DEVELOPMENTS OF THE CURRENT PERFORMANCE-BASED SEISMIC DESIGN CODE FOR BUILDINGS IN TAIWAN

4th Internatonal Conference on Earthquake Engneerng Tape, Tawan October 12-13, 2006 Paper No. 223 DEVELOPMENTS OF THE CURRENT PERFORMANCE-BASED SEISMIC DESIGN CODE FOR BUILDINGS IN TAIWAN Qang Xue 1 Cha-We Wu 2 Cheng-Chung Chen 3 Kuo-Chng Chen 4 ABSTRACT Performance-based desgn has been wdely recognzed as an dealzed method for future sesmc desgn n practce. A draft sesmc desgn code for buldngs by ntroducng performance-based desgn methodology has been proposed n ths research. Frst of all, the current code provsons have been examned accordng to the theoretcal bass of PBSD. Several key ssues have then been nvestgated to clarfy why, what and how to ncorporate the methodology of PBSD n domestc sesmc desgn code currently. In the proposed draft code, transparent sesmc desgn objectves and crtera for buldngs of dfferent use group have been establshed quanttatvely. Drft lmt for each consdered earthquake level s clearly establshed. Ste feasblty requrement, conceptual desgn scope and basc rule have been modfed consderng use group. A performance objectve orented prelmnary desgn procedure s proposed. Drect dsplacement-based desgn procedures have been summarzed n the appendx. Methods and procedures for sesmc performance evaluaton have been presented. Fnally n ths paper, what we have learned through the appled examples wll be summarzed. We beleve that performance-based sesmc desgn technology wll brng a new era to engneerng practce wth ncreased confdence and relablty on sesmc safety. Keywords: Performance, Desgn, Code INTRODUCTION Tawan s manly located n an unstable tectonc zone between the Phlppne Sea and Eurasan Plates. It has a very rch sesmc hstory. Sesmc safety has been our major concern. Smlar as n other countres, strength and ductlty are the most mportant factors n the local sesmc desgn code. Servceablty under low earthquake level only s requested to check. After the severe damage and a large amount of lfe and economc loss caused by the Chch earthquake n Sep. 21 1999, the need of performance-based sesmc desgn (PBSD) (Xue, 2000) of structures has been recognzed by professonals. Snce then, n addton to keepng an eye on the nternatonal technology development n ths area, researchers n local research organzatons such as Snotech Engneerng Consultants Incorporaton and Natonal Center for Research on Earthquake Engneerng and n the unverstes such as Natonal Tawan Unversty, etc. have put a lot of effort on performance-based sesmc evaluaton of exstng structures and performance-based sesmc desgn of new structures. Semnars or workshops have also been held to ntroduce such concept to engneerng practce. In 2003, more and more projects regardng sesmc evaluaton and retroft desgn are requested to ntroduce performance- 1 Research Fellow and SE Secton Chef, Cvl Engneerng Research Center, Snotech Engneerng Consultants Inc., Tawan, qxue@snotech.org.tw 2 Research Fellow, Cvl Engneerng Research Center, Snotech Engneerng Consultants Inc., Tawan 3 Research Fellow and IT Secton Chef, Cvl Engneerng Research Center, Snotech Engneerng Consultants Inc., Tawan 4 Manager, Cvl Engneerng Research Center, Snotech Engneerng Consultants Inc., Tawan

based sesmc desgn methodology. On recognzng an ncreasng need n performance-based sesmc desgn gudelne or provsons for local engneers to follow, n 2003, Archtecture and Buldng Research Insttute (ABRI), Mnstry of Interor, sponsored Snotech Engneerng Consultants Incorporaton to propose a tentatve framework of performance-based sesmc desgn code, whch must be currently applcable to engneerng practce. The results of ths study were approved by an advsory commttee, whch was composed of professonal specalsts, before they are ncluded n the fnal report. Followng the framework, ABRI sponsored another project for the same research group to propose a tentatve sesmc desgn code for buldngs by ntroducng performance-based desgn methodology. The research ncludes two parts, Part 1, Provsons, Commentary and References and Part 2, Examples. Ths paper focuses on part 1. Sesmc performance of a buldng s assocated wth both structural and non-structural systems and elements. In ths research, a non-structural system s desgned to accommodate acceleraton or drft after completon of the structural system desgn as n the current desgn code. No economc loss has been employed as a quantfed desgn crteron. Ths research focused on the structural performance. Issues regardng non-structural systems, sesmc evaluaton of exstng buldngs, buldngs ncludng base solaton system and energy dsspatve devces, desgn revew, constructon qualty assurance and buldng management and mantenance are out of the scope of ths paper. Due to paper length lmt, examples are not gven. Rather, some conclusons are drawn. KEY ISSUES The major dfference of PBSD from the tradtonal desgn methodology s summarzed as the followng 6 ssues. 