Analysis of Production Laws of Hydrate Reservoirs via Combined Heat Injection and Depressurization Based on Local Thermal Non-Equilibrium

Abstract

Natural gas hydrate is a kind of low-carbon and clean new energy, so research on its efficient extraction in terms of theory and technology is particularly important. Combined thermal injection and depressurization is an effective method for extracting natural gas hydrate. In this study, the classical local heat equilibrium model was modified, and a pore-scale fully coupled unsteady heat transfer model for hydrate reservoirs was set up by considering multiple forms of heat flow accompanying hydrate’s decomposition and gas–liquid flow. Based on this model and the basic geological information of the X2 hydrate reservoir in the western Pacific Ocean, a numerical model of gas hydrate extraction using combined heat injection and depressurization was constructed to simulate the production performance of the hydrate reservoir. The results were fully compared with the results obtained by the depressurization method alone. The results indicated the hydrate extraction via a combined heat injection and depressurization would have a cumulative gas production of 31.609 million m³ and a cumulative water production of 1.5219 million m³, which are 72.57% higher and 31.75% lower than those obtained by depressurization alone, respectively. These study results can provide theoretical support for the industrial extraction of gas hydrate in seas.

1. Introduction

The demand for oil and natural gas is growing in all countries around the world, driven by population growth and rapid economic development [1,2]. Natural gas hydrate has been attracting attention around the world as a clean new energy source with a wide distribution, low carbon emissions and a high energy density [3,4]. Research indicates that global reserves of natural gas hydrates are in the order of 3000 to 5000 trillion cubic meters, and that the organic carbon content of natural gas hydrates is twice that of conventional fossil fuels [5,6].

Since over 90% of natural gas hydrates occur in the sea [7], there is an urgent need to strengthen the research on the theory and technology behind the efficient extraction of natural gas hydrates in the sea in order to provide backup techniques for its commercial extraction. Natural gas hydrate cannot be extracted by the traditional methods used for fluid mineral deposits such as oil and natural gas because of its reservoir-forming properties and gas-producing characteristics. At present, the main extraction methods used in hydrate reservoirs include depressurization [8], thermal stimulation [9], CO₂ replacement [10] and inhibitor injection [11], as well as the combinations of them [12]. The principles of gas production from hydrate reservoirs can be summarized as follows: (1) changing the in situ temperature and pressure conditions of the reservoir to force the hydrate to decompose or (2) changing the phase equilibrium conditions of the hydrate to force its decomposition under in situ temperature and pressure conditions. So far, in Mallik [13], Canada; Alaska [14], USA; Nankai Trough, Japan [15,16]; the South China Sea; and other regions [17,18,19,20,21,22], the thermal stimulation method [13], depressurization method [15,16,17,18,23], CO₂ replacement method [14,24] and solid fluidization method [19] have been used in eight hydrate production tests. These field tests show that the depressurization method is the most effective method at breaking the hydrate’s phase equilibrium. But as hydrate decomposition needs a large amount of heat and hydrate reservoirs have poor thermal conductivity in general, relying solely on the energy of the formation itself often leads to a sharp decline in the gas production rate [25,26,27,28]. Meanwhile, increasing the pressure dropping magnitude to increase production often leads to a number of problems such as sand production and the secondary generation of hydrate [29,30].

