Experimental Simulation of the Self-Trapping Mechanism for CO 2 Sequestration into Marine Sediments

: CO 2 hydrates are ice-like solid lattice compounds composed of hydrogen-bonded cages of water molecules that encapsulate guest CO 2 molecules. The formation of CO 2 hydrates in unconsolidated sediments signiﬁcantly decreases their permeability and increases their sti ﬀ ness. CO 2 hydrate-bearing sediments can, therefore, act as cap-rocks and prevent CO 2 leakage from a CO 2 -stored layer. In this study, we conducted an experimental simulation of CO 2 geological storage into marine unconsolidated sediments. CO 2 hydrates formed during the CO 2 liquid injection process and prevented any upward ﬂow of CO 2 . Temperature, pressure, P-wave velocity, and electrical resistance were measured during the experiment, and their measurement results veriﬁed the occurrence of the self-trapping e ﬀ ect induced by CO 2 hydrate formation. Several analyses using the experimental results revealed that CO 2 hydrate bearing-sediments have a considerable sealing capacity. Minimum breakthrough pressure and maximum absolute permeability are estimated to be 0.71 MPa and 5.55 × 10 − 4 darcys, respectively.


Introduction
Carbon capture and storage (CCS) technology is essential for rapid CO 2 mitigation. The geological storage of carbon dioxide (CO 2 ) is a highly effective, long-term mitigation solution for the large quantities of CO 2 emissions [1,2]. For these reasons, to date, CO 2 geological sequestration (CGS) technology has been developed by several leading countries. However, most of existing CGS methods worldwide require particular geological structures to work, such as a highly pervious rock formation (e.g., sandstone layer) imbedded in impermeable layers (i.e., cap-rocks). This requirement leads to CGS application difficulties such as a shortage of proper sites, challenges in the long-range transport of CO 2 , deep drilling and injection, and restricted storage capacity, which substantially increases the cost of using CGS. To overcome these limitations, several CGS methods that do not need cap-rock, such as carbonated water injection (CWI), have been suggested [3][4][5][6].
CO 2 can be stored in unconsolidated sediments under CO 2 hydrate-bearing sediments. CO 2 hydrates are ice-like solid lattice compounds composed of hydrogen-bonded water cages that encapsulate guest molecules of CO 2 . CO 2 hydrates are formed in the seabed under low temperatures and high pressures [7,8]. Previous studies on natural gas hydrate-bearing sediments [9] and preliminary studies on CO 2 hydrate-bearing sediments [10,11] have shown that CO 2 migration is significantly hampered by the formation of gas hydrates, resulting in a self-trapping mechanism. Furthermore, the self-preservation response of CO 2 hydrates slows the CO 2 hydrate dissociation process [12], which serves to mend unintended fractures of CO 2 hydrate-bearing sediments, thereby severely diminishing the transport of CO 2 fluids [13,14]. Thus, it has been suggested that CO 2 hydrates can be used as primary or secondary safety factors for CO 2 geological storage in marine unconsolidated sediments [15][16][17]. Furthermore, unconsolidated sand sediments have advantages over consolidated rocks (e.g., sandstones) in that the CO 2 storage capacity of the former is higher than that of the latter due to the high porosity of unconsolidated sandy sediments (40-60%). In addition, the CO 2 injectability of unconsolidated sand sediments is superior because of their high permeability (0.1-10 darcys) resulting from wide and well-connected pore spaces. Tohidi et al. (2010) performed experimental CO 2 leakage simulations through each type of unconsolidated sediment (glass-bead, sand, and sand-clay mixture), and using electrical resistance measurements and a CO 2 concentration analysis they confirmed the existence of the self-trapping effect of CO 2 hydrates [11]. Massah et al. (2018) demonstrated the sequestration of CO 2 through horizontal injection into a laboratory scale reservoir and revealed the large storage density of CO 2 hydrate formations [18]. Gauteplass et al. (2018) described CO 2 hydrate formation caused by liquid CO 2 injection into cold, water-saturated sandstone and reported that hydrate formation in the pore space resulted in blockage of CO 2 flow under most conditions [19]. However, more direct and comprehensive experimental data including temperature-pressure relations, elastic wave velocity, and dissociation tests are required for a better understanding of the behavior of CO 2 and CO 2 hydrate formation in unconsolidated sediments.
The objective of this study was to simulate CO 2 geological storage into marine unconsolidated sediments using CO 2 hydrates as a cap-rock. A large reaction cell was used to experimentally verify the CO 2 self-trapping mechanism in marine sediments and to evaluate the behavior of CO 2 -stored unconsolidated sediment during CO 2 hydrate formation and dissociation.

Soil Used
The strata of unconsolidated marine sediments typically consist of multiple layers of a different sediment types such as sand-rich sediment layers and fine-grained sediment layers. The permeability of fine-grained sediments is very low (i.e., 10 −3 -10 −7 darcys; [20]), therefore, fine-grained sediments can be practically considered as impermeable layers, which obstruct the upward flow of CO 2 . Meanwhile, sand-rich sediments are suitable as CO 2 storage host sediments because of their relatively high permeability (0.1-10 darcys), while they are ready for permeation of CO 2 . The effect of the self-trapping mechanism on a sand-rich layer is, therefore, important to CO 2 geological storage into unconsolidated sediments. In this study, fine sand (Ottawa F110; mean particle size = 120 µm, specific gravity = 2.65, permeability = 5-6 darcys, quartz 99%) was used as the host sediment sample.

