2.1 Sample preparation

The bentonite used in the study was sourced from the Gaomiaozi area of Inner Mongolia. It looks like a light gray powder, primarily composed of montmorillonite (accounting for 58.3% by mass). More fundamental properties of the bentonite were presented in the study conducted by Wen (2006). The granite used was obtained from the Beishan area in Gansu province. To prepare the Gmaomiaozi (GMZ) bentonite powder with a target water content of 10.55%, the powder was exposed to controlled relative humidity of 43% until it reached mass equilibrium. Subsequently, the bentonite powder was accurately weighed and placed into molds for compaction, aiming for a dry density of 1.7 g/cm3. Two specimens were prepared by compacting the powder at a constant loading rate of 0.1 mm/min. When the loading piston reached the set position, the axial force was maintained constant and left to stand for 1 h to reduce the axial rebound resulting from unloading. The compacted specimens had a diameter of 25 mm and a height of 10 mm. Granite was machined into three cylindrical specimens with a diameter of 50 mm and a height of 10 mm. Bores with a diameter of 26 mm were drilled in the centers of two of these specimens, creating hollow granite disks. Two interfaces were generated: One of the disks was left as drilled, creating a smooth interface with the subsequently inserted bentonite specimen. Circular grooves with depths between 1 and 2 mm were cut in the inner surface of the second tube, creating a grooved interface with the bentonite specimen.

2.2 Experimental design

The water injection and gas breakthrough experiments were carried out under two different boundary conditions, flexible and semirigid, as shown in Figure 3. The flexible condition was previously used in experiments with bentonite samples (Guo et al., 2022). However, in this study, it was used for the cylindrical granite specimen. The semirigid condition was applied to the granite–bentonite systems, where the compacted bentonite specimens were placed within the hollow granite tubes. The gap between granite and bentonite was measured to be approximately 1 mm. The experimental details, including those from the study with bentonite specimens (Guo et al., 2022), are summarized in Table 1.

2.3 Water infiltration experiments

2.4 Gas breakthrough experiments

During the experiments, the temperature was maintained at 295 K to reduce the influence on argon viscosity. The viscosity of argon under different pressures is shown in Table 2 (Younglove & Hanley, 1986).

3.1 Water infiltration tests

Figure 5 shows the water volume with time for the three specimens. The flow rate increases sharply at the start of the water injection and later transitions to a linear function of time.

The saturated water permeability values are summarized in Table 3, together with the previously published values for pure bentonite. For intact granite, Kw is 1.53 × 10−19 m2; for the SBG specimen, it is 8.19 × 10−20 m2; and for the GGB specimen, it is 1.57 × 10−19 m2, slightly exceeding the saturated water permeability of SBG. In comparison, bentonite specimens have a Kw in the range 1.64 × 10−20–2.26 × 10−20 m2 (Guo et al., 2022). In the case of the granite–bentonite sample, the bentonite rapidly swells by absorbing water, effectively filling gaps within the granite. However, it is noteworthy that the Kw of intact granite is close to that of the granite–bentonite sample. The specimen with grooves shows the highest Kw, suggesting that even though bentonite can fill the gaps between the materials when saturated, its superior water retention properties lead to the granite being the preferred pathway for water infiltration. The process of carving grooves into the granite may lead to damage, making water more likely to pass through the micro-cracks in the granite.

3.2 Gas migration experiments

3.2.1 Gas migration in the intact granite sample

The pressure development at the inlet and the outlet of the saturated granite specimen under varying gas injection pressures is presented in Figure 6. The gas flow rate at the downstream side is calculated as the time derivative of the pd and denoted by Qg. Initially, the pressure at the inlet decreases nearly linearly and Qg gradually increases. As the pressure difference between the two ends of the specimen decreases, the rate of pressure reduction at the inlet also gradually decreases, eventually leading to equal pressures at both ends of the specimen. The oscillations in the upstream pressure are mainly caused by temperature changes. At a gas pressure of 1.5 MPa, the rate of pressure change at the outlet increases, resulting in a reduced pressure difference between the two ends and a subsequent decrease in the rate of pressure reduction at the inlet. Upon opening the valve at the outlet, the pressure difference between the two ends increases again, signifying that gas is entering the granite specimen and initiating drainage during the pressure rise phase. The increase in gas pressure leads to the expansion of internal microfractures. However, after a certain duration, the pressure at the inlet decreases to 1.14 MPa, indicating that the gradually decreasing gas pressure is insufficient to cause further expansion and the formation of continuous cracks within the granite specimen.

