Gas-Liquid-Solid Migration Characteristics of Gas Hydrate Sediments in Depressurization Combined with Thermal Stimulation Dissociation.

The exploitation of natural gas hydrate is always hindered by the migration of water and sands due to gas production. Depressurization combined with thermal stimulation is an effective method for hydrate dissociation. This paper reported the influence of gas-liquid-solid migration on morphological change of hydrate sediments in natural gas production using visualization method. Different backpressures combined with thermal stimulation methods were applied to simulate natural gas hydrate exploitation. Pressure compensation was first employed to study sediment recovery features. The expansion rate of a porous medium layer under combined dissociation and different backpressure (4.5, 3.5, 2.5, 1.5, and 0.1 MPa) was discussed. A 176% hydrate sediment expansion rate was found after the combined dissociation at 0.1 MPa. In addition, it was observed that the height of the water layer above the porous media after pressure compensation was gradually reduced with a decrease in backpressure and eventually disappeared at 0.1 MPa. It was also found that the disappearing water layer caused an anomalous memory effect phenomenon. Expansion and subsidence of sediments provide a better reference for hydrate exploitation and geological safety.

Introduction

Natural gas hydrate (NGH) is an icelike crystal that hosts water molecules to form a lattice via hydrogen bonding, with gas molecules filling the lattice.[1] The formation of NGH requires a certain temperature and pressure, so hydrate resources are mainly stored in permafrost and submarine sediments.[2−11] Worldwide, the available volume of NGH is estimated at 21 × 1015 m3 and is proposed to be the largest source of hydrocarbons on earth.[1] It is assessed that the total carbon content in NGH is approximately more than double that in other fossil fuels.[12] Therefore, the importance of NGH exploitation increasingly emerges.

Hydrate, a special phase-change material driven by both temperature and pressure,[13] has a wide application (such as carbon dioxide capture and separation, separation of near-boiling-point hydrates, hydrogen/gas storage, and seawater desalination).[14,15] Therefore, the characteristics of hydrate dissociation need to be emphatically studied in application and exploitation. Depressurization has been recognized as the most feasible, economical, simple, and technically effective method for utilizing NGH resources,[16−26] and it has been applied in the Nankai Trough (Japan), Mallik (Canada), and Shenhu Area (China).[26−28] However, ice formation and reformation of hydrate can be caused by the lack of energy supply in the depressurizing dissociation.[29] Hence, depressurization combined with thermal stimulation has been regarded as the most promising gas production technology.[18,30]

Gas–liquid–solid migration is one of the most important factors influencing hydrate dissociation.[31] The dissociation of hydrates produces free water and gas, and the flow of free water and gas results in irregular hydrate sediment structure (including hydrate, free water, gas, and porous media). The dissociated gaseous methane significantly increases the pressure inside the pore space, and the methane also increases the interactive force between the fluid and hydrate-bearing porous media; these phenomena may result in the reformation of hydrate[32−34] and the deformation of hydrate-bearing sediments.[35,36] Hydrate reformation could be caused by free-flowing gas in sediments under local temperature. Hydrate reformation can reduce the permeability of sediments and result in the clogging of the penetration path;[37−39] the deformation of hydrate-bearing sediments counteracts gas–liquid–solid migration. In addition, pore-scale phenomena, such as capillary phenomena and surface adsorption effects, influence the mechanisms and efficiency of gas recovery, especially in gas–liquid flow.[40] The flow generated via the decomposition causes the migration of small sediment particles.[41] However, it remains unclear why and how gas–liquid–solid migrate in sediments.

There have been some studies on hydrate mining by means of combined dissociation, but the visualization method failed to present in gas–liquid–solid migration. Previous research on natural gas exploitation has always focused on the gas production efficiency of hydrate without analyzing the morphological changes of sediments after exploitation. We analyze the influence of combined dissociation on from the perspective of gas–liquid–solid migration and discuss the formation recovery status for the first time by means of pressure compensation. This is a meaningful experimental study on the influence of hydrate dissociation on strata.

