Experimental examination of new separated turbulent flow validation test geometry

Abstract

Computational fluid dynamics (CFD) is a powerful tool for aeronautic design, however current simulations are challenged by turbulent separated flows due to their complexity. A new series of experiments are required to provide insights into the simulation challenges of this flow and to assist with the validation of future turbulence models. In this thesis, initial qualification experiments are conducted for a newly proposed separated flow geometry. This validation test used an elongated hill, “speed-bump”, shaped model, which was chosen because it allowed for the largest feasible Reynolds number, is longer in the spanwise direction making it more applicable to wing-like applications, is tapered at the sides to minimize side wall effects, and allows for adjustment of the flow separation via multiple parameters (i.e. incoming boundary layer, bump height, and splitter plate position). Tests were conducted over a range of Reynolds numbers, 58, 000 ≤ Re_h ≤ 301, 000, where h is the height of the bump. Wind tunnel qualification showed high uniformity and less than 0.2% freestream turbulence, ensuring the flow was suitable for the remainder of the study. The upstream boundary layer was examined to ensure a clean inflow onto the bump. Multiple tripping devices were tested and a strip of 240-grit sandpaper was chosen for future tests. Once the trip was chosen, the experimental momentum thickness Reynolds number was used to determine the corresponding simulation inflow length, which was set to 1.5L. China clay flow visualization and surface pressure measurements were used to examine the separated region. China clay revealed the presence of surface vortices surrounding a strongly separated three-dimensional flow. Static surface pressure measurements also indicate slight three-dimensionality at the centerline as well as insensitivity of the separation region at higher Reynolds number. However, at lower Reynolds numbers, separation appeared to be greatly reduced. A comparison of experimental and 2D RANS simulation data showed the magnitude of the simulated pressures were similar to the experimental data, however they exhibited an opposite Reynolds number trend in the separated and recovery regions. This suggests the proposed validation geometry is a suitable challenge for current RANS models of turbulent separated flows.