337

Pushover procedure for s e i s m i c analysis of buiIdin g s

W K Tso and AS Moghadam McMaster University, Canada

Summary
This article is a state-of-the-art review of the simplified inelasticanalytical procedure commonly referred to as the pushover analysis to estimate the seismic response of buildings. The basic steps of a pushover analysis are outlined. The choices available to users in each step are discussed and past studies involvingthe pushover procedure are briefly reviewed. Finally, the accuracy of the method compared to inelasticdynamic analysis is presented using examples.

Progress in Structural Engineeringand Materials I998Vol I(3): 337-344

A current trend in new seismic design code developmentis the incorporation of performancebased design methodology[i-~,51. In this methodology, every building is designed to have the desired levels of seismicperformance corresponding to different specified levels of earthquake ground motion. For example, a building would be designed for immediate occupancy at one level of ground shaking and for possible damage but not collapse at a higher level of ground motions. To achieve this goal, engineers need information regarding the distribution of forces and deformationin the building elements during earthquakes.Elastic analyses are insufficientbecause they cannot realistically predict the force and deformationdistributions after the initiation of damage in the building. Inelastic analytical procedures become necessary to identify the modes of failure and the potential for progressive collapse. The need to perform some form of inelastic analysis is already incorporated in many building codesyl. Inelastic time-history analysis using an ensemble of representativeground motions is probably the most realistic analytical approach for evaluating the performance of a building. However, the inelastic time-history analysis is usually too complex and timeconsuming in the design of most buildings. As a compromise, a simplified procedure commonly
Abbravlatiom CM = cenrreofms CQC = mmplete quadratic equation DMRF = duailemanent-resisting frame FEMA = Fedeni Emergenq Managementb n c y (US) MDOF = mukidepedfrredom = mlnforcd concrete RC

referred to as the pushover analysis was developed to supply the necessary information to the designers. Depending on the information sought, there exists a wide interpretation of how the procedure should be carried out. It is the objective of this article to provide an overview of the procedure and a summary of the different variations of analyses used under the generic name of pushover analysis. Finally, the accuracy of the pushover procedure is illustrated using examples.

Basic concepts
The performance-based design methodology requires the proper matching of two basic quantities, the seismic capacity and the seismic demand. Demand is a description of the earthquake ground motion effects on the building. Capacity is a representation of the ability of the building to resist the seismic effects. The performance is measured by the manner with which the capacity is able to handle the demand. In a pushover procedure, the seismic capacity and demand are estimated explicitly in two separate steps.
I Estimating seismic capacity

A sequence of inelastic static analyses is performed on the building, when it is subjected to a set of monotonically increasing lateral loads. Such a loading

SDOF = single degree of freedom SRSS = square mot of rum of squares

338
sequence is representative of 'pushing' of the building, hence the term, pushover analysis. The building is pushed until its displacementreaches some predetermined limits. The predetermined displacement limit may be set based on different criteria such as the maximum allowablestorey drift or ductility limits. Alternatively,the building may be pushed so that it becomes unstable and reaches its collapse state. One important product of the pushover analysisis the base shear versus top displacement relationship, commonly referred to as the capacity curve of the building. This curve gives an overall summary of the capacity of the structure. Information such as the initial elastic stiffness, the initiation of first yielding, the stage of rapid stiffnessdeterioration, and the ultimate strength can all be inferred from the capacity curve. In addition, the damage pattern of the building at any post-yield level can be found by examining other response parameters such as interstorey drift ratios, hinge locations and member ductilities. The weak links and undesirable characteristicsin the building such as soft storey conditions, strength and stiffness discontinuities can readily be detected from the damage pattern. A judgment is then formed as to the acceptability of the behaviour of the building. This first step essentially establishes the seismic capacity of the building, and is often the only step taken in a study using the pushover procedure.
2 Estimating seismic demand

