Agent-based modeling for household level hurricane evacuation

In this paper, we present an agent-based model for hurricane evacuation which captures the interactions among evacuees during the evacuation process. Both normal and greedy household agents are considered. The normal agents simply follow the shortest route toward their destinations regardless of the congestion on each route segment. The greedy agents on the other hand make choices between the shortest route and the least congested route according to a certain probabilistic rule, taking into account the evolving network conditions. In the experiments, we simulate different scenarios on a test network and present the results and analysis. The simulation results indicate that greedy agents can sometimes reduce the individual evacuation times by detouring and avoiding congestions, but the greedy behavior makes the whole evacuation inefficient.