Optimized biodiesel synthesis from an optimally formulated ternary feedstock blend via machine learning-informed methanolysis using a composite biobased catalyst

Abstract

This work investigated biodiesel production from a blend of waste cooking oil (65.46 % WCO), waste palm oil (18.56 % WPO), and waste animal fat (15.98 % WAF) using a biobased heterogeneous catalyst. The catalyst was comprised of Ca (47.80 %), K (11.06 %), Mg (4.11 %), and Al (2.31 %). Biodiesel was produced via transesterification, and yield prediction and optimization were carried out using machine learning models and nature-inspired optimization algorithms. The catalyst's surface area and pore volume were 288.1 m2/g and 0.159 cm3/g respectively. The optimum biodiesel yield of 98.31 % was achieved at 61 °C, with 2 wt% catalyst concentration, 149.98 min reaction time, and a methanol-to-oil ratio of 6.01:1. The most influential input was the methanol-to-oil ratio, as revealed by global sensitivity analysis (GSA). The catalyst remained active for six cycles, and the produced biodiesel met quality standards. This study emphasizes the importance of machine learning and optimization algorithms in heterogeneous catalyzed biodiesel production.