In this paper, to elucidate the influence of gas–liquid–solid migration on hydrate dissociation, NGH exploitation was simulated by varying the backpressure combined with thermal stimulation, and the recovery features of sediments were studied using a pressure compensation method. Gas–liquid–solid migration mechanisms were proposed based on the observed change in sediment morphology. The expansion rate of sediment under various backpressures was compared and discussed. The effects of gas–liquid–solid migration on pore pressure and pore evolution were also analyzed.

Materials and Methods

Figure shows the experimental system. The formation and dissociation of hydrate sediments were carried out in a high-pressure stainless steel cylindrical vessel (two 70 mm diameter sapphire windows; effective volume of 1160 mL; withstand pressure of 20 MPa). A thermostatic bath (XT5718RC-E800L, Xutemp, Hangzhou, Co., Ltd. with an accuracy of ±0.1 K and temperature varying from −15 to 50 °C) was used to regulate the temperature of the vessel. Two resistance thermometers (Pt-1000 with an accuracy of ±0.2%) were inserted into the vessel. The pressure in the cylindrical vessel was probed using a Unik 5000 pressure transducer that has a pressure limit of 25 MPa and a precision of 0.25% FS.

Figure 1

Schematic diagram of the experimental apparatus.

The hydrate formation process was performed as follows. First, 550 g of glass beads with an effective volume of 180 mL (BZ-02, porosity of 37.2% to simulate the porous media layer) were used to fill the vessel, and then 200 g of deionized water (200 ppm sodium lauryl sulfate solution) was injected into the vessel to mix the porous media. The overhead volume in the vessel is 780 mL. Then, the vessel was purged three times with methane gas and evacuated to eliminate the influence of air in the hydrate sediment samples. Next, the bath was set and kept at 15 °C, and methane was slowly injected into the vessel until 7.7 MPa. When the pressure and temperature became almost constant, the bath was set at 2 °C to fulfill the formation of hydrate sediment samples. After the pressure and temperature had been maintained constant for approximately 10 h (pressure decrease <0.01 MPa h–1), the hydrate sediment sample formation was considered to be finished. The synthesized amount of methane hydrate was about 1.5 mol by computing, and the calculation method is reported in ref (24).

The experimental process consisted of two main stages: (1) methane hydrate (MH) dissociation by changing the backpressure combined with thermal stimulation and (2) pressure compensation (after the completion of hydrate dissociation, the gas was re-injected into the vessel). Hydrate sediment sample preparation was required prior to hydrate decomposition (the preparation process was given in advance). During the hydrate dissociation period, the backpressure regulator was opened and the temperature of the thermostatic bath was set at 15 °C at same time. The temperature and pressure profiles and sediment morphologies were recorded in the dissociation. The dissociation was considered to be finished when no gas discharged from the backpressure equipment, and the temperature and pressure in the vessel had stabilized (pressure decrease <0.01 MPa h–1). After dissociation, the backpressure regulator was shut off. Then, the pressure compensation was carried out and kept the vessel pressure at 7.7 MPa.

The formation and dissociation of hydrate sediment samples and the pressure compensation were tested five times. The vessel backpressure in the dissociation was 4.5, 3.5, 2.5, 1.5, and 0.1 MPa. The sediment morphology was recorded by a digital camera (EOS 6D, Canon Company, lens model EF24-105 mm f/4L IS USM), and the temperature and pressure were measured via data acquisition equipment.