Combined thermal injection and depressurization is an effective method for hydrate development. After studying the influence of different production well systems on hydrate production performance according to the geological conditions of hydrate reservoirs in the permafrost of the Qilian Mountains, Liang et al. [31] and Li et al. [32] concluded that the well spacing and heat injection rate were the key factors affecting the gas production efficiency of hydrate. The research of Feng et al. [33] showed that injecting 90 °C hot water resulted in a peak gas production rate about 2.02 times that of injecting 28 °C hot water, but that the hydrate’s energy utilization efficiency was significantly lower when injecting 28 °C hot water. Moridis et al. [34] found that the injected energy was mainly used to heat the formation, and only part of it was directed to enhancing the hydrate’s decomposition kinetics; when almost all of the hydrate along the flow channel between a water injection well and a production well decomposed, the increase in the water injection temperature and salt concentration resulted in a much smaller rise in the gas production rate. Yu et al. [26] proved that the injected water flow not only boosted the kinetics of hydrate decomposition in the porous medium but also drove the free gas to the production well by constructing a three-dimensional model of hydrate production in a horizontal well via heat injection. But they found that a part of the free gas was displaced to the outer boundary of the formation, giving rise to secondary hydrate there. Formation permeability plays a dominant role in hydrate development. High formation permeability is more conducive to multiphase flow, convective heat transfer and pressure propagation in the reservoir. The anisotropy of permeability is conducive to reducing water production but could also have a negative impact on gas production. Through simulating the development systems of dual production wells and “one injection well and one production well”, Ma et al. [12] reached the finding that, compared with depressurization alone, when the water injection temperature was increased from 20 °C to 80 °C, the gas productivity increased by 4.20% to 30.40%, but the energy utilization efficiency decreased from 10.22 to 1.53; a heat injection in the late stage of production could not improve the gas production rate noticeably, and stopping the water injection would instead increase the relative permeability of the gas in the formation. Through numerical simulation, Zhao et al. [35] concluded that a hot water injection could increase cumulative gas production by 1.47 times compared to depressurization, but only 20.4% of the injected energy was directed to hydrate decomposition, and 52.7% of the injected energy was directed to raising the formation’s temperature. After examining the long-term gas production characteristics of the test production of hydrate, Yu et al. [36] suggested that although the 10-year average gas production rate could be increased from 7.71 × 10² m³·d⁻¹ to 2.88 × 10³ m³·d⁻¹ by heat injection, it still could not meet the production capacity requirements for hydrate’s commercial development. After analyzing the effects of hydrate reservoir properties (permeability, porosity and rock density), water injection parameters (water injection rate and salt concentration) and development mode (combined heat injection and depressurization, hot water huff and puff) on the gas production characteristics of hydrate reservoirs, Vishal et al. [37] concluded that the water injection rate was the most critical factor affecting energy utilization efficiency; but when the permeability of the reservoir was between 50 and 100 × 10⁻³ μm², the increase in permeability could not significantly enhance gas production due to the limited energy available.

The above theories and research show that numerical simulation is important and can provide theoretical support for hydrate development. But under the dynamic phase change of hydrate, together with complex heat flow forms, the reservoir presents local thermal non-equilibrium characteristics between its fluids and solids; that is, there is an non-negligible temperature difference between the fluid and solid. Although a model based on the assumption of local thermal equilibrium can still well match the production profiles of water and natural gas in experiments or field tests by adjusting the model parameters during historical fitting, it does not really reveal the actual occurrence of local thermal non-equilibrium phenomena in the system. In the absence of sufficient historical production data, ensuring the prediction accuracy of the model would also be a challenge.

In this study, as a modification of the typical local heat equilibrium model, a pore-scale fully coupled unsteady heat transfer model was built by considering multiple forms of heat flow associated with hydrate’s decomposition and gas–liquid flow. Based on this model and the basic geological information of a gas hydrate reservoir in the western Pacific Ocean, a numerical model of gas hydrate extraction via combined heat injection and depressurization was made to simulate the production performance of a gas hydrate reservoir, and the results were fully compared with the results of extraction by depressurization alone. This study can provide theoretical support for the industrial development of gas hydrate at sea.

2. Model Building

2.1. The Difference between a Fluid–Solid Local Thermal Non-Equilibrium (LTNE) Model and Fluid–Solid Thermal Equilibrium (LTE) Model

Heat transfers in porous media are involved in many engineering fields, such as the recovery of energy in deep strata [38], electronic equipment cooling [39] and chemical catalytic reactors [40]. At present, these models can be divided into two kinds, local thermal equilibrium models (LTE) and local thermal non-equilibrium models (LTNE), according to whether the transient heat transfer process between the fluid and solid in the porous medium is considered. As shown in Figure 1, the LTE model considers the heat transfer between the fluid and the solid substrate to reach equilibrium instantaneously during the flow of the fluid in the porous medium; that is, the temperature between the fluid and the solid is equal at all times, which is usually only applicable to steady-state or transient problems [41] with stable convective heat transfer. In contrast, the LTNE model maintains that there is a large thermal resistance between the fluid and the solid, and the transient heat transfer process between the fluid and the solid in the porous medium should be considered. Therefore, when constructing an LTE model, only one energy equation is needed to characterize the temperature field distribution in the porous medium, which can be approximately regarded as the average fluid–solid temperature, but when building an LTNE model, energy conservation equations for the fluid and solid need to be built separately, and the solution to these equations can be found using the transient heat transfer equation between the fluid and solid.

