Numerical modelling of H 2 storage with cushion gas of CO 2 in subsurface porous media: Filter effects of CO 2 solubility

(cid:1) An extra 30% volume of CO 2 is required when considering CO 2 solubility. (cid:1) Back produced H 2 purity depends strongly on the hydrodynamics of H 2 and CO 2 . (cid:1) “ Filter ” effects due to CO 2 solubility are identiﬁed for the ﬁrst time. (cid:1) CO 2 e H 2 mixing zone is expanded due to CO 2 solubility in gravity-stable regimes. (cid:1) Best H 2 recovery performance occurs in the gravity-dominated scenario.


Introduction
The target of "Net Zero" has been set out to limit the increase in global temperature below 1.5 C compared with the preindustrial level [1].The term "Net Zero" means completely negating the amount of greenhouse gases produced by human activities to mitigate climate change [2,3].Clearly, this goal fundamentally relies on the energy transition from fossil fuels to renewable energy technologies such as solar and wind energy [4].
Currently, the application of renewable energy technology is restricted by their nature of intermittency and uncertainty, which puts energy security and reliability at risk [5].Therefore, appropriate energy storage technologies must be developed to progress the deployment of renewable energies.One of the viable solutions is the use of hydrogen (H 2 ) as an energy storage vector [6,7].H 2 has very high energy density (33.3 kWh/kg) and produces only water when combusted [8,9].The electricity produced from renewable energies can be used to split water to H 2 (and oxygen) via electrolysis, i.e., the concept "power-to-gas" [10].As a clean fuel, H 2 is then combusted to provide energy/heat when the demand increases [11,12].Such flexibility can secure the energy supply and unlock the potential of renewable energies.However, the success of this process requires large storage capacity for the significant amounts of H 2 involved at grid scale [13,14].This is because H 2 has a very low mass density and is difficult to compress and liquefy [15].
Underground gas storage (UGS), which potentially has tremendous storage capacity, is increasingly considered as a necessary part of the supply chain for H 2 energy [16,17].Typical subsurface storage can be categorized into 3 types: salt caverns, depleted hydrocarbon reservoirs and saline aquifers.In fact, H 2 storage in salt caverns has been successfully applied in the past, primarily for use in oil refining and ammonia production [18].Due to the advantage of high operational efficiency, some researchers proposed to progress H 2 storage in salt caverns for energy transition [19,20].However, this option is subject to the sufficient thickness of salt deposits and therefore its availability is much limited [21e23].On the other hand, saline aquifers and depleted gas reservoirs, which have wider availabilities and larger storage capacities, are considered more suitable for decarbonizing the energy sector at scale [24].In this research, our study is based on a synthetic aquifer.We will test the feasibility of H 2 storage in depleted reservoirs in future work.
Gas storage in subsurface geological formations is not a new technology in the energy industry [25,26].Cyclic injections and withdrawals of natural gas (working gas) were widely applied to smooth gas supplies to meet daily/seasonal fluctuations of energy demand [27].In addition to the working gas, a certain amount of cushion gas (usually the same as the working gas), may need to be injected in advance to provide pressure support for the back-production of working gas.This cushion gas is not extracted and remains in the subsurface formations.Unlike natural gas storage, H 2 generation, such as via electrolysis, still requires very high capital and operation expenses [28,29].Hence, a different type of cushion gas may be a more cost-effective operational strategy for H 2 storage [30].For this reason, this study tests the idea of using CO 2 as cushion gas, which provides an extra benefit of directly reducing CO 2 emissions.
Recently, a range of techno-economic case studies were conducted to examine the feasibility of H 2 storage based on specific fields [31e33].These field-scale studies identified the practical and economic factors determining the viability of H 2 storage.However, very limited studies can be found in literature focusing on fundamental understandings of the hydrodynamic flow behaviour of H 2 with CO 2 as cushion gas in the subsurface porous media.Much work remains to be done to identify and understand the key factors influencing the migration and distribution of fluids [34,35].
The flow behaviour in subsurface porous media is controlled by the interactions between the fluid characteristics, medium properties and various driving forces [36].Depending on the relative contributions of these factors, the flow regimes can be categorized according to scaling theory [37e39].For example, in viscous-dominated flow the displacement process is dominantly controlled by the viscous instability whereas in the gravity-dominated regime, strong gravity segregation may be observed [40].
In a previous study [41], the key flowing features occurring in viscous-dominated to gravity-dominated scenarios have been systematically demonstrated in the context of H 2 storage.Scaling theory was applied to assess the various flow regimes in terms of the force balances (viscous/gravity mainly) in systems of different sizes.The effects of the flow regimes and the related hydrodynamics on the final purity of the back produced H 2 was studied.In this paper, we extend this earlier work by determining the impact of CO 2 solubility on the purity of the produced H 2 from the storage reservoir.CO 2 solubility is in fact one of the key trapping mechanisms for underground CO 2 storage [42e44].We demonstrate how the interactions between mechanisms of CO 2 dissolution, reservoir heterogeneity and multiphase flow behaviour, affect the H 2 recovery performance at specific levels of impurity.A range of very fine scale simulations, which are essential and necessary to capture the fluid behaviour in detail, are conducted.We present two extreme scenarios, namely the viscous-dominated and gravity-dominated flow regimes, with an aim of providing insight to optimize the operation strategy to improve the H 2 recovery performance.The outcome of this study together with our previous work, should also be useful to determine the required experimental work for projects of H 2 storage.The complete dataset of this numerical study can be found at https://doi.org/10.17861/615bdf45-a92b-4ff4-8f9e-efa6608ed030.

