Supercritical water gasification of lignocellulosic biomass: Kinetic modeling enhanced by Bayesian optimization

A new kinetic model for the supercritical water gasification of lignocellulosic biomass is presented. This model accurately simulates the production of gases up to C 3 products, and distinguishes condensable products as either aqueous condensate or tars. The reaction orders were treated as hyperparameters, which were tuned via Bayesian optimization. The model was validated against experiments conducted in a continuous laboratory plant operating within a broad range of conditions, including variations in temperature (823–973 K), pressure (240–300 bar), biomass content (1.3–6.6 wt%), K 2 CO 3 addition (0–3750 ppm), and residence time (6–48 s). The proposed model provides insight into the behavior of the reactive system and is useful for applications such as scale-up, process optimization, and process design. • New kinetic model for SCWG of lignocellulosic biomass predicts gas products up to C 3 . • Reaction orders tuned via Bayesian optimization • Model validated in a broad operating window with the cross-validation method • Obtained insight into the behavior of the reactive system