Evaluation of correlation between engineering demand parameters of structures for seismic system reliability analysis: Fid-Bau Portal

Evaluation of correlation between engineering demand parameters of structures for seismic system reliability analysis

Kang, Chulyoung / Kwon, Oh-Sung / Song, Junho

Highlights • A new framework is developed to incorporate the correlation between EDPs of structures. • EDP correlation is affected by correlations of intensity measure and EDP residuals. • Incremental dynamic analysis is employed to estimate variance and correlation of EDP residuals. • Regional loss assessment demonstrates the improved accuracy of the proposed framework. • A regression model is proposed to estimate variances and correlations of EDP residuals efficiently.

Abstract As urban communities get sophisticated and complex, it is becoming increasingly important to assure the resilience of infrastructure systems against earthquakes, which may cause many casualties and critical socioeconomic losses. To this end, seismic system reliability analysis, such as regional loss assessment and network reliability analysis, needs to be performed with various uncertainties in seismic motions and structural responses properly considered. Previous studies pointed out that it is essential to incorporate the correlation between the failures of components for accurate seismic system reliability analysis. This paper proposes a new framework to evaluate the correlation between the failures of structures based on their engineering demand parameters (EDP). The proposed framework first defines EDP of structures as the sum of a regression function of the selected intensity measure (IM), and the residual termed “EDP residual”. The parameters of the regression model of EDP on IM are estimated based on incremental dynamic analysis (IDA) results. The variances and correlations of EDPs are estimated by incorporating ground motion prediction equation (GMPE) and statistical analysis of EDP residuals from IDA curves. Numerical examples of regional loss assessment demonstrate that the proposed method can correct the significant bias in estimating the probability of total loss, which is caused by an improper evaluation of the correlations. To facilitate applications of the proposed method, this study also proposes a method to predict the correlation between EDP residuals without performing IDA.