Abstract
It has been shown that there is a large eddy simulation (LES) evolution, the ideal LES, that guarantees accurate single-time statistics and at the same time produces the most accurate short-time dynamics of the simulated turbulence. In optimal LES, models are constructed by formally approximating ideal LES using stochastic estimation. In this paper, optimal LES modeling is applied to the turbulent flow in a channel, using statistical data from a direct numerical simulation to form the stochastic estimates. Due to the data requirements, the modeling process pursued here does not directly yield generally applicable LES models; instead, the current study provides information on the required characteristics of subgrid models for wall-bounded turbulence. In the channel flow, and other wall-bounded flows, the strong inhomogeneity near the wall introduces several complications. There is a mean subgrid stress that must be represented, and there are subgrid contributions to turbulent transport. It is found that formulating the optimal LES models to reproduce a priori the important terms in the Reynolds stress transport equations was necessary to produce accurate LES. Models formulated this way were found to produce good predictions of mean velocity, wall shear stress and turbulent intensities. The results of this study indicate that for inhomogeneous flows, the subgrid model must represent processes other than transfer of energy to small scales. Specifically, it must represent subgrid transport, mean subgrid Reynolds stress and subgrid intercomponent transfer due to pressure strain.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call