Plume Migration of Different Carbon Dioxide Phases During Geological Storage in Deep Saline Aquifers

This study estimates the plume migration of mobile supercritical phase (flowing), aqueous phase (dissolved), and ionic phase CO2 (bicarbonate), and evaluates the spatial distribution of immobile supercritical phase (residual) and mineral phase CO2 (carbonates) when CO2 was sequestered. This utilized a simulation, in an anticline structure of a deep saline aquifer in the Tiechenshan (TCS) field, Taiwan. All of the trapping mechanisms and different CO2 phases were studied using the fully coupled geochemical equation-of-state GEM compositional simulator. The mobile supercritical phase CO2 moved upward and then accumulated in the up-dip of the structure because of buoyancy. A large amount of immobile supercritical phase CO2 was formed at the rear of the moving plume where the imbibition process prevailed. Both the aqueous and ionic phase CO2 finally accumulated in the down-dip of the structure because of convection. The plume volume of aqueous phase CO2 was larger than that of the supercritical phase CO2, because the convection process increased vertical sweep efficiency. The up-dip of the structure was not the major location for mineralization, which is different from mobile supercritical phase CO2 accumulation.


INTRODUCTION
Storing carbon dioxide (CO 2 ) geologically is a technology that benefits from extensive petroleum engineering experience from dealing with hydrocarbon production and storing natural gas (IPCC 2005;IEA 2008).Suitable formations for CO 2 storage include depleted oil and gas reservoirs, coal seams, and saline aquifers.Compared with other geological media, deep saline aquifers have the largest storage capacity and can be found in most of the world's sedimentary basins.This is an advantage in terms of CO 2 transport from emission sources (IPCC 2005;Bachu 2008).
Several trapping mechanisms act to prevent the buoyant CO 2 from coming back to the atmosphere when CO 2 is stored in a saline aquifer.These mechanisms include stratigraphic and structural, residual gas, solubility, ionic, and mineral trappings (Pruess et al. 2003;Nghiem et al. 2004Nghiem et al. , 2009a, b;, b;Kumar et al. 2005;Bachu 2008).Corresponding-ly, CO 2 mobile and immobile supercritical, aqueous, ionic, and mineral phases exist simultaneously in the aquifer.
When stored in a saline aquifer, CO 2 is always less dense and less viscous than the groundwater, making it buoyant and mobile in the aquifer (Bachu 2003).The mobile supercritical CO 2 plume will migrate upward, under the impermeable cap rock (Bachu 2008).Over time, other and more secure trapping mechanisms take over.In residual gas trapping, the CO 2 is trapped in the rock pore space by capillary pressure (Flett et al. 2004;Bennion and Bachu 2005;Kumar et al. 2005;Juanes et al. 2006;Ide et al. 2007;Bachu 2008;Nghiem et al. 2009a, b, c;Qi et al. 2009).There is no "plume migration" in immobile supercritical phase (residual) CO 2 because it is trapped and fixed in the rock pores.The "spatial distribution" of immobile supercritical phase (residual) CO 2 is investigated in this study.
Much of the injected CO 2 will eventually dissolve into the saline water.This process, which further traps the CO 2 , is called solubility trapping (Kumar et al. 2005;Bachu 2008;Nghiem et al. 2009a, b, c).The plume migration of the dissolved CO 2 (or aqueous phase CO 2 ) is different from that of mobile supercritical CO 2 , because solubility trapping forms a denser fluid and causes a concentration difference in the saline water, which may then sink to the bottom of the formation (Ennis-King and Paterson 2005).
The dissolved CO 2 reacts with formation water and then dissociates into bicarbonate (HCO 3 ) and carbonate (CO 3 2 ) ions.This chemical reaction is the ionic trapping mechanism (Gunter et al. 2004;Kumar et al. 2005;Bachu 2008;Nghiem et al. 2009a, b).The dissociated bicarbonate ions are referred to as ionic phase CO 2 in this study.The plume migration might be similar to that of aqueous CO 2 because the ionic phase CO 2 is formed from the existing aqueous phase CO 2 .
Depending on the rock formation, the ionic CO 2 may react chemically with the surrounding rocks to form stable minerals.Known as mineral trapping, this process can store CO 2 permanently and provides the most secure form of storage for the CO 2 (Bachu et al. 1994;Gunter et al. 1997Gunter et al. , 2004Gunter et al. , 2008;;Nghiem et al. 2004Nghiem et al. , 2009a, b;, b;Rochelle et al. 2004;Xu et al. 2004;Gaus et al. 2005;Thibeau et al. 2007;Bachu 2008).Similar to residual CO 2 , there is no plume migration in the mineral phase CO 2 (i.e., carbonates) because it is precipitated or bound to the rock matrix.The "spatial distribution" of the mineral (solid) phase CO 2 is investigated here.
The plume migration and the spatial distribution of different of CO 2 phases provide essential information for monitoring, risk assessment, and management issues when a CO 2 storage project is developed.The U.S. Environmental Protection Agency (EPA) Underground Injection Control Program requires that the permit applicant define an Area of Review (AOR) in which all penetrations intersecting the injection formation or its confining layer be identified and be determined to have been properly plugged and abandoned when a proposed injection operation has the potential for contaminating underground sources of drinking water through wells, faults or other pathways that penetrate an injection zone (Nicot et al. 2008;EPA 2014).The AOR for a CO 2 storage project should be defined and investigated before CO 2 is injected into a reservoir.The AOR is traditionally evaluated from the plume migration of mobile supercritical CO 2 phase.The integrated AOR evaluation should include all CO 2 phase because different CO 2 phases have different migration behaviors.
A CO 2 storage project is currently preparing to launch in Taiwan.An onshore saline aquifer of the Yutengping sandstone formation in the Tiechenshan (TCS) field in northwestern Taiwan is a potential site for this project.This study evaluates the AOR from the plume migration of mobile supercritical, aqueous, and ionic CO 2 , and from the spatial distribution of the immobile supercritical and mineral phase CO 2 when CO 2 is stored in the anticline structure of a deep saline aquifer in the TCS field.

