© Z. Xia et al., published by IFP Energies nouvelles, 2021

1 Introduction

Natural gas hydrates (mainly methane hydrates) are crystalline solids stabilized under low-temperature and high-pressure environments, which have been considered a potential energy supply since the mid-1960s due to their wide distribution and enormous reserves [1]. Therefore, the effective development of hydrate reservoirs (i.e., the methane recovery) is of great significance.

Currently, depressurization, heat stimulation, chemical inhibitor stimulation and gas exchange (CO₂) are the most investigated methods for hydrate reservoir development [2]. Among these methods, the depressurization method is acknowledged as the most potential one due to its simplicity and efficiency. Many studies in the depressurization method have been conducted, such as sensitivity analysis of laboratory-scale parameters [3], heat transfer behavior [4], hydraulic fracturing enhanced depressurization [5], decreasing bottom-hole pressure [6], hydrate production trials in the Mallik area [7], Eastern Nankai Trough [8] and South China Sea [9], and relevant simulation studies [10–12].

The gas exchange method was initially proposed by Ohgaki et al. [13]. When the environment temperature was lower than 10 °C (283 K), the CO₂ hydrate was more stable than the CH₄ hydrate (Fig. 1). The CH₄ hydrate will dissociate, and the CO₂ hydrate will form if the reservoir condition falls into the shadow area in Figure 1. Therefore, CO₂ injection into hydrate reservoirs can enhance CH₄ recovery and sequester CO₂ (the well-known greenhouse gas) under a suitable P-T range. This is the main reason why the gas exchange method has received special attention in recent years. Lee et al. [14] investigated the influence of the CO₂ addition on CH₄ + C₃H₈ hydrates from thermodynamic, microscopic and kinetic aspects, and found high replacement degree at the high CO₂ pressure. Gharasoo et al. [15] developed a model of CO₂ injection into a gas-hydrate-filled pressure vessel, and evaluated the effective gas hydrate dissociation rate into gas and water by fitting the experimental results. Shagapov et al. [16] developed the mathematical model of the warm CO₂ injection into a CH₄-saturated hydrate reservoir and concluded that the CO₂ hydrate formation heat had a great impact on the CH₄ hydrate dissociation rate. Khasanov et al. [17] also developed a similar mathematical model of liquid CO₂ injection into the hydrate reservoir and analyzed four different regimes during the process. All these investigations showed the feasibility of the methane recovery from hydrate reservoirs through CO₂ injection.

However, the effect of pure CO₂ injection into the hydrate reservoir may be quite limited, and experimental studies have shown that the existence of N₂ in the CO₂ stream can improve the sweeping region and recovery efficiency. Koh et al. [18] investigated the fluid flow characteristics and exchange mechanisms of continuous CO₂/N₂ injection into a one-dimensional high-pressure reactor containing the hydrate, and they found that the exchange efficiency was inversely proportional to the gas injection rate. Yang et al. [19] investigated the dissociation of the CH₄ hydrate stimulated with flue gas flooding through a sand pack model, and they concluded that the flue gas injection was potential for hydrate production and CO₂ sequestration. Li et al. [20] investigated the fracture-filled CH₄ hydrate dissociation by CO₂ and CO₂/N₂ injection, respectively, and they found that the presence of N₂ could enhance the final accumulative methane recovery ratio and exchange rate. In addition to N₂, Sun et al. [21] also studied the hydrate dissociation using the CO₂/H₂ mixture injection, and they observed good CH₄ hydrate recovery ratio and CO₂ storage capability.

Although the gas exchange method is promising, the low production rate or efficiency is the main disadvantage of this method. Merey et al. [22] reviewed the CH₄–CO₂ exchange studies and pointed out that the combination of depressurization and gas exchange was promising for hydrate reservoir development. Okwananke et al. [23] experimentally studied the enhanced-depressurization for CH₄ hydrate dissociation by injection of compressed air and nitrogen, and they found that the CH₄ hydrate was quickly dissociated and concluded that the gas injection was a potential approach to improve conventional depressurization method for the hydrate reservoir development. Based on extensive experimental and numerical investigations, the Iġnik Sikumi Field production trial was carried out to assess the CH₄ production potential with CO₂/N₂ stimulation in 2012. In this production trial, a huff and puff method was used which was a combination of depressurization and gas exchange, and the “bulk exchange” may have occurred [24].

