Drag reduction by means of dimpled surfaces in turbulent boundary layers

Direct force measurements and particle image velocimetry (PIV) were used to investigate the drag and flow structure caused by surfaces with patterns of shallow spherical dimples with rounded edges subject to turbulent boundary layers. Drag reduction of up to 4 % is found compared to a flat surface. The largest drag reduction was found at the highest tested Reynolds number of 40,000 (based on dimple diameter). A favorable trend promises further improvements at higher Reynolds numbers. PIV revealed the absence of significant separation inside the dimples but did show the existence of a converging/diverging flow in the upstream and downstream dimple half, respectively. This leads to the rejection of theories proposed by other authors concerning the mechanism responsible for drag reduction. Instead, a fundamental dependence on pattern orientation is observed. Furthermore, preliminary Reynolds-averaged Navier-Stokes (RANS) simulations have been compared with the PIV data. Although the large-scale mean flows show good agreement, the numerical simulation predicts no drag reduction. As the RANS approach is inherently incapable of resolving effects on the behavior of small-scale turbulence structure, the origin of drag reduction is attributed to effects on the small-scale turbulence, which is not resolved in the simulations. It is argued that dimples, when placed in well-designed patterns to create the necessary large-scale flow structure, lead to drag reduction by affecting the turbulent structures in the boundary layer, possibly in a way similar to spanwise oscillations of the wall.