STRUCTURALANALYSISAND CAD
evaluate the inelastic behaviour of buildings, most of the past studies involved the pushover of a typical 2-D lateral load-resistingelement or elements of a building. It is implied that the damage potential to the building can be inferred from the damage pattern of the element(s).These works can be referred to as applicationsof 2-D pushover analysis (eg refsi7-91). These investigationswere on buildings possessing a high degree of symmetry in which the effects of torsion are negligible. For more complex buildings such as buildings that are asymmetrical in plan, the top displacementsof the lateral load-resistingelements at different locations are different due to torsion. The damage pattern of these elements can be obtained using 3-D pushover analyses. In a 3-D pushover analysis, the target displacement represents the seismic displacementof the centre of mass (CM)on the roof of the building. The lateral loads are applied at the CM of the floors for the pushover process~io]. A model incorporatingboth the elastic and inelastic properties of the entire building is needed to implement this procedure. The effort needed to prepare the input data, and the postprocessing of the output information for easy interpretation using this approach can be considerable. An alternative way of applying the pushover procedure to asymmetricalbuildings is to determine multiple target displacements,one for each frame or wall element where the damage pattern needs to be investigated. The multiple target displacementscan be determined from a 3-D elastic dynamic analysis of the building. Then, the damage pattern can be found by performing 2-D pushover analyses on the resisting elements.This alternative requires only the inelastic modelling of the specific frames or walls of interest, and avoids the necessity to create an inelastic model of the entire building.

If evaluationof the damage potential of the building when subjected to a specific level of ground shaking is needed, then the second step in a pushover procedure is to establish the seismic demand on the building by determining a target displacement. A target displacement is an estimation of the top displacement of the building when exposed to the specified level of ground shaking. This is usually done using some form of dynamic analysis as will be discussed later. To evaluate the damage potential of the building at the specified level of ground shaking, the target displacement should first be determined, and then a pushover analysis conducted in which the building is pushed u n t i l its top deflection matches the target displacement.The damage estimates from the pushover analysis at the target displacement level are considered representativeof the structural damage to the building due to ground shaking.

THE LATERAL LOAD DISTRIBUTION

The lateral load distribution used in the pushover can affect the accuracy of the results. Ideally, the distribution should represent the lateral inertial force distribution that causes the most severe responses to the building. Such a distribution depends on, among other variables, the severity of the shaking, the dynamic characteristicsof the building and the frequency contentsin the ground motion. There is no general agreement on the best choice of lateral load distribution to be used in the pushover process. Currently, two types of load distribution are used. They are fixed load distribution and variable load distribution.
Fixed load distribution

Choices in pushover analysis

The overall steps of a pushover analysis are straight forward. However, there are variations within each step that need to be discussed.

MATHEMATICAL MODELLING
PUSHOVER ANALYSIS

OF THE BUILDING FOR

While the objective of using pushover analysis is to

In the fixed load distribution, the distribution is determined a priori and remains unchanged during the pushover. Some of the fixed distributions used are as follows.

PUSHOVER PROCEDURE FOR

0 A single concentrated horizontal force at the top

SEISMIC ANALYSIS

339

floor 0 Uniform load distribution on all floors 0 Triangular or standard code load distribution 0 A load distribution proportional to the product of the mass vector and the fundamental mode shape 0 Lateral force distribution based on a linear elastic dynamic analysis or response spectrum analysis of the building
Variable load distribution

used to obtain the maximum response of the SDOF system.

2 Capacity spectrum approach

To allow for the changes in inertial forces with the level of inelasticdeformation, some researchershave proposed adaptive load patterns to be used in the pushover. The load distribution changes as the building is deformed to larger and larger displacements. The following are some of the variable load distributions. 0 A distribution proportional to the product of the mass vector and the fundamental mode shape is used initially until first yielding takes place. Then, for each load increment beyond yielding, the forces are adjusted to be consistentwith the deflected shape in the inelasticstate. The load distribution is based on the product of the current floor displacementsand masses. 0 A distribution based on mode shapes derived from secant stiffnesses at each load step 0 A distribution proportional to storey shear resistances at each step These adaptive load distributions require more computationaleffort. However, their superiority over the more simple fixed load distributions has not been establishedp~.

