Experimental Fluid Mechanics
This thesis presents an extensive experimental data set that systematically examines the impact of surface morphology on wall-flows and could serve as benchmark data for validation of numerical simulations and other drag-prediction models. Experiments were conducted in the fully-rough regime on surfaces with large relative roughness height (h/delta = 0.1). The surfaces were generated by distributed LEGO™ bricks of uniform height, arranged in different configurations. Measurements were made with both floating-element drag-balance and high-resolution 2D and 3D particle image velocimetry on six configurations with different frontal solidity, lambda_F , at fixed plan solidity, lambda_P , and vice versa, for a total of twelve rough-wall cases. Results show, for the first time, the individual effect of frontal and plan solidities on various quantities in a controlled experiment. The bulk drag behaviour is shown to be different for variation of the solidities. This reaches a peak value for lambda_F = 0.21, while it monotonically decreases with increasing lambda_P in contrast to previous results obtained by examining the flow over cube roughness, where the effects of the solidities are coupled. The upper boundary of Townsend's similarity hypothesis is also investigated. While mean velocity profiles are found to conform to outer-layer similarity, the higher-order quantities show a lack of collapse for almost all cases considered. This suggests that Townsend's hypothesis, for these rough surfaces with a high relative roughness height (h/delta = 0.1) does not hold - this conflicts with previous results obtained on simplified cube roughness. The use of proper orthogonal decomposition to infer spatial similarity of flows over different wall morphologies is also demonstrated, where some universal characteristics of the turbulence across rough-walls are identified. Hairpin vortex packets inclination is confirmed to be insensitive to a change in the frontal and plan solidities and so are two-point velocity correlations in both streamwise and wall-normal directions. Moreover, POD modes are indistinguishable in both shape and size across the different roughness morphologies.
Figure: Filed patterns and associated flow regimes for different roughness configurations. Published in https://doi.org/10.1017/jfm.2015.552.
Marco Placidi and Bharath Ganapathisubramani
On the effect of surface morphology on wall turbulence