Two-phase flow visualization under reservoir conditions for highly heterogeneous conglomerate rock: A core-scale study for geologic carbon storage

Abstract

Geologic storage of carbon dioxide (CO₂) is considered a viable strategy for significantly reducing anthropogenic CO₂ emissions into the atmosphere; however, understanding the flow mechanisms in various geological formations is essential for safe storage using this technique. This study presents, for the first time, a two-phase (CO₂ and brine) flow visualization under reservoir conditions (10 MPa, 50 °C) for a highly heterogeneous conglomerate core obtained from a real CO₂ storage site. Rock heterogeneity and the porosity variation characteristics were evaluated using X-ray computed tomography (CT). Multiphase flow tests with an in-situ imaging technology revealed three distinct CO₂ saturation distributions (from homogeneous to non-uniform) dependent on compositional complexity. Dense discontinuity networks within clasts provided well-connected pathways for CO₂ flow, potentially helping to reduce overpressure. Two flow tests, one under capillary-dominated conditions and the other in a transition regime between the capillary and viscous limits, indicated that greater injection rates (potential causes of reservoir overpressure) could be significantly reduced without substantially altering the total stored CO₂ mass. Finally, the capillary storage capacity of the reservoir was calculated. Capacity ranged between 0.5 and 4.5%, depending on the initial CO₂ saturation.

Introduction

Atmospherically released carbon dioxide (CO₂) is considered to be a major deriver behind climate change, and as such geologic CO₂ storage is considered a key technology in climate change mitigation strategies^1,2. Both industry and research communities are currently evaluating the safety and feasibility of long-term CO₂ sequestration, and a number of pilot- and demonstration-scale projects have been conducted as a part of this effort to test, monitor, and verify technologies in various subsurface geological environments^3,4,5.

Previous research related to geologic CO₂ storage has until now focused on evaluating sandstone formations, as their relatively high porosity and permeability suggest greater economic viability than other rock formations. For this reason, the reservoir lithology of most pilot-scale (e.g., Frio, Nagaoka, Ketzin and Otway) and demonstration- or commercial-scale projects (e.g., MGSC Decatur, Sleipner, Snøhvit, In Sala and Gorgon) has been sandstone^3,5. The average reservoir porosity has been 5–35% and reservoir permeabilities have ranged from as low as 5 mD (In Salah) to 5,000 mD (Sleipner).

One exception was the MRSCP Gaylord project (a Midwest regional carbon sequestration partnership), where the targeted storage formation consisted of dolomite. This was characterized by interbedded, laminated algal dolomudstone, represented by a mean porosity and permeability of 13% and 22.6 mD, respectively⁶. In an effort to assess the geologic storage potentials, basalt formations have also been considered as alternative storage formations^7,8.

Although a number of CO₂ storage projects are being conducted worldwide, the need for additional safe storage verification experiments using various geological formations cannot be overemphasized. The fact that only limited data has been collected during the injection and post-injection phases demonstrates the necessity of additional field assessments of the processes leading to plume stabilization and long term trapping⁹. Followed by on-going international efforts, the first onshore pilot-scale (~10,000 ton) CO₂ storage project is being conducted in South Korea at the Janggi sedimentary formation, located on a south-eastern portion of the Korean Peninsula (Fig. 1(a)). One distinctive feature differing from other pilot-scale projects is that the targeted formation is a conglomerate, consisting of gravel and cobbles in a silty sand matrix. Figure 1(b) shows a thickness-permeability cross-plot of values selected from various existing storage projects. This plot reveals the injectivity at Janggi falls between 0.1 and 1 darcy-meter, similar to the estimated injectivities at the Nagaoka and In Salah sites.

Figure 1

(a) The location of onshore Janggi pilot site for CO₂ storage in South Korea and the representative lithologic log. (b) Thickness-permeability cross-plot for the Janggi pilot site, along with data from various existing storage projects.

In addition to field-demonstration studies, well-controlled laboratory experiments are essential for obtaining a fundamental understanding of physical^10,11 and chemical processes^12,13 occurring in potential storage formations, as well as predicting their CO₂ trapping capacities¹⁴. A number of studies have conducted reactive transport experiments to understand the complex processes of CO₂-water-mineral interactions and their coupling effects on the CO₂-water multiphase flow at in-situ reservoirs conditions^15,16,17,18. These experimental data can be utilized to validate the numerical studies and build-up confidence in the modeling tools. For example, Smith et al.¹⁸ used detailed experimental data to constrain three-dimensional reactive transport models in order to describe and predict the evolution of pore-space and permeability during geologic CO₂ storage in carbonate reservoirs.

