Micro-to macro-structural investigations of faulting in deep South African gold mines

Abstract

Deep and high stress South African gold and platinum mines (1000 m –4000 m) are important for sustainable economic development. These mines are critical for foreign currency earnings, creating employment, alleviation of poverty and reducing inequality. However, deep level mining operations are subject to seismicity, falls of ground (FOG) and rockbursts, which pose a risk to the safety of mineworkers, infrastructure and people living in mining regions. On the other hand, deep mines offer a unique and cost-effective opportunity to examine the physics and mechanics of earthquake processes at some seismogenic depths. A reliable scientific method has been developed in the last decade, to drill into ‘active faults’ at seismogenic depths. Analysis of the core has shown that it is possible to resolve fundamental seismological questions in earthquake and faulting processes. This study is part of the international collaboration under the International Continental Scientific Drilling Programme (ICDP) Drilling into Seismogenic zones project (DSeis). As the continuation of the Science And Technology Research Partnership for Sustainable development (SATREPS) project, a 5-year Japanese –South African collaborative project named "Observational studies in South African mines to mitigate seismic risks", drilled into seismogenic zones in several deep South African gold mines. The ICDP-DSeis project identified three active seismogenic zones:(a) the Moab Khotsong mine ML5.5 earthquake; (b) the Savuka mine M 3.5 earthquake; and, (c) the Cooke 4 mine M 1-2earthquakes. The PhD research focused on two seismogenic zones: Cooke 4 mine and Moab Khotsongmine. At Cooke 4 mine, the research examined seismicity associated with the extraction of a highly-stressed 1-km-deep shaft pillar intersected by large normal faults. At Moab Khotsong mine, the research focused on the largest mining-related earthquake ML5.5 that occurred below the mining levels. The ML5.5 fault had an uncommon left lateral strike-slip mechanism. Specific to this PhD thesis, the micro (microns –millimetres) to macro(metres –kilometres) mechanics and physics of mining-induced and mining-related earthquakes in deep and highly stress Witwatersrand Basin gold mines was examined. The study endeavoured to answer open seismological questions such as: (1) What influences fracture patterns and geometry, both in space and in time?; (2) What is the role of geology?; (3) How do earthquakes propagate along shear zones?; and (4) How do earthquakes scale (i.e., fracturing dependency on scale)? Answering these fundamental seismological questions will help to mitigate the risks posed by seismicity, falls of ground and rockbursts. Systematic and detailed comprehension of fracture patterns forming in the hanging wall and ahead of the stope face will assist miners recommend tailor-made support systems that will prevent the falls of ground in stopes. In addition, understanding rupture propagation and the influence of gouge will help mining practitioners appreciate the origin of unstable rupture propagation and factors controlling the severity of seismicity and rockbursts in deep mines. This will assist in the development of early warning systems and evacuation plans. The thesis is based on the total of five manuscripts: two manuscripts were published in peer-reviewed international journals, while three manuscripts have still to be submitted to journals. The first manuscript describes the development of fractures ahead of the stope in both space and time. An integrated fracture model obtained from underground mapping, petrographic analysis using thin sections, rock mechanics and high-resolution microseismic analysis at Cooke 4 mine is presented. Underground mapping and petrographic analysis reveal that the shaft pillar is largely comprised of quartzites, pebbly quartzites, argillaceous quartzites and conglomerates, which are characterised by different uniaxial compressive strengths (UCS), with quartzite being the strongest. In addition, underground mapping reveals that the shaft pillar is characterised by several minor to macro-scale discontinuity sets. Microseismic data revealed the formation of shear fractures ahead of the mining excavation (van Aswegen (2013) referred to as ‘Ortlepp shears'), and that the shear fracture 'turning-point' occurs in the soft strata (weak hanging wall lavas). The integration of these datasets enabled the development of the fracture model for different geotechnical zones (i.e., weak/soft lava hangingwall and quartzite/conglomerate footwall). This fracture model has significant safety implications. The second manuscript, examines how earthquakes propagate along these fractures occurring ahead of the stope, and what controls rupture propagation. The ICDP-DSeis team drilled into this seismogenic zone, and successfully recovered fragile ruptures, fault rocks and fault gouge material using a specialised triple-tube core barrel. Several dynamic friction-experiments were performed under room-dry conditions, some with gouge recovered from deep drilling, and others without fault gouge