Detection of safe areas in flood as emergency evacuation stations using modified particle swarm optimization with local search

Abstract

Emergency evacuation throughout and after the flood is a crucial task to mitigating more instantaneous impacts, whilst refining social resilience for longer-term recovery. To enhance the evacuation process, the determination and prediction of safe areas before a flood is necessary. Indeed, the safe area or shelter in place could play two roles during the flood; as temporary shelters and as meeting points (station) for gathering before evacuation. This paper aims to determine the safe area according to the spatial and environmental characteristics of the urban extent (accessibility, topography, congestion, and land use). The main contribution is finding safe areas using modified particle swarm optimization (MPSO) with local search (LMPSO). The proposed method recognizes the optimal location of temporary shelters as evacuation stations. It has been implemented in Districts 3, 6, and 7 of Tehran, the capital of Iran. The comparison between the achieved results of MPSO and LMPSO demonstrated that the LMPSO is more efficient than the modified version. Since LMPSO is less sensitive to local minima and converged to minimum cost faster than MPSO, and the distribution of optimum locations of safe areas has been balanced, so all the population could benefit from these stations. The comparison among the results of MPSO, LMPSO, GA, ACO and genetic simulated annealing algorithms justified the efficiency of LMPSO too.