The base shear and roof displacement values at each point on the capacity curve are transformed into spectral acceleration and spectral displacement values to obtain the capacity spectrum. A reduced response spectrum is created by adjusting the response spectrum to a damping value appropriate to the level of anticipated deformation. The capacity spectrum is superimposed on the reduced response spectrum curve of the design level earthquake. The intersection of these two curves gives an estimate of the target displacement.
3 Elastic dynamic analysis approach

This approach is based on the equal seismic displacement principle between elastic and inelastic systems.For buildings in the medium to long period range, it has been shown that the top displacement of a building is essentiallythe same, whether the building is responding in its elastic state or being excited into the inelastic range. For these buildings, the target displacement can be determined by an elastic dynamic analysis of the building. The analysis may take the form of time-history analysis or response spectrum analysis, depending on the form of the input ground motion information available.
Discussion

THE DETERMINATION OF TARGET DISPLACEMENT There are three approaches to establish the target displacement.
I Single degree of freedom (SDOF) approach

Traditionally, the target displacement is determined based on the seismic response of an equivalent SDOF system. Different suggestions exist to derive the equivalentSDOF for the building. Common to all the formulations is the assumption that the building will respond in a single mode. Therefore, the deflected shape of the building remains unchanged during the response and can be represented by a single shape vector. This assumption allows the transformation of the multi degree of freedom (MDOF)building into a generalized SDOFsystem. The load-deflection curve of the resulting SDOFsystem takes the form of the capacity curve of the building. Once the equivalent SDOF system is constructed, its response to any ground motion can be obtained by dynamic analysis. The absolute maximum displacement of the SDOF system is then taken as the target displacement. Alternatively,the response spectrum approach can be

There are some advantages in using the elastic dynamic analysis approach. First, there is no need to model all the elastic and inelasticproperties of the entire building to create the equivalent SDOF system. For many buildings, some form of elastic static analysiswill be done for preliminary and possibly for final design purpose. In this case, the additional effort to carry out the elastic dynamic analysis to establish the target displacement is minimal. Second, this approach is particularly convenient if one wants to avoid the need to perform 3-D pushover analyses for plan-eccentricbuildings. The elastic dynamic analysis approach can be used to determine multiple target displacements,one for each resisting element in which the damage assessment is sought. The evaluation of the damage pattern of a plan-eccentric building is then reduced to a number of 2-D pushover analyses performed on the resisting elements of interest.

Research on pushover analysis

A brief review of selective past studies is given here to show the pushover procedure, as it is applied to different studies. This review is not exhaustive so readers can obtain additional information from the references cited. The variety of load distributions used in the pushover analysis is exemplified in the following
0COMTRUCTION RESEARCH COMMUNICATIONS LIMITED I998 ISSN 13654556

