Gas Permeability and Production Potential of Marine Hydrate Deposits in South China Sea

: The permeability of marine sediments is critical to the gas production assessment of hydrate reservoirs. In this work, the sample of natural marine sediments was obtained from Shenhu Area of South China Sea at the depth of 1600 m, and the gas permeability of the sample was measured in the laboratory under various conﬁning pressures. The porosity of the sample decreased from 41.82% to 29.54%, and the e ﬀ ective gas permeability of the sample decreased from 2.638 × 10 − 16 m 2 to 0.872 × 10 − 16 m 2 as the conﬁning pressure increased from 0 to 23 MPa. The gas permeability of the natural sediments was determined to be 1.535 × 10 − 16 m 2 with conﬁning pressure of 15 MPa and porosity of 32%. The deformation of sample in longitudinal direction was sensitive to the conﬁning pressure, and the compressibility in the radial direction was limited. On the basis of the experimental results, the particle size term in the classical Kozeny–Carman equation was revised by correction factor N, and the experimental results ﬁtted well with the curves with N = 2.40. Moreover, the gas production potential at the site of W–17 in the Shenhu Area was numerically studied based on the measured gas permeability. The simulation results showed that the cumulative volume of produced gas was approximately 1.46 × 10 6 ST m 3 after 30 years. A new enlarged permeable well wall method was proposed to improve the poor gas production caused by low permeability. The cumulative gas production increased by 2.7 times as the permeable well wall radius increased from 0 to 5 m. Regarding the gas and water production rates, water production increased with the increase of permeable well wall radius. Spatial distributions of hydrate and gas saturations, pressure, and temperature were investigated as well.


Introduction
Natural gas hydrates (NGHs) are solid crystalline compounds composed of water and gas molecules, and the gas molecules are encaged inside the lattices of water cavities [1]. NGHs in nature are mainly concentrated in the deep ocean sediments and permafrost, where the necessary high pressures and low temperatures exist [2]. NGH have attracted global attentions because of its huge reserves and wide distributions, and NGH is considered as one of the most promising alternative energy sources in the 21st century [3,4]. In addition, the gas production from NGH deposits is considered to have a far higher impact on global economy than the impact of other unconventional natural (<1 mD), such as the natural sediments used in this study, gas may be the only media to go through the compacted sample because of its low diffusivity [31]. For the gas permeability measurements, the steady-state method was employed. The steady-state method was a steady injection state follows Darcy's law, and the pressure variation in the sample was given by the simplified gas flow formula as shown in Equation (1) [32,33]. Figure 1 shows the sketch map of northern South China Sea and the drilling expedition location of W-17 [34]. The drilling results showed that the NGH existed in silt clay formation, and it was difficult to exploit this kind of NGH because of its low-permeability characteristics [29]. The production test site was located in the middle of the southeastern continental slope of Shenhu area. The bottom of the hydrate bearing layer (HBL) was at the depth of 1572 m, which was basically close to the depth of our obtained samples (1600 m). Since the position of the obtained sample sediments was close to the site of W-17, it was believed that the geological environment of the sample sediments was more likely similar to the sediments at the site of W-17. Hence, the gas production characteristics were numerically simulated based on the porosity, hydrate saturation, temperature, and pressure conditions of W-17 sediments. The measured gas permeability (1.535 × 10 −16 m 2 ) of natural sediments with the confining pressure of 15 MPa was used as the intrinsic permeability of the hydrate bearing layer. Considering that the improving of percolation capacity was the key to the production of low permeability reservoirs, an enlarged permeable well wall was used to promote the pressure propagation speed and distance. The enlarged permeable well wall filled with hydrophobic sand was surrounding the vertical production well in the production interval [35]. Hence, the production potential of hydrate deposits at site of W-17 was evaluated based on the measured gas permeability of natural sediments. In this work, we aimed at evaluating the production potential of Shenhu Area of South China Sea based on the measured gas permeability of the natural marine sediments. First, the gas permeability of natural sediments was measured by the steady-state method under different confining pressures. After that, according to the porosity, hydrate saturation, temperature, and pressure conditions of W-17, the gas production performance at the site of W-17 was numerically studied based on the experimental results. Considering the low gas production of simulation results, we tried to adopt an enlarged permeable well wall to improve the pressure propagation speed and distance. The gas and water production and gas-to-water ratio were analyzed with different radii of enlarged permeable well wall. Spatial distributions of the characteristic parameters hydrate and gas saturations, pressure, and temperature were investigated as well.

