Shelter location-allocation for tsunami using floating population and genetic algorithm

Recently, large and small earthquakes have occurred in the Korean peninsula. In this sense, Korea is no longer considered as an earthquake free zone. Especially, it is necessary to respond quickly to earthquake tsunami which may be caused by the influence of neighboring countries with large earthquakes. Since the occurrence of tsunamis can cause great casualties, it is very important to allocate the location of the shelter in case of an earthquake. Although many researches on shelter allocation have been conducted in various ways, but most of them have been analyzed based on administrative district resident data, resulting in a lack of reality. In this study, floating population data were used to reflect reality in case of emergency situations, and genetic algorithm, which produce good results among the heuristic algorithms, was used to select shelter locations. The number of evacuees was used as a objective function of genetic algorithm and the optimal solution was found through selection, crossover and mutation. As a result of the research on Busan Haeundae-Gu, selected as a research area, allocating eight shelters was the most efficient. The location of the new shelters was selected not only in residential areas but also in major tourist areas whose results can not be derived from administrative district resident data alone, and the importance of utilizing the floating population data was confirmed through this study.