Prediction of Calorific Value of Coal by Multilinear Regression and Analysis of Variance

Abstract

Abstract The higher heating value (HHV) of 84 coal samples including hard coals, lignites, and anthracites from Russia, Colombia, South Africa, Turkey, and Ukrania was predicted by multilinear regression (MLR) method based on proximate and ultimate analysis data. The prediction accuracy of the correlation equations was tested by Analysis of variance method. The significance of the predictive parameters was studied considering R2, adj. R2, standard error, F-values, and p-values. Although relationships between HHV and any of the single parameters were almost irregular, MLR provided a reasonable correlation. It was also found out that ultimate analysis parameters (C, H, and N) played a more significant role than the proximate analysis parameters (fixed carbon (FC), volatile matter (VM), and ash) in predicting the HHV. Particularly, FC content was seen inefficient parameter when elemental C content existed in the regression equation. The elimination of proximate analysis parameters from the equation made the elemental C content the most dominant parameter with by-far very low p-values. For hardcoals, adj. R2 of the equation with three parameters (HHV = 87.801(C) + 132.207(H) − 77.929(S)) was slightly higher than that of HHV = 11.421(Ash) + 22.135(VM) + 19.154(FC) + 70.764(C) + 7.552(H) − 53.782(S).