Numerical simulations on thermal plumes with κ-ε types of turbulence models

Building fire field models or application of computational fluid dynamics (CFD) for studying fire environment is used extensively for both academic research and practical safety design. There are three key elements in a field model: turbulence modelling, discretization of the conservation equations, and algorithms in solving the velocity-pressure linked equation. All three parts must be evaluated carefully in order to give a good model. In this paper, four turbulence models were assessed by simulating fire-induced thermal plumes where experimental data are available. Experiments on pool fires and heptane spray fires reported in the literature are taken as the example. Since the k-epsilon turbulence model is widely used for its robustness and simplicity, the standard k-epsilon turbulence model and its three other modified forms were tested. These were the low-Reynolds number (LRN) k-epsilon model, Chen-Kim modified k-epsilon model (CK model) and renormalization group (RNG)-derived k-epsilon model. Numerical results are compared with the experimental data. It is found that a more complicated form of k-epsilon model might not give better results. An alternative approach is to tune the parameters concerned on a simpler model. (c) 2005 Elsevier Ltd. All rights reserved.