Synthesis of biodiesel from the composite oil and process parameters optimization using artificial neural network

Abstract

This study aims to investigate the effect of catalyst utilization (wt%), methanol to composite oil molar ratio (mole) and reaction thermal state (°C) on the biodiesel yield percentage of transesterification of novel composite oil (equal volume % of Castor oil, Mahua oil and dairy waste scum oil). Artificial neural network (ANN) based on central composite design was implemented to investigate the influence of catalyst utilization (1–3 wt%), methanol to composite oil molar ratio (6–12 molar) and reaction thermal state (55–65 °C) as the independent variables. Levenberg-Marquardt algorithm was used to predict the responses, biodiesel yield %. The fairness of the ANN prediction of biodiesel yield % is tested by means of coefficients of determination of training phase, validation set, test set and combined training, validation and test. The results revealed that the optimum catalyst utilization, methanol to composite oil molar ratio and reaction thermal state were determined as 3 wt%, 9 molar, and 65 °C for the maximum biodiesel yield of 94 ± 1% from the composite oil. The ASTM D6751 standard is employed to assess the quality of biodiesel derived from a blend of oils. The findings of this study indicate that the newly developed mixed oil is a viable feedstock for the production of biodiesel.