Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Seismic behavior and design of friction concentrically braced frames for steel buildings Tremblay, Robert 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1994-894539.pdf [ 12.79MB ]
JSON: 831-1.0050464.json
JSON-LD: 831-1.0050464-ld.json
RDF/XML (Pretty): 831-1.0050464-rdf.xml
RDF/JSON: 831-1.0050464-rdf.json
Turtle: 831-1.0050464-turtle.txt
N-Triples: 831-1.0050464-rdf-ntriples.txt
Original Record: 831-1.0050464-source.json
Full Text

Full Text

SEISMIC BEHAVIOR AND DESIGN OFFRICTION CONCENTRICALLY BRACED FRAMESFOR STEEL BUILDINGSbyROBERT TREMBLAYB.Sc.A., Université Laval, 1978M.Sc., Université Laval, 1988A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESDepartment of Civil EngineeringWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAOctober 1993(c) Robert Tremblay, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of Ct’%/ L çThe University of British ColumbiaVancouver, CanadaDate D C2YZDE-6 (2/88)ABSTRACTThis study explores the possibility of improving the seismic response of concentrically braced frames(CBFs) by including friction connections at the end of the bracing members. These connections willslip at a predetermined load level in order to absorb and dissipate by friction most of the energyinput by earthquake ground motions, and then avoid yielding and buckling of the bracing members.The characteristics of CBFs and other alternative systems are reviewed and a FCBF system is proposedwhich includes bolted brace connections with slotted holes in the gusset plates. 42 connection samplesincluding various faying surface materials and bolting configurations were subjected to dynamicallyapplied cyclic loading. These tests revealed that a stable response can be achieved by using propersliding material together with an appropriate bolt clamping force level. The behavior of a full-scalebraced frame assembly undergoing severe interstorey drifts was also investigated. The results of thequasi-static tests performed showed that the system behaves in a very predictable and satisfactorymanner.An analytical study including nonlinear dynamic analyses of typical single- and multi-storey FCBFswas then performed in order to develop design guidelines for their stability under seismic loading.For single-storey buildings with a rigid roof diaphragm, design spectra were developed for predictingthe ductility demand and the threshold of instability. For single-storey structures with a flexible roofdiaphragm, a case study including 48 buildings revealed that this type of structure experiences moresignificant nonlinear response than the equivalent SDOF system.Analyses of eight braced frames varying from 2 to 12 storeys in height showed that instability inmulti-storey FCBFs occurs in a collapse mechanism involving only a few stories. Such phenomenacan be inhibited by providing the columns in the building with sufficient bending strength and stiffness.Values were proposed for the stiffness which appeared to yield a stable response of the frames.Columns designed solely for gravity loads were found to have sufficient strength for sustaining themoments likely to develop during ground motions. Design criteria for the brace induced column loadsand for the loads acting in the collecting elements were also proposed.IiTABLE OF CONTENTSABSTRACT iiTABLE OF CONTENTS iiiLIST OF TABLES vLIST OF FIGURES vilACKNOWLEDGEMENTS xviChapter 1 INTRODUCTION 11.1 Motivations 11.2 Objectives, scope and limitations 5Chapter 2 ON IMPROVING THE SEISMIC RESPONSE OF CBFs 92.1 Current earthquake-resistant design philosophy 92.2 Conventional CBFs 122.3 Improving the seismic response of CBFs 19Chapter 3 FCBF: PROPOSED SYSTEM AND DESIGN PROCEDURE 343.1 Introduction 3432 Proposed FCBF system 473.3 Design of the proposed FCBF system 573.4 Discussion 78Chapter 4 TESTING THE PROPOSED FCBF SYSTEM 864.1 Experimental program 864.2 Material 944.3 Static slip tests 1054.4 Dynamic cyclic tests 1104.5 Full scale braced frame testing 151mChapter 5 SEISMIC RESPONSE AND STABILITY OF FCBFs 2065.1 Introduction 2065.2 Stability of single-storey structures 2115.3 Stability of multi-storey structures 248Chapter 6 CONCLUSIONS 3946.1 Summary and fmdiugs 3946.2 Conclusions and recommendations 4026.3 Needs for further research 405BIBLIOGRAPHY 409APPENDIX AAPPENDIX BNEED FOR DUCTILE FRAMING SYSTEMS 420CURRENT NBCC SEISMIC PROVISIONS FORBASE AND STOREY SHEARS 428APPENDIX C VARIATION OF THE BOLT CLAMPING FORCEUPON SLIDING OF FRICTION CONNECTIONS 431APPENDIX D DEMAND ON FRICTION SLIDING CONNECTIONS 435D.1 Procedure 435D.2 Results 436APPENDIX E DESCRIPTION OF THE GROUND MOTIONACCELEROGRAMS 443APPENDIX F SELECTION AND SCALING OF GROUND MOTIONACCELEROGRAMS 453F.1 Scaling accelerograms 453F.2 Selecting accelerograms for regions with given a/v ratios 455F.3 Selecting site-specific accelerograms 456APPENDIX G CHARACTERISTICS OF THE SINGLE-STOREYSTRUCTURES WITH FLEXIBLE ROOF DIAPHRAGM 475APPENDIX H CHARACTERISTICS OF THE MULTI-STOREYSTRUCTURES 478lvLIST OF TABLESTable 4.1 Properties of the studied bolt-disc spring assemblies 94Table 4.2 Faying surface materials: nominal properties and cost factor 97Table 4.3 Faying surface materials: Specified chemical requirements (%) 97Table 4.4 Strength of the faying material used in the connection samples 99Table 4.5 Results of the calibration of the bolts 104Table 4.6 Static slip tests: specimens and results 106Table 4.7 Dynamic tests: specimens of group 1 (ordinary structural steel) 115Table 4.8 Dynamic tests: specimens of group 2 (high strength steels) 127Table 4.9 Dynamic tests: specimens of group 3 (with cobalt-base insert plates) 135Table 4.10 Full scale braced frame tests: main values of the slip resistance 159Table 5.1 Design data for the studied sites 238Table 5.2 Dimensions of the studied buildings 244)Table 5.3 Post-slip storey shear stiffness and storey drifts 261Table 5.4 Mean and maximum values of the ratio of the elastic to the NBCC baseshears for the small and large buildings (for ground motions NAVO3,04,12and LAVO2,03,12,13,15) 270Table 55 Summary of the analyses performed 276Table 5.6 Post-slip storey shear stiffness for the observed three-storey failure modes 294Table A.1 Design data for studied locations in Canada 422Table E.1 Records HAVO1 to HAV15 445Table E.2 Records TAVO1 to IAV1S 446Table E.3 Records LAVO1 to LAV15 447Table E.4 Records NEOl to NFO6 448Table E.5 Records SSO1 to SSO5 448Table F.1 Ranges of period (sec) for which each group of E/Q records was retained ineach seismic region 456Table F.2 Parameters for the attenuation relationships 461Table F.3 Design intensity parameters for the studied sites 461vTable F.4 Accelerograms selected for single-storey structures with flexible roof diaphragm 464Table G.1 Small size single-storey buildings: member and diaphragm properties 475Table G.2 Small size single-storey buildings: periods and stability coefficient 475Table G.3 Medium size single-storey buildings: member and diaphragm properties 476Table G.4 Medium size single-storey buildings: periods and stability coefficient 476Table G.5 Large size single-storey buildings: member and diaphragm properties 477Table G.6 Large size single-storey buildings: periods and stability coefficient 477Table H.1 Bracing bent members 478Table H.2 Axial load and shapes for columns of the frame 479viLIST OF FIGURESFig. 1.1 Typical braced steel frame for buildings 7Fig. 1.2 Typical CBF configurations 8Fig. 2.1 Braced Frame Systems with substitute dissipating energy mechanisms 32Fig. 2.2 Conventional CBF versus proposed FCBF systems 33Fig. 3.1 Proposed FCBF system 80Fig. 3.2 Typical sliding friction connections 81Fig. 3.3 FCBF for inverted V chevron and K bracing configurations 82Fig. 3.4 Idealized model of the components of the sliding connections 83Fig. 3.5 Design of the gusset plates 84Fig. 3.6 Bending of the bracing members upon sliding of the connections 85Fig. 4.1 Studied sliding connection 164Fig. 4.2 Disc spring washers 165Fig. 4.3 Measured load deformation relationships of the disc spring washers a)AM602130, b) SP-522155 166Fig. 4.4 Experimental setup for calibration of the bolts 167Fig. 4.5 Measured load deformation relationships for the bolts 168Fig. 4.6 Static slip tests: specimens 169Fig. 4.7 Static slip tests: a) connection sample upon testing, b) measured load sliprelationships for series 1 and 2 170Fig. 4.8 Static slip tests: measured load-slip relationships for a) series 2, b) series 3and4 171Fig. 4.9 Static slip tests: typical damage to the faying surfaces (specimens SF-05 andSF-02A) 172Fig. 4.10 Dynamic slip tests: original specimens and testing setup 173Fig. 4.11 Dynamic slip tests: bolt-disc spring assemblies 174Fig. 4.12 Dynamic slip tests: loading and connection sample 175Fig. 4.13 Dynamic slip tests: a) variation of the bolt length after tests at 1.0 Hz, b)interior plate of sample D-01 after testing 176VIIFig. 4.14 Dynamic slip tests: slip load time history and hysteresis for specimen D-O1 177Fig. 4.15 Dynamic slip tests: slip load time history and hysteresis for specimens D-03,D-04 and D-09 178Fig. 4.16 Dynamic slip tests: a) interior plate of sample D-02 after testing, b) interiorplate of sample D-11 after testing 179Fig. 4.17 Dynamic slip tests: slip load time history and hysteresis for specimens D-02and D-05 180Fig. 4.18 Dynamic slip tests: slip load time history and hysteresis for specimens D-1O 181Fig. 4.19 Dynamic slip tests: slip load time history and hysteresis for specimens D-06,D-07 and D-08 182Fig. 4.20 Dynamic slip tests: slip load time history and hysteresis for specimens D-11and D-12 183Fig. 4.21 Dynamic slip tests: slip load time history and hysteresis for specimens D-13and D-14 184Fig. 4.22 Dynamic slip tests: slip load time history and hysteresis for specimens D-15,D-16 and D-17 185Fig. 4.23 Dynamic slip tests: slip load time history and hysteresis for specimens D-18,D-19 and D-20 186Fig. 4.24 Dynamic slip tests: slip load time history and hysteresis for specimens D-21,D-25 and D-26 187Fig. 4.25 Dynamic slip tests: a) interior plate of sample D-21 after testing, b) interiorplate of sample D-27 after testing (two runs) 188Fig. 4.26 Dynamic slip tests: slip load time history and hysteresis for specimen D-27and D-28 189Fig. 427 Dynamic slip tests: slip load time history and hysteresis for specimens D-30,D-32 and D-33 190Fig. 4.28 Dynamic slip tests: a) interior and insert plates of sample D-32 after testing,b) interior plate of sample D-38 after testing 191Fig. 4.29 Dynamic slip tests: slip load time history and hysteresis for specimens D-31,D-34 and D-36 192Fig. 4.30 Dynamic slip tests: slip load time history and hysteresis for specimens D-38and D-41 193Fig. 4.31 Dynamic slip tests: a) interior and insert plates of sample D-39 after testing,vmb) interior and insert plates and bolt of sample D-42 after testing 194Fig. 4.32 Dynamic slip tests: slip load time history and hysteresis for specimen D-39 195Fig. 4.33 Dynamic slip tests: slip load time history and hysteresis for specimen D-42 196Fig. 4.34 Full scale braced frame testing: testing setup 197Fig. 4.35 Full scale braced frame testing: details of the connections of the bracingmember 198Fig. 4.36 Full scale braced frame tests: photos of the connections of the bracing member 199Fig. 4.37 Full scale braced frame tests: damage to faying surfaces of specimen F-01 a)during the 26th cycle of3rd run, b) after testing 200Fig. 4.38 Full scale braced frame tests: storey shear time histories and hysteresis(storey shear vs storey drift) for specimen F-01 201Fig. 4.39 Full scale braced frame tests: storey shear time histories and hysteresis(storey shear vs storey drift) for specimen F-02 202Fig. 4.40 Full scale braced frame tests: slip load time history and hysteresis (brace loadvs connection slip) in 1st run for specimens a) F-01, b) F-02 203Fig. 4.41 Full scale braced frame tests: model for computing the bending moments inthe frame 204Fig. 4.42 Full scale braced frame tests: bending moments in the bracing member in 1strun for specimens a) F-01, b) F-02 205Fig. 5.1 Modeling of a steel framed single-storey building with a rigid roof diaphragm 304Fig. 5.2 SDOF system subjected to a static horizontal external load 305Fig. 53 SDOF system subjected to a seismic ground motion 305Fig. 5.4 Load deformation relationships Fs(u) and V(u): a) linear elastic response, b)nonlinear inelastic response 306Fig. 5.5 Linear seismic response of SDOF systems: amplification of the peak absoluteacceleration due to gravity loads 307Fig. 5.6 Linear seismic response of SDOF systems: amplification of the peak relativedisplacement due to gravity loads 308Fig. 5.7 Nonlinear response of an initially deformed SDOF system to a horizontalpulse 309Fig. 5.8 Nonlinear response of a SDOF system to ground motion IAVO1: a) relativedisplacement for Uy / Ue = 0.40, b) relative displacement for Uy / Ue = 0.30,Ixc) accelerogram 310Fig. 5.9 Nonlinear response of SDOF systems to ground motions IAVO1 and IAVO2:a) peak displacement vs strength, b) ductility vs stability coefficient, c) criticalstrength vs period 311Fig. 5.10 Critical strength of elasto-plastic SDOF systems for different seismic groundmotions 312Fig. 5.11 Occurrence of failures of nonlinear code designed SDOF systems for differentground motion ensembles, values of the stability coefficient and strength levels 313Fig. 5.12 Effects of a post-slip stiffness on the peak relative displacement (mean +SD) of SDOF systems subjected to seismic ground motions: a) a = 1 .1 0,b)a = 1.50 314Fig. 5.13 NBCC seismic response factor, S 315Fig. 5.14 Actual period vs NBCC period for typical steel framed single-storey buildingswith a rigid roof diaphragm 315Fig. 5.15 Anticipated ductility and failure line for single-storey buildings with a rigidroof diaphragm for T(jesjg,j = 0.25 sec and for: a) Uy / Uec = 0.15 and a/v= 0.50, b) Uy / Uec = 0.15 and a/v = 0.71, c) Uy / Uec = 0.15 and a/v =1.00, d) Uy / Uec 0.15 and a/v = 1.41, e) Uy / Uec = 0.15 and a/v = 2.00,f)uy/Uec = 0.20 anda/v = 0.71 316Fig. 5.16 Anticipated ductility for single-storey buildings with a rigid roof diaphragmfor T(ksig,, = 0.25 sec and for different strength levels, values of the stabilitycoefficient and a/v ratios 319Fig. 5.17 Statistics of the ductility demand for single-storey buildings with a rigid roofdiaphragm for T(kSj = 025 sec, Uy / Uec = 0.25, 0 = O.O2S T2 anda/v = 0.71 321Fig. 5.18 Characteristics of the studied single-storey buildings with a flexible roof diaphragm: a) calculated periods, b) stability coefficient 322Fig. 5.19 Modeling single-storey braced frame buildings with a flexible roof diaphragm 323Fig. 5.20 Typical response of single-storey buildings with a flexible roof diaphragm subjected to ground motion IAV13: a) building LSH2, b) building SSH1 324Fig. 5.21 Peak ductility: computed values 325Fig. 5.22 Peak ductility ratio of the computed to the expected values (mean + SD) 326xFig. 523 Peak ductility: ratio of the values for buildings with heavy roof to the valuefor building with light roof 326Fig. 524 Peak total drift at mid span 327Fig. 5.25 Peak roof diaphragm deformation at mid span 328Fig. 5.26 Peak bending moment at mid span of roof diaphragm 329Fig. 5.27 Typical envelope and amplification of the bending moment in the roof diaphragm 330Fig. 5.28 Typical envelope and amplification of the shear force in the roof diaphragm 331Fig. 5.29 Approximate mode shapes and stability coefficients for the first three modesof a MDOF beam-column system 332Fig. 5.30 Computed vs NBCC fundamental period for multi-storey braced frames 333Fig. 5.31 Second and third period vs fundamental period for multi-storey braced frames 333Fig. 5.32 Estimated vs computed stability coefficient for the studied braced buildings:a) building with large columns, b) buildings with small columns 334Fig. 533 Storey shear vs shear storey drift for: a) columns pinned at each storey, b)columns continuous over two storey 335Fig. 5.34 Bending of columns continuous over two storeys upon slippage of the bracesliding connections at the bottom level 336Fig. 5.35 Modeling multi-storey braced frame buildings 337Fig. 5.36 Nonlinear response of a multi-storey braced frame subjected to groundmotion LAV14: a) with columns pinned at each storey, b) with columns continuous over three storeys 338Fig. 5.37 Gravity load effects for: a) a single-storey instability mode, b) a two storeyinstability mode 339Fig. 5.38 Post-slip storey shear stiffness for different single-storey instability modes 340Fig. 539 Post-slip storey shear stiffness for different two-storey instability modes 341Fig. 5.40 Gravity load effects as a function of the slenderness and stiffness of thebuilding columns 342Fig. 5.41 Capacity of building columns for combined axial loads and bending moments 342Fig. 5.42 Pseudo-acceleration response spectra of the selected accelerograms for theseismic region with a/v equal to a) 1.41, b) 0.71 343Fig. 5.43 Characteristics of the studied buildings 344Fig. 5.44 Computed elastic base shear for the studied buildings 345xlFig. 