Methodology
The central purpose of this study is to conduct a detailed sensitivity analysis to identify the significant factors driving the flow behaviour of H 2 (working gas) and CO 2 (cushion gas), and thus the recovery performance of the H 2 and its purity.To achieve this, we performed a series of flow simulations in a 2D vertical system, using a fully compositional reservoir simulator-CMG/GEM [45].The modelling methods are mostly inherited from our previous work [41].All the simulation configurations can be found in the data repository with the link given above.Here, we only outline the main input to which we refer later in this article when explaining our results.

Fluid model
Table 1 lists the modelled fluid properties based on the Peng-Robinson Equation of State (EOS) under initial reservoir conditions.The molecular diffusion was neglected as this study is based on a highly heterogeneous system, where convective dispersion dominates the flows.Note that the CO 2 is injected at supercritical conditions, but is referred to as "gas" in this paper.

Synthetic geo-model
As shown in Fig. 1, a correlated random permeability field (i.e.CRF) was created using Petrel [48] to mimic reservoir heterogeneity.Other realizations with the fixed geostatistical setting were also tested, and they essentially have very similar reservoir response.Therefore, we select this model as a representative case to demonstrate the flow behaviour of H 2 and CO 2 , driven by various factors including viscous instability, gravity force, CO 2 solubility etc.

Operation strategy
From our previous work [41], we have identified that the two key end member flow regimes encountered are, viscousdominated and gravity-dominated flow regimes, although the flows can be in the transitional region where these forces are balanced.In this work, we extend that study to analyse how the CO 2 solubility influences the flow behaviour and thus the H 2 recovery in both scenarios; in particular, we focus on the issue of back produced H 2 purity.Operational strategies, which trigger the viscous-dominated and gravitydominated flow regimes respectively, are outlined in Table 2 and Table 3.The operation rates in the viscous-dominated case are 100 times greater than that in the gravity-Table 1 e Fluid modelling using PR-EOS under initial reservoir conditions.

Equation of State (EOS)
Cubic PR-EOS [46] with data by Ref. [47] Initial reservoir conditions T ¼ 51.9 C; P ¼ 15,000 kPa i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 7 ( 2 0 2 2 ) 2 8 9 5 6 e2 8 9 6 8 dominated case.CO 2 is injected as a cushion gas first and then followed by the injection and production of H 2 .The injected volume of CO 2 is twice as much by volume as H 2 under reservoir conditions.Both the injection and the production wells were horizontally perforated across the top and controlled by the flow rate at reservoir conditions.We assumed that the fracture pressure of the formation was never exceeded during the operation.The bottom of the model was set to mimic a large aquifer that allows fluid flow.
Viscous-dominated scenario: Gravity-dominated scenario: Relative permeabilities and capillary pressure In the system configured here, the period controlled by the two-phase flow mechanisms is primarily occurring at the displacing front, i.e., between CO 2 and water in our case.Fig. 2 shows the relative permeability curves for CO 2 /brine, which are utilized here for modelling two-phase flow based on the experimental data produced by Ref. [49].Capillary pressure is modelled using a synthetic Leverett J-function created in our previous work [41].As the processes of gas injection and production involve strong flow reversals, the issue of gas trapping may arise and therefore is considered here.The Land constant (C) is set to 2.44 and this leads to a maximum trapped gas saturation (S grmax ) of 0.2 [50,51].

