjeted to transverse shear by hanging some of the parameters, suh as diameter, spaing, or both (Porter et al. 199) Bent Portion Strength Currently-available FRP reinforing bars are fabriated using thermosetting resin matries and onsequently annot be bent on site. Bends and hooks, when required, must be produed by

46 2 Chapter 2: Literature Review the bars manufaturer during the fabriation proess. It is possible to obtain bends and hooks in virtually any geometry from urrent FRP rebar manufaturers (Figure 2.12), although minimum bend radius is typially larger than for steel bars due to signifiant weakening of FRP bars around tight orners. Typial minimum allowable bend radius for FRP bars are.5 to 4 times the bars diameter and these bends are aompanied by up to 50 perent redution in the tensile strength of the bar at the bend (ISIS 2007). Figure 2.12 Bends in GFRP bars reinforement Many studies have reported a signifiant redution in the tensile strength of bent FRP bars and stirrups at the loation of the bend. For instane, Shehata et al. (2000) reported that the bend strength generally dereases with dereasing bend radius and an be as low as 5% of the tensile apaity of straight portions of the bar. Other investigation onduted by Morphy et al. (1997) on 16 speimens with different types of CFRP stirrups, it was reommended limiting the strength of CFRP stirrups to 50% of the ultimate straight bar apaity for design. The CSA S806 (2012) reommends that bend portion strength shall be taken equal to 40% of the straight portion tensile strength while the ACI 440.1R (2015) gives the following equation for bend portion strength estimation (SI units): f FRPbend ( 0.05 r / d 0.) f f (2.8) b b Fu fu where ffrpbend is the bend portion strength, rb is the bend portion s radius, db is the bar diameter, ffu is the ultimate tensile strength for straight portion.

47 2.4. FRP RC Strutural Systems for Seismi Fores Strutural Frame Systems Moment-resisting frames are retilinear assemblages of beams and olumns that are well detailed in a standard way to be able to resist gravity and seismi loads. The philosophy behind the seismi design of frame systems is to provide them with suffiient dutility by whih they an dissipate the ating seismi energy (Mady et al. 2011). However, due to the elasti nature of FRP up to failure, onerns have been raised about its appliability as internal reinforement in earthquake resisting frames that require the inelasti behavior (dutility) of reinforement. To address this issue, many investigations have been onduted to validate the feasibility of FRP RC frame in earthquake regions. A brief summary for the implemented works, ordered hronologially, is given as following. The first investigation was onduted by Fukuyama and Masuda (1995) on a half-sale threestory onrete frame totally reinfored with braided aramid fiber polymer reinforing bars. The frame had 1800 mm story height, and 500 mm span (between olumn enters). The seleted reinforement ratios for the beams were 0.64% and 0.48% at the bottom and top; respetively. Meanwhile, the main reinforement ratio of the olumns was 1.47%. The speimen was tested under reversed yli loading applied at the mid-height of the third floor up to failure. It was reported that the frame remained elasti up to a drift angle of 1/50 rad, and no substantial derease in strength took plae after rupture of some main beam bars due to the high degree of indeterminay of the frame. Due to the elasti behavior of FRP bars in tension, the unloading branh of hysteresis loops was observed to aim towards the origin with negligible residual deformations. This indiated that rehabilitation of FRP RC frame would be muh easier than the ase if it was reinfored with steel. Said and Nehdi (2004) tested two full-sale beam-olumn joint speimens; one reinfored with steel while the other with GFRP grids in order to investigate their performane under the event of an earthquake. Beams and olumns ross setion were similar and measured mm. The GFRP RC speimen was designed to have a similar flexural apaity to that of the ontrol steel RC speimen, thus induing a omparable level of joint shear input. The GFRP RC

48 4 Chapter 2: Literature Review speimen showed a satisfatory drift apaity, assuming a minimum drift requirement of % as reommended in the literature for dutile frame buildings. However, the joint showed very low plastiity features resulting in lower energy dissipation ompared to that of the onventional steel RC beam-olumn joint. Sharbatdar and Saatioglu (2009) onduted an experimental program inluding testing two types of reinfored onrete elements; square olumns and retangular beams. The elements represent portions of olumn and beam elements between rigidly attahed adjoining members and the points of ontra-flexure, as antilever speimens. Four speimens of eah element were onstruted and reinfored with arbon FRP bars and arbon FRP grids as longitudinal and transverse reinforement, respetively. The studied parameters were the shear span length and the spaing of transverse reinforement. The olumns were tested under onstant axial load and lateral yli load while the beams were tested under yli load only. The test results indiated that FRP RC onrete beams and olumns an attain a lateral drift ratio up to %, while essentially remaining elasti in spite of the softening indued by raking. It was onluded that using FRP as flexural and shear reinforement in onrete frames is feasible. Given the elasti nature of FRP bars; however, it was suggested that FRP RC frames should be designed based on elasti member behavior (i.e. the seismi redution fator is equal to the unity) while taking advantage of relative flexibility of FRP material and the assoiated elongations in the fundamental period of strutures with potentially redued seismi demand (Figure 2.1). Figure 2.1 Effet of the lengthening of period on design fore levels (Sharbatdar and Saatioglu 2009)

49 5 Hasaballa (2009) tested four full-sale exterior T-shaped beam-olumn joints. The first speimen was reinfored with longitudinal and transverse steel bars and served as a referene speimen. The seond was similar to the first; however, GFRP bars were used in the longitudinal diretion. The other two speimens were reinfored with GFRP bars and stirrups with different reinforement ratio. Eah speimen was simulating a beam-olumn onnetion of an exterior bay in a multi-bay, multistory reinfored onrete moment-resisting plane frame. It was onluded that although steel RC speimen was able to dissipate energy in the order of 2 to times that of the GFRP speimens, the residual strains in the GFRP flexural reinforement at the 4.0% drift ratio were muh lower than in steel RC speimen. This indiated that the joint will regain its original shape after removing the loads, thus requiring minimum amount of repair. Mady et al. (2011) ontinued the investigation that started by Hasaballa (2009). A total of five full-sale beam-olumn joint prototypes were onstruted and tested under yli loading up to failure. The first test speimen was reinfored with onventional steel bars and stirrups and used as a ontrol speimen. The seond speimen was reinfored with GFRP bars and steel stirrups. The remaining three speimens were totally reinfored with GFRP bars and stirrups with different reinforement ratios. The findings revealed that GFRP RC joints an be designed to satisfy both strength and deformability requirements. The tested GFRP RC onrete beam-olumn joints safely ahieved 4.0% drift apaity with insignifiant damage. The obtained drift apaities were more than the 2.5% required by the NBCC (NBCC 2005) and the.5% required by the ACI 74.1 (2005). The results also demonstrated that inreasing the beam reinforement ratio an enhane the ability of the joint to dissipate the seismi energy through utilizing the inelasti behavior of onrete as long as the joint is safe under the applied shear stresses. Tavassoli et al. (2015) arried out an experimental program inluding nine irular olumns. The main investigated variables were the axial load level, the type of GFRP bars, and size and spaing of GFRP spirals. The speimens were tested under onstant axial load and lateral yli displaement exursions. The axial load was either 0.28Po or 0.42Po, where Po is the nominal axial load apaity of the olumn. Two types of GFRP bars were used. The diameter

50 6 Chapter 2: Literature Review of spiral was either 12 mm or 16 mm and spaed at a distane ranged from 50 mm to 275 mm. Experimental results in the form of moment-versus-urvature and shear-versus-tip defletion hystereti responses and various dutility parameters were presented and ompared with results of similar steel RC irular olumns whih onduted by Sheikh and Khoury (199). The results showed that onrete olumns reinfored with GFRP bars and spirals an behave in a manner that has stable post-peak response and ahieve high levels of deformability. The results also indiated that due to the large stiffness of GFRP bars than steel beyond yielding, GFRP RC olumns performed in a more stable manner than the ompanion steel RC speimens. Ali and El-Salakawy (2016) tested eight full-sale retangular olumn prototypes under ombined lateral yli quasi-stati and onstant axial loading. The test speimens represented the lower portion of first-story olumns between the footing and the ontra-flexure point. The test parameters inluded longitudinal reinforement type and ratio, level of axial load, and stirrup spaing. Test results showed that the drift apaity of GFRP reinfored onrete (RC) retangular olumns at failure ranged between 8.5 and 12.5%, whih highly exeeds the limitations of North Amerian building odes. This indiates that the deformability of GFRP- RC olumn prototypes suessfully replaed the dutility in steel RC olumns in dissipating the seismi energy in the presene of onstant axial load. Furthermore, there was insignifiant strength degradation before failure due to the well-onfined onrete ore. Naqvi and El-Salakawy (2016) studied the effet of using lap splies in GFRP-RC retangular olumns subjeted to yli-reversed loads. The experimental program omprised six speimens; five reinfored with GFRP bars while the last speimen reinfored with steel and served as a referene. The test parameters inluded lap splie length of longitudinal reinforement and transverse reinforement spaing. Test results indiated that a splie length of 60 times the diameter of the longitudinal olumn bar was adequate in transferring the full bond fores along the splie length. In addition, lap-splied GFRP-RC olumns with losely spaed transverse reinforement showed stable hysteresis response and ahieved high levels of deformability, whih far exeeded the limitations of the North Amerian building odes.

51 Shear Walls Shear walls are broadly used as the first line of defense against earthquake exitations and they have many advantages over moment-resisting frames. Compared to the studies onduted in FRP RC resisting frames; however, very little researh has been done in investigating the feasibility of shear walls reinfored with GFRP in regions prone to earthquake exitations. The first attempt was onduted by Yamakawa and Fujisaki (1995) on seven wall speimens reinfored with CFRP grids. The speimens were tested under reversed yli lateral loading while simultaneously subjeted to onstant axial load. All speimens exhibited early degradation in lateral load apaity when 1% drift was ahieved assoiated with low energy dissipation. The authors attributed the poor performane to three main reasons: (1) the CFRP grids were not able to arry ompressive stress and therefore experiened frature under low ompressive stresses, (2) there was a need to design adequate development length to prevent the reinforing bars from pulling out of the wall base, and () the CFRP grid reinforement did not provide onrete onfinement. Mohamed (2014a) started another study to investigate the yli behavior of shear walls internally reinfored with GFRP bars. From the learned lessons in Yamakawa and Fujisaki study, the authors strove to avoid all brittle failures that might our in shear walls and that would prevent them from reahing their apaities. Four large-sale shear walls one reinfored with steel bars (as referene speimen) and three totally reinfored with GFRP bars were onstruted and tested to failure under quasistati reversed yli lateral loading. The main parameter is the wall s aspet ratio (height/length). The three GFRP RC walls, G10, G12, and G15 had aspet ratio of.5, 2.9 and 2., respetively. The steel speimen ST15 served as a referene for G15, hene the it had the same onrete dimensions and axial stiffness. The reported test results learly showed that properly designed and detailed GFRP RC walls ould reah their flexural apaities with no strength degradation, and that shear, sliding shear, and anhorage failures were not major problems and ould be effetively ontrolled. It was also reported that GFRP RC walls showed a reoverable and self-entering behavior up to allowable drift limits before moderate damage ourred and ahieved a maximum drift level meeting the limitation of most building odes.

52 8 Chapter 2: Literature Review Interation of flexural and shear deformations of the tested shear walls was also investigated (Mohamed et al. 2014b). The ontribution of flexural and shear deformation to the total deformation of the walls showed that, at early stage of loading, flexural deformations dominated the response. At higher levels of lateral drift; however, the shear deformations beome relatively pronouned, although the fatored shear strength is 0% higher than the ultimate flexural apaity of the shear walls. The results also showed that using GFRP better ontrolled shear distortion ompared to using steel. The steel RC wall, espeially after yielding of the vertial bars, experiened nearly twie as muh distortion as the ompanion speimen reinfored with GFRP. The authors explained that due to the elasti nature of GFRP bars, shear strain was uniformly distributed along the walls height, resulting in less shear deformations than those experiened in steel RC shear wall in whih yielding of steel bars intensified the shear strains at the yielding loation as shown in Figure Wall height (m) G15 G12 G10 ST Shear strain (%) Figure 2.14 Measured shear strain at ultimate load (Mohamed et al. 2014b) The non-orrodible nature of FRP bars and its effetiveness in ontrolling shear distortion enourages for new investigation to study the feasibility of using suh material in squat walls in whih these problems are frequently enountered. To the author knowledge; however, information on the seismi behavior of FRP RC squat walls is nonexistent. Therefore, this researh intended to fill the gap of knowledge by providing detailed experimental test results for this behavior.

53

54 CHAPTER EXPERIMENTAL PROGRAM.1. Introdution The details of the experimental program that inluded nine large-sale squat walls are presented in this hapter. The design, onstrution and testing of the speimens at the Strutural Engineering Laboratory at the University of Sherbrooke are disussed in details..2. Test Matrix Nine large-sale squat walls with an aspet ratio (height to length ratio) of 1. were onstruted and tested under quasi-stati reversed yli lateral loading up to failure. One speimen was reinfored with onventional steel bars and served as a ontrol speimen; the remaining eight were entirely reinfored with GFRP bars. The preliminary design and details have been onduted aording to aording to CSA A2. (2014) and CSA S806 (2012) for steel RC wall and GFRP RC walls, respetively. It should be noted that sine there is no seismi provisions in the CSA S806 (2012), similar methodologies that are being using in the ompanion ode was followed. The speimens measured 1500 mm in length, 2000 mm in height, and 200 mm in thikness. The walls thikness satisfied the CSA A2. (2014) Clause requirement for the minimum thikness. Eah speimen was ast with an integral mm heavily reinfored foundation; funtioning as anhorage for the vertial reinforement and to enable the speimen to be fixed to the laboratory floor. Figure.1 shows the onrete dimensions of the test speimens. Two boundary elements with equal width and breadth ( mm) were plaed at eah end of the horizontal length. The longitudinal and transverse reinforement ratios at the boundary elements were kept onstant in all speimens; 1.4% and 0.89%, respetively. The longitudinal reinforement involved 8 No. 10 (#) GFRP or steel bars and laterally tied against premature bukling by using transverse reinforement No. 10 spiral (#) GFRP or steel ties with spaing of 80 mm whih is approximately the maximum spaing permitted in CSA S806 9

55 40 Chapter : Experimental Program (2012). Two layers of web reinforement were used in all walls aording to CSA A2. (2014) Position of Dywidag bars fixing the foundation to the laboratory floor the laboratory floor All dimensions in mm Figure.1 Overall dimensions of test walls The speimens were arranged to investigate the influene of the following parameters on the behavior: 1. Reinforement type (GFRP, steel); 2. Horizontal web reinforement ratio (0.0%, 0.51%, 0.79%, 1.58%, and.56%);. Existene of vertial web reinforement (0.0%, 0.59%); 4. Use of bidiagonal web reinforement; and 5. Use of bidiagonal sliding reinforement. The first speimen G4-80 was designed to have almost equal flexural and shear apaities. The horizontal web reinforement onsisted of two layers of No. 1 (#4) GFRP bars spaed at 80 mm. The vertial web reinforement onsisted of two layers of No. 10 (#) GFRP bars spaed at 120 mm. Sliding shear was prevented by adding one layer of bidiagonal No. 10 (#) GFRP bars aross the potential sliding plane at an angle of 45, spaed at 100 mm and suffiiently anhored on eah side of the shear plane. The anhorage length for both the vertial reinforement and bidiagonal sliding reinforement was equal to the development

56 41 length speified in CSA S806 (2012) multiplied by 1.25 to aount for the yli effet as suggested by Mohamed et al. (2014a). The steel RC wall (S4-80) served as a ontrol speimen for G4-80, so both had idential reinforement onfigurations and ratios. It should be mentioned that this onfiguration ensured S4-80 would fail in flexure, sine its shear strength against either sliding or diagonal tension was muh higher than its flexural strength. Three speimens; G4-250, G4-160, and G6-80 were reinfored identially to G4-80; however, with different horizontal web reinforement ratios; 0.51%, 0.79%, and.58% using No. 1 (#4) GFRP bars with spaing of 250, 160 or No. 19 (#6) with spaing 80 mm, respetively. Two other speimens were onstruted with either vertial or horizontal web reinforement, G-V, and G-H, respetively, to test the absene of horizontal or vertial web reinforement on the behavior. Speimen G-V was reinfored with vertial web reinforement idential to that used in the presribed four GFRP RC speimens while speimen G-H was reinfored with horizontal web reinforement idential to speimen G Another speimen, GD, was designed to investigate the effet of using bidiagonal web reinforement instead of horizontal and vertial web reinforement. The speimen was reinfored with two layers of bidiagonal No. 10 GFRP bars spaed at 100 mm with angle of 45 relative to the longitudinal axis of the wall. The web reinforement details were seleted suh that GD had a flexural apaity similar to its ounterpart speimen G4-80. It should be mentioned that although GD had less web reinforement than G4-80, GD s predited shear strength was almost twie that of G4-80 due to the absene of bent portions, whih are responsible for the lower strength of horizontal web reinforement; more details is disussed in Ch. 4. In an attempt to investigate the sliding-shear resistane of GFRP, speimen G4 was built idential to G4-80 but without bidiagonal slidingshear reinforement. The base for all speimens was reinfored with 25M Grade 60 deformed steel bars spaed at 100 mm in eah diretion in both upper and lower levels. All speimens had a onrete over of 25 mm. The used reinforement ratios in eah speimen are listed in Table.1. Figure.2 shows reinforement details of the tested speimens.

