Predictive Modeling of Hydrogen Wettability Shifts in Saudi Arabian Basalt Due to Organic and Nanofluid Aging

Abstract Subterranean hydrogen storage on a large scale is an essential component of the value chain for the hydrogen economy and a prerequisite for the successful replacement of carbon-based fuels. Recent research has concentrated on the wettability of rock-H2-brine systems, as measured by contact angles, because of the effects it has on underground hydrogen storage fluid flow, H2 migration, and recovery efficacy. However, the contact angle data sets that have been reported are highly inconsistent. Furthermore, in comparison to the contact angle data for quartz, shale, mica, and calcite, the literature provides a scarcity of information regarding the contact angles of H2/brine on Saudi Arabian basalt (SAB). This study focuses on accurately modeling the wettability behavior of a ternary system comprising H2, Saudi Arabian basalt (SAB), and 0.3 molar NaCl brine solution under various physio-thermal conditions (298 – 323k and 0.1 – 20 MPa) in presence of organics (stearic acid-aged; 10−2 mol/L) and nanofluids (0.05, 0.1, 0.25, and 0.75 wt%; SiO2) using various machine learning techniques, including, Bayesian ridge, Extra trees, CatBoost, Gradient Boosting, Extreme Gradient Boosting, and Random Forest. A comprehensive dataset, derived from laboratory experiments conducted under realistic pressure and temperature conditions was utilized. To enhance the understanding of the dataset, various graphical exploratory data analysis techniques were employed. The model's generalization capabilities were improved through k-fold cross-validation and grid search optimization. The machine learning models were trained to predict the advancing and receding contact angles of the Saudi Arabian basalt/H2/brine systems. Statistical evaluation and graphical analysis were used to assess the models' reliability and performance. The results demonstrated that the machine learning models accurately predicted the wettability behavior across different operating conditions. The XGB model achieved highest accuracy, evidenced by low average absolute percent relative errors and high R2 values. The investigation into feature importance revealed that pressure exerted the most significant influence on the contact angles within the SAB/H2/brine system. Accurate predictions of wettability behavior can enhance the estimation of H2 geo-storage capacities and ensure containment security in large-scale geo-sequestration projects.