340
works. Miranda[ii] has used a fixed triangular distribution. Fajfar & Gaspersic[slused the product of the mass distribution and a fixed displacementshape. The displacement shape is different for different types of buildings. Uniform load distribution is one distribution among many distributions recommended by FEMA for seismic rehabilitation of buildings131. The FEMA reference suggests that a load distribution based on modal forces, combined using either the SRSS or CQC combination rules, be used for cases when higher modal effects become important. Among suggestionsregarding the use of variable load distribution for pushover analysis, one can include the use of load distribution proportional to the deflected shape of the building[u],load distribution proportional to mode shapes derived from secant stiffnessesat each load step[u],and load distribution proportional to story shear resistances at each step[irl. Gergely & Hamburger[is] pointed out that improvementof the force distribution during pushover analysis may become necessary for irregular buildings. Regarding the target displacementestimation, many researchershave proposed the use of equivalent SDOF systems to approximate the response of the building[ia-231. Miranda[ii]suggested the use of a displacementshape corresponding to target displacementin pushover analysisfor deriving the SDOF system. Because the target displacementis not known a priori, an iterative procedure would be involved. Fajfar & Gaspersic[a]suggested the use of fixed displacement shapes in deriving the equivalent SDOF system. The response spectrum analysiswas then applied to the equivalent SDOF system to obtain the target displacement.Some researchers suggested that the target displacementwhich resulted from the equivalent SDOF system should be treated as baseline informationto be further modified to include other effectq". Modificationsto include effects caused by multi-modal responses[25-z7],eg of MDOF systems[25-z71, soft soilp81 and 3-D effectim], have been proposed. Most of the studies in the literature are on the applicationof 2-D pushover analysis for walls and frames in a planar building. Moghadam & Tso[i0,30] extended the procedure to asymmetricalbuildings. The target displacementis estimated using an equivalent SDOF system based on the loaddisplacement relationship at the CM on the roof of the building. The pushover is carried out by performing 3-D inelastic analyses, applying the lateral loads at the CM of the floors of the building. A simplified 3-D pushover procedure for asymmetricbuildings has been developed by Kilar & Fajfar[si].In this method, the building is modelled as a collectionof planar macro-elements.An approximate relationship between the global base shear of the building and its top displacement is computed via a step-by-step analysis. This study focuses on finding the capacity
0CONSWKTION RESEARCH COMMUNICATIONS LIMITEDI998I S S N I3654556

STRUCTURAL ANALYSIS AND CAD

curve and does not include a method for estimating the target displacement.An alternative method uses the results of 3-D elastic dynamic analysisof the asymmetricalbuilding to obtain both the target displacementsand the lateral load distributions for individual resisting elements of interest. The pushover is performed on the elements using 2-D pushover analyses[sz]. An extension to this method is the direct use of elastic response spectrum analysis instead of elastic time-history dynamic analysisto obtain target displacements[x~]. The capacity spectrum method developed by Freeman et a11341 has been used for the evaluation of seismic performance of buildings[ss]. The Applied TechnologyCouncil[r*] has recommended this method for use in the seismic evaluation and retrofit of concretebuildings. If only global response parameters such as deflections and storey drifts are of interest, the method can be simplifiedby consideringan appropriate deflection shape of the building (eg the shape vector used for deriving the equivalent SDOF system) and normalizing the top deflection of this deflection shape to the target displacement[i7,36,37]. This directly gives the seismic floor displacement and storey drift estimates of the building. However, this simplificationwould not provide estimatesof local response parameters such as ductility demands on the resisting elements. Finally, a discussionof the potential and limitations of the pushover procedure has been provided by Lawson et al~ssq.

Examples
To illustrate its accuracy, the pushover procedure is applied to estimate the seismic responses of two seven-storeyRC ductile moment-resisting frame (DMRF)buildings subjected to an ensemble of 10 ground motion records having similar spectral shapes as input. The ground motions are assumed to be unidirectional and come from the Y-direction. The lateral load-resistingelements in that direction consist of three identical ductile moment resisting frames as shown in Fig. 1.Each frame has three bays and uniform properties along its height. It is designed for a base shear of 400 kN, which is about 20% of the elastic base shear as determined from the National Building Code of Canadap91. The design of the beams and columns in the frames follows the strong columnweak beam capacity design philosophy. It is assumed that the x-directionlateral load-resistingelementswill not contributeto resistance to excitationsin the ydirection in these buildings. One building is a symmetricbuilding (BuildingS) and the other is an asymmetricalbuilding (BuildingA). In Building S, the floor masses are uniformly distributed so that the CM of each floor coincide with the geometric centre. In Building A, the mass distribution of each floor causes

