Partial Surrogate Modeling Approach for Multicomponent Adsorption Process Simulation and Optimization using Ideal Adsorbed Solution Theory

Multicomponent adsorption equilibria strongly influence the performance of adsorption-based carbon dioxide separation processes. Ideal Adsorbed Solution Theory (IAST) provides a practical framework to predict competitive adsorption from pure-component isotherms, but its application in process simulation and optimization is often hindered by the high computational cost associated with implicit equilibrium calculations and numerical integration. This study develops a partial surrogate model that reproduces IAST predictions while substantially reducing computation time. The surrogate replaces only the equilibrium calculation rather than the full process model; it preserves the governing mass and energy balances and cycle operation, is straightforward to validate against equilibrium reference data, and can be reused across different process configurations and operating conditions with minimal reimplementation. A multilayer perceptron was trained using equilibrium data generated from IAST calculations based on fitted pure-component isotherms for the metal–organic framework UTSA-16. The surrogate model was successfully integrated into a two-bed temperature swing adsorption model, enabling dynamic process simulations that capture time-dependent competitive adsorption behavior. Finally, the surrogate-integrated temperature swing adsorption superstructure was optimized as a large-scale nonlinear programming problem.