During the unloading stage, the value of Qg decreases. However, in comparison with the loading stage, the rate of pressure increase at the outlet is higher at the same gas injection pressure. When pu decreases to 0.5 MPa and the outlet valve is opened, the pressure at the inlet suddenly drops. This suggests that some damage has occurred within the granite during the pressure loading and unloading process. However, due to the low gas pressure, some of the expanded pores or fractures that might have formed could have recovered under the influence of the confining pressure, once again impeding the migration of gas.

The argon content at the outlet end of the intact granite is shown in Table 4. At a gas injection pressure of 0.5 MPa, the argon content fluctuates in the range of 3.38 × 10−4–7.79 × 10−5 mL/s. Over time, there is a noticeable decrease in the argon content. When the gas pressure is increased to 1 MPa, the argon content increases by an order of magnitude and water columns flow out from the outlet end. This phenomenon indicates that gas is gradually entering the sample, displacing pore water through a viscous–capillary two-phase flow. During this phase, the argon content undergoes more stable changes, indicating the formation of a consistent and stable gas migration pathway within the sample.

Table 4. Experimental results of IG at the outlet end.

Gas injection pressure (MPa)	0–120 min argon content (mL/s)	Experimental phenomenon
0.5	3.38 × 10−4–7.79 × 10−5	Water and gas mixture expelled at the downstream side
1.0	5.66 × 10−3–8.45 × 10−3	Water column expelled at the downstream side
1.5	6.49 × 10−2–7.22 × 10−2	Water and gas mixture expelled at the downstream side
1.0A	2.54 × 10−2–3.10 × 10−2	/
0.5A	1.02 × 10−3–2.39 × 10−3

During the unloading stage, there is no discharge of water at the outlet, and the argon content decreases, but remains higher than that during the loading stage. The partial pressures of argon and water are shown in Figure 7. At a pressure of 0.5 MPa, the water content at the outlet is slightly higher than that of argon. This is a result of water residue within the pipeline from the water injection process, potentially obstructing the flow of argon to the outlet. With the gradual increase of gas pressure, free water within the specimen is discharged, causing the argon's partial pressure to steadily increase to a stable level. This stabilized partial pressure of argon relative to water is two orders of magnitude higher, indicating that argon dominates the composition at the outlet.

3.2.2 Gas breakthrough in the grooved granite–bentonite sample

The results for the GGB specimen are shown in Figure 8. When pu = 1 MPa (Figure 8a), the pressure at the outlet end of the specimen increases slowly initially. However, after 6 h, it increases rapidly, and the pressure at both ends equilibrates after 11 h, with a minimal pressure difference of 7.3 kPa. When pu = 3 MPa, the pressure at the outlet end increases sharply, with equilibration around the 1-h mark. During the unloading stage, the rate of pressure reduction at the inlet decreases, and the time required to reach equilibrium between the two ends increases due to the decrease in gas injection pressure. When the gas pressure is reduced to 1 MPa, the pressure curve at the inlet fluctuates again. This suggests that the cracks within the specimen are continually opening and closing under the combined influence of surrounding pressure and gas pressure, leading to unstable gas flow.

The argon content at the outlet end is presented in Table 5. At an injection pressure of 1 MPa, the argon content ranges from 2.63 × 10−5 to 1.41 × 10−4 mL/s and water columns are discharged through the hose. As the gas pressure gradually increases to 3 MPa, the argon content increases to 10−2 mL/s, while water columns continue to be discharged through the hose. During the unloading stage, the argon content decreases with the pressure reduction, but remains higher than that during the loading stage at the same injection pressure. Additionally, a small volume of a gas–water mixture is released from the hose.

Table 5. Experimental results of GGB at the outlet end.