Apparatus
As shown in Figure 2, the apparatus used for the gas permeability tests consists of a confining cell, a constant-pressure pump, a buffer vessel, and a newly developed interface to ensure airtightness, etc. The confining pressure (P C ) is supplied by the constant-pressure pump, and its value is recorded on a computer through a digital pressure gauge. The buffer vessel with the inner volume of 400 mL made of 316 L stainless steel is used to provide a steady gas flow. The volume of buffer vessel is expressed by V 0 . The gas used for the gas permeability tests is pure argon (more than 99%). Argon as an inert gas is considered to be an ideal gas to measure the gas permeability of the sediment, because its greater molecule mass could effectively decrease the magnitude of potential leaks. In the natural marine sediments used in this work, the adsorption of methane gas cannot be ignored, because the methane gas adsorbed in the sediments will reduce the flow rate. Therefore, argon gas is used instead of methane gas to determine the gas permeability of the sediments. Two high-precision manometers are installed at both sides of the buffer vessel. The inlet pressure labeled as P 1 is recorded once per second during the gas permeability measurement. The outlet pressure labeled as P 0 is set to be the atmospheric pressure. In addition, all tests are carried out in a constant temperature room at 293.15 K.

Materials and Sample Preparation
In this study, the natural sediments used as the porous media were dried for 120 h at the temperature 393.15 K and then pulverized. The purpose of removing the moisture content from sediments is to prevent smectite in sediments from swelling with water. The solid density and volume-weighted mean diameter of the sample were 2.42 g/cm 3 and 6.491 µm, respectively. As shown in Figure 3, the particle size distributions and volume-weighted mean diameters measurement results of the natural sediments, which were measured using the Mastersizer 2000E produced by Malvern Instruments Ltd., United Kingdom. The particle size distributions of the sample were measured three times, and the three measurements were in good agreement. The sediments with volume-weighted mean diameter less than 1.0 µm were approximately 10%, and only very few sediments with the volume-weighted average diameter greater than 15.0 µm. Sample I and sample II with a mass of 46.95 g and 47.05 g were placed into the cylinder tank, respectively, and then installed the piston to preliminarily compact the sample. The axial pressure of the piston cylinder was 15 MPa and the compression time was 2 h. The resulting heights of sample I and sample II were 17.25 mm and 17.22 mm, respectively. The resulting diameters of sample I and sample II were 49.61 mm and 49.80 mm, respectively. Hence, the initial porosities of sample I and sample II were 41.82% and 42.04%, respectively.

Method to Measure Gas Permeability
Darcy's law was employed to measure the gas permeability due to the continuous gas flow through the sample under steady conditions, and the effective gas permeability K eff could be applied as follows [23,31,36]: where µ g was the viscosity of argon (2.2× 10 −5 Pa·s), Q g was the average gas volume rate, h was the length of sample, A was the sample cross-section, and P m was the average of upstream gas pressure at the inlet. As shown in Figure 2, the inlet of the cylinder sample was subjected to a given pressure P 1 . The outlet was subjected to constant atmospheric pressure of P 0 . There was a pressure decrease (∆P) after a period of time (∆t) under the driving force of the pressure difference. By assuming a quasi-static flow and the ideal gas state equation [33], the average gas volume rate (Q g ) and the average of upstream gas pressure at the inlet (P m ) were calculated with the following equations, respectively: Combining Equations (1)-(3), the effective gas permeability K eff could be expressed as follows:

The Relative Displacement
The sample was placed in triaxial cell with confining pressure of P C . The confining pressure was gradually increased from 0 to 23 MPa, and the pressure points of experiments were 2 MPa, 5 MPa, 8 MPa, 10 MPa, 15 MPa, 20 MPa, and 23 MPa, respectively. Figure 4 shows the radial and longitudinal deformation of sample I and sample II as confining pressure increased from 0 to 23 MPa. At the end of the experiments, the radial deformations of sample I and sample II were 1.724 mm and 1.465 mm, and the longitudinal deformations of sample I and sample II were 1.960 mm and 2.381 mm. In other words, sample I was compressed by 3.48% in the radial direction and 11.37% in the longitudinal direction. Sample II was compressed by 2.94% in the radial direction and 13.83% in the longitudinal direction. Hence, the longitudinal deformation was very sensitive to the confining pressure, and the compressibility of sample in the radial direction was not obvious. The radial deformation curves of sample I and sample II were close to a straight line as confining pressure increased from 0 MPa to 23 MPa. The longitudinal deformation was evident as the confining pressure increased from 0 to 2 MPa, and then decreased with the increase of confining pressure.

The Variation of Inlet Pressure during the Experiment
The measurement methods were similar under different confining pressure conditions, so only the determination of the measurement point at a confining pressure of 15 MPa was given. Figure 5 shows the inlet pressure (P 1 ) changes of sample I and sample II during gas permeability measurement with the confining pressure of 15 MPa. The whole measurement process could be divided into two stages. In stage I, also named preparation stage, the gas flow from the buffer vessel into the triaxial cell was in a stable state. In this preparation stage, it took a long time for the pressure to stabilize and remaining constant, and we only drew 35 s after stabilization. In stage II, the control valve was opened (point A and D), and there was a pressure decrease (∆P) of inlet pressure (P 1 ) after a period of time (∆t) under the driving force of the pressure difference. Namely, the outlet pressure (P 0 ) of the triaxial cell was set to be atmospheric pressure. In other words, with the increase of time from 35 s to 120 s, the inlet pressure (P 1 ) of sample I decreased from 2.04 MPa (point D) to 1.75 MPa (point F), and the inlet pressure (P 1 ) of sample II decreased from 2.17 MPa (point A) to 1.86 MPa (point C). That was to say, the inlet pressure of sample I and sample II decreased by 0.29 MPa and 0.31 MPa in 85 s, respectively. The ∆P/∆t ratio of two samples was very close. Hence, the gas permeability results measured at the point B and E were determined to be the value of the effective gas permeability with the confining pressure of 15 MPa. The same method was applied in other confining pressure measuring points.  Figure 6 shows the effective gas permeability and corresponding porosity change results. The two sets of measurement results were consistent. The experimental results showed that the gas permeability and the porosity were very sensitive to the confining pressure. The experimental results could be divided into three stages. In stage I, the gas permeability of the samples decreased rapidly as the confining pressure increased from 2 MPa to 5 MPa, which was due to the compression of the sample. In stage II, the gas permeability of both samples gradually slowed down as the confining pressure increased from 5 MPa to 15 MPa; namely, the gas flow through the sample was more uniform and stable than that in the stage I. In stage III, with the confining pressure increased from 15 MPa to 23 MPa, the rate of gas permeability reduction was obviously higher than that of the stage II, because the samples structure changed, and the sample was more compact. With the confining pressure increased from 0 MPa to 23 MPa, the porosity of sample I decreased from 41.82% to 29.54% and the porosity of sample II decreased from 42.04% to 28.60%, respectively. Also, the effective gas permeability of the sample I decreased from 2.638 × 10 −16 m 2 to 0.872 × 10 −16 m 2 and the effective gas permeability of the sample II decreased from 2.642 × 10 −16 m 2 to 0.841 × 10 −16 m 2 as the confining pressure increased from 2 MPa to 23 MPa. Taking the measurement point of 15 MPa confining pressure as an example, the effective gas permeability of sample I and sample II were 1.535 × 10 −16 m 2 and 1.505 × 10 −16 m 2 and the porosity of sample I and sample II were 32.00% and 31.71%, respectively.

