Geology and Geophysics

A single float sink test destroys the sample and gives one deterministic answer, so it cannot show how washability changes with particle size distribution. Simulating washability from petrographic images and tomography allows many size distributions to be tested statistically from one sample. Running simulation alongside float sink work during exploration lets the prediction error of each be compared.

Existing thermal imagery of the Witbank coalfield was traced to its holders and assessed for format, quality, resolution and coverage, then brought into one geographic information system with supporting metadata. Satellite thermal bands give a regional picture only; detailed mapping of coal fires or acid drainage discharge needs high resolution aerial surveys.

A three dimensional model of the northern Witbank and Highveld coalfield was built from more than 1100 borehole logs, with strata sorted into 13 major groupings and surface topography taken from satellite elevation data. It runs on open source visualisation software rather than a commercial package. Delays limited coverage to about a quarter of the coalfield.

Microgravity, resistivity based methods including self potential, and time domain electromagnetics are identified as able to detect and delineate shallow old workings, with controlled source audio magnetotellurics useful below about 30 metres. How well each performs depends on target depth, the material filling the void and the overburden, and has never been quantified. Field trials were planned to close that gap.

Micro-gravity, electrical resistivity and time domain electromagnetics located the boundary between old workings and virgin ground, but no ground technique tested could resolve individual bords and pillars at the depths typical of the Witbank and Highveld coalfields. Airborne thermal infrared surveys worked well as a low cost regional reconnaissance tool for shallow sinkholes, fires and possible pollution plumes.

A rating and ranking system was applied to South African coalfields and, separately, to the coal bearing countries elsewhere in the Southern African Development Community. The Waterberg ranked highest in South Africa and is the coalfield most worth detailed investigation as Witbank output declines. Botswana ranked highest among the other member states on contained resources and exploitation potential.

Tonnage estimates for the remaining low phosphorus resource could not be produced, because the phosphorus analyses supplied by mining companies did not state what had been measured. Distribution trends were still mapped. Coal at or below 0.010 per cent phosphorus is widespread in the Witbank Coalfield, restricted to the far northwestern Highveld, and follows a northwest to southeast band in eastern Ermelo.

Electrical and electromagnetic ground surveys were run over a Waterberg profile with borehole logs and three dimensional seismic data as control, to test whether cheaper methods can match seismic reflection for locating coal and faults. Resistivity tomography at two electrode spacings, frequency and time domain electromagnetics, controlled source audio magnetotellurics and magnetics were all acquired. Results are preliminary.

Geotechnical factors affecting secondary extraction were identified from interviews with collieries already re-mining pillars and with people experienced in stooping, then quantified and ranked against the different aspects of the work. No single factor governs the outcome. Factors interact, so choosing a method and working safely in old workings depends on managing them together across disciplines.

Shallow coal suitable for small scale opencast working was mapped in the northern Witbank and Highveld coalfields, using a limit of 30 m from surface to the top of the first seam. Most outcrop areas examined proved already mined out, and depth to the top seam increases southwards, though some ground remains available. Obtaining resource data was the main obstacle.

Sub-seams within the No 2 Seam were characterised at a single test site using lithology, seam thickness, proximate analysis, petrography and geophysical logs, and a paleoenvironmental model was built from them. Wireline logs separated coal from interburden but could not resolve the individual zones, because the physical differences are too subtle. Better sedimentological models gave better reserve and resource estimates.

Government and industry aeromagnetic data over the Witbank Coalfield were merged into one set of maps and GIS compatible files, and roughly 1200 dyke traces were interpreted from them. Purchase of a 200 m line spacing government survey removed the need for infill flying. One participating company valued advance knowledge of a single underground dyke at about R15 million.

Dolerites in the study area differ in mineral content, geometry, thickness, grain size and degree of alteration. Plagioclase textures indicate fractionation at different depths in the crust for the Witbank and Sasolburg intrusions, and the olivine content points to separate magma sources for two of the sills. Fine crystalline dolerite, including chilled margins and bifurcations, alters most readily.

Testing of 28 coal and coal-shale samples showed the coals to be more liable to spontaneous combustion than the shales, though reactive coal-shales still self-heat and pyrite in them speeds the process. Higher carbon, moisture, hydrogen, volatile matter and nitrogen with lower ash all raised proneness to self-heating, so proximate and ultimate analysis can serve as a screening tool.