Large Eddy And Unstructured Grid Methods For Heat Transfer Applications In Propulsion Systems

Large eddy simulation methods have been developed to compute compressible turbulent flows with heat transfer. The formulation employs preconditioning to permit efficient use in the low Mach number regime. Efficient algorithms have been developed for both vector and parallel processing computer systems. Simulation results have been obtained using a dynamic subgrid-scale model for a channel heated and cooled at temperature ratios as high as 3. Results were also obtained for a channel with uniform wall heat flux heating and cooling with wall to bulk temperature ratios as high as 1.5 for heating and 0.56 for cooling. Density rms fluctuations as high as 9% have been observed for very low Mach number flows with heat transfer. Mean and rms velocity and temperature data were observed to collapse toward incompressible results when plotted in semi-local coordinates. Preliminary results have also been obtained for flow in a rib-roughened channel. The large eddy simulation algorithms were extended to include the use of unstructured and zonal embedded grids. Particularly, the use of zonal embedded grids show much promise for extending the Reynolds number range for which large eddy simulation can be employed with existing computers.