Impact of operational parameters on water invasion in underground hydrogen storage: A numerical investigation of the Ahuroa field, New Zealand

The problem

New Zealand's wind generation is seasonal, higher in winter and spring than in summer and autumn (Nottage, 2010). Surplus electricity could make hydrogen by electrolysis (Chang and Rajuli, 2024), to be stored until it is needed (Arsad et al., 2022). If every technically viable use switched to hydrogen, national demand could reach 3 million tonnes a year (Vatankhah Ghadim et al., 2024). New Zealand's gas storage supplies about a tenth of annual gas demand. Storing the same share of hydrogen demand means about 0.3 million tonnes. At the surface this would take about 100 of the most advanced cryogenic tanks (Nicol et al., 2022). Underground, several depleted gas reservoirs could meet it.

In a depleted gas field, years of production leave the gas-water contact, the depth below which the rock pores hold water, above its original depth. This raises the risk of water invasion, where the gas-water interface rises beneath a well as gas is withdrawn. Water reaching a well is produced with the gas and makes hydrogen withdrawal less efficient (Sainz-Garcia et al., 2017). How operating choices affect water invasion in hydrogen storage has received little study (Lysyy et al., 2021; Luboń and Tarkowski, 2020).

What we did

We built a reservoir model of the Ahuroa gas field, 14 km northeast of Stratford in Taranaki. The field produced 1.22 × 10^9 Sm³ (standard cubic metres) of gas and 6.1 × 10^4 m³ of water before it was converted to natural gas storage in 2008. We ran the model in OpenGoSim, an open-source reservoir simulator, and calibrated it against the field's production rates and well pressures. We ignored chemical and biological reactions and gas dissolution in water.

Our baseline case starts with 90 days of injecting cushion gas, gas left in the reservoir to maintain its pressure, at 5 × 10^5 Sm³/day. Ten cycles follow, each injecting hydrogen at 10^6 Sm³/day for 90 days and then withdrawing gas at the same rate for 90 days. A cycle therefore injects and withdraws 9 × 10^7 Sm³, about a fourteenth of the gas the field produced in its lifetime (our arithmetic). We then varied six operating parameters one at a time and repeated the baseline with natural gas.

What we found

The baseline produced 3.44 × 10^4 m³ of water over ten cycles, a little over half of the field's production-era total (our arithmetic). Hydrogen gave a lower water production rate than natural gas throughout. We attribute this to the different compressibility of the two gases, that is, how much a gas's volume changes with pressure. For the same injected volume, hydrogen keeps the pressure at the bottom of the well higher (Poling et al., 2020). However, hydrogen holds about a third of the energy of natural gas per unit volume (Poling et al., 2020). Storing the same energy needs larger volumes of hydrogen. Cycling larger volumes may worsen water invasion.

Cushion gas injection rate made little difference (totals of 3.36 to 3.55 × 10^4 m³). Cushion gas volume had a larger effect. Injecting cushion gas for 30, 90 and 150 days gave 4.52, 3.44 and 2.74 × 10^4 m³ of water. This is because the lower reservoir pressure left by a smaller cushion persists throughout cycling and raises water production. However, cushion gas can account for nearly 50% of the investment in hydrogen storage in a depleted reservoir (Lord et al., 2014).

In our model, doubling the withdrawal rate doubled the water production rate. This can be attributed to faster pressure drops during withdrawal. Halving the injection rate had a smaller effect than changing the withdrawal rate. Withdrawal stage length had the largest effect in our sensitivity analysis. Extending the 90-day withdrawal stage to 93 or 95 days raised the water production rate at the end of the first cycle by 5.7 and 7.2 m³/day. This is because a longer withdrawal takes out more gas, by volume underground, than was injected. The longer stages also raised hydrogen recovery from 89% to 91% and 92.3%. However, we suggest caution once the withdrawn volume approaches the injected volume.

The effect of the working gas volume, the gas injected and withdrawn in each cycle, depended on whether runs were compared over the same time or the same number of cycles (widget below). Over the same 1530 days, longer stages left a higher water production rate at the end of withdrawal (40.7, 45.2 and 52.1 m³/day for 60, 90 and 120 day stages), but total water varied only from 29,010 to 29,420 m³. Over ten cycles, total water rose from 25,470 to 34,430 to 42,050 m³. Larger working gas volumes also produced a wider range of well pressure within each cycle.

We also modelled a case after water breakthrough, when rising water has reached the well. Over five cycles, each of four strategies prevented further increases in water production. Injecting 10% more hydrogen than was withdrawn in each cycle (our arithmetic) was the most effective.

Limitations

Our calibration is uncertain, because the field has few wells and partial production records. We set the minimum well pressure to 8 MPa (megapascals), below the 16 MPa permitted at Ahuroa, to test water invasion. Our reservoir is uniform, and we did not consider well configuration. Our main runs neglected hysteresis in relative permeability. Neglecting it can lead to underestimated water production (Bahrami et al., 2023). Future work could test these results with measured relative permeability for mixed gases and with hysteresis included.

Glossary

Cushion gas. Gas left in the reservoir throughout storage to maintain reservoir pressure.

Working gas. The gas injected and withdrawn in each storage cycle.

Gas-water contact. The depth below which the pores of the reservoir rock are filled with water.

Water invasion. The rise of the gas-water interface beneath a well as gas is withdrawn, also called water coning.

Water breakthrough. The point at which rising water reaches a well and starts to be produced with the gas.

Compressibility. How much a gas's volume changes when its pressure changes.

Sm³. A standard cubic metre, the volume a gas would occupy at surface temperature and pressure.

Hydrogen recovery. The ratio of the hydrogen produced to the hydrogen injected.

Relative permeability. How easily gas or water flows through the rock when both are present, compared with when it is the only fluid.

Relative permeability hysteresis. The tendency of gas and water to flow differently when gas is entering the rock than when it is leaving.