To investigate the possibility of simultaneous CH₄ production and CO₂ sequestration in a field scenario, this paper proposed a cyclic CH₄ hydrate production method stimulated by CO₂ and N₂ mixture based on the production trial mentioned above. An actual-parameter based simulation model was established, the CH₄ hydrate production characteristics were evaluated, and a sensitivity analysis was carried out.

2 Model description

2.1 Production method

The injected gas in this paper is a mixture of CO₂ and N₂, and the existence of a N₂ hydrate is ignored because it is difficult to form under typical hydrate reservoir conditions (Fig. 1). The cyclic production scheme [25] is adopted to conduct this research, which is widely used in the heavy oil production by high-temperature steam stimulation in the petroleum industry. The whole production process is composed of several production cycles and each production cycle includes the injection stage, the soaking stage and the production stage. The well is opened in the injection and production stage, and during the soaking stage, the well is shut in in order that the injected gas and pressure disturbance can originate in the depth of the reservoir.

2.2 Simulator selection

The methane recovery from the hydrate reservoir is a complicated process involving phase transition, heat and mass transfer, multi-phase flow, etc. Numerical simulation is an effective way to investigate the possible hydrate production behaviors and several codes/simulators have been used to carry out relevant analyses. In this paper, the CMG-STARS [26] was selected to conduct the simulation study. This simulator was used to simulate hydrate reservoir production in the Mount Elbert area [27], and it obtained long-term production characteristics similar to other tested simulators. In addition, it has also been used for sensitivity studies of CH₄ hydrate production [28], economic studies [29], gas production behavior [30], and CO₂ hydrate formation in depleted gas reservoirs [31]. The base model was established using CMG-STARS by referring to the reservoir and operation parameters of the Iġnik Sikumi Field production trial.

2.3 Fundamentals

2.3.1 Mass and energy conservation

Assuming that the hydrates in the reservoir are pure CH₄ hydrates, the system is composed of the immobile phase (hydrate phase) and the mobile phases (aqueous and gas phases), and the flows of the gas and aqueous phases follow Darcy’s law. The relationships among the components and phases in this study are shown in Table 1. For each single cell, the spatially discretized conservation equations of each component and the energy are shown in equations (1), (2) and (3), respectively.

For the conservation of the flowing component i (i.e., CH₄, CO₂, N₂ or H₂O):

$$\frac{\mathrm{\partial }}{\mathrm{\partial }t}\left[V_f({\rho }_A{S}_A^e{w}_i+{\rho }_G{S}_G^e{y}_i)\right]=\sum _{n=1}^{n_f}\mathrm{}\left[{\left(\frac{A}{l}\right)}^e{k}^e\left(\frac{k_{\mathrm{rA}}}{{\mu }_A}{\rho }_A{w}_i\Delta p_A+\frac{k_{\mathrm{rG}}}{{\mu }_G}{\rho }_G{y}_i\Delta p_G\right)\right]+V_b\sum _{n=1}^{n_t}\mathrm{}\left(s_{\mathrm{ni}}^{\mathrm{\prime }}-s_{\mathrm{ni}}\right)r_n+q_i,$$(1)

where V_f is the volume occupied by the mobile phases; V_b is the apparent volume of the cell; ρ_A and ρ_G are the density of the aqueous and gas phases, respectively; w_i and y_i are the mass percent of component i in the aqueous and gas phases, respectively; n_f is the number of neighboring cell faces; ${\left(\frac{A}{l}\right)}^e$ is the ratio of the effective area and distance between interfaces; k^e is the effective permeability at the interface; p_A and p_G are the pressure of the aqueous and gas phases, respectively; n_t is the number of chemical reactions; s^′_ni and s_ni are the product and reactant stoichiometric coefficients of component i, respectively; r_n is the volumetric reaction rate; and q_i is the mass source/sink from the injection/production wells.

