Measuring and Correlating CGA Data at Particle Topsize

Coal Grain Analysis (CGA) is gaining some traction as an alternative way to characterise a coal, and it is usually undertaken on a sample of 1 mm topsize, as per a conventional petrographics analysis. In prior work, samples of up to 4 mm topsize have been successfully analysed using CGA. This project sought to investigate whether CGA could be applied to particles up to 31.5 mm topsize, not for routine analyses, but to examine whether analyses undertaken on particles of large topsize would be reflected by the standard analysis.

It is not possible to routinely use the CGA method directly on topsize samples, because of the number of individual analyses that would be required to cover a sufficient number of large particles to make the analysis statistically significant.

It was thought that CGA could be undertaken on individual particles up to 31.5 mm in size, and the only limitation would be the time and cost to extend that to multiple particles. However, multiple issues arose upon trying to analyse individual large particles:

• Cracks within particles were almost impossible to distinguish automatically using the ParIS (image analysis) software, leading to false identification of mineral matter;
• Water emanated from the cracks after particle polishing was complete, leading to false component identification;
• Relief (raised and lowered components after polishing) was accentuated for large particle surfaces;
• Gouging was exacerbated by individual components being dragged from within a large particle during polishing. This rarely happens for a routine CGA since each (more liberated) individual particle is held in the block by the resin;
• ParIS processing times were extremely long (days rather than hours) during the particle segmentation and component identification validation/ correction process. This was thought to be due to the fact that the software holds information “live” for all pixels in all particles affected by the segmentation at the time. When particles are small, that is a small number of pixels, but for larger particles, the number of pixels is extreme.

In summary, despite concerted efforts to complete the project, these issues proved to be insurmountable, and the final recommendation from the project is that CGA should always be limited to a maximum particle size of 4 mm.