Structural Trapping
The critical pressure and critical temperature of CO 2 are 7376 kPa and 304.2K (31°C), respectively.When CO 2 is injected into an aquifer deeper than 800 m, it is in a supercritical state (Bachu 2003).The density of the injected supercritical CO 2 is approximately 5160 kg m -3 (or 32.9 lb ft -3 ) at reservoir conditions of 16858 kPa (or 2445 psi) and 72.2°C (or 162°F), which is lower than that of saline formation water (10039 kg m -3 or 63.9 lb ft -3 at the same reservoir conditions).The flow behavior of supercritical CO 2 in an aquifer is similar to that of a fluid (Nghiem et al. 2009a, b).Buoyancy drives the injected CO 2 to migrate upward until an impermeable cap rock traps it.Thus, the structural trapping mechanism needs a cap rock to prevent mobile CO 2 from leaking out of the storage reservoir (Bachu 2008).

Residual Gas Trapping
Residual gas trapping is the one of the important processes for trapping CO 2 .The mechanism converts CO 2 into an immobile phase in the pores via the capillary effect and imbibition.The imbibition usually occurs at the rear of the plume of supercritical CO 2 after the injection stops.The classic Land's model (Land 1968) (Fig. 1) was used in this study to calculate the residual gas (CO 2 ) saturation [Eqs.(1) and (2)], as follows (Nghiem et al. 2009a occurs.C = Land's coefficient, S g, max = the maximum gas saturation, S gt, max = the maximum residual gas saturation.When the gas saturation S * gi reverses and decreases at drainage curve, the gas relative permeability follows the imbibition curve k rg i (red curve).

