A location-allocation model and algorithm for evacuation planning under hurricane/flood conditions

The location of shelters in a region threatened by a hurricane can greatly influence the highway network clearance time, i.e. the time needed by evacuees to escape from origin points to safe areas. The studies conducted to date for developing hurricane plans have not implicitly considered the impact of shelter locations on evacuation times. The purpose of this research is to develop both a planning and an operational computer-based tool through a particular location-allocation model. This model selects a set of candidate shelters from among a given set of admissible alternatives in a manner feasible to available resources, and prescribes an evacuation plan which minimizes the total congestion-related evacuation time. An extraneous flow is also superimposed on the network in order to represent the traffic of evacuees not using the designated shelters as destinations. The model formulated is a nonlinear mixed-integer programming problem, for which we develop a heuristic and (two versions of) an exact implicit enumeration algorithm based on the generalized Benders' decomposition method. Computational experience is provided against a set of realistic test problems formulated on the Virginia Beach network. Potential avenues for future research are also explored.