Prediction of higher heating value of biochars using proximate analysis by artificial neural network

The biochars obtained from the pyrolysis of biomass at different conditions have the potential to be used as biofuels. Thus, as a critical fuel property, the higher heating value (HHV) of biochars must be determined to decide on their application area. However, oxygen bomb calorimeters that are employed for HHV determination are expensive. Also, analysis is time-consuming, needs specialists, and can suffer from experimental errors. Although some model equations are available for solid fuels (biomass, coal, etc.) to calculate HHV, biochar has different properties, and a new model is required. This study aims to form an artificial neural network (ANN) model in order to estimate HHV of biochars by using simple proximate analysis data of 129 different biochars. The experimental and the predicted model results showed good agreement that the ANN model presented the highest regression coefficient of 0.9651 and the lowest mean absolute deviation of 0.5569 among all models previously reported in the literature.