Estimating CO2 solubility in brine with Artificial Intelligence : a comparative study

Estimating CO2 solubility in various solutions such as brine is one of the crucial tasks in CO2 absorption phenomenon used in desperate chemical industries. Different techniques such as experimental procedures, models, and correlations have been introduced to estimate the mentioned parameter. However, experimental methods are very expensive and time consuming. Although there are some physical and thermo-dynamical models developed for this purpose, they are restricted with certain pressure and the temperature ranges and are also associated with mathematical complications. Hence, these models are not reliable for all pressure-temperature conditions and they are difficult to be implemented. Therefore, developing a cheap, fast and rigorous technique to estimate the CO2 solubility in brine solution seems ineluctable. Therefore, to overcome the aforementioned problems, an Artificial Neural Network (ANN) is employed in this study to estimate the CO2 solubility in the brine solution and its results are compared with Duan and Sun’s (2003) model, and Henry’s correlation (2008). Both statistical analysis and regression plot unveils that the developed ANN-based model outperforms the indicated physical model and correlation in estimating the CO2-brine solubility data.

Keywords:

CO2 solubility estimation , Artificial intelligence , Artificial Neural Networks