Results and Discussion

Hydrate Sediment Dissociation in Different Backpressures Combined with Thermal Stimulation

A summary of the experimental conditions and findings during the combined dissociation is shown in Figure . The characteristics of hydrate dissociation were studied by controlling the backpressure at 4.5, 3.5, 2.5, 1.5, and 0.1 MPa. The pressure in the vessel was approximately 3.5–4 MPa before the hydrate dissociation in Figure b. The hydrate dissociation was mild when the backpressure was 4.5 MPa. Hydrate was initially dissociated by heating before the pressure reached the backpressure. When the pressure reached 4.5 MPa, the dissociated gas was released through the backpressure pipeline. At the 3.5 MPa backpressure, the difference between the pressure in the vessel and the backpressure was small, so the driving force was small when the gas was released, and the temperature did not drop sharply. Moreover, as shown in Figure a, there is a temperature buffering region under the combined dissociation at 3.5 and 4.5 MPa. In this region, the temperature and pressure were below those of the methane hydrate phase equilibrium and hydrate dissociated. The heat absorbed by hydrate dissociation was almost the same as that from the external water bath into the vessel, resulting in the temperature curve remaining horizontal. The hydrate was dissociated dramatically when the backpressures were 2.5 and 1.5 MPa. The intensity of hydrate dissociation mainly depended on the pressure driving force (ΔP), which was 2 and 3 MPa in the combined dissociation at 1.5 and 2.5 MPa, respectively. The excess free gas was quickly released and the pressure sharply decreased, which caused an instant drop in temperature. When the temperature went down to a point, the temperature curve was in temperature buffering. The point was defined as the freezing point (ice was formed), which provided the heat for gas release and hydrate dissociation in the temperature buffering region. The same mechanism of temperature buffering was proposed by Chong et al.[42] When the hydrate dissociation was at the backpressure of 0.1 MPa, the faster gas release caused by a larger pressure driving force resulted in the temperature decreasing to −7 °C. The temperature then rose as the hydrate decomposed, and a temperature buffer occurred when the temperature reached the freezing point. During the combined dissociation, the backpressure was always lower than the vessel pressure. Therefore, the combined dissociation was mainly controlled by the difference between the pressure in the vessel and the backpressure. A larger pressure difference resulted in a faster gas release. The pressure difference was the largest within 0–10 min of the dissociation. The speed of the gas release would affect the temperature in the vessel and determine whether there was ice formation. The pressure difference determined the rate at which the gas was released, thus affecting the temperature increase process.

Figure 2

(a) Temperature changes of the combined dissociation at different backpressures. (b) Pressure changes of the combined dissociation.

Volume Expansion of the Porous Media Layer in the Combined Dissociation

Figure shows the sediment morphologies in the first hydrate formation. It was found that the hydrate first formed in the water phase above the porous medium layer and then spread over the vessel window. As the formation continued, hydrates also formed in the porous media layer. The state of complete hydrate formation can be reached at 300 min. As shown in Figure , the combined dissociation at 3.5 MPa was taken as an example. Some of the gas generated by the hydrate dissociation in the porous media would be stored in the porous media in the form of pores. These pores increased the height of sediment. The formation morphologies of hydrate sediments before the combined dissociation are given in the Supporting Information (Figure S1).

Figure 3

Sequential images of the first hydrate formation process.

Figure 4

Sequential images of the combined dissociation at 3.5 MPa.

Moreover, the sediment morphology significantly changed after the combined dissociation at various backpressures. Figure (I)b–f shows the final sediment morphology after the combined dissociation. At the backpressure of 4.5, 3.5, 2.5, and 1.5 MPa, the height of the porous media layer increased from 30 to 37 mm and the expansion rate reached about 126%. The expansion rate can be calculated as He/Hi × 100% (Hi is the initial height of the porous media layer and He is the final height of the porous media layer after the combined dissociation at various backpressures). Note that the height of the porous media rose from 30 to 53 mm and that the expansion rate reached 176% after the combined dissociation at 0.1 MPa. The increase in the height of the porous media layer is defined as the volume expansion (VE). Top hydrate should not affect the expansion of sediment. Meanwhile, compared with the initial state in Figure (I)a, a number of pores in the porous media appeared after the dissociation. As shown in Figure (I)f, the number of pores after the combined dissociation at 0.1 MPa is more than others. The experimental results indicated that pore evolution is a key factor for VE.

