Optimization of CO₂ absorption into MDEA-PZ-sulfolane hybrid solution using machine learning algorithms and RSM.

Area of Science:

• Chemical Engineering
• Computational Chemistry
• Environmental Science

Background:

• Carbon dioxide (CO₂) absorption is crucial for mitigating greenhouse gas emissions.
• Hybrid amine solutions offer synergistic advantages over single solvents for CO₂ capture.
• Accurate modeling is essential for optimizing CO₂ absorption processes.

Purpose of the Study:

• To model and simulate CO₂ absorption in hybrid amine solutions.
• To compare the predictive accuracy of various machine learning algorithms and Response Surface Methodology (RSM).
• To optimize CO₂ absorption using a genetic algorithm (GA).

Main Methods:

• Utilized seven machine learning models: MLP, RBF, LightGBM, XGBoost, Random Forest, ExtraTrees, and Adaboost.
• Employed Response Surface Methodology (RSM) for comparative analysis.
• Applied a genetic algorithm (GA) for parameter optimization.

Main Results:

• MLP and RBF models achieved high R² values (0.9982 and 0.9975, respectively), demonstrating superior predictive performance.
• Neural networks showed better accuracy and generalization compared to RSM.
• CO₂ absorption increased with pressure and time, but decreased with temperature and solvent concentration.

Conclusions:

• Machine learning, particularly MLP and RBF, provides highly accurate predictions for CO₂ absorption in hybrid amine systems.
• The study identified optimal operating conditions for maximizing CO₂ loading through GA-driven optimization.
• Findings support the use of advanced computational methods for designing efficient CO₂ capture technologies.