Home > Research > Large Eddy Simulation

Large Eddy Simulation

One promising “top-down” approach to predicting turbulent flows in a number of engineering applications simplifies the computing by separating the scales. Professor Charles Meneveau is studying this particular approach to modeling turbulence with a method known as Large Eddy Simulation (LES), in which the equations of motion are solved explicitly for all scales larger than some given threshold (the grid-scale). Motions smaller than these (the sub-grid scales) are parameterized by a set of models that depend on various simplifying assumptions about the small-scale dynamics. In contrast to other “top-down” modeling approaches, such as Reynolds averaging, LES does not rely on averaging all the turbulent eddies but only the smaller ones, thus making it capable of capturing much more accurately the dynamics taking place in turbulence.

This method, while elegant in principle, is inherently difficult because little is actually known about the physics of the flow at the small scales. Without good experimental data to test different sub-grid-scale model possibilities, their accuracy remains questionable. Prof. Meneveau uses carefully controlled wind-tunnel experiments to test the assumptions and models that determine how the small-scale physics is represented in LES.

In a recent experiment, Prof. Meneveau and postdoctoral scholar Hyung Suk Kang placed an electrically heated metal cylinder horizontally in the Corrsin Wind Tunnel. Downwind of the hot cylinder, they placed an array of probes that measured both velocity and temperature. As the turbulent eddies that formed in the wake of the cylinder became smaller and smaller, the flow lost its structure, and the velocity field became more and more random, or isotropic (about equal in all directions). But to their great interest, the statistical data they gathered indicated that the temperature field did not follow suit; rather, it retained some sense of the larger-scale spatial orientation even as it “cascaded” into smaller scales. This effect had not been captured correctly with the sub-grid scale models, which assumed that temperature was also isotropic at the sub-grid scale.

For an in-depth look at some applications of LES modeling to specific engineering problems, see Energy and the Environment, and Aerospace and Marine Systems areas.