1) Mult-level sesmc hazard s consdered wth emphass on the transparency of performance objectves. 2) Buldng performance s nterpreted n terms of both structural and nonstructural damage. Buldng performance s guaranteed through lmted elastc or nelastc deformaton n addton to strength and ductlty. 3) Sesmc desgn s orented by performance objectves. 4) An analytcal method, through whch the structural behavor, partcularly the nonlnear behavor s ratonally obtaned, s employed. 5) The desgned buldng wll meet the prescrbed performance objectves relably wth accepted confdence. 6) The fnal desgn result must ensure the mnmum lfecycle cost. Snce performance-based desgn covers a wde range of felds, ntroducng performancebased desgn methodology nto sesmc desgn code for buldngs must be done stage-by-stage. The current verson focused on the precedng frst 4 ssues through whch relablty of such desgned buldngs to meet the performance objectves s beleved to ncrease. PERFORMANCE OBJECTIVES AND CRITERIA The purpose to establsh ndvdual performance objectve for buldngs of dfferent sesmc use group s to provde a platform for nformaton exchange between the owners and the desgners so that they can acheve a common vew about the structural performance under dfferent level of earthquake exctaton. Therefore, transparency of the desgn objectves s very mportant. In the current sesmc desgn code, each performance goal of an mportant buldng s adjusted by ncreasng the consdered earthquake level through an mportance factor whle the expected performance level s mantaned. Ths s the case even for a performance goal wth respect to an earthquake level, whch s equal to 2500 years of return perod. Wthout sesmc hazard analyss, desgner may not know the return perod of such amplfed earthquake level. Even f the return perod of such an earthquake level s avalable, no body can tell the correspondng ntensty level exactly because 2500 years of return perod corresponds to the maxmum ntensty level for most places n Tawan (Loh and Hwang, 2002). Referred to the latest code or supportng reports n the USA (IBC2003, 2003, FEMA450, 2003), Japan (JSCA, 2000), New Zealand and Australa (AS/NZS 1170) (Kng and Shelton, 2004) and Canada (NBCC) (Hedebrecht, 2004), a good alternatve to adjust the performance goal of an mportant buldng s to decrease the expected performance level whle keepng the consdered earthquake level. No mportance factor s needed. Three sesmc hazard levels (Fg. 1) are consdered and can be dstngushed by return perod, probablty of exceedence or correspondng ste ntensty. Performance

has been classfed as 5 levels, Operatonal (OP), Immedate Occupancy (IO), Damage Control (DC), Lfe Safety (LS) and Collapse Preventon (CP). For each buldng, a performance objectve s composed of three performance goals, each of whch s constructed by one consdered sesmc hazard level and an expected performance level. Performance objectves for sesmc use group I II III are depcted n Fg. 1. Fgure 1. Performance Objectves The purpose to establsh performance crtera s for engneerng desgn. Performance crtera are the quantfed acceptance crtera to meet the performance objectve nterpreted by both quantfed hazard levels and quantfed performance levels. Each sesmc hazard level s quantfed by an elastc desgn spectrum. Each performance level s quantfed by parameters assocated wth strength, stffness and ductlty. Regardng strength, OP corresponds to an elastc behavor. Overstrength must be ensured for other performance levels and no large strength degradaton beyond ductlty lmt. No weak story exsts and the structure has enough vertcal capacty. Regardng ductlty, the concept of nelastc dsplacement demand rato (IDDR) (SEAOC, 1999) s employed. IDDR stands for the rato of nelastc dsplacement demand over the ultmate nelastc dsplacement