| There are many challenges to modeling one- and two-way coupled high Reynolds number particle-laden flow using LES. As the particle mass loading is increased, a major challenge is accurately predicting turbulence modification by particles. To date, no LES computation has demonstrated the ability to predict turbulence modification by direct comparison to data. This thesis attempts to systematically evaluate and improve particle-laden LES (PLES) techniques through careful comparison to data acquired in the fully developed channel flow experiments of Paris and Eaton, 2001 and Benson and Eaton, 2003. The experiments were conducted in air with moderate loadings of spherical glass beads, and showed substantial turbulence modification with particle mass loadings on the order of 20% and greater.; The first step was to test the single-phase performance of the LES to insure that the numerical methods and grid resolution accurately predicted the flow in the absence of particles. Excellent agreement between the LES code, experiments, and DNS by Moser et al. validated the single-phase performance of the code. Next, the LES was modified to compute one-way coupled particle motion using a simplified form of the particle equation of motion. The basic PLES code gave reasonably good agreement with the smooth-wall particle velocity results from the Benson and Eaton, 2003 experiment. A model to simulate subgrid-scale dispersion was then added following the development by Oefelein, 1997. Using this model the PLES was able to closely match the Benson and Eaton, 2003 smooth-wall particle mean and RMS velocity measurements.; As a final step, the PLES was modified to compute two-way coupled particle-fluid interactions using a conventional force-coupling scheme. It was shown that the conventional model was not able to predict the level of turbulence attenuation measured by Paris and Eaton, 2001. When the mass loading was increased by a factor of 10 compared to the experimentally measured value, however, the PLES matched the turbulence attenuation very well. This indicates that the widely used point-force coupling scheme underestimates the effect of particles on turbulence. The precise reason for this still remains unclear. |