Solubility Trapping
Solubility trapping causes both mobile and immobile supercritical CO 2 to change into an aqueous phase via the dissolution process.Denser CO 2 -saturated water will be formed and it will tend to sink to the bottom of the formation.The convection effect will force the fresh water to replace the CO 2 -saturated water.Consequently, more supercritical CO 2 dissolves into the water.
CO 2 solubility in brine can be modeled as a phase-equilibrium process.To calculate the quantity of aqueous CO 2 , the equality of fugacities in the gas and aqueous phase used is as follows (Nghiem et al. 2009a, b): In this study, the fugacity of CO 2 in the gas phase ( f , g CO2 ) was calculated using the Peng-Robinson equation-of-state (EOS) (Peng and Robinson 1976), and the fugacity of CO 2 in the aqueous phase ( f , aq

CO2
) was modeled with Henry's law (Li and Nghiem 1986), as follows: where y , aq

CO2
= the mole fraction of CO 2 in the aqueous phase and H CO2 = Henry's constant of CO 2 , which is a function of pressure, temperature and salinity.

Ionic Trapping
The H + and HCO 3 or CO 3 2 ions are dissociated when the injected CO 2 dissolves in water.The main chemical reactions related to CO 2 sequestration are as follows: where CO 2(aq) = the CO 2 that is dissolved in the aqueous phase (from solubility trapping).
Chemical equilibrium reactions were used in this study to model this fast and reversible intra-aqueous chemical reaction (ionic trapping mechanism) (Nghiem et al. 2009a, b).The chemical equilibrium reactions were governed by chemical equilibrium constants (Bethke 1996;Nghiem et al. 2009a, b), as follows: where R aq = the number of intra-aqueous chemical equilibrium reactions, K , eq a = the chemical equilibrium constant for the aqueous reaction a , and Q a = the activity product for the aqueous reaction a .
The values of K , eq a for several aqueous reactions were studied by Kharaka et al. (1988) and Delaney and Lundeen (1990).The activity product (Q a ) is calculated by (Nghiem et al. 2009a, b): where n aq = the number of aqueous components, a k = the activity of component k, and v , k a = the stoichiometry coefficients of the chemical equilibrium reactions.
The activities a k are the product of the molality (m k , moles per kg of H 2 O) and the activity coefficient ( k c ) of component k.An efficient model for calculating the ionic activity coefficients is the B-dot model for the non-ideal solution (Bethke 1996) or the Pitzer (1987) model for a highsalinity solution (Nghiem et al. 2009a, b).

Mineral Trapping
The ions that are dissociated through the chemical equilibrium reaction will react with the minerals in place, and with other ions in the solution, and will lead to the precipitation of carbonate minerals or the dissolution of formation minerals.The typical geochemical reaction for the precipitation or dissolution of calcite (CaCO 3 ) is: Geochemical reactions occur between minerals and aqueous components, and are reversible.The dissolution or precipitation of minerals follows the reaction rate (r b ), given by (Bethke 1996;Nghiem et al. 2009a, b): where r b = the reaction rate for a given mineral b , R mn = the number of mineral reactions, A b t = the reactive surface area, k b = the rate constant of the mineral reaction, K , eq b = the chemical equilibrium constant of the mineral reaction, and Q b = the activity product of the mineral reaction.
The ratio of the activity product to the chemical equilibrium constant ( Q K , eq b b ) is the saturation index (SI) of the reaction and is used to evaluate the dissolution or precipitation path (Nghiem et al. 2009a, b).The changes in the moles of minerals through dissolution or precipitation can be estimated after the geochemical reaction occurs.

