A joint demand and supply management approach to large scale urban evacuation planning: Evacuate or shelter-in-place, staging and dynamic resource allocation

Urban evacuation management is challenging to implement as it requires planning and coordination over a large geographical area. To address these challenges and to bolster evacuation planning and management, joint supply and demand management strategies should be considered. In this study, we explore and jointly optimize evacuate or shelter-in-place (SIP), dynamic resource allocation (DRA), and staging decisions for an efficient evacuation plan that minimizes total risk exposure of the population threatened by a sudden onset disaster. We introduce a Cell Transmission Model-based mathematical formulation and propose an exact solution methodology based on Benders decomposition. We further enhance the effectiveness of the algorithm by solving the Benders subproblem using a network flow based formulation on a time-expanded-network, and generating valid inequalities based on DRA decisions and for time-feasible solutions and develop an effective branch-and-cut algorithm to solve the master problem. We conduct extensive numerical experiments using realistic instances to test the effectiveness of the algorithm and to derive managerial insights. We find that considering evacuate or SIP, staging, and DRA decisions jointly contributes significantly to the effectiveness of the evacuation operations. A zone-based approach where some zones are ordered to evacuate while others shelter-in-place is superior to other approaches where an evacuate or SIP decision is given for all population at risk.