MODELING PARTICLE DISPERSION IN A TURBULENT, MULTIPHASE MIXING LAYER

This paper describes the application of a Stochastic-Separated-Flow (S SF) algorithm for the numerical modeling of turbulent dispersion of gl ass particles injected in a two-dimensional, incompressible mixing lay er. Although it is recognized that unsteady, large eddy structures pla y an important role in the determination of the vorticity field in tur bulent shear layers, the computations reported here made use of time-a veraged turbulence models for the closure of the gas field because mos t often these practical turbulence models are used in industrial appli cations to model the transport of heavy particles. The comparison of v arious techniques used to sample fluctuating properties from the gaseo us held is discussed. Two distinct turbulence models were used to esti mate turbulence quantities of the continuous phase. These models were the traditional eddy viscosity (k-epsilon) model and a Multiple-Time-S cale (MTS) model, which allows for the partition of the turbulence kin etic energy spectrum into two distinct time scale ranges. In addition a set of post-processor algebraic relations was considered to study th e effect of turbulence anisotropy on the particle behavior. These alge braic relations were used to circumvent the effects of the isotropic a ssumption embedded in the formulation of one-point closure turbulence models. It was found that using second-order algebraic relations, or u sing more simplistic Boussinesq-hypothesis-based cross-correlations, h as minor effects on the determination of the Eulerian time-averaged pa rticle properties. However, the cumulative effects of the gas fluctuat ing velocities on the dispersion predictions were significant for the mixing layer studied here. Also, it was observed that the MTS model di d not give better gas phase predictions compared to the standard k-eps ilon model. (C) 1998 Elsevier Science B.V. All rights reserved.