GEOLOGICAL SETTING AND ENGINEERING DATA
This case study area is located in the Taihsi basin, in northwest Taiwan.The Taihsi basin is situated near the Taiwan mountain ranges with the eastern boundary of the deformation front (Fig. 2).To the North and South, the Taihsi basin is bounded by the Kuanyin uplift and Penghu platform, respectively (Fig. 2).The west of the Taihsi basin connects with the Taiwan Strait shelf.The thickness of the Taihsi basin is about 8 km.The formation dip angle is about 4° from west to east, which is caused by the orogenic load of the Taiwan mountain ranges (Lin 2001;Lin et al. 2003).
The Cenozoic sediments of the Taihsi basin, which have the highest capacity for CO 2 storage in Taiwan (Lin 2007), can be divided by three evolution episodes: the Paleocene-Eocene syn-rift, the Oligocene-Miocene post-breakup, and the Latest Miocene-Recent foreland basin (Lin et al. 2003).The sedimentary formations in the Oligocene-Miocene postbreakup period, from bottom to top, are the Wuchihshan, Mushan, Piling, Shihti, and Nankang formations, which are the major gas reservoirs and hydrocarbon source rocks in Taiwan (Fig. 3).The sedimentary formations in the latest Miocene-Recent foreland basin are the Kueichulin, Chinshui, Cholan, and Toukoshan formations, which are saline aquifers (Fig. 3).
The Kueichulin formation was selected for the CO 2 storage project in the Taihsi basin, because its depth interval is suitable for the storage of CO 2 in its supercritical state (Bachu 2003;Lin 2007).The Kueichulin formation is divided into three parts: Yutengping sandstone, Shihliufen shale, and Kuantaoshan sandstone (Fig. 3).The Yutengping sandstone is the top layer and is overlain by the impermeable Chinshui shale (Fig. 3).The thickness of the Chinshui shale is approximately 300 m and its permeability is extremely low, less than 10 -4 md.The Chinshui shale is the caprock and will prevent CO 2 leakage after CO 2 has been injected into the Yutengping sandstone for storage.The Yutenping sandstone lies on top of the Shihliufen shale, which is treated as the impermeable lower boundary of the CO 2 storage formation.
The selected potential storage site is an anticline trap with a closure depth of 1600 m (Fig. 4).The formation parameters of Yutengping sandstone (Table 1) were collected from the available drilling reports, core analyses, and well logging interpretations from Taiwan's CPC Corporation.
Formation water samples from the Yutengping sandstone were laboratory analyzed and their composition Table 2 was corrected using a geochemical aqueous equilibrium model, Solmineq.88(Kharaka et al. 1988;GAMS 2011b).XRF (X-ray fluorescence) and XRD (X-ray diffraction) were used to analyze the volume percentage of rock minerals (Table 3) was analyzed form a formation rock sample.
A geochemical reaction path model, GAMSPath (GAMS 2011a), was used to analyze the intra-aqueous chemical reactions and geochemical reaction paths of rockbrine-CO 2 in a preliminary theoretical study, in which the injection of CO 2 into brine was simulated.The aqueous phase CO 2 [CO 2(aq) ] was formed and the bicarbonate (HCO 3 ) ions were then dissociated in the solution.Consequently, the pH value of the solution was lowered.This caused variations in the kaolinite and muscovite early in the simulated period.A small amount of muscovite was dissolved and kaolinite was precipitated.However, anorthite was dissolved continuously in the low pH solution, which dissociated the calcium (Ca 2+ ) and aluminum (Al 3+ ) in the solution.The reaction between the bicarbonate (HCO 3 ) and calcium (Ca 2+ ) finally led to the precipitation of calcite (CaCO 3 ).Kaolinite, another secondary mineral, was precipitated because of the dissociated aluminum (Al 3+ ).Based on the geochemical reaction analysis result we concluded that there were ten rock-brine-CO 2 interactions: five intra-aqueous chemical reactions in the aqueous phase and five geochemical mineral reactions in the mineral phase (Table 4).