57 42 Chapter : Experimental Program Wall Table.1 Reinforement details f' (MPa) Reinforement Ratio ρl (%) ρt (%) ρv (%) ρh (%) ρd (%) ρs (%) S G G G G G-V G-H GD G f' = onrete ompressive strength; ρl = boundary longitudinal-bar reinforement ratio; ρt = boundary-tie reinforement ratio; ρv = web vertial-bar reinforement ratio; ρh = horizontal web reinforement ratio; ρd = bidiagonal web reinforement ratio; ρs = bidiagonal sliding-shear reinforement ratio. No.1 or No.19 steel or S Horizontal web reinf. A Reinforement details for S4-80, G4-250, G4-160, G4-80, and G No.1 steel or GFRP Vl. boundary No.10 steel or 120 Vertial web reinf. 200 Hz. No.1 or No. S Vl. 120 No No. 10 steel or Sliding reinf. A Θ = 45 Se A-A Horizontal reinforement details No. 1 = 80 mm for S4-80 No mm for G4-250 No mm for G4-160 No mm for G4-80 No mm for G No.10 Vl. boundary 200 Figure.2 Reinforement details

58 4 8 No.1 GFRP Vl. boundary 8 No.10 GFRP Vl. boundary No Vertial web reinf. No. 1 GFRP@ 250 Horizontal web reinf. GFRP 100 GFRP 100 A A Θ = 45 A A Θ = 45 Reinforement details for G-V Reinforement details for G-H 8 GFRP No GFRP No. 10 Boundary Element Boundary Element GFRP 120 Vertial Reinf. GFRP 100 GFRP 80 Θ = 45 Horizontal Reinf. A A A A Reinforement details for GD Reinforement details for G4 Vl No No.10GFRP 8No.10 GFRP 80 mm Vl. boundary Hz No.10 GFRP 80 mm 8 No.10 GFRP Vl. boundary Sliding reinf. Se A-A (G-V) Sliding reinf Se A-A (G-H) 200 Diagonal Web Reinf. No No.10 GFRP Vl. boundary Hz No.1 80 Vl No No.10 GFRP Vl. boundary Se A-A (GD) Se A-A (G4) 200 Figure.2 Reinforement details (ontinued)

59 44 Chapter : Experimental Program.. Material Properties The test speimens were ast using normal-weight, ready-mixed onrete with a target 28-day ompressive strength of 40 MPa. Table.1 gives the atual onrete ompressive strength (f') based on the average of at least three mm ylinders for eah onrete bath on the day of speimen testing. An average onrete tensile strength of.5 MPa was obtained from the split-ylinder tests. Speimen S4-80 ontained No. 10 (#) and No. 1 (#4) grade 420 deformed steel bars: No. 10 (#) for vertial and retilinear spiral reinforement; No. 1 (#4) for horizontal reinforement. The GFRP reinforing bars onsisted of three diameters of Grade III sand-oated bars (CSA S ): No. 10 (#) for vertial and retilinear spiral reinforement and No. 1 (#4) and No. 19 (#6) for horizontal reinforement. The longitudinal tensile properties of the GFRP bars were determined by testing five speimens aording to ASTM D7205 (2011), in the ase of the straight bars, and test method B.5 in ACI 440.R (2004), in the ase of the bent bars. The manufaturer provided the properties of the steel bars. Table.2 lists the material properties of the reinforing bars. Figure. shows the vertial, horizontal, and retilinear spiral GFRP reinforement. Bar Designate d Bar Diameter (mm) Table.2 Reinforement mehanial properties Nominal Area 1 (mm 2 ) Immersed Area (mm 2 ) Tensile Modulus of Elastiity 2 (GPa) Tensile Strength 2 * (MPa) Average Strain at Ultimate (%) Straight bars No. 10 GFRP No. 10 steel fy = 420 εy = 0.2 No. 1 steel fy= 420 εy = 0.2 Bent No. 10 GFRP retilinear spiral Straight Bent Bent No. 1 GFRP horizontal bar Straight Bent Bent No. 19 GFRP horizontal bar Straight Bent fy: steel yielding strength, εy: steel yielding strain. 1 Aording to CSA S807 (CSA, 2010) 2 Tensile properties were alulated using nominal ross-setional areas. *Guaranteed tensile strength: Average value standard deviation (ACI )

60 45 No. 1 GFRP No. 19 GFRP Horizontal bars Retilinear spiral (No. 10 GFRP) Vertial bar (No. 10 GFRP) Figure. GFRP reinforement.4. Speimens Constrution The speimens were onstruted at the Strutural Laboratory in the Department of Civil Engineering at the University of Sherbrooke. Efforts have been made to fabriate and ast the speimens in upright position to ensure realisti onstrution site onditions. The onstrution started with assembling of the steel base age in wooden formwork (Figure.4). Then the wall ages were instrumented and assembled to the base steel age (Figure.5) followed by preparation of wall formwork (Figure.6). Casting of the speimens was implemented in two stages. The base was ast first. The rest of the speimen was ast 2 days later, without taking any speifi measures for uring at the onstrution joint in order to represent the onstrution praties used for an atual reinfored onrete (RC) building site (Figure.8). One day later, the speimens were wrapped with wet burlap and plasti sheets and ured for 7 days then left to the date of testing. Figure.9 shows the speimens after onstrution and uring. Figure.4 Prepared formwork and age of the base

61 46 Chapter : Experimental Program S4-80 G4-80 GD Figure.5 Assembly of the wall age to the base age Figure.6 Assembly and alignment of wall formwork Figure.7 Casting the base

62 47 Figure.8 Casting the wall Figure.9 Speimens after uring.5. Preliminary Design of Speimens The predited flexural strength for the investigated walls was alulated based on plane setional analysis. The analysis was based on strain ompatibility, internal fore equilibrium, and the ontrolling mode of failure. The alulation was arried out onsidering the unonfined and onfined onrete setion. The ontribution of the GFRP longitudinal bars in ompression to the flexural strength was onsidered by assuming its strength in ompression is 50% of its tensile strength as suggested by Deitz et al. (200) with keeping the modulus of elastiity is onstant. The ontribution of the diagonal web reinforement or sliding-shear

63 48 Chapter : Experimental Program reinforement was obtained by onsidering their omponents in the longitudinal diretion of the wall. Regarding the shear apaity, due to the absene of seismi provisions for GFRP RC squat walls in CSA S806 (2012), we adopted the onept provided in CSA A2. (2014) for steel. Clause in CSA A2. (2014) speifies that, under seismi loading, the shear fore is to be resisted by horizontal web reinforement only with no reliane on the onrete ontribution to shear strength. This is due to the effet of riss ross shear raks and the subsequent degradation of onrete shear resistane. The shear strength of the steel RC wall was determined as follows: V r s Av f ydv ot Vss (.1) s where Vr is the shear strength, Vss is the shear resistane provided by the horizontal web reinforement, s is the material resistane fator for steel, Av is the area of horizontal web reinforement within the distane s, fy is the speified yield strength of the horizontal web reinforement, dv is the effetive shear depth equal to the greater of 0.9d or 0.72 lw but not less than 0.8 lw, θ is the angle of inlination of diagonal ompressive stresses to the longitudinal axis of the wall and equal to 45, and s is the spaing of the horizontal web reinforement measured along the longitudinal axis of the wall. The shear apaity of horizontal web reinforement in the GFRP RC walls was alulated as follows: V r 0.4 f Av f fud v ot M f dv V f Vsf, l, l (.2) s 2E A f f where, f is the material resistane fator for GFRP; εl is the longitudinal strain at mid-depth of the setion; Mf and Vf are the bending moment and shear fore at the ritial setion for shear, respetively; and θ ranges from 0 to 60. For speimen GD, whih had bidiagonal web reinforement, the shear resisted by the diagonal web reinforement was omputed from the general formula as follows:

64 49 V r f Av f fud v (ot ot) sin Vsf (.) s where α is the angle between the diagonal bars and the longitudinal axis of the member. It should be noted that CSA S806 (2012) urrently limits the stress level in the FRP transverse-reinforement stirrups to (1) avoid failure at the bent portion of the FRP stirrup, (2) ontrol shear-rak widths under servie load, and () maintain the maximum size of the diagonal raks at ultimate state so as to not seriously diminish shear transfer by aggregate interlok. Based on the available geometri requirement for the bent portion, the ultimate strength of the bent portion ranged between 0.4 and 0.45 fu. CSA S806 (2012) onservatively reommends the ultimate strength be equal 0.4fu. To limit the allowable rak-width size under servie load, CSA S806 (2012) limits the maximum strain in the stirrups to 5000 μstrain. This level of strain was also found to satisfy the third ondition of maintaining shear transfer through aggregate interlok (Razaqpur and Spadea 2015). Sine our study ignored the shear apaity arried by onrete (in terms of aggregate interlok and dowel ation) and due to the elasti nature of GFRP bars, whih should enable the raks to realign and lose after an earthquake, the seond and third riteria were omitted. Hene, the ultimate strength of the GFRP was limited to 0.4fu (Eq. 4.2). In speimen GD, whih ontained bidiagonal web reinforement, the bar stress was limited to fu due to the absene of bent portions (Eq. 4.). Constrution joints in squat walls may drastially degrade under yli loading and beome the weakest link in the hain, resulting in premature sliding-shear failure. Clause in CSA A2. (2014) gives the following equation for alulating the resistane to sliding-shear failure: os v f (.4) r s v y f where vr (MPa) is the sliding-shear strength; λ is a fator to aount for low-density onrete, is the resistane fator for onrete, is the ohesion stress, μ is the oeffiient of frition, fy is the yield strength for the steel, λ( + μσ) shall not exeed 0.25 f', f is the resistane fator for steel, and αf is the angle between the shear-frition reinforement and the shear plane.

65 50 Chapter : Experimental Program Aording to Clause , the values of σ are: N vf y sinf, A g A vf v (.5) Av where Avf is the area of the shear-frition reinforement, Av is area of onrete setion resisting shear transfer, and N is the un-fatored permanent ompressive load perpendiular to the shear plane. Sine the surfae of onrete base was not intentionally roughened, the sliding resistane was alulated based on a ohesion stress () of 0.25 MPa and a frition oeffiient (μ) of 0.60, as speified in Clause in CSA A2. (2014). It should be noted that the frition oeffiient of 0.6 was speified by the ode based on Mattok s observation (1977). Mattok found that, when onrete is ast against hardened onrete without roughening, sliding-shear resistane is primarily due to reinforement dowel ation. Test results learly indiated that the slidingshear strength was very lose to the shear yield strength of the shear-transfer reinforement (0.58 ρ fy). Due to the lak of provisions in CSA S806 (2012) for alulating sliding shear, we opted to alulate it for GFRP RC walls based on CSA A2. (2014), as given in Equations.4 and.5. The ontribution of all vertial GFRP reinforement rossing the onstrution joint was, however, ignored in terms of dowel shear resistane due to the lower strength and stiffness of GFRP bars in the transverse diretion [ACI 440.1R (2015)]. Hene, the GFRP RC speimens were found to be suseptible to sliding-shear failure. Therefore, one layer of bidiagonal sliding-shear reinforement was used at the potential sliding plane in all speimens exept G4, in whih the effet of bidiagonal reinforement was investigated..6. Test-Setup Figure.10 provides the layout of the test setup. All speimens were tested laterally as a vertial antilever with a fore applied through a rigid steel loading beam, designed to transfer lateral loads aross the top of the wall. The test speimens were tested without axial load. This is beause the axial ompressive stress in squat walls indued by gravity load generally represents an insignifiant perentage of the produt of onrete ompressive strength and

66 51 gross area. Additionally, adding axial loading will inrease the sliding resistane of GFRPreinfored squat walls whih was questionable at the preliminarily design. Hene, the author opted to study sliding resistane provided exlusively by GFRP bars while onservatively omitting the effet by axial load. The test setup of the squat walls is summarized in the following steps: 1. The base was stritly aligned and fixed to the laboratory floor by four pre-stressing 66 mm diameter Dywidag bars (high strength steel bars) to prevent uplifting and/or horizontal sliding during the appliation of lateral loading 2. Lateral load was applied to the wall speimen with an MTS hydrauli atuator with a maximum apaity of 1000 kn and a maximum stroke of ±250 mm. The atuator was jointly onneted to the transfer steel beam and transmitted lateral load to the speimen through a 50 mm thik steel bearing plate. Lateral loads were applied at 2550 mm above the base of the wall.. Out-of-plane braing was provided at the level of transfer steel beam to prevent twisting of the wall speimen during testing. The out of plane supporting onsisted of two steel beams attahed to the reation wall with two bearing rollers at the end attahed to the transfer steel beam.

67 52 Chapter : Experimental Program Atuator Out of plane braing Reation wall Transfer steel beam 2550 Dywidag bars Base Squat-wall speimen Rigid floor (a) Shemati drawing Lateral braing Atuator Steel beam Squat-wall speimen Base Dywidag bars (b) Shemati drawing Figure.10 Test-setup

68 5.7. Loading Proedure The loading proedure for all speimens followed the reommendations of the Amerian Conrete Institute (ACI) Committee 74 Report on the aeptane riteria for testing reinfored onrete strutural elements under Slowly Applied Simulated Seismi Loads [ACI 74.2R (201)]. The walls were yled twie at eah displaement level with inrements of 0.1% up to 0.5% lateral drift, followed by inrements of 0.25% up to 2.5%, and then inrements of 0.5% to failure. Figure.11 gives a typial sequene of displaement yles. Lateral drift (%) Cyle number Figure.11 Loading history of testing program Lateral displaement (mm).8. Instrumentations For eah speimen, a total of 22 eletrial resistane strain gauges are attahed to the reinforing bars (boundary and web reinforement) at ritial loations to measure strains as shown in Figure.12. Two longitudinal bars of the main boundary reinforement are instrumented at the extreme ompression and tension fibers at two height levels of the walls; 200 mm from the base, and at height h = lw/2. Three vertial web bars were instrumented at three height levels; 200 mm from the base, lw/2, and lw. This pattern gives the distribution of the vertial strains both along the horizontal length of the walls and along the instrumented

69 54 Chapter : Experimental Program bars. Similarly, horizontal web reinforement was instrumented with strain gauges at the same three height levels of walls. This pattern of gauging gives the distribution of horizontal strains at three different levels and along the instrumented bars. Further four strain gauges were attahed to sliding reinforement at sliding plane. For speimen GD, the diagonal web reinforement was instrumented at three levels; 200 mm from the base, lw/2, and lw, in a pattern similar to that in the ompanion speimen with horizontal and vertial web reinforement. 200 l w / l w / 2 l w l w Figure.12 Stain gauges instrumentation Deformation response was monitored by a series of linear variable displaement transduers (LVDTs) as shown in Figure.1. Two LVDTs were installed to measure the lateral displaement at two levels; one at the wall tip, while the other at a height equal to the wall length (lw). One LVDT was mounted between the wall and base to measure sliding displaement. One LVDT was used to measure the sliding between the base and laboratory rigid floor, if any. Two LVDTs are mounted lose to the boundary elements to measure the onrete strain and wall urvature. Two LVDTs were installed at the wall ends to measure the axial deformations of the boundaries, by whih the rotation between wall and base an be alulated. Shear distortion was alulated based on the X-onfiguration LVDTs inlined at 45 and attahed of a square panel with dimension of lw lw. For the sake of safety, further

70 55 two LVDTs were attahed to the upper steel beam to measure the top out-of-plane deformation. The flexural and shear raks width were also measured through two LVDTs mounted at the first two major raks. The foregoing system of measurements made it possible to estimate the flexural, shear, and sliding omponents of the wall deformation, as disussed in the following setions. These estimates were also based on a series of strain gage measurements at various positions along the reinforing bars. Lateral top displaement Out-of-plane displaement Lateral displaement at h=l w Diagonal LVDT Vertial LVDT h= lw Conrete strain Wall-base sliding Base-floor sliding Figure.1 LVDTs instrumentation

71

72 CHAPTER 4 EXPERIMENTAL BEHAVIOR OF GFRP- REINFORCED CONCRETE SQUAT WALLS SUBJECTED TO SIMULATED EARTHQUAKE LOAD Foreword Authors and Affiliation o Ahmed Arafa: PhD andidate, Department of Civil Engineering, University of Sherbrooke. o Ahmed Sabry Farghaly: Postdotoral Fellow, Department of Civil Engineering, University of Sherbrooke, and Assoiate Professor, Assiut University, Egypt. o Brahim Benmokrane: Department of Civil Engineering, University of Sherbrooke, Sherbrooke. Journal: Journal of Composites for Constrution, ASCE Paper status: Submitted on Otober 6, 2016 Referene: Arafa, A., Farghaly, A. S., and Benmokrane, B., 2016, Experimental Behavior of GFRP-Reinfored Conrete Squat Walls subjeted to Simulated Earthquake Load, Journal of Composites for Constrution, ASCE. 56

73 57 Contribution in thesis: This hapter inludes the test results of speimen S4-80, G4-80, G6-80, G4, and GD. The feasibility of using GFRP RC squat walls as a lateral seismi system is disussed. The main differenes in behavior between steel and GFRP RC walls in term of rak pattern and failure mode, drift apaity and ultimate strength as well as the energy dissipation, are presented. The effet of using either bidiagonal web reinforement or sliding reinforement is also evaluated. Additionally, the preliminary design of speimens was assessed. Abstrat This study addressed the feasibility of reinfored-onrete squat walls totally reinfored with glass-fiber-reinfored-polymer (GFRP) bars ahieving the strength and drift requirements speified in various odes. Using nonorrodible GFRP bars represents an effetive method for overoming deterioration due to orrosion problems. The previous experimental studies on GFRP-reinfored mid-rise shear walls showed that GFRP reinforement an ontrol shear deformation, whih is a major problem in steel-reinfored squat walls. Five full-sale onrete squat walls with an aspet ratio (height to length ratio) of 1. one reinfored with steel bars (as a referene speimen) and four totally reinfored with GFRP bars were onstruted and tested to failure under quasi-stati reversed yli lateral loading. The reported test results learly show that properly designed and detailed GFRP-reinfored onrete squat walls an reah high deformation levels with no strength degradation. The results also show that the ahieved drift satisfies the limitation in most building odes. Aeptable levels of energy dissipation, ompared to the steel-reinfored squat wall, were observed. The promising results an provide impetus for onstruting onrete walls reinfored with GFRP and onstitute a step toward using GFRP reinforement in suh lateral-resisting systems. Keywords: GFRP bars, onrete walls, hystereti response, energy dissipation

