Discrete particle simulation of packing, fluidization and heat transfer of ellipsoids

Packing, fluidization and heat transfer of particles are involved in many industry processes. Many variables affect these processes, among which, particle size and shape are of most importance. Understanding its fundamentals is of paramount importance to the formulation of strategies for process development and control. To consider the effect of particle shape, ellipsoids are often used as they can represent particle shapes from flat to elongated. Discrete element method (DEM) is usually used to study the packing of fine particles. The combined approach of computational fluid dynamics (CFD) and DEM is favourable to study the gas fluidization of fine powders. Base on this approach, heat transfer behaviour in fluidized beds can then be considered. This thesis represents an effort in this area and four major components are included. The effects of particle size and shape on packing structure of fine ellipsoids are investigated. The results indicate that the porosity-aspect ratio curve shows a “W” shape for coarse particles, but the cusp at aspect ratio of 1.0 varies from convex to concave when particle size decreases. A correlation between the porosity, particle size (or force ratio) and aspect ratio is established. The results show that particles become less ordered when particle size becomes smaller. The contact and force network, force transmission and probability distribution are also investigated. With the introduction gas flow, the effects of particle size and shape on flow behaviour in fluidization of fine ellipsoids are investigated. At the macro-scale, “chain phenomenon”, as a special shape of agglomerates, exists in expanded and fluidized beds for fine prolate spheroids. In expanded beds, there is an obvious pressure drop fluctuation before pressure drop levels off at bed weight per unit area, and when the aspect ratio deviates from 1.0, the fluctuation amplitude becomes higher. The correlation between minimum fluidization velocity and particle size and aspect ratio has been established. At the micro-scale, it shows that fine particles show vortex flow for different particle shapes in fluidized beds. Flat or elongate particles tend to flow with small project area in the flow direction of fluid to reduce flow resistance. Focus is then given to the formation process of expanded bed with different particle shapes. Further, the effect of particle shape on heat transfer in packed and bubbling fluidized beds is examined. Conductive heat transfer models for ellipsoids are established. It indicates that in packed beds with stagnant fluid, ellipsoids exhibit larger effective thermal conductivity than spheres. In fluidized beds, ellipsoids have lower convective heat fluxes but higher conductive heat exchange rates than spheres. Prolate spheroids have larger convective heat transfer coefficient than those of spheres and oblate spheroids. Finally, to improve the computing efficiency and realize large scale simulation, the GPU-based DEM is developed. The performances of DEM on GPU and CPU are compared, and two different GPU parallel methods are compared. To realize industry scale simulation, the GPU-based DEM model considers arbitrary wall geometry and complex wall movements. Lastly, multiple GPUs technology is used for further acceleration and to deal with large granular system.