Scalable risk assessment of large infrastructure systems with spatially correlated components

Abstract

Risk assessment of spatially distributed infrastructure systems under natural hazards shall treat the performance of individual components as stochastically correlated due to the common engineering practice in the community including similarities in building design code, regulatory practices, construction materials, construction technologies, and the practices of local contractors. Modelling the spatially correlated damages of an infrastructure system with many components can be computationally expensive. This study addresses the scalability issue of risk analysis of large-scale systems by developing an interpolation technique. The basic idea is to sample a portion of components in the systems and evaluate their correlated damages accurately, while the damages of remaining components are interpolated from the sampled components. The new method can handle not only linear systems, but also systems with complex connectivity such as utility networks. Two examples are presented to demonstrate the proposed method, including cyclone loss assessment of the building portfolios in a virtual community, and connectivity analysis of an electric power system under a scenario cyclone event.