Abstract
The majority of processes in the chemical and allied industries involve the storage and conveyancing of granular material, the physics of which is still not particularly well understood. Whilst some non-invasive techniques have been developed, much experimental work unfortunately interferes with the fields being investigated. For this reason and in conjunction with increasing computing power, there has been an increase in simulation based studies. Granular dynamics simulations, being based upon inter-particle interaction laws, give the potential to investigate assemblies at the "micro-level" and have been successful in modelling process conditions in a number of granular flow situations. To date, most analyses of these simulations are essentially static in nature involving "time snapshots". However, in a granular dynamics simulation there is a wealth of data available on a time referenced basis which has the potential to allow a quantitative analysis of the dynamics of assembly evolution. This dissertation describes the development and application of a toolkit for postsimulation analysis. However, the utilities within the toolkit would be equally applicable to large experimental data sets should such data sets exist. The application of the toolset focuses largely on the dynamics of heap evolution in both 2D and 3D with some supportive 3D work on hopper discharge. A major part of the work involves the application of time series techniques (including the wavelet transform) in the context of variable coupling during avalanching. Segregation by self-diffusion receives particular attention and a new mechanism is proposed by which segregation by particle size takes place in the boundary layer of a low impact feed heap displaying a clear velocity gradient during discrete avalanching. Periodic lateral surging is shown to enforce mixing for a high impact feed, a phenomenon which appears to switch off below a certain feed impact. Segregation by self-diffusion is also shown to take place with the conical section of a 3D discharging hopper (in the presence of a velocity gradient) which is in agreement with experimental studies. In the context of the discrete defining events which characterise the evolution of a granular assembly, the discrete wavelet transform is shown to be capable of identifying "time constants" in the absence of real periodicity. A flexible colour coded display capability (for both 2D and 3D) is shown to be helpful in investigating the dynamics in conjunction with the time series methods developed.