Gas injection pressure (MPa)	0–120 min argon content (mL/s)	Experimental phenomenon
1.0	1.41 × 10−5–2.63 × 10−4	Water column expelled at the downstream side
2.0	1.01 × 10−3–1.99 × 10−3
3.0	3.28 × 10−2–6.49 × 10−2
2.0A	3.22 × 10−3–5.84 × 10−2	Water and gas mixture expelled at the downstream side
1.0A	1.12 × 10−3–2.89 × 10−3	/

The partial pressures of argon and water are shown in Figure 9. Figure 9a indicates that the low gas pressure leads to delayed water discharge from the vertical tube, causing a blockage in the gas flow channel. Consequently, at an injection pressure of 1.0 MPa, the water partial pressure exceeds that of argon. Although the argon partial pressure increases with injection pressure, it eventually peaks before decreasing. This trend could be due to reduced air pressure, which fails to adequately support pore expansion and results in the partial closure of fissures within the specimen. During the unloading stage, the argon partial pressure initially increases and then stabilizes, signifying the establishment of a stable flow channel inside the specimen.

3.2.3 Gas breakthrough in the smooth granite–bentonite sample

The results obtained with the SGB specimen are shown in Figure 10. At a gas pressure of 1.0 MPa, the pressure at the outlet gradually increases to a final pressure of approximately 0.6 MPa. When the gas pressure is 2.0 MPa, the outlet pressure increases rapidly, leading to an equilibrium pressure difference of 6.9 kPa between the two ends after 3 h. Upon opening the outlet valve, the pressure at the inlet end rapidly decreases and eventually stabilizes at 0.34 MPa. As the gas injection pressure gradually increases, the duration of the rapid pressure increase at the outlet gradually shortens.

During the unloading stage, the pressure changes at both ends follow a pattern similar to that in the loading stage, with the value of Qg decreasing as the pressure drops. The time required for gas breakthrough increases, but at a gas injection pressure of 2.0 MPa, the pressure breakthrough occurs earlier than at the previous pressure. Additionally, Qg slightly increases, which can be attributed to the ongoing injection of gas, leading to further expansion of micro-fractures within the specimen. When the gas pressure is reduced to 1.0 MPa during injection, the pressure at the inlet rapidly decreases upon opening the outlet valve. This indicates the presence of connected gas migration pathways within the specimen.

The argon content at the outlet end is presented in Table 6. At a gas pressure of 1.0 MPa, water columns flow out through the valve outlet and the argon content decreases over time. However, no gas breakthrough is observed. When the gas pressure increases to 2 MPa, a water column flows out rapidly from the hose and the argon content stabilizes at about 10−2 mL/s. This indicates the formation of a connected gas pathway within the specimen and the occurrence of gas breakthrough. As the gas pressure increases further, the argon content at the outlet end shows an increasing trend. During the unloading stage, the argon content decreases as the gas pressure decreases and there is no water column outflow at the outlet.

Table 6. Experimental results of smooth granite tube (SGB) at the outlet end.

Gas injection pressure (MPa)	0–120 min argon content (mL/s)	Experimental phenomenon
1.0	3.39 × 10−3–2.56 × 10−4	Water column expelled at the downstream side
2.0	1.45 × 10−2–3.38 × 10−2
2.5	2.36 × 10−2–3.39 × 10−2
3.0	4.49 × 10−2–6.81 × 10−2
3.5	6.34 × 10−2–8.69 × 10−2	Mixture of water and gas bubbles expelled at the downstream side
3.0A	5.44 × 10−2–7.21 × 10−2
2.5A	5.61 × 10−2–6.33 × 10−2
2.0A	2.23 × 10−2–3.39 × 10−2
1.0A	5.39 × 10−3–6.77 × 10−3	/

The partial pressures of argon and water are shown in Figure 11. At an injection pressure of 1 MPa, the partial pressure of water initially exceeds that of argon, indicating that the water content is higher than the argon content during this period. However, a few hours later, the partial pressure of argon surpasses that of water, showing a first-rising-and-then-declining trend, which is attributed to the reduction in gas pressure. With the increase in injection pressure, gas breakthrough occurs and the argon partial pressure increases to 1.33 × 10−4 Pa. Subsequent experiments reveal that the pressure change curve is similar, suggesting that a connected gas migration pathway has formed within the specimen after gas breakthrough.