Core-flooding is a technique that has been used to conduct various laboratory experiments on core samples under conditions closely imitating those of subsurface environments¹⁹. Due to this advantage, core-flooding tests have been utilized for characterizing the dynamic behaviour of CO₂ under reservoir conditions. Such tests have focused on resolving various problems related to important factors controlling the behaviours of two-phase fluids, including permeability heterogeneity and multiphase flow^20,21,22, measuring relative permeabilities^23,24,25, interfacial tension^26,27, and capillary pressures^28,29,30. Additionally, the core-flooding experiments have been applied to assessing residually trapped CO₂ amounts under reservoir conditions^{23,25,31,32,33}. Still other research topics have included multiphase fluid behaviour in fractured porous media^34,35, salt-precipitation dynamics^36,37 and the non-equilibrium dissolution of CO₂ in a heterogeneous core^32,38.

Despite these numerous core-scale studies, there has been little investigation on how CO₂ migrates in a highly heterogeneous porous media. This study presents, for the first time, a two-phase flow visualization under reservoir conditions for a highly heterogeneous conglomerate core that has been obtained from the real CO₂ storage site. The three-dimensional (3D) distribution of clasts within the matrix was determined utilizing X-ray computed tomography (CT). Additionally, real-time X-ray scanning techniques (conducted during the CO₂ and brine flow tests) captured the dynamic distribution of CO₂ saturation in the conglomerate core. Multiphase transport simulations were also conducted to assess the effects of heterogeneity on pressure build-up as well as spatial variation of CO₂ saturation. Finally, based on the initial-residual (IR) characteristic curve, the storage potential of the conglomerate formation is discussed.

Results and Discussion

Heterogeneity characterization

A core-plug obtained from the Janggi conglomerate formation was used to investigate the spatial distribution of clasts within a silty sand matrix. Figure 2(a) presents two cross-sectional CT images taken along the longitudinal axis at different angles. In addition, Fig. 2(b) shows the binary image generated from the 8-bit CT images, where the boundaries between the clasts and matrix are distinguished more clearly and discontinuities within the clasts are highlighted. The clast fractioning profiles along the longitudinal core-axis are shown in Fig. 2(c). Nine cross-sectional images, taken along the white-dashed lines appearing in Fig. 2(a), are presented in Fig. 2(d). In the upstream region (30 mm) of the core, the constituents are relatively homogeneous and only a small fraction of clasts were preserved (f < 0.3). In the midstream portion (70 mm), both medium size of clasts and matrix existed together, whereas larger clast sizes were present in the downstream region (100 mm). Figure 1(e) shows the three-dimensional distribution of clasts in the core-plug.

Figure 2

(a) Two CT cross-sectional images along the longitudinal axis taken from different angles. (b) Binary image of the core-plug for clarification of the discontinuities in the clasts. (c) The profile for clasts fraction along the longitudinal core-axis. (d) 2D cross-section images along the white-dashed lines shown in (a). (e) 3D distribution of clasts in the core-plug.

The pore size distribution were analysed for both the matrix and clasts (Fig. 3(a)). The clast regime was characterized by a skewed uni-modal pore-size distribution dominated by micropores (<1 μm diameter). The silty sand matrix regime demonstrated a bi-modal pore size distribution dominated by micropores (<1 μm diameter) and mesopores (>2 μm diameter) (Fig. 3(a)). The bi-modal pore size distribution characteristics of the matrix were also reflected in the shape of the capillary pressure (P_c) curve, appearing as a superposition of two uni-modal pore size distribution characteristic curves. The P_c (S_w = 0.5) of the clasts was more than two orders of magnitude greater than that of the matrix (Fig. 3(b)).

Figure 3

(a) Pore size distribution for the silty sand matrix and clasts. (b) Capillary pressure (P_c) curves as a function of brine saturation.