PUSHOVER PROCEDURE
2 4 Frame 1

FOR SEISMIC ANALYSIS

34 I

m
Frame 3

Example I:symmetric building(Building S)pz] Fig. 2 compares the target displacements obtained using two approximate approachest o the corresponding top displacement of the building, based on inelastic dynamic time-history analysis. Each point represents the correlation for one earthquake record in the ensemble of records. The target displacements based on the equivalent SDOF approach are shown in circles and those using linear dynamic analysis are shown in squares. The mean target displacementsusingthe two approachesare shown in the Fig. 2 as 0 and 17, respectively. Fig. 2 shows that both approacheslead t o good estimates of the mean target displacement and both approachesare comparable in accuracy. The mean seismic responses of Building S Over the ensemble of ground motions are shown in Fig. 3. Since Building S is symmetrical, the responses are identical i n all three frames. The responses for one frame are presented in Fig. 3. In each plot, three curves are shown representing results obtained from: 0 a 2-D pushover analysis of the frame t o the mean target displacement based on an equivalent SDOF system analyses, 0 a 2-D pushover analysis of the frame using the mean target displacement based on dynamic elastic analyses of the building, and 0 the mean of the responses computed using inelastic dynamic analyses of the building.

Building S

Fig. I Plan of BuildingS and BuildingA

the CM of the floor to shift a distance of 2.4 m from frame 2 towards frame 3. Therefore, Building A is mass eccentric, and has a constant floor eccentricity equal to 10%of plan dimension. The mean of the maximum inelastic dynamic timehistory responses of the buildings, computed using the computer code CANNY[@], are used as references to check the accuracy of the pushover analysis results. The response parametersused for the comparison are: 0 the displacements, 0 the interstorey drifts, 0 the column ductility demands, and 0 the beam ductility demands in the edge frames.

A triangular distribution of static loads is used in both pushover analyses. A comparison ofthe three curves in each plot shows that both pushover analyses give a good estimate of the parameters of interest for the frame.

Example 2: asymmetric building(BuildingA)[33] The top seismic displacementsofframes i n Building Aare different due t o a torsional response. Therefore, three target displacementsneed t o be determined, one for each frame. These target displacementsare calculated using the elastic dynamic analysis approach. Specifically, an elastic responsespectrum analysis i s carried out and the modal contributions t o the top displacementsof Frame I, and also Frame 3, are combined usingthe CQC combination rule. The mean seismic responses of the edge frame at the flexible side of Building Aover the ensemble of ground motions are shown in Fig. 4. The two curves that are shown in each plot represent results obtained from: 0 a 2-D pushover analysis of the frame t o the target displacement shown as adotted line, and 0 the mean of the responses computed using inelastic dynamic analyses ofthe building shown as a solid line. The load distribution used i n the pushover analysis is taken from the load distribution in Frame 3 based on the response spectrum analysis of the building. A comparison of the two curves in each plot shows the accuracy ofthe pushover analysis t o estimate the parameters of interest for an edge frame ofthe asymmetrical building. Similar accuracy i s obtained for the responses of Frame I, the frame at the other edge of the building.

0.25

g .5
Y

/ I
I Elorticdynamic analysis

0.2
0.15

-:

-~

0 Et.

=
Y

0.1 --

0.05 -:

I-"
' , ' ; , ' , ' I 4 ,
L L

I " , ,

: ' " ' I , ' -

0.05

0.1 0.15 0.2 0.25 Maximum roof displacement from inelasticdynamic analysis (m)

0.3

Fig. 2 Correlationof target displacementfor symmetric building.

