A high-throughput computational screening of potential adsorbents for a thermal compression CO2 Brayton cycle

Abstract

By employing heat rather than mechanical work to compress the working fluid, the thermal compression CO₂ Brayton cycle (TC-CBC) has been considered as a promising pathway to the efficient utilization of low-grade thermal energy. However, finding reasonable adsorbents to efficiently realize the thermal compression process via the CO₂ adsorption–desorption loop has become a significant challenge to the development of such an innovative system. To solve the dilemma, high-throughput computational screening based on grand canonical Monte Carlo (GCMC) simulations and machine learning (ML) have been conducted to identify promising adsorbents from 1625 metal–organic frameworks (MOFs) for the TC-CBC. Results demonstrate that the thermodynamic efficiency and output per unit mass adsorbent of the system with a low-temperature heat source at 393 K can reach up to 9.34% and 21.84 kJ kg⁻¹, respectively. MOFs with large surface area, pore volume, porosity, and moderate pore size have exhibited high thermodynamic performances. In addition to the low-temperature heat source, a high-temperature heat source is also considered in the analysis. The elevation of the thermodynamic performance is observed to be dependent on the structural properties of MOFs. With a random forest algorithm, a rapid and accurate prediction of thermodynamic performances for the innovative cycle is achieved.