This thesis is concerned with the development of a model to predict the dispersion of solid particles in a turbulent gaseous environm ent. The motion of particles in a turbulent flow is relevant to a number of areas of engineering including, for exam ple, combustion where the m otion of liquid fuel droplets in a combustion chamber is of interest, and atmospheric flows where one may be interested in calculating the dispersion of pollutants leaving a smokestack.

The challenge in this work is to properly account for the effect of the turbulence on the momentum of the particle:​ the fluctuating turbulent gas-phase velocities impose a random force on the particle hence changing the particle equation of m otion from an ordinary differential equation to a random or stochastic differential equation.

The approach taken here is to approximate the random force acting on the particle as a Gaussian white noise random process such that the particle equation of motion is treated as a stochastic differential equation with a white noise forcing function. By applying the theories of stochastic mathematics, inform ation on the particle velocities is obtained hence allowing for particle position and concentrations to be determined.

The validity of the model was examined by comparing model predictions to analytical or experimental results for particles released into a number of fundam ental flows including laminar, uniform flows of homogeneous isotropic and grid-generated turbulence, and round jets. In addition, a number of particle sizes were considered ranging from very light ‘fluid’ particles, which essentially follow the gas-phase turbulence, to heavy particles which have a limited response to the turbulence. Predictions from the model were found to compare favourably with analytical and experimental results for the flows considered. Further, the new model was compared to one which models the effect of the gas-phase turbulence on the particle concentration as a gradient diffusion process. From this analysis a clear understanding of the restrictions im plicit in the gradient diffusion model was obtained.