Uncertainty and Sensitivity Analysis of Bleed Modeling in Shock/Turbulent Interactions

Abstract

Bleed is often used to alleviate the detrimental effects of shock-wave/boundary-layer interactions in high-speed vehicles. Simulations play a crucial role in successfully implementing this control methodology; however, high-fidelity approaches are prohibitively expensive for parametric design studies. Thus, simplified models are typically employed. The errors introduced by these models are difficult to quantify, especially under off-design conditions, which results in compounding uncertainties in realistic configurations. To address this knowledge gap, a three-dimensional Reynolds-averaged Navier–Stokes simulation with discrete bleed holes is compared to a two-dimensional simulation with a bleed boundary condition. Notable differences in boundary-layer shape and turbulence parameters are observed, which are then used to perform sensitivity analyses on an impinging shock response. The response quantity of interest is the shock-reflection angle, for which a variation of more than 3 deg is observed. The most sensitive input uncertainties are found to be the inflow shape factor and the eddy viscosity magnitude, which have not been directly explored in existing models. Therefore, this work assimilates the simulation results to account for the observed disagreement between existing experimental and numerical campaigns. Additionally, the identified sensitive parameters inform future model development efforts, thus aiding in improving the design capabilities of high-speed inlets.