A novel computational strategy to estimate CO2 solubility in brine solutions for CCUS applications

[Display omitted] •Developed a Machine-learning-based framework to estimate CO2 solubility in brine.•Validated the workflow & model against experimental data for various salt mixtures.•Model predictions on CO2 solubility are accurate and within 2–7% relative error.•The modelling framework is predictive and can be extended to any salt mixture.•Main contributor/feature and ionic properties were identified for CCS applications. Estimation of CO2 solubility in brine is crucial to various CCUS (carbon capture utilization and storage) applications, especially for engineering design of the physical/chemical processes. In this work, we developed machine learning based workflow to calculate CO2 solubility in brine at various combinations of salt mixtures, pressure, and temperatures. Most importantly, the performance of predictive models and workflow were tested against extensive experimental observations and key features of brine components with significant contributions were determined.