Revisiting Correlations for Biomass Calorific Value: The Case of Extractive-rich Feedstocks

Accurate prediction of the calorific value of lignocellulosic biomass is fundamental for evaluating its energy potential and optimizing thermochemical conversion processes. Classical stoichiometric correlations based on elemental composition (C, H, and O) are widely used to estimate the higher heating value (HHV). Nevertheless, preliminary laboratory observations suggest that these conventional models may not adequately represent the behavior of biomass with high extractive content, due to the influence of lipidic and volatile compounds that modify the decomposition and energy release during torrefaction. This study evaluated and recalibrated literature-based HHV correlations by incorporating experimental data obtained from the torrefaction of pequi seeds with extractives (PS) and extractive-free pequi seeds (PSW). By combining elemental analysis, proximate analysis, and statistical modeling, the research refined the predictive relationship between biomass composition and the heating value, providing a more accurate framework for extractive-rich feedstocks. The original literature-based correlations showed limited accuracy for pequi seed biomass, particularly for samples with distinct extractive and ash contents. The proximate analysis-based model presented the poorest performance, with systematic HHV underestimation for PS and relative errors of 18.03 to 27.16%. After parameter modification, prediction errors decreased substantially, with RMSE values decreasing from 1.08 to 6.59 MJ kg.1 in the original models to 0.38 to 0.73 MJ kg.1 for PS and 0.09 to 0.20 MJ kg.1 for PSW, corresponding to RMSE reductions of 51.9 to 88.9% and 88.9 to 97.0%, respectively. These results indicate that feedstock-specific modified correlations better represent the calorimetric behavior of pequi seed biomass during torrefaction.