University of Minnesota
Aerospace Engineering and Mechanics
Spring 2001 Seminar Series
A Radical Approach to Large Eddy Simulation of Turbulence
Large eddy simulation (LES is a very promising technique for the prediction
of turbulent fluid flows. However, there remains in LES, the problem of
modeling the unsimulated small scales. In this talk, LES subgrid modeling is
approached as a formal optimization problem. A rigorous result shows that the
best-possible deterministic subgrid model has several good features,
particularly it guarantees that the single-time statistics of the simulated LES
will be correct. Unfortunately, this ideal LES model is uncomputable, though it
can be approximated. The study of formal approximations to the ideal yield many
interesting results regarding the nature of subgrid models and LES. Among these
are that in many cases, the exact subgrid term appears to be largely
stochastic. Also, in strongly inhomogeneous flows, the subgrid term makes an
important contribution to the spatial transport of resolved scale kinetic
energy, in addition to dissipation, and this must be represented by the model.
Finally, when casting LES in the context of finite volume representations, the
resulting optimal LES evolution schemes look neither like standard finite
volume representations or like representation of common LES models.
Friday, February 23, 2001
**Refreshments served after the seminar in
107 Akerman Hall.
Disability accomodations provided upon request.
Contact Kristal Belisle, Senior