Multiphase flow experiments

This experimental study aimed to characterize the spatio-temporal distribution of CO₂ saturation build-up during the multiphase flow tests. Two different tests were conducted: (i) CO₂ injection into a brine-saturated core, simulating the CO₂ injection period (drainage) conditions; and (ii) brine injection after the CO₂ injection test, reproducing the post-injection period (imbibition) conditions.

CO₂ saturation distribution snapshots revealed the propagation of CO₂ saturation throughout the conglomerate core with increases in pore volume (PV) (Supplementary Figs S1 and S2). Figure 4(a) presents the vertically averaged CO₂ saturations from the five different locations (30, 60, 70, 110, and 120 mm) delineated in Fig. 2(a). Although the averaging is smoothing out the fluctuation of CO₂ saturation related to small-scale heterogeneity, it clearly presents the increasing tendency within the core sample. All profiles showed logarithmic increases with the PVs. The CO₂ saturation profile at 30 mm increased slowly when approaching the asymptote during the flow tests; however, the other profile (such as at 110 mm) approached their asymptotes more rapidly, implying that CO₂ saturation was quickly stabilized throughout the experiment. It is interesting to note that CO₂ saturation at the downstream side of the core was greater than that at the upstream side in early stages (<1.0 PV). The spatial inversion of CO₂ saturation distribution was attributed to the heterogeneous pore space along the conglomerate core, including the presence of complex discontinuities in clasts (Fig. 2(b)). On the upstream side where the matrix was dominant, CO₂ saturation was low due to the presence of relatively large sized pores, which accommodated the injected CO₂. Equal amounts of CO₂ flowing through the downstream side, dominated by clasts, showed elevated CO₂ saturation because clast discontinuities provided only small pore-spaces.

Figure 4

(a) Vertically averaged CO₂ saturation at five different locations (30, 60, 70, 110, and 120 mm) versus pore volume during the CO₂ injection test. (b) Three subdomains (20 × 20  mm) of the core located at upstream (domain I), midstream (domain II), and downstream (domain III) region that are mainly composed of sand, small clasts and large clasts, respectively. (c) Snapshots of CO₂ saturation at three different domains during the CO₂ injection test. (d) Vertically averaged CO₂ saturation at five different locations (30, 60, 70, 110, and 120 mm) versus pore volume during the brine injection test. (e) Snapshots of CO₂ saturation at three different domains during the brine injection test.

To perform a detailed assessment of the spatial distribution of CO₂ saturation, three subdomains (20 × 20  mm) were selected, one each located in the upstream (domain I), midstream (domain II), and downstream (domain III) regions as shown in Fig. 4(b). These domains primarily composed of sand, small clasts, and large clasts, respectively. Domain I (primarily sand) was characterized by a relatively homogeneous CO₂ saturation distribution over time (Fig. 4(c)) with a small (sub-core) scale heterogeneous CO₂ saturation observed due to the effect of the bi-modal pore-size distribution preserved in the matrix (Fig. 3(a)). Domain II, which was composed of irregularly spaced small clast fragments, demonstrated heterogeneous CO₂ saturation caused by irregularly spaced small clast fragments. Low CO₂ saturations resulted from the presence of clasts, whereas high CO₂ saturations were observed in high porosity sand. Highly heterogeneous CO₂ saturation was observed in domain III, caused by a dense network of discontinuities within the large clasts where multiple discontinuities with relatively large apertures (110 mm) were located, as seen in Fig. 4(c). This discontinuity network provided well-connected pathways for CO₂ flow, resulting in a rapid increase in CO₂ saturation.

The vertically averaged CO₂ saturation was also plotted for the brine injection tests (Fig. 4(d)). As discussed in Oh et al.³², the dynamic transition of CO₂ trapping could be assessed during the imbibition test. Three regimes of CO₂ trapping mechanisms were characterized: (i) immediate mobile CO₂ displacement by injected brine (R1), (ii) immobile CO₂ preservation as residual trapping (R2), and (iii) the gradual dissolution of residually trapped CO₂ into the fresh brine (R3). The dynamic transition of CO₂ trapping at 30 mm displayed initial decrease in CO₂ saturation from 0.35 to 0.27 during the early stage (0.01–0.1 PV) primarily resulted from the mobile CO₂ displacement (Regime 1). In Regime 2, the CO₂ saturation was relatively consistent and stabilized at 0.27 between 0.1 and 1.0 PV, which implies the preservation of immobile CO₂ (residually trapped). This is possible because the brine that enters the core will initially dissolve a small amount of CO₂ at the inlet of the core and becomes fully-saturated preventing it from dissolving any more CO₂. However, as more fresh brine was injected into the core, the residually trapped CO₂ dissolved into fresh brine (Regime 3).