2.2. Building of the Unsteady Heat Transfer Model Considering Fluid–Solid Thermal Non-Equilibrium in the Porous Medium

2.2.1. Assumptions in the Model Building

To construct this local thermal non-equilibrium model involving hydrate phase changes, the following assumptions were made:

(1) Hydrate, water and sand were incompressible, and only the variation of gas density with temperature and pressure was considered;

(2) The synthetic hydrate sediment was homogeneous, with uniform porosity, permeability and phase saturation. Although considering the heterogeneity of the three-phase sediment can make the solution more accurate, assumptions based on homogeneous sediment can obtain results with acceptable accuracy [42].

(3) It was assumed that the fluids (water and free gas) had the same temperature. Similarly, hydrate and sand have basically consistent in situ temperature change rates under the influence of the same heat flow. Therefore, only the local thermal non-equilibrium effect between the fluid and the solid phase was considered [39,41,43].

2.2.2. Energy Equation of the Solid Phase

For a solid phase with a hydrate phase change, besides the heat conduction and convective heat transfer between fluids involved in the traditional model, the heat absorption of the hydrate during its phase change also plays a key role in the evolution of the temperature field. According to the law of energy conservation, the energy equation of the solid phase was obtained [44]:

$${\left(\rho C\right)}_{\mathrm{eh}}{\displaystyle \frac{\partial \left[\left(1-{\varphi }_{\mathrm{wg}}\right)T_{\mathrm{e}}\right]}{\partial t}}=\nabla \cdot \left[\left(1-{\varphi }_{\mathrm{wg}}\right)k_{\mathrm{eh}}\nabla T_{\mathrm{e}}\right]-\Delta E_{\mathrm{h}}{\dot{m}}_{\mathrm{h}}+h_{\mathrm{sf}}{A}_{\mathrm{sf}}\left(T_{\mathrm{f}}-T_{\mathrm{e}}\right)$$

(1)

where ϕ_wg is the finite porosity of the gas and liquid flow in the hydrate reservoir, ϕ_wg = ϕ (1 − S_h) and ϕ is porosity of the sand body; (ρC)_eh is the equivalent product of the density and specific heat capacity of the solid phase, J∙m⁻³ ∙ °C ⁻¹, k_eh is the equivalent thermal conductivity of the solid phase, J∙m⁻¹∙s⁻¹ ∙ °C ⁻¹, and these two thermophysical parameters of the solid phases can be solved by weighting the volume fraction; T_e is the temperature of the solid phase, °C; t is the production time, s; $∆E_{\mathrm{h}}$ is the decomposition heat of hydrate, J∙kg⁻¹;$\dot{m_{\mathrm{h}}}$ is the rate of hydrate decomposition in the porous medium, which can be solved using the classical kinetic model of hydrate decomposition and the mass transfer area of hydrate phase change, kg∙m⁻³∙s⁻¹; h_sf is the convective heat transfer coefficient between the solid phase and fluid, W∙m⁻² ∙ °C ⁻¹; A_sf is the specific surface area in the porous medium, m⁻¹; and T_f is the fluid temperature, °C.