5.45 Slenderness ratio of the columns 346Fig. 5.46 Post-slip storey shear stiffness supplied by the a) large columns, b) small columns 347Fig. 5.47 Effects of the gravity loads on the post-slip storey shear stiffness for the a)single-storey collapse mode b) two-storey collapse mode 348Fig. 5.48 Effects of the continuity of the columns on the storey shear vs shear storeydrift (Uc = 0.15) 349Fig. 5.49 Effects of the continuity of the columns on the storey shear vs shear storeydrift (Uy = 0.20) 351Fig. 5.50 Increase in the total capacity of absorbing energy due to the continuity of thecolumns 352Fig. 5.51 Increase in the capacity of absorbing energy due to the continuity of the columns (at a shear storey drift equal to 2% of the storey height 353Fig. 5.52 Exact vs simplified post-slip shear storey stiffness for a two-storey columnsegment upon sliding in one storey 354Fig. 5.53 Exact vs simplified response of building No.3 subjected to two differentground motions 355Fig. 5.54 Number of occurrences of instability failures in series 0 and 1 356Fig. 5.55 Number and mode of instability failures in series 1,2 and 4 357Fig. 5.56 Statistics of the peak brace ductility demand for building no. 1 in series 0 and2 358Fig. 5.57 Statistics of the peak brace ductility demand for building no. 2 in series 0 and2 359Fig. 5.58 Statistics of the peak brace ductility demand for buildings no. 3 to 5 in seriesOand2 360Fig. 5.59 Statistics of the peak brace ductility demand for buildings no. 6 to 8 in seriesOand2 361Fig. 5.60 Statistics of the peak storey shear drift for building nos. 1 and 2 in series 0and2 362Fig. 5.61 Statistics of the peak storey shear drift for building nos. 3 to 8 in series 0 and2 363Fig. 5.62 Statistics of the peak bending moment in the columns for building no. 1 inseries 2 364xiiFig. 5.63 Statistics of the peak bending moment in the columns for building no. 2 inseries 2 365Fig. 5.64 Statistics of the peak bending moment in the columns for buildings nos. 3 toSinseries2 366Fig. 5.65 Statistics of the peak bending moment in the columns for buildings nos. 6 to8 in series 2 367Fig. 5.66 Variation over the height of the accumulated energy dissipated (mean) forbuildings nos. 1 to 2 in series 2 368Fig. 5.67 Variation over the height of the accumulated energy dissipated (mean) forbuildings nos. 3 to 8 in series 2 369Fig. 5.68 Peak brace induced column loads (mean + SD) for buildings nos. 1 to 2 inseries 2 370Fig. 5.69 Peak brace induced column loads (mean + SD) for buildings nos. 3 to 8 inseries 2 371Fig. 5.70 Peak brace induced column loads (COV) for buildings nos. 1 to 8 in series 2 372Fig. 5.71 Statistics of the peak seismic loads for building no. 1 in series 2 373Fig. 5.72 Statistics of the peak seismic loads for building no. 2 in series 2 374Fig. 5.73 Statistics of the peak seismic loads for buildings nos. 3 to 5 in series 2 375Fig. 5.74 Statistics of the peak seismic loads for buildings nos. 6 to 8 in series 2 376Fig. 5.75 Statistics of the peak brace ductility demand for buildings nos. 1 to 2 in series3 377Fig. 5.76 Statistics of the peak brace ductility demand for buildings nos. 3 to 5 in series3 378Fig. 5.77 Statistics of the peak brace ductility demand for buildings nos. 6 to 8 in series3 379Fig. 5.78 Peak brace ductility demand and peak shear storey drift (mean + SD) forbuildings nos. 1 to 2 in series 3 relative to series 2 380Fig. 5.79 Peak brace ductility demand and peak shear storey drift (mean + SD) forbuildings nos. 3 to 5 in series 3 relative to series 2 381Fig. 5.80 Peak brace ductility demand and peak shear storey drift (mean + SD) forbuildings los. 6 to 8 in series 3 relative to series 2 382Fig. 5.81 Statistics of the peak bending moment in columns for buildings nos. 4,6 and8inseries3 353XIIIFig. 5.82 Statistics of the peak seismic loads for buildings nos. 4, 6 and 8 in series 3 384Fig. 5.83 Peak seismic loads and peak bending moment in columns (mean + SD) forbuildings nos. 4, 6 and 8 in series 3 relative to series 2 385Fig. 5.84 Statistics of the peak brace induced column loads for buildings nos. 4, 6 and8inseries3 386Fig. 5.85 Statistics of the peak brace ductility demand for buildings nos. 1 and 2 inseries 4 387Fig. 5.86 Statistics of the peak brace ductility demand for buildings nos. 3 to 5 in series4 388Fig. 5.87 Statistics of the peak brace ductility demand for buildings nos. 6 to 8 in series4 389Fig. 5.88 Peak brace ductility demand and peak shear storey drift (mean + SD) forbuildings nos. 1 to 2 in series 4 relative to series 2 390Fig. 5.89 Peak brace ductility demand and peak shear storey drift (mean + SD) forbuildings nos. 3 to 5 in series 4 relative to series 2 391Fig. 5.90 Peak brace ductility demand and peak shear storey drift (mean + SD) forbuildings nos. 6 to 8 in series 4 relative to series 2 392Fig. 5.91 Statistics of the peak bending moment at the base of the columns for buildings nos. 1 to 8 in series 4 393Fig. A.1 Ratio of the factored base shear due to seismic and wind loads 423Fig. A.2 Ratio of the factored storey shear due to earthquake and wind loads 427Fig. D.1 Peak ductility demand 439Fig. D.2 Number of full hysteresis 440Fig. D.3 Average frequency of yielding 441Fig. D.4 Peak frequency of yielding 442Fig. E.1 Pseudo-acceleration response spectrum of records HAVO1 to HAV15 449Fig. E.2 Pseudo-acceleration response spectrum of records IAVO1 to IAV15 449Fig. E.3 Pseudo-acceleration response spectrum of records LAVO1 to LAV15 450Fig. E.4 Pseudo-acceleration response spectrum of records NF01 to NFO6 450Fig. E.5 Pseudo-acceleration response spectrum of records SSO1 to SSO5 451Fig. E.6 a) typical artificial ground motion, b) pseudo-acceleration response spectrumof records ARO1 to ARO5 452Fig. F.1 Probability of exceedance for intensity levels higher than the NBCC value forxiva) PHA, b) PHV for seven Canadian sites 465Fig. F2 Ductility demand on elasto-plastic SDOF systems 466Fig. F3 Occurrence of failure of elasto-plastic SDOF systems 467Fig. F.4 Distribution of the contribution to the seismic risk for Victoria (site 1) for a)PHA., b) PHV 468Fig. F.5 Distribution of the contribution to the seismic risk for Vancouver (site 2) fora) PHA, b) PHV 469Fig. F.6 Distribution of the contribution to the seismic risk for Ottawa (site 3) for a)PHA., b) PHV 470Fig. F.7 Distribution of the contribution to the seismic risk for Montréal for a) PHA,b) PHV 471Fig. F.8 Distribution of the contribution to the seismic risk for Québec (site 4) for a)PHA, b) PHV 472Fig. F.9 Distribution of the contribution to the seismic risk for Prince Rupert (site 5)for a) PHA, b) PHV 473Fig. F.10 Distribution of the contribution to the seismic risk for Whitehorse (site 6) fora) PHA, b) PHV 474xvACKNOWLEDGEMENTThis study was supported by the Natural Sciences and Engineering Research Council of Canada(NSERC Grant), the Gouvernement du Québec (Fonds pour la Formation de Chercheurs etl’Aide a la Recherche) and the University of British Columbia (University Graduate Fellowships).The author wishes to thank the following organizations for their generous assistance in providingthe material for the experimental part of this study: Empire Iron Works Ltd., Delta, B.C., andCoast Steel Fabricators Ltd., Port Coquitlam, B.C., for supplying structural steel and bolts, SpaeNaur Inc., Kitchener, Ont., for samples of disc spring washers, and Haynes International, mc,.Kohomo, Illinois, for providing the cobalt-base filler plate material.The assistance and the involvement in this project of the technical staff of the Department of CivilEngineering was also greatly appreciated. The author would like to express his gratitude to DickPostgate, Howard Nichol and Bernie Merkli for their remarkable contribution. Many thanks alsohave to be directed to Harald Kulimanu, graduate student, who designed and built the full scaletesting frame used in this study and who skillfully executed several welding works.The author wishes to thank Dr. Pak Ko, from the Institute for Mechanical Engineering, Tribologyand Mechanics, of the National Research Council of Canada, Vancouver, B.C. for his valuableopinions and suggestions related to the tribological aspect of the studied connection. The authorexpresses his thankfulness to Professor S.F. Stiemer, his supervisor, who provided continuous support and inspiration throughout the project, and to his wife Yolande Dugré whose love and support have made the conducting of this research work possible.xviChapter 1INTRODUCTION1.1 MotivationsEfficiency of CBF systems:Resistance to lateral loads in steel framed buildings can easily be provided for by adding inclinedbracing members which can transfer storey shears between successive floors. Such braced frames aremost often of the simple type of construction (CSA 1989), with the beams simpiy supported at theirends. Consequently, the structural framework without the bracing members contributes very little inresisting the lateral loads and such contribution, if any, is generally ignored in the design process.Braces in multi-storey structures are preferably arranged to form with the adjacent beams and columnsvertical cantilever trusses anchored to the foundations (Fig. 1.1) (the figures are inserted at the endof each Chapter). This pattern generally suits well the architectural constraints while providingdesirable continuity of the lateral load resisting system throughout the building height. By takingadvantage of the ability of the roof and floor deck, deck-slab or slab systems to carry horizontal loadsin their own plane, the lateral stability of an entire structure can typically be ensured by only a fewbracing bents in each of the principal directions of the building. For higher resistance to in-planetorsional moments, these bracing bents are preferably layed out along the outer edges of the building.In Concentrically Braced Frames (CBFs), the bracing members meet at the beam-to-column joints,or at a single point within the length of the beams or the columns. Therefore, lateral loads primarilyinduce only axial forces in the members. Fig. 1.2 shows typical configurations of CBF bracing bents.It can be seen that the bracing members generally extend one floor level in height and can be designedto resist either only tension forces (dashed lines) or both tension and compression forces (solid lines).Typical bracing bays range between 6 m and 12 m in width and the practical maximum height-to-widthratio is about 7 (Schueller 1977). Thus, the use of single bent CBFs is generally limited to low- andmedium-rise buildings (up to 20 stories).1Taller structures can also be braced efficiently with CBFs using multi-storey framing configurationsin the building facades or by providing belt or cap perimeter stiffening trusses. Such applications are,however, very uncommon since the vast majority of buildings fall within the low- and medium-heightranges for which combining a hinged frame with vertical trusses as described above is considered tobe the most cost-effective solution when compared to other common systems such as MomentResisting Frames (MRFs) or Shear Walls (HERA 1989).Indeed, the analysis and the design of CBFs are rather simple since the functions of carrying thegravity loads and resisting the lateral loads are accomplished by two different groups of elements inthe structure. Only members within the bracing bays may carry both gravity and lateral loads. Furthermore, both the gravity and the lateral load carrying mechanisms are simple, which makes theirdesign straightforward and less prone to errors. Standard shear beam-column connections are usedthroughout the framework and special detailing is limited to only a few elements located within thebracing bays (gusset plates). Consequently, the fabrication, installation and checking of the wholeframework is easier, thus faster and cheaper. The system also permits shop welded/field boltedconnections which allow quick and all-weather assembly.In MRFs, lateral loads are resisted by bending of the columns and the beams which are rigidlyconnected. This system is by far less efficient, especially for stiffness, than the truss action exhibitedby CBFs. The design of MRFs is also more complex and lengthy since the gravity and lateral loadcarrying systems are combined. Most of the time, the structure is statically indeterminate and aniterative analysis-design process is required. Further, beam-column connections in MRFs are custommade and involve a rather significant amount of detailing, together with substantial field welding atcritical locations.Shear walls, either cast-in-place concrete or steel plated, resist lateral loads similarly to CBFs sincethey also form cantilevered elements fixed at their base. When using concrete shear walls, however,one loses the advantages offered by having a whole prefabricated steel superstructure: uniformity,speed of installation, guaranteed strength, etc. For steel plate shear walls, no accepted design ruleshave been made readily available as yet, and recent studies on this system showed that it would notbe viable for new constructions (HERA 1989).2The Concentrically Braced Frame system also represents an attractive solution for upgrading theresistance to lateral loads of existing buildings. Additional strength and stiffness can readily andefficiently be provided for by inserting steel diagonal bracing members within existing frameworks.In such case, the modifications to the structure are generally limited to the attachment details forthe bracing members and to the beams and columns within the bracing bents which may requiresome local reinforcement to resist the additional axial loads induced by the truss action. This techniquehas recently been found to be very well suited, with minor adjustments, for other framing systemssuch as reinforced concrete frames (Jam 1984; Badoux and Jirsa 1990; Goel and Lee 1990; Bush etal. 1991; Goel and Wight 1993).CBFs for earthquake resistance:For earthquake loading, CBFs also exhibit beneficial features. Their higher stiffness is seen to contribute in limiting building damage and occupant discomfort under minor and moderate events. Moreimportantly, if properly designed, braced frames will experience inelastic action only in their bracingmembers under a severe earthquake, which means that the gravity