Figure 5

(I) (a) Initial sediment morphology. (b–f) Final sediments morphology after the combined dissociation at various backpressures. (b: 4.5 MPa, c: 3.5 MPa, d: 2.5 MPa, e: 1.5 MPa, and f: 0.1 MPa). (g) Pore channel in the combined dissociation at 0.1 MPa. (II) (a–c) Schematic diagram of the pore channels evolution under the combined dissociation at 0.1 MPa. (d–g) Schematic diagram of the pores evolution.

According to observations, the hydrate dissociation was divided into violent eruptive and quiet vadose stages. In the violent eruptive stage, many pores emerged inside the sediments due to the randomness of hydrate decomposition sites during early dissociation. These pores were occupied by free gas and water. The increment of hydrate decomposition rate increased the pore pressure. When the pore pressure was eventually higher than the sediment effective stress, a portion of the pores evolved to channels in Figure (II)a,b. A pore channel was observed in the experiment in Figure g and the Supporting Information (Movie S1). Unevolved pores gradually enlarged with the increase in pore pressure in Figure (II)d–g. Using the modified Peng–Robinson equation, the theoretical pore pressure could exceed 2000 MPa.[43] The pore pressure created a sufficiently large force against the boundary particles and pushed the porous media layer to “grow” gradually.

The evolution of pores was accompanied by the migration of the particles. Particle migration rests with geometrical constraint, i.e., the relative size of the migratory particles with respect to the pore throat size in the host sediment skeleton. If the size of sediment particles is less than the diameter of the pore channels, then the particles would migrate theoretically by the pressure driving force. With the increase in the hydrate decomposition rate, continuous production of water and gas caused the local pressure to increase, augmenting the pore channels. The augmented pore channels provide the macroconditions for particle migration, and the pressure difference caused by the local high pressure supplied the driving force, resulting in sediment particle migration. Meanwhile, the free gas and water in the pore channels migrated along the direction of the pressure difference.

The migration velocity increase caused by the fast decomposition rate changed the capillary force and resulted in fluid retention,[12] which, in turn, affected the gas–liquid–solid migration in the porous media and led to the partial blockage of the pore channels. In other words, pore channels were converted to pores.

As the rate of hydrate decomposition decreased, the hydrate dissociation converted to the quiet vadose stage in the later period, and the local pressure was reduced. The pressure difference caused by the local pressure was unable to overcome the sediment effective stress at this time. The flow of gas was impeded by water and sediment particles in the pore channels. That is, the produced gas and liquid were trapped in the porous media layer. The pore channels became relatively loose and filled with water and residual gas. Finally, as shown in Figure (II)b,c, the pore channels are blocked by the gravity of the sediment particles. Overall, large amounts of pores and blocked pore channels together caused the VE of the porous media layer. Overburden sediment is the ubiquitous environment of natural gas hydrate. Different types of overburden sediments have different stiffness. Pores are generated by hydrate dissociation in overburden sediments with different stiffness, which change the volume of sediments.[43] Pore pressure also increased with increasing stiffness for formations.

Furthermore, when the pore pressure was higher than the backpressure (0.1 MPa), the pores and blocked pore channels were not broken, and the VE was more obvious, consistent with observation. The expansion rate was at a maximum after the combined dissociation at 0.1 MPa. However, a higher backpressure (4.5, 3.5, 2.5, and 1.5 MPa) could break the pores during dissociation, and VE was not obvious, whereas a small number of pores still existed under the higher backpressure due to a higher pore pressure than backpressure, which resulted in lower expansion rates in Figure (I)b–e.

Therefore, pore evolution was divided into three stages: (1) production of pores; (2) conversion of a portion of pores to pore channels; and (3) conversion of some pore channels to pores and clogging of others. The local pressure difference was the difference between the local high pressure area caused by the accumulation of decomposing gas and the relatively low pressure area, which was mainly to provide the driving force for the gas–liquid–solid migration. The pore channel, evolved from the gradually increased pore, was to provide the path for the migration. These pores and pore channels not only change the morphology of the sedimentary layer but also have a significant influence on the sediment heat-transfer characteristics. In addition, a high backpressure could break the pores and the pore channels, resulting in a nonobvious VE.