capacty. Acceptable IDDR values assocated wth structural system performance levels OP, IO, DC, LS, CP are 0, 0.2, 0.4, 0.6 and 0.8, respectvely. For structural members, no less than 80% elements are needed to meet the same crtera. Regardng stffness, the maxmum nter-story drft rato (IDR) s consdered to lmt buldng lateral dsplacement. It s not easy for everybody to agree wth the same drft lmt. In ths research, based on the references such as SEAOC (1999), JSCA (2000), IBC2003 (2003), FEMA450 (2003), AS/NZS 1170 (Kng and Shelton, 2004) and NBCC 2005 (Hedebrecht, 2004) and accordng to the local sesmc hazard analyss results and opnons from the advsory commttee of ths project, IDR lmt as n Table 1 s suggested prelmnarly. Structural system type n Tawan s not dscussed n ths paper and can be referred to the local sesmc desgn code (CPRMI, 2005) or Xue et al. (2005). It should be noted that lmt of the maxmum nter-story drft rato s used as performance assessment crteron but may result n an unreasonable demand partcularly for a less ductle structural system f a drect dsplacement based desgn method s used for prelmnary desgn. Table 1. Allowable maxmum nter-story drft rato wth respect to each performance level Structural system Performance Level CP LS DC IO OP Systems wth brck shear walls 0.009 0.007 0.007 0.007 0.005 Other systems 0.025 0.020 0.015 0.010 0.005

An obvously sgnfcant change arsng from the current performance crtera s an ncrease n structural stffness demand for buldngs of hgher sesmc use group or n other words, for more mportant buldngs. It s mportant to pont out that the code suggested performance objectve and crtera provdes the mnmum protecton to the buldngs. Increase of performance objectve s allowed upon the owner s acceptance. SITE FEASIBILITY Ste feasblty study s to ensure that the establshed performance objectve can be met on the constructon ste. Smlar as the current code, ths research focus on the determnaton of very weak sol layer and lquefacton potental of sandy sol so that the coeffcent of sub-grade reacton can be reduced by a reducton factor D E. Under EQ1 level earthquake, lquefacton s not acceptable. For EQ2~EQ3 levels, sol lquefacton may occur as long as structural performance s mantaned. The allowable degree of lquefacton s lmted by the maxmum allowable value of D E, whch s set to be 0, 1/3 and 2/3 for buldngs of sesmc use group I, II and III, respectvely. If ths lmt has been exceeded, sol mprovement must be appled. For buldngs of sesmc use group II and III, back off dstance from type 1 actve fault and other hazard such as landslde must be taken nto consderaton. CONCEPTUAL DESIGN Basc conceptual desgn rules are gven wth emphass on redundancy and unform contnuty of strength, stffness and ductlty, respectvely. Structural systems have been classfed as four types, namely, the load-bearng walls, frame systems, moment resstng frames and dual systems. Ductlty capacty of each system s represented by the ultmate ductlty rato R accordng to ther relatve capacty to dsspate energy after yeldng. Buldng heght lmt, horzontal and vertcal rregularty, expected energy dsspaton and yeldng mechansm have been defned. Flexblty s gven to allow desgn of specal structural systems based on relable technology. PRELIMINARY DESIGN Performance based desgn mples that the desgn procedure s orented by the performance objectve. It can be acheved through ether an ndrect or a drect dsplacement-based desgn method. Snce approaches of the latter method are not mature enough to be appled drectly to varous structures, they are summarzed n the appendx for reference only before detal parametrc and case studes. Alternatvely, the former method, whch employs a force-based desgn procedure combned wth a check on the nter-story drft rato, s adopted. Performance objectve orented desgn by force-based method Smlar as tradtonal force-based desgn method, fundamental perod of the structure system s gven by emprcal formula. Sesmc desgn force s estmated by an elastc desgn force reduced by a reducton factor F u assocated wth an allowable ductlty rato correspondng to each performance level R a, PL. Varous format of elastc desgn spectra are gven for each refned sesmc zone. Newmark-Hall (1982) reducton factors have been used wth slght modfcaton accordng to local sesmc hazard analyss n the present desgn code and n ths research.

