PREDICTION OF HIGHER HEATING VALUE OF BIOMASS: STATISTICAL AND EMPIRICAL MODELLING USING MULTIVARIATE REGRESSION AND RESPONSE SURFACE METHODOLOGY

This study developed experimental and statistical predictive models to determine the higher heating value (HHV) of lignocellulosic biomass using proximate, ultimate, and structural composition analysis data. Multiple linear regression (MLR) and response surface methodology (RSM) were utilized to establish a mathematical relationship between important compositional parameters and HHV. Results indicate that carbon, hydrogen, fixed carbon, and lignin contents positively influenced HHV, whereas increased oxygen content decreases energy density. The RSM model gives a better prediction accuracy as compared to the MLR model because it takes into consideration the interaction and nonlinear effects of biomass constituents. The lignin inclusion as a structural parameter contributes greatly to the strength of the model. The suggested method is fast and economical compared to bomb calorimetry and can be applied to screen biomass, select feedstock, as well as optimize combustion, gasification, and pyrolysis systems