Large-eddy simulation and filtered mass density function approach to non-equilibrium turbulent combustion modeling

Turbulent combustion modeling is a challenging task due to the wide range of time and length scales of the physical processes involved. As government agencies tighten the regulations on pollutant emissions and efficiency of combustion devices, turbulent combustion modeling is expected to play an increasing role in the design of the next generation of internal combustion engines, turbine combustors, furnaces and burners. Among the most arduous task is the modeling of finite rate chemistry in non-premixed combustion. In the non-equilibrium combustion regime, chemistry and flow motion interact strongly and simplifying assumptions customarily invoked to reduce the complexity of the modeling task become unrealistic. At present, state-of-the-art industrial simulation methodologies do not capture such effect a priori, despite it being a key process in many combustion devices as far as emissions and efficiency are concerned. Large-eddy simulation coupled with a filtered mass density function (FMDF) approach has been considered as a promising method to capture non-equilibrium effects, making it possible to predict the occurrence of slow chemistry effects, extinction and possibly reignition.;In this work we show that an implementation of such approach is feasible, albeit computationally very expensive. Due to the high-dimensionality of the differential equations which need to be solved, a Monte Carlo, particle-based, technique is adopted. We present a brief summary of the mathematical foundation of the methodology. We explain the implementation details and illustrate our work on supporting methodologies that allowed to reduce computational wall-clock time, permitting to execute a simulation in a reasonable time frame. It is found that a hybrid large-eddy/FMDF approach is capable of reproducing available experimental data on Sandia/TUD piloted methane-air non-equilibrium jet flames. We investigate the sensitivity of our predictions to the choice of mixing model for subgrid-scale mixing, observing that one out of the three mixing models commonly used in turbulent combustion modeling resulted in unphysical flame extinction. A probability of extinction conditioned on the local fuel/air make-up (mixture fraction) is computed and it is found in excellent agreement with experimental data. The performance of mixing models and their impact on subgrid-scale statistics is further investigated by stochastic modeling of the ignition of a homogeneous hydrogen-air mixture under temperature stratification and turbulent mixing. It is found that mixing models predictions are in excellent agreement with detailed simulation results provided that an accurate estimate of scalar dissipation rate is available.