Experiment Setup
The experimental design simulated CO 2 injection into a shallow marine sediment (i.e., high water pressure, low temperature) and CO 2 hydrate formation. The experimental design used in this study is shown in Figure 1. A cylindrical and rigid-wall reaction cell was made of an aluminum alloy (duralumin, AA2024). The inner diameter of the cell was 20 cm, the height of the interior was 100 cm, and the internal volume was 31.4 L. The reaction cell was originally developed for an experimental simulation of thermal stimulation on gas hydrate-bearing sediments [21]. Water and liquid CO 2 were injected from the bottom of the reaction cell using a water pump and gas booster. Pressure inside the reaction cell was controlled using a back-pressure regulator at the top of the cell. The quantities of CO 2 gas and water that flowed out of the reaction cell were measured using a water substitution system.
Various types of sensors were installed at predetermined layers (every 10 cm) within the reaction cell as shown in Figure 2. The cell contained five T-type thermocouples for temperature measurements of the cell interior, five pressure transducers for fluid pressure measurements, five pairs of piezoelectric ceramic disks (diameter: 20 mm) for compressional wave (P-wave) measurements at layers A1-A5, and four pairs of electrodes for electrical resistance measurements at layers B1-B4. For P-wave velocity measurements, square-shaped pulses with amplitude of 10 V (peak-to-peak) were used for excitation, and the input frequency ranged from 1 to 10 kHz. The electrodes were connected to an LCR meter in order to measure the electrical resistance (frequency = 50 kHz).
Cool water was circulated through copper tubes that coiled around the reaction cell. The temperature inside the reaction cell was controlled by two water coolers, which had different temperatures (i.e., Water cooler 1 at 3 • C, and Water cooler 2 at 15 • C). Figure 3 shows the temperature gradient of the inside of the reaction cell, which was formed by two separated cooling systems. The CO 2 hydrate stability zone was developed in the middle part of the reaction cell (i.e., height of 0.35-0.75 m in the reaction cell). measurements at layers A1-A5, and four pairs of electrodes for electrical resistance measurements at layers B1-B4. For P-wave velocity measurements, square-shaped pulses with amplitude of 10 V (peak-to-peak) were used for excitation, and the input frequency ranged from 1 to 10 kHz. The electrodes were connected to an LCR meter in order to measure the electrical resistance (frequency = 50 kHz). Cool water was circulated through copper tubes that coiled around the reaction cell. The temperature inside the reaction cell was controlled by two water coolers, which had different temperatures (i.e., Water cooler 1 at 3 °C, and Water cooler 2 at 15 °C). Figure 3 shows the temperature gradient of the inside of the reaction cell, which was formed by two separated cooling systems. The CO2 hydrate stability zone was developed in the middle part of the reaction cell (i.e., height of 0.35-0.75 m in the reaction cell).     measurements at layers A1-A5, and four pairs of electrodes for electrical resistance measurements at layers B1-B4. For P-wave velocity measurements, square-shaped pulses with amplitude of 10 V (peak-to-peak) were used for excitation, and the input frequency ranged from 1 to 10 kHz. The electrodes were connected to an LCR meter in order to measure the electrical resistance (frequency = 50 kHz). Cool water was circulated through copper tubes that coiled around the reaction cell. The temperature inside the reaction cell was controlled by two water coolers, which had different temperatures (i.e., Water cooler 1 at 3 °C, and Water cooler 2 at 15 °C). Figure 3 shows the temperature gradient of the inside of the reaction cell, which was formed by two separated cooling systems. The CO2 hydrate stability zone was developed in the middle part of the reaction cell (i.e., height of 0.35-0.75 m in the reaction cell).      . Temperature distribution of the high-pressure cell obtained from the preliminary test result with distilled water. The maximum CO2 hydrate equilibrium temperature is approximated using the second quadruple point of CO2-water mixture (the intersection of the water-CO2 vapor-CO2 liquid, and water-CO2 hydrate-CO2 vapor equilibrium line) because the water-CO2 hydrate-liquid CO2 equilibrium line is essentially vertical in the pressure-temperature diagram [22].

Experimental Procedure
The experiment involved three procedures: (1) The preparation of a water-saturated sample; (2) the injection of the CO2 liquid; and (3) the depressurization of the cell. When the CO2 liquid was injected from the bottom of the cell, the injected CO2 moved upward in the water-saturated sediment sample due to its buoyancy. Then, CO2 hydrates formed within the CO2 hydrates stable zone, which was located at the middle part of the cell (refer to Figure 3). The depressurization test was performed after CO2 hydrate formation to evaluate the behavior of CO2 during hydrate dissociation. These procedures are detailed within this subsection.    . Temperature distribution of the high-pressure cell obtained from the preliminary test result with distilled water. The maximum CO2 hydrate equilibrium temperature is approximated using the second quadruple point of CO2-water mixture (the intersection of the water-CO2 vapor-CO2 liquid, and water-CO2 hydrate-CO2 vapor equilibrium line) because the water-CO2 hydrate-liquid CO2 equilibrium line is essentially vertical in the pressure-temperature diagram [22].

Experimental Procedure
The experiment involved three procedures: (1) The preparation of a water-saturated sample; (2) the injection of the CO2 liquid; and (3) the depressurization of the cell. When the CO2 liquid was injected from the bottom of the cell, the injected CO2 moved upward in the water-saturated sediment sample due to its buoyancy. Then, CO2 hydrates formed within the CO2 hydrates stable zone, which was located at the middle part of the cell (refer to Figure 3). The depressurization test was performed after CO2 hydrate formation to evaluate the behavior of CO2 during hydrate dissociation. These procedures are detailed within this subsection.   Figure 3. Temperature distribution of the high-pressure cell obtained from the preliminary test result with distilled water. The maximum CO 2 hydrate equilibrium temperature is approximated using the second quadruple point of CO 2 -water mixture (the intersection of the water-CO 2 vapor-CO 2 liquid, and water-CO 2 hydrate-CO 2 vapor equilibrium line) because the water-CO 2 hydrate-liquid CO 2 equilibrium line is essentially vertical in the pressure-temperature diagram [22].