RESERVOIR SIMULATION MODEL CONSTRUCTION
The numerical simulation method was used in this study to estimate the plume migration of the supercritical, dissolved, and ionic CO 2 to evaluate the spatial distribution of the residually and mineralogically trapped CO 2 in a saline aquifer.The fully coupled geochemical equation-ofstate model (GEM) simulator was used with the GEM-GHG module, a commercial reservoir simulator developed by CMG (Computer Modelling Group Ltd.), which is capable of modeling all trapping mechanisms (Nghiem et al. 2009a, b;CMG 2011a).
The numerical geological model was constructed by dividing the structure into 6365 grids and assuming an open boundary on the edge grid.Each grid is uniform in size (about 228.6 × 228.6 × 41 m), and we divided the reservoir into 5 layers to illustrate the vertical migration path.The rock and fluid properties (formation parameters, fluid PVT data, and relative permeability curves) and formation initial conditions (initial formation pressure, reservoir temperature, formation water analysis data, and rock mineral compositions) were sequentially assigned to each grid block.The specific operation (by designing injection rates or injection pressures) and completion (perforation intervals) conditions were then used to create a model of a well for injecting CO 2 into the aquifer.
To simulate the residual gas trapping, the maximum gas saturation (S g, max ) and maximum residual gas saturation (S gt, max ) (Table 1) were used to calculate the Land's

Parameter (unit) Value
Formation Top (m) 1300 -1600 Reservoir Size (m 2 ) 6.65E7 coefficient (C) [Eq.( 2)] and to estimate the residual gas saturation (S * gt ) from the Land's model [Eq.( 1)].The hysteresis effect on the gas relative permeability was modeled.The solubility trapping mechanism was modeled from the phase-equilibrium process (Nghiem et al. 2009a, b).The Peng-Robinson EOS and Henry's law parameters, which were used to calculate the fugacities of CO 2 in the gas and aqueous phases, were derived from a phase property program: WinProp (CMG 2011b).
This case study was simulated to inject one million tons of CO 2 per year, to create a commercial site for CO 2 storage, with an injection period of 20 years, to create an oil development project life cycle, under the constraint of injection pressure lower than the fracture pressure, for the case of a vertical well at the down-dip of an anticline structure.The total simulation period for studying the plume migration and spatial distribution of the different phases of CO 2 at various times was 1000 years.

RESULTS
When CO 2 is injected into an aquifer, the plume of supercritical CO 2 tends to move upward because of the CO 2 buoyant force (Figs.5a, b).Consequently, the plume is stopped by the cap rock because the cap rock is impermeable.Supercritical CO 2 then flows under the cap rock, following the structure inclination and migrating to the structure up-dip (Fig. 5c).
In the post-injection period, the supercritical CO 2 continuously migrates to the structure up-dip and accumulates in the trap (Fig. 5d).Some supercritical CO 2 is trapped in the pores and becomes immobile at the rear of the moving plume.Solubility trapping also reduces the size of the supercritical CO 2 plume phase, especially in the area of the immobile supercritical CO 2 (Figs.5e, f).
During CO 2 injection period supercritical CO 2 is continuously injected into the aquifer to drain the formation water away from the wellbore.In this drainage process, there is no residual CO 2 (or immobile supercritical CO 2 ) formed (Figs.6a, b).At the end of the injection period, a few pockets of residual CO 2 are formed around the rear side of the plume (Fig. 6c).
In the post-injection period, supercritical CO 2 is no longer injected and imbibition occurs.Residually trapped CO 2 is formed behind the moving supercritical CO 2 plume (Fig. 6d).The amount of residual CO 2 reaches a maximum at the simulation time of 100 years (80 years after the end of CO 2 injection).The solubility trapping mechanism then works to reduce the immobile supercritical CO 2 volume because the trapped CO 2 dissolves into the water (Fig. 6e).At the simulation time of 1000 years, the greater part of the immobile supercritical CO 2 is dissolved into the formation water (Fig. 6f).
The simulation results show that the residual gas trap-ping mechanism is a fast and safe mechanism in the postinjection period because it fixes a great deal of mobile supercritical CO 2 and provides a good environment for the solubility trapping mechanism to operate.When mobile or immobile supercritical CO 2 contacts the formation water, CO 2 dissolves into the water and aqueous CO 2 forms.The simulation results show that the aqueous plume migration CO 2 (Figs.7a -c) is similar to that of supercritical CO 2 (Figs.5a -c) in a 20-year injection period.However, the aqueous CO 2 plume size is larger than that of supercritical CO 2 because of the convection effect.Because CO 2 -saturated water is heavier than the original formation water, it tends to sink to the bottom of the formation (Figs. 7a -c), while the supercritical CO 2 tends to move

