Files in this item



application/pdf9512293.pdf (10MB)Restricted to U of Illinois
(no description provided)PDF


Title:Stochastic estimation of near-wall closure in turbulence models
Author(s):Bagwell, Ted Glyn
Doctoral Committee Chair(s):Adrian, Ronald J.
Department / Program:Engineering, Mechanical
Engineering, Nuclear
Physics, Atmospheric Science
Discipline:Engineering, Mechanical
Engineering, Nuclear
Physics, Atmospheric Science
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Engineering, Mechanical
Engineering, Nuclear
Physics, Atmospheric Science
Abstract:Linear stochastic estimation (LSE), a best mean square estimator, is used to formulate boundary closure models for the numerical large eddy simulation (LES) of turbulence. The stochastic estimation method resolves coherent structure of the wall layer using empirical correlation functions, and provides an accurate estimation of the wall layer characteristics. As in standard large eddy simulations, a wall shear stress model is developed and applied to a direct numerical simulation (DNS) of channel flow at $Re\sb\tau$ = 180 and to large eddy simulations at $Re\sb\tau$ = 640. The a priori error estimates in DNS, performed at $Re\sb\tau$ = 180, show that the LSE model is more accurate than previous models. The direct numerical simulations with wall shear stress boundary conditions show that the method is accurate outside of the near-wall layer. The large eddy simulations, performed at $Re\sb\tau$ = 640, show that the wall shear stress model will close the LES equations, however the simulations also show that the method is sensitive to the Reynolds number extrapolation method employed upon the correlations. A pseudo-boundary condition is formulated for large eddy simulation, where the computational boundary is applied in the logarithmic layer. In this case, the streamwise and spanwise boundary shear stresses plus the wall normal velocity are estimated in terms of horizontal velocity events in the channel. The a priori error estimates in DNS at $Re\sb\tau$ = 180 show that the pseudo-boundary will be most accurate for the boundary plane located thirty wall units from the wall, with events that are 10 wall units above the boundary plane. It is also found that the wall normal velocity estimate was dominated by continuity effects. While both the direct numerical simulations and the large early simulations must be augmented by auxiliary information, the cases do show that a computation may be sustained without resolving the near wall region.
Issue Date:1994
Rights Information:Copyright 1994 Bagwell, Ted Glyn
Date Available in IDEALS:2011-05-07
Identifier in Online Catalog:AAI9512293
OCLC Identifier:(UMI)AAI9512293

This item appears in the following Collection(s)

Item Statistics