Large Eddy Simulations of High Speed Flows

Abstract

RANS models are required for the prediction of scalar fluctuations and turbulent transport in the high speed flow regime. These models will have application, for example, in missile exhaust plume signature analyses, scramjet combustors and other important areas. However, experimentally derived scalar fluctuation data needed to develop these models for the high speed flow regime is not readily available due to the inability of relevant experimental measurement techniques (e.g. hot wires) to cope with this flowfield environment. This issue poses significant difficulties for model development in this flow regime. Researchers have used different values for the turbulent Prandtl and Schmidt numbers but no consensus has been reached as to what these values have to be for high speed flows. To address this difficulty, a two part program has been initiated to fill the data gap and thus facilitate model development. Part I of this program involves the collection of LES data over a wide range of conditions. Part II involves the use of these data to evaluate and develop RANS tools to improve predictive capabilities. This paper presents results and findings of Part I of this program. Several flow fields of relevance to the problems mentioned above are studied. These include classical unit problems such as high and low Mach number shear layers, boundary layers and separated flows such as compression corner flows. In the process we are gradually extending the applicability of LES to more complex flows and at the same time enabling RANS model development by facilitating flow databases in the high speed flight regime. The findings of this study elucidate the effects of compressibility on the character of mean scalar profiles, variations in turbulent Prandtl number, and on scalar rms fluctuations.

High Performance Computational Modeling of Unsteady Surface Loads in Complex Weapons Bays

“…CRAFT CFD ® is a multi-block structured flow solver. The spatial numerical scheme implemented uses a highly accurate fifth-order upwindbiased reconstruction procedure [6][7][8] in conjunction with Roe's approximate Riemann solver. The fifth-order reconstruction scheme permits the resolution of waves using minimal number of points, while also minimizing the numerical dispersion and dissipation errors.…”