The Revision of Kozeny-Carman Equation
The classical Kozeny-Carman (KC) equation of the permeability-porosity relationship was applied to groundwater flow, chemical engineering, and other fields. The KC equation could be simplified as the following formula [37][38][39]: where k was the KC constant, and we employed k = 5 as the empirical KC constant in this work; φ was the porosity of the sample and D was the mean diameter of the sediments (D = 6.491 µm).
The surface of particles in sediment was not smooth, and the particles were not spherical. Hence, the particle size term in the KC formula was revised by a correction factor (N) in this paper. The revised formula was expressed as follows: Figure 7 shows the relationship between the effective gas permeability and the porosity of the samples. The mean diameter (D = 6.491 µm) was used as the volume-weighted mean diameter of the sediments. Three kinds of N values were used to verify the experimental results. It fitted well with the experimental results when N value was 2.40. Furthermore, all the experimental results suited at the curves of N = 2.35 and N = 2.45. Therefore, it could be considered that the particle size correction factor of marine sediments at the depth of 1600 m below the sea level in the Shenhu Area of South China Sea was about 2.40.  Figure 8 shows the geometry schematic diagram of simulation domains. The enlarged permeable well wall was surrounding the vertical production well in the production interval. There were three hydrate-bearing layers (HBLs) between the overburden (OB) and underburden (UB). The thickness of HBL-I, HBL-II, and HBL-III were 35 m, 15 m, and 27 m, respectively. The bottom of HBL-III was at the depth of 1572 m below sea level. The OB and UB had the same thickness of ∆Z = 20 m, which was large enough to provide precise estimates of pressure and heat transfer in the hydrate deposit [7,20]. The radius and height of the cylindrical simulation domains were 145 m and 117 m, respectively. Considering the symmetry, only half of simulation domains were used for simulation calculation. The vertical production well was positioned at the center of the cylindrical simulation domains and parallel to the axial direction. The radius of production well (r w ) was 0.1 m, the production well was assumed to have the permeability of k = 5.0 × 10 −9 m 2 (5000 Darcies), the porosity was φ = 1.0, and the capillary pressure was P cap = 0 [10,40]. The core analysis and well logging data at the site of W-17 indicated HBL-II and HBL-III were the most active section, where the pressure and temperature were closest to NGH phase equilibrium curve [29]. Hence, the HBL-II and HBL-III were employed as production intervals, because both of them were ideal gas hydrate production layers. As shown in Figure 8, the enlarged permeable well wall filled with hydrophobic sand was surrounding the vertical production well in the production interval. Furthermore, the enlarged permeable well wall was realized by the directional hydraulic fracturing technology, and the porous hydrophobic materials was injected into the artificial crack to form a relatively high-permeability zone with specific radius. The intrinsic permeability inside the enlarged permeable well wall was assumed to have k = 5×10 −13 m 2 (0.5 D).

