Optimal Equation of State Selection for Modeling CO2 Density with Impurities for Carbon Capture, Usage and Storage Applications

Summary This study develops a data-driven algorithm to predict CO2 stream density, across different flow regimes, by analyzing 134,204 density data points for CO2-rich mixtures. Fourteen equations of state (EoS), including cubic, virial, physical, and multiparameter models, were evaluated to assess impurity effects from fossil fuel-based power generation. Poorly performing EoS were eliminated using Pareto plots while decision trees with Gini and Entropy indices classified performance based on pressure, temperature, and CO2 purity. The study highlights the superior performance of multiparameter EoS while recognizing the situational advantages of others. We present a dynamic algorithm that adapts to phase transitions and critical conditions, improving CO2 flow predictions as it compares to using a single EoS model.