2.2.3. Energy Equation of the Fluid

As fluids in the hydrate’s decomposition involve complex gas–liquid two-phase flow, they have more complex forms of heat flow than the solid phase. The flow, conduction, convection and other heat flow forms of fluids in a porous medium would be reflected by the enthalpy change, so the specific enthalpy of the fluid was taken as the research object in this section, and the energy equation of the fluid was constructed as follows [45]:

$${\displaystyle \frac{\partial \left({\varphi }_{\mathrm{wg}}{\rho }_{\mathrm{f}}U\right)}{\partial t}}=-\nabla \left({\rho }_{\mathrm{f}}H{v}_{\mathrm{f}}\right)+\nabla \cdot \left({\varphi }_{\mathrm{wg}}{k}_{\mathrm{f}}\nabla T_{\mathrm{f}}\right)+h_{\mathrm{sf}}{A}_{\mathrm{sf}}\left(T_{\mathrm{e}}-T_{\mathrm{f}}\right)$$

(2)

In this equation, ρ_f is the density of the fluid, kg∙m⁻³; U is the internal energy of the fluid, J; H is the specific enthalpy of the fluid, J∙kg⁻¹; v_f is the flow velocity of the fluid m∙s⁻¹; and k_f is the thermal conductivity of the fluid, J∙m⁻¹∙s⁻¹∙°C ⁻¹. The mass balance equations used in this study and the further derivations of Equation (2) used in the simulations are detailed in Appendix A.

2.2.4. Interfacial Convective Heat Transfer Equation

In heat transfer models under local thermal non-equilibrium, the closed relationship between the fluid and solid’s energy governing equations is often constructed through the interfacial convective heat transfer equation. Therefore, an accurate description of the transient interfacial heat transfer process is the key to the local thermal non-equilibrium model. Due to the high complexity of the pore throat channels, specific surface area and solid walls in a porous medium, it is difficult to strictly characterize the interfacial heat transfer process between a fluid and solid using accurate mathematical models. Therefore, in most of the current studies, the interfacial heat transfer characteristic is characterized by constructing empirical correlations related to the Nusselt number (Nu) of the pore fluid [39,41]. In this section, the interfacial heat transfer model proposed by Kuwahara et al. [46] was used in the calculations, and its expression is as follows:

$$Nu={\displaystyle \frac{h_{\mathrm{sf}}{d}_{\mathrm{p}}}{k_{\mathrm{f}}}}=\left(1+{\displaystyle \frac{4\left(1-{\varphi }_{\mathrm{wg}}\right)}{{\varphi }_{\mathrm{wg}}}}\right)+{\displaystyle \frac{1}{2}}{\left(1-{\varphi }_{\mathrm{wg}}\right)}^{0.5}{\mathrm{Re}}^{0.6}{\mathrm{Pr}}^{1/3}$$

(3)

In this equation, Nu is the Nusselt number; d_p is the mean diameter of the sand grains (D50), m; Re is the Reynolds number; and Pr is the Prandtl number.

2.2.5. Auxiliary Equations

In addition to the central equations of the model mentioned above, a series of auxiliary equations are needed to form a complete unsteady heat transfer model. The auxiliary equations of the model detailed in this section mainly include a hydrate phase equilibrium and decomposition kinetics model [40], multiphase percolation model [47], solid–liquid heat transfer area model, physical property model of hydrate formation (effective permeability, two-phase relative permeability, capillary pressure, etc. [48,49]) and a thermal physical property equation of the components of a hydrate reservoir (methane, water, hydrate and sediment) [50].

3. Numerical Solution

3.1. Simulation of a Hydrate Reservoir in the X2 Sea Area

The water depth of the X2 sea area in the western Pacific Ocean is 1000–1500 m. The hydrate reservoir in this sea area mainly occurs in unconsolidated sediments of Quaternary argillaceous silt with a very low permeability (<10 × 10⁻³ μm²), making it much more difficult to exploit than the sandstone hydrate reservoirs in Japan and India. The development of hydrate in this sea area faces a series of technical problems related to drilling horizontal wells in shallow soft formations in deep water, such as wellhead stability, reservoir stimulation and sand control and the directional drilling of a horizontal well.

