Can a hydrodynamic model predict the flow evolution of a hydrogen plume in a depleted natural gas reservoir?

Hydrogen storage in depleted natural gas reservoirs is a promising solution for storing excess renewable energy on a timescale longer than can be accommodated by salt cavern storage. However, commercial exploitation of the technology in question awaits the resolution of key challenges and uncertainties. Chief among these from a fluid mechanics point of view is to resolve the connection between hydrogen leakage and the mixing of hydrogen and cushion gas. Reduced-order-models examining this question have been developed i.e. by Sheikhi and Flynn (2024), however their work relies much more heavily on fluid mechanics than it does on thermodynamics. It is therefore unclear the extent to which their model predictions are accurate when compared with e.g. the numerical output from reservoir-level simulation packages. Addressing this knowledge gap is the key focus of the present study. To this end, we use OpenGoSim and CMG to numerically resolve hydrogen injection directly below an interbed layer of reduced, but still finite, permeability. The resulting comparison demonstrates that the theoretical model predicts, with generally good accuracy, the overall shape of the gravity current plus the amount of hydrogen that dispersively mixes into the surrounding cushion gas. However, reduced-order-model fidelity suffers when the injection time is long, the draining layer is thin and the interbed layer admits a relatively large drainage. This observation highlights areas of future improvement for the reduced-order-model, which can otherwise be applied, with great computational efficiency, in screening candidate reservoirs for hydrogen storage.