Proceedings of the Second International Energy 2030 Conference,
November 4-5, 2008, Abu Dhabi, U.A.E.

CFD Modeling of Large-Scale Pool Fires

Arnaud Trouvé
University of Maryland, USA

Abstract
We review in this study current numerical and modeling challenges found in a computational fluid dynamics (CFD) treatment of large-scale pool fires. The numerical challenge comes from the need to suitably resolve flame geometries that are significantly more complex than those found at laboratory scales. The flame geometries found in large-scale pool fires correspond to multiple, relatively small flames and the increased small-scale activity results in more severe computational grid requirements. The modeling challenge comes from the need to provide adequate descriptions of flame extinction and soot processes. Flame extinction, soot mass leakage across the flame and the subsequent accumulation of cold soot in the overfire region are identified as key physical ingredients that will determine the performance of fire models in large-scale pool fire simulations. Current CFD capabilities are illustrated using the Fire Dynamics Simulator (FDS, developed by the National Institute of Standards and Technology, USA). FDS is applied to a series of numerical simulations corresponding to open, wind-free, gaseous pool fires of different sizes H, 0.4 ≤ H ≤ 40 m. The simulations reveal the fundamental change in flame structure that is observed as H is increased from laboratory- to large-scales. Predictions of combustion efficiencies, soot yields and radiant fractions show limited success, however. Some modifications in the FDS combustion and soot models are proposed in order to enhance the modeling capability. These modifications are based on the assumptions that: (1) flame extinction in pool fires correspond to slow mixing conditions combined with radiation cooling; (2) flame extinction is the dominant mechanism responsible for soot mass leakage across the flame. FDS simulations performed with these modifications show improved performance and provide some encouraging support to the assumptions made.