The extraction of the hydrate reservoir in the X2 sea area of the western Pacific was simulated in this study. A geothermal well in this area is 4000 m deep vertically and has a casing program as follows: 30″ casing × 1225 m + 26″ casing × 2200 m + 12-1/4″ casing × 3400 m + 9-5/8″ casing × 4000 m + 8-1/2″ openhole × 3000 m. The casing program of the hydrate production/water injection wells is 30″ casing × 1225 m + 26″ casing × 1286.8 m + 12-1/4″ casing × 1512.7 m + 9-5/8″ casing × 2121.23 m. Other important parameters used in the simulation, such as the reservoir’s physical parameters, initial temperature and pressure distribution, wellbore parameters and thermophysical parameters were taken from relevant studies [18,20,51] and are listed in

Table 1. The main parameters of the hydrate reservoir’s development.

Parameter	Value	Parameter	Value
Water depth	1225 m	Sea surface temperature	20 °C
Hydrate reservoir pressure	15.45 MPa	Hydrate reservoir temperature	17.10 °C
Permeability of the caprock layer and bottom layer	1.5 × 10−3 μm2	Permeability of the hydrate layer	2.38 × 10−3 μm2
Thickness of the caprock layer and bottom layer	20 m	Thickness of the hydrate layer	45 m
Porosity of the caprock layer and bottom layer	0.31	Hydrate layer porosity	0.373
Mixture layer permeability	6.63 ×10−3 μm2	Gas layer permeability	6.8 × 10−3 μm2
Mixture layer thickness	25 m	Gas layer thickness	20 m
Mixture layer porosity	0.346	Gas layer porosity	0.347

.

3.2. Numerical Model of the Hydrate Reservoir’s Development

A three-dimensional model 275 m × 200 m × 130 m was built to simulate the hydrate reservoir’s development. It consisted of a top caprock layer, a bottom caprock layer, a hydrate layer, a mixture layer and a free gas layer. To make full use of the free gas in the mixture layer and obtain a lower bottom-hole flowing pressure, a production wellbore about 0.108 m in radius was set at the upper part of the three-phase mixture layer during the second hydrate production test. Figure 2 shows the schematic diagram of the physical model and the gridding of the gas hydrate reservoir exploited by combined heat injection and depressurization in the sea area. Due to the low permeability of the top and bottom caprock layers, they were regarded as 30 m thick each, and the shallow formations and the seawater section were not taken into account in this study, as this was enough to accurately simulate the mass and heat transfer between the hydrate reservoir and the outer boundary layers. The upper and lower boundaries of the model were constant in temperature and pressure, and the wellbore was the pressure boundary. Due to its shallow burial depth, the hydrate reservoir was deemed good in its pore connectivity. Its initial pore pressure was calculated using the hydrostatic column pressure gradient of seawater, and its initial temperature was calculated using the geothermal gradient.

With a consideration of the complexity and strong nonlinearity of the coupling process of heat flow–hydrate decomposition dynamics in the formation and the calculation speed, the non-uniform gridding shown in Figure 2 below was used to refine the grids near the hydrate reservoir and wellbore to meet the convergence requirements of the reservoir flow parameters such as effective porosity, permeability, hydrate saturation, gas and liquid flow rate, etc., with strong nonlinear dramatic changes in the models during the hydrate development process. In the depth direction of the horizontal well (X axis), the grids were even; in the depth direction of the sea bed (Y axis), the closer to the wellbore, the denser the grids were. The grids became bigger gradually in the x direction, with the maximum grid Δx = 4 m, which still met the accuracy requirements of the numerical simulation. In the vertical direction (Z axis), as the top and bottom caprock layers had a single-phase flow, coarser grids were used, while, for the hydrate layer, mixture layer and gaseous hydrocarbon layer, especially the parts near the horizontal wellbore with a strong nonlinear flow and heat transfer and hydrate decomposition occurring, the grids were refined to meet the simulation requirements.

4. Analysis of Production Laws

4.1. Production Patterns of Fluids during Hydrate Extraction

In the base case of this study, the hydrate extraction strategy adopted was as follows: first, dual horizontal wells were used in the depressurization recovery to make full use of the energy of the formation itself, and after one year of depressurization production one of the production wells was converted into a hot water injection well to extract hydrate via combined heat injection and depressurization, making use of geothermal energy.