The allowable ductlty rato or the maxmum usable ductlty assocated wth each performance level s calculated by Eq. 1. Ra, PL = 1+ IDDRPL ( R 1) (1) where R s the prescrbed ductlty capacty for a gven structural system and IDDR has been defned before. Subscrpts a and PL stand for allowable value and consdered performance level, respectvely. Snce R has been taken by consderng some safety margn, t corresponds to 80-90% of the ultmate nelastc dsplacement capacty. Accordngly, durng prelmnary desgn, acceptable IDDR values assocated wth structural system performance levels OP, IO, DC, LS and CP are revsed to be 0, 2/9, 4/9, 2/3 and 1, respectvely. For stes wthn Tape basn, they are revsed to 0,1/6,1/3,1/2 and 1, respectvely consderng ncreasng number of earthquake cycles. Analyss and Desgn The same as n the current desgn code, elastc analyss can be used at ths stage. Statc analyss or modal spectrum analyss can be selected accordng to buldng heght and system rregularty. Element desgn can be conducted n the usual way accordng to the exstng desgn specfcatons and tools. Prelmnary Check on The Inter-Story Drft Rato Regardng concerns arsng from complexty of nelastc analyss durng detal sesmc performance assessment n the next step, a smple method for prelmnary check on the nter-story drft rato s employed so that defcency assocated wth stffness can be notced at an earler stage. Concept of modal spectrum analyss (Goel and Chopra, 2004; FEMA440, 2004) has been employed. N modes wth 90% modal mass partcpatng are consdered. Alternatvely, N=3 s allowed. Under earthquake level EQ2 or EQ3, for each mode wth perod of T, nelastc spectral dsplacement S d assocated wth the correspondng performance level s calculated through Eq. 2. S d = R a, PL F u T 2π 2 S ae (2) 1 where S ae s the elastc desgn spectral acceleraton correspondng to T. For the frst mode, R a,pl s equal to R, estmated by Eq. (1) wth revsed IDDR. For other mode, the same value may be assumed or a PL 1 a, PL R =1. S d also represents the nelastc target dsplacement of the equvalent sngle degree of freedom (ESDOF) system under mode. Modal superposton s appled by ether SRSS or CQC rule to obtan the dsplacement of ESDOF system. Fnally, the maxmum nter-story drft rato s calculated by Eq. 3 (SEAOC, 1999). ( h k ) IDR = x * k (3) where h n, k 1 and k 2 are heght at roof level, factor of effectve heght, factor of dsplacement shape functon defned n SEAOC (1999). It should be notced that ths approxmate method s very rough and s most sutable for a structure whose behavor s domnated by ts frst mode. For more flexblty, prelmnary desgn accordng to the current force-based desgn method s allowed as long as the performance crtera are acceptable after detal performance assessment. It s suggested to use together wth prelmnary check on the nter story drft rato. DETAILED SEISMIC PERFORMANCE ASSESSMENT Although the prelmnary desgn s somewhat objectve orented, the assumed sesmc ductlty n 1 2

capacty may not be developed n the desgned structure due to consderaton of szng standard and constructon convenence. Besdes, nonlnear behavor s not explctly consdered n the procedure. Real behavor of the structure may be dfferent from the desgn model. Therefore, a detal sesmc performance assessment s needed to ensure, although not wth 100% relablty, that ts sesmc performance meet the prescrbed performance crtera and objectve. Analyss Method for Performance Evaluaton In ths research, the mnmum allowable analyss method to be employed for performance evaluaton s suggested n Table 2. Table 2. The mnmum allowable analyss method for performance evaluaton Sesmc hazard level Statc analyss s allowed n the prelmnary desgn Dynamc analyss must be used n the prelmnary desgn T 3. 5T s T > 3. 