Developing Robust Facility Reopening Processes Following Natural Disasters

In the wake of a natural disaster, quickly reopening services and amenities is critical for community well-being and resiliency. However, in many situations, residents’ access to essential services, such as gas stations, grocery stores, and medical clinics, is limited by the severity of the disaster and the quality of the recovery plan. To improve the equity of access throughout the recovery period, and thus the speed with which a community recovers normalcy, we formulate an integer programming model to determine an optimal repair and reopening schedule for supermarkets after a natural disaster. We take into account census block level population data, driving distances between census blocks and stores, estimated repair times, and estimated work crew availability. While solving the nominal integer program can provide an ideal reopening plan, in practice, the proposed strategy could be far from optimal if repairs take longer than expected. To account for uncertainty in repair time and work crew estimates, we construct a robust counterpart to the nominal model. We demonstrate the value of using robust optimization by applying our models to data from Hurricane Florence in Wilmington, NC.