Author(s): Hao-Ming Hsu; Chang-Ying Lee; Ming-Che Hu
Linked Author(s):
Keywords: Agent-based modelling; Evacuation; Network flow analysis; Phase-adaptive optimisation; Urban flooding
Abstract: Intensifying rainfall and sea level rise under climate change are increasing the frequency of urban floods, making timely evacuation planning critical. To support evacuation decision making under urban flooding, this study develops a multi objective phase-adaptive evacuation optimisation model. A network flow analysis of the road systems is conducted, jointly considering link capacities, travel distances, network nodes and shelter capacities, and the risk associated with road segments located in potentially affected areas. We first run hydraulic simulations, and flood arrival times, water depths and inundation extents are translated into risk scores and safe time windows. These hazard variables are then used as inputs to the optimisation model for evacuation risk assessment. The optimisation seeks to minimise total evacuation time, risk exposure and shelter imbalance, while generating recommended routes from evacuees’ origins to shelters. An Agent-Based Model (ABM) is then employed to simulate evacuation for verification and calibration. Discrepancies between the theoretical optimum and ABM outcomes are compared across multiple scenarios and fed back to iteratively refine the optimisation. The study provides optimised evacuation networks, critical bottlenecks and evacuation strategy improvement to support local governments in evacuation planning and emergency response under urban flooding.
Year: 2026