particles. These experiments were conducted from low to high slip velocities (~1.0 mm/s to 1200 mm/s). The experiments reveal a difference in the dynamic friction weakening behaviour (e.g., the 'no fault gouge' rock-on-rock experiments show a sudden reduction in fault strength at a significantly lower slip velocity than the specimens with the gouge material). The study shows that the presence of the fault gouge and its formation during rupture propagation controls fault rupture propagation processes along underground brittle shear fractures (i.e., the presence of the fault gouge causes a fault to temporarily strengthen at lower slip velocities). These results of friction experiments have implications for the severity of rockbursts. The third manuscript, examines the architecture, physics and mechanics of the unusual ML5.5 strike-slip earthquake that occurred below the Moab Khotsong mine. This is compared to the shear fractures with the common normal-slip displacement that occur ahead of the stope at Cooke 4 mine. In a serendipitous circumstance, the ML5.5 earthquake occurred within an area mapped by the 2-D reflection seismic data acquired by Anglo Gold Ashanti in 1992 for gold exploration. This legacy data was re-processed using modern techniques and re-interpreted to make clear the architecture of geological structures that hosted the ML5.5 earthquake. Similar to Cooke 4 mine, seismicity was recorded by underground seismic sensors near the source region. This enabled the DSeis team to spatially delineate the ML5.5 earthquake rupture, and drilled into this active seismogenic zone with its upper edge at a depth of 3.5 km. The DSeis team successfully drilled and recovered the fragile core, host rocks, rock breccia, sampled hypersaline brine, dolerite sills and a lamprophyre dyke. The fault rupture zone was characterised by the lamprophyre dyke, sandy fault gouge material and a core-loss zone. Interestingly, the re-processed and re-interpreted legacy 2-D reflection seismic data revealed that the geological structure that hosted the ML5.5 earthquake crosscut the Transvaal Supergroup strata, and potentially the base of the Karoo Basin. In addition, the legacy 2-D reflection seismic data revealed several dolerite sills characterised by high seismic amplitude reflections, while seismic attribute analysis was able to image vertical structures such as faults and/or dykes. The integration of the reflection seismic data and mine seismicity data, including the DSeis drill core (geological core logging) and geophysical borehole logging (gamma, density and velocity), allowed a full description to make clear a near-vertical structure, striking NNW-SSE. The relatively younger and highly-altered lamprophyre dyke intruded into this existing fault and older dolerite dykes. The fault was reactivated as a left-lateral strike-slip fault. The X-ray diffraction (XRD) analysis and measurements of the particle size distribution of the fault gouge material were conducted. The fault gouge consists predominantly of talc and biotite minerals (> 40%) and the particle size approach a minimum of ~ 10 μm.The abundance of talc and biotite minerals may control the friction weakening mechanism, acting as a lubricant during rupture propagation. The study conducted friction experiments on the Moab Khotsong sandy fault gouge sample at a slip velocity of ~100 mm/s and normal stress of 2 MPa. These experiments revealed that the peak friction coefficient of ~1.0 decreased exponentially to steady-state friction of ~0.66 over a slip weakening distance of ~9.1 m. The friction weakening behaviour and steady-state friction values are comparable with results obtained from normal faults occurring ahead of the mining stope at Cooke 4 mine at similar slip velocity and normal stress. The investigation shows that there may be three active processes that influence the ML5.5 earthquake seismogenic zone: (a) the architecture of the fault zone (e.g., intersection of lamprophyre dyke and dolerite sills); (b) the mechanical process induced by tectonic and/or mining-related stresses; and (c) the chemical processes caused by hypersaline fluids. This is different from Cooke 4 mine normal faults that are largely controlled by the mechanical processes induced by mining-related stress. The fourth manuscript, will be developed to a paper at a later stage. The Cooke 4 shear fractures induced by mining activities are compared to shear ruptures induced by triaxial laboratory experiments conducted up to 20 MPa confining stress, similar to underground conditions at Cooke 4 mine. The experiments and underground observations showed clear similarities - characterised by rough rupture surfaces and fault gouge material. The energy required to create a new fracture surface was computed and found that very little energy (0.15% and 0.56% of the total energy budget) was consumed during the creation of a new fracture. This was found to be in the same order of magnitude as the seismic efficiency. The fifth draft manuscript (see Appendix) discussed and developed new criteria that could be used to identify shear ruptures in drill core. The primary focus was on El Teniente mine and the Cooke 4 mine drill cores