Reducing energy and economic costs in post-combustion carbon capture with intelligent amine selection system: A case study on the Abadan power plant

Abstract

Global warming has become a pressing global concern, manifesting in rising temperatures and significant adverse effects on human health. In response, various technologies, including carbon capture solutions, have been developed to mitigate carbon emissions. Among these, post-combustion carbon capture (PCC) is particularly noteworthy for its compatibility with existing industrial emission sources. However, PCC systems face significant energy challenges, primarily due to the lack of optimal solvents for specific applications, complicating solvent selection in a diverse market. To address this, we utilized multiple machine learning models to predict and optimize amine solvent selection for a PCC unit at the Abadan power plant. The models tested included KNN, logistic regression, Gaussian processes, decision trees, random forest, AdaBoost, gradient boosting, support vector classification, and artificial neural networks (ANNs). Among these, the random forest model demonstrated the highest performance, achieving a mean area under the curve (AUC) of 0.97, identifying monoethanolamine (MEA) as a potential candidate, though not the final selection. Subsequent simulation and economic analysis were conducted using Aspen HYSYS v11 to compare MEA with diglycolamine (DGA). The results revealed that DGA was the most cost-effective option, reducing total utility system costs by 35.76 % compared to MEA due to its lower solvent regeneration requirements. This study highlights the synergy between machine learning and process simulation tools like Aspen HYSYS in optimizing PCC systems for enhanced economic and environmental performance.