Abstract
Consequence analysis is an important part of the risk assessment technique, where fire consequence models have to be used to predict the consequences of a given fire scenario inside a compartment. As fire models use several input parameters to construct the fire scenario of concern, sensitivity analysis is considered suitable for application to determine how the fire model output depends upon these input parameters. CFAST is one of the most common fire models. To investigate its sensitivity, a number of simulations were conducted considering a single room fire scenario representing a fire in a cabin aboard a Ro-Pax vessel. Four selected input parameters were varied about a base scenario. Both small and large variations were studied. The results indicate that fire growth parameter and peak of heat release rate HRR are the most important input parameters influencing most of CFAST outputs. Simple response-surface correlations were then used to study the sensitivity of CFAST outputs to only the HRR with two door statuses (open/closed). The results demonstrated that CFAST outputs are more sensitive to changes in HRR when the cabin door is closed, i.e., when the fire is ventilation controlled.