Modeling the Evacuation of Large-Scale Crowded Pedestrian Facilities

Abstract

This paper presents a simulation modeling framework for evaluating the performance of large-scale crowded pedestrian facilities during emergency evacuation. The framework adopts a microsimulation assignment approach implemented in a cellular automata platform. It captures the different behavioral rules that govern the dynamics of evacuees' decisions, including exit gate choice, path choice, frequency of path updating, and the evacuees' tolerance of congestion. The framework is first used to investigate the sensitivity of evacuation performance to the behavioral rules that were introduced. Next, it is used to evaluate evacuation performance in a large pedestrian facility, with a capacity of about 50,000 pedestrians, that frequently operates at extreme congestion levels. Several experiments are designed to investigate the impact of different model parameters on evacuation performance. The results illustrate the impact that the evacuees' behavior can have on the performance of the evacuation process. In particular, overall performance of the evacuation process can be improved if evacuees are trained to make a better choice of gate, so that they weigh both the proximity of a gate and the level of congestion around that gate. Furthermore, when evacuees have the opportunity to learn about better exit opportunities, either on their own or through a guidance system, they can frequently update their exit decisions and the overall evacuation performance improves. In addition, the evacuation throughput generally improves as evacuees become less tolerant of congestion.