CO₂ Sequestration in a Carbonate Saline Aquifer: An Investigation into the Roles of Natural Fractures and Well Placement

Abstract

CO₂ sequestration is considered one of the main pillars in achieving the ongoing decarbonization efforts. A myriad of CO₂ sequestration projects targeted sandstone reservoirs since carbonate reservoirs appeared to be unpropitious due to their geological complexity and unfavorable mineralogy and properties. This study investigates CO₂ sequestration potential in a carbonate saline aquifer while considering various geological complexities by capitalizing on numerical simulation. A synthetic anticline reservoir model examined the optimum well location and landing zone for CO₂ sequestration. Additionally, the model evaluated the role of natural fractures in the migration path of CO₂ plume and geochemical reactions throughout the storage process. The study demonstrates that placing the injection well away from the top of the structure in a low-dip region while injecting in the bottom interval would yield the optimum design. After applying a plethora of analyses, geological complexity could impede the migration path of CO₂ but eventually produce a similar path when injected in a similar region. The geochemical interactions between the injected CO₂ and reservoir fluids and minerals reduce the free and trapped CO₂ quantities by dissolving calcite and precipitating dolomite. Furthermore, natural fractures impact the CO₂ quantities during early times only when the fractures cross the top layers. Similarly, the CO₂ migration differs due to the higher permeability within the fractures, resulting in slightly different CO₂ plumes. Consequently, the role of natural fractures should be limited in carbon storage projects, specifically if they do not cross the top of the reservoir. This study reflects a unique perspective on sequestering CO₂ while capturing the roles of natural fractures and well placement in depicting the migration path of the CO₂ plume. A similar systematic workflow and holistic approach can be utilized to optimize the subsurface storage process for potential formations.

1. Introduction

Greenhouse gas (GHG) emissions have been increasing significantly in the past 30 years, with carbon dioxide (CO₂) representing the majority of these emissions [1,2]. Even though CO₂ is considered one of the least abundant gases in the atmosphere, excess CO₂ concentrations could lead to some adverse effects on the Earth’s ecosystem. Therefore, policymakers have recommended a myriad of decarbonization strategies worldwide [3]. The net-zero strategy, where any emitted GHG should be either naturally absorbed or sequestered, reflects one of the strategies for controlling the CO₂ concentrations in the atmosphere. The recommended strategies have the potential to ultimately control the GHG levels, particularly the CO₂, in the atmosphere [4,5].

CO₂ sequestration should partially help to reduce the CO₂ level in the atmosphere to reach the desired decarbonization goals [6]. The sequestration process refers to capturing the excess CO₂ from the atmosphere and injecting it into suitable storage in the subsurface. Afterward, the injected CO₂ can be stored permanently in the formation while maintaining healthy storage conditions [7]. The Sleipner carbon capture and storage project exemplifies a great CO₂ storage example, which captured and stored more than 16 metric megatons of CO₂ from 1996 to 2016 [8]. Storing CO₂ into the subsurface has been showing presumably effective results in controlling the CO₂ levels in the atmosphere. Consequently, decarbonization strategies capitalize on CO₂ sequestration as one of the potentially feasible techniques for reaching net zero [4,5,6,9].

Generally, the injected CO₂ interacts with the formation fluids and minerals in the storage process; these interactions play a major role in trapping the CO₂ in the formation. The CO₂ trapping materializes via two mechanisms: physical trapping mechanisms and geochemical trapping mechanisms. Stratigraphy and structure trapping and hydrodynamics trapping represent the main physical trapping mechanisms. In the stratigraphic and structure mechanism, the injected CO₂ becomes trapped by a physical seal such as a caprock or a fault. Hydrodynamics, on the other hand, refers to trapping the CO₂ within the pores during its migration due to buoyancy. Furthermore, geochemical trapping has the potential to develop if chemical interactions between the injected CO₂ and the formation’s rocks or fluids occur. CO₂ mineralizes in the formation when the CO₂ interacts with the rocks while it dissolves into the in situ brine [7,9,10,11]. Each trapping mechanism contributes differently to the overall CO₂ subsurface storage across time. Typically, mineral trapping requires a significantly longer time to occur compared to the other trapping mechanisms, which have the potential to take place within the first year of injection [7]. Since each trapping mechanism has its unique features and timeframe, it will be crucial to understand the expected trapping mechanisms in the targeted formation to safely optimize the injected CO₂ quantity.

