Direct numerical simulation of rough wall pipe flows

© 2016 Dr. Leon Chan Zen HsienDirect Numerical Simulations (DNSs) were conducted in a turbulent rough wall pipe with three-dimensional sinusoidal roughness using a conforming grid. The wealth of information provided by these simulations is used to gain a better understanding of rough wall turbulent flows from both a practical and fundamental point of view. A comprehensive range of simulations were conducted where the wavelength and roughness semi-amplitude of the roughness elements are systematically investigated. Two roughness parameters were investigated, namely the average roughness height k_a+ and the effective slope ES. An empirical roughness function model was then developed based on these roughness parameters and is found to reasonably estimate the drag of a variety of rough surfaces even in the transitionally rough regime. The shape, inclination angle of the turbulent structures and the dominant energetic length scales in the flow was found to be affected by ES. In addition, ES can have a dominant influence on the turbulent flow in the fully rough regime and dictate the degree to which outer-layer similarity is observed. For all of the rough cases tested, Townsend's outer-layer similarity is observed when considering the mean and higher-order statistics. There is also good collapse in the contours of the streamwise premultiplied energy spectra of the rough cases with the smooth wall in the outer region of the flow. Despite the large blockage ratio of the largest roughness, good collapse is observed in the outer-layer due to the symmetry of an internal flow where the boundary layer thickness is fixed and the shape of the total stress profile is well-defined. Finally, a rough wall turbulent flow is simulated in the minimal-span channel. This proof of concept shows that accurate and direct determination of drag is obtainable at a fraction of the computational resources required for a full-span channel