load carrying function of thestructure is preserved and that structural damage is restricted to but a few elements in the entireframework. In seismic design, post-elastic analyses may be required to determine the ultimate strengthand the distribution of the internal forces for lateral load resisting systems responding nonlinearly.For CBFs, these can easily be performed because of the rather direct load paths involved and thelimited load redistribution that takes place upon inelastic response. In most situations, such analysescan easily be carried out manually.Unfortunately, bracing members have been found to perform poorly in absorbing and dissipatingseismic input energy through inelastic response: progressive slackness, degradation of the compressivestrength, premature fracture, etc. Moreover, braced structures inherently exhibit a limited redundancyand are then exposed to undesirable concentration of ductility demand as well as being vulnerableto possible material or construction defects. For these reasons, which will be discussed in more detailslater, the system becomes less attractive for earthquake resistance and practitioner engineers are mostoften reluctant in using CBFs in seismically active regions.In view of the many advantages offered by CBFs, considerable efforts have been devoted in the lasttwo decades to enhance their seismic response. Extensive research work has focused on theimprovement of the inelastic response of the bracing members and on a better understanding of the3behavior of CBFs. From these emerged more appropriate and comprehensive seismic design provisionsin codes. However, the design process gained in complexity, and the efficiency and the reliability ofthe braces still remain a matter of concern. Further, in many cases, these design provisions have alsobeen found to result in a high level of brace overstrength that has to be accounted for in the designand which offsets the economical benefits of the system.Other studies concentrated on the development of substitute means for energy absorption and dissipation in braced structures. In most cases, this goal has been achieved since the system developedexhibited superior inelastic response when compared to the conventional CBF system. Unfortunately,these modified systems have been found rather complex and therefore more difficult to be handledwithin normal office design procedures. Further, most of them are less efficient in resisting lateralloads, both for strength and stiffness. One of these newly proposed systems, referred to herein as theFriction Concentrically Braced Frame (FCBF) system, appeared, however, very promising since itdoes not exhibit these shortcomings while standing very close to the basic CBF system: a connectionis provided at the end of the braces which would slide at a predetermined load level and then dissipatethrough friction the seismic input energy.Friction had long been known as an efficient mechanism for dissipating energy and increasing dampingin systems subjected to dynamic loading. In steel structures, the hysteretical response exhibited bybolted connected parts slipping relative to each other had been investigated (Brown 1968; Vitelleschiet al. 1977) and higher damping had then been attributed to bolted steel frames for inclusion indynamic analysis (Werner and Reddy 1975; Newmark and Hall 1982).More recently, many analytical and experimental studies had shown that providing dedicated energydissipation frictional devices acting in parallel with lateral load resisting systems, such as momentframes or shear walls, could significantly improve (reduce) the seismic response of these systems (Palland Marsh 1981, 1982; Austin and Pister 1985; Filiatrault and Cherry 1987; Olson 1987; Whittakeret a!. 1989a; Aiken 1990; Filiatrault 1990). Although most of the tests performed involved rathersophisticated devices and/or small slip loads, they revealed that stable hysteresis could be exhibitedthrough friction upon cyclic loading. Tests by Constantinou et a!. (1989) on frictional assemblies forseismic bridge isolation also indicated that friction energy dissipators could respond in a very predictable and reliable manner, provided that suitable faying surface material was used.4Surprisingly, very few studies addressed the behavior of a FCBF like system, where the frictionalmechanism is to act in series in the lateral load resisting system. Roik et a!. (1988) and FitzGeraldet al. (1989) had performed dynamic cyclic tests on samples of bolted connections between unfinishedsteel connected parts with slotted holes. Results were encouraging in a sense that stable hystereticbehavior was obtained in both cases. Test of a scaled model of a single storey braced frame includingmultiple stage bracing members with sliding connections, also performed by Roik et aL, showed goodagreement between the measured and computed response of the frame.In these tests, however, the slip loads considered were relatively low and the conditions imposed forthe testing of the connection samples were deemed to be less critical than what could be expectedduring actual seismic events: 1 mm relative slip displacement (Roilc et aL) and 0.25 Hz frequency(FitzGerald et aL). Also, the systems proposed by these researchers involved a rather complexmulti-stage sliding mechanism which would unlikely be implemented in practice.1.2 Objectives, scope and limitationsThe objectives of this research project were to come up with an efficient FCBF system, which couldbe readily implemented in practice, and to propose appropriate and simple design procedures for it.Preferably, the proposed system was to be such that it could be entirely designed, fabricated andinstalled by organizations currently involved in the structural steel industry (engineering firms, steelfabricators, etc.). Thus, it was to be as close as possible to the conventional CBF system and torequire only techniques and materials currently employed for the fabrication of CBFs. Its performancewas to be verified experimentally under the most critical conditions likely to be encountered.The proposed design procedures were to be similar to, or based on, those currently in use in Canada.They were also to particularly account for the susceptibility of the CBF system to dynamic instability(large localized interstorey drift) due to its inherent low redundancy.Although some of the fmdings of this study could eventually be applicable to other situations, thisproject was limited to the study of concentrically braced steel frames for which lateral loads areprimarily resisted by individual bracing bents anchored at their base, as described previously. Inaddition, only the case of normal buildings, founded on stiff soils or rock, with uniform mass, stiffnessand geometry was investigated.5To achieve the aforementioned objectives, the following tasks were undertaken:Evaluation of the appropriateness of a FCBF system for the resistance to ea,thquake groundmotions. As the system had been proposed only very recently, it was decided to start thisresearch by examining its possible advantages and shortcomings when compared to currentCBF and alternative systems. In Chapter 2, the seismic response of conventional CBFs isinvestigated and the main weaknesses are pointed out and discussed. Thereafter, the mostpromising ways that have been proposed for improving their response are briefly reviewed andcompared to the FCBF system.Proposal for a FCBF system. Based on the fmdings and recommendations of previous researchworks, a suitable FCBF system was to be proposed, together with design guidelines. This ispresented in Chapter 3. Throughout the discussion, the questions and matters of concern relativeto the performance of the system or to the appropriateness of the design methods are outlined.Erperimental verification of the performance of the proposed system. An experimental study hasbeen performed in order to assess the performance of different sliding connections and thebehavior of a whole bracing bay assembly under cyclic loading. This study is included in Chapter4. Static slip tests and dynamic cyclic slip tests were performed on sample connections in orderto fmd the more appropriate combination of contact materials and clamping force. The testson the braced frame assembly permitted the obtaining of valuable information on the overallresponse of the system undergoing severe interstorey drifts.Investigation of the stability of braced frames. This topic is addressed in Chapter 5. The case ofsingle-storey structures was investigated first. Analyses of Single Degree of Freedom (SDOF)systems were performed for deriving design charts for the ductility demand for the case wherethe roof diaphragm is perfectly rigid. An analytical study on typical single-storey buildings withflexible diaphragms have been carried out subsequently in order to determine if the designspectra developed for the case of rigid diaphragms could be applied to such structures. Formulti-storey frames, the possibility of taking advantage of the continuity of the columns overtwo or more storeys in order to avoid the concentration of energy dissipation was explored.Simple design guidelines obtained from idealized models were proposed and validated throughnumerical analyses.6SAlt t t t t titBUILDING PLAN-_CROSS SECTION NA A1Fig. 1.1 Typical braced steel frame for buildings.tttttt t7zFig 1.2 Typical CBF configurations.CHEVRON BRACINGVVDBF CATEGORY NDBF CATEGORYTENSION-COMPRESSIONX-BRACINGTENSION-ONLY K-BRACINGX-BRACINGN/SPLIT-X BRACING/N2-BAY CHEVRON BRACING TENSION-COMPRESSIONDIAGONAL BRACINGDIAGONAL BRACING//NNTENSION-ONLYDIAGONAL BRACING8Chapter 2ON IMPROVING THE SEISMIC RESPONSE OF CBFs2.1 Current earthquake-resistant design philosophyThe main objective of the earthquake-resistant design provisions included in modern building codesis essentially to prevent collapse of building structures when exposed to major but rare earthquakeevents. Experiences during past earthquakes have shown that some lateral load resisting systems couldsatisfy this ultimate limit state even if their capacity was well below the force level that would havebeen reached if they had remained elastic during the earthquake. This fact has been recognized inestablishing code provisions, and design strengths lower than the elastic force level have thereforebeen prescribed for these systems, provided that they can exhibit a proper inelastic response.Basically, a proper inelastic response implies maintaining good energy absorbing and dissipatingcapabilities over the entire duration of the expected earthquake ground motions. Proper energyabsorption has traditionally been achieved by means of a ductile and tough system which can undergolarge inelastic excursions without loss of strength, failure and, ultimately, collapse. The systems mustalso be capable to maintain its initial stiffness and strength over successive inelastic load reversals inorder to dissipate efficiently the seismic input energy for the duration of the event.In the 1990 edition of the National Building Code of Canada (NBCC) (NRCC 1990a), the NationalModel Code document for new buildings in Canada, the design strength for a given system is obtainedby dividing the expected elastic forces by a force modification factor R. This factor varies from 1.0for non-ductile structural systems to 4.0 for ductile structural systems. For steel framed buildings, anylateral load system qualifies for R equal 1.5 because of the inherent ductility of the material. ForCBF systems, two special categories are defmed in the NBCC: the Braced Frames with NominalDuctility (NDBF) for which R equals 2.0, and the Ductile Braced Frames (DBF) with an assignedR value of 3.0.When subjected to the same ground motion, systems designed for lower seismic loads (higher Rvalues) are likely to undergo larger and more numerous inelastic excursions. In order to ensure anadequate post-elastic behavior for those systems, the CAN/CSA-S16.1-M89 (CSA 1989), Limit States9Design of Steel Structures, the Standard for the design, fabrication and erection of steel structuresin Canada, now includes specific seismic design and detailing provisions based on a capacity designphilosophy. Obviously, more stringent rules apply to the DBF system.In the capacity design approach, elements or mechanisms of the lateral load resisting system arechosen in the frame where the inelastic response is to take place during severe earthquakes. Thechoice is made in such a manner that the structure can still carry the applied gravity loads whileresponding inelastically. These critical elements, also called weak elements, are designed to merelysatisfy the design strength prescribed by the NBCC and are properly detailed to form efficient energydissipation mechanisms that can survive cyclic inelastic loadings for the duration of the design basedearthquake.Upon yielding, the critical elements impose forces equal to their actual ultimate capacity to thesurrounding, non-critical, elements of the lateral load resisting system. These non-critical elements,or strong elements, must be protected against premature failure in order to ensure the viability ofthe whole lateral load resisting system and, particularly, of the chosen energy dissipation mechanisms.For example, non-critical elements may be part of the gravity load carrying system, exhibit limitedductility and/or play a role of prime importance in the system (e.g. exterior columns of bracing bents)and their protection is mandatory in order to maintain the integrity of the framework. Thus, noncritical elements must be designed for the maximum forces that can be imposed to them during theseismic event, regardless of the intensity of the shaking. That includes the forces developed by thecritical elements together with those due to the gravity loads acting on the structure.Selecting a system which qualifies for lower seismic design loads seems at first economically advantageous since it would likely result in a lighter framing and lower forces applied to the foundations.As shown in Appendix A, the relative magnitude of wind and seismic effects on buildings of variousdimensions across Canada is such that the use of ductile systems having a R factor higher than 1.5represents, on that basis, an attractive solution for a wide range of projects.In addition to that immediate economical benefit, designing a structure which can adequately respondinelastically to earthquake ground motions presents other advantages. For example, there is a growingconcern about erecting structures with non-ductile lateral load resisting systems in regions where windloads dominate over seismic loads. Indeed, it is now believed that many such regions can be struckby ground motions of intensity many times higher than the code seismic intensity parameters (SEAOC101990). Though the probability of occurrence of such events is rather small, it is likely that the needfor proper ductile systems be further extended in the near future to a much wider range of buildingsand locations than indicated by the results obtained in Appendix A.Buildings responding inelastically to earthquake ground motions also undergo reduced peak accelerations, and the damage potential to non-structural elements and to the building content consequentlydiminishes. Furthermore, response of intermediate and longer-period structures is most likely to bereduced because the inelastic action shifts their fundamental period away from the range of frequencycontaining most of the energy of ground motions.These advantages may, however, be annihilated by the requirements imposed by, and the possibleconsequences of, resisting earthquakes