Introduction

Natural gas hydrate (NGH) is a new energy source widely distributed on submarine slopes and permafrost zones, where gas molecules are encased in polyhedral cages constructed from water molecules connected by hydrogen bonds1. Due to its huge reserves, high energy density, and clean combustion, is considered a promising potential energy resource that plays a vital role in the future climate environment2.

Since the 1980s, a worldwide wave of research on NGH resource investigation and development technologies has arisen. So far, more than 230 NGH-rich areas have been discovered in 79 countries, and approximately 97% of the global NGH resources are deposited in submarine sediments3. Due to the complex marine environments, geological conditions, and phase behavior, NGH development remains at the exploratory stage, with only four marine NGH production trials conducted in the Nankai trough of Japan4 and Shenhu Area of the South China Sea5. In contrast to the sandy deposits in the Nankai trough, Shenhu hydrates are hosted in clayey-silty sediments with low reservoir permeability and permeable boundary layers6. Such host environments contain more than 90% of the global NGH resources and are known as challenging hydrates7. This challenge is twofold: first, the low reservoir permeability significantly increases the difficulty of NGH dissociation and gas recovery2,7,8,9,10, and second, the permeable boundary layers exacerbate the risk of methane leakage11,12,13,14,15,16. As a result, searching for effective and reliable extraction methods has been a pressing issue.

The general idea of NGH extraction is to break its phase equilibrium state in situ and subsequently recover the decomposed gas, with depressurization, inhibitor injection, thermal stimulation, and gas replacement as primary methods. Of these, depressurization is generally considered the most promising because of its relative ease of implementation, with productivity controlled by pressure relief potential as well as heat and mass transfer17. However, for challenging hydrates, pressure propagation, fluid flow, and enthalpy available are suppressed by the low reservoir permeability and open boundaries. Su et al. (2012) found that lowering the permeability of boundary layers promoted pressure drop propagation and that the increased gas volume improved the gas-phase relative permeability, resulting in higher productivity10. Sun et al. (2015) pointed out that permeable underburden was superior to that of permeable overburden as warmer bottom water intrusion favored heat supply18. Bhade and Phirani (2015) found that the NGH reservoir with an unconfined aquifer resulted in ineffective depressurization19. Li et al. (2021) concluded that the completion section should contain the underburden at a reservoir permeability of 1–100 mD20. Ning et al. (2022) found that reducing the permeability of boundary layers strengthened the synergistic pressure-relief effect of multi-branch wells and increased long-term gas production21. The above-mentioned studies indicate that the boundary effect can not be ignored in challenging hydrate extraction, that is, permeable boundary layers are conducive to the environmental enthalpy available, but also weaken pressure propagation and result in high water production, a bottom-water intrusion phenomenon that should be avoided in conventional gas reservoirs. As a result, challenging hydrate extraction by depressurization alone makes it hard to reach the desired production performance.

The combination of depressurization and thermal stimulation has been demonstrated to be a more efficient extraction method. The heating means include fluid-free methods such as electricity and magnetism, as well as fluid-containing methods such as thermal fluid injection. The heating range of fluid-free methods is limited due to the low thermal conductivity of sediments, whereas the fluid-containing methods with convective heat transfer are considered more efficient22. Particularly, injection-production well patterns such as two-, three-, and five-spot exhibit tantalizing productivity and have been proposed for commercial development by Japan23,24. This method may be applicable to sandy NGH reservoirs; however, for challenging hydrates, the decomposed gas and injected fluid may migrate to the boundary layers driven by the high injection pressure, resulting in ineffective thermal stimulation and greatly increasing the risk of methane leakage. Numerical investigation on Shenhu hydrates showed that the gas escaping under a unit horizontal well section reached 15080 m3 at a well spacing of 100 m16. In our previous studies, the escaped gas was successfully recovered by placing two capture wells at the boundary layers. However, due to the small spacing between the capture and injection wells, premature water flooding greatly weakened the thermal stimulation effect and resulted in high water yield16,25. Consequently, an urgent issue is how to enhance gas productivity while controlling water and avoiding gas leakage.

