Thesis abstract


Multivariate statistical techniques have been applied to study interrelationships among 12 variables within a set of 277 coals representing whole-seam channel, column, and core samples obtained from each of the 6 coal provinces of the United States, and varying in rank from lignite through anthracite. The data are maintained in a computer­ized data base at The Pennsylvania State University Coal Research Sec­tion. The variables selected are components of the elemental analysis (carbon, oxygen, organic sulfur, hydrogen, and nitrogen), selected com­ponents of the proximate analysis (volatile matter and moisture), calo­rific value, reflectance of vitrinite, and the relative proportions of the 3 maceral groups (total vitrinite, inertinite, and liptinite group macerals).

Factor analyses performed on the entire data set and on subsets separated on the basis of rank, geographic location, and by cluster analysis indicated that rank is the most important factor in determining the amount of variation of each data set. The rank-dependent variables for the entire data set are carbon, reflectance, oxygen, volatile matter, calorific value, and moisture. The maceral groups account for the next greatest source of variation. Organic sulfur is independent of the first 2 factors and is the third most important source of variation.  Cluster analyses indicated that the most significant partitioning produces 4 groups which are differentiated primarily on the basis of rank, maceral composition, and organic sulfur content. Factor analyses of the individual groups provide insights into the coalification processes of these more homogeneous coal associations. Carbon, oxygen, reflectance, and calorific value vary together and are strongly affected by the processes of, coalification within each of the groups.  At about medium volatile bituminous rank, volatile matter and hydrogen also become sensitive rank indicators. Nitrogen enrichment in the low ranks is offset by nitrogen depletion in the high ranks. Factor analyses per­formed on 5 classifications of coals grouped by only the ASTM rank cri­terion yield similar results, but high-organic sulfur Interior Province coals of high volatile B and C bituminous ranks complicate the analyses of those rank groupings.

Analysis of variance indicated the following: Eastern Province coals grouped according to stratigraphic position differ significantly in only 4 variables (carbon, hydrogen, volatile matter, and nitrogen); the coals lower in the stratigraphic section (Pottsville Group) have attained a higher rank than coals higher in the section (Monongahela Group).      Interior Province coals of the high volatile bituminous rank range are significantly higher in moisture and lower in reflectance than coals of similar rank from the other provinces; within the medium vola­tile rank range, Interior Province coals have normal moisture contents but are still significantly lower in reflectance. For post­ Pennsylvanian coals grouped by age, Paleocene age coals are significant­ly lower in rank than Cretaceous and Eocene age coals, a result which is attributed to a geographic sampling bias. The coals from the Eastern, Interior, and Rocky Mountain Provinces for which more than 1 sample was obtained from individual mines have a variance within the mines which is inhomogeneous when measured by selected variables which were independent in the factor analysis, precluding pooling of the data to investigate sources of variation between both mines and provinces.