How to accurately fast-track sorbent selection for post-combustion CO2 capture? A comparative assessment of data-driven and simplified physical models for screening sorbents

Abstract

The recent discovery of a multitude of hypothetical materials for CO2 capture applications necessitated the development of reliable computational models to aid the quest for better-performing sorbents. Given the computational challenges associated with existing detailed adsorption process design and optimization frameworks, two types of screening methodologies based on computationally inexpensive models, namely, data-driven and simplified physical models, have been proposed in the literature. This study compares these two screening methodologies for their effectiveness in identifying best-performing sorbents from a set of 369 metal-organic frameworks (MOFs). The results showed that almost 60% of the MOFs in the top 20 best-performing materials ranked by each of these approaches were found to be common. The validation of these results against detailed process simulation and optimization-based screening approach is currently underway. © 2023 Elsevier B.V. Author keywords adsorption; machine learning; metal-organic frameworks; modelling and optimization; post-combustion CO2 capture

How to accurately fast-track sorbent selection for post-combustion CO2 capture? A comparative assessment of data-driven and simplified physical models for screening sorbents