Experimental Procedure
The experiment involved three procedures: (1) The preparation of a water-saturated sample; (2) the injection of the CO 2 liquid; and (3) the depressurization of the cell. When the CO 2 liquid was injected from the bottom of the cell, the injected CO 2 moved upward in the water-saturated sediment sample due to its buoyancy. Then, CO 2 hydrates formed within the CO 2 hydrates stable zone, which was located at the middle part of the cell (refer to Figure 3). The depressurization test was performed after CO 2 hydrate formation to evaluate the behavior of CO 2 during hydrate dissociation. These procedures are detailed within this subsection.

Water-Saturated Sample Preparation
To simulate deep-marine sediments, a water-saturated fine sand (i.e., Ottawa F110 sand) sample was prepared. First, the sample was mixed with distilled water and packed into the cell by hand tamping. The cell was fully filled with moist sand and the porosity of the sample was 0.40. Then, the cell was slowly flushed with distilled water at pressure of~0.5 MPa for several hours to remove remaining air bubbles. During the water flushing, a cooling process was initiated. After the completion of water flushing, the cell was pressurized to 5.5 MPa with distilled water and left for about 16 h to stabilize the temperature throughout the sample.
Our primary goal was to study the self-trapping effect induced by CO 2 hydrates, and because of that, distilled water was used as pore water instead of saline water (it allows easier hydrate formation). However, this decision means that the electrical and geochemical behavior in this experiment was different from the real behavior in the marine sediments.

Injection of CO 2 Liquid
Liquid CO 2 was introduced into the cell from the bottom using the gas booster. The injection pressure was 5.6 MPa, and the backpressure at the top of the cell was 5.0 MPa. Thirty-four hours later, the injection pressure and backpressure increased to 6.2 and 5.6 MPa, respectively. The CO 2 liquid was injected for more than 17 days while the temperature, pressure, P-wave velocity, and electrical resistance were measured. The formation of CO 2 hydrates during the CO 2 injection process was expected (refer to Section 3).

Depressurization
The cell was depressurized stepwise using a back-pressure regulator while the inlet valve was closed. Each depressurization step was 0.5 MPa for more than 16 hours. Temperature, pressure, P-wave velocity, and electrical resistance during the depressurization process were measured in the same manner as during CO 2 injection process.

Liquid CO 2 Injection Process
When the CO 2 liquid was injected from the bottom of the cell, a CO 2 liquid plume moved upward because of a buoyancy force. Eventually, the CO 2 liquid front reached the CO 2 hydrate stability zone, and then CO 2 hydrates formed. The CO 2 hydrate-bearing sediment layer then obstructed the upward flow of the CO 2 liquid. While water and CO 2 were consumed in the CO 2 hydrate formation process in the hydrate stability zone, the CO 2 hydrate-bearing sediment layer prevented CO 2 supply. Thus, a pressure difference between the upper and lower part of the cell appeared. The P-wave velocity and electrical resistance monitoring results indicated the formation of CO 2 hydrates and the blockage of the CO 2 flow. Detailed experimental results are shown in the following sections. Figure 4 shows the pressure of each layer in the cell over time. The CO 2 liquid was injected at 1300 min after data logging started. The pressure in each layer was scattered until 5000 min because of a difference in pressure between the injection pressure and backpressure (i.e.,~0.6 MPa, refer to Section 2.3.2), and volume change of pore fluids due to CO 2 dissolution into pore water. The pressure in each layer was very similar to one another because the pore space was well connected throughout the sample. CO 2 hydrates started to form when the injected CO 2 reached the CO 2 hydrate stability zone (herein, between layers A3 and A4; Figure 3). Then, the pressure in layers A1, A2, and A3 rapidly dropped to 3.3 MPa after 5000 min while the pressure of layers A4 and A5 were nearly constant and identical to the injection pressure. The difference in pressure between the upper and lower part of the cell was induced by the sealing (i.e., pore clogging) effect of the CO 2 hydrate bearing-sediments layer. The sealing capacity of the CO2 hydrate-bearing sediment layer gradually increased during the growth of CO2 hydrates in the pore space of the sample. Meanwhile, water and CO2 were consumed during CO2 hydrate formation. For a constant-volume process, the consumption of CO2 and water during CO2 hydrate formation leads to a pressure decrease because the molar volume of the CO2 hydrates is smaller than the original molar volume of the consumed fluids. For the lower part of the cell (represented by layers A4 and A5) the pressure was preserved because CO2 was supplied continuously from the bottom of the cell during the experiment. For the upper part of the cell (represented by layers A1, A2, and A3), however, the pressure decreased because the CO2 hydratebearing sediment layer prevented the CO2 to be supplied from the lower part of the cell. Figure 5 shows the pressure-temperature evolution during the CO2 injection test. Note that layers A2 and A3 were in the CO2 hydrate stable condition while the others were not. The pressure of the upper part of the cell (i.e., layers A1, A2, and A3) decreased when the sealing capacity of the CO2 hydrate-bearing sediment layer increased to a level that prevented flow. Note that there were no CO2 hydrates in layer A2 because the injected CO2 did not reach it, even though this layer is in the CO2 hydrate stability zone. Meanwhile, the lower part of the cell (i.e., layers A4 and A5) maintained a constant pressure level. Then, the pressure of the upper part of the cell gradually increased. There are two mechanisms for the pressure recovery of the upper part of the cell: (1) The uppermost CO2 hydrates dissociated with the pressure decrease. Therefore, pressure was recovered restrictively via emitted CO2 and water from the CO2 hydrates, and (2) CO2 hydrate saturation is limited by capillary pressure, which is determined by the pore size [23].

