Joint pre- and post-disaster planning to enhance the resilience of critical infrastructures

Pre-disaster planning and post-disaster scheduling are common strategies to increase the resilience of critical infrastructure. In this paper, we propose a new two-stage stochastic optimization model to simultaneously determine the locations for building stations of restoration teams before disasters and their routing of conducting restoration tasks after disasters. Two layers of interdependent infrastructures are explicitly considered in the proposed model: The first layer for the transmission of flows to satisfy customers’ demands (e.g., a power transmission or telecommunication network) and the second layer for the transportation of restoration crews (i.e., the transportation nework). We seek to minimize the total cost, including the station building cost and the penalty cost incurred by performance loss. To deal with the computational complexity of the proposed problem, we propose a Column-generation-based Lagrangian Relaxation algorithm with a dynamic programming algorithm embedded to solve the included pricing problem. Extensive computational experiments, including the real-world infrastructures (i.e., French natural gas network), are conducted to illustrate the merits of the proposed approach.