Evaluating the Thermophysical Behaviours of H₂-CH₄ Mixtures for Underground Hydrogen Storage: Implications for Nigerian Gas Reservoirs

Underground hydrogen storage (UHS) has attracted a significant amount of interest recently due to the global demand for cleaner energy. This research investigates potential impurities in hydrogen systems during subsurface storage. It analyses the thermophysical properties of hydrogen-methane (H₂-CH₄) binary mixtures under varying temperature and pressure conditions using high-fidelity datasets developed using the GERG-2008 equation of state. This finding offers critical insights on thermophysical properties like density, viscosity, and heat capacity across a range of compositions. Possibly phase transitions were noticed at a certain pressure-temperature range. Using the same datasets, multiple machine learning models were deployed to predict key properties such as gas density, viscosity, and heat capacity across a range of subsurface P-T conditions similar to Nigeria's Niger Delta reservoirs. After extensive evaluation and hyperparameter tuning, the model achieved excellent predictive performance. Linear Regression was most effective for density and viscosity, while Random Forest Regression performed best for heat capacity prediction. The results from machine learning provide evidence that when models are trained on robust thermodynamic simulations, they can accurately predict hydrogen gas properties when mixed with impurities in the subsurface. This approach improves efficiency and can potentially reduce the cost of hydrogen-site screening. It can also enhance decision-making in Nigeria's energy transition and beyond.