Model-driven AI analysis and optimization of a dynamic continuous stirring tank reactor system for biofuel production by lignin bio-oil upgrading

Abstract

This study develops an integrated mechanistic-data-driven AI framework to model a biofuel production process by the conversion of lignin-derived bio-oil with the hydrodeoxygenation of isoeugenol (I-HDO). A transient continuous stirring tank reactor (CSTR) model is developed utilizing experimentally supported kinetics and is then employed to generate an extensive database. Multiple artificial neural networks are evaluated to recognize the reactor behavior, among which the double-layer cascade feedforward scheme trained with the Broyden-Fletcher-Goldfarb-Shanno algorithm (DLCFF- trainbfg ) delivers the highest accuracy (an average absolute relative deviation of 1.04%) without violating physical constraints. Statistical error analyses further confirm the reliability of this model, which is subsequently utilized to conduct a sensitivity analysis of the process. Sensitivity analysis reveals the dominant influence of feed temperature, flow rate, and concentration, as well as catalyst load on steady-state establishment, reaction mechanism and product distribution, product yields, biofuel quality, and hydrogen requirement. Then, the DLCFF- trainbfg model is integrated with a genetic algorithm to determine optimal operating conditions that yield the highest production rates of propylcyclohexene and propylcyclohexane and heating value in tandem with the lowest hydrogen consumption. The optimized condition resulted in a rapid development to a steady-state regime (3.58 h) and a high-energy-density (48.66 MJkg) biofuel stream rich in propylcyclohexene and propylcyclohexane (hydrocarbon yield of 66.23 wt%), while minimizing oxygenates yield (1.63 wt%) and hydrogen consumption (1.272 m3hr). The results confirm that the hybrid AI-assisted approach accurately describes I-HDO behavior, provides computationally efficient optimization, and offers a scalable pathway for the design and intensification of lignin-derived biofuel production.