Sediment Recovery Features after Pressure Compensation

To study sediment recovery features, gas pressure compensation was carried out and the vessel pressure was kept at 7.7 MPa after each combined dissociation.

The sediment morphology after pressure compensation is shown in Figure a–e. The height of the water layer above the porous media (WLPM) after pressure compensation was gradually reduced with the decrease in backpressure. A comparison of the morphology of the sediments in Figures (I)b–f and 6a–e shows that the difference is the number of pores in the porous media. After pressure compensation, the number of pores decreased for different backpressures. Therefore, the reduction of the WLPM was attributed to the pores breaking. Under the combined dissociation at different backpressures, low pressure pores increased with the decrease in backpressure, and the pore pressure was lower than the compensating pressure. After pressure compensation, as the backpressure decreased, more pores were destroyed, more water entered the porous media layer, and the water layer eventually disappeared for a backpressure of 0.1 MPa. Meanwhile, comparison of the morphology of the sediments in Figures (I)f and 6e shows that an obvious subsidence in the sediments was found in the pressure compensation after the combined dissociation at 0.1 MPa, indicating that the pore channels and pores causing the obvious VE had been damaged. Therefore, for 1.5 MPa, the sediment morphology has no obvious change. The backpressure of 1.5 MPa has a larger driving force for hydrate dissociation.[42] The combined dissociation at 1.5 MPa can be used as a reasonable reference for hydrate exploitation and geological safety.

Figure 6

Sediment morphology after the pressure compensation at different backpressures (a: 4.5 MPa, b: 3.5 MPa, c: 2.5 MPa, d: 1.5 MPa, and e: 0.1 MPa).

As shown in Figure , the porous media layer height is basically the same after pressure compensation. The difference was the distribution of free water in the porous media pores. The water completely filled the porous media layer after pressure compensation for the combined dissociation at 0.1 MPa compared with others at 4.5, 3.5, 2.5, and 1.5 MPa. At this time, the distribution of water was such that the water saturation of the lower porous media layer was high and that in the upper layer was low, as shown in Figure e. Therefore, the heat transfer characteristics of the sediments enhanced due to water filling in the disappearing water layer (DWL). The larger compensated pressure destroyed the pores in the sediment, allowing the water to enter into the porous media completely. This process was accompanied by the entry of gas, resulting in the tight gas–water contact.

Overall, pores were damaged during pressure compensation and water entered the porous media layer. The height of the WLPM after pressure compensation was gradually reduced with the decrease in backpressure. In addition, considering the effects of backpressure on the efficiency and geological safety of gas hydrate extraction, the combined dissociation at 1.5 MPa has little effect on the morphology of hydrate sediments and acts as a larger driving force for hydrate dissociation.

Anomalous Memory Effect Phenomenon under the Disappearing Water Layer

Based on the special distribution of free water after pressure compensation for the backpressure of 0.1 MPa, hydrate reformation was performed to further study the effects of a DWL on hydrate reformation. Two hydrate formation experiment groups were tested. One was carried out under the DWL and the other with an absent DWL (the distribution of water was the same as the initial state in Figure a). Each group has four runs of hydrate formation and dissociation, and the experimental results correspond to Figures and 8. Both MH formation and dissociation were visually investigated via closed mode (the temperatures of formation and dissociation were set at 2 and 15 °C, respectively). The closed mode means that the vessel remained closed during MH formation and dissociation without the gas being discharged, and the driving force only depended on temperature.

Figure 7

(a–d) Sediment morphologies before MH formation under the DWL (a: before first formation, b: before 1st re-formation correspond to after the first closed dissociation, c: before 2nd re-formation correspond to after the second closed dissociation, and d: before 3rd re-formation correspond to after the third closed dissociation). (e) Four runs of the hydrate formation.