Domain Discretization, System Properties, and Initial Conditions
The simulation results were calculated by the parallel version of the TOUGH + HYDRATE (T + H) code. The geologic system was arranged on the basis of hydrate deposits at the site of W-17 in South China Sea [29]. Table 1 shows the system properties and initial conditions used in this simulation model. The HBL-II and HBL-III were used for gas production and the HBL-I was used as natural barriers. As shown in Figure 9, the radial direction R was divided into 81 grids, the discretization along the radial direction was non-uniform, and the first grid was 0.1 m followed by logarithmical increase. The axial direction Z was divided into 194 grids; a very fine discretization (∆Z = 0.5 m) was used in HBLs and a coarser discretization (∆Z = 1.0 m) was used in OB and UB. This fine discretization was sufficient to accurately describe the heat and pressure transfer and spatial distributions of characteristic parameters in the deposit. Hence, the cylindrical domain system was discretized into 81 × 194 = 15,714 grid blocks in (R, Z) direction. The number of active grid blocks was 14,140, and the remaining grid blocks were the inactive steel and boundary elements, which included the uppermost and lowermost layers with the constant pressure and temperature conditions.
The initial conditions used in the model were calculated by the methods described by Li et al. [11,20]. The base of the HBL-III was at the depth of 1572 m under the sea level. The initial pressure and temperature at the bottom of the HBL-III were initialized to be 15.8 MPa and 289.38K, respectively (as shown in Table 1). The mean effective porosities of HBL-I, HBL-II, and HBL-III were 35%, 33%, and 32%, respectively. The pore space within HBL-I was filled with liquid water and solid NGH (mean hydrate saturation S H = 34%). The pore space within HBL-II was filled with liquid water, solid NGH (mean S H = 31%), and free hydrocarbon (mean gas saturation S G = 7.8%). The pore space within HBL-III was filled with liquid water and free hydrocarbon (mean S G = 7.8%). The basic parameters were based on field measurement [29].  Figure 10 shows the cumulative volume (V p ) and volumetric rate (Q p ) of methane production from the well. The depressurization method was employed in all four cases, and the production pressure P W was set to be 3.0 MPa in each case. In these four cases, the V P curves were all characterized by a continuous rise during the 30-year production period, and the Q p curves showed a rapid decline from each highest value in a short time. In the case without the enlarged permeable well wall, the total V p was about 1.46 × 10 6 m 3 under the standard state (ST) during the 30-year production period, the maximum Q p was approximately 600 m 3 per day in the first few days, and then gradually decreased to 89 m 3 per day after 30 years. In the case with the enlarged permeable well wall radius of 5 m, the total V p was about 3.89× 10 6 ST m 3 , the maximum Q p was more than 1800 m 3 per day in the first few days, and then gradually decreased to 215 m 3 per day at the end of 30-year production period. The results indicated that the V p increased by 2.7 times, and the Q p increased by 2.4 times as the permeable well wall radius increase from 0 to 5 m. In other words, the larger the permeable well wall radius, the larger amount of gas and the volumetric rate we would obtain, which indicated that the influence of enlarged permeable well wall on hydrate dissociation was obvious. Meanwhile, it showed a stable production rate after 4500 days, which indicated that the hydrate dissociation process tended to be steady. Figure 11 shows the cumulative mass of produced water (M W ) and the mass rate of water production (Q W ) from the production well. In these four cases, the M W curves were all characterized by a continuous rise during the 30-year production period, and Q W curves showed a rapid decline from each highest value in a short time; this phenomenon was consistent with the analysis of gas production. In the case without the enlarged permeable well wall, the M W curve was close to a straight line, which indicated that the water production rate was basically constant. However, in the case with the enlarged permeable well wall radius of 5 m, the M W curve had a tendency to exaggerate, which indicated that the water production would continue to increase. At the end of 30-year production period, the total M W increased from 4.6 × 10 7 kg to 2.0 × 10 8 kg as the permeable well wall radius increased from 0 to 5 m, which indicated that the total M W increased by 4.3 times. Meanwhile, the Q W was 110 kg per day in the case without enlarged permeable well wall, and increased by 2.2 times in the case with the enlarged permeable well wall radius 5 m. Comparing the increase of Q W and M W , the M W was mainly produced in the early stage of production, which indicated that the effect of enlarged permeable well wall on the early stage of production was better than that in the later stage of production. Figure 10. The cumulative volume (V p ) and volumetric rate (Q p ) of gas production. Figure 11. The cumulative mass of produced water (M W ) and the water production rate (Q W ). Figure 12 shows the evolution of the gas-to-water ratio (R GW ) during gas production from the production well with different circumferential radii. The R GW was calculated to be R GW = 1000 V P /M W . In these four cases, the R GW curves rose sharply in the first few days and then decreased to a lower level gradually. As shown in Figure 12, the maximum R GW of four cases was 106 (well wall radius 3 m). In the first 2000 days, the R GW of the case with the enlarged permeable well wall radius of 3 m was the highest among the four cases, and then the R GW of the case with well wall radius of 1 m became the largest. The simulation results indicated that the ideal maximum value of R GW was between the cases with the enlarged permeable well wall radius from 1 to 3 m. Meanwhile, R GW maintained stable after 4500 days, which was consistent with the Q p mentioned above. The R GW in the case with the enlarged permeable well wall radius of 5 m was 19.2 at the end of 30-year period, which was only a half of that in the case with the enlarged permeable well wall radius of 1 m. This proved that the production potential with the enlarged permeable well wall radius of 1 m was more promising in the relative criterion. Figure 12. Gas-to-water ratio (R GW ) produced from the production well. Figure 13 shows the evolution of the spatial distributions of S H during the 30-year production period in the case with the enlarged permeable well wall radius of 5 m. The selected time points were t = 60 days, 1 year, 5 years, 10 years, 20 years, and 30 years, corresponding to Figure 13a-f, respectively. The S H spatial distributions over time provided a measure to evaluate the hydrate decomposition compared with the initial HBL. In initial state, the pore space within HBL-I (-20 m < Z < -55 m) was filled with liquid water and solid NGH (mean S H = 31%). The pore space within HBL-II (-55 m < Z < -70 m) was filled with liquid water, solid NGH (mean S H = 31%), and free hydrocarbon (mean S G = 7.8%). In the first few 60 days, a small cylindrical region with the hydrate dissociated around the enlarged permeable well wall was formed, and the hydrate was basically not dissociated within HBL-I. However, at the end of 30-year production period, the radius of hydrate decomposition region within HBL-I and HBL-II were about 20 and 35 m, respectively. Meanwhile, secondary hydrate formation occurred in the interface of the HBL-I and HBL-II, which was taken as natural barriers to prevent the low-temperature water invasion from the OB. The hydrate decomposition rate at the bottom of the HBL-II was larger than that of the upper part, which mainly due to the influence of the geothermal gradient.  Figure 14 shows the evolution of the spatial distributions of S G during the 30-year production period in the case with the enlarged permeable well wall radius of 5 m. In initial state, the pore space within HBL-III (-70 m < Z < -97 m) was filled with liquid water and free hydrocarbon mean (S G = 7.8%). The mean S G within the HBL-II was employed as 7.8% in the initial state as well. The originating undissociated hydrate between the interface HBL-I and HBL-II functioned as a barrier to prevent the gas and fluid flow because of its low permeability. In the early stage of about 10 years, the gas saturation in the production well could maintain a relatively high level, and then gradually decreased to a level close to the hydrate layer saturation. The free gas saturation within the HBL-I was very limited after 1 year because of the very little hydrate decomposition, and then the hydrate near the production well began to decompose gradually. At the end of the 30-year production period, the radius of hydrate dissociated cylindrical region at the interface of HBL-I and HBL-II was about 30 m, but the height within the HBL-I was only close to 15 m, which indicated the hydrate dissociation of HBL-I was still at a low level. The gas saturation had a significant increase between the HBL-II and HBL-III, which indicated the free gas of HBL-III gradually diffused to the HBL-II with the increase of production period.