Temperature and Pressure
In the central part of the CO2 hydrate-bearing sediment layer, CO2 hydrates grew until the CO2 hydrate saturation reached maximum CO2 hydrate saturation. Thus, the consumption of CO2 and water diminished because any additional formation of CO2 hydrates was restricted. Finally, the pressure of the upper part of the cell increased to 5.2-5.5 MPa. However, the pressure of the upper part of the cell was still lower than that of the lower part of the cell. The repetitive ascending and descending pressure could be due to the continuous repetition of the CO2 formation and dissociation process at the CO2 hydrates front.   The sealing capacity of the CO 2 hydrate-bearing sediment layer gradually increased during the growth of CO 2 hydrates in the pore space of the sample. Meanwhile, water and CO 2 were consumed during CO 2 hydrate formation. For a constant-volume process, the consumption of CO 2 and water during CO 2 hydrate formation leads to a pressure decrease because the molar volume of the CO 2 hydrates is smaller than the original molar volume of the consumed fluids. For the lower part of the cell (represented by layers A4 and A5) the pressure was preserved because CO 2 was supplied continuously from the bottom of the cell during the experiment. For the upper part of the cell (represented by layers A1, A2, and A3), however, the pressure decreased because the CO 2 hydrate-bearing sediment layer prevented the CO 2 to be supplied from the lower part of the cell. Figure 5 shows the pressure-temperature evolution during the CO 2 injection test. Note that layers A2 and A3 were in the CO 2 hydrate stable condition while the others were not. The pressure of the upper part of the cell (i.e., layers A1, A2, and A3) decreased when the sealing capacity of the CO 2 hydrate-bearing sediment layer increased to a level that prevented flow. Note that there were no CO 2 hydrates in layer A2 because the injected CO 2 did not reach it, even though this layer is in the CO 2 hydrate stability zone. Meanwhile, the lower part of the cell (i.e., layers A4 and A5) maintained a constant pressure level. Then, the pressure of the upper part of the cell gradually increased. There are two mechanisms for the pressure recovery of the upper part of the cell: (1) The uppermost CO 2 hydrates dissociated with the pressure decrease. Therefore, pressure was recovered restrictively via emitted CO 2 and water from the CO 2 hydrates, and (2) CO 2 hydrate saturation is limited by capillary pressure, which is determined by the pore size [23]. 3.1.2. P-Wave Velocity Figure 6 shows the results of the P-wave velocity measurements during the CO2 injection process. Before the water injection process, the cell was partially saturated, and the P-wave velocity was about 900 m/s. When the sediment sample was saturated by distilled water, the P-wave velocity of all the layers was about 1600 m/s. Then, when the CO2 liquid was injected, the P-wave velocities of the lower part of the cell (i.e., layers A4 and A5) decreased because the bulk modulus of the CO2 liquid was much lower than that of the water [24,25]. Meanwhile, the P-wave velocity of layer A3   In the central part of the CO 2 hydrate-bearing sediment layer, CO 2 hydrates grew until the CO 2 hydrate saturation reached maximum CO 2 hydrate saturation. Thus, the consumption of CO 2 and water diminished because any additional formation of CO 2 hydrates was restricted. Finally, the pressure of the upper part of the cell increased to 5.2-5.5 MPa. However, the pressure of the upper part of the cell was still lower than that of the lower part of the cell. The repetitive ascending and descending pressure could be due to the continuous repetition of the CO 2 formation and dissociation process at the CO 2 hydrates front.
3.1.2. P-Wave Velocity Figure 6 shows the results of the P-wave velocity measurements during the CO 2 injection process. Before the water injection process, the cell was partially saturated, and the P-wave velocity was about 900 m/s. When the sediment sample was saturated by distilled water, the P-wave velocity of all the layers was about 1600 m/s. Then, when the CO 2 liquid was injected, the P-wave velocities of the lower part of the cell (i.e., layers A4 and A5) decreased because the bulk modulus of the CO 2 liquid was much lower than that of the water [24,25]. Meanwhile, the P-wave velocity of layer A3 gradually increased because of the stiffening effect induced by CO 2 hydrate formation. This P-wave velocity increase in layer A3 indicates that the CO 2 hydrate bearing-sediment layer is between layers A3 and A4. On the other hand, the P-wave velocity of layer A1 and A2 did not change during CO 2 injection. This is evidence that the CO 2 hydrate-bearing sediment layer prevents any upward flow of the CO 2 liquid.  Figure 7 shows the normalized electrical resistance (R/R0) during the CO2 injection process, where R is the measured electrical resistance and R0 is the initial electrical resistance of the distilled water-saturated sample at each layer (i.e., layers B1-B4). For the lower part of the cell (i.e., layers B3 and B4), the electrical resistance decreased with CO2 injection because of the dissolution of CO2.
Electrical resistance increased in-situ in the marine sediments because the conductive pore water    Figure 7 shows the normalized electrical resistance (R/R 0 ) during the CO 2 injection process, where R is the measured electrical resistance and R 0 is the initial electrical resistance of the distilled water-saturated sample at each layer (i.e., layers B1-B4). For the lower part of the cell (i.e., layers B3 and B4), the electrical resistance decreased with CO 2 injection because of the dissolution of CO 2 .
Meanwhile, the electrical resistance of the upper part of the cell (i.e., layers B1 and B2) showed minor changes during CO2 injection. This is additional evidence demonstrating that CO2 liquid did not reach the upper part of the cell. Based on the change of the pressure, P-wave velocity and electrical resistance, we can presume that the CO2 hydrate formation front is located between layers A3 and B3. Meanwhile, CO2 hydrates formation was not observed in the electrical resistance data. The electrical resistance of in-situ water-saturated sediments increased when CO2 hydrates formed because the electrical resistance of CO2 hydrates is higher than pore water [29,30]. In this study, however, the change in electrical resistance was insignificant in spite of the presence of CO2 hydrates, because distilled water was used as pore water in this experiment. This is the one of the limitations of this experiment.  Figure 8 shows the pressure of each layer in the cell over time. The pressure of the cell was reduced step-wise using a back-pressure regulator. Pressure differences between the upper and lower part of the cell remained, even though CO2 was vaporized. This pressure discrepancy indicated that the sealing capacity of the CO2 hydrate-bearing sediment layer was preserved. When the CO2 hydrate dissociated completely, the pressure of each layer became equal. Electrical resistance increased in-situ in the marine sediments because the conductive pore water (i.e., brine) was replaced by CO 2 , which is a nonpolar molecule. This electrical resistance increase induced by the CO 2 replacement was weakened by the dissolution of CO 2 and the surface effect of the mineral grains [26]. For typical brine, the effect of CO 2 dissolution on electrical resistance is negligible because the concentration of salt (i.e., NaCl) is much larger than the ionic concentration increased by CO 2 dissolution [27]. However, if the salt concentration is low (e.g., onshore sediments), electrical resistance of in-situ sediments can decrease during the CO 2 permeation [28]. In this experiment, the effect of dissolved CO 2 was dominant on electrical resistance because the pore water was distilled.