through inelastic response. Firstly, a capacity design procedureis more lengthy and thus more costly than the traditional design approach which is based on a linearresponse of structures. Among other things, it requires the following additional steps: i) a suitablechoice and proper detailing of the critical elements, ii) a careful examination of the relative strengthof the elements of the whole lateral load resisting system and iii) an investigation of the post-elasticresponse of the system.Secondly, building a ductile system may result in supplementary construction costs since the additionaldetailing required for the various elements of the structure obviously increases the fabrication andinstallation efforts. Also, upon inelastic action, over-sized critical elements will eventually overloadthe structure down the lateral load path. Therefore, any attempt to save on the cost by standardizingthe critical elements over a given region of the structure would be somewhat cancelled out by theadditional material required by the non-critical elements affected. Further, more careful checking atthe construction site and, possibly, maintenance over the building lifetime, is mandatory in order tomake sure the system will behave in the manner assumed in the design.Among the possible consequences of relying on inelastic behavior is that building structures designedwith lower seismic loads will likely suffer damage to a certain extent and will then need to be repairedand even replaced. That, together with the costs implied by the fact that the building cannot be usedfor the time of repair, can represent a significant loss for the owner.Consequently, in addition to being reliable and capable of sustaining the expected ground motions,a proper ductile system must present additional advantages to represent a viable solution. The energydissipation mechanism (critical elements) must be simple to design, build, install and checked. It must11also qualily for significantly reduced design seismic loads. In order to minimize the overstressing ofthe non-critical elements, each critical element should easily be given exactly, or very closely, itsrequired strength. Moreover, the ratio of their expected, or actual, ultimate capacity to their designstrength should be kept to a minimum. Finally, damage should be constrained to a few elements inthe structure, preferably not part of the gravity load carrying system.2.2 Conventional CBFs2.2.1 Seismic response and designSince lateral loads primarily induce axial forces in the members of CBFs, inelastic action in thissystem has to be associated with longitudinal deformation of the members. Concentrating cyclicinelastic axial deformations in the connections is not suitable as the resulting large local strains wouldlikely result in premature failures. Beams and columns are not good candidates either for inelasticresponse: under compressive loads, buckling of these members would occur which, in turn, wouldendanger the integrity of the gravity load carrying system of the frame. On the other hand, bracingmembers play no, or little and non-essential, role in carrying gravity loads in a structure, and mostof the lateral loads pass through the braces on their way down to the foundations. These membershave then been traditionally retained to form the critical elements in CBFs.Upon application of lateral loads to bracing bents, an overturing moment and storey shear develop.The former is resisted through axial loads by the exterior columns of the bents and the storey shearsare taken up by the bracing members and, possibly, by the beams depending upon the bracingconfiguration (Fig. 1.2). Therefore, while the elastic response of a bracing bay involves both bendingand shear deformations, as this is the case for shear walls, the global inelastic response of bracedstructures is of the shear beam type, similar to that exhibited by moment resisting frames.Storey shears in tension-only bracings is assumed to be entirely resisted by tension acting braces only.Slender braces, which buckle elastically at low compression loads, are then used in both directions(Fig. 1.2). Thus yielding, and thereby energy dissipation, can only take place through stretching ofthe braces. Upon cyclic inelastic loading, braces accumulate permanent elongation and the framegradually loses its lateral stiffness and its strength near the undeformed position as the length of thebraces increases. Therefore, as the inelastic action progresses, the energy absorption and dissipation12capabilities continually degrade near the rest position and the horizontal deformation of the framekeeps increasing. Further, at each half-cycle, impact loads are induced when the previously buckledbraces suddenly tighten up (Ghanaat 1980).Such poor performance of tension-only systems, when undergoing inelastic response, has been traditionally compensated by specilring higher strength levels (higher design seismic loads). In the S16.1,the use of this bracing system is restricted to the NDBF category (R = 2.0) or to the case where avalue ofR equal to 1.5 is considered in the design. In the former case, beams, columns and connectionsmust also be designed for amplified brace loads (10%) to minimize the risk of premature fracturedue to impact loading.Recent research works (Gugerli and Goel 1982; Jam et al. 1978, 1980; Kahn and Hanson 1975; Maisonand Popov 1980; Popov and Black 1981; Zayas; 1980) have shown that stockier braces could exhibitinelastic post-buckling behavior with plastic hinges forming near their mid-span and possibly at theirends, depending upon their support conditions. Supplementary energy can therefore be dissipatedthrough inelastic bending in these hinges when the buckled braces are shortened further after bucklingand thereafter straightened up upon load reversal. Such inelastic behavior also results in a smoothertransition between the buckled and tensioned states and the undesirable impact loads observed intension-only systems are then eliminated.These studies also revealed, however, that while the tensile resistance of the braces remains fairlyconstant under inelastic cyclic action, their compressive strength deteriorates after the initial buckling,as more deformation is imposed, and upon re-application of the compressive load in repeated cyclesof loading. The higher rates of strength degradation were observed immediately after the first bucklingand after the first cycle of loading, respectively. Such weakening has been attributed to the residualmid-span lateral deflection induced by previous compressive cycles as well as to the reduction of themodulus of elasticity of the steel (Bauschinger effect) in the plastic hinge regions.In these studies, it was also found that bracing members, while stockier, still accumulate residualelongation upon cyclic loading, due to the limited amount of axial inelastic shortening, and thereforeexhibit a more pronounced laterally deformed shape after each cycle. Occurrences of local bucklingin plastic hinge regions have also been observed due to the high compressive axial and bending strainsexperienced in these zones. Local buckling leads to a reduction of the bending capacity of the bracesand thereby of their energy dissipation capability.13The S16.1 recognizes the benefits of using stockier braces and allows bracing systems with tensionand compression acting braces to classify under the DBF category (R = 3.0), provided that suitablebracing configurations are used and that the bracing members are adequately sized. For example, tomitigate the possible effects of the asymmetrical response of the braces (one-sided structures) andto provide the system with some redundancy (Redwood and Channagiri 1991), the storey shear hasto be nearly equally shared between tension and compression braces. In Fig. 12, bracing schemeswhich meet the S16.1 requirements for the DBF category are identified. To ensure a minimum levelof energy dissipation by the bracing members, limits have been included in the S16.1 for the maximumoverall brace slenderness ratio and minimum width-to-thickness ratios are prescribed for the flatelements of the brace cross-section (minimum compactness).For the DBF category, a reduced factored compressive strength also has to be considered in thedesign of the braces, unless the extra capacity of the tension acting braces can compensate for thereduction. Specific capacity design based provisions are also given for the design of the non-criticalelements within the bracing bents (connections, beams and columns). These have to be sized forbrace forces equal to the actual brace capacity, including the compressive strength degradation ifmore critical, together with the specified gravity loads acting on the structure. However, these forcesdo not need to exceed twice the prescribed seismic brace loads plus the effects of the specified gravityloads, which approximately corresponds to designing for seismic loads computed with R equal to 1.5.This limitation makes some sense since buildings designed with a R factor of 1.5, with no specialdetailing, are implicitly assumed to possess sufficient inherent ductility to survive adequately designlevel earthquake ground motions.2.2.2 Weaknesses and disadvantagesThe DBF framing category would obviously represent the logical choice for a CBF, either for takingadvantage of the lower prescribed seismic loads or simply to enhance the ductility of the frame incase of extreme ground motions. However, such application is most likely to be limited in practicebecause of the extra costs associated with a ductile system, as discussed in section 2.1, and, perhapsmore important, because the seismic performance of the system still remains questionable. The formeris mainly due to the brace overstrength likely to be present in DBF framings. The second argumentmainly involves the efficiency and the reliability of the bracing members as the energy dissipating14mechanism. In addition, the low redundancy of CBFs also represents a serious matter of concernfrom practitioners when selecting a lateral load resisting system. These three aspects are commentedin the following.Accounting for brace overstrength:Typically, the factored storey shear resistance in CBFs of the DBF category, as provided by the braces,exceeds the prescribed design storey shear. The discrete choice of sections, drift limitations,requirements for wind loadings, limitations on the brace slenderness or on the width-to-thicknessratio, etc. contribute in selecting larger braces than required for seismic loads. Further, the actualultimate storey shear capacity that can be developed upon large inelastic storey shear drift alsogenerally exceeds the value obtained by dividing the factored storey shear resistance by the resistancefactor. Higher brace resistance in tension than in compression, shedding of the brace gravity loads,if any, to the columns, actual yield stress higher than the nominal value, strain hardening, increasedyield stress due to dynamic loading, lower actual effective length for compression braces, etc. maycontribute to such second level of overstrength.For a symmetrical DBF framing having its factored storey shear resistance equal to the appliedfactored storey shear, it can be shown that the ratio of the brace yield load to the brace load dueto the factored loads typically varies between 1.25 to 2.75 for stocky and slender braces, respectively(assuming Fy = 300 MPa and factored compressive strength as per S16.1-M89, ci. 13.3.1). For thesame frame, if only the contribution of the reserve strength of the tension braces is considered, theratio of the ultimate storey shear capacity to the applied factored storey shear then varies between1.11 and 1.75 for stocky and slender braces, respectively. These ratios can be much higher when theother aforementioned design criteria come into play (Osteraas and Krawinlder 1990; Tremblay et al.1991a, 1991b). In the case of low-rise structures or in the upper stories of higher structures, valuesup to 10 can easily be encountered for the second ratio.It can therefore be seen that, in many cases, the brace forces to consider in the design of the noncritical eiements of the frame are likeiy to exceed the upper bound established by the S16.1. In suchsituations, there is no economic advantages in designing for R equal to 3.0, especially when consideringthe other costs involved for providing a ductile system (more laborious design, additional detailing,more careful checking, etc.) and, in order to be consistent in the design, when the definition of thenon-critical elements is extended to include the other structural components of the lateral load resisting15system for which inelastic response is undesirable: floor diaphragms, diaphragm to steel framingconnections, collector beams which drag the loads to the bracing bents, anchorage to the foundationsand the foundations themselves. If those elements were not designed as strong elements, the intendedenergy dissipation in the braces could not be mobilized.In the future, this economical argument probably will have wider consequences since the upper limiton the brace forces is now seen by many as being unconservative (Carpenter and Tavangar 1990;Osteraas and Krawinkler 1990; Redwood et at. 1991; Tremblay et at. 1991a, 1991b), and higher valuesare currently used in other codes. For instance, in the New Zealand Standard (SANZ 1989), thislimit corresponds to the forces generated by fully elastic response of the frame. Further, no over-strength factor needs to be applied to the brace resistance to account for the strain hardening, asthis is the case for the Ductile Moment-Resisting Frame system for which critical elements have tobe assumed to apply relevant loads to the other elements equal to 1.2 times their unfactored yieldresistance. Therefore, it is thought that higher brace forces may be prescribed for the design of thenon-critical elements in future editions of the S16.1 (Redwood and Jam 1992), which would likelyrestrain more the use of the DBF framing category.Bracing members for dissipating energy:The energy absorption and dissipation capacity of a bracing member is closely related to its compressive strength. As it can be observed on typical hysteresis obtained experimentally (e.g. Popov andBlack 1981; Perotti and Scarlassara 1991), this dependance increases upon cyclic load reversals as thebrace gradually elongates. Therefore, even in framing systems for which the storey shear is resistedby tension and compression braces acting in parallel, the efficiency of the system in dissipating energybecomes strongly related to the compressive strength of the braces as cyclic loading progresses.Consequently, this efficiency is adversely affected by any degradation of the resistance in compressionof the braces.It has been shown that even quite stocky braces can experience significant degradation of theircompressive strength upon increasing negative axial deformation applied past the initial buckling orupon application of cyclic loading. For example, post-buckling strength ranging from 30 to 60% ofthe initial compressive strength have been measured after as few as 2 or 3 cycles of loading for braceswith a kl/r between 60 and 100 (Popov and Black 1981; Jain et at. 1978, 1980). Therefore, the capacityto dissipate seismic input energy of DBF framings may be well below that of an elasto-plastic system16designed for the same load, and it will likely degrade substantially with the number of cycles (Fig.2.2). Such performance will be more remarkable if one makes use of the extra capacity exhibited bythe tension brace and does not consider the prescribed