Water plugging is an effective technology for water control in the development of oil and gas reservoirs. The general technology idea is to block the water-bearing layer or water flow channel with sealing and plugging materials such as cement, gels, polymers, fibers, and Fuzzy-ball, to reduce the water-phase relative permeability26,27,28. In the case of challenging hydrates, the open boundary conditions are similar to those of oil and gas reservoirs containing edge or bottom water. A viable option is to create artificial barriers by injecting sealing and plugging materials at the boundaries, thus cutting off the hydraulic connection between the reservoir and aquifer29,30,31,32. Furthermore, the low-permeability sealing zones could increase the flow resistance of decomposed gas and hot water to the boundary layers, which may be a promising technology for the safe and efficient development of challenging hydrates.

This study proposed a novel extraction method combining boundary sealing and hot water injection to address issues such as CH4 leakage, boundary water intrusion, insufficient reservoir sensible heat, as well as poor gas production in challenging hydrate development, and the stimulation potential of boundary sealing or/and hot water injection was numerically evaluated, with Shenhu hydrates at the SH2 site as the research object. Furthermore, the effects of key factors including well pattern types, boundary permeability (kb), injection pressure (Pi), and hot water temperature (Ti) on hydrate decomposition and gas–water production were investigated. The findings of this study could provide important guidance for the commercial development of challenging hydrates.

Numerical modeling

Geological background

Shenhu Area, located in the northern land slope of the South China Sea, with a seawater depth of 900–1500 m, is one of the most active areas for marine hydrate exploration and exploitation. Since 2007, three cruises have been conducted, and numerous hydrate samples have been recovered, with hydrate saturation (Sh) of 20–48% and CH4 content exceeding 99.7%6,33,34,35. Among them, the NGH reservoir at the SH2 site has the highest Sh value of 47.3% and a maximum reservoir thickness of 43 m, and thus has development potential and represents the typical storage characteristics of Shenhu hydrates36. The core analysis further revealed that the permeabilities of the reservoir-cover sediment were 10 mD, and consisted of less than 2% sand, 60–80% silt, and 15–45% clay37,38. As a result, challenging hydrates at the SH2 site were taken as the research object to explore the stimulation effect of boundary sealing and hot water injection.

Development plans and simulation scheme

Horizontal well patterns were used here to obtain considerable drainage area. In addition, due to the limited thickness of the hydrate-bearing layer (HBL), arranging one or two wells vertically can fulfill the NGH decomposition requirements. Consequently, the mainstream horizontal well patterns including three-spot and five-spot were adopted22,39,40, as shown in Fig. 1.

Figure 1
figure 1

Schematic diagram of development plans.

In plan I (Fig. 1a), a three-spot well pattern was used, wherein all horizontal wells were located in the middle of the hydrate-bearing layer (HBL) and producing at a constant bottom-hole pressure of 4.5 MPa, which was consistent with field trial production condition4.

In plan II (Fig. 1b), the difference from plan I is that the middle horizontal well was used for hot water injection, to evaluate the thermal stimulation potential.

In plans III and IV (Fig. 1c,d), low-permeability sealing zones were created at the boundaries by using sealing and plugging materials. The thickness of the sealed boundaries was assumed to be 1 m, and the permeability was set to 0.0001–1 mD to analyze the stimulation effect of boundary sealing.

In plans V–VIII (Fig. 1e–h), various well patterns were designed, including three-spot with middle injection and up-down production (Fig. 1e), up injection and down production (Fig. 1f), down injection and up production (Fig. 1g), as well as five-spot with central injection and surrounding production (Fig. 1h), to obtain the ideal well pattern.

The specific simulation scenarios are listed in Table 1.

Table 1 Simulation scenarios.

Numerical code, initial and boundary conditions

TOUGH + HYDRATE (T+H), developed by Lawrence Berkeley National Laboratory, is a code for the simulation of hydrate geologic systems. It is written in standard FORTRAN 95, and can be run on computing platforms (Macintosh, workstation, PC) for which such compilers are available41. Currently, T + H has been widely used in the simulation of various CH4-hydrate deposits because of its powerful function in describing methane hydrate phase behavior, heat exchange, and fluid flow in porous media. Moreover, its reliability has been fully examined42,43,44,45,46, and thus was employed in this study.