Spatial Distributions of P
Figure 15a-f shows the evolution of the P spatial distributions over time in the case with the enlarged permeable well wall radius 5 m. The pressure gradient propagated from the HBL to the OB and UB in the middle to late production periods. In 60 days, the hydrate dissociation caused by depressurization only affected a small area near the production interval. The pressure of OB was basically constant, which indicated that OB as a barrier layer played a very good protective role to prevent the invasion of the upper seawater. At the end of the 30-year production period, the pressure distribution was basically stable, which indicated that the gas and water two-phase flow process in the whole hydrate reservoir was relatively stable. Meanwhile, the pressure propagation radius within the HBL-I reached 60 m, which indicated that the hydrate in the HBL-I was also in a decomposed state. Comparing the spatial distributions of S H and P, the pressure of hydrate was lower than the condition of phase equilibrium for a period of time before decomposition. That was to say, the driving force of depressurization broke down the original condition of phase equilibrium and stability, which resulted in the decomposition of hydrate reservoirs.  Figure 16 shows the evolution of the T spatial distributions over time in the case with the enlarged permeable well wall radius of 5 m. The evolutions of temperature T provided the effects of the environmental impacts. As shown in Figure 16a, the temperatures around the production well within the HBL-II decreased sharply, which was due to the sensible heat adsorption by the hydrate decomposition. In the first five years of the production period, there was an obvious temperature drop area around the production well within the HBL-II, and then it gradually rose to a level close to the original formation. The fluctuation of temperature was mainly concentrated within the HBL-II, which indicated that the HBL-II was the main hydrate decomposition layer. The rapid decrease of temperature within the HBL-II absorbed a lot of heat, which was the main reason for the formation of secondary hydrate. The flow of free gas from the HBL-III into the production well brought a lot of heat, which played an important role in preventing the formation of secondary hydrate in the production interval. Therefore, simultaneous exploitation of the HBL-II and HBL-III was an effective way to solve the secondary hydrate problem.