Temperature and Pressure
Meanwhile, the electrical resistance of the upper part of the cell (i.e., layers B1 and B2) showed minor changes during CO 2 injection. This is additional evidence demonstrating that CO 2 liquid did not reach the upper part of the cell. Based on the change of the pressure, P-wave velocity and electrical resistance, we can presume that the CO 2 hydrate formation front is located between layers A3 and B3. Meanwhile, CO 2 hydrates formation was not observed in the electrical resistance data. The electrical resistance of in-situ water-saturated sediments increased when CO 2 hydrates formed Minerals 2019, 9, 579 9 of 18 because the electrical resistance of CO 2 hydrates is higher than pore water [29,30]. In this study, however, the change in electrical resistance was insignificant in spite of the presence of CO 2 hydrates, because distilled water was used as pore water in this experiment. This is the one of the limitations of this experiment. Figure 8 shows the pressure of each layer in the cell over time. The pressure of the cell was reduced step-wise using a back-pressure regulator. Pressure differences between the upper and lower part of the cell remained, even though CO 2 was vaporized. This pressure discrepancy indicated that the sealing capacity of the CO 2 hydrate-bearing sediment layer was preserved. When the CO 2 hydrate dissociated completely, the pressure of each layer became equal.  Figure 9 shows the pressure-temperature relationship during the depressurization test. The pressure of the cell dropped with the pressure release using the back-pressure regulator. When the pressure of layers A4 and A5 reached the CO2 vapor pressure, their pressure and temperature relation moved along the CO2 vapor pressure (Figure 9a). The path of the pressure and temperature relationship of layers A4 and A5 was similar to that of the isometric process because the flow of fluids was obstructed by the remaining CO2 hydrate-bearing sediment layer (Figure 9b). Then, the pressure and temperature relationship of layer A2 moved along the CO2 hydrate equilibrium line (Figure 9c). This is evidence that the CO2 hydrates re-formed and dissociated in layer A2. In the previous CO2 injection process, CO2 liquid did not reach layer A2 because of the sealing effect of the CO2 hydratebearing sediment layer, which was located between layers A3 and B3. However, CO2 was supplied to layer A2 when the existing CO2 hydrates partially dissociated by depressurization. CO2 hydrates then re-formed because layer A3 is in the CO2 hydrate stability zone. Then, CO2 hydrates in layer A3 dissociated by additional depressurization. During the dissociation of CO2 hydrates in layer A3, a self-preservation effect was observed in the pressure and temperature relationship as seen in previous experimental studies on CO2 hydrate dissociation [12,31]. Finally, CO2 hydrates completely dissociated with step-wise depressurization (Figure 9d).   Figure 8. Pressure of the cell with lapsed time during depressurization. Figure 9 shows the pressure-temperature relationship during the depressurization test. The pressure of the cell dropped with the pressure release using the back-pressure regulator. When the pressure of layers A4 and A5 reached the CO 2 vapor pressure, their pressure and temperature relation moved along the CO 2 vapor pressure (Figure 9a). The path of the pressure and temperature relationship of layers A4 and A5 was similar to that of the isometric process because the flow of fluids was obstructed by the remaining CO 2 hydrate-bearing sediment layer (Figure 9b). Then, the pressure and temperature relationship of layer A2 moved along the CO 2 hydrate equilibrium line (Figure 9c). This is evidence that the CO 2 hydrates re-formed and dissociated in layer A2. In the previous CO 2 injection process, CO 2 liquid did not reach layer A2 because of the sealing effect of the CO 2 hydrate-bearing sediment layer, which was located between layers A3 and B3. However, CO 2 was supplied to layer A2 when the existing CO 2 hydrates partially dissociated by depressurization. CO 2 hydrates then re-formed because layer A3 is in the CO 2 hydrate stability zone. Then, CO 2 hydrates in layer A3 dissociated by additional depressurization. During the dissociation of CO 2 hydrates in layer A3, a self-preservation effect was observed in the pressure and temperature relationship as seen in previous experimental studies on CO 2 hydrate dissociation [12,31]. Finally, CO 2 hydrates completely dissociated with step-wise depressurization (Figure 9d).  Figure 10 shows the results of the P-wave velocity measurements taken during the depressurization process. The sealing effect of the original CO2 hydrate-bearing sediment layer reduced because some portion of the original CO2 hydrates dissociated. Thus, the P-wave velocity of layer A1 decreased because CO2 intruded the upper part of the cell. On the other hand, the P-wave velocity of layers A2 and A3 suddenly increased because CO2 hydrates formed using the CO2 supply from the lower part of the cell. Note that layers A2 and A3 were in the CO2 hydrate stability zone until 3300 minutes (refer to Figure 9). Meanwhile, the P-wave velocity of layers A4 and A5 decreased because CO2 vaporized during depressurization. When the pressure was lower than the equilibrium pressure of the CO2 hydrates (i.e., 4000-6000 min, refer to Figure 9d), the P-wave velocity of layers A2 and A3 suddenly decreased because reformed CO2 hydrates in these layers dissociated.  (d)  Figure 10 shows the results of the P-wave velocity measurements taken during the depressurization process. The sealing effect of the original CO 2 hydrate-bearing sediment layer reduced because some portion of the original CO 2 hydrates dissociated. Thus, the P-wave velocity of layer A1 decreased because CO 2 intruded the upper part of the cell. On the other hand, the P-wave velocity of layers A2 and A3 suddenly increased because CO 2 hydrates formed using the CO 2 supply from the lower part of the cell. Note that layers A2 and A3 were in the CO 2 hydrate stability zone until 3300 minutes (refer to Figure 9). Meanwhile, the P-wave velocity of layers A4 and A5 decreased because CO 2 vaporized during depressurization. When the pressure was lower than the equilibrium pressure of the CO 2 hydrates (i.e., 4000-6000 min, refer to Figure 9d), the P-wave velocity of layers A2 and A3 suddenly decreased because reformed CO 2 hydrates in these layers dissociated.  Figure 11 shows the normalized electrical resistance (R/R0) during the depressurization process. The results of the electrical resistance measurements indicated that CO2 intruded the upper part of the cell. For layers B1 and B2, electrical resistance exhibited complex behavior due to the formation of CO2 hydrates and the movement of CO2 gas bubbles. However, the electrical resistance decreased generally because of CO2 dissolution. As stated before, electrical resistance of distilled watersaturated sediment decreased with CO2 intrusion because of dissolved CO2. Meanwhile, the electrical resistance of the lower part of the cell (i.e., layers B3 and B4) was barely affected by depressurization because the pore water of the lower part of the cell was already saturated by dissolved CO2.  Figure 11 shows the normalized electrical resistance (R/R 0 ) during the depressurization process. The results of the electrical resistance measurements indicated that CO 2 intruded the upper part of the cell. For layers B1 and B2, electrical resistance exhibited complex behavior due to the formation of CO 2 hydrates and the movement of CO 2 gas bubbles. However, the electrical resistance decreased generally because of CO 2 dissolution. As stated before, electrical resistance of distilled water-saturated sediment decreased with CO 2 intrusion because of dissolved CO 2 . Meanwhile, the electrical resistance of the lower part of the cell (i.e., layers B3 and B4) was barely affected by depressurization because the pore water of the lower part of the cell was already saturated by dissolved CO 2 . Minerals 2019, 9, x FOR PEER REVIEW 13 of 19