The top and bottom of the model were set as constant temperature–pressure boundaries as sufficient cover thickness was considered. Since the sediments are permeable, it is assumed that pore seawater is connected, allowing good pressure–temperature transformation. As a result, reservoir pressure and temperature can be derived from the hydrostatic pressure gradient and ground temperature gradient, respectively. Other model parameters are given in Table 2.

Table 2 Model parameters.

Result analysis

Stimulation by boundary sealing

Figure 2 shows the production dynamics of depressurization at various kb values. The gas release rate from hydrate decomposition (Qd) decreased gradually at kb of 0.1–10 mD, whereas at kb of 0.0001–0.01 mD, it showed a rapid rise, followed by (~ 150 days) a sharp decline and then (~ 1440 days) a slow decline (Fig. 2a). The maximum Qd increased from 46 to 168 m3/d as kb decreased from 10 to 0.0001 mD, indicating that reducing kb was conducive to hydrate decomposition. This mechanism can be revealed by the reservoir physical field distribution in Fig. 3. As can be seen, the low-pressure zone and the magnitude of the pressure drop in the HBL increased with decreasing kb (Fig. 3a–f), especially when kb reduced to 0.01 mD, the whole HBL was effectively depressurized, resulting in the hydrate dissociation along the borehole changed to volumetric dissociation (Fig. 3g–x). The stimulation effect of boundary sealing was mainly effective in the first 1440 days. After that, the temperature of the HBL decreased significantly due to the heat absorption of hydrate dissociation (Fig. 3g–l), and thus, hydrate decomposition was inhibited by limited reservoir sensible heat. The hydrate decomposition rate (Rd, defined by the ratio of the mass of the decomposed hydrate to the mass of the original hydrate) over 5760 days increased from 51 to 69% as kb decreased from 10 to 0.0001 mD (Fig. 2b). Therefore, hydrate decomposition is encouraged by sealing boundaries, and reducing kb to less than 0.001 mD is recommended here.

Figure 2
figure 2

Production dynamics of depressurization at various kb values: (a) gas release rate Qd; (b) hydrate decomposition rate Rd; (c) gas production rate Qg; (d) cumulative gas production Vg; (e) water production rate Qw; (f) gas-to-water ratio Rgw.

Figure 3
figure 3

Reservoir physical field distributions for 1440 days of depressurization: (a–f) pressure P; (g–i) temperature T; (m–r) hydrate saturation Sh; and (s–x) gas saturation Sg.

The evolution of the gas production rate (Qg) was similar to that of Qd (Fig. 2c). One difference is that there is no initial rise phase of Qg at kb of 0.01 mD due to the retardation of gas production. As kb decreased from 10 to 0.0001 mD, the maximum Qg increased from 50 to 75 m3/d. After 1440 days, Qg was the lowest at kb of 0.0001 mD as the whole depressurized HBL cannot provide enough pressure difference for gas flow, resulting in substantial gas hold-up (Fig. 3j–x). The cumulative gas production (Vg) over 5760 days increased first and then decreased with decreasing kb, with a maximum value of 14.25 × 104 m3 at kb of 0.01 and 0.001 mD (Fig. 2d). Consequently, there exists a critical kb value for gas enhancement, and 0.01–0.001 mD is suggested.

The water production rate (Qw) increased rapidly and then (~ 2880 days) slowly at kb of 0.1–10 mD, whereas it decreased sharply and then stabilized at kb of 0.0001–0.01 mD (Fig. 2e). The sources of produced water include free water in the HBL, hydrate dissociation water, and boundary water in the boundary layers. It can be concluded that boundary water intrusion was inhibited by decreasing kb and thus reducing water yield. The gas-to-water ratio (Rgw, defined by the ratio of cumulative gas production to total water production) decreased at kb values of 0.1–10 mD, with a maximum value lower than 15; whereas increased linearly at kb of 0.0001 mD, finally reaching a value of 142 (Fig. 2f). Consequently, boundary sealing is effective for water control.

