Exploring cascaded adsorption cooling cycles with MOF − hydrocarbon integration: An accelerated computational framework using Bayesian Optimization and Monte Carlo

Abstract

Rising global temperatures have intensified the demand for air conditioning, yet conventional vapor compression cycles (VCCs) impose significant energy loads on power grids, further exacerbating climate change. Adsorption-based cooling cycles (ARCs) present a sustainable alternative; however, their widespread adoption is hindered by low efficiencies due to suboptimal sorbents and low-vapor-pressure refrigerants. In this study, we explore cascaded ARC systems incorporating metal–organic frameworks (MOFs) and hydrocarbon refrigerants (propane and isobutane) as a viable alternative to VCCs. A three-level accelerated Bayesian optimization framework was employed on the CoRE MOF Database to identify high- performance MOFs. We demonstrate that simulating ~ 5% of the database using the Bayesian optimization approach is sufficient to identify the top-performing MOFs. Further we analyse the top 30 MOFs obtained from the Bayesian optimization sampling to understand the influence of structural properties on the COP of the cycle. We have also compared the prediction accuracy from Bayesian Optimisation (BO) compared to the conventional approaches emphasizing the significance of BO process in materials discovery. Our results demonstrate a maximum coefficient of performance (COP) greater than 1 for both refrigerants. Specifically, the COPs for isobutane and propane were improved by 100% and 42%, respectively, compared to their single-stage ARC counterparts, thereby enhancing the practical viability of these systems.