Figure 4(e) shows CO₂ saturation snapshots from the three different domains during the brine injection test. In domain I where the matrix dominated, CO₂ was distributed uniformly at the end of drainage test. Brine displacement tests induced similar uniform distributions in residually trapped CO₂, which eventually dissolved into the brine. However, in domains II and III, a larger fraction of pores was initially filled with CO₂, resulting in more snap-off and trapping as brine invaded the pore space. Under these conditions, CO₂ dissolution was highlighted only within the matrix and discontinuities.

Multiphase transport simulations

The above experiments were conducted with fluid injection at the left boundary, but CO₂ injection from the opposite boundary would cause difference in the distribution of CO₂ saturation due to nature in highly heterogeneous conglomerate core. We used TOUGH2³⁹ simulator to model multiphase and multicomponent fluid transport to assess directional effect of fluid flow on pressure build-up as well as CO₂ saturation distribution along the core sample. A two-dimensional model was constructed based on the CT image of the core sample to reflect the distribution of clasts. Then, two different cases were simulated (case 1: left-to-right flow; case 2: reverse flow).

Figure 5 presents simulation results of pressure (a and f) and CO₂ saturation map (b and g), profiles for CO₂ pressure (c and h) and saturation (d and i) along the center of the model domain, and pressure build-up (e and j) for the two different cases. Both cases showed a significant pressure drop between points A and B, which lie on a boundary of a large sized clast. The heterogeneous pressure distribution resulted in different CO₂ saturation within the model for the two cases. For example, case 1 showed relatively high CO₂ saturation (S_g~0.5) at point A and low CO₂ saturation (S_g~0.3) at point B, while case 2 showed relatively high CO₂ saturation (S_g~0.5) at point B. This implies that the distribution of CO₂ saturation is dependent on the flow direction. The pressure difference between the inlet and outlet (ΔP) showed ~150 kPa during the early stage of injection and stabilized at ~30 kPa for both cases (Fig. 5(e and j)).

Figure 5

Simulation results for the two different flow directions (case 1: left-to-right flow (a–e); case 2: right-to-left flow (f–j)). The results presents pressure (a and f) and CO₂ saturation map (b and g), profiles for CO₂ pressure (c and h) and saturation (d and i) along the center of the model domain, and pressure build-up (e and j).

CO₂ mass at different capillary numbers

A dimensionless capillary number (${{\rm{N}}}_{{\rm{c}}}$), representing the relative contributions of viscous versus capillary forces, was set to reflect the reservoir conditions. Two CO₂ injection tests (qco₂ = 1.0 and 0.1 ml/min) were performed in this study, producing associated ${{\rm{N}}}_{{\rm{c}}}$ values of ~10⁻¹ and ~10⁻², respectively. The high ${{\rm{N}}}_{{\rm{c}}}$ case was observed in a transition regime between the capillary and viscous limits, whereas the low ${{\rm{N}}}_{{\rm{c}}}$ case was under capillary dominated flow conditions. The transient pressure difference (ΔP) values across the core for the two different cases are shown in Fig. 6. The ΔP build-up under a high injection rate (${{\rm{N}}}_{{\rm{c}}}$ ~ 10⁻¹) reached 2,300 kPa in early injection stages before stabilizing to ~700 kPa; however, the low injection rate (${{\rm{N}}}_{{\rm{c}}}$ ~ 10⁻²) displayed a small buildup of ΔP build-up (~150 kPa) in early injection stages before ΔP stabilized to ~65 kPa for the remainder of the multiphase flow experiment.

Figure 6

The transient pressure difference (ΔP) across the core for two injection rates (q = 1.0 and 0.1 ml/min).