Stimulation by hot water injection and boundary sealing

Figure 4 exhibits the production dynamics of injection-production scenarios at various kb values. As can be seen, the evolution of Qd at kb of 0.1–10 mD presented four phases (Fig. 4a): first, with the dissociation of hydrate around the injection well, the hot water spread area increased gradually, leading to a gradual increase in Qd; when the thermal decomposition front extended to the boundary layers, the high injection pressure was released (Fig. 5a–c); then entered the second phase (~ 360 days), and Qd gradually decreased due to a substantial amount of hot water transport to the boundary layers (Fig. 5g–i); in the third phase (~ 800 days), Qd was stabilized at about 40 m3/d with stable thermal stimulation; once the hot water broke through to production wells, (a phenomenon known as water flooding in water-driven reservoirs and corresponding to the breakthrough time Tbt), it entered into the fourth phase (~ 4000 days), and Qd decreased dramatically as hypertonic channels were formed (significant increase in effective reservoir permeability after hydrate between injection and production wells was decomposed). The difference is that when kb was reduced to 0.01 mD, Qd decreased initially as the stagnant gas inhibited the hot water injection. Subsequently (~ 200 days), Qd increased with the injection of hot water. Interestingly, when the thermal decomposition front reached the boundary layers, the high injection pressure (Fig. 5d–f) was maintained and the hot water loss was avoided (Fig. 5j–l), and therefore, Qd continued to increase after a slight fluctuation until it reached the peak value (80 m3/d). After that (~ 3000 days), Qd decayed sharply after Tbt. Overall, the influence of kb on the final Rd was limited (Fig. 5m–r) because hydrate decomposition cycles (Thd) were all approximately 3500 days (Fig. 4b). Thus, boundary sealing has a significant influence on hydrate decomposition behavior in the injection-production scenarios but hardly affects Thd because it mainly controlled by the thermal breakthrough.

Figure 4
figure 4

Production dynamics of injection-production scenarios at various kb values: (a) Qd; (b) Rd; (c) Qg; (d) Vg; (e) Qw; (f) Rgw.

Figure 5
figure 5

Reservoir physical field distributions for 2880 days of depressurization and hot water injection at various kb values: (a–f) P; (g–i) T; (m–r) Sh; (s–x) Sg.

The evolution of Qg at different kb values was similar, as shown in Fig. 4c. Initially, Qg was stabilized at 18 m3/d and was contributed by depressurization. Subsequently, thermal decomposition gas was driven to the production wells (~ 1440 days), and thus Qg increased sharply. The maximum Qg at kb of 0.0001–0.01 mD was higher than 95 m3/d, while it was lower than 75 m3/d at kb of 0.1–10 mD. The mechanism for this is twofold: first, the high injection pressure (Fig. 5a–f) improved the water driving force; and second, hot water loss (Fig. 5g–l) and gas escape (Fig. 5s–x) was avoided. The final Vg initially increased and then decreased with reducing kb, with a maximum value of 24.20 × 104 m3 and a gas recovery period (Tgr) of 5600 days at kb of 0.0001 mD (Fig. 4d). Thus, hot water loss and gas escape were addressed by reducing kb to 0.0001 mD.

With Tbt as the node (~ 2880 days), the change in Qw can be divided into low- and high-yield phases afterward, as shown in Fig. 4e. For Rgw, it first decreased and then increased, with a low level lower than 15 at kb of 0.1–10 mD; however, when kb reduced to 0.01 mD, it increased after the thermal decomposition gas was produced, and decreased after Tbt (Fig. 4f). The final Rgw increased from 10 to 26 as kb reduced from 10 to 0.0001 mD. As a result, boundary sealing is still effective for water control in injection-production scenarios.

Influence of well patterns

Figure 6 shows the production dynamics under various injection-production well patterns. In the three-spot well patterns, Qd, Qg, and Qw were the higher in Scene No.12, and Tbt was also the shortest (Fig. 6a,c,e). The mechanism is that the pressure gradient between the injection and production wells is greater under a smaller well spacing, which is more favorable for hot water penetration, hydrate decomposition, as well as gas and water recovery. Compared with the three-spot, the five-spot was more conducive to pressure propagation and hot water penetration, which further promotes hydrate decomposition and gas recovery. The Thd and Tgr of the five-spot were 3878 and 4890 days, shorting at least 903 and 680 days, respectively, compared to the three-spot (Fig. 6a–d). The final Rgw of the five-spot was 30, at least 5 higher than those of the three-spots (Fig. 6f). Therefore, the five-spot well pattern is the most ideal in the current simulations.

