Large eddy simulation of turbulent combusting flow

Advances in the performance of nearly all kinds of power plants including various types of engines and central power stations depend crucially on improved control and optimization of the turbulent combustion process. Besides the experimental research, which encounters considerable difficulties in measuring all relevant quantities of turbulent combusting flows, the numerical simulation represents a powerful tool to provide deeper insight into the dynamics of turbulent combustion. To date the concept of Large eddy simulation (LES) has proven successful in the computation of many types of turbulent flow. In contrast to the computationally unfeasible Direct Numerical Simulation (DNS) the LES computes only the large scale ("filtered" or grid-scale) motions directly, while the effects of the small scale motions occurring at the so-called subgrid-scale (SGS) are modeled. It is the objective of the present project to extend LES to the case of turbulent non-premixed combustion assuming finite chemistry and heat release. To this end a flow configuration of practical relevance and moderate complexity has been chosen: a methane-air co-flowing jet combustor. It is the first time that LES is rigorously applied to turbulent combusting flow involving fast but finite chemical reactions, the so-called flamelet regime, and heat release. Due to the strong interactions among the turbulent fluctuations in the scalar quantities of the reactive species and the chemical reaction processes inherent in turbulent combustion the development of an appropriate subgrid-scale model will be the key issue of the project. Using recent experimental and computational data this goal should be obtainable now. The first few steps already made into this direction performing LES of turbulent mixing layers with infinitely fast chemistry were very promising. To assure that the computations are accurate the results of the LES will be appraised by comparison to experimental data obtained in an experimental program accompanying the project. Since the proper modeling of the turbulent small scale motions is essential to a realistic simulation of any turbulent flow configuration, the improvements made in SGS modeling for the present LES of combustion apply to practically all kinds of turbulent, also chemically inert, flows. Beyond their relevance to the particular method of LES the results of the project may help to develop more realistic turbulence models used with other computational methods like the Reynolds-averaged simulations (RANS), as well.