Figure 7 displays the CO₂ saturation distribution maps for the two injection rates (qco₂ = 1.0 and 0.1 ml/min) and the CO₂ saturation contrast (ΔSco₂ = S_q=1.0−S_q=0.1) maps between the high and low injection rate cases. The measured ΔSco₂ displayed a strong dependency on core heterogeneity, showing negative values on the upstream side (implying lower CO₂ saturation with a greater injection rate (S_q=1.0 < S_q=0.1)). The high injection rate created excessed pressure on CO₂ (nonwetting phase), enabling CO₂ to transport across the high P_c region on the midstream side. This had a negative impact on the build-up of CO₂ mass on the upstream side. At low injection rates, however, additional CO₂ occupied the pore-spaces on the upstream side because CO₂ transport across the high P_c region on the midstream side was hindered. The CO₂ mass within the core was calculated for both ${{\rm{N}}}_{{\rm{c}}}$ cases to be 2.24 and 2.18 g, respectively, once a steady-state condition was reached. This demonstrated a 10× reduction in injection rate caused only a 2% decrease in CO₂ mass, suggesting that injection rates, which may cause reservoir overpressure when high, can be significantly reduced without substantially altering the total CO₂ mass.

Figure 7

CO₂ saturation maps for the two different injection rates (qco₂ = 1.0 and 0.1 ml/min) and the contrasts of CO₂ saturation (ΔSco₂ = S_q=1.0 − S_q=0.1) between the high and low injection rate cases.

Storage capacity

During the post-injection stage, capillary forces at the pore scale are responsible for breaking up the CO₂ phase into the form of blobs or ganglia, becoming effectively immobile^40,41. The capillary trapping potential was evaluated in these conditions based on the performed multiphase flow experiments. The initial-residual (IR) characteristic curves have commonly been used to describe the residual trapping potentials of rock, with numerous developed models describing this relationship. This study applied the Land model⁴², one of the earliest and most widely used trapping models.

Figure 8(a) presents the relationship between the initial (Sco_2,i) and residual saturations (Sco_2,r), where Sco_2,i refers to the initial CO₂ saturation prior to brine injection testing whereas Sco_2,r refers to the residually trapped CO₂ saturation. For analysis, highly heterogeneous conglomerate core was divided into 50 equal subdomains. Then, based on the brine injection testing, the residual CO₂ saturation achieved at 0.1 PV (R2) was determined (Fig. 4(d)), and subsequently, the amount of capillary trapped CO₂ was assessed using the Land model given by ${{\rm{S}}}_{{{\rm{CO}}}_{2,{\rm{r}}}}=({{\rm{S}}}_{{{\rm{CO}}}_{2,{\rm{i}}}})/(1+{\rm{C}}\cdot {{\rm{S}}}_{{{\rm{CO}}}_{2,{\rm{i}}}})$. Here, C is the dimensionless constant known as the Land coefficient. In this study, the calculated C was calculated as 0.9, showing a monotonic increase in the residual saturation as a function of Sco₂,_i. This was in agreement with numerous previous works investigating water-wet porous medias.

Figure 8

(a) A trapping curve relating initial CO₂ saturation (Sco_2,i) to residual CO₂ saturation (Sco_2,r). The Sco_2,i refers to the initial CO₂ saturation prior to imbibition test while the Sco_2,r refers to the residually trapped CO₂ saturation. Red circles represent the relationship between the Sco_2,i and Sco_2,r within equally divided subdomains; error bars represent one standard deviation. (b) Trapping capacity (φSco_2,r) versus initial CO₂ saturation (Sco_2,i); colors represent different ranges of porosity.

The capillary-trapping capacity, defined as φSco_2,r (φ is porosity) is a rescaling of the traditional IR saturation plot^43,44. This quantity is of great importance in geologic CO₂ storage applications because it states how much CO₂ can be stored securely per unit volume of a rock. Figure 8(b) presents the relationship between Sco_2,i and φSco_2,r at different porosities (blue, red and orange circle symbols). The IR relationships of other core samples (Berea and Pohang sandstones) are plotted alongside for comparison. Core sample details are described in the Supplementary Table S1. The trapping capacities of both Berea and Pohang sandstones showed more or less linear increases with Sco_2,i, whereas the φSco_2,r values for heterogeneous conglomerate cores were widely spread and dependent on the size of porosity. Although the overall capacity of the conglomerate core was smaller than those of the Berea or Pohang sandstones, the conglomerate high porosity region displayed a similar capacity to the sandstones (e.g., φSco_2,r ~4.5% at Sco_2,i = 0.3).

