Development and Validation of a Thickened Flame Modeling Approach for Large Eddy Simulation of Premixed Combustion

Abstract

The development of a dynamic thickened flame (TF) turbulence-chemistry interaction model is presented based on a novel approach to determine the subfilter flame wrinkling efficiency. The basic premise of the TF model is to artificially decrease the reaction rates and increase the species and thermal diffusivities by the same amount, which thickens the flame to a scale that can be resolved on the large eddy simulation (LES) grid while still recovering the laminar flame speed. The TF modeling approach adopted here uses local reaction rates and gradients of product species to thicken the flame to a scale large enough to be resolved by the LES grid. The thickening factor, which is a function of the local grid size and laminar flame thickness, is only applied in the flame region and is commonly referred to as dynamic thickening. Spatial filtering of the velocity field is used to determine the efficiency function by accounting for turbulent kinetic energy between the grid-scale and the thickened flame scale. The TF model was implemented into the commercial computational fluid dynamics code FLUENT. Validation in the approach is conducted by comparing model results to experimental data collected in a laboratory-scale burner. The burner is based on an enclosed scaled-down version of the low swirl injector developed at Lawrence Berkeley National Laboratory. A perfectly premixed lean methane-air flame was studied, as well as the cold-flow characteristics of the combustor. Planar laser induced fluorescence of the hydroxyl molecule was collected for the combusting condition, as well as the velocity field data using particle image velocimetry. Thermal imaging of the quartz liner surface temperature was also conducted to validate the thermal wall boundary conditions applied in the LES calculations.

An Attempt at Large Eddy Simulation for Combustor Modeling

“…That is, the effect of imposed turbulence perturbation on the LES results is negligible for the present configuration. A Similar observation is also reported by Strakey & Eggenspieler [8], where a lean premixed swirl burner was studied with LES, and the inlet flow perturbation was generated by the same method with a turbulent kinetic energy of 5% of the mean flow. Fig.…”

“…*Address correspondence to this author at the Gas Turbine Laboratory, Institute for Aerospace Research, the National Research Council of Canada, 1200 Montreal Road, M-10, Ottawa, Ontario, K1A 0R6, Canada; Tel: 613-993-9235; Fax: 613-952-7677; E-mail: leiyong.jiang@nrc-cnrc.gc.ca There are a fairly large number of publications in LES related to gas turbine combustion such as [2][3][4][5][6][7][8][9]. However, simulations with detailed quantitative validation against well-defined experimental results are rare, with few exceptions.…”

Toward finite-rate chemistry large-eddy simulations of sooting swirl flames

“…Fulton et al 2016for example obtained reasonable LES-QLC results using an average mesh spacing which was half the laminar flame thickness. Another example is the LES-QLC simulation of Strakey and Eggenspieler (2010) which showed a good agreement to OH-measurements and little differences compared to LES-ATF (Colin et al, 2000) and LES-EBU (eddy-break-up model) (Spalding, 1971) simulations. This good performance has been attributed to a relatively fine grid.…”

“…While this approach is not applicable to (U)RANS (Duwig et al, 2011), some authors report that it can be a reasonable approximation for LES if the flame is 'adequately' resolved (Duwig et al, 2011;Edwards et al, 2012;Fulton et al, 2016;Hodzic et al, 2017;Lourier et al, 2015;Potturi and Edwards, 2015;Ranjan et al, 2016;Strakey and Eggenspieler, 2010;Zhang et al, 2015Zhang et al, , 2010. Fulton et al 2016for example obtained reasonable LES-QLC results using an average mesh spacing which was half the laminar flame thickness.…”

Investigation of temporal variation of combustion instability intensity in a back step combustor using LES

“…The effect of combustion product is taken into account as S comb,k . E and F are the efficiency function and flame thickening factor and Ω is flame sensor from Dynamically thickened flame model (Legier et al, 2000, Charlette et al, 2002, Strakey and Eggenspieler, 2010, Proch and Kempf, 2014, Rittler et al, 2015. This model ensure that there are at least 8 grid points across the flame thickness.…”