reduced post-buckling strength for the designof the braces.The reliability of the mechanism can also be questioned, especially when the bracing members aremade from rectangular tubular (RHS) or double angle sections. Early cracks and fractures wereobserved under repeated axial loads in regions of plastic hinging of RHS braces (Foutch et a!. 1987;Fukuta et a!. 1989). This phenomenon has been attributed to the high concentration of strains causedby local buckling and to the reduced ductility of the material resulting from the cold forming process.The limit on the wall-to-thickness ratio prescribed by the S16.1 can reduce considerably the severityof local buckling and thereby improve the fracture life of bracing members. However, this limit isstill 35% higher than the value recommended by Tang and Goel (1987) for RHS. Indeed, a recentexperimental study on X bracing made from rectangular tubings conforming to current specifications(Kullman 1993) revealed the occurrence of premature fracture of the braces in the hinge regions.Suggestion has been made to impose stringer limits in the future (Redwood and Jam 1992). Fillingthe tubes with plain concrete has also been proposed for reducing the local buckling (Kawano andMatsui 1988; Liu and Goel 1988). It is thought, however, that using either thicker wall tubes orconcrete-filled RHS is only a means for delaying fracture rather than a guarantee against its occurrence. Indeed, Tang and Goel formulated an empirical equation to predict the fracture life of rectangular hollow sections, based on their cross section and effective length, and failures of filled tubeswere observed by Liu and Goel during their experiments.Similar behavior has also been observed in built-up bracing members for which additional provisionsin the S16.1 imposes a closer spacing of the stitch connectors in order to delay the occurrence oflocal buckling. In the case of back-to-back double angle members designed to buckle in the plane ofthe outstanding legs, recent experiments (Aslani and Goel 1991) however demonstrated that membersdesigned according to these S16.1 requirements still exhibit severe early local buckling followed byfractures.Such fracture of braces may be crucial as its occurrence in actual structures subjected to sustainedground motion may lead to the collapse of the entire framework. To the author’s knowledge, no suchbehavior has been reported for W (or I) shaped braces. However, it is worth mentioning here that17the two types of bracing members discussed here have been, and still are used extensively, if notexclusively in low- and medium-rise buildings, because of their high efficiency in compression (HSS)and their simple connections to gusset plates (double angle). The utilization of the latter type is,however, generally restricted to small buildings because of its limited capacity in compression.The performance of the bracing members may also be strongly influenced by their connections tothe framework. Unfortunately, so far, very little guidance has been made available to engineers inorder to guarantee a proper behavior of these connections. Existing guidelines mainly describe thedesired performance rather than including specific design provisions (e.g. AISC 1990). Further, theyare mostly based on tests performed on rather small brace-gusset assemblies (e.g. Astaneh-Asl andGoel 1984; Astaneh-Asl et al. 1985; Goel and Xu 1989). Therefore, there also still exists a significantlevel of uncertainties as to the efficiency and the reliability of this important link in CBFs.Low redundancy of CBFs:Redundancy in a lateral load resisting system means that the system includes surplus elements forresisting the loads. This generally implies that alternative load paths exist in case of failure of oneof these elements, which may help preventing the entire failure of the system. Obviously, this is adesirable feature for earthquake resistance when one considers the uncertainties related to the intensityof the ground motion and the high demand expected for some of the components of the system.Redundancy often also leads to an overall overstrength of the lateral load resisting system, that is anactual higher total capacity beyond the prescribed minimum seismic design strength, which then resultsin tougher structures.These two expressions of the redundancy in structures are of prime importance in classifying lateralload resisting systems for R values (BSCC 1988). For example, MRFs exhibit both features and qualifyfor R equal 4.0. On the other hand, concentrically braced structures typically include very few bracingbays because of their inherent efficiency in resisting lateral loads. Unless the drift or the concentrationof forces at the foundation level becomes a problem, there is no economical justification for providingmore bracing bents than merely required. Most often, architectural constraints restrict the numberof elements to the bare minimum required to ensure in-plane torsional stability.Further, in each bracing bent, lateral loads are resisted by only a few elements. The worst caseobviously corresponds to tension-only bracing configurations (R = 1.5 or 2.0) which exhibit very little18overstrength. For the DBF category, the fact that the load is shared between tension and compressionbraces together with the resulting degree of overstrength discussed earlier contributed in classifyingthis system under R equal to 3.0.Thus, the inherent low-redundancy of CBF structures is already partly accounted for by the lowervalues specified for the R factor. Nevertheless, it must be kept in mind that they rely on very fewelements acting mostly in series (bracing members, bracing connections, anchorage to the foundations,etc.). Failure of any of these elements as a result of defects or low-cycle fatigue (brace fracturediscussed above) can have catastrophic consequences. it is expected that the concept of systemreliability will become a more important matter in design in the future (Schiff et al. 1989) and thatmore specific provisions related to the safety will be incorporated in codes (Galambos 1989). Theintroduction of a factor, which would penalize systems with critical non-redundant elements and rewardthose redundant structures, may be even less favorable to CBFs.Another consequence of the low redundancy of braced frames is the rapid formation of a collapsemechanism (BSSC 1988; Elsesser 1987; Mueller 1984). In tension-only configurations, for example, acollapse mechanism is formed instantaneously upon yielding of the braces since the system has nooverstrength. As the bracing bents exhibit very little capability for redistributing vertically the lateralloads after occurrence of inelastic action in the bracing members at a given level, large interstoreydrifts may take place at that level, regardless of the response at other levels. Such situation may leadto instability and collapse of the structure.In DBF framings, the phenomenon is somewhat delayed by the presence of the tension-acting braceswhich continue to offer resistance to storey shear deformations past the buckling of the compressionbraces. Upon yielding of the tension braces, however, the conditions favorable to dynamic instabilitythen prevail. The lower the level of storey shear overstrength, the more likely and more rapidly theformation of a complete collapse mechanism will be. Therefore, the problem is to be more acute atthe bottom stories of taller structures where stockier braces are used, the design of which beinginfluenced to a lesser extent by the other aforementioned design criteria that could lead to highoverstrength.192.3 Improving the seismic response of CBFsAmong the three shortcomings exhibited by CBFs which were discussed in the previous section, it isthe one related to the poor inelastic response of the bracing members that has mobilized most ofthe research efforts devoted in the last two decades for enhancing the seismic response of CBFs. Incomparison, the problem of reducing the unnecessary high overstrengths in order to make the systemmore attractive economically has not received any direct attention from researchers. It has beenaddressed indirectly, however, within the works carried out on the inelastic performance of bracingmembers. Similarly, the problems associated to the low redundancy of CBFs did not involve significantresearch efforts. The main contributions on that topic were studies on the possible benefits ofcombining CBFs to other lateral load systems, typically MRFs.The main fmdings of the works on the energy dissipation mechanism and on the behavior ofCBFs-MRFs combined systems are presented in the following sub-section. Subsequently, the possibleadvantages and disadvantages of a FCBF system are compared to these recent developments.2.3.1 Enhanced energy dissipation mechanisms2.3.1.1 Improving the inelastic response of the bracing membersOne of the avenues was to solve the problem directly by searching means for improving the performance of the braces themselves. Most of the fmdings have been included in recent codes and werealready presented (limiting the brace slenderness, maximum width-to-thickness ratio, etc.). Althoughthis area represents a very active research field and more refmed code provisions are expected in thefuture, it is thought that the mechanism of absorbing and dissipating energy through cyclic yieldingand buckling of ordinary structural steel shapes can hardly be improved to reach a level of reliabilityand efficiency sufficient to allow lower seismic design loads and encourage wider acceptance andbroader utilization.In this line of thought, an alternative approach has been proposed by Chen and Lu (1990) whichconsisted in preventing the buckling of the bracing members, still the main energy dissipatingmechanism. The braces were made of steel plates inserted in a precast concrete wall panel and werefree to slide. Two different materials were used for the braces to ensure proper inelastic behavior.Although the experimental and analytical results demonstrated the efficiency and the reliability of the20mechanism, the system requires a concrete wail panel as well as rather complicated detailing, bothfor the braces and the concrete panel. This makes the system less attractive for broad use. Moreover,the durability and the long term behavior of the sliding mechanism which involved epoxy painting,silicone grease and shrinkable tubing would require further investigation. Substitute means for dissipating energyA second route that has been followed to overcome the low efficiency and reliability of the bracingmembers was to avoid the problem by using an energy dissipation mechanism other than the braces.Basically, these mechanisms involve inelastic action in i) other existing structural components of theframework or ii) in dedicated devices inserted within the bracing members or between the bracingmembers and the framework. In both cases, the bracing members no longer are the critical elements.Numerous systems have been proposed for each scheme, and only some of the most representativeones are presented and commented in the following.In the first group, the eccentrically braced frame system (EBF) is reviewed in some details as itexhibits interesting features which were considered in the development of the FCBF system. A similarsystem, the Ductile Link CBF (DLCBF) is briefly introduced. For the second group, theDisposable-Knee-Bracing (DKB) system, energy absorbing devices located within the braces, theY-shaped braces (YBF) and ADAS systems are presented. Viscous dampers are also briefly discussedfor sake of completeness. All these systems are schematically illustrated on Fig. 2.1.EBF and DLCBF systems:The EBF system introduced by Roeder and Popov (1978) is a braced frame where one end of thebraces intentionally connects the beams away from the columns. Under strong ground motion, theshort length of beam, called the weak link, acts as the critical element and yields either in bendingor in shear, depending upon its length. Shorter links are preferred since they produce stiffer framesand yield predominantly in shear, which has been found as a more efficient mechanism for dissipatingenergy (Popov et a!. 1987). In chevron bracing, the weak link is formed by an eccentricity deliberatelyleft along the beams between the opposing braces. The EBF system has been studied extensively andspecific design provisions have been included in current codes. It is now commonly used in activeseismic regions.21Tests performed on link samples and reported by Kasai and Popov (1986) showed that links behavein a fairly stable manner under increasing cyclic shear deformation, without any significant strengthor stiffness degradation, until the imposed shear deformation reaches a point beyond which bucklingof the web occurs together with a deterioration in the capacity. Stable post-buckling inelastic behaviorwas however observed if the amplitude of the deformations was reduced. The critical shear angle isfunction of the aspect ratio of the link and the slenderness of the web, and can therefore be offsetby the addition of web stiffeners within the link.The EBF system exhibits a similar degree of redundancy to CBFs. Under monotonically increasinglateral loads, the behavior of EBFs is very much similar to that of a tension-only CBF system, i.e.nearly elasto-plastic if strain-hardening is neglected. Based on previous considerations, CBFs of theDBF category therefore typically exhibit at least the same level of overstrength than EBFs do.However, due to the better response of the critical elements, the system has been assigned a R factorequal to 3.5 in the NBCC.Among the other code requirements, it is worth mentioning that limitations on the maximum linkbeam rotation expected under the design ground motion level are imposed in order to avoid bucklingof the link web. The expected link rotation is directly related to the maximum expected storey drift,the latter being taken as the elastic drift under the prescribed seismic loads multiplied by the R factor,as prescribed in the NBCC. Also, for the design of the bracing members, a design brace load equalto 1.5 times the one that corresponds to the factored resistance of the link has to be considered.Although the system is expected to better resist earthquake induced ground motions, it exhibits somedisadvantages when compared to CBFs: it requires a additional design efforts since lateral loads nolonger induce only axial loads in the members; larger bracing members are required because thedesign brace forces have to account for the actual strength of the link and to include the aforementioned bending moments (Gillies 1991); larger beams may be required to sustain the effects ofthe prescribed seismic loads; larger columns are needed if rigidly connected to the beams, as this isrequired for the configuration shown in Fig. 2.1; complex and critical detailing, especially within thelink and at the rigid beam-to-column connection when needed; expected damage to the gravity loadcarrying system (beams) after an earthquake; reduced stiffness due to the eccentricities in the membersof the bracing bent; critical storey drift beyond which the strength of the system deteriorates.22The DLCBF system proposed by Carpenter and Tavangear (1990) is very similar to the EBF systemas existing beams in the frame act as weak links (in flexure in this case). The behavior of the systemresembles that of coupled shear walls (Paulay and Priestley 1992). The link beams can be detailedas in MRFs in order to behave in a ductile manner. Since they are rigidly connected to the columns,substantial bending moments are induced in the