Turbulent premixed flames play a major role in modern internal combustion engines and gas-turbine combustors. As the trend is to reduce pollutant emissions and to increase efficiency, operating conditions at the limit of the engine stability are currently sought. To promote an efficient engine design process through accurate large-eddy simulations (LES), a better understanding of turbulent premixed flames in their regime of operation, typically the thin reaction zones regime, is desirable. In this talk, a recent numerical framework and results from direct numerical simulations (DNS) of premixed flames with heavy hydrocarbon fuels (with 35- to 200-species chemical kinetics mechanisms) at such high turbulence intensity will be presented.

These simulations were made possible by the development of a novel efficient semi-implicit time-integration scheme targeted for the simulation of turbulent reacting flows. Time-integration schemes previously used in the literature are either computationally expensive, may introduce lagging errors, or are inapplicable to large hydrocarbon fuels. The large series of DNS performed with this numerical framework provides important insights for the improvement/development of state-of-the-art LES closure models. The most striking result is that the structure of large hydrocarbon/air flames is significantly more affected by turbulence than that of methane/air flames. For instance, a n-heptane/air premixed turbulent flame can exhibit a 40% reduction in conditional mean burning rate (vs. progress variable) compared to its laminar counterpart, while very limited reduction is observed with a methane/air flame at the same turbulence intensity. It will be shown that these burning rate reductions are strongly affected by turbulent mixing and differential diffusion. It will be evidenced that, while effects of differential diffusion and turbulence intensity should be included in LES closure models, pressure and unburnt temperature can be accounted for by considering the turbulence intensity at the reaction zone (as opposed to the unburnt gas). Finally, it will be shown that premixed turbulent flames conserve a flamelet-like structure (in progress variable space) up to turbulence intensities significantly higher than predicted by the Peters or Borghi regime diagrams. As such, the importance of the dissipation rate and the diffusion of the progress variable as opposed to strain rate and curvature for source term closure modeling will be highlighted.