Abstract Proposal Number: CTS-9803385 Principal Investigator: Meneveau The focus of this study is on the use of Large Eddy Simulation (LES) in the numerical prediction of turbulent scalar transport, combustion, compressible flows, etc.. For such applications it becomes crucial to accurately represent not only the momentum transport caused by small scales, but also the transport of scalar quantities, such as heat or species concentration. It has recently become clear that small scale isotropy and universality is a more questionable assumption for a transported scalar field than for the velocity field. Large-scale motions maintain a significant effect on the small-scale structure of the scalar field over surprisingly large length-scale separations. One of the main objectives of this research is to study the implications of these phenomena on SGS fluxes, dissipation of scalar variance, etc. in the context of LES and subgrid modeling for scalar transport. To genuinely capture effects of scale separation, high Peclet number flows must be examined. Experiments will be in a high Reynolds and Peclet number turbulent wake behind a heated cylinder. The data will be collected by means of hot and cold-wire arrays which will be developed specifically for the purpose of studying variables of interest to LES (such as subgrid fluxes, resolved strain rates, etc..). The probe arrays will allow us to analyze the data by means of two and three-dimensional filtering, as opposed to only one-dimensional filtering that is now possible with single probes. Since the SGS stress and heat flux are highly fluctuating stochastic variables, comparison between models and real SGS variables (such as SGS stresses, scalar flux, and dissipation of kinetic energy) must be statistical in nature. In the past, comparisons have been based on either global averaged quantities, or on point-wise comparisons. More refined analysis techniques based on conditional averaging will be developed that will allow a comparis on between real and model stresses in significantly more detail than is possible with global averages, but in a statistically more meaningful way than by comparing individual realizations. This research is expected to significantly contribute to the foundation of LES and associated SGS modeling, based on real data at high Reynolds numbers. 2 1
