| The purpose of this study was to apply a new technique, automated reflectance microscopy, to the petrographic characterization of coals and to use the data in a quantitative model for the prediction of yields in coal liquefaction.; Channel samples of 104 coals were collected from three different physiographic provinces within the United States in order to include as much of the natural variation inherent to coals as possible. These provinces contained coals derived from differing parent materials, differing geologic ages and differing post-burial histories. All the coals were in the high volatile bituminous (ASTM) rank range.; Nine properties of each sample were measured automatically by Rapid Scan including vitrinite reflectance, vitrinite content, inertinite content, pyrite content, mean pyrite intercept length, range of pyrite intercept lengths, number of pyrite particles per unit scan length, mean coal intercept length and the range of coal intercept lengths. One other variable, conversion (or the yield of soluble hydrocarbons) was determined chemically in the laboratory from samples which had been hydrogenated in a 'tubing bomb' apparatus. Conversions ranged from 37.6% to 81.9% with a mean value of 63.2%.; The resulting data matrix was subjected to a series of statistical analyses. The univariate analysis described the frequency distribution for each variable. The bivariate analysis revealed the existence of interdependencies among variables and thus indicated that more than one property was necessary to predict yield in liquefaction effectively. Finally, the multivariate technique of principal components analysis identified six linearly independent sources of information within the data matrix. Three of these, vitrinite reflectance (coal rank), pyrite content and petrographic composition were related significantly to yield in liquefaction. The remaining three components represented the size distributions of coal and pyrite particles and were related to aspects of the sample preparation process but not to yield.; The final predictive equation derived through multiple regression was found to be:; YIELD = 3.00 S(,p) + .12 VIT - 42.00 R(,o) + 78.50; where S(,p) is the pyritic sulfur content (% vol.), VIT is the vitrinite content (% vol.) and R(,o) is the vitrinite reflectance (%) under oil immersion. Application of an automated system such as Rapid Scan may allow monitoring of coal feeds on a routine basis and further expand our knowledge about the roles of coal properties in liquefaction processes. |
