Abstract
M.Sc. (Geology)
The product yield problem during coal beneficiation refers to the failure of a processing plant
to produce expected product yield. Two well-known case studies from southern Africa
(Limpopo Coalfield, South Africa and Moatize Coalfield, Mozambique) are good examples
of where coal processing plants obtained lower than expected product yield results. This
product yield problem resulted in the devaluing of the project in the Moatize Coalfield, whilst
the plant in the Limpopo Coalfield was shut down. Both case studies demonstrated a lack of
sufficient research and experiments to enhance the understanding of the washability
characteristics of the coals. The product yield problem experienced in these coalfields
demonstrates the need for the application of advanced techniques, including simulations, to
provide better coal washability detail.
This study was aimed at achieving a better understanding of the washability characteristics of
coals obtained from three southern African coalfields by using different applied techniques.
The different techniques were applied in a novel situation to improve the beneficiation of
southern African coals. The applied techniques were digital imaging, petrography, image
analysis, and X-ray Computed Tomography (XCT). Washability simulation was performed
on the images generated for various particle size distributions (PSDs) using the newly
developed software.
Sectioned lump coals and cores were photographed using a handheld digital camera and
petrographic microscope. The block mounted polished samples were prepared on uncrushed
coal lumps/cores that were 5 cm × 5 cm in size. The preparation was done this way in order
to determine in situ sizes and distributions of the coal components. This enabled the
estimation of suitable crush sizes for the different coals in order to liberate the minerals and
macerals. XCT affirmed that 2D digital photographs of the polished block samples were
representative of the original 3D sample as very little lateral variation was observed in the
coal samples over a short distance.
Image analysis was performed on the digital and XCT images of the coal samples using
automated image recognition software. The software can characterize different coal
lithotypes based on colour and texture. Combined with the petrographic composition, this
method improved the level of detail observed in the hand specimens. The accuracy of this
method is directly proportional to the quality of the image produced. Images obtained from...