Modified from Tro & Moghadamp]

0CONSTRUCTION RESEARCH COMMUNICATIONS LHITED I998 ISSN 1365-0556

342

STRUCTURAL ANALYSIS AND CAD

0.0s

0.1

0.1s

0.2

F
I
0

0.00s

0.01

0.01s

0.02

mean beams

--

(d)
0
0T-t

1
0 -

displacement by SDOF displacement by eiutic dynamic

inelastic dynamic anatyaes

Fig. 3 Symmetric buildingresponses: (a) displacements, (b) interstoreydrifu, (c) column dudity, (d) beam ductility.
Modifiedfrom Tso & Moghadam[n]

mean columns

7
6
5

mean beams

---

4 -- 1
3 --o

* --r
0

I --

.,
0 1 2 3 4 5

(d 1
1 2 3 4 5 6

"

Inelastic dynamic h Y -

Fig. 4 Asymmetric buildinrframe 3 responses: (a) displacements. (b) interstoreydrifts, (c) column d u d t y , (d) beam ductility.
Modified from Moghadam & Tso[33]

@ CONsnwCnON kSFARCH COMMUNICATIONSLIMITED I998 lSSN 13654556

PUSHOVER PROCEDURE FOR SEISMIC ANALYSIS

Conclusions
One main assumptionimplicit in the pushover procedure is that the building will respond in a single mode. Therefore, the procedure is suitable for applying to regularbuildings. For irregular buildings such as buildings with significant setback, the contributionto responses can come from more than one mode. The accuracy of the procedure for these irregular buildings has not been established.While most of the applicationsare for regular planar buildings, the procedure can give equally good results for plan-eccentricbut otherwise regular buildings. The pushover analysis has been developed as a simplified procedure to provide informationto the designers on the inelastic performance of buildings when subjected to earthquake excitation.Compared with the more sophisticated inelastic dynamic analysis procedure, pushover analysis is simpler to carry out. More important to the design profession is that the volume of output data using pushover analysisis more manageable and easier to understand. Despite its limitations, it is currently recognized as a viable method for estimating seismic responsesin earthquake engineering.It should be emphasized that given the many possible variations in ground motion, the uncertainty in material properties and the difficultiesof mathematical modelling of structural elements, the interpretation of the pushover analysis results requires sound engineeringjudgment.

