APPLICATION OF AUTOMATED IMAGE ANALYSIS TO THE STUDY OF MINERAL MATTER IN RAW AND PROCESSED COALS.

摘要

Automated Image Analysis (AIA) and Scanning Electron Microscopy (SEM) was developed and applied to the characteriza- tion of mineral matter in two series of processed coals. Fundamental factors for the application of image analysis to the characterization of minerals in coal which were addressed include development of chemistry definitions for classification of minerals in coal, sampling design, characterization of mineral matter mass distributions by size and type for both raw and processed coals, and AIA overestimation of pyritic sulfur.;A formula was developed for designing AIA analyses so that adequate sample area and particles may be analyzed to produce reliable and reproducible results. An example was also presented of using actual particle count to determine precision on a category- by-category basis.;AIA was applied to the characterization of mineral matter in samples of 200 mesh Illinois No. 6 (Illinois), Pittsburgh (West Virginia), Adaville No. 11 (Wyoming), and Dietz No. 1&2 (Montana) coals before and after float-sink cleaning, and to the characteriza- tion of samples of Illinois No. 6 and Pittsburgh No. 8 coals before and after cleaning with the TRW Gavimelt molten caustic (NaOH-KOH) process. AIA provided a method to identify statistically significant differences in mineral matter features among the coals studied, such as the distribution among mineral phases (e.g., predominant mineral phases being different from one coal to another), the difference in size distribution (i.e., especially after cleaning), and the ability to monitor the formation of new phases associated with chemical cleaning.;Factors leading to AIA overestimation of pyritic sulfur by as much as 50% were evaluated. Area inflation due to threshold setting was shown to cause overestimation of less than 10%. Preferential settling of pyrite particles was shown to be a minor effect (6%) for 200 mesh coals and can be avoided by proper sample preparation. Porosity in.;Two methods of developing a chemistry definition file were described. A priori class definition and autoclassification were discussed as complementary procedures for writing an initial definition file, and guidelines were suggested for evaluating and revising chemistry files.;large particles of pyrite was found to be the most significant reason for overestimation of pyrite.;*DOE report IS-T-1204. This work was performed under contract No. W-7405-Eng-82 with the U.S. Department of Energy.