The mechanism of cave mining propagation still requires a better understanding to be attained outside the industry-accepted Duplancic conceptual model. While this model suggests a continuous damage profile to be followed when an orebody is undercut in cave mining operations, the research of Cumming-Potvin (2018) describes an extended conceptual model to cave propagation which highlights a different failure mechanism whereby discontinuous damage occurs through the advancing events of parallel fracturing termed `fracture banding'. In this dissertation, a physical modelling approach was adopted in an attempt to simulate the process of cave mining propagation at various stress states in order to observe the resulting failure mechanism. Four centrifuge tests were conducted utilising manufactured artificial rock material sample panels that were subjected to various ratios of horizontal to vertical stress. Before the applicability of this material to represent actual rock found in cave mines could be deemed adequate, a full characterisation of the properties of this artificial rock material was performed. The material testing included uniaxial compression strength tests, triaxial tests and Brazilian disc tests. These values were compared to typical corresponding parameters of various rock types in order to establish a suitable range of scale factors. Particle Image Velocimetry (PIV) was integrated into determining the critical extensional strain for the material which was used to establish a strain-based failure criteria for the artificial rock material using the model developed by Stacey (1981). Once the artificial rock material was manufactured to replicate the characteristics rock by a standard deemed acceptable; a scale factor range of 12 - 9 291 was achieved using an absolute critical extensional strain value of 0.014 %. The following set of conditions were achieved when conducting physical modelling: negligible horizontal confinement using sand with zero overburden pressure, minimal horizontal confinement with zero overburden pressure, maximum horizontal confinement with minimal overburden pressure, and lastly maximum horizontal confinement with maximum overburden pressure following the same proportion in lateral earth pressure (K ratio) as the previous test. Even though the results, with regard to the geometry of cave formation, in each of the tests were different; all four tests displayed an indication of `fracture banding'. In terms of the different geometries that had formed during cave progression, it was found that models with lower K ratios showed a higher development of the caving mechanism (at the same vertical stress), whilst models with higher K ratios suppresses this. Moreover, models conditioned with larger vertical stresses saw cave formation forming in a longer time period, but ultimately exhibited caves with a larger perimeter and area once the full undercut width had been reached. Total collapse occurred when the cave advanced vertically reaching the top surface of the sample through the formation of a `chimney'. On further investigation of mapping strains in models throughout common time-steps, the results showed that regions of high minor principal strains from PIV analysis of cave propagation correlated well with identified fractures. In all test cases, cracks propagated at the point of the material matching or exceeding the absolute critical extensional strain value in either new cracks opening via cave-back progression or reaching pre-existing fractures. In most cases, regions of these tensile strain values were observed to have been bisected by visual cracks in models which suggest that the perceived behaviour of parallel fracturing is tensile in nature.
Dissertation (MEng (Structural Engineering))--University of Pretoria, 2020.