Session F02: Turbulent Combustion II: Combustion Modeling

Evaluation of a conditional presumed subfilter PDF model in LES of turbulent nonpremixed sooting flames

In turbulent nonpremixed flames, soot evolution is strongly affected by soot-turbulence-chemistry interactions. In particular, soot nucleation and growth happen in fuel-rich regions, and soot is then rapidly oxidized before being transported by turbulence into fuel-lean regions (in non-smoking flames). Additionally, different soot evolution processes are dominant over different mixture fraction ranges. Therefore, a presumed subfilter PDF conditioned on mixture fraction is proposed to account for this non-uniform distribution of soot in mixture fraction space. In this model, the sooting mode of the presumed subfilter PDF is locally activated only at fuel-rich mixture fraction values where growth is locally faster than the oxidation. This model is implemented within a Large Eddy Simulation (LES) framework and applied to a series of turbulent nonpremixed sooting flames. Compared to the previous unconditional presumed subfilter PDF model, the soot volume fraction values are significantly increased and in significantly better agreement with the experimental measurements.   

Mixing of confined reacting co-axial jets with disparate viscosity

Mixing of miscible liquids with disparate viscosity is important in a number of industrial processes including reacting flows. When the reaction rate is much faster than the rate of mixing (Da >>1), the reaction becomes mixing limited. The relevant scale, Batchelor, for the reaction is much smaller than the turbulent dissipative scale for liquids (Sc ≈ 1000). Therefore, large-eddy simulation (LES) is the practical approach whereas DNS is inaccessible. However, subgrid-scale (SGS) modeling becomes challenging since there is large viscosity difference between the liquids and the mixture involves reacting fluids. In this study, we focus on mixing of two miscible liquids of different viscosity in a co-axial jet mixer. The experimental setup includes PIV and PLIF to resolve the velocity field and the mixture fraction. The computational approach is based on LES and the results with regular Smagorinsky, dynamic and dynamic mixed SGS models will be presented with comparison to experiments for the viscosity ratio of one. The experimental setup and the computational results for viscosity ratio of up to 1000; and the challenges in the SGS modeling of the reacting flows will be presented.   

Abstract Withdrawn

Regime independent multi-scale models such as the linear-eddy mixing (LEM) model has been developed in the past as a subgrid model large-eddy simulations (LES). In this two-scale approach finite-rate kinetics and reaction-diffusion interactions are included without any filtering within the LES cells thereby enabling a prediction of the filtered reaction rate rather than modeling it on the LES level. In this study, the LEMLES methodology is extended to use well-known flamelet/progress variable (FPV) approach within a compressible (high pressure) LEM model. Simulations of premixed flame-turbulence interaction in a channel for various regimes (corrugated flamelet to broken reaction zones) for a range of pressure is carried out and results compared to the conventional LEMLES method and prior DNS data. The subgrid FPV model is also analyzed to assess the assumptions and closures used in conventional LES using the filtered FPV approach.