EXPERIMENTAL AND NUMERICAL STUDY OF A TWO STREAM, PLANAR, TURBULENT MIXING LAYER (VORTEX, EDDY, ENTRAINMENT)

In recent years the ideas of turbulence have changed substantially. Since the famous shadowgraphs of Brown and Roshko (1974), turbulent flows have not been viewed as completely random motion but as having a well defined two dimensional structure. Previous research has shown that these structures control the rate of entrainment and, therefore, have a significant impact on combusting flows.; The present study is a continuation of an investigation of turbulent reacting flows. Previous findings indicate that the basic fluid mechanical mechanisms are established in the density stratified layer and the effects of combustion are superimposed onto this case. This study compares a uniform density two stream mixing layer to a mixing layer with an initial density difference. Mean velocity and turbulence intensity profiles indicate that the density ratio is less important at high velocity ratios, and that the uniform density layer is less sensitive to velocity ratio than the layer with a density difference. Velocity spectra show peak frequencies at all velocities ratios in the uniform density layer but not at the higher velocity ratios in the density difference case.; Since the importance of the structures to the entrainment process is well-documented, the modeling efforts emphasized the accurate prediction of the large scale eddies. The random vortex technique is appropriate for this purpose because it solves the instantaneous Navier-Stokes equations and can, therefore, predict evolution of these structures which appear on short time scales.; The random vortex technique recasts the Navier-Stokes equations in terms of vorticity. The no-slip boundary condition is met by generating vortex elements along the solid boundaries. The elements are tracked each time step without the use of an Eulerian grid, so the computational time is spent only in the regions of vorticity concentration.; The numerical results indicate that the random vortex technique qualitatively reproduces the instantaneous behavior, but the mean and turbulence quantities deviate from the experimentally observed behavior.