Solubility model
The solubility of gas in the aqueous phase is modelled based on Henry's Law (Equation ( 1)).The variation of Henry's law constant with respect to pressure and temperature is modelled by Equation ( 2) [52e54].The effect of salinity is not included to simply the process.Note that the modelling method of Henry's law is assumed that the gaseous phase and the aqueous phase are in thermodynamic equilibrium.In practice, non-equilibrium CO 2 dissolution may occur, depending on the interface area between CO 2 and water and the operation timescale [55].The grid block residence times of the fluids are more than adequate for equilibrium to be established in the gravity-stable case, but there may be some deviation from equilibrium in the viscous dominated case.However, this is beyond the scope of this study and a detailed analysis of possible nonequilibrium effects may be required in the future.Henry's law where f iw is the fugacity of the component i (CO 2 here) in the aqueous phase; y iw denotes the mole fraction of component i in the aqueous phase; H i is Henry's Law constant; H i 0 is the reference Henry's Law constant; V ∞ is the partial molar volume of component i in the aqueous at infinite dilution [56]; P is its partial pressure in the gas phase under equilibrium conditions.P i 0 is the reference pressure; R is gas constant; T is temperature.According to Ref. [57], H 0 i is set to 3.3 Â 10 À4 mol/ m 3  $Pa at a reference temperature of 298.15K under atmospheric pressure.We also compared our simulated results with the data in literature and the accuracy is acceptable, as shown in Fig. 3 [58].
On the other hand, hydrogen solubility in water is much lower than CO 2 and therefore is ignored here [57].In fact, we did conduct numerical tests with H 2 solubility activated, in which negligibly small changes in flow behaviour were observed.However, this does not exclude the necessity of taking H 2 solubility into account for H 2 storage in real systems.This is because even small amount of dissolved H 2 may potentially trigger biological and geo-chemical reactions, such as methanation and sulfur-reducing processes [59].In a future publication, we will address how these reactions influence the recovery performance of H 2 and their potential impacts on the flow assurance.

Pseudo-tracer tests
A pseudo tracer analysis was conducted to uncover some of the flow mechanics due to CO 2 solubility in the viscousdominated scenarios.These tests entail injecting a certain amount of CO 2 tracer, which had the exact same EOS data and solubility as CO 2 .All the other simulation parameters were the same as the corresponding ones in previous regular simulations.The details of the tracer tests are outlined as below [60].

Injection location
The CO 2 tracer was injected into a grid block in one of the areas bypassed by H 2 .These zones, which are occupied by CO 2 but bypassed by H 2 , can be identified in the preliminary flow simulations.

Injection timing
To minimize the impacts of the tracer on the overall flow behaviour, the tracer injection was not conducted at the beginning of the simulation.Instead, the CO 2 tracer was injected slightly before the end of the H 2 injection, by which time the major bypassed zones have already formed.

Injection amount
A very small amount (almost negligible) of tracer was injected into the chosen bypassed grid block at a rate of 1.04e-8 PV/min for 15 min (1.56e-7PV in total).We found that such a slow and small amount of injection has negligible impacts on the overall flow behaviour.However, the flow behaviour of the CO 2 tracer can be used to represent the actual flow trajectory of the CO 2 in bypassed areas during back production.

Results and discussion
A series of simulations were conducted and analysed to understand the impacts of CO 2 solubility on the flow behaviour.
To clarify the purpose of each step of the analysis and the logic of this research, the results are presented in the following three steps (#1 to #3) outlined in Table 4.Note that the CO 2 solubility in water must lead to a reduction in the gas volume of CO 2 occupying in the pore space under reservoir conditions.
To make the simulations with and without CO 2 solubility consistently comparable, the total amount of CO 2 injection in Cases with CO 2 solubility activated is increased to compensate for its dissolution in water.Due to the non-uniform gas distribution and pressure variations throughout the system, the total amount of the dissolved CO 2 cannot be determined a priori and therefore a series of trial-and-error tests was conducted.It turns out that an approximately 30% extra volume of CO 2 is required to achieve the same distance travelled by the gas displacing front at the end of gas injection.As seen in Fig. 4 and Fig. 10, the interface between gas and water (G/W) is indicated by two red arrows.