4.1 Dependence of gas permeability on gas injection pressure

It can be observed from Figures 6, 8, and 10 that when the pressure difference between the two ends of the specimen gradually reduces, the pressure decline at the inlet end slows down. When the valve at the outlet end is opened, the pressure difference between the two ends increases again, causing the rate of pressure decrease at the inlet to increase. The permeability of the samples follows this pattern. The results are shown in Figure 12, where k1 and k2 represent the gas permeability before and after the outlet valve is opened, respectively.

Figure 12a shows that for the three samples, k1 ranges between 10−19 and 10−21 m2. For the intact granite specimen, k1 increases from 2.47 × 10−20 to 1.76 × 10−19 m2 when the gas injection pressure increases from 0.5 to 1.5 MPa. Subsequently, k1 decreases to 3.28 × 10−20 m2 when the gas injection pressure is reduced to 0.5 MPa. For the GGB specimen, k1 increases from 1.68 × 10−20 to 4.47 × 10−19 m2 when the gas injection pressure increases from 1.0 to 3.0 MPa. When the gas injection pressure is decreased, the value of k1 also decreases to 2.19 × 10−19 m2. For the SGB specimen, k1 increases from 3.84 × 10−21 to 8.39 × 10−20 m2 during the loading phase and decreases to 3.37 × 10−21 m2 during the unloading phase. Notably, at low gas pressures (0.5–1.0 MPa), the SGB specimen has the lowest permeability before gas breakthrough. This observation underscores the effectiveness of utilizing a combination of granite and bentonite as a gas-blocking method.

As the gas pressure is gradually increased, gas breakthrough occurs, causing a rapid increase in permeability. Figure 12b shows the dependence of k2 on gas pressure. Notably, for the intact granite specimen, no gas breakthrough occurs at a gas pressure of 0.5 MPa and k2 is equal to k1. The first breakthrough is observed when pu = 1.0 MPa and k2 is 2.92 × 10−19 m2, which is proportional to the injection pressure. Upon unloading the gas pressure to 0.5 MPa, the value of k2 decreases to 1.85 × 10−19 m2, indicating a one-order-of-magnitude increase in permeability compared to the initial 0.5 MPa. This phenomenon is a result of the increased gas injection pressure, causing the continuous discharge of water within cracks. Simultaneously, it leads to the expansion of these minute cracks, ultimately resulting in an increased permeability of the specimen. For the GGB specimen, there is no change in k2 when the gas pressure is 1.0 MPa. Gas breakthrough occurs when the specimen is injected with a gas pressure of 2 MPa and k2 increases by one order of magnitude to 2.90 × 10−19 m2, indicating enhanced gas migration paths. During the unloading stage at a gas pressure of 3.0 MPa, both k1 and k2 have magnitudes of 10−19 m2. This is attributed to the presence of primary fissures in the assemblage of specimens. With an increase in gas pressure, water within the fissures is discharged, gradually forming gas migration pathways. The gas pressure is the dominant factor of gas migration and the permeability is positively correlated with the gas injection pressure.

The breakthrough pressure of the SGB specimen is also 2 MPa, but at the time of breakthrough, k2 reaches 1.26 × 10−18 m2, which is three orders of magnitude greater than the permeability before breakthrough. With the increase of gas pressure, k2 gradually increases to 1.03 × 10−17 m2. During the unloading stage, k2 gradually decreases to 1.68 × 10−18 m2, which is still in the same order of magnitude. It can be deduced that at this time, the gas migration paths have not changed significantly, and the primary influencing factor of permeability is the gas injection pressure.

The variation in Qg at different gas injection pressures is shown in Figure 13. At a gas injection pressure of 1 MPa, Qg for the CG, GGB, and SGB specimens is 0.043, 0.070, and 0.007 MPa/h, respectively. Gas breakthrough occurs only in the intact granite, but Qg for the GGB specimen is the highest. This implies that some damage occurs when grooves are carved into the inner wall of the granite, facilitating gas migration from the granite section. In both the granite and the GGB specimens, Qg tends to increase as the gas injection pressure increases. In contrast, the SGB specimen shows a declining trend in Qg when the gas pressure increases from 2 to 3 MPa. This could be attributed to the continuous discharge of water in the pipe at the outlet end as the injection pressure increases. Water has lower compressibility than gas, leading to the transfer of lower pressure to the pressure gauge's end.

