Enhancing Seismic Fragility Analysis of Structural System: Developing Intensity Measure and Ground Motion Selection Algorithm

Abstract

Structural collapse is the dominant cause of deaths and injuries under seismic excitation. Thus, collapse prevention of building during strong earthquake is the most important design objective of modern seismic design provisions to promote life-safety and to prevent socio-economic losses. In order to ensure an acceptably small likelihood of structural collapse under the earthquake load, nonlinear dynamic analysis coupled with probabilistic seismic hazard analysis is needed. However, nonlinear structural responses under seismic excitation vary greatly even if ground motions are scaled to get the same level of intensity measure (e.g., ground motions are scaled to get the same spectral acceleration at first mode period of structure). Furthermore, a large set of ground motions are needed for comprehensive reflection of hazard characteristics at a given site, which incurs high computational cost during dynamic analyses. To reduce the variability of structural responses as well as the number of ground motion time series used in nonlinear stochastic analyses, the study aims to develop a new seismic intensity measure by combining a cumulative IM, e.g. Arias intensity (Arias 1970) and a peak IM, e.g. spectral acceleration, and a new algorithm about selecting ground motion time series for IDA. To this end, various techniques of statistical methods such as linear regression, clustering analysis, and best subset selection method are employed. In order to demonstrate the proposed intensity measure (IM) and algorithm, nonlinear dynamic analyses are performed using a validated computational model of ductile steel frame structure and one of the reinforced concrete (RC) structural frames modeled by Haselton et al. (2011). It is found that using a developed IM and ground motion selection algorithm, one can obtain a reliable estimation on the collapse potential of structure using far less number of ground motion time histories with uncertainty reduced.