Machine learning techniques for modeling chemical absorption in CO2 capture process

Post-combustion carbon capture (PCC) technologies play an important role in the reduction of CO2 emissions to address climate challenges. This process is usually simulated simulation software based on first-principle models, which calculate physical properties directly from basic quantities such as mass and temperature. Using models requires a long computation time, makes optimization control difficult. In this study, machine learning algorithms, eXtreme Gradient Boosting (XGBoost) Support Vector Regression (SVR), are investigated potential alternative modeling approaches. XGBoost ensemble algorithm that decision tree optimized by gradient boosting. SVR fits best line within predefined or threshold error value. These two algorithms used build predict rate (CR) specific reboiler duty (SRD) monoethanolamine-based PCC process. By using XGBoost, verification result shows R2 (a statistical measure represents fitness model) predicting CR 91.7% SRD 80.8%, while 87.9% 87.2% individually. addition, take 0.022 seconds 0.317 respectively 1318 cases, first-principal model needs 3.15 1 case. The data-driven built employed for further optimization, aims find operating point have higher lower SRD. Particle swarm (PSO), stochastic technique movement intelligence swarms, implemented optimization. optimal conditions 72.2% 4.3 MJ/kg each. computations faster with incorporated technique. Thus, application techniques demonstrated successfully.