Experimental studies and machine learning-based prediction models: Influence of Reaction time on the product yields and composition during thermochemical conversion of sewage-water grown microalgae

Chlorella sorokiniana was grown in an open raceway pond of ∼10000 L volume, utilizing sewage water as its nutrients/growth media. Pilot-scale hydrothermal liquefaction (HTL) of the harvested algal biomass was conducted in a 50 L stainless steel batch reactor. The role of reaction time (0.5, 1, 1.5, 2, and 2.5 h) was investigated on the product yield as well as composition distribution. The highest bio-oil and biochar yield of 34.04% and 15.74% were attained at 250 °C for 2 and 0.5 h, respectively, recording the maximum higher heating value (HHV) of 43.27 MJ/kg for bio-oil. The chemical composition and quality of the products were mapped based on advanced characterization techniques. A larger number of lower molecular weight hydrocarbons in bio-oil were obtained at 0.5 and 2 h reaction time. Longer HTL reaction time also led to a decrease in crystallinity of biochar. It is evident that reaction time can manipulate the quality and quantity of the products, clearly indicating the importance of reaction time optimization for ideal composition and higher energy recovery. The study also introduces a novel integration of AI-ML driven models (Random Forest, XGBoost, Support Vector Machine (SVM), k-Nearest Neighbors (kNN), and Linear Regression) with thermochemical conversion data of sewage-grown microalgae, enabling accurate prediction of product composition and quality. From the results, it is clear that Random Forest (test R2 of 0.96) demonstrated the highest predictive accuracy, followed by XGBoost (test R2 of 0.93), while Linear Regression (test R2 of 0.81) showed comparatively lower performance due to its limited ability to capture non-linear relationships in the dataset. By applying machine learning (ML) for yield analysis, the research offers a data-driven framework that enhances insight beyond conventional experimental methods. Such technological aspects for biofuel production would significantly aid in achieving the Sustainable Development Goals (especially SDGs 6, 7, 11 & 13) for a bio-circular-green economy.