columns. The braces can however be designed foraxial loads only, as in CBFs. For the lateral stiffness, this system is more efficient than EBFs (noeccentricity) and standard CBFs (multiple bracing bents joined one another in shear). it is thereforemore suitable for taller structures. For low- and medium-rise structures, it is thought that the additional costs due to the need for stronger beams and columns, as well as their connections, wouldmake the system not viable.DKB system and energy absorbers:Although the details are somewhat different, the basic principle of the Disposable Knee Bracingsystem (Aristizabal-Ochoa 1986) and of systems with energy absorbing devices (internal small framesor rings) introduced within the diagonal bracing members, is the same that of EBFs and, consequently,their overall response would be very similar. In these systems, however, energy is to be dissipatedthrough inelastic bending of the additional members acting in series with the braces.Inserting energy absorbing devices within the bracing members presents the advantage, when compared to the other systems, that no bending moments are induced in the beams and columns. Furthermore, when using an internal rectangular frame as shown on Fig. 2.1, the deformation mode ofthe internal frame is such that light tension-only braces could be used without experiencing theproblems associated with such bracing members in conventional CBFs. The latter only applies forlow-rise or small buildings since the size of the braces required in larger structures will be such thatcompressive loads will likely develop in the braces. Therefore, braces will have to be designed astension-compression members in order to guard against premature failure.The DKB system has not yet been studied experimentally but an analytical investigation by Wijantoet al. (1988) confirmed the potential of the system for low-rise structures. These authors also reportedon tests performed on energy absorbers which indicated a very stable hysteretic behavior over sustained cyclic loading. Other tests on energy absorbers have been carried out by Jurukovski andSimeonov (1988). In this experiment, the ring was a rectangular rigid frame made from square RHSsections and was about one third the dimensions of the bracing bay. The four braces, also square23tubes, were welded to the corners of the ring. Although the authors claimed for a very high efficiencyin absorbing and dissipating energy, local buckling and cracking of the tubes of the ring were observedwhich led to significant deterioration in strength and stiffness (severe pinching in the hysteresis) afteronly a few cycles.Even if both systems may at first appear very attractive, these tests warn that standard design andfabrication procedures are very unlikely to result in a desirable performance, especially when oneconsiders the type (flexural) and the relatively high concentration of inelastic demand in the yieldingelements. It is believed that considerable research is still required before such systems could beimplemented in the industry. Further, unless tension-only braces are used, the brace end connectingthe yielding element would require some support in order to prevent instability of the system in thedirection normal to the plane of the bracing bent. Efficient out-of-plane support will most likelyconflict with architectural or operational functions of the building.Y-shaped bracing and ADAS systems:Among the different bracing configurations (Fig. 12), the inverted V single bay chevron scheme ispreferred by many because of its many advantages: high storey shear strength and stiffness uponlinear response (White and Salmon 1987; Tremblay 1991); braces connect at mid-span of the beams,which means that beams have an intermediate support for gravity loads, a shorter unsupported lengthand reduced axial loads due to lateral loads; easier fabrication because its requires fewer and simplerconnections; allows door openings in the bracing bent. Unfortunately, it exhibits a poor performancewhen braces undergo cyclic inelastic action. The beams cannot sustain the vertical unbalanced forcesthat develop after buckling of the braces, and the storey shear strength and stiffness deteriorate rapidly(Khatib 1988).To improve the response of this bracing configuration, shear link devices were proposed to transferthe horizontal storey shear between the beams and the braces at the intersection of these members.Seki et al. (1988) studied the behavior a Y-shaped braced frame (YBF) system which includes a steelpanel that yields in shear. The performance of such system as obtained by tests, is very similar tothat of the EBF system: a stable hysteretical response with the ultimate capacity governed by bucklingof the shear panel element.24Another device, made from hourglass-shaped mild steel plates mounted in parallel which yield inbending, the ADAS structural element (Scholl 1990), serves the same purpose. The latter, which wasoriginally developed for nuclear power plant piping supports (Stiemer and Godden 1980; Stiemer eta!. 1981), is seen more as an additional element for increasing the overall damping of new or existingstructures which already have their own lateral load resisting system. Extensive testing of the systemhas been carried out (Whittaker et a!. 1989b) and the device was shown to exhibit a very stablehysterestic response upon sustained cyclic yielding reversals. Failure of such device eventually occurredby fracture of the plates after many cycles of large inelastic deformations. At maximum displacement,the strength developed by tested devices reached approximately 1.8 times their nominal yield load,which gives an indication on the design loads to be considered for the surrounding non-critical elements. Such device has already been used in practice (Fierro 1992).As for the previous systems, both systems would exhibit a nearly elasto-plastic response undermonotonically increasing lateral load when used in CBFs. The ADAS device was found, however, toexhibit some post-yielding stiffness that was mainly attributed to strain hardening. Because of the sizeof the devices, significant bending moments may be induced in the beams and a lateral support islikely to be required at the base of the elements in order to prevent out-of-plane instability. The useof the two systems is limited to chevron bracings but could alternatively be installed in K-bracingswith the devices inserted vertically at the brace to column connections.Viscous dampers:The addition of shear-type viscous dampers to braced frames have also been studied (Aiken et a!.1990; Bergman and Hanson 1988; Soong and Mabmoodi 1990; Zhang and Soong 1989). The dampersconsisted of two rectangular pads of viscoelastic material, sandwiched between three steel plates,which are loaded in shear. Dampers are commonly installed at one end of the bracing members inorder to act in series with them. Thus no bending moments are introduced in the beams and thecolumns. As for the ADAS system, viscous dampers with the bracing members are mainly installedin structures which already have another lateral load resisting system. Experimental investigations andnumerical simulations have shown that the relative displacements and the absolute accelerations canbe reduced significantly by using such dampers.25In all the systems presented earlier, hysteretic damping was to be produced by yielding of steelcomponents only during severe earthquakes. The main advantage of viscous dampers is that theadditional damping is available at any load level, and thus can mitigate any vibrations induced excitation the building can experience: human activities, machinery, wind, traffic, any level of earthquakeground motion, etc.However, the material is sensitive (strength and stiffness) to temperature (strongly), excitation frequency, shear strain level and, in certain cases, to cumulative energy dissipated. Therefore, the designof such system cannot be performed on the basis of a simple static analysis of the structure, as thisis the case with the previous systems, and would unlikely be performed within normal design officeprocedure. Also, because of the variability in the strength, the prediction of the maximum strengthdeveloped by the dampers for the design of the non-critical elements would require some judgement.Summaiy:The following features of the reviewed systems with substitute energy dissipation mechanisms can besummarized with respect to the three shortcomings of CBFs. Viscous dampers were not included asthey will most likely be used in parallel with other systems.Overstrength. In all systems, the critical elements can be sized to merely withstand the effects of theprescribed seismic loads. Therefore, for the non-critical elements, the extra-strength to consider onlyincludes the difference between the actual and the factored resistance of the critical elements. Asseen, such values will typically vary between 1.5 and 1.8, which is comparable and, in some cases,less than the values encountered for CBFs. In some systems, design of the critical elements may haveto include forces due to gravity loads (e.g. beams of EBFs and DLCBFs) which will be redistributedto other elements upon yielding of the critical elements. Such loads may result in some additionaloverstrength.Energy dissipating mechanism. In all the tested systems, the critical elements were found to exhibit adesirable stable hysteretic behavior. Beyond a given level of deformation, however, failure occurs andthe strength and stiffness degrade rapidly. Similar behavior is expected for systems that have not beentested.26Redundancy. All the systems exhibit a degree of redundancy similar to that of CBFs, with the totallateral loads being resisted through a single chain of elements. In all cases, the response undermonotonicafly increasing loads is nearly of the elasto-plastic type, with some post-yield stiffness dueto strain hardening. In fact, CBFs of the DBF category, with both tension- and compression- actingbraces, would probably exhibit higher post-yield storey shear strength than any of the systems reviewed.Compared to CBFs, these systems present other advantages and disadvantages:Advantages. In all systems, post-earthquake damages are expected to be concentrated in the devices,and beams, columns and braces would be preserved. In all cases, the storey shear strength and stiffnessare partly uncoupled since the strength of the critical elements can be chosen independently of thesize of the braces (braces should however be large enough to sustain loads associated with the strengthof the critical elements). This allows more flexibility in the design.Disadvantages. All systems require an increased design effort. Additional material and/or detailing isalso required for the energy dissipation mechanisms. In all systems but the one with energy absorbingdevices within the braces, heavier members are required in the bracing bents due to the bendingmoments induced. In all cases, the design load for the braces is higher and the stiffness of the bracingbent is reduced. Some systems (DKB, energy absorbers, YBF) may need encumbering out-of-planesupports in order to ensure the stability of their energy dissipation mechanisms.2.3.2 Combining CBFs to MRFsCombined systems may be formed by individual systems connected through the floor diaphragm, suchas perimeter MRFs acting together with bracing bents located within the building, or by co-planardual systems where the bracing bents are part of the MRFs (Amin et al. 1990).The advantages of combining CBFs and MRFs have been recognized for many years: more reliableand stable response due to the efficiency of the CBF system in providing lateral stiffness and strengthtogether with the higher redundancy and superior hysteretic behavior of the MRF system. As mentioned before, both systems behave as shear beams upon inelastic action. Because of its higher shearstiffness, the CBF first undergoes inelastic response whereas the MRF still provides some post-yieldstrength and stiffness (Carpenter and Tavangar 1990). Such behavior may help in mitigating large27interstorey drifts and the likelihood of instability. In multi-storey structures, the continuity betweenthe members of MRFs also contributes in having a more uniform vertical distribution of the ductilitydemand throughout the height of the building (SEAOC 1990).The traditional design approach for dual systems has been to consider the MRF as a back-up system.The lateral loads were assumed to be entirely resisted by the CBF, whereas the MRF was designedfor the loads distributed to each system in proportion to their relative stiffness. A minimum participation of the MRF was specified (25%) to ensure a minimum degree of redundancy. For such asystem, the design seismic loads were lower than those specified for a CBF acting alone. Althoughthe complexity of the framework was increased by incorporating a MRF, which implied additionaldesign, fabrication and instaliation efforts, the expected improvements in the seismic performance ofbuildings were obtained at a reasonable cost. In many cases, the overall cost was even reduced becauseof the lower seismic loads and because the participation of the MRF was minimum and not basedon stiffness criteria.Although still in use in the other countries (SEAOC 1990), such approach is no longer allowed inCanada. In order to take advantage of reduced seismic loads, R equals 4.0 for example, the MRFhas to take all of the lateral loads and to incorporate all the detailing prescribed for a ductile response,exactly as if the CBF was not to contribute to the energy absorption and dissipation during a severeground motion. Such design approach will undoubtedly result in safer structures but the cost certainlywill exceed that of a MRF or CBF acting alone. In absence of inciting savings, it is very unlikely thatsuch a combined system be broadly used.Findings of recent studies performed on co-planar dual systems with chevron bracing systems (Khatib1988; Whittaker et al. 1990) also tend to support such design approach. In both cases, importantdynamic interactions were observed between the two systems when subjected to ground motions andthe concept of statically adding the contribution of the two systems was invalidated. Further, Whittakeret al. formulated rather severe recommendations when compared to the traditional design approach:higher strength for the design of the CBF (approximately R equal to 1.5), in order to limit the inelasticresponse in the braces; MRFs to be designed for 40-50% of the CBF base shear, in order for theMRF to appropriately supplement the CBF in case of brace failure; minimum stiffness for MRFs(up to 50% of that of the CBF) in order to increase the inelastic action in the MRFs at low levelof drift.28These recommendations seem fairly conservative since both the MRF and the CBF would have tobe designed for higher loads than those currently prescribed for the individual systems. Nevertheless,they clearly demonstrate the increasing concerns about the seismic response of CBFs and the needsfor increasing the participation of MRFs to the response of combined systems. As said, such approachis unlikely to make the system economically viable. Further, a higher contribution of MRFs in theenergy dissipation would also result in structural damage extending to the gravity load carrying systemof the building (beams and/or columns), as opposed to having only the braces damaged in the caseof CBFs acting alone.2.3.3 Friction Concentrically Braced FramesA braced frame with sliding brace friction connection is schematically shown on Fig. 22. Such FCBFsystem enters