Pattern Recognition Analysis of Structure in Manipulated Channel Flow Simulations

Abstract

The possibility of applying Pattern Recognition Analysis (PRA) techniques, originally developed for analysing experimental multipoint hot-wire probe data, to Large Eddy Simulation output for fully-developed plane and manipulated channel flows is discussed. It is apparent that patterns can be detected in the Simulated streamwise velocity fluctuation fields which are similar to the ensemble-averaged patterns derived from PRA of experimental data for plane and manipulated boundary layers. This means that the detailed velocity vector and pressure field information provided by the Simulations can be used to examine the typical 3D structure of the large eddies producing the patterns. In practice it appears that these eddies have a similar vortex-loop structure to that inferred from the more limited experimental data for both plane wakes and boundary layers. However their scales are such that they appear more like squatter, wake ´horse shoe´ vortices than the thinner ´hairpin´ eddies found in boundary layers. Furthermore, just as the existence of horse-shoe vortices in the wake seems to be closely associated with stronger entrainment, the eddies in each half of the channel flow appear to be reinforced by cross-flow from the opposite wall region. When manipulators are introduced into the flow these not only influence the flow near the channel walls, but also serve to damp such cross-flow interactions. There would thus appear to be rather more similarity between internal and external flow manipulation than has previously been suspected; as is made clear by further comparison with plane and manipulated boundary layer Simulation results. This finding has important implications for the scaling of internal manipulator parameters, as well as attempts to predict their optimum values using turbulence models developed for external flow application. These points are discussed together with the possibility of using the Simulations to resolve outstanding questions regarding the manner in which internal flows recover from manipulation, and how more beneficial control of the flow structure might be achieved.Key wordsSimulationsPattern RecognitionManipulated Flow Structure