Figure 3 shows the curves of the gas production rate and cumulative gas production over an 8-year production period. It can be seen from the figure that the gas production rate of the new method proposed in this paper showed a trend of increasing first and then decreasing second. The whole hydrate development process can be subdivided into five stages, with the turning points of trends in the gas production rate as the dividing points:

(1) The initial stage of depressurization (0–50 days): when the reservoir was just opened, the formation’s energy and the hydrate in the near-wellbore formation were sufficient, so that with the depressurization alone the gas production rate gradually increased and reached a peak of about 4.50 × 10⁴ m³·d⁻¹ after 50 days.

(2) The remaining depressurization period (from the 50th day to 1 year): with the almost complete decomposition of the hydrate in the near-wellbore area and the drop of the formation temperature, together with the extremely low permeability of the hydrate reservoir, dominated by argillaceous siltstone, the pressure drop range was very limited, so the hydrate decomposition rate and natural gas production rate declined sharply.

(3) The early stage of heat injection (from the end of the 1st year to 2.6 years): Combined heat injection and depressurization was started 1 year after production began; at this time, the number of production wells was reduced to 1, so the gas production rate had a sudden drop at the beginning of this stage, and then the gas production rate decreased continuously to its lowest point of about 0.35 × 10⁴ m³ · d⁻¹ at about 2.6 years into production. The reasons are twofold. On the one hand, the hydrate near the wellbore had been basically decomposed completely by the end of the depressurization production stage, with the hydrate left in very low saturation, so the hot water injected had a very limited strengthening effect on hydrate decomposition before it reached the deep reservoir with higher hydrate saturations; on the other hand, the distance between the two wells was 150 m, and the gas produced by decomposition around the water injection well was not displaced into the production well by the water, so the water injection had little effect on the gas production rate in the early days of this stage and the gas production rate still declined gradually, although the rate of decline gradually slowed down.

(4) The middle stage of heat injection (2.6–6.1 years of production): In this period, a large amount of hot water was injected into the hydrate reservoir, and a large amount of gas was produced and displaced into the production well, so the gas production rate began to increase gradually. At 2.66 years of production, the gas production rate of the combined method began to exceed that of the depressurization method alone, and the gas production rate reached the peak of its second stage at 6.1 years of production (about 1.09 × 10⁴ m³·d⁻¹).

(5) The later stage of heat injection (6.0–8.0 years): as almost all the hydrate within the heat injection range decomposed, the gas production rate began to drop slowly until all the hydrate in the whole reservoir was completely decomposed.

It can be seen from the cumulative gas production curve that the cumulative gas production of the new method exceeded that of the depressurization method alone 3.34 years after the beginning of production and that the cumulative natural gas production of the depressurization method alone was 1831. 66 × 10⁴ m³, while that of the new method was 3160.90 × 10⁴ m³. Compared with the depressurization method alone, the cumulative gas production of the new method increased by 72.57%.

Figure 4 shows the curves of the water production rate and cumulative water production over the 8-year production period. It can be seen from the figure that the water production trend of the new method present in this paper is just the opposite of the gas production trend, with a decreasing section first, then an increasing section, then a decreasing section, and then finally an increasing section again. Although injecting hot water into the hydrate reservoir would increase the water production rate of a single well (at an average increase of 36.51%), the total water production rate of depressurization extraction alone would be higher, as this kind of extraction had two production wells. During the 8-year production cycle, the cumulative water production of the extraction by depressurization alone was 222. 98 × 10⁴ m³, while the cumulative water production of the new method was only 152. 19 × 10⁴ m³; 31. 75% lower than that of depressurization alone.

To sum up, the extraction of hydrate using combined geothermal heating and depressurization would have a cumulative gas production 72.57% higher and cumulative water production 31.75% lower than depressurization alone over an 8-year production cycle. This also proves that the new method is a huge upgrade of the existing method which could make the commercial development of hydrate economically viable.

