A Risk‐Based Approach to Improving Disaster Relief Asset Pre‐Positioning

Abstract

It is well‐recognized that pre‐positioning assets in advance of a natural disaster is important for supporting timely relief provision after a disaster occurs. In addition to preparing for imminent disaster events, humanitarian relief organizations also maintain such asset allocations over time to support future responses to recurrent events such as wildfires or winter storms. This study is motivated by the authors’ work with one such relief organization, the American Red Cross in Colorado and Wyoming, and it discusses a new mixed integer linear programming model that can help to more effectively pre‐position needed assets to open emergency shelters. The new model incorporates a measure of risk into its objective function, which helps address the need for generating equitable solutions, and the model explicitly considers reallocation of assets to nearby locations after a disaster event occurs. Model performance is first analyzed in the context of the Red Cross’ actual asset allocation problem and the results show a decrease in residual risk from that of historical allocations of more than 33%. The general applicability of the approach is then illustrated in the context of several different extensions to the base model and a series of important takeaways are presented and discussed.