343

[8] QiX Displacementresponse and earthquake-resistance design for reinforcedconcrete structures usinga displacement control approach. PhD Dissertation, Universityof Californiaat Berkeley, California 1989. [9] Fajf&rP,Gaspersic P & Drobnic D. A simplified nonlinear methodfor seismic damage analysis ofstrucwres. ProceedingsWorkshop on SeismicDesign Methodologierfor the N u t Generation ofcodes, Bled, Slovenia, 1997. Rotterdam: Balkema. 1997. [I 0 1 MoghadamAS & Tso WK. Damage assessmentof eccentric mukistoreybuildings using 3-D pushoveranalysis. Proceedings IIth World Conference x f o r d : Elsevier. 1996. on Eonhquoke Engineerha Mexico. Paper N o 997. O [I I] Miranda E. Seismicevaluationand upgrading of existing buildings. PhD Dissertation, University of Californiaat Berkeley, California I99 I. [I21 Fajfar P & Firchinger M. N2 a methodfor non-linear seismic analysis of regular buildings. Proceedings 9th World Confemncc on Earthquake Engineering. Japan:JapanAssociationfor Earthquake Disaster Prevention. 1988. &IS. I I I- I 16. [I31 Eberhard MO & Sozen MA. Behaviodmed methodto determine ofthe Structural Division,ASCE design shear in earthquake-resirt;lntwls.~ournal 1993: II9(2): 6 19-640. [I41 BracciJM, KunnathSM & Reinhorn AM. Seismic Performance and RetmfK Evaluation of ReinforcedConcmte Structures.Journol ofStructural Engineering(ASCE) 1997: 123(1):3-10. [I 51 Gergely P & Hamburger RO. Simplifiedmethods for evaluationof rehabilitated buildings. ProceedingsUS-ltdy Workshop on Guiddinesfir Seismic Evaluation and RehabilitotjanofUnminfirced MasonryBuildings, ReportNo 94402 I . Buffalo, New Yo& National Center for Earthquake Engineering Research. State Universityof New York. 1994. [IS] BiggsJM.Introductionto swctuml dynamics. New Yo& McGraw-Hill. 1964. [171 Sddii M & Sozen MA. Simple nonlinear seismic analysis of R/C strucwres.loumalofthe Structural Division, ASCE I98 I: 107: 937-952. [181 LinJ & Mlhin SA. Effect of inelasticbehavior on the analysisand design of earthquake resistant structures. Report UCB/EERG85/08.Berkeley, California: Earthquake EngineeringResearchCenter, Univenity of California 1985. [191 HartJD& Wilson E L simplified earthquake analysisof buildings includingsite effects. Report UCBISEMM-t99/23. Berkeley, California: Departmentof Civil Engineering, Universityof California. 1989. D O ] AndersonJC. Dynamic response of buildings. In: Naeim F (ed) Seismic design handbook. Chapter 3. New Yo& Van Nostrand Reinhold. 1989. pI] QiX & MoehlcJR Displacementdesign approach for reinforced concretestructures subjected t o earthquakes. Report UCB/EERG9 1/02,Berkeley California: Earthquake EngineeringResearchCentre, Universityof California I99 I. p Z ] Bernal D. Instability of buildings subjected to earthquakes.Jwmolofthe StructumlDivision,ASCE 1992: I18: 2239-2260. [23] Montanr FJ& Alarcon E. Simplifiedcomputational methodfor nonlinearseismic analysis of bridges. EuropeanEarthquake Engineering1996 2: 14-23. [24] Krawinkler H . New trends in seismic design methodology. In: Duma G (ed) Proceedings 10th Eur+an Confemce M EarthquakeEngineering,Vienna Vol2. Rotterdam: Balkema 1995.82 1-830. [25] Nasrar AA & Krawinkler H. Seismic demands for SDOF and MDOF systems. Report No 95. Stanford, California:JohnA Blume Earthquake Engineering Center, Sfanford University. I99 I. [ZS] Bazzurro P & Cornell CA. Seismic r i s k non-linear MDOF structures. Proceedings I 0th World Conferenceon Earthquake Engineering, Madrid, 1992.563-568. [27] SeneviratnaGDPK& Krawhkler H. Strength and displacement S National demandsfor seismic design of structural walls. Proceedings5th U Conferenceon Eonhquoke Engineering, Chicago. Illinois. Vol2. 1994. I 8 I 190. [ZS] Rahnama M & Krwinkler H .Efferrc of sok soils and hysteresis modelson seismic design spectra. Report No 108. Stanford, California:JohnA Blume Earthquake EngineeringCenter, Stanford University. 1993. [29] Bertem RD I Bertero W . Tall reinforcedconcrete buildings: conceptual earthquake-resirrantdesign methodology.Report UCBIEERG92/I6. Berkeley, California: Earthquake Engineering ResearchCenter, Universityof Californi. 1992. [30] MoghadamAS & Tso WK. 3-D pushover analysisfor eccentric buildings. Promdings 7th Canadian Conferenceon Earthquake Engineering(7CCEE), Montreal, Canada, 1995.285-292. [3 I] Kilar V & Fajfar R Simple push-over analysisof asymmetric buildings. journalofEarthquake Engineering and Structural Dynamics I997 26 233-249. [32] Tro W K & MoghadamAS. Seismic response ofasymmetrical buildings using push-overanalysis. ProceedingsWorkshop on Seismic DesignMethodologiesfor the N u t Generationofcodes. Bled, Slovenia, 1997. Rotterdam: Balkema 1997. [33] Moghadam AS & Tso WK. Pushover analysisfor asymmetrical multistorey buildings. Paper submiaedto 6th U S Nationol Conkmnce on Earthquoke Engineering, Seattle, Washington, 1998. (in press).