74 58 Chapter 4: Experimental Behavior Of GFRP-Reinfored Conrete Squat Walls Subjeted 4.1. Introdution Squat walls are defined as strutural walls with a height to length ratio less than 2.0, whih is widely used as the primary seismi-fore-resisting omponent in low-rise strutures suh as nulear failities and industrial buildings. Squat walls also frequently serve as bridge piers and abutments (Paulay et al. 1982). Beause of their low aspet ratio, squat walls generate high shear fores at their bases to develop strutural flexural strength. This makes shear apaity a major issue in squat walls design sine the dutile behavior indued by the inelasti flexural yielding annot be ahieved (Paulay et al. 1982, Kuang and Ho 2008, Whyte and Stojadinovi 2014). Experimental investigations of squat walls have demonstrated that their behavior is dominated by inelasti shear deformations (shear distortion or sliding shear), indiating that these deformations develop and signifiantly inrease with the onset of flexural-reinforement yielding (Saatioglu 1991, Massone et al. 2009, Takahashi et al. 201, Luna et al. 2015). These deformations, in turn, rapidly degrade strength and stiffness with subsequent shear-indued damage. Thousands of bridges and parking garages in North Ameria that use squat walls need repair and rehabilitation, or omplete replaement due to orrosion problems. Using fiber-reinforedpolymer (FRP) bars as the main reinforement in onrete strutures in harsh environments is beoming a widely aepted solution to override orrosion issues [ACI 440 (2007)]. Sine glass-frp (GFRP) bars are relatively less expensive ompared to the other ommerially available FRP bars, GFRP bars in reinfored onrete strutures have found their way into numerous appliations suh as bridge dek slabs, beams, and olumns (El-Salakawy et al. 2005; Kassem et al. 2011, Tobbi et al. 2014). With the requirement of designing a multistory building with adequate strength and stiffness using GFRP reinforement, Mohamed et al. (2014a) reently onduted an experimental study to investigate the feasibility of using GFRP bars to reinfore mid-rise shear walls that would resist lateral loads. The reported test results learly revealed that properly designed and detailed GFRP-reinfored walls ould reah their flexural apaities with no strength degradation and with reasonable deformability in an inelasti stage. It was also found that

75 59 using elasti materials (GFRP bars) distributed the shear strain along the wall height, resulting in ontrolled shear deformation relatively to the loalized shear deformation in the yielding zone experiened by steel-reinfored shear walls (Mohamed et al. 2014b). Controlling shear deformation and solving orrosion problems in mid-rise shear walls with GFRP bars alls for investigations into the appliability of GFRP bars for reinforing squat walls, in whih these issues dominate. It is worth mentioning that none of the urrent FRP design odes and guidelines [CSA S806 (2012), ACI 440.1R (2015)] provide any reommendations about the seismi design of GFRP-reinfored squat walls. This study aimed at assessing the behavior of GFRP-reinfored squat walls under quasi-stati reversed yli loading to simulate seismi loading. The investigation foused mainly on the assessment of failure haraters, drift apaity, ultimate strength, and hystereti response. Doumentation on the amount of energy dissipation attained by the tested walls is also presented and the predition of ultimate strength is assessed Experimental Program Test Matrix of Speimens Five large-sale squat walls with an aspet ratio (height to length ratio) of 1. were onstruted and tested under quasi-stati reversed yli lateral loading up to failure. One speimen (ST4-80) was reinfored with onventional steel bars and served as a ontrol speimen; the remaining four (G4-80, G6-80, GD, and G4) were entirely reinfored with GFRP bars. The walls were designed and detailed aording to CSA A2. (2014) and CSA S806 (2012). The speimens measured 1500 mm in length, 2000 mm in height, and 200 mm in thikness. Eah speimen was ast vertially to ensure realisti onstrution-site onditions and onneted to a rigid base. Figure 4.1 shows the onrete dimensions and reinforement details of the test speimens.

76 60 Chapter 4: Experimental Behavior Of GFRP-Reinfored Conrete Squat Walls Subjeted steel or GFRP No.10 Boundary element No.10 steel or 120 vertial reinf. No.1 or No.19 steel or GFRP horizontal reinf. No.10 steel or 100 Θ = a) Conrete dimensions 8 GFRP No.10 Boundary Element GFRP 100 Θ = 45 8 GFRP No.10 Boundary Element GFRP 120 Vertial Reinf. GFRP 80 Horizontal Reinf. ) Reinforement onfigurations for GD d) Reinforement onfigurations for G4 Hz. Steel or GFRP Vl No.10 steel 8 No.10 steel or No.1 or No.19@ 80 or 120 GFRP Vl boundary Diagonal Web Reinf. No No.10 GFRP Vl. boundary Sliding reinf. e) Hz. Cross setion in ST4-80, G4-80, and f) Hz. Cross setion in GD 200 Hz No.1 80 Vl No No.10 GFRP Vl boundary g) Hz. Cross setion in G4 200 Figure 4.1 Conrete dimensions and reinforement details

77 61 All speimens were reinfored with the same longitudinal and transverse reinforement ratios in the boundaries. The longitudinal reinforement in the boundaries onsisted of 8 No. 10 (#) bars (steel or GFRP) and were laterally tied with No. 10 (#) retilinear spiral transverse reinforement (steel or GFRP), and spaed at 80 mm along the total wall height, whih is approximately the maximum spaing permitted in CSA S806 (2012). Two layers of web reinforement were used in all walls aording to CSA A2. (2014). Speimen G4-80 was designed to have almost equal flexural and shear apaities. The horizontal web reinforement onsisted of two layers of No. 1 (#4) GFRP bars spaed at 80 mm. The vertial web reinforement onsisted of two layers of No. 10 (#) GFRP bars spaed at 120 mm. Sliding shear was prevented by adding one layer of bidiagonal No. 10 (#) GFRP bars aross the potential sliding plane at an angle of 45, spaed at 100 mm and suffiiently anhored on eah side of the shear plane. The anhorage length for both the vertial reinforement and bidiagonal sliding reinforement was equal to the development length speified in CSA S806 (2012) multiplied by 1.25 to aount for the yli effet as suggested by Mohamed et al. (2014a). The steel-reinfored wall (S4-80) served as a ontrol speimen for G4-80, so both had idential reinforement onfigurations and ratios. It should be mentioned that this onfiguration ensured S4-80 would fail in flexure, sine its shear strength was muh higher than its flexural strength. Speimen G6-80 was idential to G4-80 exept that No. 19 (#6) horizontal web reinforement was used. Speimen GD was designed to investigate the effet of using bidiagonal web reinforement instead of horizontal and vertial web reinforement. The speimen was reinfored with two layers of bidiagonal No. 10 (#) GFRP bars spaed at 100 mm with angle of 45 relative to the longitudinal axis of the wall. The web reinforement details were seleted suh that GD had a flexural apaity similar to its ounterpart speimen G4-80. It should be mentioned that although GD had less web reinforement than G4-80, GD s predited shear strength was almost twie that of G4-80 owing to the absene of bent portions, whih are responsible for the lower strength of horizontal web reinforement; more details are disussed in speimen design setion. To investigate the sliding-shear resistane of GFRP, speimen G4 was built idential to G4-80 but without the bidiagonal sliding-shear

78 62 Chapter 4: Experimental Behavior Of GFRP-Reinfored Conrete Squat Walls Subjeted reinforement. The base for all speimens was reinfored with 25M Grade 420 deformed steel bars spaed 100 mm in eah diretion in both upper and lower levels. All speimens had a onrete over of 25 mm. Table 4.1 lists the reinforement details of the tested walls. Wall f ' (MPa) Table 4.1 Reinforement details and alulated apaities of the walls Reinforement Ratio ρ l ρ t ρ v ρ h ρ d ρ s P u (kn) V r (kn) V s (KN) Unonfined V fun (kn) P u /V fun Confined V fon (kn) P u /V fon S G G GD G f ' = onrete ompressive strength; ρ l = boundary longitudinal-bar reinforement ratio; ρ t = boundary-tie reinforement ratio; ρ v = web vertial-bar reinforement ratio; ρ h = horizontal web reinforement ratio; ρ d = bidiagonal web reinforement ratio; ρ s = bidiagonal sliding-shear reinforement ratio; P u = experimental ultimate strength; V r = predited shear strength; V s = predited sliding-shear strength; V fun = predited flexural strength for unonfined setion; V fon = predited flexural strength for onfined setion Material Properties The test speimens were ast using normal-weight, ready-mixed onrete with a target 28-day ompressive strength of 40 MPa. Table 4.1 gives the atual onrete ompressive strength (f') based on the average of at least three mm ylinders for eah onrete bath on the day of speimen testing. Speimen S4-80 ontained No. 10 (#) and No. 1 (#4) grade 420 deformed steel bars: No. 10 (#) for vertial and retilinear spiral reinforement; No. 1 (#4) for horizontal reinforement. The GFRP reinforing bars onsisted of three diameters of Grade III sand-oated bars [CSA S807 (2015)]: No. 10 (#) for vertial and retilinear spiral reinforement and No. 1 (#4) and No. 19 (#6) for horizontal reinforement. The longitudinal tensile properties of the GFRP bars were determined by testing five speimens aording to ASTM D7205 (2011), in the ase of the straight bars, and test method B.5 in ACI 440.R (2004), in the ase of the bent bars. The manufaturer provided the properties of the steel bars. Table 4.2 lists the material properties of the reinforing bars. Figure 4.2 shows the vertial, horizontal, and retilinear spiral reinforement and a typial assembled age.

79 6 Bar Designated Bar Diameter (mm) Table 4.2 Reinforement mehanial properties Nominal Area 1 (mm 2 ) Immersed Area (mm 2 ) Tensile Modulus of Elastiity 2 (GPa) Tensile Strength 2 * (MPa) Average Strain at Ultimate (%) Straight bars No. 10 GFRP No. 10 steel fy = 420 εy = 0.2 No. 1 steel fy= 420 εy = 0.2 Bent No. 10 GFRP retilinear spiral Straight Bent Bent No. 1 GFRP horizontal bar Straight Bent Bent No. 19 GFRP horizontal bar Straight Bent fy: steel yielding strength, εy: steel yielding strain. 1 Aording to CSA S807 (CSA, 2010) 2 Tensile properties were alulated using nominal ross-setional areas. *Guaranteed tensile strength: Average value standard deviation (ACI ) Retilinear spiral (No. 10 GFRP) No. 1 GFRP No. 19 GFRP Horizontal bars Vertial bar (No. 10 GFRP) Assembly of wall age Figure 4.2 GFRP reinforement and wall age

80 64 Chapter 4: Experimental Behavior Of GFRP-Reinfored Conrete Squat Walls Subjeted Speimen Design The predited flexural strength for the investigated walls was alulated based on plane setional analysis. The analysis was based on strain ompatibility, internal fore equilibrium, and the ontrolling mode of failure. The alulation was arried out onsidering an unonfined onrete setion (onrete ompressive stain equal to aording to CSA S806 (2012) and CSA A2. (2014)), and a onfined onrete setion (onrete ompressive stain equal to and for steel and GFRP RC walls, respetively) (Mohamed et al. 2014a). The ontribution of the longitudinal GFRP bars in ompression to the flexural strength was onsidered by assuming its strength in ompression was 50% of its tensile strength, as suggested by Deitz et al. (200), while the ontribution of the diagonal web or sliding-shear reinforement was obtained by onsidering their omponents in the longitudinal diretion of the wall. Regarding the shear apaity, owing to the absene of seismi provisions for GFRP-reinfored squat walls in CSA S806 (2012), we adopted the onept provided in CSA A2. (2014) for steel. Clause in CSA A2. (2014) speifies that, under seismi loading, the shear fore is to be resisted by horizontal web reinforement only, with no reliane on the onrete ontribution to shear strength due to the effet of riss ross shear raks pattern and the assoiated degradation of onrete shear resistane. The shear strength of the steel-reinfored wall was determined as follows: V r s Av f ydv ot Vss (4.1) s where Vr is the shear strength, Vss is the shear resistane provided by the horizontal web reinforement, s is the material resistane fator for steel, Av is the area of horizontal web reinforement within the distane s, fy is the speified yield strength of the horizontal web reinforement, dv is the effetive shear depth equal to the greater of 0.9d or 0.72 lw but not less than 0.8 lw, θ is the angle of inlination of diagonal ompressive stresses to the longitudinal axis of the wall and equal to 45, and s is the spaing of the horizontal web reinforement measured along the longitudinal axis of the wall.

81 65 The shear apaity of horizontal web reinforement in the GFRP-reinfored walls was alulated as follows: V r 0.4 f Av f fud v ot M f dv V f Vsf, l, l (4.2) s 2E A f f where, f is the material resistane fator for GFRP; is the longitudinal strain at mid-depth of the setion; Mf and Vf are the bending moment and shear fore at the ritial setion for shear, respetively; and θ ranges from 0 to 60. For speimen GD, whih had bidiagonal web reinforement, the shear resisted by the diagonal web reinforement was omputed from the general formula as follows: V r f Av f fud v (ot ot) sin Vsf (4.) s where α is the angle between the diagonal bars and the longitudinal axis of the member. It should be noted that CSA S806 (2012) urrently limits the stress level in the FRP transverse-reinforement stirrups to: (1) avoid failure at the bent portion of the FRP stirrup, (2) ontrol shear-rak widths under servie load, and () onstrain the maximum size of the diagonal raks at ultimate state to not seriously diminish shear transfer by aggregate interlok. Based on the available geometri requirement for the bent portion, the ultimate strength of the bent portion ranged between 0.4 and 0.45 fu. The CSA S806 (2012) onservatively reommends the ultimate strength be equal 0.4fu. To limit the allowable rakwidth size under servie loads, CSA S806 (2012) limits the maximum strain in the stirrups to 5000 με. This level of strain was also found to satisfy the third ondition of maintaining shear transfer through aggregate interlok (Razaqpur and Spadea. 2015). Sine our study ignored the shear apaity arried by onrete (in terms of aggregate interlok and dowel ation) as well as the elasti nature of GFRP bars, whih should enable the raks to realign and lose after an earthquake, the seond and third riteria were omitted. Hene, the ultimate strength of the GFRP was limited to 0.4fu (Equation 4.2). In speimen GD, whih ontained bidiagonal web

82 66 Chapter 4: Experimental Behavior Of GFRP-Reinfored Conrete Squat Walls Subjeted reinforement, the bar stress was limited to fu owing to the absene of bent portions (Equation 4.). Constrution joints in squat walls may drastially degrade under yli loading and beome the weakest link in the hain, resulting in premature sliding-shear failure. Clause in CSA A2. (2014) gives the following equation for alulating the resistane to sliding-shear failure: os v f (4.4) r s v y f where vr (MPa) is the sliding-shear strength; λ is a fator to aount for low-density onrete, is the resistane fator for onrete, is the ohesion stress, μ is the oeffiient of frition, fy is the yield strength for the steel, λ ( + μσ) shall not exeed 0.25 f', f is the resistane fator for steel, and αf is the angle between the shear-frition reinforement and the shear plane. Aording to Clause , the values of σ and ρ are: N vf y sinf, A g A vf v (4.5) Av where Avf is the area of the shear-frition reinforement, Av is area of onrete setion resisting shear transfer, and N is the unfatored permanent ompressive load perpendiular to the shear plane. Sine the surfae of onrete base was not intentionally roughened, the sliding resistane was alulated based on a ohesion stress () of 0.25 MPa and a frition oeffiient (μ) of 0.60, as speified in Clause in CSA A2. (2014) It should be noted that the frition oeffiient of 0.6 was speified by the ode based on Mattok (1977). He found that, when onrete is ast against hardened onrete without roughening, sliding-shear resistane is primarily due to reinforement dowel ation. Test results learly indiated that the sliding-shear strength was very lose to the shear yield strength of the shear-transfer reinforement (0.58 ρ fy).

83 67 Due to the lak of provisions in CSA S806 (2012) for alulating sliding shear, we opted to alulate it for GFRP-reinfored walls based on CSA (2014), as shown in Equations 4.4 and 4.5. The ontribution of all vertial GFRP reinforement rossing the onstrution joint was, however, ignored in terms of dowel shear resistane due to the lower strength and stiffness of GFRP bars in the transverse diretion [ACI 440.1R (2015)]. Hene, the GFRP-reinfored speimens were found to be suseptible to sliding-shear failure. Therefore, two layers of bidiagonal sliding-shear reinforement were used in the potential sliding plane in all speimens exept G4, in whih the effet of bidiagonal reinforement was investigated. Table 4.1 provides the predited strength in flexure, shear, and sliding shear of the tested speimens. It should be noted that the material-redution fators and safety fators were only used in the design to reflet unertainties in material harateristis as well as ontrator quality. In the experimental study, however, all resistane fators were set to unity sine the material harateristis were aurately determined and the speimens were produed in the laboratory under high quality ontrol Test Setup and Proedure Figure 4. provides the layout of the test setup. All speimens were tested laterally as a vertial antilever with fore applied through a top omprised of a speially fabriated steel load-transfer assembly. Sine the axial ompressive stress in squat walls due to gravity load generally represent an insignifiant perentage of the produt of onrete ompressive strength and gross area, the test speimens were tested without axial load. The lateral yli displaement was applied at 2550 mm above the base of the wall with a hydrauli atuator with a maximum apaity of 1000 kn and a maximum stroke of ± 250 mm. Out-of-plane braing was employed at the level of the steel transfer beam to prevent out-of-plane displaement during testing. The seismi loading in this study was applied in several steps under displaement-ontrol mode throughout the test. Eah loading step onsisted of two idential displaement yles with inrements of ±2 mm up to 10 mm, followed by inrements of ±5 mm up to 50 mm, and thereafter inrements of ±10 mm up to failure. Figure 4.4 gives a typial sequene of displaement yles.

