A post-disaster resource allocation framework for improving resilience of interdependent infrastructure networks

Abstract Extreme events can greatly impact the functionalities of infrastructure networks. The ability of an infrastructure network to restore its before-the-event functionality after the occurrence of an extreme event, being recognized as one of the most important aspects of resilience, is also one area where the resilience of the infrastructure network can be improved. To expedite the recovery of the infrastructure network after an extreme event, it is essential to allocate limited repair crews properly to disrupted infrastructure facilities. The loss and recovery of infrastructure functionalities are further complicated by infrastructure interdependencies, which can lead to the propagation of the impact of an extreme event to all infrastructure facilities in the network. The interdependent nature of infrastructure networks also complicates the process of optimally allocating repair crews and the effectiveness assessment of an allocation strategy. This paper proposes a methodological framework, by combining agent-based modeling and reinforcement learning, to assess the effectiveness of a repair crew allocation strategy and optimize the strategy after an extreme event, with which the impact of the extreme event can be better mitigated.