For the conservation of the solid component i (i.e., CH₄·NH₂O or CO₂·NH₂O):

$$\frac{\mathrm{\partial }}{\mathrm{\partial }t}\left[V_v{c}_i\right]=V_b\sum _{n=1}^{n_f}\mathrm{}\left(s_{\mathrm{ni}}^{\mathrm{\prime }}-s_{\mathrm{ni}}\right)r_n,$$(2)

where V_v is the volume occupied by the mobile and immobile phases and c_i is the volumetric concentration of the solid component i.

For energy conservation:

$$\frac{\mathrm{\partial }}{\mathrm{\partial }t}\left[V_f({\rho }_A{S}_A^e{U}_A+{\rho }_G{S}_G^e{U}_G)+V_v{c}_S{U}_S+V_r{U}_r\right]=\sum _{n=1}^{n_f}\mathrm{}\left[{\left(\frac{A}{l}\right)}^e{k}^e\left(\frac{k_{\mathrm{rA}}}{{\mu }_A}{\rho }_A{H}_A\Delta p_A+\frac{k_{\mathrm{rG}}}{{\mu }_G}{\rho }_G{H}_G\Delta p_G\right)\right]+\sum _{n=1}^{n_f}\mathrm{}{\left(\frac{A}{l}\right)}^e{\lambda }^e\Delta T+V_b\sum _{n=1}^{n_r}\mathrm{}{H}_{\mathrm{rn}}{r}_n+q_{e,}$$(3)

where V_r is the rock volume (solid inert matrix, rock grains); c_S is the total solid concentration; U_r is the internal energy per rock volume; U_A, U_G and U_S are the internal energy of the aqueous, gas and solid phases, respectively; H_A, H_G and H_rn are the enthalpy of the aqueous phase, gas phases and reaction n, respectively; λ^e is the effective thermal conductivity at the interface; T is the temperature; and q_e is the heat source/sink from the injection/production wells.

2.3.2 Permeability model

When the hydrate exists, the fluid phase permeability in the porous medium will decrease [32]. Currently, there are two methods to treat the permeability variation in the presence of the hydrate [33], namely, the variable absolute permeability method and the constant absolute permeability method. For the first method, the reservoir absolute permeability is modeled as a function of the hydrate saturation; in addition, the mobile phase relative permeability changes with the effective phase saturation. In the second method, the reservoir absolute permeability is constant, while the mobile phase relative permeability changes with the actual phase saturation. We adopted the first method to conduct our research. The definitions of the fluid phase permeability, effective phase saturation and actual phase saturation are shown in equations (4), (5) and (6), respectively:

$$k_{\beta }=k_a{k}_{r\beta },$$(4)

where k_β is the effective permeability of phase β; k_a is the absolute permeability of the hydrate reservoir; and k_rβ is the relative permeability of phase β:

$$S_{\beta }^e=\frac{V_{\beta }}{V_f},$$(5)

where S^e_β is the effective saturation of phase β.

$$S_{\beta }=\frac{V_{\beta }}{V_v},$$(6)

where S_β is the actual saturation of phase β.

(1) Absolute permeability model (Carman-Kozeny-type formula)

$$k_a=k_{a0}{\left(\frac{{\phi }_f}{\phi }\right)}^m{\left(\frac{1-\phi }{1-{\phi }_f}\right)}^2,$$(7)

where k_a0 is the reservoir absolute permeability without the existence of the gas hydrate; φ and φ_f are the reservoir porosity and fluid porosity, respectively; and m is the model parameter, which is set to 4.3413 by transforming the Civan’s permeability-porosity relationship from Anderson et al. [34].

(2) Relative permeability model

The flows of the mobile phases follow Darcy’s law, and the relative permeability models are shown in equation (8):

$$\begin{array}{c}{k}_{\mathrm{rA}}={\left[\frac{S_A^e-S_{\mathrm{irA}}}{1-S_{\mathrm{irA}}}\right]}^{n_A}\\ k_{\mathrm{rG}}={\left[\frac{S_G^e-S_{\mathrm{irG}}}{1-S_{\mathrm{irA}}}\right]}^{n_G},\end{array}$$(8)

where k_rA and k_rG are the relative permeability of the aqueous and gas phases, respectively; S_irA and S_irG are the irreducible aqueous and gas saturation, respectively; n_A is the model parameter, which is set to 5.04 [35]; and n_G is the model parameter, which is set to 3.16 [35].

