As shown in Figure 1, the HPRWJ tool consists of filters, a one-way valve, centralizer, rotation controller, and self-oscillation nozzles. While constructing NWAFs using the HPRWJ technology, the tool is lowered into the borehole at the perforation intervals using the connecting tubing. Then, high-pressure clean water (or water with added sands, expansion inhibitor, or clay stabilizers) is pumped through the tubing, filters, one-way valve, and nozzles to generate jets. Simultaneously, the nozzles are rotated by the rotation controller and the rotary speed is adjusted by the dampener. The impurities in the perforations become loose and will be removed with the back flow under repeated jet impacts. Then axisymmetric cylindrical artificial fractures are obtained near the wellbore in the selected position. The tool is able to move up and down to the next position by the draw work of the tubing. Therefore, a series of axisymmetric cylindrical artificial fractures are obtained very easily. Furthermore, fracture shapes can be kept by adding sand to the jet water or replacing the jet water with coagulable and permeable grouting mixtures (Yuan et al., 2021a; Li et al., 2021). This article focuses on the impacts of NWAFs on the hydrate production performance, the geomechanical effect on fractures’ permeability, and the geomechanical response associated with the fractures that are not considered during gas production. However, the geomechanical effect and response will be further investigated.

The Shenhu Area is the first NGH reservoir exploited in China. It is located in the middle part of the northern slope between the Xisha Trough and the Dongsha Islands in the South China Sea (Figure 2) (Wu et al., 2011). During GMGS-1 in 2007, three of the five sites (sites: SH2, SH3, and SH7) cored in this area were detected of having high concentrations of NGH. The thickness of the hydrate-bearing sediments (HBS) is 10–43 m and the bottom of the HBS is situated directly above the base of the gas hydrate stability zone, which is suitable for exploitation (Wang et al., 2011). Core sampling and well logging analysis of the SH7 site indicated that the NGH exist at depths of 155–177 m below the seabed, with a water depth of 1,108 m. The sediment porosity and NGH saturation are 33–48% and 20–44%, respectively (Li et al., 2011a). The overburden and underburden layers are similar to HBS but lack of hydrates sedimentary. In situ measurements show that the bottom temperature is approximately 3.7°C with a geothermal gradient of 43.3°C∙km⁻¹. The main component of the gas is methane (96.10%–99.91%), with other gases mainly consisting of C₂H₆ and C₃H₈. In addition, the sediments are mainly composed of silty clay and clay silt with a millidarcy-range intrinsic permeability.

In this study, the TOUGH + HYDRATE V1.5 software was used to investigate the effects of NWAFs on the gas production of NGH reservoirs (Moridis, 2014). The accuracy of this software has been tested by massive studies at different hydrate sites around the world. The following assumptions were made when using this simulator (Moridis, 2014): (1) Darcy’s law is valid in the model domain, (2) the mechanical dispersion of dissolved gases and inhibitors is ignored, (3) the movement of the geologic medium is not described, (4) the aqueous phase is not allowed to disappear when salts are present, (5) the dissolved inhibitors do not affect the thermophysical properties of the aqueous phase, and (6) the inhibitor is a non-volatile component. The governing equations for its multiphase flow and heat convection and conduction processes are given as follows.

1) Components and phases:

Four phases, namely, solid-hydrate (H), aqueous (A), gaseous (G), and ice (I) phases are continuously distributed in the pore of HBS. The four components, including water (w), methane (m), hydrate (h), and salt (i), are partitioned among four possible phases. For simplicity, the κ and β indicators are used in the subsequent equations to define these components and phases, respectively.

According to the thermodynamic state of HBS, the quantity of formed hydrate or the quantity of released methane gas is determined by the following reaction: C H 4 + N H H 2 O = C H 4 ⋅ N H H 2 O , (1)

2) Mass balance:

The governing equation for the flow of multicomponent fluid mixtures determined based on the mass balance is as follows: d d t ∫ V n M κ d V = ∫ Γ n F κ ⋅ n d Γ + ∫ V n q κ d V , (2) where M ^κ , F ^κ , and q ^κ are the mass accumulation, flux, and source/sink ratio of component κ, respectively.