Discussion: Simple Analysis on the Sealing Capacity of CO2 Hydrate-Bearing Sediments
There are two major sealing mechanisms for CO2 structural trapping. One is the capillary seal, which occurs by capillary pressure between CO2 and water in pores. The other sealing mechanism is the permeability seal, which is related to the laminar flow velocity of CO2 in pores due to a pressure gradient. In view of the two sealing mechanisms, a simple analysis on the sealing capacity of the CO2 hydrate-bearing sediment was performed using the experimental results.

Capillary Sealing Capacity
Capillary pressure is the difference in pressure across the interface between two fluids. In petroleum reservoirs, capillary pressure between oil and water in rock pores is responsible for trapping oil [32,33]. In the same manner, capillary pressure between water and CO2 can trap CO2. For a given pore structure, the CO2 breakthrough pressure (PC * ) induced by capillarity can be described using the Young-Laplace equation: * = * , (1) where γ is interfacial tension between water and CO2, θ is wetting angle, and d * is the critical pore throat diameter. Several researchers have measured various temperatures and pressures for the interfacial tension between water and CO2 [34][35][36].
In this study, the CO2 hydrate-bearing sediment layer could maintain about 0.71 MPa of pressure difference between the upper and lower part of the cell (Figure 12). Thus, the minimum breakthrough

Discussion: Simple Analysis on the Sealing Capacity of CO 2 Hydrate-Bearing Sediments
There are two major sealing mechanisms for CO 2 structural trapping. One is the capillary seal, which occurs by capillary pressure between CO 2 and water in pores. The other sealing mechanism is the permeability seal, which is related to the laminar flow velocity of CO 2 in pores due to a pressure gradient. In view of the two sealing mechanisms, a simple analysis on the sealing capacity of the CO 2 hydrate-bearing sediment was performed using the experimental results.