84 68 Chapter 4: Experimental Behavior Of GFRP-Reinfored Conrete Squat Walls Subjeted Lateral braing Atuator Steel beam Squat-wall speimen Dywidag bars Base Figure 4. Test setup Lateral drift (%) Cyle number Lateral displaement (mm) Figure 4.4 Loading history Instrumentation Deformation response was monitored with a series of linear variable displaement transduers (LVDTs) and strain gauges (Figure 4.5). One LVDT was installed to measure the lateral displaement at the top of the walls, another at the onstrution joint to monitor sliding between the wall and base, and a third to maintain a rigid onnetion to the laboratory floor. Two LVDTs were mounted lose to the boundary elements to measure onrete strain. For the

85 69 sake of safety, two more LVDTs were attahed to the upper steel beam to measure the top outof-plane deformation. Crak width was also measured with two LVDTs mounted at the first two major raks. Lateral top displaement Out-of-plane displaement Conrete strain Wall base sliding Base floor sliding Figure 4.5 Instrumentation 4.. Test Results and Disussion General Behavior and Mode of Failure Figure 4.6 depits the typial rak propagation for the tested walls. Initially, a few raks predominantly horizontal propagated in the lower part of the walls. With further loading, these raks aquired some inlination in the entral zone of the web due to shear stresses and propagated up to one-third of the wall height. As loading ontinued, new shear raks tended to propagate loser to the top of the wall. These raks were steeper than those propagated in the lower part of the wall. Their inlination and width were, however; signifiantly lower lose to the boundary element. This ontrol stems from the presene of heavy reinforement and onfinement in the boundary elements, whih are known to favorably affet the shear apaity of squat walls (Salonikios et al. 1999, Kassem 2015). As larger displaements were imposed, flexural shear or shear raks originating from eah side ontinued to progressively extend down to the opposite side with inreased inlination and interseted eah other forming a rissross pattern. With inreased displaement, over splitting gradually initiated at the most ompressed fibers of the boundary, as shown in Figures 4.7a and 4.8a for the steel- and GFRP

86 70 Chapter 4: Experimental Behavior Of GFRP-Reinfored Conrete Squat Walls Subjeted RC squat walls, respetively, assoiated with gradual spalling of the onrete over (Figures 4.7b and 4.8b) for the steel- and GFRP RC squat walls, respetively. At this stage, yielding of the longitudinal reinforement in the steel RC squat wall beame signifiant along both sides of the wall and loalized at the plasti-hinge zone, whih extended above the height of the bidiagonal sliding reinforement. As a result, a notieable horizontal rak was observed (Figure 4.7) above the bidiagonal sliding reinforement going through the entire wall length. Consequently, a loalized sliding-shear deformation was learly apparent (Figure 4.7d) assoiated with exessive deterioration of the onrete over in this region, leading to progressive degradation of lateral strength. The substantial deterioration assoiated with the in-plane bukling of longitudinal bars (Figure 4.7e) aused failure as this stage, followed by a drasti drop in lateral strength. Figure 4.8f shows the speimen at the end of testing. It should be mentioned that, due to the additional shear resistane provided by the diagonal ross sliding reinforement, the most damaged wall setion was pushed away from the base wall interfae, where the interation between moment and shear is the largest. Conrete deterioration above Longitudinal bar bukling Conrete rushing Conrete rushing a) S4-80 b) G4-80 () G6-80 Conrete rushing Conrete rushing (d) GD (e) G4 Figure 4.6 Crak pattern (note: the raks with bold line are the major deteted shear raks)

87 71 a b d e f Figure 4.7 Failure progression of speimen S4-80: (a) vertial over splitting, (b) spalling of onrete over, () onrete deterioration above sliding reinforement, (d) spalling of the deteriorated onrete, (e) bukling of longitudinal reinforement ausing failure, (f) speimen fae at failure Contrary to S4-80 speimen, the GFRP RC speimens ontinued arrying loads with no strength degradation. In G4-80, G6-80, and G4, the failure started with gradual deterioration and splitting of the onrete in the ompressed boundary element; flexural failure was then imminent. The speimens ultimately failed in flexural ompression (Figure 4.8), assoiated with rupture of the GFRP ties at the bent portion (Figure 4.8d) and frature in the ompressed longitudinal reinforement (Figure 4.8e). In the last speimen, GD, the failure progression was quite different; a few yles before failure, the speimen experiened out-of-plane bukling of some diagonal ompression bars at the lower part of the wall. As a result, progressive spalling of the web onrete over in the viinity of the bukled bars was evidened (Figure 4.9a). Speimen GD eventually failed beause of exessive out-of-plane-bukling of the diagonal ompression bars assoiated with onrete-ore rushing in the ompression zone, as shown in Figure 4.9b,, and d. Table 4. summarizes the performane parameters during failure progression of the test speimens.

88 72 Chapter 4: Experimental Behavior Of GFRP-Reinfored Conrete Squat Walls Subjeted a b d e Figure 4.8 Failure progression of speimens G4-80, G6-80, and G4: (a) vertial over splitting, (b) spalling of onrete over, () onrete rushing ausing failure, (d) rupture of GFRP tie, (e) frature of longitudinal bars a b d Figure 4.9 Failure of speimen GD: (a) out-of-plane bukling, (b) onrete rushing, () rupture of GFRP tie, (d) frature of longitudinal bars

89 7 First flexural rak Table 4. Wall failure progression Stage Wall P P/P u Δ d (%) S G G GD G First shear rak Yielding Vertial splitting Exess. Cover spalling Exessive bukling Out-of-plane bukling Conrete deterioration Conrete rushing S G G GD G S G G GD G S G G GD G S G G GD G S G G GD G S G G GD G S G G GD G S G G GD G Note: P = applied lateral load (kn); P u = ultimate lateral load (kn); Δ = lateral top displaement (mm); d% = lateral drift (= Δ/h w 100).

90 74 Chapter 4: Experimental Behavior Of GFRP-Reinfored Conrete Squat Walls Subjeted Hystereti Response Figure 4.10 gives the applied yli load versus the lateral drift, and the points illustrating the speial events of damage during the loading proess. The hystereti response of eah squat wall showed reasonable symmetri lateral load top-drift relationships for loading in the positive and negative diretions until failure ourred at one end. The response of speimen S4-80 was initially linear up to the formation of the first rak. During early loading, the longitudinal reinforement at the boundary yielded at a drift ratio of 0.4%, followed by widening in the hystereti loops, in addition to a gradual derease in overall stiffness. Thereafter, the hystereti loops got gradually wider with larger ommened residual displaement. At 1% lateral drift, onrete-over splitting was observed; lear over spalling ontinued at a lateral drift of 1.25%, at whih point the wall ahieved its ultimate lateral load of 54 kn. This was followed by strength degradation due to loalized sliding-shear deformations. It is lear that the reloading proess in the opposite loading diretion exhibited initial softening, followed by gradual stiffening, whih is attributed to the loalized shear deformations. As bar bukling ourred orresponding to a 2% drift ratio, the wall s strength remarkably dereased to 82% and 55% of the ultimate load in the positive and negative loading diretions, respetively. For the GFRP RC walls, the speimens exhibited initial stiff behavior up to the initiation of the first flexural rak, at whih point a redution in stiffness was observed. This was followed by a gradual degradation in stiffness as raks propagated. The unloading/reloading urves seemed to demonstrate linearity due to the linear nature of the GFRP bars. At a lateral drift ranging from 1.% to 1.5%, raks propagation stabilized and the onrete over split, resulting in a slight widening of the hystereti loops. This defined the beginning of the inelasti stage in the wall and the speimens lost self-entering. Suh response has been previously observed in mid-rise GFRP RC walls (Mohamed et al. 2014a).

91 75 Lateral drift (%) Lateral drift (%) S4-80 G4-80 Load (kn) Load (kn) Load (kn) Top displaement (mm) Top displaement (mm) Lateral drift (%) Lateral drift (%) G6-80 GD Load (kn) Top displaement (mm) Lateral drift (%) Top displaement (mm) Load (kn) G4 Initial yielding (S4-80 ) Cover splitting Exessive over spalling Bars bukling (S4-80 ) Conrete deterioration Out-of-plane bukling (GD) Conrete rushing Top displaement (mm) Figure 4.10 Hystereti response

92 76 Chapter 4: Experimental Behavior Of GFRP-Reinfored Conrete Squat Walls Subjeted Spalling of the onrete over gradually initiated after over splitting and beame more signifiant at a lateral drift ranging from 1.75% to 2%. Nevertheless, the onrete ore remained intat due to the onfinement at the boundary elements. As a result, G4-80, G6-80, and G4 experiened no strength degradation up to drift ratios of 2.75%, 2.9%, and 2.6%, orresponding to ultimate loads of 912, 95, and 740 kn, respetively. This stage was followed by a gradual degradation in strength up to %,.1%, and 2.75% drift for G4-80, G6-80, and G4, respetively. At this point, the lateral strength dereased to 88%, 87%, and 89% of the peak apaity, respetively, followed by a sudden drop in apaity as a result of onrete rushing. The diagonally web RC wall (GD) was similar to the other GFRP RC squat walls, exept that the diagonal web-bars bukled and the web onrete over spalled at a lateral drift of 2.25%, followed by a lear ability to sustain further displaement but in softer manner up to a lateral drift of 2.9%. At that point, a drop in strength indued by exessive out-of-plane bukling and onrete rushing was observed. Speimen GD attained an ultimate lateral load of 804 kn. The NBCC (2010) and CSA A2. (2014) require that strutural walls be able to maintain strutural integrity at least three-quarters of their ultimate apaity through peak displaements equal to a story drift ratio of 2.0%. The four GFRP RC squat walls safely reahed this drift ratio without strength degradation; indiating the feasibility of using GFRP RC squat walls in earthquake regions Steel-versus GFRP RC Walls The rak patterns of S4-80 and G4-80 were generally similar, although G4-80 exhibited more distributed and intensive raks (Figure 4.6). This ould be attributed to two main reasons: (1) the differene in bond harateristis between GFRP and steel bars; (2) the different deformation levels experiened by the speimens. Figure 4.6 also learly shows that the angle between the inlined raks and longitudinal axis of the speimen was larger in G4-80 than in S4-80 (ranging from 55 ο to 59 ο and 5 ο to 40 ο in G4-80 and S4-80, respetively) due to the higher axial rigidity of speimen S4-80 (Razaqpur and Spadea. 2015). The raks in G4-80 tended to realign and lose between load reversals with negligible residual rak width (the rak width at zero loading) up to a lateral drift of 1.5%, orresponding to onrete-over

93 77 splitting, while S4-80 exhibited signifiant residual rak width, espeially after the steel reinforement yielded. This an be attributed to the elasti nature of GFRP bars and onsidered an advantage in their use. Moreover, it provides evidene of satisfatory bond between the GFRP reinforement and onrete. Furthermore, while S4-80 had smaller rak widths than G4-80 during early loading, rak widths substantially inreased and beame remarkably larger in S4-80 after yielding ourred. The measured maximum flexural-rak width at failure was 6 mm and 2.1 mm in S4-80 and G4-80, respetively. It is worth mentioning that, despite S4-80 and G4-80 having the same reinforement ratios and onfigurations, S4-80 exhibited extensive sliding shear failure while G4-80 failed in flexural ompression and almost the full flexural apaity for the onfined setion was ahieved, as will be disussed later. In S4-80, a major ontinuous rak along the wall length formed above the sliding reinforement and remained open even under ompressive reversal loading beause the longitudinal reinforement yielded. The shear stress was therefore transferred along this rak primarily by longitudinal-reinforement dowel ation, sine the frition-resistane fores maintained by aggregate interlok were degraded under yling. Due to the relatively flexible nature of this mehanism, the sliding-shear deformations loalized in this zone and aused onrete deterioration with subsequent degradation of lateral-loading apaity before the full flexural apaity was ahieved. Suh interation between flexural and sliding-shear deformations has been doumented in a number of experimental studies on steel RC squat walls (Paulay et al. 1982, Salonikios et al. 1999, Whyte and Stojadinovi 2014). In G4-80, the elasti nature of GFRP bars helped the raks realign and lose between load reversals, and spread deformations along the wall height rather than loalizing them in the plasti-hinge zone. As a result, the speimen was able to ahieve its flexural apaity with no sign of sliding distress. Figure 4.11 shows the envelope urve S4-80 and G4-80. Initially, both speimens had similar stiffness until initiation of the flexural rak. Due to the low modulus of elastiity of GFRP bars ompared to steel bars, G4-80 exhibited a softer response than S4-80 until their responses interseted at a lateral drift of 1.5%, orresponding to 99% and 56% of ultimate load for S4-80 and G4-80, respetively. Thereafter, S4-80 experiened signifiant strength degradation

94 78 Chapter 4: Experimental Behavior Of GFRP-Reinfored Conrete Squat Walls Subjeted due to the loalized shear deformations up to failure at 2% lateral drift. Conversely G4-80 s strength kept inreasing almost linearly to ahieve an ultimate load and drift apaity higher than S4-80, with ratios equal to 71% and 50%, respetively. The higher ultimate load and drift ratio ahieved by G4-80 indiates the aeptable behavior of GFRP RC squat walls in resisting lateral loads. It should be mentioned that although the softer behavior of G4-80 ompared to S4-80 would inrease the displaement demand, it ould be onsidered as advantage of using GFRP bars. This is due to the fat that the lower strutural stiffness results in a longer natural period of vibration and, onsequently, lower seismi fore demand (Sharbatdar and Saatioglu 2009) Lateral drift (%) % Load (kn) S4-80 G Initial yielding Cover splitting Cover spalling Bar bukling Conrete deterioration Conrete rushing Top displaement (mm) Figure 4.11 Envelope urves: Steel vs. GFRP Effet of Reinforement Configurations The initial flexural or shear raking load was not affeted by using different horizontal webreinforement ratios, hanging from orthogonal to diagonal web reinforement, or removing the diagonal sliding-shear reinforement (Table 4.). The studied parameters, however, appeared to signifiantly ontrol the shear-rak width as the measured maximum rak width was 0.51, 0.4, and 0.98 mm for G6-80, GD, and G4-80, respetively (The loations of the measured maximum shear rak are indiated in Figure 4.6). The greater diameter of horizontal web reinforement in G6-80 ompared to G4-80 aused less shear strain to develop,

95 79 leading to ontrolled rak width. The diagonal web reinforement in GD was almost perpendiular to the shear-rak diretion, so it ated primarily in diret tension, while the horizontal and vertial web reinforement in G4-80 interseted the shear raks at 0 to 60 ο, so it tended to at essentially as dowels (Salonikios et al. 1999, Sittipunt and Wood 1995). G4 exhibited larger rak widths in its lower part than G4-80 due to the absene of bidiagonal sliding reinforement. Figure 4.12 shows the envelope urve for the GFRP RC speimens. Inreasing the horizontal web reinforement in G6-80 with respet to G4-80 did not signifiantly affet drift apaity and ultimate strength as the differene was less than %. This minor differene ould be attributed to the shear deformation, whih was greater in G4-80 than in G6-80, ausing greater deformability of the onrete. The lak of bidiagonal sliding reinforement, however, signifiantly affeted the ultimate apaity; sine G4 had an ultimate apaity 2% lower than G4-80. This is due to the fat that sliding reinforement ontributes not only to sliding shear but to flexure as well. It is interesting to note that, while GD had approximately 47% less total web reinforement than G4-80, both speimens exhibited similar behavior up to 2.25% drift. However, after this stage G4-80 kept inreasing up to a drift ratio of %, while GD exhibited softer behavior, failing at similar drift ratio of 2.9% but at a lower ultimate apaity. The differene in onrete ompressive strength between G4-80 and GD is eliminated by alulating the shear stress, whih was found to be 0.6 ' f and 0.56 ' f MPa, respetively. Hene, GD s redued lateral apaity at failure ould be attributed to the out-of-plane bukling of the diagonal web reinforement. Transverse links ould be an effiient solution to delay or eliminate the out-of-plane bukling; this might, however, raise onstrution issues. Other solution ould also be rendered by prestressing the diagonal bars. In suh ase, the tension indued in the diagonal bars would allow exploiting the advaned mehanial properties of GFRP. The advantage of using diagonal web reinforement is ontrol the shear-rak width with less reinforement so as to attain a apaity similar to that when using an orthogonal grid. In addition, straight diagonal reinforement is muh easier to fabriate than horizontal

96 80 Chapter 4: Experimental Behavior Of GFRP-Reinfored Conrete Squat Walls Subjeted reinforement with bent ends. Therefore, from eonomi and design points of view, using diagonal reinforement as web reinforement might be attrative. Lateral load (kn) Lateral drift (%) G6-80 GD G4 G4-80 Cover splitting Cover spalling Out-of-plane bar Conrete deterioration Conrete rushing Top displaement (mm) Figure 4.12 Envelope urves: Top displaement (mm) 4.4. Predition of Ultimate Strength The predited ultimate strength of the tested walls in flexure, shear, and sliding, using the methodology disussed in setion 4.2., are listed in Table 4.1. The experimentally obtained apaity for the GFRP RC squat walls that exhibited flexural failure is signifiantly larger than the predited flexural strength assuming an unonfined onrete setion (onrete ompressive strain = aording to the CSA S806 (2012) and CSA A2. (2014). This an be attributed to the effet of onfinement at the boundary elements, whih remarkably improved onrete ompressive strain and onsequently improved the flexural strength of GFRP RC elements. This effet has been reognized in many studies (Tavassoli et al. 2015; Mohamed et al. 2014a; Ali and El-Salakawy 2016). Figure 4.1 shows that the measured ompressive strain at the GFRP RC walls toes are more than double the value speified in the CSA odes. In ontrast, due to the sliding shear deformations, the maximum measured onrete ompressive strain in S4-80 is muh lower than that in the ompanion speimens.

