Architecture of a Classification System to Evaluate Fault Slip Risk in a Mining Environment

Thumbnail Image
Vatcher, Jessica Lauren
Classification , Numerical Stress Analysis , Structural Geology , Fault Slip , Geomechanics , Mining , Fault Behaviour
As the depth of mining increases, so does the risk of fault slip related rockbursts. Currently, there is no way to evaluate this risk, however the need for such a system is clear. Fault behaviour in mining environments is the result of a complex interaction between the mining system and the geological system. Although numerous models exist, the wide spectrum of fault behaviour cannot be fully explained. Additionally, these models are phenomenological, resulting in a disconnect between observable parameters and the models of faults. Fault behaviour is dependent upon the strength of the fault, the stresses acting along the fault, the boundary conditions and fault-system stiffness. Significant work exists in the field of earth science attempting to relate properties of the geological system to fault behaviour. In mining environments, these relationships become increasingly difficult to determine due to the time variable nature of mining activities. In order of importance, the following factors influence fault behaviour: excavations, tectonic history and in situ stress, fault system, fault zone geometry, pore pressure, fault zone slip surface and core, blasting, fault zone damage zone and wall rock and temperature. Numerical stress analysis models were created to evaluate the influence of excavations, tectonic history and in situ stress and the fault system on fault behaviour. Excavations were placed in various locations in a fault system. Results showed that there was no clear relationship between excavation location and fault behaviour; small perturbations in the initial state caused significantly different outcomes. The architectures of many classification and decision support systems were evaluated for purposes of a fault slip classification system. Due to the chaotic nature of fault behaviour and the time variable nature of the factors that influence fault slip, a typical classification system is not an appropriate architecture. Instead, it is recommended that a fault slip risk identification system be created, allowing for the incorporation of historical and live data to create a real time response. Artificial neural networks, numerical stress analysis, data from the identified important factors, and seismic data is recommended to form the basis of the fault slip risk identification system.
External DOI