Conclusions

Although sandstone formations are considered to be the most appropriate for CO₂ sequestration, other types of geologic formations are being investigated for this purpose. For example, the target rock formation in MRSCP Gaylord project was dolomite⁴⁵, and the Midale project (an extension of the Weyburn project) demonstrated the possibility for CO₂ containment in a fractured reservoir⁵. Basaltic formations have also been considered as alternative storage formations^7,8. In South Korea, a conglomerate formation is being tested for geologic CO₂ storage.

This study reported a thorough core-scale investigation of a highly heterogeneous conglomerate formation. The rock heterogeneity was investigated with the use of X-ray CT and MICP analysis revealing that the targeted storage formation was composed of a silty sand matrix and irregularly spaced clasts of various sizes. Laboratory multiphase flow tests with flow visualization under reservoir conditions provided important information on CO₂ migration behaviours as well as the storage capacity of the highly heterogeneous conglomerate. Variations in clast spatial configurations gave rise to different storage features. Dense networks of discontinuities within clasts were observed to provide well-connected CO₂ flow pathways, causing a rapid increase in CO₂ saturation, and potentially helping to reduce overpressure. Two flow tests, one in the capillary-dominated condition and the other in a transition regime between the capillary and viscous limits, revealed that a 10× reduction in injection rate caused only a 2% decrease in the CO₂ mass. This implies the heterogeneity positively influenced pressure management because high injection rates, which may cause overpressure in the reservoirs, could be significantly reduced without substantially altering the total stored CO₂ mass. We note that although chemical reactions are not considered in this study, the complex process of CO₂-water-mineral interactions may play important roles on multiphase flow in subsurface reservoirs and trapping mechanisms.

Among the four potential trapping mechanisms, capillary trapping is a rapid and effective mechanism for rendering an injected fluid immobile and reducing the need to ensure caprock integrity. The capillary-trapping capacity of the conglomerate core investigated in this study was calculated using imbibition test results and the IR characteristic curve. While the Berea and Pohang sandstone cores showed approximately linear increases in capacity with initial CO₂ saturation, the Janggi conglomerate displayed significant variations due to its highly heterogeneous porosity distribution. The conglomerate capacity ranged between 0.5 and 4.5%, depending on the initial saturation (0–0.4): for an initial saturation of 0.4, the conglomerate capacity ranged between 3–4%. Despite the fact that the trapping analyses showed lower overall capacities relative to homogeneous sandstone, the high porosity region of the conglomerate core displayed a similar capacity to sandstones, implying conglomerate formations could be an alternative for CO₂ storage formation. CO₂ injection and comprehensive monitoring in conglomerate formations will validate the feasibility of CO₂ storage in the highly heterogeneous conglomerate formations.

Methods

Field site and geological setting

An onshore pilot-scale geologic CO₂ storage project was launched in South Korea in 2011. The primary tasks (Phase I, 2012–2015) were the site-selection and initial characterization of potential pilot-scale storage sites. The Janggi conglomerate, composed of both clasts and silty sand matrix, was chosen as the target storage formation for the project. More recently, Phase II (2015–2017) was completed, including the drilling of seven boreholes to investigate the lithology of targeted and capping formations. The representative lithological log for this is illustrated in Fig. 1(a)⁴⁶. The average porosity and permeability of the storage formation were 16.1% and 8.5 mD, respectively⁴⁷. The overlying Seongdongri formation, composed of thick mudstone and massive dacitic tuffs, forms the regional sealing formation for CO₂ storage, and a monitoring well (~40 m up-gradient from the injection well) was drilled. A comprehensive monitoring system was installed including a down-hole pressure and temperature measurements, vertical seismic profile (VSP), electrical resistivity tomography (ERT), distributed temperature sensing (DTS) and distributed acoustic sensing (DAS) along the well via an optical fiber, and U-tube downhole sampler.

X-ray micro CT

The dimensions of the core-plug obtained from the Janggi conglomerate formation were 130 × 48 mm (length × diameter). X-ray computed tomography (CT), a non-destructive and non-invasive method, was used to investigate the spatial distribution of clasts within the silty sand matrix. The CT images used in this study were obtained at 140 kV and 200 μA. A total of 2,000 cross-sectional images were taken at 0.065 mm intervals with a micrometer-order pixel size. Subsequently, two-dimensional (2D) images were stacked to generate 3D core sample images.