2.3.3 Capillary pressure model

The capillary pressure between the gas phase and the aqueous phase is shown in equation (9):

$$p_c=-p_{c0}{\left[{\left(\frac{S_A^e-S_{\mathrm{irA}}}{1-S_{\mathrm{irA}}}\right)}^{-1/\lambda }-1\right]}^{1-\lambda },$$(9)

where p_c is the capillary pressure; p_c0 is the model parameter, which is set to 10⁴ Pa [35]; and λ is the model parameter, which is set to 0.77437 [35].

2.3.4 CH₄/CO₂ hydrate dissociation/formation model

Two chemical reactions shown in equations (10) and (11) are considered:

$$\mathrm{C}{\mathrm{H}}_4\cdot \mathrm{N}{\mathrm{H}}_2\mathrm{O}\leftrightarrow \mathrm{C}{\mathrm{H}}_4+\mathrm{N}{\mathrm{H}}_2\mathrm{O}+\Delta {\mathrm{H}}_1,$$(10)

$$\mathrm{C}{\mathrm{O}}_2\cdot \mathrm{N}{\mathrm{H}}_2\mathrm{O}\leftrightarrow \mathrm{C}{\mathrm{O}}_2+\mathrm{N}{\mathrm{H}}_2\mathrm{O}+\Delta {\mathrm{H}}_2.$$(11)

where ΔH₁ and ΔH₂ are the enthalpy of CH₄ hydrate dissociation/formation and CO₂ hydrate dissociation/formation, which are set to 56.84 kJ mol⁻¹ [36] and 65.22 kJ mol⁻¹ [36], respectively.

(1) Hydrate dissociation rate model

In 1987, Kim and Bishnoi proposed the widely used CH₄ hydrate dissociation kinetic model based on experimental results [37, 38]. In their model, the dissociation rate was proportional to the hydrate particle surface area or dissociation area and to the difference in the methane fugacity at the equilibrium and dissociation pressures. The fugacity can be approximated with an equivalent pressure when setting the fugacity coefficient equal to 1.0 [39].

We assume that both the dissociation and formation of the CH₄ hydrate and CO₂ hydrate follow the Kim-Bishnoi model. The expression of the CH₄/CO₂ hydrate dissociation rate can be expressed as follows: [40]

$$\frac{\mathrm{d}{c}_{\mathrm{Hyd}}}{\mathrm{d}t}=\left(\frac{k_d^0{A}_{\mathrm{HS}}}{{\rho }_w{\rho }_h}\right)\left({\phi }^2{\rho }_w{\rho }_h{S}_A{S}_H{p}_e\right)\exp \left(-\frac{\Delta E_d}{\mathrm{RT}}\right)\left(1-\frac{y}{K\left(p,T\right)}\right),$$(12)

where c_Hyd is the mole quantity of the CH₄/CO₂ hydrate per unit volume; $k_d^0$ is the intrinsic dissociation rate constant of CH₄/CO₂ hydrate (Tab. 2); A_HS is the reaction specific area, which is 750,000 m²/m³ (i.e., hydrate particles are assumed to be regular spheres with a diameter of 8 μm [41]); ρ_w is the density of the aqueous phase, 1000 kg/m³; ρ_h is the density of the CH₄ hydrate or CO₂ hydrate, 919.7 kg/m³ or 1100 kg/m³ [40]; ΔE_d is the activation energy of the dissociation reaction, which is 81 kJ/mol [41] and 102.88 kJ/mol [42] for the CH₄ hydrate and CO₂ hydrate, respectively; R is the gas universal constant; p_e is the equilibrium pressure; y is the mole fraction of CH₄/CO₂ in gas phase; and K is the equilibrium ratio, as follows:

$$K=\left(\frac{a_1}{p_g}\right)\exp \left(\frac{a_2}{T-a_3}\right),$$(13)

where p_g is the gas phase pressure; a₁, a₂ and a₃ are model parameters, and they are calculated based on the experimental results of Adisasmito et al. [43] (Tab. 3), where the average fitting error is 0.46% and 0.39% for the CH₄ hydrate and CO₂ hydrate, respectively.