Geochemical mineral reactions
Calcite H Ca HCO  aq  2   3 2 Table 4. Major intra-aqueous chemical reactions and geochemical mineral reactions considered in this study.upward to the top of the structure (Figs.5a -c).During the CO 2 injection period, the difference in migration behavior between the aqueous CO 2 (Fig. 7c) and supercritical CO 2 (Fig. 5c) plumes is easily seen, especially in the vicinity of the wellbore.
In the post-injection period, convection transfers a large amount of aqueous CO 2 to the bottom of the formation.The aqueous CO 2 finally accumulates in the down-dip of the structure (Figs.7d -f), where it is safe and distant from the cap rock.The plume migration of aqueous CO 2 shows that the risk for CO 2 leakage can be lowered when the solubility trapping mechanism operates continuously.
The simulation results show that the plume migration of ionic CO 2 (HCO 3 ) (Figs. 8a -f) is very similar to that of aqueous CO 2 (Figs.7a -f).This is because the ionic CO 2 is formed from the aqueous CO 2 and formation water (H 2 O) reaction.In other words, the aqueous CO 2 plume is the cradle for creating ionic phase CO 2 .During CO 2 injection the ionic CO 2 (HCO 3 ) (Figs. 8a -c) plume migration is similar to that of supercritical CO 2 (Figs.5a -c).The convection effect, which dominates the plume migration of aqueous CO 2 , also affects the plume migration of ionic CO 2 .In the postinjection period convection causes a large amount of aqueous CO 2 to sink to the bottom of the formation (Figs. 7d -f).Consequently, the ionic CO 2 (HCO 3 ) plume has the same flow behavior (Figs.8d -f).Similar to aqueous CO 2 , ionic phase CO 2 (HCO 3 ) tends to sink and accumulate in the down-dip of the structure.Ionic CO 2 is also distant from the cap rock.Based on the simulation results the ionic trapping mechanism changes the phase of CO 2 into a safer ionic phase and generates flow behavior that is low-leakage-risk and allows safe storage.
The precipitation of carbonates is because of the reaction between the ionic CO 2 (HCO 3 ) and rock minerals.In this case study, anorthite dissolved in the low pH environment yielded calcium (Ca 2+ ).Calcite (CaCO 3 ) was precipitated from the reaction between the bicarbonate (HCO 3 ) and the calcium (Ca 2+ ).Calcite (CaCO 3 ) is the major mineral phase CO 2 in this case study.Our simulation results show that the pH value of the formation water was lowered after CO 2 was injected into the formation, and that the variations in the kaolinite and muscovite were caused first.A small amount of muscovite was dissolved and kaolinite was precipitated.The simulation results were identical to the observations from the preliminary GAMSPath theoretical study (GAMS 2011a).
The simulation results show that there is no calcite precipitation during the 20-year injection period (Figs.9a -c).Moreover, the precipitation of kaolinite and the dissolution of muscovite did not substantially affect the formation porosity.Based on the simulation results, there were almost no changes in formation porosity during the injection period.
In the post-injection period calcite precipitation was the predominant reaction during mineralization.After 100 years a small amount of calcite precipitated into the formation pores where the ionic CO 2 plume had formed (Fig. 9d).The ionic CO 2 plume was the source of the calcite precipitation (Figs.8e, 9e).The up-dip of the structure was not the major location of mineralization (Fig. 9f) because the source (bicarbonate) of the mineralization tended to sink to the bottom of the structure.Moreover, the dynamic changes in the formation porosity were related to the mineral CO 2 (CaCO 3 ) spatial distribution.At the end of the simulation (1000 years), the formation porosity had decreased by 0.2% near the wellbore.The wellbore vicinity is where most damage to the formation porosity occurred.

