Quantitative comparison of FDS and parametric fire curves with post-flashover compartment fire test data

This paper presents a comparison of two parametric fire modelling techniques (Eurocode 1, and the BFD curve method) and one field model (fire dynamics simulator) against large-scale post-flashover test data. Using a method of the product moment correlation coefficient, it is shown that the BFD curve predictions are most closely representative of reality. For the computational test data, two grid resolutions are adopted in the FDS field model, the finer of which having comparable results in terms of regression analysis to the BFD method. Both the field model and the BFD curve method were found to give better predictions compared to the Eurocode method over the duration of the test. However, a direct comparison of the maximum gas temperatures shows the field model to be poorer in its predictive capability than the parametric methods, under-predicting the maximum gas temperatures. In addition, a more in-depth analysis of the FDS predictions indicates that by considering simply average compartment temperatures the more inaccurate spatially specific temperature predictions were disguised. This study provides useful quantitative data on the three techniques presented and discusses more general issues concerning fire modelling.