Professional Documents
Culture Documents
www.elsevier.com/locate/engstruct
Review Article
Performance-based design in earthquake engineering: state of
development
*
Ahmed Ghobarah
Department of Civil Engineering, McMaster University, Hamilton, Ontario, Canada L8S 4L7
Received 28 February 2001; received in revised form 17 March 2001; accepted 17 March 2001
Abstract
The design objectives in current building codes address life safety, control damage in minor and moderate earthquakes, and
prevent collapse in a major earthquake. However, the actual reliability of the design in achieving the objectives is not known. There
is a general agreement among researchers and professionals that future seismic design needs to be based on achieving stated multiple
performance objectives. Future seismic design practice will be based on explicit performance criteria that can be quantified, consider-
ing multiple performance and hazard levels.
There are several challenges to be addressed before procedures for performance-based design can be widely accepted. The
development in performance-based design in seismic engineering will be directed towards the definition of performance objectives,
a general design methodology, issues of ground motion modeling, and demand and capacity evaluations. 2001 Elsevier Science
Ltd. All rights reserved.
Keywords: Performance based-design; Performance objectives; Earthquake hazard; Design criteria; Design procedures; Evaluation; Challenges;
Future trends
0141-0296/01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 0 1 ) 0 0 0 3 6 - 0
A. Ghobarah / Engineering Structures 23 (2001) 878–884 879
that includes traditional methods of seismic design with summary of trends and challenges, and review the most
significant upgrades. important contributions in the field.
There have been different interpretations of what is
meant by performance-based design [3–5]. The most
appropriate definition is that performance-based design 2. State of development
refers to the methodology in which structural design cri-
teria are expressed in terms of achieving a set of per- There is increasing agreement among researchers and
formance objectives. The state of practice and trends in professionals that future seismic design needs to be
performance-based design and performance evaluations based on achieving multiple performance objectives.
were discussed in several recent articles [6–12]. However, there are divergent viewpoints on the meaning
Performance-based design and displacement-based of performance-based design and its methods of
design have been used interchangeably. This is based on implementation. Three documents are credited with lay-
the idea that performance objectives can be related to ing the foundation for performance-based design con-
the level of damage to the structure, which in turn can cepts: SEAOC Vision 2000 [3]; ATC 40 [4]; and FEMA
be related to displacements and drift. However, this 273 and 274 [5]. The documents attempted to develop
assumption is an oversimplification since the level of procedures that can be used as seismic provisions in
damage is influenced by several other parameters such as building codes.
the accumulation and distribution of structural damage, The goal of SEAOC Vision 2000 [3] is to develop the
failure mode of elements and components, the number framework for procedures that lead to design of struc-
of cycles and duration of the earthquake, and the acceler- tures of predictable seismic performance and is able to
ation levels as in the case of secondary systems. An accommodate multiple performance objectives. The
attempt to develop a procedure to correlate damage of document presents the concepts and addresses the per-
various structural systems to drift, taking into account formance levels for structural and nonstructural systems.
various ground motion characteristics, was made through Five performance levels are described with specified
the use of a damage index [13]. For effective design limits of transient and permanent drift. It is suggested
criteria, the correlation between damage and drift must that capacity design principles should be applied to guide
be calibrated against the performance of structures in the inelastic response analysis of the structure and to
actual earthquakes. In fact, displacement-based design designate the ductile links or forces in the lateral-force-
may be thought of as a subset of performance-based resisting system. Possible design approaches include
design. The performance target can be any response various elastic and inelastic analysis procedures such as:
parameter attached to a certain threshold. A single (1) conventional force and strength methods; (2) dis-
design parameter such as displacement or drift may not placement-based design; (3) energy approaches; and (4)
adequately control all performance objectives for struc- prescriptive design approaches.
tural and nonstructural systems [14]. For example, force- In the Applied Technology Council ATC 40 document
or stress-based criteria are more appropriate for short [4], performance-based design refers to the methodology
period structures, when trying to achieve pre-yield limit in which structural criteria are expressed in terms of
state, than displacement-based criteria. achieving a performance objective. The document is lim-
Performance-based earthquake engineering (PBEE) is ited to concrete buildings and emphasizes the use of the
a more encompassing concept that includes design, capacity spectrum method. The procedure involves
evaluation and construction engineering [15]. The determining the capacity and demand spectra. To con-
Pacific Earthquake Engineering Research (PEER) Center struct the capacity spectrum, the force–displacement
has PBEE as its principal focus for research and devel- curve of a point on the structure is determined using
opment. Some progress has been made over the past nonlinear static (pushover) analysis. The forces and dis-
three years in the development of a global framework placements are converted to spectral accelerations and
and in various research aspects. spectral displacements using an equivalent SDOF sys-
The general methodology for performance-based tem. The demands of the earthquake are defined by
design may include various approaches. In one approach, highly damped elastic spectra. At the performance point,
traditional force-based analysis is conducted and, after the seismic capacity is assumed equal to the demand,
the design is completed, the deformation and damage which provides an estimate of acceleration (strength) and
may be estimated and checked against established dis- displacement (demand). The probability of occurrence of
placement limits. Other approaches may start by estab- the earthquake may be related to the risk of occurrence
lishing the displacement or drift associated with a certain of the associated damage state. Not all the components
performance, proportion the structure and then conduct of the procedure are well established. For example, an
the response analysis [16,17]. attempt was made to develop relationships between duc-
The objective of this study is to evaluate the state of tility and damping using perfect, hardening and softening
development of performance-based design, present a models [18]; however, further research and development
880 A. Ghobarah / Engineering Structures 23 (2001) 878–884
3. Performance objectives
4. Design evaluation
Performance objectives are statements of acceptable
performance of the structure. The performance target can Acceptable procedures for design evaluation include:
be specified limits on any response parameter such as (1) elastic analysis; (2) component-based elastic analysis
stresses, strains, displacements, accelerations, etc. It is procedure; (3) simplified nonlinear analysis methods;
appealing to express the performance objective in terms
of a specific damage state or the probability of failure Table 2
against a prescribed probability demand level [3,13]. Proposed earthquake hazard levels
Various documents [3–5] promote the same concepts but
differ in detail and specify different performance levels. Return
Earthquake frequency period in Probability of exceedance
Some of the suggested performance levels can be years
grouped in equivalent categories as listed in Table 1.
