Optimization on Spatial Distribution of Shelter through Dynamic Evacuation Simulation of High Density Urban Area-Xinjiekou Case

Abstract

The optimal location of emergency shelters is essential for boosting the evacuation efficiency and resilience of cities. For determining evacuation demands and residents' evacuation allocations, a two-stage model for the optimization of shelter site that utilizes high-precision day and night population data derived from mobile signalling has been developed. In the initial phase, static network analysis served as the foundation for identifying the site arrangements with the smallest number of additional shelters via the enhanced set-covering model. The second-stage optimization determines the best location scheme with the minimum evacuation time using a modified social force model, along with the population's spatial heterogeneity, age structure, and evacuation route selection. Xinjiekou was chosen to assess evacuation efficiency. The utilization rates of shelters and the amount of congestion on the roads before and after optimization are assessed. The outcome demonstrated a 76.5% reduction in evacuation time, a vast improvement in the usage efficiency of shelters, and a significant reduction in congestion along the evacuation route. This method of location optimization marks a break from the usual empirical top-down research paradigm in favor of a bottom-up numerical simulation of the real evacuation process, which satisfies all local residents' and government's needs.