97 81 Given the neessity of onsidering the onfinement effet in GFRP RC squat walls, their flexural apaity was realulated onsidering the onfinement effet using the experimentally reorded onrete ompressive strain. The ratios between the experimentally obtained ultimate strengths to their predited analytial values, listed in Table 1, generally reflet that using plane setional analysis gave a reasonable estimate of the ultimate flexural strength as the ratios at the ultimate levels are within a 5% differene. The maximum differene an be observed in speimen GD, whih ould be attributed to the out-of-plane bukling of web reinforement and subsequent over loss. While G4 was expeted to fail due to sliding-shear, it did so identially to G4-80 (gradual flexural ompression); while the sliding displaement was negligible (less than 1% of the top displaement). This indiates that the assumption the shear resistane along unintentionally roughened shear plane is equal to the transverse shear strength of the rossing reinforement at right angles (CSA S806) is not ompletely valid for GFRP RC squat walls. Hene, this suggests that the mehanism of aggregate interlok in the flexural ompression zone ould be engaged with the dowel ation to resist the applied shear stresses as the flexural raks that formed between the wall and the base tended to realign and lok up in the ompression zone with load reversal due to the elasti nature of the GFRP bars. Further experimental study should be onduted to address this differene in behavior and to expand understanding of the sliding-shear resistane for joints reinfored with GFRP bars under tension ompression load reversals The maximum measured strain in the straight portion of G4-80 s horizontal web reinforement (4840 με) was muh lower than the rupture strain in the bent portion (8000 με aording to CSA S806-12). The experimentally obtained ultimate strength of G4-80 was, however, slightly higher than the predited value. This indiates that the shear stresses were not arried only by the horizontal web reinforement, as indiated in CSA A2. (2014). Further investigation is thus needed to determine the load-transfer mehanisms in GFRP RC squat walls.

98 82 Chapter 4: Experimental Behavior Of GFRP-Reinfored Conrete Squat Walls Subjeted S4-80 G4-80 Load (kn) Load (kn) LVDT 1 LVDT 2 LVDT 1 LVDT 2 Conrete strain G6-80 GD Conrete strain Load (kn) LVDT 1 LVDT 2 Load (kn) LVDT 1 LVDT 2 Conrete strain Conrete strain G4 (+) P (-) LVDT 1 LVDT 2 Load (kn) LVDT 1 LVDT 2 LVDT 1 LVDT 2 Conrete strain Figure 4.1 Conrete strain 4.5. Energy Dissipation Figure 4.14 shows the umulative energy dissipation of the tested speimens, whih is a ommon index to desribe the ability of a struture to dissipate imposed seismi energy. The umulative energy dissipation was alulated by summing up the dissipated energy values in onseutive load displaement loops throughout the test. The residual displaement values

99 8 were also plotted against the orresponding umulative dissipated energy values in eah yle (Figure 4.14b). As expeted, at all drift levels, S4-80 exhibited the highest energy dissipation but with substantial residual displaement. Conversely, due to the elasti behavior of the GFRP bars, the GFRP RC walls exhibited almost linear behavior with limited residual deformations. Although G4-80 had a pinhed behavior, it provided energy dissipation similar to S4-80 at the moderate-damage load level orresponding to onrete-over splitting. The dissipated energy was 17.4 and 14.8 kn.m, orresponding to 7. and 1.08 mm residual displaement for S4-80 and G4-80, respetively. The higher strength ahieved by G4-80 ompared to S4-80 signifiantly ontributed to the steel- and GFRP RC squat walls attaining omparable levels of dissipated energy. Similarly, the energy dissipated by S4-80 and G4-80 at onrete-over spalling was similar: 1 and 28 kn.m, orresponding to 1.2 and 2.7 mm residual displaement, respetively. At this stage, a notieable opening in the hystereti loops of S4-80 and G4-80 an be seen, resulting in an inrease in the rate of dissipated energy. That was due to exessive yielding of the steel in S4-80 and due to onrete plastiity in G4-80. The maximum ahieved energy dissipation at failure was 94 and 5 kn.m, orresponding to 26 and 4.6 mm residual displaement for S4-80 and G4-80, respetively. Figure 4.14 shows similar umulative dissipated energy of approximately 16 kn.m by all the GFRP RC walls up to onrete-over splitting. The effet of removing sliding reinforement in G4 or hanging the web reinforement onfiguration in GD was apparent after the drift level orresponding to onrete over splitting, while no signifiant effet was observed due to inreasing the horizontal web reinforement in G6-80. The maximum aumulated energy dissipation was 5, 56, 70, and 7 kn.m orresponding to 4.6, 5.4, 7.1, and 8.6 mm residual displaement for G4-80, G6-80, GD, and G4, respetively. The fat that G4 dissipated more energy than G4-80 an be attributed to the lak of bidiagonal sliding reinforement, whih allowed G4 to experiene higher deformations during unloading. The higher energy dissipation in GD ould be attributed to the damage indued by out-of-plane bukling and subsequent onrete-over spalling.

100 84 Chapter 4: Experimental Behavior Of GFRP-Reinfored Conrete Squat Walls Subjeted Cumulative energy dissipation (kn.m) Lateral drift (%) S4- Bar bukling G4- Conrete rushing 80 G6- GD G Yielding Cover spalling Cover splitting Top displaement (mm) (a) Cumulative energy dissipation Cumulative energy dissipation (kn.m) Conrete rushing Cover spalling Yielding Cover splitting Bar bukling Top displaement (mm) (b) Residual displaement S4- G4- G6- GD G4 Figure 4.14 Energy dissipation The disussion demonstrates the apability of GFRP RC squat walls to dissipate energy through the inelasti behavior of onrete. Additionally, the disussion reveals that a GFRP RC squat wall ould potentially be restored after an earthquake event, while there will likely be tehnial diffiulties in repairing steel RC walls owing to exessive permanent deformations Conlusions This hapter presented a test program aimed at studying the appliability using GFRP RC squat walls as seismi-fore-resisting elements. The results of testing five steel- and GFRP RC squat walls under simulated earthquake loading are presented. Based on the analysis of the experimental results, the following onlusions were reahed: 1. The GFRP reinforement provided the laterally loaded squat walls with stable behavior through the hystereti response, as no strength degradation or signs of premature shear failure was observed in omparison to the steel RC squat wall.

101 85 2. The elasti nature of the GFRP bars helped the raks realign and lose between load reversals and distribute shear deformations along the wall height. This assisted in avoiding premature sliding failure in GFRP RC walls.. GFRP ties at the boundary elements played a signifiant role in delaying rushing of the onrete ore with a subsequent inrease in ultimate apaity. 4. The GFRP-reinfored squat walls exhibited reasonable levels of energy dissipation assoiated with relatively small residual displaements ompared to the steel-reinfored wall as a result of the plasti deformations of the onrete 5. Bidiagonal web reinforement was more effetive than onventional web reinforement in ontrolling shear-rak width. Eliminating the out-of-plane bukling in the diagonally reinfored wall ould be ahieved by using transverse link or by prestressing the diagonal grid, but this might lead to onstrution issues. 6. The GFRP RC wall demonstrated their ability to dissipate energy through plasti deformations of onrete. 7. Using horizontal web reinforement exeeding the amount required to arry the ultimate flexural strength had no effet on either the ultimate strength or the drift ratio. On the other hand, it learly ontrolled the shear-rak width. 8. Using bidiagonal sliding reinforement is not neessary in GFRP RC squat walls; further investigation is needed to address the elasti nature of GFRP bars. 9. The shear stresses indued by the lateral load were not arried only by the horizontal web reinforement as indiated in CSA A2. (2014). Further studies, therefore, are required to larify this point. 10. All of the GFRP RC walls safely attained the maximum allowable drift ratio required by both the NBCC (2010) and CSA A2. (2014) with no strength degradation, while the residual deformation was insignifiant. This indiates the appliability of GFRP RC squat walls in resisting lateral loads in low to moderate earthquake-prone regions.

102

103 CHAPTER 5 Effet of Web Reinforement on the Seismi Response of Conrete Squat Walls Reinfored with Glass-FRP Bars Foreword Authors and Affiliation o Ahmed Arafa: PhD andidate, Department of Civil Engineering, University of Sherbrooke. o Ahmed Sabry Farghaly: Postdotoral Fellow, Department of Civil Engineering, University of Sherbrooke, and Assoiate Professor, Assiut University, Egypt. o Brahim Benmokrane: Department of Civil Engineering, University of Sherbrooke, Sherbrooke. Journal: Engineering Strutures Paper status: Submitted on May 11, 2017 Referene: Arafa, A., Farghaly, A. S., and Benmokrane, B., 2017 Effet of Web Reinforement on the Seismi Response of Conrete Squat Walls Reinfored with Glass-FRP Bars, Engineering Strutures Journal. Journal: Journal of Composites for Constrution Paper status: Submitted on May 1, 2017 Referene: Arafa, A., Farghaly, A. S., and Benmokrane, B., 2017 (submitted tehnial Note) Predition of Flexure and Shear Strength of Conrete Squat Walls Reinfored with Glass-FRP Bars, Journal of Composites for Constrution, ASCE. 86

104 87 Contribution in thesis: This hapter inludes the test results of speimen G4-250, G4-160, G4-80, G6-80, G-V, and G-H. It is aimed at evaluation the effet of web reinforement onfigurations (horizontal and/or vertial) and the horizontal web reinforement ratio. Emphasis is given to the effet on ultimate strength, mode of failure, drift apaity. Doumentation of strain distribution in the horizontal and vertial diretion is presented. Evaluation of the ultimate strength using the ACI 440.1R-15 and CSA S guidelines and odes is also disussed. Some reommendations that assist in a reasonable estimation of ultimate strength are also given. Finally, the effiieny of onfinement on the behavior is disussed. Abstrat Six full-sale onrete squat walls reinfored with glass-fiber-reinfored-polymer (GFRP) bars were tested to failure under quasi-stati reversed yli loading. The test parameters were the onfiguration of web reinforement (horizontal and/or vertial) and the horizontal web reinforement ratio. The test speimens experiened different mode of failures as a funtion of the web reinforement. The horizontal web reinforement was found to signifiantly inrease the ultimate strength as long as the failure was dominated by diagonal tension but had no signifiant effet if in exess of what was needed for flexural resistane. Both horizontal and vertial web reinforement was shown to be essential for rak reovery between load reversals and for ontrolling shear-rak width as well as for enhaning the onrete ontribution to the lateral shear resistane. The preditions based on FRP Canadian and Amerian design odes and guidelines were investigated and ompared to the test results. The results demonstrated that, in the Canadian ode, the onrete ontribution to the shear resistane should be onsidered and, in the Amerian design guidelines, the 45 shear-rakangle assumption should be modified for squat walls with different properties and should onsider the inrease in onrete shear ontribution after initiation of the first shear rak. The onfinement at boundary elements has been shown to signifiantly improve ultimate flexural strength that should not be omitted in the design. Keywords: Conrete; GFRP bars; squat walls; shear strength; web reinforement; seismi

105 88 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls 5.1. Introdution A onsiderable amount of work has been devoted to understanding the behavior of steel RC onrete squat walls (height-to-length ratio 2.0). The behavior of reinfored squat walls is different from that of slender walls due to their relatively larger magnitude of shearing and normal stresses. Test investigations have revealed that, by the onset of flexural reinforement yielding, shear deformations either shear distortion and/or sliding are ativated and loalized along the yielding zone and then begin to dominate the behavior, ausing rapid strength and stiffness degradation with subsequent premature shear failure (Paulay et al. 1982; Saatioglu 1991; Sittipunt et al. 2001). There is no onsensus among researhers about the influene of web reinforement in squat walls on shear strength. Some researhers have reported that using proper amount of horizontal reinforement restrained the diagonal tension, thereby inreasing the shear strength (Beekhuis 1971; Pilakoutas and Elnashai 1995; Hidalgo et al. 2002). In ontrast, other experiments have shown that horizontal web reinforement has no impat, whereas the shear strength signifiantly inreased as a funtion of vertial web reinforement (Barda et al. 1977; Lefas et al. 1990; Emamy Farvashany et al. 2008). The methods in ACI 18 (2014) and CSA A2. (2014) for estimating the shear strength of squat walls only onsider the amount of horizontal reinforement. Nevertheless, both odes reognize that vertial web reinforement is essential to maintain the equilibrium of internal fores. Both odes also require that minimum horizontal and vertial web reinforement should be provided to ontrol rak propagation and width. In pratie, squat walls are being used in low-rise strutures suh as parking garages and overpass bridges, whih are exposed to severe environmental onditions in northern limates that ause the orrosion of steel reinforement. The use of glass-fiber-reinfored-polymer (GFRP) bars as a viable alternative reinforing material has grown to obviate orrosion issues while providing an aeptable level of performane [ACI 440R (2007), ACI 440.1R (2015)]. Mohamed et al. (2014a, b) investigated the appliability of using GFRP as internal reinforement for earthquake-resistant systems suh as mid-rise shear walls. The test results demonstrated the potential of GFRP reinforement for distributing shear deformations along

106 89 the wall height, owing to its elasti nature, resulting in ontrolled shear distortion relative to the steel RC wall. This result motivated a new study to evaluate the feasibility of using GFRP bars in squat walls, in whih these problems are dominant. As a part of the ongoing experimental program, two squat walls with a height-to-length ratio of 1. were tested: one was reinfored with steel bars, the other with GFRP bars (Arafa et al. 2016b). The test results learly showed the stable behavior of the GFRP RC squat wall through its hystereti response, sine it evidened no strength degradation or signs of premature shear failure ompared to the steel RC one. The results also demonstrated that the attained drift ratio satisfied the limitation in most building odes. Nevertheless, FRP has not been adopted yet by the relevant design odes and guidelines [CSA S806 (2012), ACI 440.1R (2015)] as internal reinforement for squat walls under seismi loads. This paper aimed at experimentally assessing the impat of web reinforement on the response of squat walls totally reinfored with GFRP bars under quasi-stati reversed yli loading. The experimental results were analyzed onsidering the rak pattern, mode of failure, drift apaity, ultimate strength, and load top-displaement hystereti response. The distribution of strains in either the vertial or horizontal diretion is doumented. The effet of horizontal and vertial web reinforement on the onrete shear resistane is disussed. Evaluation of the ultimate apaity aording to the ACI and CSA odes was also introdued Experimental Program Desription of Test Speimens A total of six large-sale retangular onrete squat walls entirely reinfored with GFRP bars were onstruted and tested in the Strutural Laboratory in the Department of Civil Engineering at the University of Sherbrooke. Eah test speimen measured 200 mm thik, 1500 mm long, and 2000 mm high. The wall thikness satisfied the CSA A2. (2014) minimum-thikness requirement in Clause Eah speimen was ast vertially to reprodue onstrution pratie, with an integral mm heavily reinfored foundation funtioning as anhorage for the vertial reinforement and to fasten the speimen to the laboratory floor. Figure 5.1 provides the onrete dimensions and reinforement details.

107 90 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls 2000 mm 1500 mm 200 mm 8 No. 10 GFRP Vl. boundary Vertial web rein. See Se A-A Horizontal web rein. See Se A-A No A A Θ = mm 1200 mm a) Conrete dimensions b) Reinforement onfiguration No.1 or No.19 Hz S No.10 Vl. No.10 GFRP ties 8 No.10 Vl. 80 mm GFRP boundary No.10 Vl. 120 No.10 GFRP ties 8 No.10 Vl. GFRP boundary No ) Se A-A for speimens G4-250, G4-160, G4-80, and G6-80 No. 1@S= 250 mm for G4-250 No. 1@S= 160 mm for G4-160 No. 1@S= 80 mm for G4-80 No. 19@S= 80 mm for G6-80 All dimensions in mm No.1 Hz No d) Se A-A for speimen G-V No.10 GFRP 80 mm 1500 # 100 e) Se A-A for speimen G-H No.10 GFRP Vl. boundary 200 Figure 5.1 Conrete dimensions and details of reinforement Two boundary elements of equal width and breadth ( mm) were plaed at eah end of the horizontal length. The longitudinal and transverse reinforement ratios at the boundary elements were kept onstant in all speimens: 1.4% and 0.89%, respetively. The longitudinal reinforement onsisted of 8 No. 10 (#) GFRP bars laterally tied against premature bukling with transverse reinforement onsisting of No. 10 (#) spiral GFRP ties spaed at 80 mm along the wall height. The experimental parameters were the effet of the horizontal and vertial web reinforement. The fous; however, was on the horizontal web reinforement sine the urrent odes for steel RC squat walls [ACI 18 (2014) and CSA

108 91 A2. (2014)] alulates the shear apaity of squat walls based on the horizontal web reinforement. Four speimens G4-250, G4-160, G4-80, and G6-80 were reinfored with different horizontal web reinforement ratios; 0.51%, 0.79%, and 1.58%, and.58%, respetively, using No. 1 (#) GFRP bars spaed at 250, 160, and 80 mm or No. 19 (#6) GFRP bars spaed at 80 mm, respetively. The vertial web reinforement was kept onstant and omprised No. 10 (#) GFRP bars spaed at 120 mm with a reinforement ratio of 0.59%. The two remaining speimens were onstruted with either vertial or horizontal web reinforement (G-V, and G-H, respetively) to test the absene of horizontal or vertial web reinforement on wall behavior. G-V was reinfored with vertial web reinforement idential to that used in the four speimens, while G-H was reinfored with horizontal web reinforement idential to that used in G The sliding shear was prohibited by adding one layer of bidiagonal No. 10 (#) GFRP bars aross the potential sliding plane at an angle of 45 spaed at 100 mm. All reinforement rossing the wall base joint were anhored to the base with a development length in ompliane with the requirements of CSA S806 (2012) multiplied by a fator of 1.25 to aount for the effet of ompression and tension yles as suggested by Mohamed et al. (2014a). Table 5.1 gives the test matrix and reinforement details of the wall speimens. Table 5.1 Conrete strength and reinforement details Test No. Wall ID f' Reinforement Ratio (%) (MPa) ρl ρt ρv ρh 1 G G G G G-V G-H Notes: f' = onrete ompressive strength; ρl = boundary longitudinal-bar reinforement ratio; ρt = boundary-tie reinforement ratio; ρv = web vertialbar reinforement ratio; ρh = horizontal web reinforement ratio.