4.2 Gas migration mechanisms under a flexible boundary

Table 7 shows the water permeability and the breakthrough pressure, and Figure 14 shows the Qg results for several specimens tested under flexible boundary conditions. Notably, the granite specimen shows the highest permeability and the lowest breakthrough pressure (1 MPa). This reflects its larger pore connectivity and reduced capillary pressure, which facilitate gas transport. In contrast, the homogeneous bentonite specimens and the bentonite specimen with an interface show similar water permeability, suggesting that the gaps between the bentonite could be effectively sealed through extended water injection and saturation. However, during gas injection, the interface-containing bentonite displays significantly higher Qg compared with the homogeneous bentonites, resulting in lower pressure and shorter time required for gas breakthrough. This highlights the higher permeability of bentonite with interfaces, which can act as gas transport channels.

At high confining pressure, exemplified by sample WI, gas migration becomes notably more challenging. Despite increasing the injection pressure to 10.7 MPa, no gas breakthrough is observed and the value of Qg remains as low as 10−4 MPa/h. Under lower pressure conditions, the pore space of the sample contracts due to the application of external pressure, resulting in a diminished growth rate at the outlet. Under this condition, the dissolution–diffusion mechanism becomes the dominant process governing gas migration. As the gas pressure gradually increases, the value of Qg increases, accompanied by the appearance of gas bubbles at the outlet. This observation suggests the concurrent presence of a viscous–capillary two-phase flow and localized pore expansion seepage as the two primary mechanisms governing gas migration.

4.3 Gas migration mechanisms under a semirigid boundary

Under a semirigid boundary, the influence of the swelling pressure on gas migration is substantial. In previous research, aluminum tubes (grooved/smooth) were used in both swelling pressure and gas breakthrough experiments on bentonite–sand specimens. The specimens showed an average swelling pressure of 7.3 MPa (Liu et al., 2014). In the case of the specimen with a smooth tube, the gas breakthrough pressure equaled or slightly exceeded the swelling pressure. In contrast, no gas breakthrough occurred in the sample with a grooved tube, even when the gas pressure was increased to 10 MPa. This suggests that gas migration through grooved or meandering interfaces is considerably more difficult, whereas smooth interfaces provide channels for gas flow.

Table 8 shows the water permeability and the gas breakthrough pressure and Figure 15 shows the Qg of several specimens tested under semirigid boundary conditions. The breakthrough pressure for both GGB and SGB specimens is 2 MPa, but it is worth noting in Figure 15 that at gas breakthrough, Qg for the SGB specimen exceeds Qg for the GGB specimen by two orders of magnitude, indicating that the migration paths of the gases are different during breakthrough. The argon content also increases by one order of magnitude. Therefore, it can be suggested that the interface between the smooth inner wall of the granite and the bentonite serves as a preferential gas migration pathway. This implies that the migration path of the GGB specimen is analogous to that of the intact granite specimen, that is, predominantly through the granite. This is corroborated by the higher saturated water permeability of the grooved inner wall specimen, which suggests that the process of formation of the grooves might have created new fractures.

In previous experiments (Liu et al., 2015), bentonite was placed inside both smooth and grooved COx argillite tubes, resulting in breakthrough pressures of 7.5 and 5.0 MPa, respectively. Notably, these breakthrough pressures are higher than those observed in the granite–bentonite specimens in this study. Additionally, the value of Qg in those experiments ranged from 1.93 × 10−5 to 2.40 × 10−3 MPa/h, which is considerably lower than the values reported in this study. These differences can be attributed to the inherent differences between granite and argillite. Argillite contains numerous clay minerals like montmorillonite, kaolinite, and ilmenite, which are hydrophilic and prone to swelling and disintegration upon water exposure. The clay minerals can trap more water compared with quartz and feldspar in granite. Consequently, the internal fissures within the argillite gradually heal over time. In contrast, granite lacks this property. Furthermore, granite has higher strength compared to argillite. Thus, under identical confining pressures, the cracks in argillite are more susceptible to compression, resulting in reduced width and gas migration pathways. Consequently, when compared to bentonite–argillite samples, bentonite–granite samples demonstrate poorer gas sealing performance.