Format Type: PDF
Read Online: 1259
Optimal LES modeling is a new approach to the development of subgrid models of turbulence. It has been found to produce accurate LES simulations when based on reliable statistical information. Now, the primary effort in optimal model development is the determination of this statistical information from theoretical considerations, with minimal empirical input. The validity of the theoretically determined statistics is being tested against experimental and DNS data. When small-scales are isotropic, Kolmogorov theory, the quasi-normal approximation and a dynamic procedure allow optimal models to be constructed with no empirical input. Such models have been found to perform well, though the dynamic procedure has not yet been tested in this context. Tests using channel flow DNS show that, except for a region very near the wall, the quasi-normal approximation is valid. Further, for the log-region, a representation for the anisotropy and inhomogeneity of the statistics is being developed. Thus, the above modeling approach can be adapted to near-wall turbulence, except for the thin viscous region. To handle this wall layer, a filtered boundary optimal LES model is being developed and tested.