$site = "publications.iafss.org"; $fullsite = "publications.iafss.org"; $basePath = "/home2/firesag5/private/data/"; ?>
This study focuses on modeling experiments carried out by the Sandia National Laboratories Underwriters Laboratories. The four computer fire models used for the study are FIRST, BRI2 (and its earlier version BRIl), FkST and COMPBRN 111. The analysis shows that for those experiments simulated as gas burner fires, FIRST (with the Hasemi and McCaffrey plumes), BRIl and COMPBRN I11 give reasonable agreement with the experiments. FAST seems to overestimate the hot layer temperature, although not to the same extent as BRI2. It is also attempted to model the fires as pool fires with FIRST; the results show marked deviations from the experimental results, which suggests further study on modeling of pool fires is required. However, the depth of the hot gas layer, the mass of hot gases flowing out through the doorway and the mass loss rate of fuel were not recorded in the experiments. These important parameters would have otherwise clarified many points and would significantly further our understanding of the fire behavior and the modeling effort. Suggestions for close cooperation between f i e researchers and testing personnel are discussed with the hope that future exercises could be more effectively carried out.