Impacts of CO 2 solubility in viscous-dominated flow scenarios
Results presented in this section are from the step #1 of the simulations listed in Table 4.In the injection sequence, there are two interfaces of interest, namely gas (CO 2 )/water and CO 2 /H 2 , when investigating the hydrodynamic flow  To investigate the impacts of CO 2 solubility on H 2 recovery in the gravity-dominated flow regime behaviour of H 2 and CO 2 in an aquifer.Fig. 4 is plotted to show the gas saturation (left side) and H 2 gas mole fraction (right side) of the viscous-dominated scenarios with/without CO 2 solubility activated.Within the system configured here, the gas displacing front is relatively stable without forming severe flowing fingers of gas in both cases (A&B).The reasons are twofold.The first is the poor gas relative permeability when the gas saturation is low, which self-stabilizes the interface between gas and water (see Fig. 2).The second is the correlation range (1 m Â 1 m) of the permeability field is much lower than the system dimensions (10 m Â 80 m) and therefore no dominant gas fingers into water arise [37,61].In fact, the relatively stable interface between CO 2 and water guarantees no early water breakthrough during back production in the system configured here.This enables us to focus on the mixing zone between CO 2 and H 2 , which is a potentially significant factor which may influence the H 2 purity during back production.
On the other hand, the interface between H 2 and CO 2 , shows evident non-uniformity, as the mole fraction of H 2 is not evenly distributed above the CO 2 cushion gas (see the right part of Fig. 4).Within the underlying heterogenous permeability field, the less viscous H 2 does not proceed and displace the cushion gas of CO 2 in a piston-like fashion.It is not surprising that gas mixing occurs at the displacing front.More importantly, the cushion gas of CO 2 may be infiltrated and even bypassed by H 2 (indicated by the black ovals in Fig. 4A).Fig. 5 shows the H 2 purity in the back produced gas for Case A and B. Clearly, the bypassed CO 2 can be quickly co-produced, which leads to early reductions in H 2 purity during the back production.In particular, if certain high purity levels are required, then H 2 is poorly recovered.For example, the required purity levels of H 2 for fuel cell and combustion are 99.97% and 98%, respectively.From a practical perspective, the purification capacity of a processing terminal may vary significantly depending on the design of the processing system [62].For example, a typical capacity of CO 2 purification based on ammonia (at an extra cost) of a processing terminal can manage up to 10% impurity of CO 2 .Therefore, we tracked the H 2 recovery given the three critical threshold values of the H 2 purity mentioned above.As seen in Fig. 6, the H 2 recovery is only 9% (purity level >99.97%), 21% (purity level >98%) and 39% (purity level >90%) in Case A.
Surprisingly, although the flow behaviour of Case B (with CO 2 solubility) looks remarkably similar to Case A (see Fig. 4), the H 2 purity level during back production is improved when CO 2 solubility is included.The back produced H 2 purity profiles are compared for both cases in Fig. 5.As shown in Fig. 6, the H 2 recovery is increased by 2% at the end of production in Case B. More importantly, an extra of 4%, 9.1 and 9.2% of H 2  can be recovered, with a purity level above 99.97%(required for fuel cell) 98% (required for combustion) and 90% (typical processing capacity), respectively.

Tracer analysis
Clearly, the CO 2 solubility must lead to some flow mechanism that positively influences the H 2 recovery but cannot be easily observed from the snapshots of the gas saturations and component concentrations.To understand how the CO 2 solubility improves the H 2 recovery performance, we conducted a tracer analysis to track the flow trajectory of the CO 2 in the areas of flow bypassed during the H 2 injection phase.These are the step #2 simulations listed in Table 4.
As explained in the methodology section, we designed a tracer test, which entails injecting a tiny amount of tracer into a representative cell in the bypassed area.Fig. 7 (right side) is plotted to show the gas mole fraction of the CO 2 tracer at the end of injection.As seen in Fig. 7, we selected the one that is closest to the top producers.This is to ensure the impact of the bypassed CO 2 can be captured before the effects of the CO 2 eH 2 mixing zone are observed at the production well.Otherwise, it is difficult to differentiate the impacts of the gas mixing (H 2 front) and the gas bypassing on the H 2 purity.The flow trajectory of the tracer is observed and compared between cases with and without including the CO 2 solubility in the simulations.Fig. 8 is plotted to compare the gas mole fraction of the tracer between two cases.As seen in Fig. 8, after the same pore volume of back-production (0.0063 PV), the tracer has reached the producer at the top in the case without CO 2 solubility (Case C) whereas the tracer distribution is much restricted in the case with CO 2 solubility (Case D).In other words, CO 2 solubility has retarded the breakthrough of the cushion gas (CO 2 ), which was bypassed during the period of H 2 injection.
The question which now arises is: what is the mechanism by which CO 2 solubility in water acts to yield the results shown above?In other words, how does the CO 2 solubility in water delay the CO 2 production into the H 2 produced stream.Fig. 9 shows the CO 2 water mole fraction of representative cells in a preferential route and in a non-preferential route in the case with CO 2 solubility activated (Case B).The preferential routes are the areas where both H 2 and CO 2 can easily flow through during gas injection.The non-preferential routes mean that areas are dominantly swept by CO 2 but mostly bypassed by H 2 during injection.The representative cell (Cell 1) of the non-preferential route is the same grid block as the one selected for the tracer analysis.The representative cell (Cell 2) of the preferential route is located to the left of Cell 1 and it has the same vertical position.The horizontal distance between of Cell 1 and Cell 2 is 1 m (10 grid blocks).We demonstrate the fluid behaviour during each stage of the operation below.

