Modeling the Post-disaster Recovery of Interdependent Civil Infrastructure Network

Abstract

Modeling the post-disaster performance of interdependent infrastructure systems contributes to strategic community resilience planning. The normal operation of the facilities in different infrastructure systems are dependent upon each other for product input or information sharing. However, when disaster happens, these dependencies would aggravate the initial damage caused by the hazards and lead to cascading failures. Thus, incorporating the dependencies among infrastructure facilities in modeling the damage and recovery of infrastructure systems under disruptive events is essential to guide the strategic pre-disaster risk mitigation and post-disaster recovery planning. The Dynamic Integrated Network (DIN) model is proposed in this study to simulate the damage and recovery of infrastructure network while considering the facility-level dependencies. The DIN model first assesses the inoperability of the network nodes and links over time to simulate the damage and recovery of the dependent civil infrastructure facilities, and then assesses the recovery and resilience of the individual infrastructure systems and the integrated network utilizing some network performance metrics. The proposed DIN model is illustrated with a hypothetical infrastructure network, consisting of interdependent power, water and telecommunication systems under a scenario hurricane hazard. The recovery simulation result from the proposed model is compared to with no interdependency considered, and with only system-level interdependencies considered. This comparative study suggests that the recovery time would be underestimated if no interdependency was considered, or be overestimated if only system-level interdependencies were considered, both of which would lead to poorly informed decision making. The DIN model is then validated through simulating the recovery of the interdependent power, water and cellular systems of Galveston City, Texas after Hurricane Ike (2008). The simulated power system recovery time is comparable to the actual time, which demonstrates that the proposed DIN model can produce comparable results to physical reality.