in the category of systems with substitute means of dissipating energy discussed in theprevious section. Basically, it is almost the same system as the one with viscous dampers. The onlydifference is that no viscous material is provided between the connecting plates. When subjected toground motion, forces are induced in the braces which eventually provoke slip of the connection andthereby energy dissipation through friction between the connecting plates.In an ideal FCBF, the slip load would remain constant upon cyclic loading (Fig. 2.2). Assuming sucha behavior, the following comparisons can be made:Overstrength. Like this was the case for the previously reviewed systems, each connection can be givena slip resistance which closely matches the effects of the prescribed factored storey shear. As somemaximum expected value of the slip resistance will have to be considered in the design of the noncritical elements, the connection overstrength will mainly depend upon the degree of variability ofthe slip resistance of the sliding mechanism. The connection overstrength would also be influencedby the effects of the gravity loads, if any, as in the previous systems.Eneigy dissipation mechanism. Stable behavior can be expected if proper friction mechanism is provided for. Within practical limits, it should be possible to provide the mechanism with the desiredrange of slip distance. This means that, as opposed to the other systems, there would be no limitationon the available ductility as imposed by the materials or the mechanism. A sliding connection, oncedesigned and built, will likely have, however, a defmed range of utilization beyond which the slip loadcould not be maintained constant.29Redundancy. A FCBF system would exhibit the same level of redundancy as any of the systemsdiscussed previously, with no post-yield strength. (In the text, the slip load of the connection is alsoreferred to as the yield load of the system as most people are familiar with that terminology andbecause some of the results of this investigation may be used for other systems with yieldingmechanisms. For instance, the subscript y is used when referring to any action, load or force, ordeformation at the onset of slip, and so on.)The system also presents the following advantages:As in CBFs, seismic loads will primarily induce axial loads in the members of FCBF bracing bents.The presence of sliding connections does not adversely affect the storey shear stiffness of the framing,as the other substitute systems do. Further, should a simple sliding mechanism be used, the additionaldesign and fabrication efforts as well as the amount of material required for the energy dissipationmechanism will be negligible when compared to conventional CBFs. Obviously, higher brace designloads will have to be considered in the design, as this is the case for the other systems, since thebraces no longer are critical elements.The energy dissipation mechanism is rather simple as there is no inelastic response of the material,thus no yielding, strength deterioration, Baushinger effects, strain-hardening, etc. The relationshipsbetween the overall ductility, the local or member ductility and the maximum slip are also straightforward. In order to be consistent with currently used defmitions, the local ductility for a brace canbe defmed as the sum of the maximum slip experienced in the connection and the axial elongationof the brace at slip, divided by the brace axial elongation at slip. From there, the expected slip distancecan easily be obtained knowing the expected storey drift or the global ductility of the system.As in the other systems, possible damage in FCBFs is expected to be limited to a few elements only:the faying surfaces of the sliding plates. Thus, no structural member is expected to suffer and aproperly designed sliding mechanism should not need to be replaced even after a severe groundmotion. Residual storey drifts after an earthquake are expected though, as if other yielding systemswere used, but these could be rectified subsequently.The system can be used either for new buildings or for retrofitting projects. In the latter case, slidingconnections could even be installed in an existing bracing, without affecting the surrounding framework. The system is also applicable to any bracing configurations, except tension-only systems because30the response would be similar to that of tension-only systems with yielding braces discussed earlier.Since the sliding connections can be given any resistance to slip, the system will suit any size ofbuilding.The system may however exhibit the following shortcomings:Many of the above statements hold only if the system exhibits a stable response upon successivereversals of sliding, which means without any significant alteration of the faying surfaces nor variationof the clamping force. Further, the resistance to slip has to be maintained uniform for the expectedlifespan of the structures, which means that the behavior of the connections should not be influencedby factors like creep, relaxation, temperature, corrosion, etc.The FCBF system still exhibits very low redundancy. As compared to CBFs, however, the expectedimprovement in the efficiency and the reliability of the energy dissipation mechanism would certainlycontribute in obtaining safer buildings, comparable to eccentrically braced structures, for example.On the other hand, as EBFs or any like systems, FCBFs will be exposed to concentration of storeydrift and, possibly, to instability problems.31SHEAR (FLEXURAL) LINKECCENTRICALLY BRACED FRAME (EBF)DUCTILE LINKED CBFIKAOEDISPOSABLE KNEE BRACINGSHEAR LINKY BRACED FRAME (YBF) I ADASCBF WITH ENERGY ABSORBERSDAMPER_\CBF WITH VISCOUS DAMPERSFig. 2.1 Braced Frame Systems with substitute dissipating energy mec1mnism.I IFLEXURAL LINK I IENERGY ABSORBERDEVICEII32IC-)ILLUU)0a0aI’ULUC)-__>>t•1>>Chapter 3FCBF: PROPOSED SYSTEM AND DESIGN PROCEDURE3.1 IntroductionThe brace sliding connection is the crucial component of the FCBF system. As stated before, theobjectives in developing such connection were related to its performance: 1) stable response uponcyclic sliding and ii) a constant resistance to slip over time. Furthermore, in order to encourage theacceptance of the system by the industry, the connection would desirably i) be simple to design,fabricate and install, ii) require low cost and readily available material only and iii) exhibit a highslip resistance per unit of material.In order to achieve these objectives, the decision was made to investigate the possibility of using astandard bolted slip-critical connection which would be modified by providing slotted holes in theconnected parts in order to allow slip to take place. This choice was based on the fact that slip-criticalconnections have been used for decades in the steel structures fabricating industry, and that recentresearch work had indicated the potential of using bolted connections for dissipating energy.Bolted slip-critical joints are used to transfer shear loads between steel parts. A clamping force isprovided by tightening the bolts and the load is transferred by friction, without relative slip of theconnected parts. This type of connection had been studied extensively and rather complete designdata was currently available. Numerous tests had demonstrated their efficiency and reliability as wellas their capability to resist sustained loading. Their fabrication, installation and checking only involvenormal procedures already implemented in the industry and the material required (bolts, steel) arestandardized and currently in use.In contrast, the knowledge on the behavior of these connections upon gross slip was scarce. A fewexperimental studies only had been carried out on bolted assemblies subjected to cyclic loading. Ingeneral, these tests showed that good performance could be achieved with very few modifications tothe standard connection configuration.34From there, a proposal was made for a FCBF system which would exhibit the desired characteristicsstated in the previous chapter. Subsequently, tentative design guidelines were prepared assuming thesystem would behave in the expected manner. In order to facilitate their implementation, these werebased on current design procedures. The proposed system and design guidelines are presented laterin this chapter. In this section, a brief description of standard bolted slip-critical joints is given, basicsof friction and wear mechanisms are presented and the research works which were considered in thestudy are briefly reviewed.3.1.1 Standard sUp-critical connectionsThe resistance to slip, V, of standard slip-critical connections is given by (Kulak et al. 1987):V3 = k,mnT (3.1)In this expression, k is the slip coefficient of the faying surfaces, m is the number of slip planes inthe connection, n is the number of bolts and T1 is the bolt preload (per bolt). The slip coefficient,k, essentially is the average coefficient of friction exhibited by the sliding surfaces, i. They slightlydiffer, however, as k accounts, among others, for the non-uniform bearing pressure around the bolts.In practice, k3 is obtained for a given plate material from slip tests where all of the parameters of(3.1) but k are known. For design purposes, an additional factor is introduced on the right-hand-sideof (3.1) to account for the uncertainty associated with the randomness of ks and T1.In typical slip-resistant connections, mild steel plates are used. They can exhibit different surfaceconditions: unfinished (mill scale conditions), sand or grit blasted, polished, painted or galvanized.As per the S16.1, high strength bolts have to be used in these joints. Either A325 high-strengthmedium carbon steel bolts (ASTM 1991b) or A490 alloy steel bolts (ASTM 1991c), which exhibit atensile strength of 830 MPa and 1040 MPa, respectively, comply with that requirement. These boltsrange between 1/2” (12.7 mm) to 1-1/2” (38.1 mm) in diameter but, typically, 3/4” (19.1 mm) to 1”(25.4 mm) bolts are most often used. (Note: Imperial size bolts were still in use in the industry atthe time of this study. In order to make this text easier to read, only the diameter in inches will begiven when imperial bolts are specified).Upon installation, bolts must be pretensioned to 70% of their ultimate capacity, which is referred toas the minimum bolt tension, Tmjn. This can be achieved by using different methods such as the35standardized turn-of-nut tightening method, the utilization of a calibrated wrench, or by usingtension-control, swedge bolts or load-indicating washers. The first method consists in giving the nuta predetermined amount of rotation from a snug-tight position. It generally is the method preferredin practice.Although slotted holes are permitted in order to facilitate erection, punched or drilled circular holes,2 mm larger in diameter than the bolts, are most often used in the connected parts. As per the S16.1,slotted holes, when used, must not exceed in length 2.5 times the diameter of the bolts, and must beentirely covered by a plate washer not less than 8 mm in thickness if the slotted holes are locatedin the outer plies of the joint. Further, based on experimental evidences (Allan and Fisher 1968), theslip resistance of connections with slotted holes is reduced by 25%. Slotted holes are generally flamecut.Extensive research has been performed on slip-critical joints in order to obtain statistical data on theslip resistance for different plate materials and surface finishes as well as for various bolt types andsizes. In the standard static slip tests performed for that purpose, a gradually increasing load is appliedto multi-bolt symmetrical but splice joints including a middle plate inserted between two exteriorplates. For structural quality steels in the clean mill scale condition, for instance, Kulak et al. reporta mean value for k of 0.33 with a standard deviation of 0.07 (more than 500 tests). Clean mill scalecondition corresponds to steel as received from the mill, free of any coating, with the loose mill scaleand dirt removed by hand wire brushing and grease or oil, originating from the fabrication process,removed with a solvent.The variation of the bolt preload, Tj, has also been investigated in order to derive design values forV. Tests revealed that the average value of the initial tension for bolts installed using the turn-of-nutmethod exceeds the prescribed Tmjn by 13% to 35%, with standard variation varying between 6%and 12%, depending upon the bolt size and type and the amount of nut rotation. For 3/4” A325 boltsand steel in the clean mill scale condition, for instance, a factor of 0.82 is to be applied to V3 asobtained by equation (3.1) with the average k3 and T1 equal to Tmjn, in order to limit the slipprobability to 5%.Upon application of the load to a slip-critical connection, it was observed that the load is initiallytransferred by friction only at the ends of the joint, because of the elastic deformation of the connectedelements. As the load is increased, the zone of friction extends toward the center of the connection.36Slip eventually starts at the ends and then progresses inward until the friction resistance is exceededover the entire contact area. The slip load of the connection is then reached and large relativedisplacement, or gross slip, takes place. Sliding stops when hole clearance is taken up and the boltscome into bearing. Similar response was observed regardless whether circular or slotted holes wereused. Joints with slotted holes however exhibited a slip-stick behavior (see next section) upon overallslip (Allan and Fisher 1968).Typical load deformation relationships are therefore nearly linear until slip is initiated at the ends ofthe plates. A that point, the slope gradually decreases as slip progresses toward the center of theconnection, until a plateau is reached when the gross slip load is attained. Finally, the connectionexhibits a rapid gain in strength as the load is resisted in bearing. Failure may occur either by shearingof the bolts, or by tearing or bearing of the connected plates. In the former case, most of the boltpreload is released when inelastic shear deformation takes place in the bolts prior to failure. Therefore,despite some additional tension load is induced in the bolts by lap plate prying action, the ultimatestrength of the connection, if governed by bolts in shear, is not dependent on the amount of initialbolt pretension.3.1.2 Basics of metallic surface interactionIn most structural engineering application, satisfactory performance of steel structures is achievedwhen slip is prevented in the connections. When, on the other hand, a system relies on a propersliding behavior of its connections for surviving strong ground motions, consideration should be givento the surface interaction phenomena likely to alter the response of these connections.In the following, basics of friction mechanisms, dissipation of energy through friction, wear mechanismsand of cold welding are summarized for further reference in the text. More information on thetribology of materials can be found in specialized references (Bowden and Taylor 1950; Rabinowicz1965; Sarkar 1980). It must be emphasized here that, as stated by many researchers in that field,though the theories and descriptions given below generally agree with observed behavior, slidingprocess is rather complex and still remains very difficult to predict as it involves many interactingphenomena.37Friction forces:For metallic surfaces, most of the friction force is attributed to adhesion, or atomic bonding, whichdevelops over small contact areas at the interface, the