The mass accumulation term M ^κ is determined by M κ = ∑ β = A , G , H , I φ S β ρ β X β κ , (3)

The mass flux term F ^κ includes the contribution from the aqueous and gaseous phases and the equation is as follows: F κ = ∑ β = A , G ( F β κ + J β κ ) , (4)

For the aqueous phase, F A κ is calculated by multiphase Darcy’s law using the following equations: F A κ = X A κ F A , F A = − k k r A ρ A μ A ( ∇ P A − ρ A g ) , (5)

For the gaseous phase, F G κ is affected by the Klinkenberg function and is determined by F G κ = X G κ F G , F G = − k ( 1 + b P G ) k r G ρ G μ G ( ∇ P G − ρ G g ) , (6)

The diffusive mass flux of component κ (κ = m,i) is calculated using Fick’s law. It is defined as J β κ = − φ S β τ β D β κ ρ β ∇ X β κ , (7)

3) Energy balance:

The governing equation for the heat flow determined based on the energy balance is as follows:   d d t ∫ V n M θ d V = ∫ Γ n F θ ⋅ n d Γ + ∫ V n q θ d V , (8) where θ denotes the heat component and M ^θ , F ^θ , and q ^θ are the heat accumulation, flux, and source/sink ratio, respectively.

The heat accumulation term includes contributions from the rock matrix and all the phases, and the equation is as follows: M θ = ( 1 − φ ) ρ R C R T + ∑ β = A , G , H , I φ S β ρ β U β + Q d , (9)

The heat flux term includes conduction and advection, and the equation is as follows: F θ = − λ θ ∇ T + ∑ β = A , G h β F β , (10)

Reaction heat Q_d of hydrate dissociation is calculated by Q d = { Δ ( φ ρ H S H Δ U H ) f o r   e q u i l i b r i u m   d i s s o c i a t i o n Δ Q H ⋅ U H   f o r   k i n e t i c   d i s s o c i a t i o n                                                 , (11)

This study investigated the stimulated production performance with special emphasis placed on NWAFs and compared these results with the production potential determined using a conventional vertical well. To this end, two different cases were defined as follows:

1) Case 1: gas production using conventional vertical well design, as shown in Figure 3B.

According to Li et al. (2002), when using the HPRWJ technology to construct NWAFs, the impact pressure, which is the pressure that the jet impacts at the sediment surface, will increase linearly with the increase of pump pressure, and the impact pressure is about 80–90% of pump pressure. Under the pump pressure of 20 MPa, the fracture’s depth and height in the consolidated sediment can be up to 1 m, 500 mm respectively and the impact pressures still reach 3.0 MPa when the radial distance increases up to 1 m. Moreover, the maximum pump pressure can reach 50 MPa, which means the jet impacting distance can reaches far more than 1 m. Otherwise, according to the study conducted by Burland (1990) and Wei et al. (2021), the shear strength of hydrate-bearing sediments and consolidated sediments is 0.5 MPa and more than 5 MPa, respectively. The bulk modulus and mechanical strength of HBS are both much lower than those of the consolidated sediment. Therefore, it is easier to create NWAFs in HBS. In this article, the fracture depth, height, and spacing in HBS is assumed to be 3 m, 500 mm, and 3 m, respectively, as shown in Figure 3C.

Figure 3A shows the schematic of the NGH reservoir model used in this study, which was established based on the SH7 site in the Shenhu area. According to the previous research results from Li Gang et al. at this site, an axisymmetric cylinder with a radius of 500.1 m and a thickness of 82 m was employed to denote the model domain, which consisted of one HBS area, one permeable top layer (the overburden), and one permeable bottom layer (the underburden), whose thickness was 22, 30 and 30 m along the vertical direction, respectively (Li et al., 2011b). According to previous studies of Moridis and Kowalsky (2007a) and Moridis et al. (2007b), the overburden and underburden layers with a thickness of 30 m each were sufficient to simulate the heat and pressure transfer during gas production.

Additionally, the inner and outer boundaries are also very important for gas production prediction (Sun X. et al., 2019). The production well with a radius of r_w = 0.1 m was located in the center of the cylinder. The perforated interval was 12 m long and located in the middle part of the HBS which can prevent free water in the overburden and underburden layers from entering into the wellbore at the beginning of the simulation (Su et al., 2010). The bottomhole pressure was set to 4.5 MPa which is similar to the field test condition in the Nankai Trough (Yamamoto and Dallimore, 2008b; Fujii et al., 2013). The top and bottom boundaries were set to boundaries with constant temperature and pressure. Meanwhile, the outside of the model (r_max = 500.01 m, the thinkness of 0.01 m improves the boundary definition accurate) was treated as a boundary where no exchange of heat and flow occurred. The wellbore porosity, permeability, and capillary pressure are assumed to be 1.0, 5.0 × 10^–9 m² and 0 respectively, treated as a pseudo-medium (Moridis and Reagan, 2011). Meanwhile, fractures were all assigned the same values except for the permeability. Based on the experiments performed by Liu et al. (2012) to analyze the effects of the width, generation method, and filled material of an artificial fracture on its permeability, the authors obtained the permeability of the fractures (≈3.0 × 10^–12 m²). While constructing NWAF using the HPRWJ technology, the tool is able to move up and down very slowly at a constant speed with a unaltered nozzle rotate speed at a fixed pump pressure, and thus, the fracture is assumed to be a thin rectangular-shape as show in Figure 3C.

