Evaluation of models for the prediction of higher heating value of biomass based on proximate analysis

Biomass is a renewable and sustainable source of energy with little greenhouse gas emissions. The higher Heating value (HHV) of biomass is a significant parameter that is used in characterising fuel quality, class and type for energy application systems. Experimental determination of HHV is expensive, takes time and not always available. This brings the need for mathematical models for HHV prediction. In this research, proximate analysis and HHV of ten common biomass samples in Nigeria were determined. The biomass considered included rice husk, rice straw, corn cob, woodchips, groundnut shell, desert date, coconut shell, palm kernel, millet straw and sugarcane bagasse. Eight linear and five non-linear correlations with good performance from the literature were employed for predicting the biomass’ HHV from proximate analysis data. The performance of the models was tested using statistical indicators. Model M1 and M7 were the best among all the tested models with average absolute error (AAE), average bias error (ABE), and root mean square error (RSME) of 3.8389%, 2.5002% and 0.8780 MJ/kg; and 3.8918%, 2.2301% and 0.8701 MJ/kg respectively. Other models also correlated relatively well with the experimental HHV with low error, though, some are good for specific biomass only. This research identifies the best models that have high accuracy and can be used for the prediction of the higher heating value of biomass samples from proximate analysis.