Capillary Sealing Capacity
Capillary pressure is the difference in pressure across the interface between two fluids. In petroleum reservoirs, capillary pressure between oil and water in rock pores is responsible for trapping oil [32,33]. In the same manner, capillary pressure between water and CO 2 can trap CO 2 . For a given pore structure, the CO 2 breakthrough pressure (P C * ) induced by capillarity can be described using the Young-Laplace equation: where γ is interfacial tension between water and CO 2 , θ is wetting angle, and d * is the critical pore throat diameter. Several researchers have measured various temperatures and pressures for the interfacial tension between water and CO 2 [34][35][36].
In this study, the CO 2 hydrate-bearing sediment layer could maintain about 0.71 MPa of pressure difference between the upper and lower part of the cell ( Figure 12). Thus, the minimum breakthrough pressure (P C * min ) can be assumed as the average pressure difference between the upper and lower parts of the cell (∆P average ), which is described in Figure 12 (i.e., P C * ≥ P C * min = ∆P average ). Then, the maximum critical pore throat diameter of CO 2 -hydrate bearing sediments (d* max ) was calculated as 132 nm using Equation (1). The values used in this calculation are summarized in Table 1. pressure (PC * min) can be assumed as the average pressure difference between the upper and lower parts of the cell (ΔPaverage), which is described in Figure 12 (i.e., PC * ≥ PC * min = ΔPaverage). Then, the maximum critical pore throat diameter of CO2-hydrate bearing sediments (d*max) was calculated as 132 nm using Equation (1). The values used in this calculation are summarized in Table 1.  If the pressure difference between the fluid interface exceeds P * C, then CO2 breaks through the interface, and laminar flow occurs [35,37,38]. Thus, in order for the capillary sealing mechanism to work, the breakthrough pressure (PC * ) must be larger than the buoyancy pressure of the CO2 plume. The buoyancy pressure (PB) that is induced by the density difference between water and CO2 can be described as where g is the acceleration of gravity, h is the thickness of the CO2-stored layer, and ρwater and ρCO2 are the density of water and CO2, respectively. In a similar manner to the calculation of d*max, the minimum buoyancy pressure (PBmin) which could be maintained by the CO2 hydrate-bearing sediment can be assumed as ΔPaverage (i.e., PB ≥ PBmin = ΔPaverage). At a similar thermodynamic condition of the experiment in this study (i.e., pressure of 6 MPa and temperature of 15 °C), ρCO2 was 784 kg/m 3 [39]. Then, the minimum thickness of the CO2-stored layer (hmin) was calculated as 335 m according to Equation (2). Thus, we can presume that the capillary trapping capacity of the CO2 hydrate-bearing sediment is high enough. The wettability (i.e., wetting angle) could be altered by the increase of the gas hydrate saturation because the solid materials contacting pore fluids are changed from sand particles to CO2 hydrates. For simplicity, this wettability alteration was not considered in this study. Further studies are required to evaluate the effect of wettability iteration on CO2 capillary sealing capacity.   [35] If the pressure difference between the fluid interface exceeds P C * , then CO 2 breaks through the interface, and laminar flow occurs [35,37,38]. Thus, in order for the capillary sealing mechanism to work, the breakthrough pressure (P C * ) must be larger than the buoyancy pressure of the CO 2 plume.
The buoyancy pressure (P B ) that is induced by the density difference between water and CO 2 can be described as where g is the acceleration of gravity, h is the thickness of the CO 2 -stored layer, and ρ water and ρ CO2 are the density of water and CO 2 , respectively. In a similar manner to the calculation of d* max , the minimum buoyancy pressure (P Bmin ) which could be maintained by the CO 2 hydrate-bearing sediment can be assumed as ∆P average (i.e., P B ≥ P Bmin = ∆P average ). At a similar thermodynamic condition of the experiment in this study (i.e., pressure of 6 MPa and temperature of 15 • C), ρ CO2 was 784 kg/m 3 [39]. Then, the minimum thickness of the CO 2 -stored layer (h min ) was calculated as 335 m according to Equation (2). Thus, we can presume that the capillary trapping capacity of the CO 2 hydrate-bearing sediment is high enough.
The wettability (i.e., wetting angle) could be altered by the increase of the gas hydrate saturation because the solid materials contacting pore fluids are changed from sand particles to CO 2 hydrates. For simplicity, this wettability alteration was not considered in this study. Further studies are required to evaluate the effect of wettability iteration on CO 2 capillary sealing capacity.

Permeability Sealing Capacity
When the buoyancy pressure (P B ) is higher than breakthrough pressure (P C * ), CO 2 flow occurs.
Fluid flow through soils finer than coarse gravel is laminar [40]. For laminar flow in CO 2 -saturated sediments the flow velocity, v, can be expressed by Darcy's law as follows: where K is absolute or intrinsic permeability of the sediments, ρ CO2 is the density of CO 2 fluids, g is the gravity constant, µ CO2 is the viscosity of CO 2 fluids, and i is the hydraulic gradient which is expressed by the difference between two hydraulic heads over the flow length. Note that the hydraulic gradient (i) is 1 for a vertical flow. Meanwhile, the average flow velocity for flow through a round capillary tube (v 0 ) can be described by Poiseuille's law as follows: where d is the diameter of the capillary tube. The flow velocity determined by Poiseuille's law (v 0 ) is the upper limit of the flow velocity (i.e., v ≤ v 0 ) in porous media because flow velocity in sediments decreases by the tortuosity of the flow channel. Therefore, the upper limit of the absolute permeability of sediments can be defined using Equations (3) and (4) as The maximum absolute permeability of CO 2 hydrate-bearing sediment (K max ) can, therefore, be calculated using the d * max , which was obtained before. K max is about 5.55 × 10 −4 darcy. This value is similar to the permeability of fine-grained sediments (i.e., 10 −3 -10 −7 darcys [20]), and can be considered as "very low" permeability [41].