It is recognized that drift levels associated with spe- Frequent 43 50% in 30 years
cific damage categories may vary considerably with the Occasional 72 50% in 50 years
structural system and construction material. An attempt Rare 475 10% in 50 years
5% in 50 years or 10% in
was made to define drift levels for different structural Very rare 970
100 years
systems and materials [3]. However, more research is Extremely rare 2475 2% in 50 years
needed, particularly in the development of realistic and
A. Ghobarah / Engineering Structures 23 (2001) 878–884 881
and (4) dynamic nonlinear time history analysis. Simpli- [3–5] attempted to provide procedures that can be used
fied nonlinear analysis methods are based on pushover as seismic provisions of building codes, these develop-
analysis to determine capacity and on design spectrum ments require much supporting research in several areas
to represent demand. Some of the recent developments [20,36]. Some of the cited reasons are the current limited
include inelastic spectra [29], yield point spectra [30] ability to accurately predict deformation demands and to
and the N2 method [31]. At each design step, design accurately predict the inelastic building behavior [37].
evaluations may involve response parameters such as the There are several sources of uncertainties inherent in
stresses, drift and deformation, structural accelerations, the performance-based design process. The expectation
ductility demand ratios, and energy dissipation in terms that the approach will produce structures with predict-
of demand versus capacity. Typical limiting values for able performance may be only achieved in probabilis-
these response parameters need to be established for tic terms.
each performance level through research including lab-
oratory testing of specific components. The limiting 5.1. Design criteria
values may be calibrated by analyzing buildings that
have experienced measurable damage in seismic events A fundamental question in performance-based design
for which strong motion records are available. is to validate the appropriateness of the selected per-
The most realistic verification process is the prediction formance levels, the specific parameters used to define
of deformation and forces from inelastic time history their minimum performance, and the seismic hazard
analysis. For the analysis to be reliable and credible, it definitions. For the case of three performance levels
is necessary to ensure that: (serviceability, damage control and life safety or collapse
prevention), three corresponding structural character-
앫 appropriate site-specific ground motion with specified istics (stiffness, strength and deformation capacity)
hazard level can be generated with confidence; dominate the performance as illustrated in Fig. 1. If more
앫 the structural model is realistic; intermediate performance levels are selected, then it
앫 the cyclic load–deformation model for each element becomes difficult to define which structural character-
is representative of the behavior; istics dominate the performance. It can be argued that
앫 analysis procedures and interpretation tools are different performance objectives may impose conflicting
reliable; and demands on strength and stiffness [13]. Much research
앫 identification of modes and sequence of element and is needed to associate the displacement or drift limits
component failure are also realistic. with the damage states and the stated general perform-
ance objectives. The displacements or drift limits are
The static nonlinear pushover analysis may provide also functions of the structural system and its ability to
much of the needed information. In the pushover analy- deform (ductility). Design criteria may be established on
sis, the structure is loaded with a predetermined or adapt- the basis of observation and experimental data of defor-
ive lateral load pattern and is pushed statically to target mation capacity. For example, near the collapse point,
displacement at which performance of the structure is the drift limits of structural walls are different from a
evaluated [32,33]. The target displacements are esti- moment-resisting frame, which suggest that different
mates of global displacement expected due to the design structural systems will undergo unequal displacements.
earthquake corresponding to the selected performance Other issues related to the damage evaluation are the
level. Recent studies addressed limitations of the pro-
cedure [34] and the selection of lateral load distribution
including adaptive techniques to account for the contri-
bution of higher modes in long period structures [35].
quantification of the relationship between building resto- of element damage. Nonlinear static pushover analysis
ration time/costs and earthquake hazard level. It is of coupled with new methods (other than SDOF-based
interest to identify the damage level at which building spectra) to determine demand, or nonlinear inelastic
restoration becomes impractical, which represents the dynamic analysis, may provide a more reliable predic-
state of irreparable damage. tion of the performance.
[34] Krawinkler H. Pushover analysis: why, how, when and when not [37] Paret TF, Saki KK, Freeman SA. Performance-based engineering:
to use it. In: Proceedings of 65th Annual Convention of the Struc- can the engineering profession deliver the goods? In: Proceedings
tural Engineers Association of California, Maui (Hawaii). Sacre- of 6th US National Conference on Earthquake Engineering. Oak-
mento (CA): Structural Engineers Association of California, land (CA): Earthquake Engineering Research Institute, 1998,
1996:7–36. [CD-Rom].
[35] Mwafi AM, Elnashai AS. Static pushover versus dynamic col- [38] Poland CD, Derrick BH. Opportunities and pitfalls of perform-
lapse analysis of RC buildings. Eng Struct 2001;23(5):407–24. ance based seismic engineering. In: Fajfar P, Krawinkler H, edi-
[36] Freeman SA. Trying to define performance-based seismic engin- tors. Seismic Design Methodologies for the Next Generation of
eering: is there a consensus? In: Proceedings of 8th Canadian Codes. Rotterdam: AA Balkema, 1997:69–78.
Conference on Earthquake Engineering. Vancouver: Canadian
Association for Earthquake Engineering, 1999:495–9.