Simulating Metro Station Evacuation Using Three Agent-Based Exit Choice Models

Metro systems are playing an essential role in transiting large passenger flows in metropolitan areas. Since commuting with Metro systems is beneficial from cost and time-consuming aspects, a vast number of people use this mode of transport. The high density of passengers in Metro stations during rush hours raises the significance of considering prevention safety planning in an emergency. Fire is the primary underground safety issue. In this paper, the evacuation of a Metro station is simulated using an agent-based framework, and three different exit choice models are compared. These models are the shortest path exit choice, a multinomial logit model, and the modified multinomial logit model with revising decisions. These approaches are verified with the previous experiments. Then, the emergency evacuation of a transfer Metro station in Tehran is simulated with different scenarios in an open-source agent-based platform. The egress time of each scenario is obtained and compared with each other. Results indicate that the modified exit choice model outperforms the two other models due to the realistic representation of human behavior. Also, the worst possible evacuation scenario's result reinforces the provision that is not allowing the evacuation of two trains at two platforms and at the same time.