Application of Coal Textural Analysis to Predict Grinding Behaviour and Product Composition in Austr

The project set out to develop a technique for the prediction of grindability parameters such as optimum liberation energy, particle size distribution and size fraction composition for pulverised fuel in advance of coal processing.   To achieve this objective, the dependence of grindability on coal 'microtexture' must be demonstrated. Coal consists of microscopic components called macerals, which are inherited from biological plant precursors such as spores, branches and leaf cuticle, and defined in Australian Standard 2856-1986.

Coal microtexture relates to the size, and associations of these macerals. Intact plant parts of uniform composition, are called 'phyterals'. These phyterals have distinct size modes for a particular coal, and may form bands of uniform character within the coal, so their size is expected to influence the grindability of the band. The size of phyteral components can be assessed by etching coal to remove the surficial gel which masks the microtexture.

Plant parts which have been broken down physically due to transport, or chemically, are called detrital or attrital, and form bands containing complex associations of macerals, glued together by matrix detrovitrinite. The aim of this project is to assess the relative ease of grinding bands which are rich in phyterals ie telovitrinite in Bright coal bands, or rich in matrix ie trimacerites in some Dull coal bands.

These macroscopic bands in bituminous coal seams which concentrate either phyterals or matrix are called lithotypes, and are recognisable by their varying brightness.

Current breakage work has shown a dependence between coal banding, breakage energy and resulting size and daughter particle composition. It has also shown that the used of excessive energy results in over breakage and a shift in the natural size segregation of maceral components. It follows that the same textural relationships will also have an impact on coal grindability.

The importance of the fundamental controls, ie the texture and composition of botanically derived maceral constituents and mineral matter, has previously been overlooked in explaining grinding behaviour.

Conclusions

• The hypothesis that coal microtexture exerts an influence over grindability is supported by the project results. Coal lithotypes containing different proportions of phyterals to matrix, or different trimacerite:monomacerite ratios, behave differently in grinding. In particular, the phyterals, which are monomacerites of telovitrinite, semifusinite and fusinite, of inherent width about 150-300 µm, are quite easily liberated, but are then comminuted to finer particle size. The matrix trimacerites, which are composites of macerated plant tissue with mean sizes of 1-10 µ m, seem to remain intact as the energies to liberate individual macerals are not reached. 

• Therefore, with increased grinding energy, we do not see predictable shifts in size modes compatible with the release of finer scale inherent maceral components. The monomacerites are over-ground, largely concentrating in the -38 µ m size fraction, by the time the trimacerites have been comminuted to p.f. size (-75 µ m). This supports the idea that grinding coal lithotypes which are largely monomacerites, along with lithotypes that are mostly trimacerites, constitutes overgrinding of the large monomacerites, inefficiency and wasted expense. As a result of this difference in grinding behaviour between phyterals and matrix, one would expect iso-rank bright lithotypes, comprising macroscopic bands of concentrated phyteral components, to have better grindability than dull lithotypes, comprising concentrated matrix-rich components, when plotted as P 80 vs energy. 
• A baffling phenomenon was the observation that for 3 coals, below an energy of 30 kWh/t, the dull lithotype has a lower P 80 than the bright lithotype. The explanation for this behaviour is not yet clear, but may be related to the size/orientation of phyterals within the coal, or to the material stiffness. 
• When plotted as P 80 vs energy, the coal lithotypes each show an inflection point, after which increased energy makes little difference to comminution. This has also been observed with other ores. The comminution energy and P 80 value at which the inflection point occurs will be a function of the texture (J Campbell, pers. comm, Jan 1996), and the same would be expected of different coals or lithotypes. 
• Petrographic composition of the size fractions prepared for the Hardgrove Grindability test shows a concentration of monomaceral components in the -600 µ m fraction which is discarded, compared to a concentration of monomaceral components in the +600 µ m fraction used in the test. The tested size fraction is therefore not representative of the whole coal sample, so that the grindability of coals and lithotypes appears more uniform than is actually the case. As a result, HGI for bright and dull lithotypes from the same seam falls within the repeatability or reproducibility of the test. 
• A comparison of HGI and grinding required to produce p.f. shows that there is a reasonable trend of agreement between HGI and grinding energy (correlation coefficient = 0.764). However, the outliers have significant grindability differences from other lithotypes within the same seam, to which HGI is not sensitive. 
• At this point of investigation, it is recommended that information additional to HGI, on the grindability of plies within a coal seam to be used for p.f., could result in significant cost savings related to milling.