Multiphase flow tests with in-situ imaging technology

The apparatus for core-flooding experiments was designed to conduct multiphase flow tests. It was comprised of fluid (supercritical CO₂ and brine) injection, core-holder, temperature controlling, confining, and back-pressure control systems (Supplementary Fig. S5). A 2D X-ray scanning system, consisting of an X-ray tube and a detector, was integrated into the core-flooding system in order to capture the real-time distribution of CO₂ saturation distributions along the core sample. The detector board had two arrays of 128 channels each, corresponding to 256 detectors with a detector pitch of 0.4 mm. The array of the X-ray detector arrays were perpendicular to the core-axis, performing scans parallel to core-axis. The data output was in 16-bit format, ranging between 0 and 65,535. The produced 2D images were converted into CO₂ saturations using the following equation: ${{\rm{S}}}_{{{\rm{CO}}}_{2}}=\,\frac{{{\rm{I}}}_{\exp }-{{\rm{I}}}_{{\rm{brinesat}}}}{{{\rm{I}}}_{{{\rm{CO}}}_{2}}-{{\rm{I}}}_{{\rm{brinesat}}}}$, where Ico₂ and I_brinesat are the values of grayscale intensity obtained from the background scans of the core, which were saturated with CO₂ and brine in the core, respectively. Finally, I_exp was the value obtained for the multiphase condition. The obtained CO₂ saturation map may include noise mainly due to the relatively small density contrast between the two fluid phases (CO₂ and brine) compared with the contrast between any one of the fluids and the rock⁴⁸. In the experiments, NaI was used as a dopant to enhance the X-ray attenuation so that the CO₂ saturation could be distinguished from the brine in the scanned image.

Two multiphase flow tests were conducted: (i) CO₂ injection into a brine-saturated core, simulating the CO₂ injection period (drainage) conditions; and (ii) brine injection following the CO₂ injection test, reproducing post-injection period (imbibition) conditions. Downstream pressure was maintained during these two experiments at 10 MPa, and the temperature was set at 50 °C to replicate the subsurface environment at Janggi pilot site. Two different injection rates (q = 1.0 and 0.1 ml/min) were applied. The capillary numbers, ${{\rm{N}}}_{{\rm{c}}}=\,\frac{{\rm{H}}}{|{\rm{\Delta }}{P}_{c}|}\frac{{\rm{\Delta }}P}{L}$ (where L and H represent the length and diameter of the core, respectively; ${\rm{\Delta }}P$ represents the pressure drop; and $|{\rm{\Delta }}{P}_{c}|$ was calculated from the capillary pressure curves), were ~10⁻¹ and ~10⁻², for q = 1.0 and 0.1 ml/min, respectively.

Pore size distribution and capillary pressure

The pore size distributions and capillary pressure curves for the silty sand matrix and clasts in the conglomerate sample were determined using a mercury injection porosimeter (MICP). To perform MICP tests, two small subsamples (~1 cm³) were taken from the conglomerate core each representing either the silty sand matrix or clasts. These subsamples were dried in a vacuum oven at 70 °C for 24 hours prior to testing. The MICP and corresponding mercury saturation measured in the mercury/air system were transformed into a capillary pressure and corresponding CO₂ saturation for a CO₂/brine system using the following relationship: $\frac{{P}_{c,C{O}_{2}}}{{P}_{c,Hg}}=\frac{{\sigma }_{C{O}_{2}}\,cos{\theta }_{C{O}_{2}}}{{\sigma }_{Hg}\,cos{\theta }_{Hg}}$, where ${P}_{c}$ (Pa) represents the capillary pressure, $\sigma$ (mN m⁻¹) represents the interfacial tension (IFT) between two fluids, and $\theta$ (deg) represents the contact angle measured in the wetting phase. The IFT of ${\sigma }_{Hg}$ = 485 mN m⁻¹ and ${\sigma }_{C{O}_{2}}$=32.6 mN m⁻¹ was used in the conversion for the geologic CO₂ sequestration conditions. The contact angles for the two systems were assumed to be equal²⁹.