Gravity flow of rock in caving mines: Numerical modelling of isolated, interactive and non-ideal draw

A fundamental understanding of the flow of rock within block caves can assist in developing strategies for the optimization of recovery from caving mines. Predictive tools for caving flows are often based upon scale approximated physical models or small scale numerical studies. The simplified nature of these models often leads to results which resemble the well known theories developed for the flow of grains within silos. Marker trial data from caving operations suggest that the flow of rock within caves is significantly more turbulent in nature often leading to regimes not well characterised. Two reasons for this difference were hypothesised; the first being that models were not able to adequately capture the high stresses within block cave operations and the second being that models were not doing justice to the size and shape distribution of particles as observed at drawpoints. This thesis explores both these areas by the construction of two high resolution models for flow developed using the Discrete Element Method(DEM). A full scale, far field model accurately captures the correct stress regime and allows examination of variables including porosity, velocity, shear banding and stress whilst a highly detailed, block aggregate model was used to investigate the effects of particle shape, size distribution and secondary fragmentation on flow. ESyS-Particle, a parallel implementation of the DEM for use on supercomputers, was adopted to study the physics of flowing rock at the appropriate representative scales and in high resolution. Calibration of the modelling media was first conducted using shear cell experiments comparing simulated bulk friction angles with that of known values for rock. It was shown that bulk friction angle is strongly dependant on the ability of particles to interlock. However, a more complex particle shape does not necessarily lead to a higher bulk friction response, a result useful for future numerical studies on granular media. The calibrated particle assembly was then used to construct a numerical flow model that could be validated against previous modelling data. In the case of the narrow-homogeneous particle size distribution, it was found that the model compares well with known theories for ellipsoidal flow. The high resolution far field model allowed for detailed analysis of isolated draw. Measurements of porosity within movement zones showed variation in the form of compressional waves. These waves that bulk and recompact the material are what ultimately cause the growth of movement zones through the collapse of the top supporting arch. The majority of bulking within movement zones was shown to occur in the shear band area of flow surrounding a very dense, unbulked plug flow region. This has significant implications for the prediction of cave propagation and muckpile porosity. The model was then extended to study interacting draw zones under different scenarios of draw. Measurements of stress were able to show the intensity of stress concentrations over drawbells due to the action of neighbouring draw. The far field model, despite being in the correct stress regime, showed similar characteristic in behaviour to bin flow theory and was not able to explain the results of the marker trials. A cluster creation method using Voronoi tessellation was then introduced to enable simulation of complex particle size and shape distributions. This represents a shift in flow modelling toward a heterogeneous material type which is better representative of its caving counterpart. Results from this model were able to demonstrate features of flow including entrainment, non-symmetrical draw, dominant shear banding and particle size segregation. These features characterise a scenario of non-ideal draw where the heterogeneity in the initial packing causes flow behaviour outside of what would normally be expected from ellipsoidal flow and more in line with the observed marker trial data. Noted also is the impact of secondary breakage on flow, which acts as a catalyst for the formation of shear bands able to cause large scale block rotations. These results elucidate toward the existence of a range of flow regimes between that of the idealised ellipsoidal flow and the complex non-ideal draw scenarios. The evidence would suggest that the flow of rock in block caves is best described using aspects from both the ideal and non-ideal flow scenarios.