Optimisation of Macroscale Functional Surfaces Through Numerical Investigation of Naturally Occurring Bedforms

Abstract

This thesis presents a detailed investigation into the potential of optimising the geometric profiles of macroscale grooves, to improve their drag reducing performance in internal laminar flow and spatially developing turbulent flow. The investigation explored whether the asymmetric profiles of naturally occurring sand ripples constitute an optimisation over the simple, symmetrical geometric profiles which have formed the typical focus for past investigations of macroscale grooves. In order to enable this analysis, the thesis developed and validated methodologies within the open source code OpenFOAM, which can overcome the bottlenecks associated with both modelling complex geometries in large-scale parametric studies, and implementing surface geometries into spatially developing turbulent flows. The first stage of the investigation developed a methodology for resolving laminar-turbulent transition in OpenFOAM using large-eddy simulation. To the authors knowledge, this work represents the first systematic validation and verification of resolved laminar-turbulent transition in OpenFOAM to investigate the combined effect of large-eddy simulation and controlled tripping. The results identify that a purely laminar boundary layer can be destabilised through imposing a period of psuedo-random, time-dependant fluctuations in the wall-normal velocity field at the wall. If the magnitude of these fluctuations match the maximum wall-normal velocity fluctuations in an equivalent boundary layer of equal thickness, then the initially period of two-dimensional instabilities is bypassed, and transition can be induced almost instantaneously downstream of the trip. Under these tripping conditions, the results expand the typical design criteria for large-eddy simulation spatial resolution, and show that typical design recommendations can sufficiently converge the flow resistance and shape factor by the start of the fully turbulent regime. Increasing this resolution by a factor of 2 achieve this convergence early on in the laminar-turbulent transitional regime. The second stage of the investigation involved an extensive parametric study of highly detailed sand ripple profiles within a periodic laminar channel flow. In all cases, the presence of both ripples and sinusoidal grooves had a negative impact on the flow resistance, typically due to a reduction in viscous forces being balanced out by the creation of a larger pressure force. The higher order details of the geometric profiles did not have a significant impact on the flow resistance, even when such details had a significant impact on promoting or delaying flow separation. The details of the geometric profile only became significant for three-dimensional ripples, when applied with a sufficient depth and and Reynolds number to manipulate the bulk flow field towards the centre of the channel, and direct high velocity flow from the centre towards the crests of the ripple profiles. The final stage of the investigation applied simplified sand ripple profiles into a wall-resolved spatially developing turbulent boundary layer, through the novel incorporation of a split-hexahedral mesh, through OpenFOAM's snappyHexMesh utility. Whilst ripples with a depth of $5\%$ of their wavelength had a negligible impact on flow resistance, deeper ripples ($15\%$) produced an increase in flow resistance which was independent of the growing ratio between boundary layer thickness and ripple depth. The local distribution of turbulent velocity fluctuations was consistent with known drag reducing phenomenon, with amplified spanwise velocity fluctuations over the shear stress spike approaching the crest, and amplified streamwise velocity fluctuations accompanying the free-shear region of flow separation downstream of the crest. It was in this free-shear region, that the streamwise resolution had the greatest impact on the accuracy of the local wall shear stress. The present approach confirms the capabilities of split-hexahedral meshes to efficiently balance the varying requirements of spatial resolution in near-wall and free-stream regions, regardless of the geometric surface profile.