Advancing Catalytic Reaction Rate Model Development with Physics-Guided Hybrid Modeling

Cu/ZnO/Al 2 O 3 catalysts are widely applied in industry, yet the influence of operating conditions and feed composition on their activity and selectivity in methanol synthesis and the water gas shift reaction remains poorly understood. The complexity of the reaction system and the scarcity of informative data hinder the development of conventional kinetic models that could guide process improvements and rational catalyst design. To address these challenges, we propose a hybrid modeling framework that integrates mechanistically and thermodynamically rigorous power law expressions─where apparent kinetic parameters are linked to surface coverages of intermediates─with data-driven modeling. A novel physics-guided strategy is introduced, whereby the hybrid model is trained with soft constraints derived from prior reaction mechanism insights. The framework is demonstrated using spatially resolved kinetic data, where catalytic activity was strongly influenced by temperature, pressure, and CO/CO 2 ratios. The hybrid model achieves satisfactory predictive accuracy across diverse conditions, with uncertainty intervals consistent with experimental variability, underscoring its robustness in capturing process trends. This study demonstrates the potential of hybrid modeling frameworks to accelerate the development of reliable reaction rate models, enabling improved optimization and control of complex catalytic reaction systems.