Experiments to generate multiple shock waves in an axisymmetric model at hypersonic speeds were conducted in a small reflected shock tunnel. Conical surfaces were used to generate shock waves inside a circular duct chosen to be representative of a scramjet combustor. These shock waves impinged on turbulent boundary layers to produce shock wave/boundary layer interactions (SWBLIs). In the process of observing this phenomenon, the commonly used empirical correlations of Korkegi were tested for accuracy, i.e. the combined pressure ratio across these shocks can be measured and compared to that predicted by these correlations. Korkegi correlates only with Mach number, and is independent of Reynolds number and on how the pressure is applied. A major contribution of this work is to examine how the details of the compression process effect separation. In this study, the history of applying the compression was varied. An analytical method was developed for theoretically estimating the onset of incipient separation using an integrated computation of the momentum flux contained in the boundary layer. By including the summed (negative) contribution of wall shear stress on the integrated momentum flux, the upstream history of the boundary layer was considered. The overall result has a form similar to the Korkegi correlations, plus an additional correction term relating to momentum loss through wall shear stress. The correction term was determined to be a second order effect, which explains why the Reynolds number independent Korkegi correlations work so well over such a large range of conditions. A hypersonic flow test condition conducive to the generation of high Reynolds number flows and turbulent boundary layer production was developed in a small reflected shock tunnel. The experimentally measured flow parameters were matched by numerical simulation using a number of in-house codes at The University of Queensland. This has allowed the unmeasured parameters which are numerically derived to be stated with greater confidence. An internal centre-body with a conical forebody was used to generate conditions of incipient separation. This provided benchmark data for comparison with subsequent experiments with multiple compressions. A semi-vertex angle of 15o was selected based on Large Eddy Simulation (LES) numerical results once the experimental and numerical static wall pressure and heat flux were matched. A two-cone experimental model, which provided for adjustment of the axial separation between the two shock systems, was tested at the same flow conditions as used in the single-cone experiments. A technique of incrementally moving the instrumentation (relative to the centre-body) and repeating the same condition to achieve high resolution in pressure and heat flux distributions with a limited number of transducers was successful. The results verified that it was possible to subject a hypersonic turbulent boundary layer to two quantified compression-expansion systems with an adjustable axial separation between them and capture the first reflected shock in a “shock trap” to remove it's influence from the second SWBLI. The data from this initial two-cone model provided non-separated pressure and heat flux data which was used as a reference to help interpret data from separated flows. The commercially available Reynolds Averaged Navier-Stokes (RANS) numerical code, CFD-Fastran, was used to help design an experimental model which produces boundary layer separation. Algebraic and two-equation turbulence models were applied to a modified two-cone model to show greater pressure rises which would produce boundary layer separation. A modified two-cone model was tested and demonstrated boundary layer separation. Three configurations with varying axial separation between SWBLIs were tested which all produced separation. The configuration that produced the largest pressure ratio and largest separation region at the second SWBLI may represent a geometry whereby the distance from the hollow cylinder inlet and the second cone may represent a critical value. The amount of viscous interaction, generated from the leading edge of the shock trap, and the proximity of the two interactions may be coupled to produce higher than expected values. It is postulated that the boundary layer momentum recovery for the configuration where the second SWBLI was furthest downstream (30 mm configuration), prevented severe separation from occurring. An in-house RANS code, elmer3, was used to simulate the flow of the modified two-cone model. An algebraic turbulence model was applied to this model and comparisons of experimentally measured static wall pressure and heat flux have given good agreement. The wall shear stress was investigated to provide further information concerning the position and size of flow reversal regions. The use of the numerical codes utilised in this study has reinforced their effectiveness for model design and comparison of experimental results.
Identifer | oai:union.ndltd.org:ADTP/279186 |
Creators | Andrew Dann |
Source Sets | Australiasian Digital Theses Program |
Detected Language | English |
Page generated in 0.0012 seconds