Estimation of coal calorific value with petrography, ultimate analysis, moisture, R_max and ash using regression and artificial neural network methods

Relationships of ultimate, maceral and group maceral analysis, ash, moisture, and Rmax of wide range of Kentucky coal samples from calorific value of 4320 to 15050 (BTU/lb) (10 to 35 MJ/kg) with Gross Calorific Value (GCV) have been investigated by single and multivariable regression also ANN methods. Two sets of input: (a) ultimate analysis, ash and moisture/Rmax (b) macerals, ash and moisture/Rmax were used for prediction of GCV by regression and input set (c) group macerals, ash and moisture/Rmax was used for perdition using regression and ANN. In single-variable regressions, significant correlations were achieved between GCV with carbon and ash. For the other parameters of the input set (a), an order of significance of N> H>Rmax> moisture> Total sulfur> O was found according to the correlation coefficients of single-variable regressions. The multivariable regression studies have shown that both of input set of (a) and (b) in linear correlation can be used on GCV prediction with very good correlation. Individual macerals were a better predictor than group macerals. It is shown that coal maceral analysis, often used in studies of coal liquefaction, grindability, washability and coking properties, can be used with ash and moisture or Rmax as reliable input for prediction of gross calorific value of coal.