No abstract

No abstract

No abstract

A computational study of sonic jet injection into a supersonic crossflow with and without combustion was performed. The gaseous injector was 3.226 mm in diameter and was injected 30 degrees to the horizontal. Simulated conditions involved sonic injection of hydrogen into a Mach 4 air cross-stream with a jet to freestream momentum flux ratio of 2.1. The numerical flow solver used was GASP v. 4.2. The inviscid fluxes were computed in three dimensions using third-order Roe Flux in the streamwise and lateral directions and third-order Van Leer flux in the vertical direction. The algorithms were chosen because of their robustness, shock resolution capabilities and efficiency. The Mentor Supersonic Transport (SST) turbulence model was used since the algorithm has good capability of solving both wall-bounded and free-shear flows. The reaction model used for hydrogen and air was a 9 species, 18 reactions model created by Drummond. The main results of this work can be summarized as: 1) modeling of combustion does not significantly alter the mixing behavior of the solution, 2) the size of the fuel plume is larger for the analysis which includes reacting flow, 3) the difference in size and shape of the plume between the reacting and non-reacting cases increases with downstream distance from injection.

Two algorithms commonly used for solving low-speed flow fields are evaluated using an unsteady turbulent flow formulation. The first algorithm is the method of artificial compressibility which solves the incompressible Navier-Stokes equations. The second is a preconditioned system for solving the compressible Navier-Stokes equations. Both algorithms have been implemented into GASP Version 4, which is the flow solver used in this investigation. Unsteady numerical simulations of 2-D flow over square cylinders are performed with comparisons made to experimental data. Formulations include both a single-cylinder and a three-cylinder configuration. Two turbulence models are also used in the computations, namely the Spalart-Allmaras model and the Wilcox k-ω (1998) model. The following output data was used for comparison: aerodynamic forces, mean pressure coefficient, Strouhal number, velocity magnitude and turbulence intensity. The main results can be summarized as follows; first, the predictions are more sensitive to the turbulence model choice than to the choice of algorithm. Second, using the Wilcox k-ω model, the preconditioned Roe algorithm produced more consistent results with experiment. Lastly, the Spalart-Allmaras model overall produced better results with both algorithms than the Wilcox k-ω model.