EVACUATION PLANNING MODEL USING CHAIN FLOW ALGORITHM: CASE OF DIFFERENT EVACUATION SPEEDS

Natural disasters (e.g., earthquakes and typhoons) are frequent in Japan. Therefore, evacuation planning must be conducted well in advance before the disaster strikes. Network models are often used in evacuation planning, particularly time-expanded networks, but their computation time is considerable. Nevertheless, evacuation plans should be developed for various scenarios, by simulations. However, the computational cost of simulating scenarios is very high. Therefore, we propose using the chain flow algorithm for simulation and demonstrate that realistic evacuation times are obtained in the case of Nagaoka City Center, Niigata Prefecture in Japan. The computation time of the chain flow algorithm was significantly faster than that of time-expanded networks. In our previous study, we had assumed an average walking speed for evacuees. In a real-life scenario, evacuees would evacuate at various walking speeds. Hence, we considered different average walking speeds in an evacuee group comprising healthy people, elderly people, and people with disabilities. The results revealed that the chain flow algorithm returned a sufficiently reasonable evacuation time and computation time, even if we considered the average walking speed of different types of people in the evacuee group.