5Ts EQ1 Lnear Statc Lnear Dynamc Lnear Dynamc EQ2 Nonlnear Statc Nonlnear Statc Nonlnear Dynamc EQ3 Nonlnear Statc Nonlnear Statc Nonlnear Dynamc Note: materal nonlnearty s the only focus n ths table. Analyss and Performance Evaluaton Lnear statc or lnear dynamc analyss ncludng modal spectral analyss and tme-hstory analyss can be carred out as usual to evaluate sesmc performance under EQ1 level earthquake. Nonlnear statc pushover analyss s usually the choce for lots of buldngs under EQ2 or EQ3 level earthquakes. In the case when nonlnear tme-hstory analyss s needed, nonlnear dynamc pushover analyss may be carred out to evaluate structural system ductlty. Concernng about complexty of nonlnear dynamc analyss, the advsory commttee suggested that smplfed modal be allowed to use n nonlnear dynamc analyss. Ths secton focused on nonlnear statc analyss. Structural Modelng Structural modelng s very mportant, sometmes may be even more mportant than selecton of the analyss method. Plastc hnge propertes of dfferent elements can be adopted from relable experments or research reports. Crackng of RC elements are consdered n the same way as suggested n FEMA356 (2000). Influence of non-structural walls such as the nfll walls to the structural deformaton must be consdered. Nonlnear Statc Pushover Analyss Based on a constant load pattern, nonlnear statc pushover analyss s an ncremental teratve method to obtan the base shear versus roof dsplacement relatonshp. In ths research, two types of load pattern are suggested to employ. One of them s proportonal to the fundamental mode. The other s a unformly dstrbuted load pattern proportonal to the story mass. Other adaptve load patterns based on relable studes are also allowed. The montorng pont s the mass center of the roof story ncludng the penthouse. The ultmate or falure state along the capacty curve may corresponds to that when the structure lost stablty, or 80% elements reach ther deformaton lmt, or more than 20% (maxmum or effectve) base shear strength reduced after yeldng, whchever s crtcal. Performance Evaluaton Performance pont s evaluated through sesmc coeffcent method (FEMA440, 2004) or capactyspectrum method (Fajfar, 1999; Chopra and Goel, 2004; FEMA440, 2004; Xue, 2001). The structural s pushed agan to the target dsplacement assocated wth the performance pont to access the behavor of both the structural system and the elements.

Performance Crtera Examnaton Performance crtera regardng all performance goals are examned regardng structural systems and elements. Structural systems Regardng vertcal load carryng capacty, the structure should not collapse due to loss of any element. Regardng lateral strength, damage and weak story mechansm can be clearly understood through hnge locaton and formaton consequence. Strength deteroraton s controlled by the defnton of falure state durng nonlnear pushover analyss. Regardng lateral deformaton capacty, the maxmum nter-story drft rato IDR and nelastc dsplacement demand rato IDDR are calculated and compared wth the acceptable performance crtera. Extremely soft story and extreme torson rregularty must be avoded. Buldng separaton dstance must be no less than ether the maxmum story dsplacement under EQ2 or 70% that under EQ3 to avod poundng. Structural elements Elastc behavor examnaton must be done for elements that need to reman elastc accordng to the damage mechansm or structural modelng. Deformaton capacty examnaton s conducted for deformaton-controlled behavor based on force-dsplacement relatonshp employed n the nonlnear statc analyss. Accordng to the performance crtera, only no less than 80% elements need to meet the same allowable IDDR as that of the structural system. Strength capacty examnaton s conducted for force-controlled behavor so that the lowest bound of strength wll not be less than the estmated force nduced by the consdered loadng. In ths research, element performance examnaton s not appled to performance goal assocated wth CP performance level where global system stablty s of the man concern. CONCLUSIONS The tentatve performance-based code ntroduces a transparent platform for the owners and desgners to exchange ther vew on the expected sesmc performance of the buldngs under dfference level of earthquake exctaton. For buldngs of dfferent sesmc use group, dfferent performance goals are specfed nstead of employng an mportance factor. Performance levels are quantfed through parameters assocated wth structural strength, stffness and ductlty. Requrement on ste sutablty wth focus on controllng the degree of lquefacton potental s gven. Conceptual desgn rules wth focus on redundancy and unform contnuty of strength, stffness and