Intrinsic Three-Dimensionality of Laminar Shock Wave/ Boundary Layer Interactions

摘要

This work has detailed the existence of a new class of shock wave/ boundary layerinteraction (SWBLI) behaviour that exhibits markedly different dynamic characteristicsto the previous models that assume 2D steady state laminar, transitional or turbulent flow.An intrinsically 3D unsteady laminar flow has been shown to establish within theseparation bubble and the process that governs this formation has been studied in detailusing high fidelity, unsteady, 3D computational fluid dynamics (CFD) simulations. Theselaminar effects have been shown to produce significant additional heating at thereattachment point of the primary recirculation vortex compared with 2D simulations andtheory. The magnitude of this additional heating has been shown to be twice that of a 2Dlaminar simulation and provides an alternative conclusion to arguments presented in theliterature regarding transition to turbulence of these SWBLI.The results detail that the formation of a secondary vortex beneath the primary vortextriggers the onset of 3D unsteady SWBLI flow. The streamline curvature that thesecondary vortex imparts on the primary vortex excites a centrifugal instability thatresults in longitudinal, counter-rotating vortices to form within the separation bubble.These destabilise the recirculation bubble and produce an unsteady 3D flow, whichdisturbs the separated shear layer. The shear layer reattachment line is found to become3D and unsteady due to the recirculation bubble unsteadiness. This adds additionalcomplexity to the typical 2D recompression process assumed for nominally 2D SWBLIflows. The 3D unsteady recompression process results in the reattachment heat fluxesincreasing by over 200% at some instances in time.The intrinsic 3D unsteady SWBLI behaviour is shown to occur at an identical Reynoldsnumber to when a secondary vortex arises within simple 2D CFD simulations. This resultoffers considerable scope for these effects to be predicted without the need forexperimental data and their effects approximated through judicious use oftransition/turbulence models.