Dense medium cyclone (DMC) is a high-tonnage device that is widely used to upgrade run-of-mine coal in the coal industry. It is known that the complicated multiphase flow in the DMC is technically difficult and expensive to experimentally investigate. In recent years, with the development of computational technology, various mathematical models based on the flow fundamentals such as computational fluid dynamics (CFD) and its combination with discrete element method (DEM) have been shown to be effective in overcoming this difficulty. In practice, the material properties fed in the DMC such as coal particle density and size distribution and their interaction are important variables affecting the DMC performance. Particularly, many problems frequently confronted in the operation of DMCs (e.g., vortex finder/spigot overloading, surging and system instability) are basically related to these variables. Although many studies have been done to help understand the underlying mechanisms, no systematic study has been made to examine the effects of these variables on DMC performance. In this work, the effects of particle density and size distribution and their interaction are systematically studied using a previously developed CFD-DEM model. In particular, Johnson s SB function, which is able to describe a wide range of distribution, is employed to represent the particle density and/or size distribution. The simulation results are analysed in terms of medium and particle flow patterns, particle-fluid, particle-particle and particle-wall interaction forces to elucidate the mechanisms. Based on the newly obtained CFD-DEM results in this thesis, a previously developed PC-based DMC model, which can readily run on a personal computer (PC), to predict the DMC performance as a function of a range of variables in an easy, fast and cost-effective way, is further developed and extended. Finally, the extended PC-based model is used to optimize the DMC designs for coal preparation through representative examples, in comparison with several typical designs in the industry. In addition, some new rules for DMC scale-up are proposed to meet industrial needs. It is shown that the developed PC-based model can indeed offer a cost-effective way to design, operate and optimize DMC process under different conditions.