Sustainable bio-oil generation from Chlorella vulgaris: Catalytic hydrothermal liquefaction and predictive modeling using RSM, ANN, and machine learning

Abstract

The catalytic hydrothermal liquefaction (HTL) of Chlorella vulgaris algal (CVA) biomass was systematically investigated to advance sustainable biofuel production by maximizing bio-oil yield and minimizing biochar generation. Potassium carbonate (K2CO3) served as an alkaline catalyst to exploit the inherent benefits of CVA, such as its high hydrogen-to-carbon ratio and superior calorific value, for enhanced biofuel conversion. Response Surface Methodology (RSM) combined with Central Composite Design (CCD) was employed to optimize key process variables, including operating temperature, residence time, biomass-to-solvent ratio, and catalyst dosage. The regression model demonstrated strong predictive accuracy for bio-oil and biochar yields (R2 = 0.9355), identifying optimal conditions with 15 g catalyst loading and a 10.573 wt% biomass-to-solvent ratio, leading to a bio-oil yield of 35.019 wt% at reduced residence time. The predictive robustness of the model was further validated using Artificial Neural Network (ANN) and Support Vector Machine (SVM) methods. The ANN demonstrated a Mean Squared Error of 0.2276 and an R2 value of 0.99, while the SVM exhibited a Root Mean Square Error of 0.537 and an R2 of 0.97. Characterization of the bio-oil through FTIR and GC-MS analysis confirmed that K2CO3 catalysis effectively reduced nitrogen content and enhanced the conversion of proteins and lipids into valuable bio-products, such as indole and pyrrolic derivatives. The resulting bio-oil was rich in phenolics, aliphatic hydrocarbons, ketones, aromatics, and carboxylic acids. K2CO3-catalyzed conversion emerges as an efficient and sustainable strategy for producing high-grade bio-oil from algal biomass, contributing significantly to renewable energy advancement under SDG 7.