junctions. Asperity interlocking and ploughingmay also contribute to the resistance to slip. In addition, the friction force may depend whether slidinghas been initiated or not.Because sliding surfaces are never perfectly flat, contact at the interface only occurs at the peaks oftheir asperities. Upon application of a normal load, the metal at these locations deforms plasticallyunder the pressure and the area of contact increases until equilibrium is reached. At each of theseso-formed junctions, an atomic bond is created. During that process, the attractive forces due to thesurface energy of the metals add up to the normal load and contribute in forming the true area ofcontact. If an incremental tangential load is then applied, additional plastic deformations take placedue to the combined action of the normal and tangential forces and the true contact area increasesfurther (junction growth). Consequently, the normal stress diminishes and the resistance to shearincreases. Eventually, the critical shear stress is reached over the extended contact area: the junctionsbreak and overall sliding begins.The strength of the bonds mainly depends upon the affinity of the two metals in contact. Much higherstrength can be developed when the two surfaces are made from the same or similar metals. This isthe reason why low-friction assemblies are often made of dissimilar materials and mild steel, whichis an heterogeneous alloy, exhibits a lower coefficient of friction than pure iron. The metallic bondsare also strongly and adversely influenced by the presence of surface films. Any metal surface exposedto air is first covered by an oxide layer (except for noble metals) and then by an adsorbed gas layer.In addition, regardless how carefully the surface has been cleaned, contaminants such as oil andgrease are most likely to be present on the surface. Very high pressure (normal load) and relativesliding may break, at least partially, those films and make possible direct metallic bonds with higherresistance to shear.The strength of the materials affects but slightly the friction force due to adhesion: if the yield strengthis increased, the critical shear stress also increases but the contact area is reduced. Similarly,strain-hardening which may occur at the junction does not influence the coefficient of friction.38However, strain-hardening in the vicinity of the junctions will likely shift the shear failure planetowards the bulk of the material. When two different materials are used, shearing will most oftenoccur in the weaker metal.The coefficient of friction has been found to be weakly dependent on the roughness of the surfacessince the contact area would be the same whether the surfaces are smooth or rough. However, veryrough and very smooth surfaces may exhibit higher coefficient of friction. In the former case, asperityinterlocking requires additional force since the two surfaces have to be separated to permit the sliding.On the other hand, very smooth surfaces tend to exhibit more pronounced junction growths.Ploughing is another component of the friction that can be of importance in metal sliding. Indeed,significant additional frictional forces can develop if a harder surface with sharp edges produces scoresor grooves in the other one. Similar behavior would also be observed if harder particles are presentbetween the two surfaces. In both cases, this contribution to friction increases with the quantity andsharpness of the surface edges or the particles. It also likely increases as the slip progresses becauseof the accumulation of the material in front of the ploughing edge.Smooth and stick-slip sliding:If the mechanical system that includes the sliding mechanism (building structure in our case) candeform and store elastically energy, a stick-slip behavior likely occurs upon sliding, which is characterized by a saw-tooth load-slip relationship. During the initial loading, before slip has occurred, mostof the work done by the external loads is stored as strain energy in the system. Upon sliding, theload falls and part of that strain energy is suddenly released into kinetic energy. As a result of thisadditional push, slip takes place at a higher rate than the applied rate of motion. Slip eventually stopsand load builds up again in the system. New junctions can form and grow as the load graduallyincreases, until it becomes large enough to initiate relative slip again. The process is then repeated.The difference between the stick and the slip forces diminishes when the velocity is increased, as thetime allowed for the plastic flow to take place is reduced.Stick-slip response can also be observed when rough and/or hard surfaces are used. In such case,the dynamic effect is due to the energy stored while the surface asperities undergo elastic deformations,which is suddenly released when slip occurs.39Energy dissipation:Almost all the work done to produce relative slip of two connected parts is converted into heat. Whenslip-stick sliding occurs, part of it can also be fed in the system on the form of vibrational energy,which produces the vibration and the noise. For practical purposes, however, it can be assumed thatall the energy is transformed into heat at the interface. The heat diffuses from the contact surfacesthrough the bulk (conduction) whose temperature then increases. Ultimately, the heat reaches theouter boundaries of the connected elements where it is transferred to the surrounding air and eniltted(convection and radiation).Therefore, during and after the sliding, the temperature rise in the connected parts will vary in spaceand in time. Obviously, it will be maximum near the faying surfaces and higher values will be observedif the rate and duration of the energy input are increased, if the specific heat of the material (quantityof energy required to increase by one unit the temperature of a unit mass of the material) is lowand if the efficiency of the various heat transfer mechanisms is low.Since the true contact area at the interface is smaller than the apparent contact area, a higher amountof heat has to be absorbed by unit of material near the surface which results in much higher temperatures, called flash temperatures. Temperature flashes are concentrated at the junctions which arecontinually broken and recreated during sliding. Therefore, they last for a very short period of time(as short as i04 sec.) and the surface temperature varies continually along the interface. Thus, evenif no sign of heating is visible or when moderate conditions of load and velocity prevail, very hightemperature can be reached which can affect the sliding mechanism. Flash temperatures increase withthe amount of heat generated, decrease if the thermal conductivity of the material is increased (heatis more easily transferred away from the surface to the bulk of material) and the maximum valuethat can be reached corresponds to the melting temperature of the lower melting point of the twosliding materials.Temperature increases in the connected parts can have a detrimental impact on their structuralproperties. For example, the modulus of elasticity as well as the yield and tensile strengths of structuralsteels decrease when the temperature is raised. Typically, all these properties are reduced by 50%when the temperature reaches about 600°C and the material exhibits less than 10% of its originalstrength beyond 1000°C (Jastrzebski 1959; Salmon and Johnson 1990).40Temperature rise may also affect the friction force. The thickness of the connected parts will likelyincrease whereas their stiffness will diminish. This may influence the bolt load and thereby the frictionforce. The reduction in the yield strength of the plates will result in a larger true contact area butin a reduced critical shear stress, and thus should not affect significantly the coefficient of friction.However, increases in temperature will facilitate the formation of an oxide layer at the surfaces whichwill most likely reduce the strength of the bonds and, consequently, the coefficient of friction.if the melting temperature is reached at the interface only, a soft layer of melted metal will formwith almost no resistance in shear and the coefficient of friction will drop significantly. If, on theother hand, the temperature within the bulk material also approaches the melting temperature, verylarge true contact area will form, together with very high coefficient of friction, and seizure of thesurfaces may occur.Wear:Wear consists in removal of material from the faying surfaces as a result of friction. Many forms ofwear can be distinguished of which adhesive and abrasive mechanisms are the most common.Adhesive wear occurs when material is pulled off one surface and adheres to the other one as aresult of shearing of the material underneath the junctions. if sliding is pursued, the wear particlesmay stay on the second surface, be transferred back to the original surface or detach and becomeloose. The stronger the bonds (alike material, breaking of the surface films) and the larger the truecontact area (higher normal load, softer material, higher temperature), the higher the wear rate.If two different material are used, most of the wear will take place in the softer one. Eventually, thesurface of the stronger material will become covered by fragments of the softer one and strongerinteraction (higher coefficient of friction) will develop as the two surfaces become similar.Abrasive wear occurs when a surface is scored or grooved by a harder surface, or by harder particleslocated between the surfaces. Debris of abrasive wear generally remain on the loose form. Togetherwith those due to adhesive wear, they may contribute further to the abrasive wear mechanism.The thickness of the connected plates will reduce due to wear. Thus, the bolt preload will decreaseand so will the friction force. On the other hand, wear debris may push outward the surfaces and41produce the opposite effect. Further, as mentioned earlier, loose strain-hardened wear debris maycontribute to the friction force (ploughing component) and adhesive wear debris stuck to the surfacemay create asperity interlocking which will increase further the friction force.Pressure or cold welding:The junctions formed upon application of the normal load actually link the two surfaces together.When hard materials like steels are used, the elastic recovery of the surrounding asperities, thosewhich only deformed elastically upon contact of the surfaces, is normally sufficient to break the bondscreated at the junctions. For softer materials, the strength of the junctions may exceed the elasticforces and cold welding occurs. Thus, as for adhesive wear, cold welding will most likely occur inpresence of stronger bonds and/or larger real area of contact. Duration of loading may also contributeas diffusion of atoms from one surface to the other may take place.3.1.3 Previous studies on joints subjected to cyclic loadingStandard bolted connections:Because of the localized slips that take place in bolted connections at a load level below that producingthe overall slip, energy can be dissipated upon cyclic loading at the service load level. Studies havebeen carried out in order to understand and assess this phenomenon and thereafter make use of itfor increasing the damping of structures. For example, Brown (1968) and Metherell and Diller (1968)carried out analytical investigations in order to determine the influence of different parameters onthe damping (geometry, velocity, etc.).Vitelleschi et al. (1977) performed experiments on symmetrical butt splice joints made of structuralsteel plates (Fy = 250 Mpa) connected with one 3/4” A325 bolt. Different faying surface conditionswere considered. Specimens were subjected to cyclic loads ranging from 0 to gross slip load. Fromtests during which the load amplitude was first decreased from 75% to 25% of the expected grossslip load, and then increased until overall slip took place (total of more than 9 000 cycles at 0.6 Hz),the loss in bolt tension was less than 10% of their initial preload for the mill scale, hot-dip galvanizedand primed surface conditions.The fmal peak cyclic loads, and corresponding slip coefficients, were, however, higher than the valuesobtained from reference static tests. The increase was 13% for the mill scale and primed conditions42whereas the galvanized specimen locked-up. During sustained dynamic tests (more than 100 000 cycleswith peak load equal to 75% of the expected slip load) specimens in the clean mill scale conditionwere found to exhibit a stable energy dissipation capability which was not altered by the number ofcycles.Friction damped braced frames:Pail and Marsh (1982) proposed to introduce bracing members with sliding connections in momentresisting steel frames to form the so-called friction damped braced frames. The objective with thissystem is similar to that pursued when using viscous dampers, i.e. increasing the damping, and thusreducing the response, of structures.The idea of using friction devices for dissipating seismic input energy was initially developed forprecast concrete panel constructions (Pail and Marsh 1979, 1980; Pall et aL 1981) and, by extension,to cast-in-place concrete structures (Pall and Marsh 1981). The devices, referred to as limited slipbolted (LSB) joints, were inserted along the vertical joints between adjacent shear wails. Each deviceincluded a pair of steel inserts, anchored on each wall, and a connecting steel plate. The latter wasto be field welded to one of the inserts and bolted connected to the other one. Slotted holes werealso to be provided in connecting plate so that relative slip could take place.Static and cyclic loading tests were performed on samples of such device including 1/2” A325 bolts.Different surface conditions were studied. For the cyclic tests, the amplitude of slip was 10 mm andapproximately 20 cycles were applied. No information is given in the references regarding the numberof bolts, the bolt preload and the frequency of excitation. For the sample in the clean mill scalecondition, the resistance to slip undergoing gross slip was uniform (approximately equal to 40 kN)over the entire displacement. Upon cyclic loading, however, the slip load varied significantly, fromapproximately 15 to 50 kN. The load seemed to degrade past the initial slip and then increase towardsthe end of the test. Also, the maximum slip loads were recorded at maximum slip distance.Although less significant, the specimens with other surface conditions also exhibited a variation ofthe slip load upon cyclic loading. The authors therefore recommended to insert a brake lining padbetween the contact surfaces, in which case the measured response was found extremely satisfactory.For thefriction damped braced frames, Pall and Marsh proposed to use a similar friction connectionat one end of tension-compression bracing members, with the brake lining pad inserted between thegusset plates and the braces. No experimental investigation of this system was performed.43Three-stage friction-grip elements:Roik et al. (1988, 1989) proposed to include three-stage friction-grip elements in structures in orderto improve their seismic response. These elements would be attached to the surrounding frame withsliding connections, including pretensioned bolts and slotted holes, in such a manner that slip wouldeventually occur in the connections upon increasing the storey shear drift. In order to avoid abrupttransition between the non-slip and the slip states, the elements would be acting in parallel and wouldbe given a different stiffness and slip load. For b