109 92 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls Material Properties The test speimens were made with normal-weight, ready-mixed onrete with a target 28-day ompressive strength of 40 MPa. Table 5.1 provides the atual onrete ompressive strength (f') based on the average of at least three mm ylinders for eah onrete bath tested on the day of wall testing. The longitudinal GFRP reinforing bars were sand-oated bars: No. 10 (#) was used for longitudinal bars, either in the boundary or in the web (ffu = 172 MPa, Ef = 65 GPa, εfu = 2.1%, Af = 71 mm 2 ) and spiral ties (for straight portions: ffu = 1065 MPa, Ef =50 GPa, εfu = 2.1 %, Af = 71 mm 2 ; for bent portions: ffu = 460 MPa). Two diameters were used as horizontal web reinforement: No. 1 (#4) (for straight portions: ffu = 1020 MPa, Ef =50 GPa, εfu = 2%, Af = 127 mm 2 ; for bent portions: ffu = 459 MPa), and No. 19 (#6) (for straight portions: ffu = 1028 MPa, Ef =50 GPa, εfu = 2%, Af = 285 mm 2 ; for bent portions: ffu = 46 MPa). The horizontal reinforement in the walls had 90 end hooks. The tensile properties of the straight GFRP bars were speified based on testing five speimens aording to ASTM D7205/D7205M (2006). The B.5 test method stipulated in ACI 440.R (2004) was used to determine the tensile properties of the bent bars. The reported tensile properties of the GFRP bars were alulated using nominal ross-setional areas Test Setup and Proedure Figure 5.2 shows the layout of the test setup. All speimens were tested laterally as a vertial antilever with a fore applied through a rigid steel loading beam, designed to transfer lateral loads aross the top of the wall. The lateral yli loading was applied at 2550 mm above the base of the wall using a 1000 kn MTS hydrauli atuator with a maximum stroke of ±250 mm. The base of eah speimen was lamped down to the strong laboratory foundations through four prestressing high-strength steel rods 66 mm in diameter to ahieve full fixation at the base level. Out-of-plane supports were pinned to the wall at the level of the transfer steel beam to prevent out-of-plane movement. No axial load was applied to the test speimens sine axial stresses in squat walls are typially insignifiant when proportioned to the produt of onrete ompressive strength and gross area.

110 9 Atuator Lateral braing Loading beam 2550 mm Pre-stressing steel rods Squat-wall speimen Base Figure 5.2 Test setup The loading was applied by gradual inreasing in the displaement of the wall tip under a quasi-stati rate of 0.01 Hz. The loading history started by applying two idential displaement yles with inrements of ±2 mm up to 10 mm (0.5% lateral drift), followed by inrements of ±5 mm up to 50 mm (2.5% lateral drift), and thereafter inrements of ±10 up to failure. Figure 5. shows a typial sequene of displaement yles adopted in this study. Lateral drift (%) Cyle number Figure 5. Displaement history Lateral displaement (mm)

111 94 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls Instrumentation The deformation response was reorded with a series of linear variable displaement transduers (LVDTs) and strain gauges. Only the instruments used in this paper are reported on (Figure 5.4a). One LVDT was installed at the wall tip to measure the lateral displaement. One LVDT was installed at the onstrution joint to monitor sliding between the wall and base. One LVDT was used to measure the sliding between the base and laboratory rigid floor. Two LVDTs were mounted lose to the boundary elements to measure the onrete strain. For the sake of safety, two additional LVDTs were attahed to the upper steel beam to measure the top out-of-plane deformation. The rak width was also measured with two LVDTs mounted at the first two major raks. Figure 5.4a shows the layout of the LVDTs instrumentation. A total of 14 eletrial resistane strain gauges were mounted on the reinforing bars (boundary and web reinforement) at ritial loations to measure strains (Figure 5.4b). Five bars were instrumented to measure the vertial strain distribution at the wall base: two longitudinal bars in the boundary elements in the extreme ompression and tension fibers and three vertial bars in the web. Three other horizontal bars loated at three height levels 200 mm, 750 mm, and 1500 mm were seleted to measure the shear-strain distribution. Eah bar was instrumented at the left, right and middle. Lateral top displaement Base-floor Wall-base Out-of-plane displaement Conrete strain Figure 5.4 Instrumentation: (a) LVDTs instrumentation; (b) strain-gauge instrumentation

112 Experimental Results and Disussion Failure Progression and Hystereti Response This setion introdues the test results of the investigated walls in referene to hystereti response and damage progression. The damage progressions involved the propagation of flexural, and shear raks, over splitting and spalling, failure mode, and the failure-assoiated damage. Table 5.2 summarizes the test results. Figure 5.5 depits the rak pattern. Figure 5.6 gives the experimentally reorded hystereti response for the lateral load at eah loading step against lateral top displaement. Figures 5.7 and 5.8 provide the final failure mode and damage-assoiated failure, respetively. Wall ID Initial Flexural Crak P d (%) (kn) Table 5.2 Summary of the test results Initial Shear Crak P (kn) d (%) Conrete Cover Splitting P (kn) d (%) Exessive Cover Spalling d P (kn) (%) Peak Capaity P (kn) d (%) Failure mode Failure P (kn) G DT G FT G FC G FC G-V DT G-H FT Notes: P = applied lateral load; d = drift ratio; DT = diagonal tension; FT = flexural tension; FC = flexural ompression In general, the initial behavior of all walls was haraterized by a flexural response as evidened by the typial amount of horizontal flexural raks propagated in the lower part of the walls at a lateral load within the range of 19 to 168 kn (Table 5.2). As a result of the flexural raks, all of the speimens exhibited a redution in lateral stiffness (Figure 5.6). At a load level of 169 to 27 kn (orresponding to 0.6% 0.44% lateral drift; Table 5.2), the flexural raks tended to inline down toward the entral zone of the web owing to the effet of shear stresses. As larger deformations were imposed, horizontal raks were aompanied by inlined raks and ontinued to initiate, forming a fan shape sine the raks varied in inlination as they approahed the wall base. This variation is attributed to moment gradient d (%)

113 96 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls along the wall height, whih affeted the orientation of the priniple stresses defining shearrak diretion. As the load inreased, new shear raks tended to propagate near the top of the wall, while the existing raks tended to widen and extend downward to the opposite boundary element with inreased inlination and interseted the raks originating from the other diretion. Clearly due to the existene of boundary elements with heavy longitudinal and transverse reinforement, the inlination and width of the propagated shear raks signifiantly dereased at the ends of the walls, suggesting that the longitudinal reinforement at the boundary elements and the onfinement ontributed to wall shear resistane. This is onsistent with test results for squat walls reinfored with onventional steel (Luna et al. 2015). θ=58 o (a) G4-250 (b) G4-160 (b) G4-80 θ=59 o () G6-80 (e) G-V (f) G-H Figure 5.5 Crak pattern (note that the indiated loations are the positions of the maximum measured shear raks) As new raks developed and existing ones extended, a gradual degradation in overall stiffness ourred as refleted in the hystereti response. Furthermore, the unloading/reloading hystereti urves seem to demonstrate linearity owing to the linear nature of the GFRP bars (Figure 5.6). Exept for G-H, as loading inreased, vertial splitting raks ourred at the most ompression zone at a lateral drift ranging from 1.2% to 1.5% and were assoiated with the initiation of over spalling (Table 5.2). After this stage, however, eah speimen experiened a different failure mode as illustrated in the following setions.

114 Drift ratio (%) Drift ratio (%) 0 1 Load (kn) Cover spalling Shear failure 60 2 G6-80 Cover splitting Cover spalling Conrete deterioration Conrete rushing Cover splitting Lateral top displaement (mm) Lateral top displaement (mm) -60 (d) G-V -60 (e) Cover spalling Lateral top displaement (mm) Flexural bars rupture - Cover spalling Conrete deterioration Conrete rushing 2 Cover splitting (b) Cover splitting () Drift ratio (%) 0 1 G G Lateral top displaement (mm) -1 Drift ratio (%) (f) 2 G-H Load (kn) Load (kn) Lateral top displaement (mm) Drift ratio (%) Load (kn) Load (kn) Shear failure Cover spalling - Load (kn) Cover splitting (a) G Drift ratio (%) Major horizontal rak Cover splitting Flexural bars rupture Lateral top displaement (mm) Figure 5.6 Lateral load versus top displaement

115 98 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls Wall G4-250 This wall experiened shear failure along a major diagonal rak forming along a distane of 1250 mm from the base and at 58 ο angle to the wall s longitudinal axis (Figure 5.7a). As the shear failure approahed, diagonal raks beame substantially wider, providing sign of shear distress. Failure then ourred suddenly at a lateral load of 678 kn, orresponding to a lateral drift of 2.65% (Figure 5.6a). While the bent portion of some of the horizontal web reinforement was straightened (Figure 5.8a), the ultimate strain orresponding to rupture of the bent portion was almost ahieved, as will be disussed below. Having the tail length extend beyond the bent portion is reommended to eliminate suh failure-assoiated behavior. Wall G4-160 The speimen experiened sudden flexural rupture in the longitudinal bars at the boundary element under tension (Figures 5.7b and 5.8b). The failure was brittle and assoiated with an explosive sound. The maximum lateral load ahieved was 708 kn, orresponding to a lateral drift of 2.8% (Figure 5.6b). Walls G4-80 and G6-80 The failure of speimens G4-80 and G6-80 was identified as flexural ompression failure and preeded by ample warning, starting with extensive deterioration of the onrete at the ompression zone (at a lateral drift of 2.75% and 2.9% for G4-80 and G46-80, respetively). Noise generated from this zone was aompanied by a gradual degradation in lateral strength (912 and 95 kn, respetively) (Figures 5.6 and 5.6d). Ultimately, onrete rushing (Figures 5.7 and d), assoiated with rupture in the GFRP ties (Figure 5.8), was prominently evident, ausing wall failure and a drop in lateral strength, followed by the sequential fraturing of the ompressed GFRP bars in the boundary element (Figure 5.8). The ultimate attained drifts were % and.1% for G4-80 and G6-80, respetively (Figures 5.6 and 5.6d). It should be mentioned that no sign of shear distress was observed up to failure. This indiates that the GFRP web reinforement was adequate in resisting diagonal tension.

116 99 Shear θ=58o Conrete Bars rupture (a) G4-250 () G4-80 (b) G4-160 Shear Bars rupture θ=59o Conrete rushing (d) G6-80 (e) G-V (f) G-H Figure 5.7 Failure modes: (a) G4-250; (b) G4-160; () G4-80; (d) G6-80; (e) G-V; (f) G-H End hooks straightened (b) G4-160 (a) G4-250 Frature of vertial bar dowel Rupture of GFRP ties Bar rupture Rupture of GFRP ties rossing shear rak (d) G-V Frature of the ompressed () G4-80 and G6-80 Major horizontal raks above sliding.reinforement (e) G-H Figure 5.8 Damage aspets Bar rupture

117 100 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls Wall G-V This speimen ahieved its ultimate strength (98 kn) at a lateral drift of 1.5%, and then experiened pronouned degradation owing to the deterioration of shear-resisting omponents. The failure ourred at a lateral drift of 2.2% by sliding along a major diagonal shear rak (Figure 5.7e). This diagonal shear rak spread along a distane of 100 mm from the base at an angle of 59 with the speimen s longitudinal axis and measured 7.20 mm wide before failure. The failure was assoiated with the progressive frature of the vertial web bars and fraturing of the GFRP ties at the boundary element rossing the shear rak (Figure 5.8d). This observation gives evidene that the reinforement in the boundary element ontributed to the wall s shear resistane. Wall G-H Due to the absene of vertial web reinforement in this speimen, a notieable major horizontal rak above sliding shear reinforement going through the entire web length was observed at a lateral drift of 1%. With load reversal, the horizontal rak remained open along the web zone even under ompression stresses of bending moment while it losed at the boundary zone due to longitudinal reinforement (Figure 5.8e). This observation reflets the neessity of using vertial web reinforement for rak reovery between load reversals. The speimen failed suddenly due to tensile rupture of the longitudinal bars at the boundary element above sliding reinforement (Figure 5.7f, Figure 5.8e) at 2.4% lateral drift, orresponding to a lateral strength of 482 kn (Figure 5.6f). It is worth mentioning that, even though the wall failed due to flexural rupture, there was substantial widening of the shear raks and shear failure was imminent as the bar rupture took plae. It should also be noted that, due to the additional flexural resistane provided by the bidiagonal sliding shear reinforement, flexural rupture of the longitudinal bars was shifted away from the base wall interfae where the bending moment is maximum. In general, as presented in Table 5.2, web reinforement (horizontal or vertial) had no signifiant effet on the first flexural and shear raking loads. The initial raks were a primary funtion of the onrete splitting stress, whih depends on onrete ompressive strength; the minor differene observed ould be attributed to the differene in onrete

118 101 strength. Figure 5.5 demonstrates that all the speimens exhibited similar tendenies with respet to the propagated flexural shear raks. The main differene, however, was the number of propagated raks. Using horizontal or vertial web reinforement or inreasing the horizontal web reinforement ratio appeared to result in more raks. This ould be attributed to two main reasons: (1) the higher ultimate load in speimens with higher web reinforement ratios resulted in more raks; and (2) the fat that the web reinforement often works as a rak initiator and onsequently influenes the shear-rak spaing. This is in agreement with experiments onduted by Barda et al. (1977) and Luna et al. (2015) on steel RC squat walls. It is worth mentioning that no sign of premature sliding or anhorage failure was observed in the speimens. This indiates that FRP bars are adequate in resisting the applied sliding fore and that the anhorage length was suffiient to transmit the wall fores to the base under the reversed yli loadings Load Top Displaement Envelope Curves Figure 5.9 presents the load top-displaement envelope urves for the test speimens. Generally, all the speimens had omparable levels of initial stiffness up to initiation of the first flexural rak. After that, the speimens experiened redued lateral stiffness and represented a raked speimen with redued moment of inertia. Different trends were observed as a funtion of the web reinforement. The higher stiffness an be attributed to either the horizontal or vertial web reinforement. Consequently, at the same load level, less lateral displaement was evidened in walls with both horizontal and vertial web reinforement. G-V exhibited a progressive deterioration in lateral stiffness as shear raks formation and extension and the assoiated dramati shear deformations. This behavior, however, was signifiantly enhaned by the horizontal web reinforement in G4-250, whih outperformed its ounterpart, G-V, by approximately 70% and 16% with respet to ultimate apaity and drift ratio, respetively. Hene, it an be onluded that horizontal web reinforement had a diret impat in inreasing the ultimate shear strength of the squat walls when diagonal tension shear failure was dominant. On the other hand, the lower stiffness of G- H relative to its ounterpart G4-250 is attributed to the absene of vertial web reinforement

119 102 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls with redued axial stiffness. It would be desirable, if the failure of G-H was pure shear, to judge the effetiveness of vertial web reinforement on wall ultimate shear strength. Sine wall strength was generally ontrolled by flexure, albeit the shear distress was signifiant, it is diffiult to judge this effetiveness. More investigation is needed to larify this point. Tentatively; however, the measured shear rak widths and evaluation of the onrete ontribution to shear resistane will be used later to give an idea of their influene on behavior. Load (kn) Drift ratio (%) Major horizontal rak Cover splitting Exessive over spalling Conrete deterioration Flexural bar rupture Conrete rushing Shear failure G4-250 G G-V G-H Lateral top displaement (mm) G4-80 G6-80 Figure 5.9 Load top displaement envelope urves Figure 5.9 also reveals that suitable horizontal web reinforement an hange the failure mode from shear to flexural while ahieving higher ultimate strength and drift ratio with subsequently higher deformability. G4-80 shows this learly, exhibiting flexural ompression failure while outperforming its ounterpart G4-250 by approximately 5% and 18% with respet to ultimate strength and drift ratio, respetively. Furthermore, while the horizontal web reinforement in G6-80 was 125% higher than that in G4-80, both speimens exhibited similar levels of ultimate strength and drift ratio (the differene did not exeed %). This behavior reveals that using horizontal web reinforement in exess of the amount required for resisting the ultimate flexural apaity had no impat on either ultimate strength or drift ratio.

120 Strains in Vertial Reinforement Figure 5.10 presents the typial vertial strain distributions along the base of the wall length at different drift levels. The values at the same drift level are onneted with a solid line in all speimens exept for G-H, in whih the strains were reorded at only two loations in the boundary elements; these values are onneted by a dashed line. It an be seen that the vertial strain varied almost linearly along the wall length at small drift levels (0.25%). After this stage, the distribution exhibited modest variations at the boundary zone either in tension or ompression. This data are useful in evaluating the existing widely used models whih are based on plane setion assumptions. Strain (με) Strain (με) G % G % 0.50% 0.50% % 0.75% % 1.00% % 1.50% Distane along wall (mm) Distane along wall (mm) G % G % % 0.50% % 0.75% 1.00% % 1.50% % Distane along wall (mm) Distane along wall (mm) Strain (με) Strain (με) Figure 5.10 Vertial-strain distribution along the wall length

121 104 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls Strain (με) G-V 0.25% 0.50% 0.75% 1.00% 1.50% Distane along wall (mm) Strain (με) G-H 0.25% 0.50% 0.75% Distane along wall (mm) Figure 5.10 Vertial-strain distribution along the wall length (ontinued) Figure 5.11 shows the envelope urves for the applied load versus the maximum measured strain in the vertial reinforement, whih is typially reorded in the first longitudinal bar near the extreme tension fibers. Initially, before raking, all of the speimens exhibited negligible strain, sine all stresses are arried by the onrete. By the onset of the flexural raks, however, a portion of the stress was transferred to the longitudinal reinforement, resulting in a sudden inrease in the measured strain in all of the speimens. After this stage, the strain in all speimens inreased almost proportionally with inreasing load due to the elasti nature of the GFRP bars. The horizontal web reinforement appeared to have no impat on the longitudinal strain readings, sine the speimens with different horizontal web reinforement ratios exhibited omparable strains at all loading levels. As expeted, however, the absene of vertial web reinforement in G-H resulted in higher longitudinal strain relative to its ounterpart speimens at the same load level. This is onsistent with the previously reported lower stiffness of this speimen. It should be noted that, for the entire test speimens exept G-V, all strain gauges on the longitudinal reinforement halted at a lateral load ranging from 60% to 70% of the ultimate strength. Therefore, the speimens exhibited higher strain than that plotted in Fig. 8. In the ase of G-V, however, the strain gauges worked properly up to the ultimate strength, at whih point the maximum measured strain was (μɛ).