Comparison with Other Materials
Estimated maximum absolute permeability (K max ) and minimum breakthrough pressure (P C * min ) are compared with measured absolute permeability (K) and breakthrough pressure (P C * ) of various sediment samples, as shown in Figure 13. We presumed that the breakthrough pressure of F110 sand increases by more than 10 2 times with CO 2 hydrate formation. The minimum breakthrough pressure (P C * min ) of the CO 2 hydrate-bearing sediments estimated in this study is comparable with that of unconsolidated clays and the shale sample. Meanwhile, the actual P C * of CO 2 hydrate-bearing sediments in this experimental simulation may be higher than the estimated P C * min because the latter was estimated conservatively using the pressure difference between upper and lower parts of the cell, instead of being measured directly. In the same manner, actual K of CO 2 hydrate-bearing sediments in this experimental simulation may be lower than estimated K max . This might be attributed to the K max being calculated conservatively using the assumption of fluid flow in a round capillary tube without any tortuosity. To determine the range of absolute permeability and breakthrough pressure of CO 2 hydrate-bearing sediments, further experimental studies are required.

Conclusions
We performed an experimental simulation of CO2 geological storage in marine unconsolidated sediments in this study. CO2 hydrates were formed during the CO2 liquid injection process, and we observed the self-trapping effect of CO2 hydrates. In addition, simple analyses were conducted using the experimental results. The feasibility of CO2 geological storage in marine unconsolidated sediments was experimentally verified using 1-m-height high-pressure cell. CO2 hydrates instantly formed in the unconsolidated sediments with CO2 introduction, and prevented any upward leakage of CO2. The main findings are summarized as follows: • CO2 hydrates formed in the CO2 hydrate stability zone of the cell during the CO2 liquid injection process. The CO2 hydrate-bearing sediment layer prevented any upward flow of CO2. This selftrapping effect was confirmed by monitoring pressure, P-wave velocity, and electrical resistance.

•
The original CO2 hydrates partially dissociated during the depressurization process, and additional CO2 hydrates instantly formed in the upper layer, which was in the CO2 hydrate stability zone. When CO2 hydrates dissociated, CO2 hydrates could re-form in the upper layer (i.e., cooler layer for marine sediments) instantly. This behavior is a positive characteristic of CO2 hydrates for use as cap-rock in CGS applications.

•
The CO2 hydrate-bearing sediment layer maintained a pressure of 0.71 MPa during the experiment. Simple analyses revealed that the capillary and permeability sealing capacity of CO2 hydrate-bearing sediments are considerably high.
Permeability and breakthrough pressure of CO2 hydrate-bearing sediments depend on the saturation of CO2 hydrates in the pore system. However, CO2 hydrate saturation was not analyzed in detail in the present study, as sufficient information regarding maximum CO2 hydrate saturation via experimental simulation was lacking. We expect that a geophysical analysis using experimental data from a denser sensor array could overcome this limitation. Meanwhile, the electrical and geochemical behavior of the CO2-containing sediments in this study was different from that in real marine sediments because distilled water was used as pore water instead of saline water. To overcome this limitation, an experiment using saline water will be performed for further study.

Conclusions
We performed an experimental simulation of CO 2 geological storage in marine unconsolidated sediments in this study. CO 2 hydrates were formed during the CO 2 liquid injection process, and we observed the self-trapping effect of CO 2 hydrates. In addition, simple analyses were conducted using the experimental results. The feasibility of CO 2 geological storage in marine unconsolidated sediments was experimentally verified using 1-m-height high-pressure cell. CO 2 hydrates instantly formed in the unconsolidated sediments with CO 2 introduction, and prevented any upward leakage of CO 2 . The main findings are summarized as follows: • CO 2 hydrates formed in the CO 2 hydrate stability zone of the cell during the CO 2 liquid injection process. The CO 2 hydrate-bearing sediment layer prevented any upward flow of CO 2 . This self-trapping effect was confirmed by monitoring pressure, P-wave velocity, and electrical resistance.

•
The original CO 2 hydrates partially dissociated during the depressurization process, and additional CO 2 hydrates instantly formed in the upper layer, which was in the CO 2 hydrate stability zone. When CO 2 hydrates dissociated, CO 2 hydrates could re-form in the upper layer (i.e., cooler layer for marine sediments) instantly. This behavior is a positive characteristic of CO 2 hydrates for use as cap-rock in CGS applications.

•
The CO 2 hydrate-bearing sediment layer maintained a pressure of 0.71 MPa during the experiment. Simple analyses revealed that the capillary and permeability sealing capacity of CO 2 hydrate-bearing sediments are considerably high.
Permeability and breakthrough pressure of CO 2 hydrate-bearing sediments depend on the saturation of CO 2 hydrates in the pore system. However, CO 2 hydrate saturation was not analyzed in detail in the present study, as sufficient information regarding maximum CO 2 hydrate saturation via experimental simulation was lacking. We expect that a geophysical analysis using experimental data from a denser sensor array could overcome this limitation. Meanwhile, the electrical and geochemical behavior of the CO 2 -containing sediments in this study was different from that in real marine sediments because distilled water was used as pore water instead of saline water. To overcome this limitation, an experiment using saline water will be performed for further study.