Even though carbonate reservoirs are considered ubiquitous, sandstone reservoirs have been utilized for CO₂ sequestration projects [10,11,12]. For instance, the Quest project, which has the potential to sequestrate up to 1.1 million metric tons of CO₂ per annum, targeted a sandstone reservoir [13]. Mineralogical and physical differences between carbonate and sandstone reservoirs drove the selection process due to their storage capabilities. The higher porosity values that usually appear in a sandstone reservoir compared to a carbonate reservoir provide them with a better storage capacity potential for CO₂ sequestration [10,11,13]. Moreover, sandstone reservoirs tend to hold a lower percentage of calcite than carbonate reservoirs. The presence of calcite enables mineral dissolution and dolomite precipitation when injecting CO₂ [14,15]. Besides the parameters and properties differences between both reservoirs, natural fractures existing in carbonate reservoirs pose additional geological complexity which makes it more challenging to depict the CO₂ plume in the subsurface [16,17]. Thus, previous CO₂ sequestration projects preferred to develop sandstone reservoirs because of their subsurface environment.

To date, the vast majority of CO₂ sequestration research highlights the potential of sandstone reservoirs to store CO₂. If a study investigates a carbonate reservoir, it would be localized and focus on a unique signature in a carbonate reservoir [8,11,12,18]. For instance, Sohal et al. examined potential storage capacity changes and plume migration due to geological heterogeneities based on the Hontomin site [18]. Since carbonate reservoirs have promising potential to store CO₂, holistic analysis that captures the subsurface geochemical reactions and trapping mechanisms should be thoroughly conducted to ensure optimum and safe CO₂ sequestration.

Reservoir simulators can model CO₂ sequestration in saline aquifers. The simulation models capitalize on compositional models using an equation of state that can be calibrated to the aquifer’s subsurface conditions. Typically, geochemical reactions between the injected CO₂ and formation rocks can be incorporated via geochemical packages like PHREEQC or TOUGHREACT [19,20,21]. Furthermore, phase equilibrium helps to compute the CO₂ solubility in the aquifer [22,23]. Figure 1 illustrates how two rock samples can undergo different geochemical reactions during CO₂ sequestration based on Licbinska et al.’s results [24]. Lyu and Voskov, Ranganathan et al., and Anchliya et al. simulated the performance of CO₂ sequestration while demonstrating the expected CO₂ distribution and plume within the aquifer (see Figure 2) [25,26,27]. Utilizing such simulation models enables accurate estimations of the subsurface geochemical interactions as well as the depiction of flow dynamics.

This study investigates CO₂ sequestration in a carbonate saline aquifer using a systematic workflow that capitalizes on numerical simulation. The workflow incorporates the impacts of natural fractures on the CO₂ plume migration path and resulting geochemical reactions, such as mineral precipitation and dissolution in the matrix and fractures, throughout the storage process. The study elucidates the role of each trapping mechanism in a carbonate saline aquifer, identifying the key characteristics that control the CO₂ sequestration potential in a similar subsurface environment. Consequently, the CO₂ sequestration project design can be optimized by locating the optimum surface and subsurface CO₂ injection interval.

2. Methodology

2.1. Model Development

A synthetic three-dimensional reservoir model was developed to examine numerous subsurface complexities in a carbonate saline aquifer. The model was structured as an anticline emulating common petroleum systems. Intentionally, the structure was designed asymmetrically, where one side had a higher dip than the other, as illustrated in Figure 3. Furthermore, the model was initialized with homogenous petrophysical properties, which are represented in

Table 1. Reservoir rock and fluids’ properties and parameters.

Property/Parameter	Value
Porosity (%)	15
Horizontal permeability (mD)	100
Vertical permeability (mD)	10
Saline water density (kg/m3)	1020
Reservoir datum depth (m)	1320
Initial datum reservoir pressure (MPa)	13

along with other essential parameters, in a regular Cartesian grid with 75 × 75 × 20 (112,500) cells.

