This thesis investigates the suitability of the discrete element method (DEM) for modelling a real granular flow. The motivation for this work is to improve current design methods for transfer chutes. As there is little theoretical knowledge about the behaviour of bulk solid flow in three dimensions, transfer chutes are currently designed by a combination of rules-ofthumb and basic applied mechanics approach where the flow of a stream of particles is essentially treated as the motion of a single particle.
DEM is a time stepping algorithm which determines the contact forces on every particle at each time step and subsequently calculates the velocities, accelerations and displacements of each body. The drawback of this method is that it requires a lot of computer resources. But with ever increasing advancements in computing power, DEM is seen as a logical step to being able to model an entire flow inside a transfer chute accurately.
The scope of this work is to investigate and evaluate a current commercially available DEM software package called PFC3D. The investigations of PFC3D included validation of the software with a single particle problem, the effects of particle size and shape, the effects of bond strengths between particles and the effects of mass and stiffness.
The major goal of this work is to test the correlation of a simulated model to an actual physical flow. To test the correlation between the DEM flow model and reality, a simple study with rice is conducted. The physical tests involve rice flowing down two slides. The rice grains would start at the top of one slide and free fall onto another. These slides could be configured at any position and angle.
The input parameters such as particle density, friction coefficients and restitution coefficients were initially tested through crude means due to the available testing equipment. Specifically, the instrument used for friction property tests (a household spring balance) was not sensitive enough to detect the changes in friction with consolidation stress. A number of commercial laboratories are able supply accurate tests results of such nature, but unfortunately they are expensive and the budget for this project was not sufficient to accommodate the outsourcing of property tests.
PFC3D only models spherical particles, but it can ‘clump’ spheres together to form arbitrarily shaped super-particles. The rice was first modelled as a clump. The tests were then simulated with PFC3D and results were generated in the form of movies and data files.
Much work was involved in getting the software to model the flow accurately but eventually the correct flow result was achieved. The flow property of friction was manipulated through a trial and error process to get the final validated result. While this may sound like much guesswork was involved, the changes in friction are consistent with bulk solid flow theory, that is, friction varies with major consolidation stress.
It must be concluded that current DEM codes such as PFC3D still require much development and evolution if they are to become a practical engineering design tool. The drawbacks of PFC3D are that it is not user friendly and it requires a lengthy period of time on a relatively high performance PC to solve simple problems.