CO2 utilization for H2-rich syngas production in a combined system: Bi-objective optimization and machine learning analysis

• A novel system used CO 2 gasification of biomass and PET to produce clean syngas. • Residual analysis confirmed the robustness of the machine learning-based models. • CGE and H 2 /CO ratio were simultaneously optimized through machine learning. • Thermodynamic and machine learning results showed high consistency and accuracy. This study aimed to mitigate environmental risks in energy production through the design of a system that generates high-quality syngas from a blend of poplar wood and polyethylene terephthalate waste. CO 2 was employed as the gasifying agent, an approach that both eliminates nitrogen dilution in the syngas stream and offers a practical pathway for CO 2 utilization from industrial emissions, thereby linking clean energy production with greenhouse gas reduction. To assess the validity and robustness of the developed models, a residual analysis was performed. Subsequently, a bi-objective optimization was conducted to simultaneously maximize cold gas efficiency and the H 2 /CO ratio. The reliability of the machine learning model was evaluated by comparing its predictions with the outcomes derived from thermodynamic simulations. The results demonstrated that the optimal operating range was within a gasifier agent to fuel of 1.95–2.15 and a water gas shift reactor agent to fuel of 1.75–1.90. In this range, the system achieved cold gas efficiencies between 97% and 98%, along with H 2 /CO ratio percentage ranging from 80% to 90%. The comparative analysis indicated that the results predicted by machine learning models showed strong agreement with those obtained from the engineering equation solver simulation software.