After generating the model’s structure, a simulation model was generated to run numerical simulations in CMG-GEM^® (version 2023.30) [28]. The simulator is considered a multidimensional equation-of-state compositional simulator that can compute subsurface multiphase flow using a finite volume method with finite differences. Miscible and immiscible fluid injection in various porous media can be efficiently simulated while considering all of the major mechanisms, such as viscosity alteration, in the process. Additionally, it can incorporate solubility and ionic, chemical, and residual trapping with any subsurface rock (e.g., mineral dissolution) or fluid changes (e.g., gas saturation). Coupling various geoscience packages with CMG-GEM^® is considered one of the advantages of utilizing the simulator. For example, fractures can be developed in a specialized fracturing software package and coupled afterward with CMG-GEM^®. Therefore, the simulator can forecast subsurface flow dynamics by smoothly integrating detailed geological, petrophysical, and geophysical features (e.g., fractures, saturations, faults, etc.), capturing various degrees of reservoir complexities.

The simulator was utilized to continuously inject CO₂ for five years in the anticline model in which five metric tons per day were injected continuously. After the five-year injection, the CO₂ storage process in the aquifer with the plume migrations and geochemical interactions was examined for 195 years (i.e., 200 years of simulation). The simulation was performed in a finite saline aquifer model with closed boundaries. Also, the model assumed that the caprock acts as a perfect seal. During the simulation, the injected CO₂ amount can be represented by free, dissolved, or trapped CO₂ quantities within the aquifer. The CO₂ solubility in the aqueous phase was calculated in the model via the Li and Nghiem method with the general Henry’s law [22]. Moreover, the saline water aquifer ionic composition, concertation, and CO₂ molecular diffusion rates in the aqueous phase were based on Machado et al. [29]. Carbonate reservoir relative permeability curves were utilized in the model based on Bennion and Bachu (see Figure 4) [30]. Additionally, Carlson’s model for relative permeability was incorporated using Land’s trapping function, with a maximum residual gas saturation of 0.25, according to Burnside and Naylor [31,32,33].

The geochemical reactions, which occurred as soon as the CO₂ reached the aquifer, played an essential role in the CO₂ distribution within the aquifer; hence, the reactions were captured in the model. The injected CO₂ interacts with the aquifer brine, which traps the CO₂ as carbonate ions, based on Equations (1)–(3), using kinetic parameters from the PHREEQC database [19]. In addition to the aquifer brine, the interactions between the CO₂ with the reservoir minerals were included using transition-state-theory-derived rate laws [21]. The study mainly focused on calcite dissolution and dolomite precipitation. Equations (4) and (5) show dolomite precipitation and calcite dissolution reactions computed in the model. The geochemical interactions reflect the solubility trapping (Equations (1)–(3)) and mineralization reactions (Equations (4) and (5)) utilized in the simulation model.

$${\mathrm{OH}}_{\left(\mathrm{aq}\right)}^{-}+{\mathrm{H}}_{\left(\mathrm{aq}\right)}^{+}={\mathrm{H}}_2{\mathrm{O}}_{\left(\mathrm{aq}\right)}$$

(1)

$${\mathrm{CO}}_{2\left(\mathrm{g}\right)}+{\mathrm{H}}_2{\mathrm{O}}_{\left(\mathrm{aq}\right)}={\mathrm{H}}_{\left(\mathrm{aq}\right)}^{+}+\mathrm{H}{\mathrm{CO}}_{3\left(\mathrm{aq}\right)}^{-}$$

(2)

$${\mathrm{CO}}_{3\left(\mathrm{aq}\right)}^{2-}+{\mathrm{H}}_{\left(\mathrm{aq}\right)}^{+}=\mathrm{H}{\mathrm{CO}}_{3\left(\mathrm{aq}\right)}^{-}$$

(3)

$$\mathrm{CaMg}{\left({\mathrm{CO}}_3\right)}_{2\left(\mathrm{s}\right)}={2\mathrm{CO}}_{3\left(\mathrm{aq}\right)}^{2-}+{\mathrm{Mg}}_{\left(\mathrm{aq}\right)}^{2+}+{\mathrm{Ca}}_{\left(\mathrm{aq}\right)}^{2+}$$

