Effects of Obstacle Length and Trailing-Edge Slope Angle on Flow Separation and Cavitation Inception

Surface-mounted protrusions, such as sensors, ribs, or roughness features, disrupt boundary layers, trigger separation, and lead to strong shear layers with localized pressure reductions, which may cause cavitation inception. The tendency of the flow to cavitate heavily depends on the nature of the separated flow, which in turn depends on the geometry of the protrusion. This study examines the impact of the geometry of simple, wall-mounted two-dimensional trapezoidal protrusions, specifically their streamwise length and trailing-edge slope, on flow separation, reattachment, and turbulence production. Experiments were conducted in a closed-loop water tunnel at a Reynolds number based on obstacle height, h, of 60000. Using high-resolution stereo particle image velocimetry (PIV), we examined flow over obstacles of constant height (10 mm), matching the boundary layer thickness, while varying length (L/h = 1 to 4) and trailing-edge slope angle (90, 45, and 30 degrees).

Short obstacles generate a single, elongated recirculation zone driven by return flow from the wake, which flows back on top of the model and directly interacts with the shear layer at the leading-edge separation region. As the length increases, the flow begins to reattach on the upper surface, forming two distinct recirculation zones: one above the model and one in its wake. The strongest turbulence amplification is observed in intermediate-length obstacles (L/h = 2 to 3), where the return flow from the wake interacts with the still partially separated flow above the model. Introducing sloped trailing edges increases the production of turbulent kinetic energy for short obstacles. However, on longer ones, the sloped geometry reduces turbulence and, in some cases, inhibits separation at the trailing edge altogether.

These results show how obstacle geometry modulates separation behavior, shear-layer interactions, and near-wall turbulence in a non-intuitive manner. In particular, we found that the effect of trailing-edge geometry strongly depends on obstacle length. The findings provide a benchmark for validating turbulence and cavitation inception models.