Abstract:
Mining at great depth is associated with many principal stability and ground control problems. One of the challenges currently facing underground mines is changes in ground conditions leading to the fall of ground (FoG) hazards. Fall of ground refers to rock material dislodged from roof and sidewalls of an underground excavation usually unintentionally induced. This study focuses on FoG characterization. A FoG database has been compiled. In order to quantify the rock mass behavior around underground excavations and fully characterize the FoG events, statistical analysis, numerical methods, rock engineering system (RES) and artificial neural networks (ANN) have been used. The Bamangwato Concessions Limited (BCL), an underground mine located in Selibe-Phikwe, Botswana was used as case study where FoG events that occurred in various stopes and other mine openings were recorded. Overall, two aspects of the FoG characterization were investigated. In the first part of this thesis, the concept of ground behaviour index was introduced to predict the ground class associated with FoG hazard. To this end, ANN was used to determine the weights of input parameters involved and the RES was employed as a tool to quantify the non linear interactions of parameters via the interaction matrices and the hazard class evaluated. Meanwhile, in the second part of the thesis, numerical modelling was implemented to analyse the modes of rock failure around the underground openings which had led to FoG. Several models were established in Rocscience software package to simulate the structure-induced and stress-induced failure modes or a combination of both that had been observed in the field. In general, the obtained results were in agreement with the field results. It is concluded that this study has enhanced the understanding of ground conditions, fall of ground characteristics and provided ground control improvements in BCL mine.
