A decision support system for post-disaster interim housing

摘要

The Northridge earthquake of 1994 displaced almost 10,000 families and destroyed major transportation infrastructure within Southern California, and Hurricane Katrina created the largest national housing crisis since the Dust Bowl of 1930, destroying over 300,000 homes and leaving over one million people seeking shelter. Numerous smaller disasters each year such as tornados, costal or inland flooding, and less severe earthquakes also destroy homes and displace families, although on a much smaller scale. Arranging housing for disaster victims ranks as a top priority after the immediate needs for food and medical care are met. This task becomes more challenging as families are displaced for a longer period of time due to increases in costs, government involvement, and expectations of the victims. In early 2009, FEMA released the first-ever National Disaster Housing Strategy which calls for improved planning and outlines the key principles and policies guiding disaster sheltering, interim housing, and restoration of permanent housing. While all three housing problems are very difficult, the provision of adequate temporary or interim housing is perhaps the most challenging. A few researchers have addressed the issue of optimal allocation of temporary housing, but have focused primarily on the first part of the problem which focuses on the selection of adequate capacity from among available interim housing alternatives. The second part of the problem, which consists of recommending housing alternatives to individual families from the pool of temporary housing units selected in phase one such that educational, healthcare, and socio-economic needs are met, has not yet been addressed to the best of our knowledge. We propose a decision support system for assigning families to housing units which addresses these needs. We develop a benchmark integer programming model for developing a balanced housing plan, and then use the model to evaluate three heuristics which could be practically applied with our system. We use a prototypical example to illustrate the model and evaluate the heuristics, and to demonstrate their appropriateness for the development of realistic real-time housing recommendations.