Large eddy simulation of turbulent diffusion flames and pool fires

Kang, Yun (2004) Large eddy simulation of turbulent diffusion flames and pool fires. (PhD thesis), Kingston University, .

Abstract

In this dissertation a study of numerical simulations of turbulent diffusion flames and pool fires is presented. In order to account for the physical coupling of turbulent mixing and combustion, the large eddy simulation (LES) technique is used. The subgrid-scale (SGS) modelling for both turbulence and combustion are examined in details and a modified version of SGS combustion model has been proposed. For SGS turbulence modelling, the dynamic approach is used. This approach allows the model coefficient to be updated temporally and spatially and it can be used to account for the energy back-scattering. For SGS combustion modelling, a conserved scalar approach, namely the laminar flamelet model is used. Due to the tiny scale of combustion, it could be infeasible for modern computers to execute a direct calculation of an industrial combustion application. By treating the fire flame as an assembly of thin flames (flamelets), the laminar flamelet model has managed to separate the chemical reaction from the turbulence mixing. The calculation of laminar flamelet approach is relatively independent of LES. The calculations of turbulence and combustion are interacted by a conserved scalar called mixture fraction. Contribution has been made by the candidate to the application and optimisations of the SGS models. Those optimisations are based on the applications to pool fires and bluff body flows. In SGS combustion modelling, the variance of mixture fraction and the scalar dissipation rate are modelled from the mixture fraction rather than solving the governing equations. This simplification has dramatically cut the computational expense and has virtually turned the 3-D look-up table to a 1-D format. During the calculation of the heat release rate, the contribution of both reactants and products are considered. For pool fires, the constant thermodynamics pressure is used to effectively establish the relation between the temperature and density fields. Pool fires with different burner diameters and various types of fuels have been simulated using LES with the above SGS modelling. All cases are studied under 3-D mode. In addition to the analysis of the distribution of mean flow quantities (temperature, density, velocities, etc), considerable effort has been directed towards the sudy of the time development and the dynamic behaviours. Different characteristics have been identified for medium and small pool fires. The dynamic approach of SGS turbulence modelling has also been applied to the simulation of bluff body flows. The simulations were carried out using the LES package called Fire Dynamics Simulator (FDS), which is developed by the researchers in the National Institute of Standards and Technology (NIST), U.S.A. During the period of the Ph.D study, the FDS codes were updated several times by both the researchers in NIST and the candidate. The update covers the combustion modelling, radiation modelling, meshing and some other aspects. The simulations were carried out on a single processor Pentium IV PC with 2G-RAM. The number of cells in each simulation was generally between 1 and 2 million with the finest grid resolution being in the order of millimetre. The predictions are compared with the experimental measurement and other established simulation data. The ability of capturing the instantaneous flow movement and dealing with the realistic geometries has made LES with appropriate SGS modelling an effective and promising tool for the numerical study of turbulence and combustion. The laminar flamelet approach of SGS combustion modelling has established progressive relationship between the modelling of turbulence and the modelling of combustion. As a recommendation, the extending use of the dynamic approach has been proposed. With more accurate determination of the model coefficient in SGS modelling, LES is expected to cope with even higher Reynolds number flows in more complicated geometries.

Actions (Repository Editors)

Item Control Page Item Control Page