(4)

$${\mathrm{CaCO}}_{3\left(\mathrm{s}\right)}+{\mathrm{H}}_{\left(\mathrm{aq}\right)}^{+}={\mathrm{Ca}}_{\left(\mathrm{aq}\right)}^{2+}+\mathrm{H}{\mathrm{CO}}_{3\left(\mathrm{aq}\right)}^{-}$$

(5)

Natural fractures were introduced in the simulation model by capitalizing on the Embedded Discrete Fracture Model (EDFM) [34,35]. The EDFM technique adds additional gridblocks to the simulation model gridblocks reflecting the fracture network. These additional gridblocks introduce a transmissibility function that captures the subsurface flow mechanics [36,37,38]. The natural fracture network was generated via a stochastic approach. Three realizations of the natural fracture network were developed, as demonstrated in Figure 5. In all realizations, the introduced fractures had a permeability tenfold the matrix permeability, which should allow the injected CO₂ to exhibit various flow patterns within the aquifer.

2.2. Sensitivity Analysis Framework

The developed simulation model was utilized to examine different well placement strategies, which include its surface location and injection zone, during CO₂ sequestration in a carbonate saline aquifer. A one-variable-at-time approach was used to analyze various scenarios. In addition to well placement, geochemical interactions and natural fractures were incorporated into the analyses, ultimately providing a deeper understanding of how to optimize CO₂ storage in a similar subsurface environment.

2.2.1. Well Placement and Injection Zone

Three main locations were analyzed in the designed anticline structure. The first well location was around the top of the anticline structure. The other two locations were placed in either high-dip or low-dip regions, which were designed to be equally distanced from the first location. Figure 6 illustrates the wells’ placement in the developed structural model. Moreover, three subsurface injection zones were examined in the analysis to assess the differences when injecting in the top, middle, or bottom of the aquifer. The wells’ placements were selected to provide insights into how the dip difference along with subsurface injection zones can presumably impact the trapping mechanisms while incorporating the geochemical reactions.

The distance from the top of the structure was investigated by placing wells away from the top with matching characteristics (see Figure 7). In addition to the distance, tortuosity was emulated in another scenario by adding flow barriers in the simulation model, as shown in Figure 8. In each scenario, the CO₂ should reach the top of the structure via various paths; hence, the traveled distance would be different, reflecting the role of displacement in each trapping mechanism.

2.2.2. Natural Fractures Network

In each realization of the natural fracture network, four scenarios were considered to assess the impact of the vertical extension of the fractures on CO₂ migration. The first scenario had the fracture extending across the entire aquifer, while the other scenarios had the fractures limited to the top, middle, or bottom zones. Figure 9 shows the differences between the vertical extensions in each scenario. In every scenario, the well location and injection zone were placed away from fractures to eliminate near-wellbore effects (see Figure 10). Initially, all the scenarios were analyzed without including geochemical reactions due to the computational runtime. The geochemical reactions were incorporated when the most influential scenario was determined. These scenarios should depict natural fractures’ unique signatures in a similar subsurface flow environment.

3. Results and Discussion

3.1. Well Placement Optimization

The well placement in CO₂ sequestration has significant importance in the storage process, which generally focuses on the well location and its injection zone. Consequently, the optimization process in the developed simulation model tackled these two aspects independently while considering the various trapping mechanisms. The optimization boils down to providing the most plausible injection strategy that yields optimum CO₂ storage.

3.1.1. Injection Location

The location of the well plays a major role in the subsurface CO₂ distribution due to the varying contribution of each trapping mechanism. Figure 11 illustrates the three main locations and the global mole fraction, which includes the dissolved and free CO₂, in the formation after 200 years of injection. The anticline structure aids in capturing the trapping mechanisms signatures; thus, it plays a significant role in understanding how each mechanism contributes to the migration of the CO₂ plume.

