| Flows through scramjet configurations are simulated using hybrid large-eddy simulation / Reynolds-averaged Navier-Stokes techniques. Present study is performed in three parts: parametric studies to determine the sensitivities of the predictions to modeling and algorithmic variations; formulation, implementation, and testing of several subgrid closures aimed at modeling filtered species production rates, which account for turbulence-chemistry interactions in a finite rate chemistry large-eddy simulation framework; and as a final assessment of the complete methodology, cavity-stabilized ethylene combustion is simulated. Throughout the present study, emphasis is placed on characterizing facility-specific effects, since they can have a significant influence on the numerical solution. In Part One, non-reactive and reactive flows through a model scramjet combustor with a wedge shaped injector are simulated. Different grids, flux reconstruction methods, reaction mechanisms, and inflow boundary conditions are used. To enhance fuel-air mixing, a synthetic eddy method is used to generate turbulence in the injector boundary layers and the hydrogen jets. The results show that in all the cases a lifted flame is predicted with varying standoff distances, heat releases, and shapes. In Part Two, the subgrid closures for modeling the filtered species production rates are tested on two different scramjet configurations with fundamentally different flow patterns and flame structures, one with the wedge shaped injector placed at the center of the combustor section (first, used in Part One), another with a three-dimensional ramp injector located on the upper wall of the combustor section (second). While the impact of these closures on the flow through the first configuration is insignificant, they have a more pronounced effect on the flow through the second configuration. Error analysis and performance quantification of these closures reveal that, relative to a baseline model, two of the closures improve the accuracy of the predictions, but the degree of improvement is quite modest. Also, from a cost-benefit perspective none of the models are a significant improvement over the 'laminar-chemistry' closure (where turbulence-chemistry interactions are ignored), for the configurations tested and the mesh resolutions employed. In Part Three, reactive flow through an ethylene fueled cavity flameholder is simulated using 14- and 22-species ethylene oxidation mechanisms, and the synthetic eddy method (used in Part Two) is used to introduce turbulence at the inflow plane of the flameholder. For an equivalence ratio of 0.15, the 14-species mechanism resulted in a flame blow-out, and the 22-species mechanism predicted a cavity stabilized flame. Results predicted using the 22-species mechanism compare well with the experimental data, especially, water mole-fraction distribution and pressure along the upper wall of the combustor. In general, the predictions show excellent agreement with experimental data within the cavity region; further downstream, experimental results suggest that the heat release is over-predicted in the simulations. |
