Data-Driven Predictive Modeling of CO2 Absorption Efficiency in Nanofluids Using Machine Learning Techniques

The rising concerns about global warming and climate change, largely driven by carbon dioxide (CO₂) emissions, highlight an urgent need for effective environmental solutions. Utilizing nanofluids in the capture of CO₂ via absorption processes has emerged as a promising and efficient approach, attracting significant attention from diverse industries. Advancing complex technologies in this area requires a sophisticated predictive system to enhance resource utilization and streamline time management. To address this need, various machine learning models, including decision tree (DT), k-nearest neighbors (KNN), support vector regression (SVR), random forest (RF), extreme gradient boosting (XGBoost), and multilayer perceptron (MLP), were applied to predict CO₂ absorption by nanofluids. An extensive data set of 3630 experimental data points, covering a range of nanofluids and diverse temperature and pressure conditions, was used to train these models. Results showed that the XGBoost model achieved the highest accuracy with a determination coefficient (R²) of 0.99585, underscoring its reliability. Validation using the leverage method confirmed that 94.0221% of the data fell within the acceptable range. Finally, the analysis of input features indicated that pressure, temperature, and solvent density had the most significant impact on CO₂ absorption in nanofluids.