On the Efficacy and Accuracy of Models for Large Eddy Simulations of Turbulent Premixed Combustion

Abstract

Combustion and turbulence are two phenomena that are challenging to build theoretical and computational models for due to the complex nature of each individual process, so turbulent combustion is perhaps one of the most vital, yet poorly understood, aspects of fluid mechanics. In particular, a deeper understanding of unresolved sub-grid and sub-filter scale effects related to these phenomena is needed to develop efficient models for simulating turbulent combustion in large eddy simulations. The ecacy of proposed adaptations of Reynolds-averaged Navier-Stokes models for capturing unresolved sub- lter interactions between combustion heat release and turbulence is studied and improvements are identi ed. Algorithms to filter direct numerical simulation databases are developed to enable the computation of sub-filter scale interactions. A priori analysis of direct numerical simulation databases of premixed turbulent hydrogen jet flames using these tools con rms that recently proposed models, parameterized by a proposed lter Damk ohler number, can account for the effects of combustion heat release on turbulence compared to conventional diffusion-driven turbulence models. Scalar and velocity statistics are combined for global sensitivity analysis of the influence of heat release on sub- lter scale quantities, leading to the establishment of pseudo-optimal values for model parameters. Furthermore, the accuracy of various reduced-ordered manifolds for combustion modeling is studied, and a posteriori evaluation, using unique time-averaging features of the tools, demonstrates the inadequacy of the proposed models in capturing shear phenomena and streamwise effects. Finally, the need for more complete and complex turbulence and chemistry models to even better represent sub-filter scale effects is established.