Referencesand recommended reading

Papers of particular interest have been marked Special interest Exceptionalinterest

[I] Applied TechnologyCouncil. A critical review of current approaches t o earthquake-resistantdesign. ReportATG34. Redwood City, California: Applied Technology Council. 1994.

p] Applied TechnologyCouncil. Guidelines for the seismic rehabilitation of buildings. ReportATC33.03. RedwoodCity, California: Applied Technology Council. 1995.
[3] Applied TechnologyCouncil. Guidelines for the seismic rehabilitation of buildings. Volume I :Guidelines: Volume II: Commentary. F E M A 273/27CBollot Version. (Prepared by the Applied Technology Councilfor the Building SeismicSafety Council). Washington D C Federal Emergency ManagementAgency. 1996. * [4] Applied TechnologyCouncil. Seismicevaluationand retmfK of concrete buildings. ReportATUO, 2 volumes. RedwoodC i . California: Applied Technology Council. 1996.
Agood source of information for the practical use of pushoveranalysis.The capacity spectrum method for estimatingthe target displacement is well explained. Several akernativetechniquesare introduced and examples are provided.

[5] SEAOC. Performance bared seismic engineering of buildings. Vision 2000. Sacramento, California: StructuralEngineersAssociation of California. 1995. [S] Faj6.w P & Gaspersic R The N2 method forthe seismic damage analysis of RC buildings.lournalofEanhquake Engineeringand StructumlDynomics I996 25: 3 146. [7] Sddii M & Sozen M. Simple and complex models for nonlinear seismic response of reinforcedconcrete structures. Universityof Illinois, StructuralResearch Series Report No 465. Urbana-Champaign. Illinois: Universityof Illinois. 1979.

0CONSTRUCTION RESEARCH COMMUNICATIONS LIMITED I998 ISSN 1365-0556

344
[34] FreemanSA, NicolettiJP& Tpll JV. Muation of edsting buildings for seismic risk- a case study of Puget Sound N m l Shipyard, Brememn, Washington. h e d i n g s f r t U S Conference on EorthquokeEngineering, 1975. [35] MahanayJA, P y e t T F ,&hoe BE I Freeman SA.The capacity spectrum methods for evaluatingstrucartal response duringthe Lorna Prieta earthquake.Proceedings Notionol Eorthquoke Confirence. Memphis, 1993.50 1-5 10. [36] Collins K R A reliabilii-based dual level seismic design procedurefor buildingsmctures. Eorrhquok specoo I995 II:4 17429. [37l Collins KR, Wen YK & Foutch DA. Imrcigation ofalternative design procedures for standard buildings. Structumf ReseorchSeries Repon No 600. UrbanaChampaign, Illinois: University of Illinois. 1995.

STRUC~URAL ANALYSIS AND CAD * [38] Lomon RS. Vance V & Knm'nkler H. Nonlinearstatic pushover
analysis: why, when and how?Proceedings5 t hU S Notionof ConFmceon Eorthgu& Engineering, Chicago, Illinois. Vol I. 1994.283-292.

The issues relatedt o feasibili, limitationsand the applicabili of pushover analyris are addressed.

[39] National Rerevch Council of Canada. NationalBuildingCode of Canada: Subsection 4. I.9. Ottawa, Canadr National ResearchCouncil of Canada. 1995. [40] Li K-N. CANNY-C a computer p r o p m for 3D nonlinear dynamic analysis of buildingrvuctures. ReKorch Repon No CE004. Singapore: National University of Singapore. 1993.

W K Tso BSc MS PhD

Professor, Department of Civil Engineering UHE 330), McMaster University, Hamilton, Ontario, Canada L8S 4L7 E-mail: [email protected]

A S Moghadam

BS MS

PhD Candidate, Department of Civil Engineering UHE 330). McMaster University, Hamilton, Ontario, Canada L8S 4L7