A dynamic programming optimization for traffic microsimulation modelling of a mass evacuation

Abstract

Abstract This study develops a novel framework to formalize the optimal utilization of all available modes of transportation , particularly transit and school buses for a mass evacuation. The study develops an “All-Mode Evacuation Decision Support Tool (AMEDST)” to determine an optimum auto-bus composition that yields an improvement in evacuation time and network congestion. The study follows a Knapsack optimization and adopts a solution algorithm called Dynamic Programming within a Python platform to optimally allocate buses to evacuees exposed to higher level of vulnerabilities. A traffic microsimulation model follows a dynamic traffic assignment process to simulate evacuation scenarios using all available modes. Results from the traffic simulation yield a vehicular traffic reduction of 3.9–7.7% and an evacuation time reduction of 9–22.7% if 5–20% of auto-based demand are served by buses. The tool will help emergency personnel evaluate alternative scenarios for making informed decisions regarding resource allocation and emergency budget policies for large-scale evacuations.