Use of the seismic reflection method to optimize safety and extraction: a case study from a South African platinum mine

Abstract

Mining operations in the Bushveld Complex platinum mines have been mostly restricted to shallow depths (<1000 m) but are migrating to intermediate depths (1000 – 2250 m). This migration is met with complexities in the local geology, prevailing rock mass conditions, thickness and dip of the orebody, all of which influence the mining procedures and standards. Mining is likely to encounter zones of rock weakness, high stress and induced seismicity. This study aims to investigate the structural complexity of the remaining part of the lease area of a platinum mine. This is an area earmarked for future operations, and thus the impact and effect these complexities have on current mining operations is studied to understand what future mine planning and ore valuation should entail. Investigation was done in two phases: the comparison of conventional to seismic attribute interpretation, and the formulation of an interpretation workflow for the seismic attributes.

Initial conventional interpretation was carried out by picking the UG2 marker horizon throughout the 3D reflection seismic data and correlating results with borehole data and synthetic seismographs. The workflow that was developed and used in this study first involved data gridding using a constrained biharmonic operator. Then various horizon seismic attributes (surface stability index, dip angle, dip-azimuth and edge detection) were used to enhance the visibility and continuity on structures (e.g. lineaments and reef terracing) not readily visible on the horizon picked using conventional techniques. Subsequently, volumetric seismic attributes (data conditioning using the dominant frequency, structural smoothing vs graphic equalizer attributes, then edge detection using variance vs chaos attributes) were applied. The outputs from variance and chaos attributes were then used as inputs to the ant-tracking algorithm. Conventional interpretation yielded expected first order results, notably horizons dipping between 9º - 12º in the NE direction, as described in the literature, but could not discern more structural detail. The dip angle, dip-azimuth and edge detection seismic attributes immediately picked up linear features trending NW – SE and a few SW – NE, possibly fault zones. It also confirmed that the dip angle found using conventional interpretation techniques, ranged between 5º to > 10º, with the steeper dips associated with regions too noisy to pick and are possible zones of depression. The ant- tracking volume from the variance attribute was chosen over the chaos attribute due to better resolution. The ant-tracked volume enhanced the visibility of gently dipping structures, initially seen in the data but not well resolved. It also enhanced steeply dipping structures that were not visible to the conventionally-processed data. In the end, it was very clear that good results are dependent on the interpretation workflow used and thus great care should always be taken before and during the use of seismic attributes as undesired features, like noise, can also be enhanced.