Machine learning in absorption-based post-combustion carbon capture systems: A state-of-the-art review

The enormous consumption of fossil fuels from various human activities leads to a significant amount of anthropogenic CO2 emission into the atmosphere, which has already massively contributed to climate change and caused harmful impacts on human life. Carbon capture and storage (CCS) technologies have emerged as short-to-mid-term solutions to reduce atmospheric CO2 concentrations. The absorption-based post-combustion carbon capture (PCC) technology is considered the most established, traditional, and operational approach compared to other CCS technologies. Modelling and optimizing the PCC process, such as operating conditions, equipment configurations, and solvent management, are time-consuming and computationally expensive. Machine Learning (ML) has gained significant attraction as a powerful tool for conducting complex computations that facilitate the training of computer algorithms to perform specific tasks with exceptional precision, which is unattainable through conventional tools. They have been used for various applications in an efficient and costeffective approach, including classification, prediction, clustering, ranking, and data optimization. In this article, we review the recent research progress on applying ML methods to PCC absorption-based technologies. This review provides a practical guide to categorizing the various ML methods used in PCC technologies based on limits, availability, and pros and cons. Finally, we propose a roadmap for community efforts to show the possible pathways and future research areas for developing the application of ML methods in PCC absorption-based technologies.