122 105 Load (kn) G4-250 G4-160 G4-80 G6-80 G-V G-H Strain (με) Figure 5.11 Maximum measured vertial strain (με) Strains in Horizontal Reinforement Figure 5.12 shows the strain distribution in horizontal bars along the wall height. Note that these values represent the maximum reorded strain along eah instrumented horizontal bar. The plot demonstrates that the maximum measured strain was reorded at a level almost equal to 1/ of the wall height. The maximum measured strain at failure was 980, 6650, 4840, 2670, and 6820 (με) for G4-250, G4-160, G4-80, G6-80, and G-H, respetively. The lower measured strain in G4-80 and G6-80 an be attributed to flexural failure with no sign of shear distress owing to the high horizontal web reinforement provided in these speimens. It should be noted that the maximum measured strain in G4-250, whih experiened shear failure assoiated with straightening of the end hooks, orresponds to 49% of the straight-bar rupture strain. That value is onsistent with the ACI 440.1R (2015) guidelines, whih antiipate failure at the bent portion at 50% of the straight-bar strength. Therefore, it an be onluded that the speimen nearly ahieved its ultimate shear apaity. Apparently, due to the elasti nature of GFRP bars, shear strain was distributed along the wall height and inreased progressively with drift levels, supporting the finding of Mohamed et al. (2014b). In ontrast, due to the effet of steel yielding in steel RC squat walls, the shear strain and deformations was loalized at the plasti-hinge zone and subsequently aelerated shear ompression failure (Sittipunt et al. 2001).

123 106 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls G4-250 G4-160 G4-80 Wall height (mm) Wall height (mm) Wall height (mm) Strain (με) Strain (με) Strain (με) 2000 G6-80 G-H Wall height (mm) Wall height (mm) % 1.00% 1.50% 2.00% Failure Strain (με) Strain (με) Figure 5.12 Horizontal-strain distribution along the wall height Figure 5.1 shows the envelop urves for lateral applied load versus the average horizontal strain for all speimens. As shown, all speimens exhibited similar initial behavior before shear raking initiated, at whih point insignifiant shear strain was reorded. As the load inreased and the first shear rak appeared, the strain in the horizontal web reinforement suddenly inreased, sine a part of shear stresses was transmitted to it from the onrete. With further loading, the shear strain progressively inreased, although at different rates depending on the parameters investigated. G-H exhibited higher strain than its ounterpart G4-250, in spite of both having the same horizontal web reinforement ratio. This is due to the absene of vertial web reinforement in G-H, whih indued a shallower ompression zone, wider raks, and smaller dowel ation, onsequently, reduing the onrete s ontribution to the

124 107 lateral shear-resistane mehanism. This redution inreased the shear load arried by the horizontal web reinforement, so that higher shear strain was reorded. This indiates that vertial web reinforement impated shear resistane by enhaning the onrete s shear resistane G6-80 G4-80 Load (kn) G4-160 G-H G Strain (με) Figure 5.1 Load average horizontal strain envelope urves Shear-Crak Width As shown in Figure 5.14, the presene of either horizontal or vertial web reinforement had a lear effet on ontrolling the shear-rak width, espeially in the ase of horizontal web reinforement. For instane, G-V and G4-250 differ in that the latter had a horizontal web reinforement ratio of 0.51%. A omparison of the rak width at the same load level in eah speimen revealed the shear-rak width of G4-250 dereased by 89%. Similarly, the differene between G-H and G4-250 is a vertial web reinforement ratio of 0.59%, yielding a derease in rak width of 41% at the same load level. A possible explanation for this ould be that the vertial web reinforement ontributes to the flexural-resistane mehanism, whih would expose them to high strains. This, in turn, would make them less effetive in resisting the transverse tensile stresses between raks; whih limits rak width. This is lear in Figure 5.15, whih shows the hystereti relationship for the measured vertial strain versus horizontal

125 108 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls strain at the same loation in G4-250 as an example. The figure learly demonstrates that the measured strain in the horizontal reinforement was signifiantly lower than that in the vertial reinforement at all load levels. Higher horizontal web reinforement (Figure 5.14) provided better ontrol over the shearrak width: the greater the horizontal web reinforement, the lower the reorded shear-rak width at the same load level. This is onsistent with the lower measured shear strain in speimens with higher horizontal web reinforement ratios. Load (kn) G6-80 G4-80 G4-160 G4-250 G-H G-V Crak width (mm) Figure 5.14 Load shear-rak-width envelope urves 8000 Vertial strain (με) Horizontal strain (με) Figure 5.15 Vertial strain versus horizontal strain at the same loation (G4-250)

126 109 The urrent version of ASCE/SEI 41 (201) suggests that shear raks to be limited to 1.60 mm to preserve the struture s integrity, making it suitable for immediate oupany after an earthquake. As mentioned earlier, the behavior of GFRP RC squat walls proved their effiieny in low-to-moderate earthquake zones where the building remains safe for oupation after an earthquake event. Therefore, it is of interest to determine how this limit ould be met in GFRP RC squat walls. Sine the shear-rak width orrelates losely to horizontal web reinforement rather than vertial web reinforement and sine the minimum horizontal and vertial web reinforement are usually provided in squat walls, the measured shear rak width was plotted against the maximum measured strain in the horizontal web reinforement (Figure 5.16) in G4-250, G4-160, G4-80, and G6-80, whih ontained both horizontal and vertial web reinforement. Generally, Figure 5.16 shows that the reinforement ratio had a negligible effet on the relation between shear-rak width and the reorded strain in horizontal web reinforement. The figure also indiates that a shear-rak width of 1.6 mm an be met by limiting horizontal strain to 5450 με, onservatively an be taken equal to 5000 με Strain (με) G4-250 G4-160 G4-80 G Shear-rak width (mm) Figure 5.16 Maximum measured horizontal strain versus shear-rak width

127 110 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls Influene of Web Reinforement on Conrete Shear Resistane Some urrent odes for steel RC strutures, suh as ACI 18 (2014), assume that the onrete ontribution to wall shear strength remains unhanged after the first shear rak initiates, regardless of the amount or distribution of web reinforement. Therefore, it is interesting to examine the onrete ontribution after the first shear rak and determine whether this ontribution would be affeted by web reinforement. In doing so, the onrete ontribution was alulated by subtrating the ontribution of the horizontal web reinforement from the total applied shear fore [ACI 18 (2014) and CSA A2. (2014)]. The ontribution of the horizontal web reinforement to the total shear resistane was alulated based on the truss analogy as follows (SI units): V sf Av f fvdv ot (5.1) s where; Vsf is the shear resistane provided by the horizontal web reinforement, Av is the area of horizontal web reinforement within the distane s, ffv is the stress in horizontal web reinforement and was alulated as the modulus of elastiity multiplied by the measured average horizontal strain shown in Figure 5.1, dv is the effetive shear depth equal to the greater of 0.9d or 0.72 lw but not less than 0.8 lw, θ is the measured shear rak angle, and s is the spaing of the horizontal web reinforement. It should be noted that, for speimens exhibiting flexural failure, θ was taken as an average measured angle of 58.5 for speimens that failed in shear (G4-250 and G-V). The onrete shear ontribution was alulated and plotted against top displaement in Figure 5.17a for all speimens. The test results indiate that the onrete ontribution was higher than that orresponding to the first shear rak at any load level. The onrete ontribution to shear at failure ranged from 51% to 110% higher than that at the onset of shear raking; the ratio is a funtion of web reinforement. Speimens with both horizontal and vertial web reinforement showed the best performane, while G-H performed the worst. The provision of horizontal reinforement (G-V ompared to G4-250) indued onfinement of the ross setion and ontrolled shear raks, while the provision of vertial web reinforement (G-H ompared to G4-250) resulted in a longer ompression zone and greater dowel ation, thereby inreasing

128 111 onrete shear strength. Figure 5.17a also demonstrates that the speimens with higher horizontal web reinforement ratios performed slightly better than those with lower reinforement ratios, whih an be attributed to the lower shear-rak width in the former. Overall, the results indiate the neessity of both horizontal and vertial web reinforement in squat walls to enhane their shear resistane. The results also reflet the exessive onservatism of the odes, whih assume the onrete shear ontribution orresponds to that at the first shear rak. Conrete-shear strength (kn) Drift ratio (%) G G4-160 G4-80 G G-V G-H (a) Top displaement (mm) Components of shear resistane (kn) 1000 (b) G4-250 G4-160 G4-80 G6-80 G-H V sf V f V sf Hz. bars V f Conrete Applied shear load (kn) Figure 5.17 (a) Conrete shear strength versus top-displaement envelope urves; (b) shearresistane omponents To further evaluate the onrete ontribution to the total wall apaity, the onrete and horizontal web reinforement ontributions were plotted against the applied shear fore in Figure 5.17b. Clearly, the onrete arried a substantial portion of the ultimate shear strength, whih ranged from 50% to 61%.

129 112 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls 5.4. Predition of Speimens Ultimate Strength The safety of squat walls reinfored with a relatively new material suh as GFRP having different harateristis than the onventional steel depends on the designer s ability to suessfully predit how suh strutures behave under seismi loading and reasonably prediting the ultimate strength and mode of failure. Whereas the foregoing setions shed the light on some key features of GFRP RC squat walls response under seismi loading in terms of ultimate strength, drift ratio, failure mode and the effiieny of web reinforement, the question still remains about prediting the ultimate strength and failure mode. Currently; however, no seismi provisions are available in the FRP design odes and guidelines [CSA S806 (2012); ACI 440.1R (2015)] due to the lak of experimental tests in FRP RC squat walls. Hene, in an attempt to predit the theoretial ultimate strength of the investigated walls, similar formulas that are being used the CSA A2. (2014) and the ACI 18 (2014) will be used onsidering the differenes between FRP and steel reinforement. A summary for these methods in flexural and shear strengths predition is given in the following setions. This is followed by the evaluation of suh equations by omparing their predition with the experimental results. It should be noted that material redution fators are intentionally omitted from the following ode expressions sine the speimen dimensions and material strengths are aurately estimated Flexural Strength Both ACI 18 (2014) and CSA A2. (2014) allow the use of plane setional analysis in prediting the flexural strength of squat walls. In implementing this method, we adopted the orresponding provisions in ACI 440.1R (2015) and CSA S806 (2012) pertaining to plane setional analysis. The main differene between them is the permissible stress and strain for the onrete ompression blok. ACI 440.1R (2015) limits the ompressive strain to 0.00 and the ompressive stress to 0.85 f', whih is assumed to be uniformly distributed over a distane a = β1 from the extreme ompressed fiber (where is the distane from the extreme ompression fiber to the neutral axis and β1 is taken as 0.85 for onrete strength (f ) up to 28

130 11 MPa). For strengths above 28 MPa, this fator is inrementally redued by 0.05 for eah 7 MPa of strength more than 28 MPa, but not taken as less than CSA S806 (2012), however, limits the ompressive strain to and the ompressive stress to (α1f'), uniformly distributed over a distane β1, where α1 and β1 are expressed as follows: ' ' f 0.67, f 0.67 (5.2) 1 1 It should be noted that the fators reported above are used when failure is initiated by onrete rushing (i.e., the onrete has attained its ultimate ompressive strain). If failure begins with the tensile bars rupturing, however, the fators are different sine onrete ompressive strain is less than ultimate. In this ase, CSA S806 (2012) provides urves to obtain α1 and β1 as a funtion of onrete ompressive strain, whih an be alulated from the strain ompatibility based on the ultimate strain of the FRP bars. This proess requires iteration until equilibrium is satisfied. This has been simplified in ACI 440.1R (2015), whih allows the use of the same as fators for ompression failure but equating the neutral-axis depth with the value of the balaned ase. The alulations in our study were made onsidering all the reinforement rossing the wall base, inluding the vertial omponent of the bidiagonal sliding reinforement. The ontribution of the GFRP bars in ompression was onsidered by assuming a ompressive strength equal to 50% of the tensile strength, while the modulus of elastiity was onsidered the same in tension and ompression (Deitz et al. 200) Shear Strength The ACI 18 (2014) shear estimation assumes a shear-rak angle of 45 aross the wall s shear depth and onsists of two superimposed shear-resisting omponents; onrete shear strength and horizontal web reinforement shear strength. The onrete shear ontribution is estimated as the shear apaity orresponding to the first shear rak propagated and was empirially obtained from experimental results, while the horizontal web reinforement ontribution is estimated by onsidering equilibrium of fores at a typial joint of a 45 truss model. In aordane with this onept, the onrete shear strength was estimated based on the

131 114 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls urrent test results. The shear stress orresponding to the first shear rak was alulated and formulated as a funtion of ' f aording to the format used in ACI 18 (2014). As shown in Table 5., the onrete shear strength ould be taken as equal to 0.12 ' f as an average value. Aording to ACI 18 (2014), in the ase of steel RC squat walls, the onrete shear strength is equal to 0.25 ' f. The lower average value measured in this study ould be attributed to the absene of axial loading. The horizontal web-reinforement ontribution was alulated based on the shear provisions in the ACI 440.1R (2015). Generally, owing to the unidiretional properties of the material, bending FRP to form stirrups substantially redues strength in the bent portion. The urrent pratie with FRP RC elements limits the ultimate stress in shear reinforement to avoid rupture at the bent portion and maintain the shear transfer via aggregate interlok. The first riterion is satisfied by limiting stress to the bent portion s ultimate strength, while the seond is satisfied by limiting the ultimate tensile strain. Considering the ACI 440.1R (2015) limitation for the ultimate stress in shear reinforement and using 0.12 ' f as onrete shear strength, the following equations were used: V sr ' Av fv Av(0.12 f ) (5.) b s where the first term is the onrete ontribution and the seond term is the horizontal webreinforement ontribution, Av is ross-setional area of the wall, f is the onrete ompressive strength, Av is total ross-setional area of the horizontal web reinforement on spaing s, b is the wall thikness, and fv is the permissible stress in the horizontal web reinforement, whih is alulated as follows: f v rb E fv ffrpbend, ffrpbend ( ) f fu f fu (5.4) d b where Efv is the modulus of elastiity of the horizontal web reinforement, ffrpbend is the bent portion s ultimate strength, rb is the internal bend radius, and db is the horizontal bars diameter.

132 115 CSA A2. (2014) provides a shear-design method based on the modified ompression-field theory (MCFT). In this theory, the angle of the diagonal shear rak is estimated as a funtion of the wall depth, longitudinal-reinforement axial rigidity, and the internal fores applied at the setion of interest. In ontrast to ACI 18 (2014), whih reognizes the shear ontribution of onrete in squat walls, the CSA A2. (2014) speifies that no dependene should be plaed on the onrete in ontributing towards squat-wall shear strength. Consequently, the shear fore should be resisted only by horizontal web reinforement. The CSA A2. (2014) provision is justified by some experimental tests that revealed the shear strength of squat wall is adversely affeted by horizontal displaement yles. Another differene relates to effetive shear depth: ACI 18 (2014) assumes that the entire length is effetive in shear resistane, while CSA A2. (2014) defines the effetive shear depth (dv) as the greater of 0.9d (d is effetive flexural depth) or 0.8 lw (lw is wall length). The shear strength of the test speimens was estimated based on the horizontal webreinforement ontribution aording to CSA A2. (2014); the CSA S806 (2012) limitations for the allowable stress in FRP shear reinforement were onsidered as follows: V sr Av fvdv ot Vsf (5.5) s M f dv V f fv 0.005E fv ffrpbend 0. 4 f fu, l 60, l (5.6) 2E A where Mf and Vf are, respetively, the moment and shear applied at the ritial setion of shear taken at the wall base; Ef and Af are the elasti modulus and the total ross-setional area of longitudinal tension reinforement at the same setion; and the other terms as desribed above Comparison of Predited Ultimate Strength to Test Results Table 5. provides the alulated ultimate flexural and shear strength for the test speimens. The lowest value of the alulated flexural and shear strengths was onsidered as the ultimate strength orresponding to failure and ompared with the test results in the same table. It should be noted that material redution fators were taken equal to unity. Generally, ACI 440.1R (2015) predited the same failure mode as the experimental results. Moreover, the f f

133 116 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls used equations yielded onservative estimations, although with different auraies in terms of failure mode. The ultimate shear strength of G4-250, whih failed in shear, was underestimated by almost 24%. While this level of onservatism might be reasonable in the design, the ACI 440.1R (2015) shear estimation method still has some shortomings that an be summarized as follows: 1. The assumption of a 45 shear-rak angle appears to be inappropriate sine the angle measured was greater than 45 ; the angle measured in G4-250 was 58 (Figure 5.7). This indiates that the horizontal web-reinforement ontribution to shear resistane was overestimated by approximately 60%. Although this shortoming did not affet onservatism in this study, whih was attributed to other shortomings, it might lead to unsafe preditions in some ases, given the fat that the shear-rak angle in FRP RC elements an signifiantly deviate from 45 and, in some ases, be as muh as 70 (Razaqpur and Spadea. 2015). 2. It is assumed that the onrete ontribution is equal to the shear strength when the first shear rak is initiated. As mentioned, however, the onrete ontribution was signifiantly higher than that (Figure 5.17b). This shortoming is lear in prediting the ultimate strength of G-V, whih is underestimated by 44%.. Exessive underestimation of the shear strain. Whereas the ultimate shear strain was limited to 4000 με, the test results showed that G4-250 attained an average horizontal strain equal to 7710 με (Figure 5.1) whih is almost double the speified value. 4. The method does not aount for reinforement at the boundary elements, whih demonstrated their effetiveness in ontrolling shear-rak width and inlination at the wall edges. Moreover, ACI 440.1R (2015) neglets the effet of other parameters that might be ruial to behavior suh as vertial web reinforement, aspet ratio, and axial loading. More investigation is required to larify their effets. Table 5. also shows that ACI 440.1R (2015) underestimated the ultimate flexural apaity of G4-80 and G6-80 by almost 6%, as it does not aount for the onfinement effet, whih will be disussed later.

