| Understanding the physics of turbulent flows over rough surfaces is of significant consequence in natural systems such as coastal and environmental flows, where surface roughness is more of a rule than an exception. In most of these natural systems, e.g. bedforms found on the ocean floor, the roughness elements are mostly random in nature, either in their distribution or geometry. As of now, we do not have an understanding of how the turbulent flow is altered due to the presence of random roughness. The focus of this work is to understand the differences between random and idealized roughness element for three-dimensional roughness geometries specific to bedforms on ocean floors.;A state of the art direct numerical simulation tool developed by Bhaganagar (2008) was used to simulate flow in a channel with wall covered with rough-surfaces. A numerical tool was developed to generate idealized and random rough surfaces. A set of metrics based on both the geometrical and statistical properties of roughness was derived to characterize both ideal and random roughness within the same framework. A parameter study was performed based on a critical parameter, namely, streamwise-spacing of the roughness elements/height of the roughness elements (lambda/h), traditionally used to characterize k- and d- type roughness.;A detailed study of the flow statistics such as mean, root mean square of velocity and vorticity fluctuations, and flow visualizations of 2-D planes was performed. The results have demonstrated (a) for the same (lambda/h ), the flow statistics and coherent structures are significantly different for random and idealized roughnesses, (b) for idealized roughness elements, lambda/h has a critical value of lambda/h around 7, where many flow parameters are maximized or minimized, and (c) for random roughness elements, the results are insensitive to lambda/h and are similar to 2D roughness. |