CO 2 injection
As seen in Fig. 9, CO 2 contacts and dissolves into the water during the CO 2 injection (black curves) until the equilibrium is reached (approximately 0.024 CO 2 mole fraction in water).This phenomenon occurs both in the representative cells of preferential and non-preferential routes selected here.As mentioned above, the period of CO 2 injection is extended by 30% to compensate for its dissolution in water compared with the case without CO 2 solubility.

H 2 injection
When the working gas H 2 starts to be injected, CO 2 water mole fraction of the representative cell in the preferential route gradually decreases until no CO 2 is dissolved in the water (blue curves).This means that H 2 not only displaces the CO 2 in the gas phase but also vaporizes the dissolved CO 2 .On the other hand, since H 2 hardly flows into the non-preferential routes, the gas phase in these bypassed areas is still pre-dominantly consisting of CO 2 .For this reason, the process of vaporization is very limited and therefore CO 2 water mole fraction does not decrease as much during the process of H 2 injection.

Back production
Unlike the process of H 2 injection, the flow during back production is a viscous-stable displacement process (since m CO2 >m H2 ; m water >m CO2 ).Therefore, gas in both non-preferential routes and preferential routes starts to flow towards producers under the pressure gradient.When CO 2 solubility is not included, the CO 2 in the bypassed areas, especially those which are close to the producer, can flow into the preferential routes in neighbouring areas via viscous crossflow and then be produced.This leads to quick reduction in H 2 purity during the process of withdrawal.However, when CO 2 solubility is activated, the CO 2 contacts and redissolves into the water (now undersaturated with CO 2 ) in the preferential routes.In other words, CO 2 solubility enables the water surrounding the bypassed zone to act as a "filter" for CO 2 .This explains the very limited distribution of CO 2 in gas phase shown in our tracer analysis.Since the CO 2saturated water is still immobile, the CO 2 production is therefore delayed.Note that with increasing gas production, the upwardsflowing CO 2 gas can come out of solution and break through to the producer, but at a later time.Then, CO 2 starts to be produced and the H 2 purity level decreases (See Fig. 5).

Impacts of CO 2 solubility in gravity-dominated flow scenarios
In the previous section, we have demonstrated the impact of CO 2 solubility on the fluid behaviour and recovery performance (i.e., the level of H 2 purity) in the viscous-dominated flow regime.We now investigate the impacts of CO 2 solubility on the flow performance in the gravity-dominated scenario.This is referred to as the step #3 simulations described in Table 4.As seen in Fig. 10, there is no evident infiltration or bypassing of H 2 in the gravity-dominated cases.Due to the density difference between H 2 and CO 2 , H 2 is almost uniformly accumulated above the cushion gas of CO 2 in both Case E (CO 2 not soluble) and Case F (CO 2 soluble in water).However, the size of the mixing zone between CO 2 and H 2 is much expanded (as indicated by the black oval), when CO 2 solubility is included.As mentioned above, H 2 not only displaces the CO 2 in the gas phase but also gradually strips the dissolved CO 2 out of the water.The extra amount of the vaporised CO 2 mixes with the injected H 2 and therefore a larger mixing zone is formed.As seen in Fig. 11, this has led to a slight reduction in the H 2 purity at the medium stage of the back production.Depending on the requirements for H 2 purity (90%e99.97%),such slight degradation in purity can lead to reductions in H 2 recovery by 3% and 6%, with a purity level above 99.97% and 98%, respectively.However, if the requirement of the purity level is reduced to 90%, the negative impacts of the CO 2 solubility is rather limited and the H 2 recovery is only lowered by approximately 1% in our cases (see Fig. 12).
Very importantly, Fig. 13 is plotted to compare the recovery performance between the two end member scenarios, namely the viscous dominated and gravity dominated flow regimes.The two compared cases shown in Fig. 13 are both activated with the CO 2 solubility in water.It is found that the overall H 2 recovery performance is much better in the gravitydominated scenarios than that in the viscous-dominated cases.Compared with the viscous-dominated case (Case B), an extra 7%, 22% and 27% H 2 recovery can be achieved in the gravity-dominated case (Case F), given the three threshold purity levels (99.97%, 98% and 90%).

