Data-driven modelling to predict interfacial tension of hydrogen–brine system: Implications for underground hydrogen storage

• Interfacial tension (IFT) between the injected hydrogen gas, cushion gas, and resident reservoir brines is crucial during underground hydrogen storage • Data-driven modelling of the IFT between a H 2 -cushion gas-brine system was conducted • Adaptive Neuro-Fuzzy Inference System (ANFIS) was coupled with Genetic Algorithm (GA) to optimize the independent variables (salinity, temperature, and pressure). • The relative sensitivity analysis result showed that salinity is the most significant input variable on the interfacial tension for most of the cases studied. This study focuses on optimizing underground hydrogen storage by addressing a critical technical challenge: accurately estimating the interfacial tension (IFT) between injected hydrogen gas, cushion gas, and reservoir brines. Underground geologic formations, with their vast capacity for hydrogen storage, have gained significant attention. However, determining the IFT which is essential for evaluating storage and withdrawal efficiencies remains a complex and time-consuming process. To overcome this challenge, a data-driven modelling approach was employed using the Adaptive Neuro-Fuzzy Inference System (ANFIS). The model demonstrated a high predictive accuracy across various conditions, achieving a correlation coefficient (R²) > 0.9 and low statistical error as indicated by the root mean square error (RMSE) < 0.8. To further enhance the model's performance, ANFIS was coupled with a Genetic Algorithm (GA) for optimizing the independent variables: salinity, temperature, and pressure. The results highlight the exceptional predictive capability of the ANFIS-GA model for complex hydrogen–cushion gas–brine systems, providing a robust, efficient alternative for IFT estimation and optimization. This methodology significantly improves the feasibility of underground hydrogen storage by streamlining critical evaluations of fluid-fluid interactions. Sensitivity analysis revealed salinity as the most influential factor affecting IFT, underscoring its importance in designing and managing storage systems.