(2) Hydrate formation rate model

Similar to equation (12), the expression of the CH₄/CO₂ hydrate formation rate (including nucleation and growth) can be expressed as equations (14) and (15).

For the hydrate nucleation process:

$$\frac{\mathrm{d}{c}_{\mathrm{Hyd}}}{\mathrm{d}t}=\left(\frac{k_f^0{A}_{\mathrm{HS}}}{{\rho }_w}\right)\left(\phi {\rho }_w{S}_A{p}_e\right)\exp \left(-\frac{\Delta E_f}{\mathrm{RT}}\right)\left(1-\frac{y}{K\left(p,T\right)}\right).$$(14)

For the hydrate growth process:

$$\frac{\mathrm{d}{c}_{\mathrm{Hyd}}}{\mathrm{d}t}=\left(\frac{k_f^0{A}_{\mathrm{HS}}}{{\rho }_w{\rho }_h}\right)\left({\phi }^2{S}_A{S}_H{\rho }_w{\rho }_h{p}_e\right)\exp \left(-\frac{\Delta E_f}{\mathrm{RT}}\right)\left(1-\frac{y}{K\left(p,T\right)}\right),$$(15)

where $k_f^0$ is the intrinsic formation rate constant of the CH₄/CO₂ hydrate (Tab. 2), and ΔE_f is the activation energy of the formation reaction, which is assumed to be equal to that of the dissociation reaction for the CH₄ hydrate or CO₂ hydrate.

2.4 Base model

2.4.1 Reservoir parameters

As shown in Table 4, the model is established based on the reservoir parameters of “C–1 sand” [24, 44] (one of four separate hydrate-bearing units according to the Ignik Sikumi well-log data) and is composed of the overlying layer (impermeable), the Hydrate-Bearing Layer (HBL), and the bottom layer (impermeable) with thicknesses of 10 m, 9 m and 10 m, respectively. The model size is 102 m × 102 m × 29 m with a grid division of 51 × 51 × 28. At the initial CH₄ hydrate saturation of 0.72 (corresponding to the concentration of 5539 mol/m³ in the base model), the effective permeability and fluid porosity of the HBL are 1.82 mD and 11.2%, respectively. The sketch of the base model are shown in Figure 2.

2.4.2 Operation parameters

The hydrate reservoir is produced by cyclic gas injection with a composition of 22.5 mol% CO₂ and 77.5 mol% N₂. The maximum Bottom-Hole Pressure (BHP) was maintained lower than the formation breakdown pressure (10 MPa) for operation safety during the gas injection stage. A total of 5 production cycles (250 d) were conducted. The main operation parameters are shown in Table 5.

3 Results and discussion

3.1 Production characteristics

Figures 3 and 4 show the production volume and the production rate of each gas component, respectively. The gas production rate changes periodically, where the highest production rate occurs at the beginning in each cycle, because the diving force (i.e. the production pressure difference) is the largest at this time. A total of 63 250 m³ (SCM) of gas is produced at the end of the simulation, where CH₄ accounts for 48.78%. The peak gas production rate of each production cycle increases with the increase of the cycle number, where the highest value reaches 8090 m³/d. From Table 6, the total production volume of CH₄ remains almost the same except the first cycle, and the percent of the CH₄ in the total gas production accounts for about 50% in each cycle. At a constant BHP of 4.5 MPa, if no gas was injected into the hydrate reservoir, there would be no CH₄ production (the curve labelled with “depressurization (4.5 MPa)” in Fig. 3) when the BHP maintains at 4.5 MPa. This is because it is higher than the corresponding equilibrium pressure of the CH₄ hydrate (4.34 MPa) at the reservoir condition. Besides, the total CH₄ produced by the cyclic production method is 50% more than that by the depressurization method at the BHP of 3 MPa (the curve labelled with “depressurization (3 MPa)” in Fig. 3). This shows the production potential of the cyclic production. Depressurization method obtains the CH₄ production by lowering the reservoir pressure which is economical and simple. However, for low-pressure hydrate reservoirs, the minimum BHP usually cannot be smaller than 2.75 MPa to avoid potential ice formation, and thus the available pressure-reduction space is quite limited, which affects the production. Under this situation, using the cyclic production can enhance CH₄ production.

