Design of supersonic inlets requires knowledge of the turbulent boundary-layer developing inside the device. When necessary, bleed is applied in the region of shock-wave/boundary-layer interaction to help stabilize the shock system and reduce the risk of separation. The present study is concerned with the discovery made in 1992 of the "bleed roughness" effect, which can cause an unexpected rapid increase in the boundary-layer growth, a decrease in near-wall streamwise velocity, and an increase in the tendency of the boundary-layer to separate.

It is shown that the term bleed roughness is misleading. The bleed roughness effect has been modeled in numerical simulations of the boundary-layer and Navier-Stokes equations through modification of both algebraic and transport-type turbulence models. Wall roughness modeling using an algebraic turbulence model yields only the qualitative trends of the experimental results because the suction is much stronger then the roughness effect. The use of an auxiliary transport equation for the eddy-viscosity by Paynter yields good results numerically but is inconsistent with some of the recent physical observation of the problem.

A low-subsonic experiment was completed and analyzed along with existing data to show that the bleed roughness effect is absent in subsonic flow but present in supersonic flow for certain geometries and conditions. It is hypothesized that an unsteady flow effect from the wall-boundary is the source of the phenomena. An unsteady numerical analysis was performed with a sinusoidal variation in the wall-boundary condition leading to a reduction in the near-wall time-averaged velocity across the whole layer. A supersonic experiment reveals that unsteady effects can be present in the flow field through a Helmholtz type resonance, but that they are not responsible for the bleed roughness. It is found that the "bleed roughness" also occurs in the absence of bleed. "Bleed roughness" affects the turbulence and thus the turbulent shear stress distribution across the whole boundary-layer. A new turbulence model is advanced based on modification of the eddy viscosity throughout the whole layer. Preliminary results are in agreement with the experiments, thus providing the first prediction capability for the "bleed roughness" effect.