4.2. Temperature and Pressure Response Characteristics of the Hydrate Reservoir

Figure 5 shows the cloud maps of the temperature field’s distribution at the end of the 1st, 2nd, 4th and 8th year of hydrate extraction. It can be seen that the temperature in the reservoir dropped gradually to a minimum of 9.34 °C due to the rapid decomposition of hydrate during the first year of depressurization production in the two wells. Compared with heat conduction, the convective heat transfer caused by fluid flow in the pores had a greater impact on the temperature field. Pore water with different temperatures in the formations above and below the production well seeped into the wellbore, resulting in the temperature field of the formation around the well having an asymmetrical distribution to its profile, with the temperature of the formation above the well lower and the temperature of the formation below the well higher. After one year of production, as the extraction switched to the mode of one hot water injection well and one production well, the temperature of the formation near the hot water injection well gradually rose, and the temperature profile gradually expanded over time. Since the hydrate saturation of the mixture layer was lower, the hydrate in the mixture layer was the first to decompose fully under the effect of the geothermal gradient; moreover, the initial permeability of the mixture layer was higher, so the hot water injected under the effect of gravity was more likely to flow into the mixture layer, thus giving rise to the temperature field distribution profile shown in the figure.

Figure 6 shows the expansion depths of the isotherms over time. It can be seen that the isotherms show two types of expansion, namely, rapid expansion and slow expansion. The isotherms, at the initial stage of heat injection, extended to the depth of the formation at a linear rate of approximately 24 m·a⁻¹, while the isotherms of 52.0 °C, 40.8 °C and 29.7 °C extended to the depth slowly, at a rate of 1.32 m·a⁻¹, 2.44 m·a⁻¹ and 6.19 m·a⁻¹, respectively. The simulation results show that the ultimate distance of the 52.0 °C isotherm expansion was about 48 m, which also proves that heat injection has little effect on the decomposition of hydrate deep in the formation. The higher the temperature is, the shorter the duration of the high-speed linear expansion of the isotherm is. The 52.0 °C isotherm only lasted for about one year, while the 29.7 °C isotherm lasted for about five years. The reason for this is that the influence range of the heat injection in the initial stage of production was relatively small, and the hydrate near the wellbore had almost all decomposed in the depressurization production stage; without heat absorption due to hydrate decomposition, the high temperature profile rapidly pushed forward, but, with the increase in the influence range of the heat injection, the heat absorption due to hydrate decomposition, and the drop of the returning fluid temperature of the geothermal layer, the expansion of the isotherm profile became slower and slower.

Figure 7 shows the cloud maps of the pressure field distribution at the end of the 1st year, 2nd year, 4th year and 8th year of hydrate extraction. Clearly, the pressure drop range during the whole production cycle was very small, and the 8 MPa isobar only extended to 18.91 m after one year of depressurization production. This is mainly because, although the hydrate in the near-wellbore zone almost completely decomposed, the permeability of the hydrate reservoir dominated by shale and silt was still very low, and the resistance to gas/water flow was still very large, significantly blocking the expansion of the pressure drop range. This is also the key reason why the productivity of depressurization method alone decreased significantly when almost all the hydrate in the near wellbore area decomposed. Figure 8 shows the pressure distribution curve between the injection and production wells. It can be seen that when one of the production wells was converted into a water injection well, the formation pressure increased as a whole, and the pressure gradient along the horizontal direction also became larger, making the natural gas production rate of the single well higher than that of the two wells. With the gradual decomposition of hydrate near the wellbore, and under the condition of water injection at a constant pressure, both the water injection rate and water production rate increased accordingly, and then the flow resistance increased accordingly, so the formation pressure near the production well increased gradually over time, while the formation pressure near the water injection well decreased gradually over time.