Conclusions
In our previous study [41], we have described and differentiated the role of various driving forces (viscous, gravity and capillary forces) during H 2 storage in subsurface porous media.Based on that work, we extend our research to analyse another important mechanism, viz., CO 2 solubility in water.With the use of very-fine scale full-compositional simulations (Dx ¼ 0.1 m, in a 10 m Â 80 m system), it is found that the CO 2 solubility can have either positive or negative impacts on the H 2 recovery performance, depending on the flow regime.Specifically, we have made the following four principal observations.1.In cases with CO 2 solubility, the volume of CO 2 injected at reservoir conditions needs to be increased by approximately 30% in our system, to produce the same gas displacing front, as occurs in the case with no dissolution.
2. In the viscous-dominated scenario, the issue of the dramatic reduction in H 2 purity is less severe, when CO 2 solubility is included.This is because the bypassed CO 2 will redissolve into water in neighbouring areas during back production, which retards the movement of CO 2 towards the producer, i.e., the so called "filter" effect for CO 2 .3. In the gravity-dominated scenario, the partitioning behaviour of CO 2 between gas and water has minor to moderate negative impacts on the H 2 recovery.H 2 not only displaces the CO 2 in gas phase but also gradually strips the dissolved CO 2 out of water.For this reason, the size of mixing zone of CO 2 and H 2 is enlarged, which leads to reductions in H 2 purity during back production.4. The overall H 2 recovery performance is much better in the gravity-dominated scenarios than that in the viscousdominated cases.This is because, in the viscousdominated scenario H 2 may infiltrate and bypass the CO 2 and lead to early and dramatic reductions in H 2 purity Fig. 12 e Comparison of H 2 recovery between Case E&F at the end of 0.075 PV production, H 2 purity above 99.97%,98% and 90%, respectively.
Fig. 13 e Comparison of H 2 recovery between Case B&F at the end of 0.075 PV production, H 2 purity above 99.97%,98% and 90%, respectively.
during back production.Compared with the viscousdominated case, an extra 7%, 22% and 27% H 2 recovery can be achieved in the gravity-dominated case, given the three important threshold purity levels (99.97%, 98% and 90%).

Fig. 1 e
Fig.1e The correlated heterogeneous permeability field used in this work.

Fig. 3 e
Fig. 3 e Comparison between simulation results of CO 2 solubility with data by Ref. [58].

Fig. 4 e
Fig. 4 e Gas saturations and mole fractions of H 2 at the end of each cycle of the base model without CO 2 solubility (Case A at the top) and the case with CO 2 solubility (Case B at the bottom).

Fig. 5 e
Fig. 5 e The purity of back-produced H 2 of the viscousdominated case (A) and the corresponding case with CO 2 solubility activated (B).

Fig. 8 e
Fig. 8 e Snapshots of tracer distribution in the viscous-dominated cases with (Case D on the right) and without CO 2 solubility (Case C on the left) at the end of tracer injection.

Fig. 9 e
Fig. 9 e CO 2 water mole fraction of representative cells: Cell 1 in a preferential route (left) and Cell 2 in a non-preferential route (right), respectively.

Fig. 10 e
Fig. 10 e Gas saturations and mole fractions of H 2 at the end of each cycle of the base model without CO 2 solubility (Case E at the top) and the case with CO 2 solubility (Case F at the bottom).

Fig. 11 e
Fig.11e The purity of back-produced H 2 of the gravity-dominated case (E) and the corresponding case with CO 2 solubility activated (F).

Table 2 e
Operational strategy for viscous-dominated scenario.

Table 3 e
Operational strategy for gravity-dominated scenario.

Table 4 e
Outline of numerical tests performed and objectives of each stage for this study.
C. Tracer test based on the Case A D. Tracer test based on the Case B To demonstrate the "filter effects" resulting from CO 2 solubility #3 E. Gravity-dominated case F. Gravity-dominated case with CO 2 solubility