Figure 6
figure 6

Production dynamics under various injection-production well patterns: (a) Qd; (b) Rd; (c) Qg; (d) Vg; (e) Qw; (f) Rgw.

Influence of injection pressure

Figure 7 shows the production dynamics at various Pi values. As can be seen, when Pi increased from 17 to 22 MPa, the maximum Qd increased from 60 to 116 m3/d, Tbt decreased from 3750 to 2640 days, and Thd shortened from 4870 to 3460 days (Fig. 7a,b); the maximum Qg increased from 75 to 140 m3/d, and Tgr decreased from 5940 to 4340 days (Fig. 7c,d); the maximum Qw increased from 7.5 to 11, and the final Rgw decreased from 32 to 28 (Fig. 7e,f). The stimulation mechanism is that improving Pi increases the hot water injection rate and water-driven force, thereby favoring reservoir heating and gas–water recovery. Consequently, increasing Pi could enhance the extraction efficiency, but simultaneously induce high water production.

Figure 7
figure 7

Production dynamics at various Pi values: (a) Qd; (b) Rd; (c) Qg; (d) Vg; (e) Qw; (f) Rgw.

Influence of hot water temperature

Figure 8 shows the production dynamics at various Ti values. Similar to increasing Pi, increasing Ti was also conducive to pressure propagation, reservoir heating, hydrate dissociation, and gas–water recovery. Specifically, when Ti increased from 30 to 80 ℃, the maximum Qd increased from 59 to 120 m3/d, Tbt decreased from 4215 to 2782 days (Fig. 8a), and Thd shortened from 6036 to 3382 days (Fig. 8b); the maximum Qg increased from 70 to 137 m3/d (Fig. 8c); Tgr decreased from 6811 to 4415 days (Fig. 8d); and the maximum Qw increased from 4.9 to 12 (Fig. 8e). Different from increasing Pi, increasing Ti provided a more significant enhancement of gas production, resulting in an increase in the final Rgw from 21 to 33 as Ti increased from 30 to 80 ℃ (Fig. 8f). The mechanism is that water injection is not significantly strengthened due to the constant Pi. Therefore, increasing Ti helps to control water while improving extraction efficiency.

Figure 8
figure 8

Production dynamics at various Ti values: (a) Qd; (b) Rd; (c) Qg; (d) Vg; (e) Qw; (f) Rgw.

Discussion

From the analysis of the above results, it is found that boundary sealing prevented boundary water intrusion, and the hydrate volumetric decomposition was realized by the effective depressurization of the whole reservoir. However, the limited environment sensible heat was insufficient for sustained hydrate decomposition, and a substantial amount of gas was retained in the reservoir due to the lack of gas driving force. Hot water injection alone can provide adequate heat for hydrate decomposition. However, when the hot water breaks through to the open boundaries, the injected hot water and free gas will leak into the boundary layers under the drive of high injection pressure. This is detrimental to energy utilization and poses a serious environmental safety risk. The combination of boundary sealing and hot water injection realized a piston-like water-driven gas extraction mode, addressing the issues of insufficient reservoir sensible heat, water intrusion, as well as gas retention and escape. Here, the synergistic effect on gas enhancement and water control is analyzed based on average Qg and Rgw at Vg of 20 × 104 m3, as presented in Fig. 9. When only boundary sealing is applied, average Qg first increases and then decreases with a maximum value of 19 m3/d at kb of 0.1 mD, and Rgw increases from 1 to 133 as kb decreases from 10 to 0.0001 mD, indicating that boundary sealing is effective for water control but not always favorable for gas recovery, thereby exhibiting a low-gas and low-water production pattern; when only hot water injection is used, average Qg increases from 11 to 45 m3/d, and Rgw increases from 1 to 10. This indicates that the gas enhancement is more significant than the increase of water yield, presenting a high-gas and high-water production pattern; when both boundary sealing and hot water injection are adopted, the average Qg increases first and then decreases with a maximum value of 53 m3/d at kb of 0.0001 mD, and Rgw increases from 10 to 26 as kb decreased from 10 to 0.0001 mD, suggesting that sealed boundaries further improve the thermal stimulation potential while reducing water yield, thereby resulting a high-gas and low-water production pattern. The values of average Qg in Scene Nos. 6 and 7 are 7 and 45 m3/d, their sum (52 m3/d) is lower than that of Scene No. 12 (53 m3/d). Consequently, hot water injection and boundary sealing play a synergistic effect like “1 + 1 ≥ 2” on gas enhancement.

