A dynamic Bayesian network model for predicting congestion during a ship fire evacuation (original) (raw)

In this paper, a new simulation model to analyze congestions in ship evacuation is introduced. To guarantee a safe evacuation, the model considers the most important reallife factors including, but not limited to, the passengers' panic, the age or sex of the passengers, the structure of the ship. The qualitative factors have been quantized in order to compute the probability of congestion during the entire evacuation. We then utilize the dynamic Bayesian network (DBN) to predict congestion and to handle the non-stationarity of the scenario with respect to the time. Considering the worst-case scenarios and running the simulation for two groups of passengers (different in sex, age, and physical ability), we demonstrate the distinct effects of these groups on the congestion. The role of decision supports (DS), such as evacuation applications and rescue team presence is also studied. In addition, the impact of congested escape routes on the evacuation time is investigated. The results of this paper are of great importance for maritime organizations, emergency management sectors, and rescuers onboard the ships, which try to alleviate the human or property losses.