(71c) Guiding the Experimental Design for Empirical Modeling of the Reverse Water Gas Shift Reaction

The Reverse Water-Gas Shift (RWGS) reaction is essential for converting carbon dioxide (CO₂) into carbon monoxide (CO), an important step in synthetic fuel and chemical production. Due to the reaction’s complex kinetics, catalyst variability, and range of reaction conditions, numerous models exist, leading to uncertainty about which model best fits a given application [1], [2]. In this work, we demonstrate how informed Gaussian Process models can efficiently represent a RWGS system. Coupled with Bayesian Optimization, we propose a Guided Design of Experiment (G-DoE) strategy that continuously incorporates new experimental data to inform subsequent experimental designs, modeling outputs like conversion and selectivity [3]. This adaptive, data-driven approach samples the design space more efficiently, achieving accurate models with fewer experiments. Our method significantly reduces experimental budget and facilitates effective process modeling and optimization, highlighting AI's potential to transform the oil and gas industry.

References:

[1] M. Schwaab, F. M. Silva, C. A. Queipo, A. G. Barreto, M. Nele, and J. C. Pinto, “A new approach for sequential experimental design for model discrimination,” Chem Eng Sci, vol. 61, no. 17, pp. 5791–5806, 2006.

[2] F. Vidal Vázquez, P. Pfeifer, J. Lehtonen, P. Piermartini, P. Simell, and V. Alopaeus, “Catalyst Screening and Kinetic Modeling for CO Production by High Pressure and Temperature Reverse Water Gas Shift for Fischer–Tropsch Applications,” Ind Eng Chem Res, vol. 56, no. 45, pp. 13262–13272, Nov. 2017.

[3] B. Malluhi, R. Fezai, C. Kravaris, H. Nounou, M. Al-Rawashdeh, and M. Nounou, “Guided experimental design for static nonparametric modeling,” Chem Eng Sci, vol. 298, p. 120327, 2024.