Reservoir simulation studies in underground hydrogen storage in a depleted gas reservoir - northwestern Germany

Abstract

To achieve a carbon-free economy in the medium term, hydrogen has been proposed as a viable solution. This requires large-scale subsurface storage options, especially, if green hydrogen produced from fluctuating renewable energy sources like wind and solar energy is considered. While H2 has already been stored successfully in salt caverns for decades, H2 storage in porous media like hydrocarbon-depleted reservoirs and saline aquifers still requires further research. We use an almost depleted gas reservoir in northwestern Germany to test various scenarios regarding withdrawal/injection cycles and different cushion gases. The case study field presents a faulted reservoir in a highly fractured rock of Upper Permian (Zechstein) age, consisting mainly of dolomite as reservoir rock and anhydrite as cap rock. A history-matched dynamic model starting in 1959 of a gas-depleted reservoir calibrated from the comprehensive information available for the reservoir site, such as density, viscosity, relative permeability, and capillary pressure, which serves as a hypothetical base case for seasonal hydrogen storage, intending to store around 300 Mio sm3. An isothermal compositional reservoir simulator with seven components is used including H2S to monitor its concentration. Eight prediction cases were simulated, excluding: diffusion, dispersion, and microbial reaction. Between each case, changes are made to the type and amount of cushion gas injected following the same injection/withdrawal cycle, mixing the cushion gas between N2+CH4, H2+N2, H2+CH4, H2+CO2, pure CH4, pure CO2, pure N2, and pure H2. Following an initial filling from only the cushion gas of 33-months of around 730000 (sm3/d). Immediately after, withdrawal begins for 2 months from the working gas of around 3600000 (sm3/d) and withdrawal/injection cycles for 3(W)/6(I) months were the amount of working gas injected increases to 1800000 (sm3/d), and with a shut-down phase for 1 month after withdrawal and 2 months after injection, for 7 times; resulting in a total H2 production over 8 cycles. The applied amounts were to avoid any spilling due to the highly-fracture nature of the reservoir. In a subsequent simulation from the case of using pure N2, the prediction time was increased to observe its changes over the next 7 years. To assess the overall recovery of hydrogen and the concentration of H2S, a volumetric and molar storage balance was analyzed. Based on the results of all the 8 simulations, at least on the first four cycles, less H2 is recovered, except if pure H2 is injected from the beginning as a filling phase. Despite this, all simulations show a greater H2 recovery for the last cycle, from 96% (pure N2 as cushion gas) to 99% (pure H2 as cushion gas). Regarding H2S, shows a diluted concentration while the storage cycles are increased, resulting lower than 2x10-5 mole fraction for the last cycle. A longer time prediction reveals that H2 recovery for the last cycle can nearly reach 100%. The next steps involve realizing a thermal simulation for the observation of the temperature effect and how can it effect the storage process, and a preliminary economic study of the storage site to determine its feasibility.