LINEAR REGRESSION MODELING OF GROSS CALORIFIC VALUE BASED ON TOTAL MOISTURE AND ASH CONTENT IN COAL BLENDS

Taufik Arief
A. M. Jannah
B. Cahyaningsih
Aisyah Minzikrina M.Rus

Multidiciplinary Output Research for Actual and International Issue (Morfai Journal)
Vol. 6 No. 2 (2026): Multidiciplinary Output Research For Actual and International Issue

Coal blending in stockpile management is widely implemented to meet market quality specifications; however, variations in coal quality parameters may significantly influence the gross calorific value (GCV). This study develops a linear regression-based predictive model to quantify the decline in GCV as a function of total moisture (TM) and ash content (AC) in multi-brand coal blending. A dataset derived from three mine-brand coals (MT-49, BB-51, and BTB-47) was used as a case study. GCV (ar) was treated as the dependent variable, while TM (ar) and AC (ar) were considered independent variables. The regression results reveal a strong negative linear relationship between TM and GCV, with a coefficient of determination (R²) of 0.6303, indicating that 63.03% of GCV variability is explained by changes in total moisture. In contrast, ash content shows a weaker relationship with GCV, with an R² of 0.1974 (19.74%). Quantitatively, an increase in TM by 1%, 2%, and 3% reduces GCV by 104.82, 209.64, and 314.46 kcal/kg (ar), respectively. Similarly, an increase in AC by 1%, 2%, and 3% decreases GCV by 94.60, 189.20, and 283.80 kcal/kg (ar), respectively. Blending simulations representing worst-case, moderate-case, and best-case scenarios were performed using the developed regression model to evaluate their impact on final calorific value (GCV). The moderate-case and best-case scenarios achieved the target specifications, confirming that moisture control plays a critical role in maintaining calorific value during coal stock blending. These findings demonstrate that a simple regression-based approach can serve as an effective decision-support tool for coal quality control and operational blending optimization.