Data-Driven Optimization of Biodiesel Synthesis from Waste Cooking Oil over Sargassum sp.-Based Catalyst Using ANN and RSM

Abstract

The continuous decline in fossil fuel reserves, together with growing environmental issues, has increased the urgency to develop sustainable alternative energy sources. Biodiesel produced from waste cooking oil (WCO) offers a promising solution; however, conventional homogeneous catalysts remain limited due to their sensitivity to water content and free fatty acids (FFA), leading to soap formation and complicated separation processes. To address these challenges, this study explores the use of a heterogeneous catalyst synthesized from Sargassum sp. macroalgae. The originality of this study is reflected in integrating marine biomass-derived catalysts with data-driven optimization techniques to enhance biodiesel production efficiency. This study seeks to predict and optimize biodiesel production from waste cooking oil (WCO) using response surface methodology (RSM) and artificial neural network (ANN). Transesterification reactions were conducted at temperatures of 50–70 °C, reaction times of 60-180 min, catalyst loadings of 1–4 wt%. ANN was developed and trained in MATLAB R2022a using the LevenbergMarquardt algorithm, while RSM modeling was employed using a Box-Behnken design (BBD). The results indicate that the maximum experimental biodiesel yield reached 96.4 wt% under optimal conditions of a molar ratio of 1:9 and 60 °C within 120 min. Comparative analysis shows that ANN provides lower prediction errors than RSM, demonstrating superior accuracy in modeling biodiesel production under varying reaction conditions. The findings highlight the effectiveness of combining marine