Optimizing Shelter Site Locations in Residential Community: A GeoSimulation and Genetic Algorithm Approach

Abstract

As disasters, both natural and human-induced, grow more frequent, effective sheltering and evacuation become crucial. While existing studies considered approaches optimizing shelter location by simulating individuals’ evacuation behaviors, most rely on pre-determined functions and overlook the heterogeneity of individuals’ behaviors. To address this limitation and account for the complexities of residents’ evacuation processes, we develop an approach that integrates Agent Based Modeling (ABM) with a Genetic Algorithm (GA), using the evacuation simulation results from ABM to optimize community shelter locations. We validate the proposed method by simulating residents’ evacuation in an assumed nighttime earthquake event within a student residential community in Shenzhen. This study provides a tool assisting in the optimization of community shelter locations in pre-disaster planning.