Heat transfer characteristics of a backward-facing step combustor

Large eddy simulations (LES) are conducted in this study to understand the convective and radiative heat transfer characteristics within a backward-facing step combustor. The Penn State backward-facing step combustor is modeled and the experimental signals are directly compared with computational results to validate physical models and numerical procedures. The baseline simulation features a wall-resolved LES of the full-combustor geometry for a lean methane/air mixture at an equivalence ratio of 0.55. A 16-species skeletal mechanism is employed with a dynamic thickened flame model to capture turbulence-chemistry interactions. A dynamic Smagorinsky model is employed to capture the subgrid-scale stress. A Monte-Carlo ray tracing based radiation solver is employed with a highly accurate line-by-line spectral database to post-process LES solutions to obtain the radiative heat fluxes. Comparison between the baseline results after accounting for experimental facility constraints show excellent agreement in radiative heat fluxes at four sensor locations. The total heat fluxes consisting of both radiation and convection is under-predicted by approximately 30%. Further parametric studies that use different spanwise dimensions, chemical kinetic models, molecular transport models, and thickening factors show that the better prediction of the temperature and flame speed of GRI-mech 3.0 can increase the prediction of convective heat transfer, while maintaining a similar comparison in the prediction of radiative heat transfer. The molecular transport model is also critical for the well-resolved LES to correctly capture the flame brush angles. The turbulence-chemistry interaction effects seem to be well-captured by the grid and have a negligible impact on the results. Compared to the reduced-span geometry that is frequently employed in backward-facing step configuration simulations, the full-span geometry is shown to be significant for capturing flame stabilization and heat transfer characteristics. Finally, limitation of this model validation study is discussed. • Direct comparison with experimental signal to reduce uncertainty in model validation. • Wall-resolved large eddy simulation in a confined geometry under reacting conditions. • High-fidelity radiation postprocessing to compare convection and radiation.