Numerical Simulation of a Dual-Mode Scramjet Using a RANS and Hybrid LES-RANS Approach

In this work, a numerical simulation study is carried out on a dual-mode, single ramp-injected, hydrogen-burning scramjet located at the University of Virginia (UVa), using both RANS and Hybrid LES/RANS simulation techniques. The solver used is NC State's REACTMB, which uses Edwards' LDFSS flux-splitting method and higher-order data reconstruction to solve reactive flows on three-dimensional structured meshes. Two chemical kinetics mechanisms are utilized: Jachimowski 1992 and Burke. In addition to laminar chemistry, three simple subgrid turbulence-chemistry closures are implemented. Efforts are also made to accommodate some facility-dependent parameters such as thermal non-uniformity of the facility inflow.;The numerical results are compared with an array of high-fidelity diagnostic information gathered from the UVa scramjet experiments. The diagnostic data available includes hydroxyl planar laser-induced fluorescence (OH-PLIF), coherent anti-Stokes Raman scattering (CARS), focusing Schlieren, stereoscopic particle image velocimetry (SPIV), and combustor wall pressure measurements. Two separate flowpath configurations are considered; for each, two equivalence ratios are simulated. Configuration A (∼33 M. cells) has no isolator and operates in scram mode for both &phis; = 0.17 and &phis; = 0.34. Configuration C (∼66 M. cells) includes an isolator and a longer combustor and transitions from scram to ram mode between &phis; = 0.17 and &phis; = 0.49.;RANS simulations of both configurations under scram conditions predict a flame anchored mainly to the injection wall of the combustor in the near-injector field, and loosely to the back face of the injector itself. There is an abrupt increase in flame temperature three to four ramp heights downstream of injection. In contrast, Hybrid LES/RANS simulations predict higher reactivity in the recirculating flow immediately behind the injector and also in the shear region between the core flow and the fuel jet. This leads to a higher pressure rise at the point of injection and a less abrupt pressure increase further downstream. The agreement between predicted average flame temperatures and CARS measurements is substantially better with Hybrid LES/RANS than with RANS; this is primarily due to the resolution of large turbulent structures and the resulting lower flame temperatures in Hybrid LES/RANS simulations. During Configuration C mode-transition simulation, the combustion shock train settles at a position slightly further downstream than experimentally observed. This results in lower predicted combustor pressures compared to experiments.;OH-PLIF data taken during scram operation indicates the presence of small turbulent structures within the flame which serve to broaden local flame structures and are not captured in the Hybrid LES/RANS simulations. Comparisons with CARS data indicate an over-prediction of average flame temperatures and species mixing rates, especially in the highly turbulent flameholding region just downstream of the fuel injector. This overprediction is attributed to solver deficiencies in regions of very high levels of turbulence and also an improper capturing of the levels and scales of turbulence entering the combustor due to a thin boundary layer (compared to mesh width) at the inflow. The use of various subgrid turbulence-chemistry closures in Configuration C provides some improvement in the prediction of flame temperatures but does not improve the prediction of turbulent mixing rates. Forcing the isolator to operate in full-RANS mode results in better prediction of the latter.;Post-simulation analysis using laminar flamelet theory is conducted for a Hybrid LES/ RANS simulation of Configuration C operating in scram mode. The subgrid instantaneous scalar dissipation rate is modeled using an estimate of the subgrid spatial variance of mixture fraction. The analysis reveals a region of high instantaneous scalar dissipation rate initially concentrated within the fuel jet and suddenly broadened by a shock / flame interaction. Also observed are resolved flame extinction (or perhaps suppressed ignition) events which occur near the injector due to high scalar dissipation rates, likely caused by the aforementioned shock / flame interaction. (Abstract shortened by UMI.).