DISCUSSION
The flow path and velocity of supercritical CO 2 is essential information for evaluating the AOR and for monitoring and safety in a CO 2 storage project.In this study, the flow path of the supercritical CO 2 was traced and the velocity of the supercritical CO 2 was estimated from the plume front moving distance of supercritical CO 2 .
The plume of supercritical CO 2 first moved northwest, turned north-northwest, and accumulated in the up-dip of the structure (Fig. 10).By investigating the plume area at the end of the injection period (Fig. 10), we saw that the average the supercritical CO 2 plume velocity was about 320 m year -1 (1050 ft year -1 ) or about 0.88 m day -1 (2.9 ft day -1 ) during the 20-year injection period.
In the post-injection period the supercritical CO 2 velocity slowed because CO 2 was no longer being injected.The trap of the structure restricted supercritical CO 2 plume extension.Based on our results, the area size of the supercritical CO 2 plume was the largest at the simulation time of 60 years (40 years after the end of injection) (Figs.11a, b).The area size of the supercritical CO 2 was about 13.9 km 2 (Fig. 11a) and the volume of the supercritical CO 2 was about 0.45 km 3 (Fig. 11b).The plume volume was relatively small.The vertical sweep efficiency affected the plume volume because the supercritical CO 2 always flowed on the top of the formation because of its buoyancy (Fig. 11b).
The size of the area and the volume of the aqueous CO 2 were also investigated and compared.The area size of the aqueous CO 2 was about 16.7 km 2 (Fig. 11c) and the aqueous CO 2 plume volume was about 2.55 km 3 (Fig. 11d).At the simulation time of 60 years the size of the aqueous CO 2 plume volume was about 5.7 times larger than that of the supercritical CO 2 .This is because convection increased the vertical sweep efficiency, which, in turn, increased the aqueous CO 2 plume volume.

CONCLUSIONS
The mobile supercritical CO 2 plume first moves upward because of its buoyancy, and then flows under the  impermeable cap rock.In this case study, the average velocity of the supercritical CO 2 plume was about 0.88 m day -1 during the 20-year injection period.
Supercritical CO 2 will become immobile at the rear of the moving plume due to imbibition.In the post-injection period, the quantity of the residual CO 2 reached a maximum and then became smaller because of the solubility trapping mechanism.Residual gas trapping is a fast and safe mechanism for trapping injected CO 2 .
The plume size of aqueous CO 2 is larger than that of supercritical CO 2 .In this study the aqueous CO 2 plume volume was about 5.7 times larger than that of the supercritical CO 2 at the simulation time of 60 years because convection increased the vertical sweep efficiency.
Convection causes ionic CO 2 to accumulate in the down-dip of the structure and distant from the cap rock.The ionic trapping mechanism changes the CO 2 into a safer ionic phase and also generates flow behavior that is low-leakagerisk and allows safe storage.
The up-dip of the structure was not the major mineralization location, because the source (bicarbonate) of the mineralization tended to sink to the down-dip of the structure.

Fig. 2 .
Fig. 2. Location of the Taihsi basin.The black dots show the location of major gas fields (based on Hsieh et al. 2013).

Fig. 4 .
Fig. 4. Anticline structure map of the Yutengping sandstone formation in the TCS field (based on CPC 2010).

Fig. 5 .
Fig. 5. Plume migration of supercritical phase CO 2 at various times: (a) initial time (the beginning of the injection period), (b) 5 years, (c) 20 years (the end of the injection period), (d) 100 years, (e) 500 years, (f) 1000 years (the end of simulation time).

Fig. 11 .
Fig. 11.Supercritical phase CO 2 and aqueous phase CO 2 plume area and volume at the simulation time of 60 years: (a) plume area of supercritical phase CO 2 , (b) plume volume of supercritical phase CO 2 , (c) plume area of aqueous phase CO 2 , (d) plume volume of aqueous phase CO 2 .

Table 1 .
Basic parameters of the Yutengping sandstone formation.

Table 2 .
Aqueous composition reported for the Yutengping sandstone.

Table 3 .
Volume percentage of minerals of formation rock.