Technical Market Support » General
A new classification method has been developed that uses imaging techniques for the microscopic characterisation of the individual grains present in ground coal samples. The grains are sorted into five grain classes, based on vitrinite abundances. This system is simpler and less subjective than the microlithotype system, which is used to manually classify coal grains.
Images were captured using a system developed for bulk analysis of coal under ACARP projects C6065 (O'Brien et al, 1998) and C8056, (Jenkins et al, 2001). This system produced reflectance-calibrated images, identified each grain within an image and processed them separately. The separation of grains into individual grain images provided an opportunity for more detailed grain-by- grain analyses.
In this project, reflectance-calibrated images of individual grains were processed to provide various textural data to enable the grain to be sorted into a class system. The grain constituents were identified by reflectance. The complexity of each grain was determined from the abundance of each constituent in the grain. The five-class system was based on the vitrinite content of each grain. The area, length and diameter of each grain were also determined. To validate the method, the bulk density was calculated for each sample from the total abundances of organic and inorganic constituents shown in the grain images and compared with laboratory results. The grain-by-grain density analysis also provided a means for estimating the sample washability characteristics.
The maceral abundance and relative density information extracted from image analysis of grains compared well with information obtained by manual petrographic analysis and other analytical methods. For the coal sample with a topsize of 300 ?m the imaging method provided a reasonable estimate of the size distribution for the coal sample but was unable to determine the size distribution of a coarse coal sample (topsize 4.75 mm) that was comprised of grains larger than the image size. The program performed best on samples that contained grains predominantly larger than 5 microns.