134 117 Wall Table 5. Ultimate strength predition Experimental ACI 440.1R-15 CSA S CSA S * Vr V exp ' A f (kn) FM V f V s V pred F.M V f V s =V sf θ V pred (kn) (kn) (kn) (kn) (kn) (kn) F.M V * * V sf V θ s V pred F.M (kn) (kn) (kn) (kn) G DT DT DT DT G FC FC DT FC G FC FC FC FC G-V DT DT DT DT G-H FT FT DT FT Notes: V r = shear raking load; A = gross setional area; V exp = experimental ultimate strength; FM = failure mode; DT = diagonal tension failure; FT = flexural tension failure; FC = flexural ompression failure; V f = flexural strength; V s = shear strength against diagonal tension; θ = angle between the shear rak and the longitudinal axis of the wall; V pred = the lower of V f and V s ; the notation * represents the predition after onsidering the onrete ontribution to shear resistane (Eq. 6). The defiieny of the 45 shear rak angle assumption was retified in CSA S806 (2012) based on the MCFT alulation method that produed a reasonable estimation of the shearrak angle of 56, ompared to an average experimental value of 58 in G However, due to omitting the onrete ontribution from the wall shear resistane and limiting shear strain (5000 με), CSA S806 (2012) unduly underestimated the ultimate shear strength. For instane, the ultimate shear strength for G4-250 was underestimated by 70% and G-V was expeted to arry a load equal to zero, whih is ontrary to the test results. Furthermore, as result of exessive onservatism, CSA S806 (2012) predited pure shear failure for G4-80 and G-H, although the former experiened flexural ompressive failure and the latter flexural tensile failure. Negleting the onrete omponent in resisting shear would be reasonable in squat walls reinfored with steel in whih the shear raks have not properly losed by the onset of plasti deformations, aordingly, by yling they expose to abrasive rubbing along their asperities. This phenomenon takes part in deteriorating the aggregate-interlok shear-resistane mehanism, whih depends on the roughness of rak asperities. This, however, was not the ase with the GFRP RC walls, in whih the raks realigned and losed between load reversals due to the elasti nature of GFRP bars. Therefore, negleting the onrete ontribution appears to be inappropriate in GFRP RC squat walls. To overome this shortoming, the shear strength was realulated, onsidering the onrete ontribution (V) alulated with the formula

135 118 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls proposed by Razaqpur and Isgor (2006) that is being used in CSA S806 (2012) for FRP RC elements: V 0.05 kmkrkaks ( ' f ) 1/ ' ' bwdv, 0.11 fbwdv V 0.22 fbwdv (5.7) 1/ 2 k V d / M ), m ( f f k 1/ ( f ), k a ( 2.5V f d)/ M f 2. 5, k s ( ) 1. 0 (5.8) 450 d r E f A vf b s ' 0.07 f w (5.9) 0.4 f fu where km represents the effet of moment on shear strength at the setion of interest, kr aounts for the effet of reinforement rigidity on shear resistane, ka aounts for the effet of arh ation, and ks aounts for the effet of member size on its shear strength (if the effetive depth is greater than 00 mm with transverse shear reinforement less than AvF; e.g., speimen G-V). Table 5. presents the realulated shear strengths aording to Eq. 5.5 in addition to Eq Considering onrete ontribution (Eq. 5.7) resulted in more reasonable yet onservative results. The shear strength underestimation in G4-250 hanged from 70% to 28%, while the predited ultimate strength of G-V hanged from 0.0 kn to 16 kn, whih is still signifiantly lower than the experimental ultimate strength. The modifiation also made it possible to predit the failure mode of G4-80 and G-V. Similar to ACI 440.1R (2015) whih omits the onfinement effet in flexural strength estimation, the ultimate flexural strengths of G4-80 and G6-80 are highly underestimated using the CSA S Confinement Influene on Wall Response Sine the test speimens were properly detailed at the boundary elements and enough transverse reinforement was provided, the onfinement effet should be taken into onsideration. This effet is evident espeially for the speimens that experiened flexural ompression failure (G4-80 and G6-80). Test observations revealed that, despite the onreteover splitting with initiation of over spalling, the speimens ontinued to sustain loading and underwent substantially large deformations. This indiates the effetiveness of the transverse

136 119 reinforement in onfining the onrete ore and delaying failure. Table 5.4 ompares the measured experimental values for lateral strength and onrete ompressive strain orresponding to onrete-over splitting to that ourring at ultimate state. It an be inferred that the onfinement improved either the ultimate strength or onrete ompressive strain by almost 61% and 171%, respetively. Therefore, negleting the onfinement effet will underestimate ultimate strength and exessively underestimate the attained ultimate urvature and displaement, whih depend on onrete ompressive strain. Given the neessity of onsidering the onfinement effet, the flexural strength of speimens G4-80 and G6-80 was realulated based on the measured ompressive strain as listed in Table 5.4. The alulations were arried out based on the onfined onrete ore but omitting the onrete over. The ratios between the experimentally obtained ultimate strengths to their predited values generally reflet that onsidering the onfinement effet yielded lose preditions of the ultimate strength with a differene of less than 4%. Wall Ps (kn) Conrete splitting 5.5. Conlusion Table 5.4 Evaluation of the onfinement effet Experimental Confinement Effiieny Predition ACI CSA S806 εs (mm/mm) Ultimate state Pu (kn) ε (mm/mm) 1 Pu 100 P s s Ppred (kn) P pred P u Ppred (kn) G G Notes: P = lateral strength; ε = onrete ompressive strain; the subsripts s and refer to onrete splitting and onrete rushing, respetively. P pred P u This hapter aimed at experimentally investigating the response of GFRP RC squat walls under quasi-stati reversed yli lateral loading, emphasizing the effet of web reinforement. Six large-sale GFRP RC squat walls with different web reinforement onfigurations were onstruted and tested to ahieve this objetive. On the basis of the test results and analysis, the following onlusions an be drawn:

137 120 Chapter 5: Effet of Web Reinforement on the Seismi Behavior of Squat Walls 1. Web reinforement had no signifiant effet on either initial flexural or initial shearraking load. The slight differenes observed an be attributed to the differene in onrete strength. 2. Horizontal web reinforement had a diret impat on inreasing squat wall ultimate shear strength and drift ratio as long as the failure was preeded by diagonal tension. Using horizontal web reinforement in exess of the amount required for resisting the ultimate flexural apaity, however, had no effet.. The presene of either horizontal or vertial web reinforement in the wall speimen signifiantly enhaned the onrete ontribution after formation of the first shear rak. However, this enhanement is onservatively negleted by the ACI 18 (2014). 4. The presene of horizontal and vertial web reinforement was essential in rak reovery between load reversals. 5. In ontrast to steel RC squat walls, the shear strain was distributed along the wall height due to the elasti nature of GFRP bars. 6. The provision of horizontal or vertial web reinforement had a lear effet on ontrolling shear-rak width, although horizontal web reinforement proved to be more effetive. 7. Limiting strain to 5000 με in the horizontal web reinforement is reommended to ontrol the shear-rak width in GFRP RC squat walls. 8. The shear estimation method adopted by ACI 440.1R (2015) yielded a onservative estimation of the ultimate shear strength. An aurate estimation of the onrete shear ontribution to aount for the inrease after the first shear rak and modifying the assumption of a 45 shear-rak angle adapted for walls with different properties might, however, be neessary. 9. The shear method in CSA S806 (2012) was shown to losely predit the shear-rak angle. Sine it neglets the onrete s ontribution to shear strength, this method unduly underestimates ultimate shear strength, whih ould lead to reinforement ongestion. Therefore, the ultimate shear strength was realulated using the onrete shear strength alulated with the Razaqpur and Isgor (2006) equation, whih yielded more reasonable yet onservative estimations.

138 The reinforement for detailing at the boundary improved onrete onfinement, improving the predited ultimate strength. Therefore, the onfinement effet should be onsidered in the design. Further studies are needed to shed more light on the effet of vertial web reinforement on ultimate shear strength and larify the effet of other parameters, suh as aspet ratio, and the effet of boundary elements on shear response.

139

140 CHAPTER 6 Evaluation of Flexural and Shear Stiffness of Conrete Squat Walls Reinfored with Glass-Fiber- Reinfored-Polymer (GFRP) Bars Foreword Authors and Affiliation o Ahmed Arafa: PhD andidate, Department of Civil Engineering, University of Sherbrooke. o Ahmed Sabry Farghaly: Postdotoral Fellow, Department of Civil Engineering, University of Sherbrooke, and Assoiate Professor, Assiut University, Egypt. o Brahim Benmokrane: Department of Civil Engineering, University of Sherbrooke, Sherbrooke. Journal: ACI Strutural Journal Paper status: Submitted on Marh, 2017 Referene: Arafa, A., Farghaly, A. S., Benmokrane, Evaluation of Flexural and Shear Stiffness of Conrete Squat Walls Reinfored with Glass-Fiber-Reinfored-Polymer (GFRP) Bars, ACI Strutural Journal. 122

141 12 Contribution in Thesis: The test results presented in hapter 4 and 5 have proven the appliability of GFRP RC squat walls in resisting lateral loads and strongly suggested the neessity of proposing a design proedure for them. One of the most important aspets in the design of strutural walls is evaluating its lateral displaement and limiting it to an aeptable threshold for servieability and other limit state riteria. This requires a proper estimation of the lateral stiffness that is the main sope of the present hapter. It should be noted that only speimens, G4-250, G4-160, G4-80 and G6-80 (that are the most pratial test speimens) in addition to the steel referene speimen are inluded in the estimation. Abstrat Estimating the flexural and shear stiffness of onrete squat walls reinfored with glass-fiberreinfored-polymer (GFRP) bars is important in order to evaluate the lateral displaement. To address this issue, five full-sale onrete squat walls, inluding four reinfored with GFRP bars and one reinfored with steel bars, were tested to failure under quasi-stati reversed yli lateral loading. Deoupling flexural and shear deformations of the tested speimens showed the ontribution of shear deformation to the lateral displaement. The shear stiffness of the raked wall an be estimated based on the truss model with an aeptable level of onservatism. The shear-rak angle and onrete shear strength were evaluated. The flexural stiffness was estimated based on the available expressions in the odes and guidelines related to the design of onrete members reinfored with FRP bars, demonstrating their adequay with walls, although they were established for beam and slab elements. Based on regression analyses of the test results, expressions that orrelate flexural and shear stiffness to lateral drift were proposed. Suh expressions would be vital in the ontext of displaement-based design. Keywords: GFRP bars, onrete squat walls, stiffness, flexural and shear deformations, seismi resistane.

142 124 Chapter 6: Evaluation of Flexural and Shear Stiffness of Conrete Squat Walls 6.1. Introdution The use of fiber-reinfored-polymer (FRP) materials has been growing to overome the usual problems indued by the orrosion of steel reinforement in onrete strutures. These investigations, however, have foused mainly on the behavior under stati-loading onditions, fousing less frequently on seismi design. The feasibility of using FRP as internal reinforement for lateral-resisting systems while preserving the stiffness and deformation apaity has beome prominent. Mohamed et al. (2014a) tested mid-rise shear walls showing the stable yli performane and high level of deformability ahieved by GFRP RC shear walls in omparison to one reinfored with steel. Mohamed et al. (2014b) indiated the potential of GFRP reinforement in distributing the shear deformations along the wall height, owing to its elasti nature, resulting in ontrol shear distortion relatively similar to that in the steel RC wall in whih shear distortion took plae simultaneously with ourrene of yielding of flexural reinforement and mobilized at the plasti hinge zone, thereby deteriorating shear resistane. The test results for the GFRP RC mid-rise walls paved the way for a new experimental series using GFRP bars in squat walls (height-to-length ratio less than 2.0) in whih the sheardeformation problem is frequently enountered (Paulay et al. 1982, Luna et al. 2015). Arafa et al. (2016b) reported experimental results on two squat walls: one was reinfored with onventional steel bars, while the seond was reinfored with GFRP bars. The GFRP RC squat wall attained satisfatory strength and stable yli behavior as well as self-entering apaity that ontributed in preventing sliding shear, whih ourred in the steel RC ounterpart. One of the most important aspets in squat-wall design is estimating wall s lateral displaement and limiting this displaement to an aeptable level. This requires an appropriate estimation of wall lateral stiffness, whih an signifiantly affet the alulation of the natural period time and the distribution of lateral fores among struture walls as well. Aordingly, estimating both the flexural and shear stiffness of GFRP RC squat walls was the main fous of this study.

143 Researh Signifiant This paper fouses mainly on estimating the flexural and shear stiffness of GFRP RC onrete squat walls as a lateral seismi element. Flexural and shear deformations were deoupled, showing the signifiant effet of shear deformation on the total displaement. The flexural and shear stiffness of the GFRP RC onrete squat walls was evaluated. Herein, these results are thoroughly disussed and ompared to the experimental results. In addition, to gain useful information within the ontext of displaement-based seismi design, expressions that diretly orrelate the squat-wall flexural and shear stiffness with lateral-drift ratio were proposed. 6.. Summary of Experimental Program and Results Five full-sale reinfored-onrete squat walls were onstruted and tested to failure under quasi-stati reversed yli lateral loading. Four speimens were entirely reinfored with GFRP bars (G4-250, G4-160, G4-80, and G6-80) and one was reinfored with steel bars (S4-80). Figures 6.1a and 1b show the onrete dimensions and reinforement onfiguration of the test speimens. The boundary elements longitudinal- and transverse-reinforement ratios and vertial web reinforement were 1.4%, 0.89%, and 0.59%, respetively, in all speimens. Four horizontal web reinforement ratios equal to 0.51%, 0.79%, 1.58%, and.58% were used in G4-250, G4-160, G4-80, and G6-80 with using two layers of No. 1 GFRP bars spaed at 250, 160, and 80 mm (9.8, 6.,.15 in) or No. 19 GFRP bars spaed at 80 mm (.15 in), respetively. Speimen S4-80 served as a referene for G4-80, so both speimens had idential reinforement onfigurations and ratios. One layer of bidiagonal No. 10 GFRP bars with spaing of 100 mm was added to prevent sliding shear. Figures 6.1 and 1d show the test setup and loading history, respetively. A series of linear variable differential transduers (LVDTs) and bar strain gauges were mounted on the speimens (Figure 6.1e). Table 6.1 provides the mehanial properties of the reinforement.

144 126 Chapter 6: Evaluation of Flexural and Shear Stiffness of Conrete Squat Walls Boundary 8 No.10 steel or GFRP No.10 steel or GFRP Vertial reinf. No.1 or No.19steel or GFRP Horizontal reinf. No.10 steel or GFRP Θ = 45 Dywidag bars Atuator Out of plane braing 2550 mm Base Loading beam Test speimen Rigid floor Reation wall a) Conrete dimensions b) Reinforement details ) Test setup Lateral drift (%) Cyle number Lateral displaement (mm) Lateral top displ. Out-of-plane displ. Lateral displ. at h=l w Diagonal LVDTs Vertial LVDTs Conrete strain Base-floor sliding h= lw Wall-base sliding d) Applied displaement e) Instrumentations Unit: mm; 1 mm = in Bar Figure 6.1 Conrete dimensions, reinforement details, test setup, load history, and instrumentation Table 6.1 Tensile properties of the reinforement Designated Bar Diameter (mm) Nominal Area 1 (mm 2 ) Immersed Area (mm 2 ) Tensile Modulus of Elastiity 2 (GPa) Tensile Strength 2 * (MPa) Average Strain at Ultimate (%) Straight bars No. 10 GFRP No. 10 steel fy = 420 εy = 0.2 No. 1 steel fy= 420 εy = 0.2 Bent No. 10 GFRP retilinear spiral Straight Bent Bent No. 1 GFRP horizontal bar Straight Bent Bent No. 19 GFRP horizontal bar Straight Bent fy: steel yielding strength, εy: steel yielding strain. 1 Aording to CSA S807 (CSA, 2010) 2 Tensile properties were alulated using nominal ross-setional areas. *Guaranteed tensile strength: Average value standard deviation (ACI )

145 127 Premature sliding failure dominated the behavior of S4-80 due to flexural reinforement yielding, whih produed a major horizontal rak (Figure 6.2a). Suh behavior, however, was prevented in G4-80, whih exhibited flexural ompression failure (Figure 6.2b). This was attributed to the elasti nature of GFRP bars, whih helped the raks to realign and lok up in the ompression zone as well as distributing shear deformations along the wall height. Figure 6. illustrates that both speimens exhibited similar initial stiffness. Due to the relatively low elasti modulus of GFRP bars, however, G4-80 experiened softer behavior than S4-80 after initiation of the first flexural rak. The two envelopes interseted at 1.5% lateral drift. Then, the strength of S4-80 deteriorated due to loalized sliding-shear deformations, while G4-80 s strength kept inreasing to ahieve an ultimate load and drift of almost 71% and 50% higher than those of S4-80, respetively. The failure of G6-80 was identified as flexural ompression failure (Figure 6.2). The failure of G4-250 ourred by sliding along a major diagonal shear rak (Figure 6.2d) due to the inadequay of horizontal web reinforement, while G4-160 experiened sudden flexural rupture in the longitudinal bars at the boundary element under tension (Figure 6.2e). Figure 6. shows that the horizontal web reinforement ratio had a signifiant effet on inreasing the ultimate strength and drift ratio. This effet, however, appears to have been insignifiant when the wall was provided with more horizontal web reinforement than required for flexural resistane (G6-80 ompared to G4-80). Overall, the observed behavior reveals the aeptable behavior of GFRP-reinfored walls as a lateral-resisting system in low to moderate earthquake regions. Table 6.2 summarizes the failure progression of the test speimens. Sliding shear Flexural ompression Flexural ompression Diagonal tension shear Flexural tension (a) S4-80 (b) G4-80 () G6-80 (d) G4-250 (e) G4-160 Figure 6.2 Failure modes of the test speimens