Figure 4 shows the schematic of the meshes used to predict the gas production from the HBS under two different conditions (i.e., cases 1 and 2). The model domain was discretized using a cylindrical coordinate system, producing 99 (in r direction) × 102 (in z-direction) = 10098 grids, including 9,876 active elements and 222 boundary elements located on the top, bottom, and inner portion of the model. In the r direction, the first grid (Δr = 0.10 m) represents the wellbore; a fine discretization (Δr = 0.25 m) was applied from the second to sixth grid; the grid size increases exponentially from the 7th to 98th grid; and the 99th grid acts as the outermost boundary. Since hydrate exploitation is a complicated process with heat and mass transfer, the grid blocks near the wellbore are relatively dense in order to accurately describe the change in the physical properties of the hydrates (Moridis and Sloan, 2007c). In the z direction, two thin layers (Δz = 0.01 m) are used to define the uppermost and lowermost boundaries; a sparse discretization (Δz = 1 m) was applied to the overburden and the underburden to improve the computational efficiency owing to the lack of hydrates in these layers; a fine discretization (Δz = 0.5 m) was applied in the HBS in order to accurately describe the physical properties of the hydrates during exploitation. Since four equations were calculated for each grid block, 39,504 (= 9876 × 4) equations were calculated in the whole simulation system in total.

The main parameters of the model were set based on the core samples’ analyses and well logs in this region (Li et al., 2013; Liang et al., 2014). The sediment porosity and permeability were set to 0.4 and 7.5 × 10^–14 m², respectively. The overburden and the underburden were fully saturated with water and the hydrate saturation of the HBS was set to 0.44. The capillary pressure model (e.g., the van Genuchten function) and the three-phase relative permeability model (e.g., the modified version of Stone’s function) were employed. They are commonly used in numerical simulations and have been validated by field hydrate production tests (Anderson et al., 2011; Moridis and Reagan, 2011). The corresponding parameters of these multiphase flow models were determined using the accessible data in the literature. The lateral-to-vertical permeability ratio (k _R /k _Z ) of HBS and the overburden and underburden layers were set to 4.0 according to the latest pressure core flood test (Yoneda et al., 2019). The system parameters and physical properties of the simulation are given in Table 1. It should be noted that the initial conditions in the whole reservoir would remain stable unless there was a disturbance from outside, such as depressurization or thermal injection.

The production performance of the two cases, including gas production rate (Q _T ), cumulative gas production (V _T ), cumulative water production (V _W ), and gas-water ratio (R _GW ), in 1 year are shown in Figures 5, 6, respectively.

In case 2, the gas production rate and cumulative gas production (Q _T and V _T ) were consistently over 6000 ST m³/d and up to 230 × 10⁴ ST m³ respectively. In case 1, they were 2000 ST m³/d and 75 × 10⁴ ST m³, respectively. These findings indicate that the hydrate reservoir productivity increased by about three times when NWAFs were constructed by the HPRWJ technology. Therefore, this technology is a promising method for improving the gas production of low-permeability hydrate reservoirs.

There are three main reasons that the HPRWJ technology promotes hydrate productivity. Firstly, high-conductivity fractures with certain morphologies can be created near the production well, thus increasing reservoir permeability and dramatically decreasing flow resistance around the production well. As a result, the efficiency of gas and water flow towards the production well can be improved. Secondly, the effective surface area for the discharge of water and gas can be increased. The effective surface area only includes the surface area of the wellbore in case 1, while it includes the surface area of the wellbore and the artificial fractures and thus increased by several times in case 2; as a result, in case 2, there was more flux at the same production pressure. Thirdly, NWAFs can enlarge the transfer distance of pressure drops, as shown in Figure 7A, and can correspondingly accelerate hydrate dissociation, as shown in Figure 7B, thus increasing the gas production rate of hydrate reservoirs.

