Optimal egress model considering bottlenecks in large multiplex buildings

In recent years, there has been a marked worldwide increase in the construction of large multiplex buildings serving a variety of functions, including offices, cafeterias, and commercial spaces. When a fire occurs in these buildings, smoke and flames spread in directions similar to the egress route, which increases the risk of large-scale casualties by creating bottlenecks in areas with high occupant density. Therefore, developing an algorithm that can minimize casualties by providing safe egress routes considering these bottlenecks is necessary. In this study, fire simulations were conducted for large multiplex buildings to analyze the correlation between fire temperature, visibility, and toxic gas concentration and to build a database. Based on this, we developed an algorithm to predict the real-time available safe egress time (ASET_i) at a specific location using an artificial neural network (ANN)-based model; the results confirmed that ASET_i can be predicted accurately. Furthermore, an algorithm was developed to estimate the number of occupants considering the bottleneck, an optimal egress route derivation system that reflects toxic gas and densely populated areas was proposed, and the reliability of the proposed model was validated.