Book Description

Hypersonic high enthalpy separated flows over two configurations, a rearward facing step and a compression corner were numerically investigated using computational fluid dynamics (CFD). Separated flow over the rearward facing step was studied at two enthalpies, 26 MJ/kg and 50 MJ/kg, corresponding to suborbital and superorbital speeds. The shock wave/boundary layer interaction and separation at the compression corner were studied at low and moderate enthalpies of 3 MJ/kg and 20 MJ/kg respectively. The numerical results have been compared to the available experimental data obtained in the X2 expansion tube facility which generates suborbital and superorbital speeds. The compression corner experimental data was obtained in the T3 free piston driven shock tunnel. In both these studies the flow medium was air. Two CFD codes, MB-CNS-2 and Eilmer-3, which primarily solve time-accurate Navier-Stokes equations were used to investigate the flow fields utilising AUSMDV flux splitting method. The perfect gas and chemical nonequilibrium calculations used MB-CNS-2, and the chemical plus thermal nonequilibrium, Eilmer-3. The chemical nonequilibrium behaviour of the gas was based on Gupta's kinetic scheme, whereas the thermo-chemical nonequilibrium was based on Park's two-temperature model. The computational flow establishments over attached and separated flows under suborbital and superorbital conditions suggested that the flow requires a longer time to establish than was available for testing in the X2 facility. Heat flux comparisons with experiments over the flat plate showed fair agreement and indicated that the boundary layer was essentially chemically and thermally frozen. In the close vicinity of the step, the presence of the Goldstein Singularity over the lip was observed and separation occurred on the face of the step. The separation moved further down the face with decrease in Reynolds number. A pressure minimum was shown to develop close to the location of separation as a result of over expansion followed by a recompression to the base pressure through a lip shock. Between separation and reattachment, the effects of chemical reactions were found to be negligible and the species were essentially frozen. A small region of thermal equilibrium prevailed close to the bottom corner of the step. Downstream of the step, the computed heat flux showed reasonable agreement with experiments. Computations with a rounded corner step configuration showed an increase in heat flux, skin friction and pressure between separation and reattachment with the increase in corner radius. In particular, the base pressure increased by nearly 30% for a fully rounded corner compared with a sharp corner in suborbital flow, while, in superorbital flow, it increased by a factor of 10. The increase in base pressure with increase in corner radius appears to be due, firstly, to the decrease in the strength of the expansion (resulting in a less steeper fall in pressure) and secondly, due to the formation of the recompression shock nearer to the base. The effects of chemistry were again found to be insignificant in the separated region. Computed flow visualisations showed that the shear layer at separation gradually became parallel to the plate as the radius increased. The compression corner results for the higher enthalpy flow showed that the separation, reattachment and recirculation regions were affected by chemical reactions. The heat flux and pressure comparisons with experiments for attached, incipiently separated and fully separated flows, suggested that CFD can reasonably predict the laminar shock boundary/layer interaction in the separated region at the corner. Significant differences were noted downstream over the ramp surface, particularly in regard to both locations and magnitudes of the peak heat flux and pressure and in the subsequent distributions of both heat flux and pressure. Under low enthalpy conditions, a similar situation was noted but chemical reactions were insignificant. The Eilmer-3 code was also used on the well known double cone configuration and the results were compared with the bench mark experimental data and previous CFD simulations based on steady solutions. The comparison of the present CFD data with these previous CFD and experimental data was fair at best highlighting not only the need for refined grids but also the difference between the time dependent and steady state approaches.