writer：Xie, Qimiao, Shouxiang Lu, Depeng Kong, and Jinhui Wang.
keywords：evacuation time, fire, Monte Carlo simulation, uncertainty, sensitivity analysis
source：期刊
specific source：Journal of Fire Sciences no. 30 (1):55-67.
Issue time：2012年

To quantify the impact of uncertain parameters on evacuation time in commercial buildings, Monte Carlo simulation with FDS + Evac is presented in this article. In addition, the partial correlation coefficients and partial rank correlation coefficients are adopted to determine the sensitivity of evacuation time to uncertain input parameters. To illustrate the effectiveness of research procedures, a hypothetical commercial building is described as a fire compartment and the data of evacuation time are obtained from the simulation results of the FDS+Evac, which is based on the social force model, developed by the VTT Technical Research Centre of Finland and fully embedded in Fire Dynamics Simulator (FDS). And the results indicate that Monte Carlo simulation with FDS + Evac can effectively quantify the uncertainty of evacuation time caused by the uncertainties associated with input parameters. Evacuation time value corresponding to the most likely occurrence scenario may be inappropriate to be selected as the typical value used in commercial building design. In addition, safety factors can be employed to deal with the uncertainties associated with input parameters of evacuation models, but it is obvious that the safety factor should be selected prudently in performance-based fire protection design. Moreover, the most significant factor is pre-movement time among uncertain input parameters considered, which is much more important than occupant density and occupant type. When the pre-movement time or effective width distributions are large enough, occupant density and occupant type can be regarded as constant at their nominal values. However, when the occupant density or occupant type distributions are large, their impact on evacuation time cannot be ignored and should be examined carefully.