During the simulated 1-year production period, the cumulative water production (V _W ) and the gas–water ratio (R _GW ) in case 2 were 55 × 10⁴ m³ and 3.7 respectively, and those in case 1 were 17.5 × 10⁴ m³ and 3.7 respectively. These numbers indicate that, in case 2, not only gas production, but also water production increased, whereas the gas–water ratio was hardly affected. Therefore, it is necessary to optimize the perforated interval before NWAFs are constructed in order to achieve the optimal gas–water ratio.

Analyzing the changes in the main physical properties of hydrate reservoirs helps to understand the characteristics of hydrate dissociation and the processes of gas and water production during NGH exploitation. The spatial distributions of the pressure, temperature, hydrate saturation, and gas saturation in case 1 and 2 after 10, 60, and 365 days are shown in Figures 7A–D, respectively.

As shown in Figures 7A,C, in case 2, the transfer distance of the pressure drop during the whole exploitation was about 30 m, which was two-fold larger than that in case 1. The hydrate dissociation range was also distinctly larger in case 2, which is the major reason why the gas production rate in case 2 was three times that of case 1 as shown in Figure 5. However, in both cases, there was no significant changes in the transfer range of the pressure drop in the early (60th day) and later stages (365th day) of gas production. This consistency was mainly due to the fact that both the overburden and the underburden layers were composed of permeable sediments, and the water in these sediments can flow into the production well after a certain period of time. As a result, the pressure drop failed to be transferred into the deep reservoir, which restricted the hydrate dissociation to some extent. Therefore, to improve the recovery of the entire hydrate reservoir, well spacing should be properly controlled when vertical wells are employed to exploit hydrate deposits with permeable boundaries.

As an endothermic reaction, hydrate dissociation consumes the heat of the surrounding areas while producing gas and water, thus affecting the temperature distribution of the hydrate reservoir. Figure 7B clearly shows that there was an obvious low-temperature area near the production well at the start stage of NGH exploitation (10 days) and that the low-temperature area in case 2 was significantly greater than that in case 1. These findings are consistent with the hydrate dissociation ranges shown in Figure 7C. Figure 7B also demonstrates that, in both cases, the low-temperature area gradually shrunk at the early and later stages of NGH exploitation (the 60th and 365th day, respectively). This occurred mainly because the water from the overburden and the underburden flowed into the production well, changing the temperature distribution of the hydrate reservoir around the well. In particular, the hot water in the underburden significantly heated the hydrate reservoir and promoted hydrate dissociation. This phenomenon is consistent with the morphology of the hydrate dissociation front illustrated in Figure 7C.

The gas released from the hydrate dissociation was either extracted from the hydrate reservoir or remained in it as free hydrocarbon gas. Figure 7D shows that the gas-bearing area in case 2 was significantly larger than that in case 1, which indirectly proves that NWAFs can effectively promote the dissociation of hydrates into gas and water.

This article aims to apply the HPRWJ technology to vertical wells. The parameters used in this technology (e.g., jet pressure, rotary speed, processing time, and the moving distance of the tubing) determine the key fracture parameters, such as fracture depth, permeability, and spacing. Therefore, the effects of the key parameters of NWAFs on gas productivity were discussed, which is very important for parameter selection when using the HPRWJ technology for constructing NWAFs. The relationships between HPRWJ parameters and NWAF parameters will be further investigated through a laboratory and numerical study.

Fractures' depth is mainly controlled by jet pressure, rotary speed, and processing time. To compare the effects of different NWAF depths on the gas productivity of hydrate deposits, NWAF depths were set to 1, 2, 3, 4, and 5 m, as shown in Table 2. The authors assumed that fracture permeability, fracture spacing, fracture height, and the length of the completion interval were 3 D, 3, 0.5, and 12 m, respectively.

Figures 8, 9 present the gas production rate (Q _TD ), cumulative gas production (V _TD ), cumulative water production (V _WD ), and gas–water ratio (R _GWD ) at different NWAF depths. When the fracture depth increased from 1 to 2, 3, 4, and 5 m, the V _TD increased from 1.54×10⁶ ST m³ to 1.91×10⁶ ST m³, 2.30×10⁶ ST m³, 2.49×10⁶ ST m³, and 2.73×10⁶ ST m³ respectively, increasing by 24.0, 49.3, 61.7, and 77.3%. The results indicate that the gas production rate increases with an increase in the NWAF depth. However, V _TD increased by 8.3% when the fracture depth increased from 3 to 4 m, and by 20.4% when the fracture depth increased from 2 to 3 m. This indicates that the increase in fracture depth corresponds with the decrease in the magnitude of gas production efficiency. Therefore, a fracture depth of 3 m is recommended for the low-permeability hydrate reservoirs at site SH7. Moreover, it has been observed that cumulative water production increases as the NWAF depth increases, and that fracture depth has no effect on the gas–water ratio.