CO₂ vertically migrates from the bottom to the top of the structure because of the density differences between CO₂ and resident water in all simulation scenarios. However, the amount of CO₂ (i.e., free, dissolved, or trapped CO₂) are quite different. The low-dip location has the potential to dissolve more CO₂ while maintaining lower free gas compared to the other scenarios. Furthermore, only a mere difference can be expected among the three scenarios for the amount of trapped gas. Figure 12 clarifies the differences between the CO₂ quantities for 200 years. Even though the low-dip scenario shows relatively higher trapped CO₂ in the early stages, the top structure scenario tends to trap slightly more CO₂ because it has higher free CO_2, providing an opportunity for the excess gas to get trapped within the pores. Since the results revealed that the low-dip scenario can potentially maximize the dissolved CO₂ while minimizing the free CO₂, it would be preferable to place the injection well in a low-dip region.

After incorporating the geochemical reactions into the simulation model for the three scenarios, the analyses suggest that all of them should obey a similar trend (see Figure 12). Even though the geochemical reactions reduce the gap between the scenarios, the difference in the CO₂ amounts resulted from the different reactions with the reservoir rock minerals. The anticipated interactions between the injected CO₂ and the subsurface minerals yield calcite dissolution and dolomite precipitation, as demonstrated in Figure 13. The low-dip scenario yields the lowest calcite dissolution and dolomite precipitation. Moreover, the free CO₂ has the most significant reduction in the top structure scenario since more calcite dissolution occurs. It is noteworthy that the majority of these processes require a relatively long-time post-injection for all scenarios. Thus, the low-dip region would be preferable to start the CO₂ sequestration development, even when including the geochemical interactions.

In addition to the dip differences, the distance between the injector and the top of the structure can influence the CO₂ storage. Placing the well further away from the top of the structure helped to dissolve more CO₂ compared to the other scenarios.

Table 2. Total moles of CO₂ for different well distances from the top of the structure.

Scenario *	Free CO2 (mol)	Trapped CO2 (mol)	Dissolved CO2 (mol)
A	6.07 × 107	3.22 × 107	1.53 × 108
B	3.86 × 107	2.99 × 107	1.75 × 108
C	2.61 × 107	2.46 × 107	1.87 × 108

* Refer to Figure 7 for the well locations.

reflects the variation in the amount of CO₂ for all scenarios. Similar to the dip, the distance impacts the distribution of the CO₂ within the aquifer. For instance, the simulation results imply that Scenario A has more than double the free CO₂ compared to Scenario C. The difference between the scenarios has been dissolved in Scenario C. Moreover, adding a flow barrier impacts the amount of CO₂ in the early stages but yields a relatively similar plume migration path after 200 years, as illustrated in Figure 14. Initially, barriers force the flow to find an easier path to flow; hence, the plume slightly expands until it overcomes the barriers. It is crucial to note that CO₂ quantities after including the tortuous path remain almost the same, which highlights the distance influence on the trapping mechanisms compared to the plume path.

3.1.2. Injection Zone

Theoretically, injecting the CO₂ into the bottom zones enables it to possibly get more exposure to the aquifer volume (i.e., higher swept volume). The developed scenarios examine if injecting at different zones could lead to potential impacts on the plume migration path. Figure 15 shows a side view of the global mole fractions for all scenarios. Unlike the top zone injection, the bottom zone allows the CO₂ to have more contact with the formation as well as the fluids, yielding a higher possibility of dissolving and trapping the injected CO₂. Eventually, the overall plume extension implies that the landing zone could define its shape, specifically when neglecting the bottom zones, due to the significant wasted opportunity around the wellbore.

3.2. Natural Fractures Network Influence

The three fracture network realizations aid in analyzing the fracture roles across different layers. The generated scenarios reflect the trapping mechanisms within the fractures throughout the storage process. The natural fractures around the wellbore region provide significantly faster corridors to transport the CO₂ from the bottom to the top of the aquifer. However, the free CO₂ has the potential to dissolve across the fracture path either laterally or vertically when it encounters fractures away from the wellbore. Figure 16 illustrates the global mole fraction in the fractures throughout the migration path. Additionally, the fractures have a noticeable impact in the earlier stages but a negligible impact on the plume shape in the long term, as demonstrated in Figure 17.

