Bayesian network-based seismic damage estimation for power and potable water supply systems

As critical infrastructures being more interdependent with each other and vulnerable to natural disasters, assessment of cascading failure in the coupled system becomes a significant task for ensuring reliability. Above all, the system interruptions caused by power outages have been paying attention to, and extensive research efforts have explored for the last decades. However, current studies tend to focus on system-level interdependency, which makes it hard to identify which component plays a critical role in the performance of infrastructure. In this context, this paper provides a method to estimate the service disruptions of power and potable water supply system after an earthquake. Two types of failures, common-cause and cascading, are modeled as a Bayesian network, which includes (a) an earthquake, (b) common-cause failure events, (c) cascading failure events, and (d) final service disruption. To quantify the probability of service disruptions under varying earthquake scenarios, a modified seismic ground motion prediction equation, fragility functions, and concept of functionality are used. The proposed Bayesian model is tested through case simulations run on data collected from Daegu City, South Korea. By presenting a total of 15 types of multi-state event nodes and all possible damage propagation causing disturbance of power and potable water supply systems, this paper gives knowledge for understanding the resilience of infrastructures in network perspectives.