Carbon dioxide solubility in polyethylene glycol polymer: an accurate intelligent estimation framework

Area of Science:

• Polymer Science
• Chemical Engineering
• Computational Chemistry

Background:

• Polyethylene glycol (PEG) is a versatile synthetic polymer with numerous applications.
• Accurate estimation of carbon dioxide (CO2) solubility in PEG is crucial for processes like carbon capture and supercritical fluid extraction.
• CO2 is increasingly utilized as a solvent or transport medium in various industrial applications.

Purpose of the Study:

• To develop advanced machine learning models for predicting CO2 solubility in PEG.
• To compare the performance of different machine learning algorithms, including Random Forest, Decision Tree, Adaptive Boosting, K-Nearest Neighbors, and Ensemble Learning.
• To identify key parameters influencing CO2 solubility in PEG and their relationships with the target variable.

Main Methods:

• Data compilation from existing literature on CO2 solubility in PEG.
• Application of outlier detection methods to ensure data quality.
• Development and evaluation of multiple machine learning models (RF, DT, AdaBoost, KNN, EL).
• Sensitivity analysis to determine the impact of input parameters (pressure, temperature, PEG molar mass) on CO2 solubility.

Main Results:

• The Decision Tree (DT) model demonstrated superior performance in predicting CO2 solubility.
• The DT model achieved high accuracy with an R-squared value of 0.991 (training) and 0.801 (testing), and low error metrics (MSE: 0.0009, AARE%: 22.58).
• Sensitivity analysis revealed that pressure and PEG molar mass have a direct relationship with CO2 solubility, while temperature has an inverse relationship.

Conclusions:

• The developed Decision Tree model provides an accurate and robust tool for estimating CO2 solubility in PEG.
• This predictive model eliminates the need for time-consuming, expensive, and tedious experimental measurements.
• The findings contribute to optimizing processes involving CO2 and PEG, such as carbon capture and polymer modification.