Relieving bottlenecks during evacuations using iot devices and agent-based simulation

Abstract

Most of the existing studies on relieving bottlenecks have aimed to develop route-finding algorithms that consider structural factors such as passages and stairs, as well as human factors such as density and speed. However, the methods in providing evacuation routes are as important as the route-making algorithms because a secondary bottleneck could occur continuously during evacuations. Even if an evacuation system provides the same routes to all evacuees regardless of their locations, secondary bottlenecks could happen following the initial bottlenecks due to people rushing toward uncrowded exits all together. To address this issue, we developed a location-based service (LBS) evacuation system prototype that provides optimized-alternative routes to evacuees in real time considering their locations in indoor space. The system was designed to relieve continuous bottlenecks, which relies on installed IoT sensors and beacon machines which detect bottlenecks and provide updated routes, separately. Next, we conducted agent-based simulations to measure the system’s effectiveness (evacuation time reduction and dispersion of evacuees) by changing the system parameters. Simulation results show the evacuation time decreased from 100 to 65 s, and the number of people who took a detour to avoid bottlenecks increased by 28.66% out of the total evacuees with this system. Since this system provides the theoretical solution for distributing evacuees, it can be flexibly employed to a disaster situation in a large and complex indoor environment by applying to other evacuation systems. Moreover, by adjusting parameters, we can derive maximum evacuation effectiveness in other indoor spaces. Future work will consider demographic features of population and multilayer building structure to draw a more accurate pedestrian flow.