The injected CO₂ and N₂ can cause the partial pressure to change in the gas component and thus affect the stability of the gas hydrate. For the CH₄ hydrate, using the 1st cycle as an example, the partial pressure of CH₄ in the gas phase and the area below/above the corresponding equilibrium pressure at the end of the injection stage are shown in Figures 5 and 6, respectively. From Figure 5, the average radius of pressure propagation into the hydrate reservoir is about 25 m. The partial pressure of CH₄ in the gas phase changes greatly due to the introduction of substitute gas molecules, and the partial pressure of CH₄ in the near-wellbore area is lower than the equilibrium pressure, leading to the dissociation of the CH₄ hydrate. At the end of the simulation, the CH₄ hydrate saturation in the near-wellbore area decreased significantly (Fig. 7), and the CH₄ hydrate was highly dissociated in the upper part of the HBL, where the injected gas preferentially sweeps due to its smaller gravity compared with water. At the end of production, the dissociation percent [45] of the CH₄ hydrate is approximately 0.759%. The maximum radius of the significantly hydrate-dissociated area is about 20 m in the plane with an average value of 5 m. Therefore, the cyclic gas stimulation method can enhance the dissociation of the CH₄ hydrate, and the depressurization mechanism (partial pressure reduction) is dominant because the CO₂ hydrate formation is quite limited as discussed below.

The partial pressure change in CO₂ can also cause the dissociation and formation of the CO₂ hydrate. At the end of the simulation, a total of 33 556 m³ (SCM) of gas mixture is injected, where 7550 m³ is CO₂. Approximately 3.5% of CO₂ is stored in the reservoir (Tab. 7), and only a relatively small quantity of CO₂ hydrate is formed. The high-concentration CO₂ is mainly in the near-wellbore area, but cyclic production can cause near-wellbore pressure to rise and drop periodically each cycle which can affect CO₂ storage. On the other hand, the partial pressure of CO₂ in the gas phase is mostly lower than the equilibrium value during the cyclic production, and this is not beneficial for CO₂ hydrate formation. Therefore, the cyclic production method may not effectively store CO₂, especially as a CO₂ hydrate.

3.2 Change of temperature, permeability and fluid porosity

During cyclic production stimulation, the CH₄ hydrate dissociation is dominant compared with the CO₂ hydrate formation, which can be inferred from the temperature distribution (Fig. 8), because the hydrate dissociation is an endothermic process. At the end of the simulation, the reservoir temperature decreases significantly with the maximum radius of 32 m, especially in the near-wellbore area where the lowest temperature approaches 1.5 °C. The similarity between the temperature distribution (Fig. 8) and the CH₄ hydrate concentration distribution (Fig. 7) shows that heat supply is quite essential for continuous hydrate dissociation. Besides the latent heat in the HBL, the latent heat contained in the overlying and bottom layers also plays an important part in the CH₄ hydrate dissociation (Fig. 8).

The effective permeability is the key factor affecting the pressure propagation, thus controlling the gas or water flow in the reservoir. It is correlated with the fluid porosity through equation (7), which is determined by the solid phase saturation, i.e. hydrate saturation. From Figures 9 and 10, the effective permeability and the fluid porosity of the hydrate reservoir increase significantly due to the CH₄ hydrate dissociation, with the highest permeability close to the reservoir absolute permeability (1D) and the highest fluid porosity close to the reservoir porosity (40%). This shows that the seepage condition improves as the production continues, and it is favorable for the water or gas flow in the hydrate reservoir during the injection or production stage.

3.3 Sensitivity analysis

The values of reservoir or operation parameters can have significant influence on the production performance of the hydrate reservoir. Using the base model, the sensitivity analyses of several parameters were conducted, including the initial CH₄ hydrate saturation, absolute permeability, intrinsic CO₂ hydrate formation rate constant, injected gas composition and production time.

