Coal washability analysis using X-ray tomographic images for different lithotypes. (1st December 2017)

• Record Type: Journal Article
• Authors: Liu, Hongping; Rodrigues, Sandra; Shi, Fengnian; Esterle, Joan; Manlapig, Emmanuel
• Is Part Of: Fuel. Volume 209(2017)
• Page Start: 162
• Page End: 171
• Publication Date: 2017-12-01
• Subjects: Coal washability -- CT scanned images -- 3D watershed method; Fuel -- Periodicals; Coal -- Periodicals; Coal; Fuel; Periodicals; 662.6
• DOI: 10.1016/j.fuel.2017.07.104
• Languages: English
• ISSNs: 0016-2361

Abstract

Highlights: A 3D watershed work flow has been developed for the coal washability analysis using X-ray CT images. The image analysis method can be applied to coal regardless of the lithotypes. Parameters used in the image binary process and the maker definition process should be different in different lithotypes. Abstract: Conventional coal washability analysis is achieved by conducting Float and Sink (FaS) test, which is sampling dependent, destructive and uses environmentally hazardous chemicals. With the improvement of modern 2D and 3D imaging technologies, image analysis methods of digital CT scanned images are being developed and trialled to predict coal washability in a more flexible, efficient and safe way. To support the validity of this approach, this study illustrates the fundamental differences in breakage and liberation behaviour of coal lithotypes at the same rank from a single seam. Nine lithotype samples were repeatedly CT scanned before and after a series of 2 m drops to emulate fragmentation during the drop shatter test. The 3D watershed segmentation method was applied to scans to monitor changes in fragment size distribution and then assign a density to each fragment to obtain coal washability. Comparison between the processed image analysis results and FaS tests conducted after 31 drops show a very good match both in particle size distribution and coal washability for all the lithotypes. Results also show that dull coal or stone illustrate better agreement between CT image analysis and FaS test due to their very low percentage of small particles which may lead to underestimation when using CT image analysis. Possible ways of improving small particles estimation would be to use an XY plane H-maxima method and low merging coefficient in the image analysis process.