Figure 8

(a–d) Sediment morphologies before MH formation in the absent DWL (a: before first formation, b: before 1st re-formation correspond to after the first closed dissociation, c: before 2nd re-formation correspond to after the second closed dissociation, and d: before 3rd re-formation correspond to after the third closed dissociation). (e) Four runs of the hydrate formation.

The four hydrate formation runs without a DWL are shown in Figure e. The first formation induction time was 138 min, and the times for three re-formations were 76, 72, and 64.5 min. The induction time computing method can be seen in ref (44). The induction time of first formation was much longer than those of the three re-formations, which is a normal phenomenon of the memory effect. Research found that the induction time of hydrate reformation was much shorter than that of the first formation, although reformed hydrates still melted at the same dissociation temperature. This phenomenon was defined as the memory effect.[45] However, as shown in Figure e, for the four hydrate formation runs under the DWL, the induction time of the first formation was only 39 min, and the times for the three re-formations were 62, 55, and 61 min. The induction time of the first formation was shorter than that of the re-formations, which is an anomalous memory effect phenomenon. This was verified by repeated experiments in the Supporting Information (Figure S2). A comparison of the sediment morphology in Figures and 8 shows that a significant difference was found with the WLPM before the first formation. As shown in Figure a, the WLPM was invisible. The WLPM appeared again after the closed dissociation in Figure b–d. Subsequently, the water layer no longer disappeared. The induction time of the first formation was obviously decreased. Nevertheless, as shown in Figure , the WLPM existed before each hydrate formation, and the memory effect was normal. Therefore, it was concluded that the anomalous memory effect was attributed to the change in WLPM.

The decrease in induction time in the first formation when anomalous memory effect occurred can be attributed to three reasons: (1) heat transfer; (2) microsize effect; and (3) change in the initial MH formation point. The heat transfer characteristics of sediments increased and promoted MH formation because of the water filling porous media. The microsize effect augmented the gas–liquid contact area and accelerated formation. The initial point of the first formation happened inside the porous media layer, which led to a heterogeneous nucleation and enhanced formation.[46] More details of the three reasons will be analyzed in a later article. The memory effect related to the change in the water layer will be the research focus in a subsequent study.

Conclusions

The MH dissociation with various backpressures combined with the thermal stimulation method and sediment recovery features using pressure compensation was investigated. It was found that the expansion rate reached 176% after the combined dissociation at 0.1 MPa. VE was caused by pore evolution, which was divided into three processes: (1) production of pore; (2) conversion of a portion of pores to pore channels; and (3) conversion of some pore channels to pores, while clogging of others. Pore evolution is the effect of gas–liquid–solid migration in the process of hydrate mining, which not only provides a theoretical basis for the process of hydrate mining but also has a significant influence on sediment heat-transfer characteristics.

In addition, the WLPM height after pressure compensation was gradually reduced with the backpressure decrease in the combined dissociation. An obvious subsidence in the sediments was found with pressure compensation after the combined dissociation at 0.1 MPa. Meanwhile, the experimental results prove that the combined dissociation at 1.5 MPa had little effect on the hydrate sediment morphology and a larger driving force for hydrate dissociation. In essence, the variation of VE and WLPM was the embodiment of the interaction among the effective stress of sediments, pore pressure, backpressure, and compensating pressure. The morphology of hydrate sediments after pressure compensation indirectly demonstrates a formation change in the recovery process, which is of great significance to geological safety.

The anomalous memory effect phenomenon was also found in hydrate reformation under the DWL. The anomalous memory effect was attributed to the heat transfer, microsize effect, and initial MH formation point change. This anomaly, which is attributed to the distribution of water, is a re-understanding of the memory effect and will redefine its mechanism. Hence, the analysis of gas–liquid–solid migration characteristics for gas hydrate sediments would be an important reference for hydrate exploitation and understanding the mechanism of hydrate formation.