Cylinder Drag Reduction Using Discrete Surface Roughness

Abstract

This thesis investigates the influence of Cylindrical Distributed Roughness Elements (polka-dots) on cylinder flow, with a focus on potential applications in sports aerodynamics. The primary goals are twofold: to explore the mechanism behind tripping and to analyse how the dimensions (height, width, spacing) of polka-dots affect flow characteristics. The research employs an experimental approach, utilising balance measurements to quantify drag within the relevant Reynolds number range experienced by the limbs of speed-skaters. Additionally, Particle Image Velocimetry (PIV) measurements are conducted to examine the boundary layer and wake flow, revealing insights into how different polka-dot geometries impact flow characteristics.

11 polka-dot configurations were tested wherein the polka-dot height, diameter and spanwise (flow-normal) spacing was varied. Two PIV domains were imaged: the boundary layer flow before and after the polka-dot (covering an azimuthal range of about 40◦ of the circular profile), and the wake domain of the cylinder (about 2 diameters into the downstream flow). The boundary layer flow images were used to characterise the flow seen by the polka-dot array, and how it is affected by changes in the polka-dot geometry. The wake domain PIV imagery was used to examine the shape and dimensions of the cylinder wake.

Among the 11 tested polka-dot configurations, 10 effectively triggered drag reduction to varying extents within the relevant regime. The minimum drag coefficient was achieved by the configuration with the polka-dots of greatest diameter. It was also seen that increasing polka-dot height is likely to cause premature separation which is further exacerbated by a narrower polka-dot spacing. In general, results indicate that shorter and wider polka-dots cause transition at lower Reynolds numbers, and a greater reduction in drag occurs when transition takes place at higher Reynolds numbers. Polka-dots placed in closer proximity initiate flow tripping earlier, while wider spacing results in more substantial drag reduction. However, it is observed that the polka-dots, when spaced closer together, see a lower flow velocity for the same polka-dot height and may lead to premature separation.

In terms of the wake width, a high linear correlation is seen between the measured wake width and the measured coefficient of drag (r² ≈ 0.9). It is also seen that for drag coefficient values close to the minimum drag coefficient value, the wake width sees minimal change. The change in wake geometry is then seen as a change in the wake tapering (downstream decrease of the wake width) and the streamwise wake length. Therefore, a larger wake imaging domain in the streamwise direction is likely to allow for a more accurate correlation of the wake geometry and the drag coefficient.

While the study offers valuable insights, several recommendations are put forth for further research. Expanding the wake imaging domain is suggested to enhance correlations with the drag coefficient, and investigating spanwise flow variations would provide deeper insight into the tripping mechanism.