|
Abstract
|
Seismic risk analysis of spatially distributed systems requires considering a large number of earthquake scenarios. Monte Carlo Simulation (MCS) technique is a common and straightforward method to generate these scenarios. However, all of the generated scenarios are not necessary for seismic risk analysis. The current study proposes a two-step method for reducing the number of redundant scenarios and the computational expenses of seismic risk analysis for expanded lifeline networks. This method uses extreme value theory (EVT) and generalised Pareto distribution (GPD) and can be applied to different lifeline networks. It was shown that the GPD is an appropriate distribution for values above the different thresholds for PGA, PGV, PSA (T¼0.3 s) and PSA (T¼1 s) data. Moreover, 24 ground-motion models (GMMs) were developed as part of the extreme scenarios selection method and Bayesian regression analysis was carried out to calculate the GMMs coefficients. The extreme scenarios selection method was applied to two lifeline networks. It was shown that in the first and second examples, 100,000 scenarios were reduced considerably, to 23 and 60 scenarios, respectively. Hazard curves of extreme scenarios and probability density functions of bridge damage indices (BDIs) confirm the prosperity of the proposed method.
|