Computational fluid dynamics (CFD) modelling of dilute particulate flows

The coal-fired power stations of the Latrobe Valley provide the bulk of the energy demands for the state of Victoria. It is for this reason that over a number of years, research in coal technology has increased, particularly in the optimization of the efficiency in the coal combustion process. In order to ensure optimized efficiency, the pulverized coal must be supplied to the furnace at the ideal flow rate and distribution between the different burner jets. For a number of years this ideal coal distribution has been lacking, so this project was designed to aid in the rectification of this maldistribution problem. In order to provide a solution for the maldistribution, a better understanding of particulate motion and behaviour was required. Through the use of Computational Fluid Dynamics (CFD), a new particle coefficient of drag model was developed for use in dilute phase particulate flows. The model was developed after a comprehensive study of the influence of surrounding particle on the drag force of particles. Through this work it was found that the location and orientation of surrounding particles could reduce the drag force of a particle by up to 80%. It was also found that particles perpendicular to the flow could increase the drag force experienced by the particle due to a “nozzle effect” or squeezing of the flow between the particles. This effect was more noticeable at very small separation distances and lower Reynolds numbers. An experimental study of free-falling particles was conducted to investigate the influence of mass-flow rate and particle size on the terminal velocity of particle streams. The experimental data was used to validate the new drag model developed earlier in a range of CFD simulations. It was found that an increase in both particle loading and particle diameter resulted in an increase of terminal velocity compared to that of an isolated particle. The inclusion of the new drag force model improved the predicted particle velocities in all cases with the best results predicted for the medium sized particles (226μm) where an error of less than 2% was produced. An experimental study of particle streams impacting an inclined surface was undertaken to investigate the influence of known particle-wall and particle-particle collision models used in conjunction with the newly developed drag model for the Lagrangian Particle Tracking method. These models were implemented into CFX4-4 and results were compared with the experimental findings. It was shown that the CFD models failed to reproduce the post wall collision dispersion seen in the experimental work unless some form of particle-particle collision is included. The new drag force model developed earlier produced more accurate particle velocities pre and post collision compared with the standard CFD model. The simulations including the new drag force model with the known collision models produced more accurate predictions for both particle velocity and main particle streams trajectories than those of the standard CFD model. Experimentally the use of different impacting wall materials produced rebounding characteristics consistent with other data available in the literature. Finally the new drag force model in conjunction with the implemented collision models were used to predict the motion of particle in straight horizontal pipe flow and flow through a 90° bend and compared to experimental results found in literature. The inclusion of the drag force model produced more accurate particle velocities within the pipe flow simulations. The inclusion of the particle collision models resulted in less gravitational settling of the particles in the horizontal pipe flow more representative of the experimental work. The various model inclusions resulted in slightly more accurate CFD predictions, but the CFD failed to accurately predict the dense rope region and the subsequent rope breakup further downstream.