4.3. Spatial Distribution of Saturation

Figure 9 shows the cloud maps of the hydrate saturation’s spatial distribution at the end of the 1st, 2nd, 4th and 8th year of hydrate extraction. It can be seen from the figure that the hydrate in the mixture layer decomposed first in the depressurization stage of the first year. The reasons for this are twofold. First, the mixture layer had a low initial hydrate saturation of only 11.7%; second, just as the simulation results for the temperature field above show, when the high-temperature water in the lower formation flowed to the production wellbore, it made the temperature of the mixture layer rise and the hydrate decomposition accelerate. Meanwhile, the low-temperature pore water in the upper layer flowed through the hydrate layer, so the decomposition rate of hydrate in the hydrate layer was slower. In addition, the hydrate layer had an initial permeability (2.38 × 10⁻³ μm²) much lower than that of the mixture layer (6.63 × 10⁻³ μm²), and thus a weaker convection of the fluid. When heat injection started on the right side, the high-temperature hot water significantly increased the reservoir temperature, thereby strengthening the hydrate decomposition in the porous medium, so the hydrate decomposition range around the water injection well on the right side was larger. However, as the influence range of the heat injection was very limited, as shown by the simulation results in the above section, the ultimate expansion distance of the 52.0 °C isotherm was only 48 m and, after 8 years of production, a large amount of hydrate remained in the upper part near the production-well side, undecomposed. Therefore, it is necessary to optimize the design of the injection–production wells’ spacing and well locations to further improve the efficiency of natural gas recovery.

Figure 10 shows the cloud maps of the gas saturation’s spatial distribution at the end of the 1st, 2nd, 4th and 8th year of hydrate extraction. It can be seen from the figure that the water injection not only enhanced the kinetics of hydrate decomposition in the porous medium, but also redirected the decomposed free gas in the formation to flow into the production well, making the productivity of the natural gas increase further. With the complete decomposition of hydrate in the mixture layer, the mixture layer did not produce free gas anymore, and the free gas zone in the hydrate reservoir became thinner gradually. It can be seen from the cloud maps of the gas saturation distribution in the 4th and 8th years that a large amount of free gas was displaced into the top caprock layer under the effect of water flooding, which not only had a negative impact on gas productivity, but also could bring about the risk of gas leakage [51,52] if the gas flowed further toward the mud line. As the overlying caprock of the hydrate reservoir has an extremely low permeability, methane leakage [51] at the seabed mudline has not been found during monitoring in short-term hydrate test production, but preventing methane leakages will be a serious safety issue [53] during the long-term commercial extraction of hydrate.

5. Conclusions

The traditional theory of local thermal equilibrium can introduce a large error into the yield assessment of hydrate extraction, which is a process drastically affected by thermal effects. In this study, a heat transfer model that considered local thermal non-equilibrium was proposed. According to the geological conditions of a hydrate reservoir and deep high-temperature geothermal layer in the X2 area of the western Pacific Ocean, a numerical model of gas hydrate extraction via combined heat injection and depressurization was constructed to simulate the extraction dynamics of hydrate reservoirs. The following main conclusions have been reached:

(1)

The whole hydrate development process can be subdivided into five stages, with the turning points of the trends in the gas production rate as the dividing points: the initial stage of depressurization (0–50 days), the remaining depressurization period (from the 50th day to 1 year), the early stage of heat injection (from the end of the 1st year to 2.6 years), the middle stage of heat injection (2.6–6.1 years of production), and the later stage of heat injection (6.0–8.0 years).

(2)

In this 8-year production cycle, hydrate extraction by depressurization alone would have a cumulative gas production of 18.3166 million m³ and a cumulative water production of 2.2298 million m³, in comparison, the hydrate extraction conducted using a combined heat injection and depressurization would have a cumulative gas production of 31.609 million m³ and a cumulative water production of 1.5219 million m³, which are 72.57% higher and 31.75% lower than those of depressurization alone, respectively.

(3)

As the hydrate reservoir, dominated by argillaceous siltstone, had extremely low permeability, the 8 MPa isobar only extended to 18. 91 m after one year of depressurization extraction, with the pressure drop range very limited. When water was injected, the pressure gradient between the injection well and production well rose significantly. At the early stage of water injection, the isotherms also expanded at an approximate linear rate of 24 m·a⁻¹, before the isotherms of 52.0 °C, 40.8 °C and 29.7 °C the expanded forward slowly at a rate of 1.32 m·a⁻¹, 2.44 m·a⁻¹ and 6.19 m·a⁻¹, respectively. The ultimate extension depth of the 52.0 °C isotherm was only about 48 m.

The results of this study provide significant guidance on the kinetics of hydrate decomposition and a detailed assessment of our capacity for hydrate extraction. As several empirical models were used, the accuracy of the existing model needs to be further validated using future field-scale experimental data.