A Stochastic Optimization Model for Positioning Disaster Response Facilities for Large Scale Emergencies

In this paper, we present a two-stage stochastic program for locating disaster response facilities. Our modeling approach is unique in the literature in that it explicitly correlates the functioning of a facility with a particular disaster scenario. In particular, in our model the functioning of a facility is directly affected by its distance from the disaster epicenter. This represents an important modeling aspect of emergency facility location that to date has been ignored in the existing literature. To demonstrate the potential contributions of our model, we present a computational case study of (earthquake) disaster response facility location for the state of California. Our computational results show the distinct changes in the optimal location of facilities. Instead of placing facilities directly on top of some of the highest-risk areas (the traditional k-median solution), our model tends to place facilities still close to population centers, but farther away from high-risk areas.