3.3.1 Initial CH₄ hydrate saturation

The initial CH₄ hydrate saturation value was set to 0.2, 0.4, 0.6 and 0.8. From Figure 11, the CH₄ production volume increases when the initial CH₄ hydrate saturation increases from 0.2 to 0.6, but the CH₄ hydrate dissociation percent and the CO₂ storage percent decrease. The production is dominated by the reserve (CH₄ hydrate saturation), but a high CH₄ hydrate saturation is not helpful for CO₂ storage due to low gas injectivity (Tab. 8). When the saturation value changes to 0.8, the influence of the poor seepage condition is more significant, and the CH₄ production volume and the total amount of injected gas decrease.

3.3.2 Absolute permeability

The absolute permeability was set to 0.01, 0.1, 1 and 10 D, and the results are shown in Figure 12. With the increase in the absolute permeability, the CH₄ production volume and the CH₄ hydrate dissociation percent increase due to the improvement of the seepage condition, while the CO₂ storage percent decreases significantly.

However, when the permeability increases to 10 D, the CH₄ production will decrease. This is because when the absolute permeability value is too high (10 D for example), the injected gas will mainly accumulate in the upper part of the hydrate reservoir because of its lower gravity, which limits its effect in enhancing the CH₄ production. Taking the 1st cycle for an example (Fig. 13), the total sweeping volume of the injected gas is more uniform and larger at the end of gas injection when the absolute permeability is 1 D. Thus, the dissociation region of the CH₄ hydrate is larger in the plane but smaller in the vertical area (Fig. 14) compared with the situation of 1 D (Fig. 7). The overall dissociation region of the CH₄ hydrate is limited when the absolute permeability is 10 D, resulting in a decline in CH₄ production.

3.3.3 Heterogeneity of absolute permeability

To investigate the influence of the vertical heterogeneity of the absolute permeability. Three situations are considered, i.e. scheme 1 (negative rhythm), scheme 2 (homogeneous), and scheme 3 (positive rhythm), as shown in Table 9. For scheme 1, the value of the absolute permeability increases from the top layer to the bottom layer of the HBL, and scheme 3 is just the opposite. The results of the sensitivity studies are shown in Figure 15.

From Figure 15, for the CH₄ production, scheme 1 is the best, followed by scheme 2 and scheme 3 in sequence. This is because the injected gas prefers to accumulate in the upper layer of the hydrate reservoir, and high permeability can assist the injected gas to take effect to increase the CH₄ production, as shown in Figure 16. From Figure 16, however, the CH₄ hydrate dissociation area is more uniform in scheme 3 than in scheme 1. That is, the negative rhythm is the most preferred scheme for CH₄ production, but the least preferred one for CO₂ storage.

3.3.4 Intrinsic CO₂ hydrate formation kinetic constant

The parameter values of hydrate kinetics are taken from bulk phase systems in this paper because these values are less accessible in porous media [46–48]. Besides, the measurement of the hydrate formation kinetic constant is much more difficult, which also exhibits a weak correlation to the temperature [49, 50]; thus, there is great uncertainty. For the reasons presented above, we carried out the sensitivity study on the intrinsic CO₂ hydrate formation kinetic constant.

The intrinsic CO₂ hydrate formation kinetic constant was set to 5, 10, 100 and 1000 k_f⁰, where k_f⁰ is the value in the base model. The high CO₂ hydrate formation kinetic constant is unfavorable for hydrate production but favorable for CO₂ storage (Fig. 17), especially when it is higher than 10 k_f⁰. This trend is observed because as the CO₂ hydrate formation kinetic constant increases, the CO₂ hydrate quick formation in the near-wellbore area weakens the CH₄ hydrate dissociation. The CO₂ hydrate formation degree is more significant than the CH₄ hydrate dissociation, and this can be deduced from the temperature distribution of Figure 18.

3.3.5 Injected gas composition

The CO₂ mole fraction in the injected gas was set to 0.3, 0.4, 0.5, 0.6 and 0.7, and the results are shown in Figure 19. The injected gas composition has little effect on the CH₄ hydrate dissociation percent and CO₂ storage percent. This result is observed because in the established model, there is no essential difference in the role of the pressure reduction by CO₂ and N₂, as there is only a small amount of CO₂ hydrate formed. The CH₄ production volume slightly decreases with the increase of the CO₂ mole fraction in the injected gas, because these small-amount formed CO₂ hydrate may influence the seepage condition. From this aspect, smaller CO₂ mole fraction in the mixture is more beneficial to the CH₄ production, and pure N₂ can have the best production performance.

