Dimpled channel flows

Abstract

Drag in pipelines is composed almost solely of skin friction drag. The most common technique to achieve drag reduction (DR) is by adding drag-reducing agents. However, in the aerospace industry, various impressive passive drag-reducing techniques have been suggested to reduce skin friction drag in the past decades. Among these techniques, dimpled surfaces form a relatively unexplored terrain. Research into reducing skin friction drag in the turbulent regime by using dimples has been performed in the aviation industry since the '80s. The excessive amount of skin friction drag in pipelines forms an intriguing challenge to break new grounds. Several researchers investigated the potential of DR of turbulent flows using dimpled channels. Even though most of these studies that found DR are disputed, studies published at the National University of Singapore (NUS) obtained positive results time and again. NUS's exact test setup was reconstructed at Delft University of Technology, which has not been done yet as far as the author of this report is aware. Identical pressure measurements were performed, yielding an absolute DR of ≈ 5%, which is slightly less than what was obtained at NUS (>7%). Flat plates return a DR between 8-15%, dimpled test plates returned at DR between 12-20%. The test plates were covered for 99.5% with diamond-shaped dimples of 100 mm long and 50 mm wide. In total, 29 pressure taps were used to determine the change in drag in an 8 m long channel of 20 mm in height. Tests were done at Reynolds numbers, based on half channel width and centerline velocity, between 6,000 and 40,000. Accurate results were perceived up to a Reynolds number of 21,000, likely due to test-setup limitations. Multiple verifications such as two-dimensionality of the ow, comparison of theoretical and experimental skin friction coefficients, and instantaneous pressure tests were used to allow for an objective analysis. The pressure measurements were also supported by 1D hotwire anemometry (HWA) tests and surface oil ow visualizations (SOFV). The majority of the investigated boundary layers were absent of anomalies. It should be mentioned that the viscous sublayer could not be captured, neither a quantitative momentum analysis was performed. However, a shift in the velocity profile, acquired at the same test location, for different test plates was observed. Furthermore, an increased velocity near the wall was observed for dimpled test plates. Surface oil ow measurements did show similar ow patterns to what was recorded at NUS. However, the actual behavior is not investigated through HWA. Hence, it cannot be confirmed nor denied if the change in drag is a consequence of these near-wall ow mechanisms. SOFV did not show irregular ow structures such as ow reversal. Finally, tests were performed with a correction for test volume increase caused by the dimples. After this correction, the drag over dimpled plates increased instead of reduced. Considering other studies on dimpled surfaces that also found an increase in drag, it is strongly believed that the positive effect of dimples in turbulent channel flows does not stem from skin friction drag but rather from the increase in channel volume. Although the precision of the results was relatively high, the accuracy of the wind tunnel was insufficient. Additional research is required to narrow down the 8-15% error that was obtained while testing at plates.