An uncertainty analysis method for passenger travel time under ship fires: A coupling technique of nested sampling and polynomial chaos expansion method

Abstract

Passenger evacuation under ship fires is very complex and largely uncertain. For ship design, conducting an uncertainty analysis of passenger travel time requires a large number of evacuation simulations, which is very time-consuming. This work proposed a new method to quantify the uncertainty of passenger travel time based on a surrogate model, which is constructed by the polynomial chaos expansion with the nested sampling technique. To illustrate the application of the proposed method, a case of a 3-storey passenger ship with different functional zones is presented. The results demonstrated that the proposed method can be employed to quantify passenger travel time uncertainty with the acceptable accuracy. Specifically, travel time approximately follows the normal distribution under uniform distributions of input parameters. Safety factor of travel time cannot be regarded as a deterministic value, while it is significantly affected by its safety level. The parametric sensitivity analysis results suggested that for restaurant and retail zone the initial passenger density plays a leading role in the safety evacuation compared with other zones, while it is ignorable for bar zone and general area. This surrogate-based uncertainty analysis method of travel time can provide a tool to facilitate passenger evacuation design under ship fires.