3.3.6 Injected gas temperature

The injected gas temperature was set to 50, 100, 150 and 200 °C, and the results are shown in Figure 20. The CH₄ hydrate dissociation percent and CO₂ storage percent vary little with the injected gas temperature. The CH₄ production volume slightly increases with the increase of the gas temperature. This is because the injected gas conveys a small amount of heat due to its low heat capacity, and the latent heat contained in the reservoir provides the most consumed heat for the CH₄ hydrate dissociation.

3.3.7 Injection time

The injection time of each cycle was set to 5, 10, 15, 20 and 25 days. From Figure 21, a longer injection time can enhance the CH₄ production volume and the CH₄ hydrate dissociation percent. More gas will be injected into the reservoir as the injection time increases, and this can induce more hydrate to dissociate. The CH₄ production volume increases about 164% as the injection time increases from 5 days to 25 days, but the increase of the CH₄ production slows down when the injection time is beyond 15 days. However, longer injection time is unfavorable for CO₂ storage and the CO₂ storage percent decreases from 6.03% to 2.39% as the injection time increases from 5 days to 25 days.

3.3.8 Production time

The production time was set to 1, 2, 3 and 4 years. From Figure 22, a longer production time can enhance the CH₄ production volume and the CH₄ hydrate dissociation percent. After a four-year production, about 2.27% of the CH₄ hydrate has been dissociated compared to the dissociation percent of 1.03% after a one-year production. The CO₂ storage percent will decrease to a stable low level as the production time increases.

4 Summary and conclusion

This paper proposed the cyclic methane hydrate production method with CO₂ and N₂ mixture stimulation, and investigated the field-scale production using this method based on a numerical model according to actual hydrate reservoir parameters. The following conclusions were obtained:

1. The depressurization mechanism is dominant in cyclic stimulation with the CO₂ and N₂ mixture; in addition, the CH₄ hydrate can be significantly produced due to the effective reduction of the partial pressure of CH₄ in the gas phase, and the CH₄ hydrate in the upper part of the HBL is highly dissociated. However, the effect of CO₂ storage is limited, and a continuous CO₂ injection (i.e., CO₂ flooding) may be better for CO₂ storage. The waste gas produced from the plant (mainly composed of CO₂ and N₂) may be directly injected into the hydrate reservoir for production.

2. In the near-wellbore area, the hydrate dissociation degree is relatively high, and the fluid porosity and effective permeability increase significantly during the cyclic stimulation process. Heat supply plays an important role in the cyclic production. The reservoir temperature decreases significantly, and the overlying and bottom layers can supply some heat for CH₄ hydrate dissociation in addition to the HBL. External heat can be added to improve the hydrate reservoir production during the process.

3. Relatively high initial CH₄ hydrate saturation, high reservoir absolute permeability, and long injection or production time are helpful in improving the CH₄ hydrate production; however, they are adverse to CO₂ storage. Too high of an initial CH₄ hydrate saturation or reservoir absolute permeability will cause a decrease in the CH₄ production due to poor seepage conditions. Relatively speaking, the injected gas temperature and injected gas composition have limited effect on the CH₄ hydrate production or CO₂ storage.

5 Comments

The findings in this paper are based on the model established from publicly available data; however, some parameter values may not be quite accurate, especially in the porous media environment, which may have a significant influence on the results. For example, the hydrate formation kinetics are not easy to determine, and do not follow an Arrhenius temperature dependence. Moreover, the dissolution of CO₂ into the aqueous phase probably has some influence on the simulation results. However, this study can provide some insights into the cyclic CH₄ hydrate production process.

Acknowledgments

This work was supported by the School-Level Scientific Research of Shengli College, China University of Petroleum (Grant No. KY2018018), the Important National Science and Technology Specific Projects of China (Grant No. 2017ZX05013-001), and the Youth Innovation Team Science and Technology Development Program of Shandong Province Higher Educational Institutions (Grant No. 2019KJA024).

References

All Tables

All Figures