The vertical extension of fractures plays a major role in the CO₂ distribution, specifically when the fractures are located away from the wellbore. The free CO₂ eventually moves to the top of the reservoir since it is lighter than the saline water. Consequently, any fracture that crosses the top layers could interrupt the migration path. It is crucial to consider that the aquifer bottom layers tend to transport trapped and dissolved CO₂ while the free CO₂ remains in the top layers, as reflected in Figure 18. In the long term, the results for all scenarios suggest that all fracture network intensities and extensions yield comparable CO₂ amounts, as shown in

Table 3. Total CO₂ moles with different fracture network realizations.

Realization	Layers	Free CO2 (mol)	Trapped CO2 (mol)	Dissolved CO2 (mol)
1	All	2.49 × 107	2.49 × 107	1.88 × 108
	Top	2.54 × 107	2.54 × 107	1.88 × 108
	Middle	2.58 × 107	2.58 × 107	1.87 × 108
	Bottom	2.52 × 107	2.52 × 107	1.88 × 108
2	All	2.52 × 107	2.51 × 107	1.88 × 108
	Top	2.50 × 107	2.50 × 107	1.88 × 108
	Middle	2.50 × 107	2.50 × 107	1.88 × 108
	Bottom	2.52 × 107	2.52 × 107	1.88 × 108
3	All	2.48 × 107	2.46 × 107	1.88 × 108
	Top	2.49 × 107	2.49 × 107	1.88 × 108
	Middle	2.50 × 107	2.50 × 107	1.88 × 108
	Bottom	2.54 × 107	2.53 × 107	1.88 × 108

. The major role of the fractures, however, can be noticeable in the short term when the fractures either extend across all the layers or are limited to the top layers, allowing them to dissolve a relatively higher volume of CO₂ and trap less volume (see Figure 19). Therefore, natural fractures should have a negligible contribution to the overall CO₂ migration path while altering the dissolved CO₂ distribution if some fractures cross in the top layers.

Incorporating the geochemical reactions in the fractured scenarios yields a comparable trend to the geochemical scenarios without the fractures (see Figure 12 and Figure 19). The dissolved CO₂ exhibits similar behavior to the low-dip scenario, where the dissolved volume is reduced compared to the scenarios that exclude the geochemical reactions. The similarities can be related to the similar well placement resulting in similar plume migration paths. Since the overall CO₂ quantities decrease, the plume size should shrink due to the interactions with the minerals. Figure 20 demonstrates the expected changes when including the geochemical reactions. Moreover, the dolomite precipitation across the aquifer shows that the fracture network expedites the dolomitization process only in the short term, as represented in Figure 21. On the other hand, the overall calcite dissolution will be higher with the presence of fractures. The impact relates to the faster exposure to the aquifer’s minerals due to the fracture network presence. Ultimately, the fracture network reduces the CO₂ quantities by capitalizing on the mineral trapping.

4. Conclusions

CO₂ sequestration will definitively contribute to reducing the GHG concentrations in the atmosphere by safely storing it in subsurface formations. The study suggests that the optimum well placement should be in a relatively low-dip region. Furthermore, the injection zones aid in defining the CO₂ plume shape since it has the potential to influence the trapping mechanisms. Due to the maximum vertical exposure, injecting in the bottom zones enables the CO₂ to optimize its distribution. In addition to the injection zone, tortuosity can impede the path of the CO₂ plume but will eventually yield a similar plume. Incorporating the natural fracture realizations into the developed model reveals that the CO₂ mainly utilizes the fractures, which extend into the top of the formation, to dissolve relatively higher volumes in the early stages. However, the overall CO₂ quantities should not significantly be impacted after 200 years. Unlike the natural fracture network, the geochemical reactions will alter the amount of CO₂ in different forms because of the calcite dissolution and dolomite precipitation.

The injection well distance and tortuosity should be further analyzed to capture the heterogeneity roles in the trapping mechanisms. Additionally, an investigation of the well’s injectivity must be incorporated for long-term injection, specifically with intermittent injection, since the mineralization could alter the injectivity. Incorporating X-ray diffraction (XRD) and mineral analysis should be considered in future work with geochemical reactions to provide a better depiction of the CO₂ plume migration. This study highlights a workflow and optimization process that can be extended to obtain optimum storage by capitalizing on detailed structural models, reservoir properties, and mineralogy.