Figure 9
figure 9

Average Qg and Rgw at Vg of 20 × 104 m3 in Scene Nos. 1–16.

In addition, for three-spot well patterns, average Qg and final Rgw are the highest (53 m3/d and 26) when the injection and production wells are located in the middle of the HBL. However, they were lower than those (58 m3/d and 30) of the five-spot, indicating the five-spot was better. Considering the multi-well synergistic effect25, increasing the complexity of the well pattern such as seven-spot and nine-spot may further improve the extraction efficiency, but simultaneously increase the drilling cost. In particular, the influence of well location on the multi-phase fluid flow is more complex, and therefore, further systematic investigation is needed.

To investigate the influence rule of injection parameters on gas–water production, the relationships of Pi and Ti with average Qg and Rgw at Vg of 20 × 104 m3 are fitted, respectively, as presented in Fig. 10. As can be seen, when Pi increases from 17 to 22 MPa, average Qg increases linearly from 47 to 66 m3/d, whereas Rgw decreases linearly from 32 to 29, indicating that water yield was enhanced more significantly. Both average Qg and Rgw increase logarithmically with increasing Ti from 30 to 80 ℃, from 40 to 65 m3/d, and from 21 to 33, respectively, indicating that gas yield is enhanced more significantly. As a result, high-pressure and high-temperature injection is conducive to gas recovery, while low-pressure and high-temperature injection favors water control. However, overall, compared to the enhancement of Pi on the average Qg, its decrease in Rgw is very limited, with fitting coefficients of 3.768 and 0.798, respectively. Consequently, high-pressure and high-temperature injection is recommended for gas enhancement and water control.

Figure 10
figure 10

Relationships of Pi and Ti with (a) average Qg and (b) Rgw at Vg of 20 × 104 m3.

Conclusions

This research proposes a novel reservoir stimulation method combining boundary sealing and hot water injection for challenging hydrates, and the stimulation effect was numerically evaluated. The following conclusion can be drawn.

  1. 1.

    Boundary water intrusion is effectively addressed by sealing open boundaries, thus promoting pressure drop propagation and realizing the volumetric decomposition of hydrates. However, its stimulation validity is limited (~ 3 years), after which a substantial amount of decomposed gas is trapped in the reservoir due to insufficient formation energy and environmental heat, resulting in productivity not always increasing with the decrease of boundary permeability.

  2. 2.

    Hot water loss and gas escape are avoided by decreasing the boundary permeability to 0.0001 mD. Furthermore, a piston-like water drive similar to secondary recovery is realized, thereby drastically improving the extraction efficiency. Furthermore, compared with the three-spot well patterns, the five-point is more conducive to gas enhancement and water control.

  3. 3.

    The enhancement of water production by increasing injection pressure is more significant, leading to a slight increase in the gas–water ratio; whereas the enhancement of gas production by increasing hot water temperature is more significant, resulting in a decrease in the gas–water ratio. As a whole, high-temperature and high-pressure injection is recommended.

  4. 4.

    This study offers a novel promising development plan for marine challenging hydrates. However, it is important to point out that to obtain universal rules, the current simulations were performed under relatively ideal geoengineering conditions, considering only a single HBL and three-spot and five-spot well patterns. Given the heterogeneity of hydrate reservoir, the effectiveness of boundary sealing, and other unconsidered complex well patterns such as seven-spot and nine-spot potentially affect mass and heat transfer behavior, further research in development plan optimization is needed, with a comprehensive consideration of these issues.