Uncovering Optimal Attached Eddies in Wall-bounded Turbulence

Abstract: Townsend's attached eddy hypothesis decomposes the logarithmic layer of high Reynolds number turbulent boundary layers as a field of randomly distributed self-similar eddies that are assumed to be attached to the wall and obey a scaling relationship with the wall-distance. The attached eddy model has emerged as an elegant and successful theory to explain the physics, structure and statistics of the logarithmic layer in wall turbulence. Building on the statistical framework of Woodcock and Marusic (2015), the present work details quantitative results on the structure of the attached eddies and their impact on velocity moment predictions. An inverse problem is posed to infer the ideal eddy contribution function that yields precise first and second order velocity moments obtained from direct numerical simulations. This ideal function is further simplified and is used to guide the proposal of a hairpin-type prototypical eddy which is shown to yield reasonable predictions of the velocity moments. This hairpin-type structure is improved upon by a) solving a constrained optimization problem to infer an idealized eddy shape, and b) inferring the circulation distribution of a hairpin packet. The associated forward and inverse modeling procedure is general enough to serve as a foundation to model the flow beyond the log layer and the codes are open sourced to the community to aid further research.