Biodiesel Production from Waste Cooking Oil using Modified Poultry Droppings as Catalyst. Response Surface Methodology, Artificial Neural Network, and Extreme Gradient Boosting as Modeling and Optimization Tools

Abstract

Poultry droppings was modified with nickel sulphate for the synthesis of biodiesel from waste cooking oil (WCO) with 6.43 % free fatty acid. The catalyst modified was characterized using Brunauer Emmet Teller (BET), Fourier Transform Infrared Spectroscopy (FTIR), Scanning Electron Microscopy and Energy dispersive X-ray (SEM-EDX), X-ray diffraction (XRD), and X-ray fluorescent analysis (XRF). Response surface methodology (RSM), artificial neural network (ANN), and extreme gradient boosting (XGB) were employed in modeling and optimizing the process conditions. Box-Behken design (BBD) of four process variables was used in designing the experiment. The biodiesel produced was characterized and its properties compared with established standards. The reusability potential of the catalyst was also assessed. The catalyst characterization revealed high surface area of 355.36 m2/g. Acidic and basic oxides such as CaO, SiO2, NiO, SO3, Al2O3 and Fe2O3 in significant quantities were found present indicating its bi-functionality. Optimum biodiesel yield of 96.82 % was obtained with methanol/oil ratio of 12.98:1, catalyst loading of 3.66 wt.%, temperature of 59℃, and reaction time of 86.45 minutes. RSM, ANN and XGB model were found to be efficient in modeling biodiesel production process with XGB performing best with the highest R2 value of 0.9915. The properties of biodiesel produced were within the acceptable limits when compared with ASTM D6751 and EN 14214 standards.