Conclusions
In this work, we performed a comprehensive work on the evaluation of gas production of hydrate deposits in the South China Sea. This work includes the measurement of gas permeability of the sediments in the South China Sea, the numerical simulation of gas and water production characteristics at the site of W-17, and the a new method of increasing the permeability around the production well to improve the gas production efficiency.
The intrinsic permeability of natural sediments was determined to be 1.53× 10 −16 m 2 with the confining pressures of P C = 15 MPa and porosity of 32.00%, and the gas permeability was very sensitive to the confining pressure.
The effective gas permeability of the natural sediments sample decreased from 2.638 × 10 −16 m 2 to 0.872 × 10 −16 m 2 as the confining pressure increased from 2 MPa to 23 MPa. The porosity of the natural sediments sample decreased from 41.82% to 29.54% as the confining pressure increased from 0 to 23 MPa.
The particle size term in the classical Kozeny-Carman equation was revised by a correction factor (N), and the experimental results fitted well with the curves when N value was 2.40.
In the case without the enlarged permeable well wall, the total V p and M W were about 1.46×10 6 m 3 and 4.6×10 7 kg under standard state (ST) after 30 years, respectively. The V p and M W increased by 2.7 and 4.3 times as the permeable well wall radius increased from 0 to 5 m, respectively.
The R GW in the case with the enlarged permeable well wall radius 5 m was 19.2 at the end of 30-year period, which was only half of that in the case with the enlarged permeable well wall radius 1 m. This proved that the production potential with the enlarged permeable well wall radius 1 m was more promising in the relative criterion.
The influence of the enlarged permeable well wall on hydrate dissociation was obvious. The larger the permeable well wall radius, the larger amount of gas and the volumetric rate we would obtain. The free gas of HBL-III gradually diffused to the HBL-II with the increase of production period. Simultaneous exploitation of the HBL-II and HBL-III was an effective way to solve the secondary hydrate problem.
Author Contributions: All the authors contributed to publish this paper. The main research idea was contributed by X.L., G.L. and P.S.; P.S. and X.L. wrote the paper; J.Z. contributed on the simulation works; J.L. prepared for the experiments. All authors revised and approved the publication of the paper.

Conflicts of Interest:
The authors declare no conflict of interest.

Nomenclature
G thermal gradient within the frozen layer ( • C/m) k intrinsic permeability (m 2 ) k eff effective permeability (m 2 ) k rA aqueous relative permeability (m 2 ) k rG gas relative permeability (m 2 ) k ΘC thermal conductivity (W/m/K) k ΘRD thermal conductivity of dry porous medium (W/m/K) k ΘRW thermal conductivity of fully saturated porous medium (W/m/K) P pressure (Pa) P B initial pressure at base of HBL (Pa) P 0 atmosphere pressure (Pa) PW pressure at the well (Pa) ∆P W driving force of depressurization, P W0 −P W (Pa) V P cumulative volume of produced CH 4 (ST m 3 ) M W cumulative mass of produced water (kg) Q P volumetric rate of methane production at the well (ST m 3 /d) Q W water mass production rate at the well (kg/d) R GW gas-to-water production ratio (ST m 3 of CH 4 /m 3 of H 2 O) R, Z cylindrical coordinates (m) ∆R discretization along the R−axis (m) ∆Z discretization along the Z−axis (m) r max maximum radius of the simulation domain (m) r w production well radius (m) r e radius of the permeable well wall (m) S phase saturation t time (days) T temperature ( • C) T B initial temperature at the base of HBL ( • C) X S salinity ∆H c combustion enthalpy of produced methane (J) Φ porosity λ van Genuchten exponent-