Biocrude production via hydrothermal liquefaction of cycas circinalis seed shell: A machine learning approach

Hydrothermal liquefaction (HTL) is a promising thermochemical method for converting biomass into bio-crude fuel. This study explores the HTL of Cycas circinalis seed shell (CSS), focusing on the impacts of reaction time, feed slurry concentration, and temperature on bio-crude yield. Experiments were conducted at temperatures ranging from 250 to 375°C, reaction times from 10 to 40 minutes, and feed slurry concentrations between 10% and 30%. A decision tree regression (DTR) model predicted the optimal bio-crude yield of 35% at 375°C, 30% feed slurry concentration, and 40 minutes, with high accuracy (R² = 0.9853, RMSE = 0.992). Results highlight temperature and time as key factors influencing bio-crude production.The bio-crude was characterized using Fourier-transform infrared spectroscopy (FTIR), thermogravimetric analysis (TGA), nuclear magnetic resonance (NMR), and gas chromatography-mass spectrometry (GC-MS). Degradation kinetics of bio-crude and CSS biomass were analyzed using the Coats-Redfern method at heating rates of 5, 10, and 20°C/min. Parameters such as activation energy (E), rate constant, pre-exponential factor (A), enthalpy, entropy, and Gibbs free energy were determined. This research advances hydrothermal technology and promotes the development of sustainable and efficient biomass conversion processes.