On the choice of ground motion intensity measure and the number of records for cloud analysis-based seismic demand assessment

Abstract

Cloud analysis has emerged as a popular tool for the seismic demand/fragility assessment of structures. The output of cloud analysis is a seismic demand model which relates an Engineering Demand Parameter (EDP) indicative of structural distress to an Intensity Measure (IM) signifying the severity of ground shaking. IMs commonly used for probabilistic seismic demand assessment are quite heterogenous with respect to their “efficiency”, i.e. their degree of correlation with a specific EDP. This feature has serious implications on the number of ground motion records that must be used to perform cloud analysis on a given structure in order to accurately describe the distribution of the EDP at various IM levels. In the current study, demand models for maximum interstorey drift (?max), based on a wide spectrum of IMs, are developed from the cloud analyses of a five-storey RC bare frame structure using a suite of fifty unscaled natural ground motion records. The method of bootstrap resampling is used to investigate the convergence of the regression coefficients in the demand model with the size of the bootstrap subsamples, each comprising of a limited subset of records drawn from the original suite with repetitions allowed. This procedure helps determine the minimum number of ground motion records necessary for the calibration of demand models without compromising its accuracy in predicting the drift demands. Results from the study indicate a strong correlation between the efficiency of various IMs and the optimal number of records required to produce reliable seismic demand models.