ductlty, respectvely are specfed. A performance objectve orented prelmnary desgn procedure s presented wth consderaton of flexblty. Instead of ensurng structural systems possessng the prescrbed ductlty capacty through ductle desgn, the expected behavor or performance of the prelmnarly desgned structures are analyzed and evaluated through analyss methods, whch may capture the real behavor reasonably. The tentatve performance-based desgn code s not to replace the current desgn code at present. Both codes are supposed to exst for a perod of tme, durng whch possble amendment may be ntroduced to the performance-based desgn code. Fnally, performance-based desgn code wll domnate after ncrementally mprovement over the present desgn code. Accordng to the examples studed n ths research, sesmc performance of the desgn structures could be easly understood and clearly nterpreted through the performance-based desgn procedures. The engneers may have more confdence n the sesmc performance of the desgned structure under dfferent level of earthquake exctaton. Prelmnary check on the nter-storey drft lmt may help fndng the stffness defcency earler n the prelmnary desgn stage and save some computatonal effort. In engneerng practce, structural detalng has been standardzed consderng constructon convenence. It s found that such a structure usually has a lower ductlty capacty than that specfed n code. Although structural ductlty s not unformly dstrbuted, structural strength and stffness are

ncreased. Therefore, the performance objectves are usually satsfed. If an dealzed detalng has been used wthout consderng constructon convenence, the ductlty capacty s n very good agreement wth the code value. ACKNOWLEDGMENTS Ths work s the result of a research project sponsored by Archtecture and Buldng Research Insttute, Mnstry of The Interor under grant No. 094301070000G1018. Specal thanks are gven to the revewers and the advsory commttee members of ths project for ther valuable comments. REFERENCES CPRMI, (2005), Sesmc Desgn Specfcatons and Commentary of Buldngs, Constructon and Plannng Agency, Mnstry of The Interor, CPRMI publsh ntroduce, Tawan. (n Chnese) Xue, Q. (2000). Need of performance-based earthquake engneerng n Tawan: a lesson from the Chch earthquake, Earthquake Engneerng and Structural Dynamcs, 29, 1609-1627. Loh, C.H. and Y.R. Hwang, (2002). Sesmc Hazard Analyss of Tawan Area consderng multple ground moton parameters, Report No. NCREE-02-032, Natonal Center for Research on Earthquake Engneerng, Tape, Tawan. IBC2003, (2003). Internatonal Buldng Code 2003, Internatonal Code Councl, CA, USA. FEMA450, (2003). NEHRP Recommended Provsons for Sesmc Regulatons for New Buldngs and Other Structures, Part 1: Provsons, BSSC, Washngton, DC, USA. JSCA, (2000). Structural Desgn by Response Control Methods, Japan Structural Consultants Assocaton, Shokokusha Publshng Co., Ltd. (n Japanese) Kng, A. and R. Shelton, (2004). New Zealand Advances In Performance-Based Sesmc Desgn, 13WCEE, Vancouver, Canada, Aug.1~6. Hedebrecht, A. (2004). Code Development Issues Arsng From The Preparaton Of The Sesmc Provsons Of The Natonal Buldng Code Of Canada, 13WCEE, Vancouver, Canada, Aug.1~6. Newmark, N.M. and W.J. Hall, (1982). Earthquake Spectra and Desgn, Earthquake Engneerng Research Insttute, Berkeley. FEMA440, (2004). Improvement of Nonlnear Statc Sesmc Analyss Procedures, (camera ready draft), Prepared by Appled Technology Councl (ATC-55 Project) for Federal Emergency Management Agency, Washngton, DC. SEAOC, (1999). Recommended Lateral Force Requrements and Commentary, Structural Engneers Assocaton of Calforna-Sesmology Commttee, USA. Goel, R and A.K. Chopra, (2004). Evaluaton of Modal and FEMA Pushover Analyss: SAC Buldngs, Earthquake Spectra, 20(1), 225-254. Fajfar, P., (1999). Capacty Spectrum Method Based on Inelastc Demand Spectra, Earthquake Engneerng and Structural Dynamcs, 28, 979-993. Xue, Q., (2001). A Drect Dsplacement-Based Sesmc Desgn Procedure of Inelastc Structures, Engneerng Structures, 23(11), 1453-1460. FEMA356, (2000). Prestandard and Commentary for The Sesmc Rehabltaton of Buldngs, prepared by ASCE, publshed by the Federal Emergency Management Agency, Washngton, DC. USA.