Optimization of biodiesel production via central composite design and machine learning approach from spent cooking oil using thermally treated Oro Anthill as a catalyst

Environmental challenges, including climate change, greenhouse gas emissions, and declining fossil fuel reserves, have driven the development of biofuels. This study introduces a heterogeneous catalyst derived from Oro Anthill through calcination in a muffle furnace at 800 °C for 4 h. The resulting catalyst exhibited high porosity, with a surface area of 150.40 m²/g, and demonstrated remarkable stability and catalytic efficiency. Using Central Composite Design (CCD) and a Machine Learning (ML) approach implemented via Python, the biodiesel yield from spent cooking oil was modeled and optimized. The CCD approach achieved a maximum biodiesel yield of 94.94 wt% under conditions of a 2 h reaction time, 60 °C, 1.25 wt% catalyst loading, and a methanol-to-oil ratio of 7:1. In contrast, the ML model predicted a higher yield of 96.18 wt% under optimized conditions of 1 h, 55 °C, 0.5 wt% catalyst, and a 6:1 methanol-to-oil ratio. The coefficient of determination (R²) for the CCD model was 0.9959, while the ML model achieved a perfect R² of 1.0, indicating superior predictive accuracy and precision for the ML approach. Both models were reliable within the tested data range. Gas Chromatography-Mass Spectroscopy analysis of the biodiesel confirmed the presence of saturated and unsaturated fatty acid methyl esters, with properties meeting ASTM biodiesel standards. This work highlights the transformative potential of converting natural biowaste into an efficient green catalyst, enabling cleaner and more cost-effective biodiesel production from spent cooking oil and driving a truly circular, sustainable future for the fuel industry.