According to laboratory test results, fracture permeability is closely related to the materials added to jet water and can reach 7.45×10² D during NWAF construction (Xu et al., 2016). To compare the effects of different NWAF permeability values on the productivity of hydrate deposits, NWAF permeability was set to 1 D, 2 D, 3 D, 4 D, and 5 D, as shown in Table 3. The authors assumed that fracture depth, fracture spacing, fracture height, and the length of the completion interval were 3, 3, 0.5, and 12 m, respectively.

Figures 10, 11 present the gas production rate (Q _TP ), cumulative gas production (V _TP ), cumulative water production (V _WP ), and gas–water ratio (R _GWP ) in cases of different NWAF permeability values. When fracture permeability increased from 1 D to 2 D, 3 D, 4 D, and 5 D, V _TP increased from 1.57×10⁶ ST m³ to 2.03×10⁶ ST m³, 2.30×10⁶ ST m³, 2.49×10⁶ ST m³, and 2.62×10⁶ ST m³, respectively, increasing by 29.3, 46.5, 58.6, and 66.9%, respectively. The results indicate that the gas production rate increases with an increase in the NWAF permeability. However, V _TP increased by 8.3% when fracture permeability increased from 3 D to 4 D, and by 13.3% when the fracture permeability increased from 2 D to 3 D. This illustrates that the increase in fracture permeability corresponds with the decrease in the magnitude of gas production efficiency. Therefore, fracture permeability of 3 D is recommended for the low-permeability hydrate reservoirs at site SH7. Moreover, cumulative water production increases as the NWAF permeability increases, and fracture permeability has no effect on the gas–water ratio.

Fracture spacing can be controlled by lifting or lowering the tubing. To compare the effects of different NWAF spacing values on the productivity of hydrate deposits, the NWAF spacing was set to 1, 2, 3, 4, and 5 m, as shown in Table 4. The authors assumed that fracture depth, fracture permeability, fracture height, and the length of the completion interval were 3 m, 3 D, 0.5, and 12 m respectively.

Figures 12, 13 show the gas production rate (Q _TS ), cumulative gas production (V _TS ), cumulative water production (V _WS ), and gas–water ratio (R _GWS ) at different NWAF spacing values. Figure 12 clearly demonstrates that NWAF spacing has significant effects on gas production rate (Q _TS ). The stable gas production rate was 8,100 ST m³/day when the fracture spacing was 1 m, whereas it was only 4,300 ST m³/day when the NWAF spacing increased to 5 m. These findings show that the gas production rate increases with a decrease in the NWAF spacing. However, V _TS increased by 53.3% when fracture spacing decreased from 4 to 3 m, and by 16.1% when fracture spacing decreased from 3 to 2 m. This means that the decrease in fracture spacing corresponds with the decrease in the magnitude of gas production efficiency. Therefore, a fracture spacing of 3 m is recommended for the low-permeability hydrate reservoirs at site SH7. NWAF spacing has no effect on the gas–water ratio.

This study’s investigation of the effects of the key parameters of NWAFs, such as depth, permeability, and spacing on gas productivity during 1 year of simulated mining led to important findings. As the fracture depth increases from 1 to 5 m, cumulative gas production can increase from 154 × 10⁴ to 273 × 10⁴ ST m³. As the fracture permeability increases from 1 D to 5 D, the cumulative gas production can increase from 157 × 10⁴ to 262 × 10⁴ ST m³. As fracture spacing increases from 1 to 5 m, cumulative gas production can decrease from 318 × 10⁴ to 126 × 10⁴ ST m³. Gas production efficiency can be significantly increased by increasing fracture depth and permeability or by reducing fracture spacing. As mentioned previously, however, marine NGH deposits often involve unconsolidated sediments that exhibit limited shear strength, especially when the hydrates are dissociated. Under these conditions, the fractures can reduce the structural stability of NGH sediments. Therefore, when using NWAFs to increase gas production, the shear strength and structural stability of NGH sediments need to be considered in the determination of fracture depth and spacing, which will be researched by the authors in the future. After fracture depth and spacing are determined, fracture permeability, which hardly affects the structural stability of HBS, should be increased as much as possible. By comparing growth gaps between gradients under variable control, the recommended NWAF depth, permeability, and spacing are 3 m, 3 D, and 3 m, respectively for the clayey silt reservoirs at site SH7, and more detailed discussions will be investigated in a later study.