gas phase transitions during methane hydrate dissociation

Journal Pre-proof Direct measurement of pore gas pressure and water/gas phase transitions during methane hydrate dissociation Mojtaba Shakerian, Armin Afrough, Sarah Vashaee, Florea Marica, Yuechao Zhao, Jiafei Zhao, Yongchen Song, Bruce J. Balcom PII:

S0264-8172(20)30079-9

DOI:

https://doi.org/10.1016/j.marpetgeo.2020.104296

Reference:

JMPG 104296

To appear in:

Marine and Petroleum Geology

Received Date: 30 September 2019 Revised Date:

6 February 2020

Accepted Date: 10 February 2020

Please cite this article as: Shakerian, M., Afrough, A., Vashaee, S., Marica, F., Zhao, Y., Zhao, J., Song, Y., Balcom, B.J., Direct measurement of pore gas pressure and water/gas phase transitions during methane hydrate dissociation, Marine and Petroleum Geology (2020), doi: https://doi.org/10.1016/ j.marpetgeo.2020.104296. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2020 Published by Elsevier Ltd.

1. Mojtaba Shakerian (The first author): performed all the design/fabrication steps, experimental measurements/tests, data analysis and interpreted the results. He also wrote the first draft of the corresponding article. 2. Armin Afrough: assisted the first author in performing some MR/MRI measurements. 3. Sarah Vashaee: provided some assistance in executing the bulk T1-T2 measurements. 4. Florea Marica: provided some assistance in executing the bulk T1-T2 measurements. 5. Yuechao Zhao: provided some assistance in interpreting of the results. 6. Jiafei Zhao: provided some assistance in interpreting of the results. 7. Yongchen Song: provided some assistance in interpreting of the results. 8. Bruce J. Balcom (The corresponding author): Ideas, writing-review and editing, supervision, project administration and funding acquisition.

1

Direct Measurement of Pore Gas Pressure and Water/Gas Phase Transitions during Methane Hydrate Dissociation Mojtaba Shakeriana, Armin Afrougha, Sarah Vashaeea, Florea Maricaa, Yuechao Zhaob, Jiafei Zhaob, Yongchen Songb and Bruce J. Balcoma

a

MRI Center, University of New Brunswick, Fredericton, NB, E3B 5A3, Canada

b

Key Laboratory of Ocean Energy Utilization and Energy Conservation of the

Ministry of Education, Dalian University of Technology, Dalian, 116024, China

'Declarations of interest: none'. Corresponding Author: Prof. Bruce. J. Balcom Telephone: 1 506 458 7938 Fax: 1 506 453 4581 Address: University of New Brunswick, Physics Dept. 8 Bailey Drive, Physics & Admin. Building, Rm 206, Fredericton, NB, E3B 5A3 Canada

2

Abstract Methane hydrate deposits worldwide are vast potential sources of natural gas. Although field tests, and many laboratory studies, of hydrate dissociation have been performed, long term gas recovery from hydrate deposits still requires a comprehensive knowledge of the pore gas pressure and related phenomena. Pore gas pressure can significantly affect the safety and efficiency of gas production from hydrate deposits. Capillary-trapped residual gas saturation, known to cause elevated pore gas pressure during methane hydrate dissociation, was measured by magnetic resonance. Elevated pore gas pressure was estimated to be 8500 psi. Different molecular species and fluid environments produced during the methane hydrate dissociation process were discriminated. The results show that the majority of gas is initially confined as capillary-trapped gas upon dissociation. The evolution of water and gas saturations was measured as a function of time. Water migration, bed failure, and crack growth, related to elevated pore gas pressure, were observed both spatially and temporally resolved. Hydrate dissociation proceeded from the sand pack exterior to the interior, in a shrinking core manner, due to heat transfer effects. It was observed that hydrate dissociation resulted in pronounced water migration toward the low-pressure surface. This study was undertaken with advanced magnetic resonance imaging (MRI) and magnetic resonance (MR) methods employing a MR/MRI-compatible metallic core holder. A hydrate-bearing sand pack, with 96% initial hydrate saturation, underwent dissociation by depressurization at 290 psi and 4°C.

3

Keywords: Hydrate dissociation, Magnetic resonance imaging, MR/MRIcompatible metallic core holder, Phase evolution, Capillary-trapped residual gas, and elevated pore pressure

4

1. Introduction The vast quantity of methane gas in gas hydrate deposits worldwide makes them an attractive energy resource [1]. Contemporary interest thus exists for evaluation of gas production technologies relevant to gas hydrate-bearing sediments. Three techniques have been proposed to exploit hydrate resources: depressurization, thermal stimulation, and inhibitor injection. All three methods decompose, or dissociate, solid hydrates to gas and water phases [2]. Existing laboratory investigations and field tests suggest that depressurization is the most reliable, productive, and economic gas recovery method [3].

The

depressurization method decreases the reservoir pressure below hydrate equilibrium conditions at the reservoir temperature [3]. Commercial gas recovery from gas hydrate deposits still encounters complex technical difficulties, economic challenges, and environmental safety hazards [2,3]. Addressing such complications requires significant effort and investigation at both the pilot and laboratory scales. Hydrate dissociation via depressurization is a multi-phase process, with complex spatial and temporal evolution, that can change the pore structure and mechanical properties of the reservoir [2,3]. This may affect the gas recovery efficiency and mechanical stability of sediment layers during gas production [4]. The volume expansion associated with hydrate dissociation can generate pore pressures of thousands of pounds per square inch (psi) in confined pore spaces [5]. The residual gas saturation, capillary-trapped in pore spaces, has been suggested as the cause of elevated pore gas pressure [5].

5

The pore pressure can decrease the sediment shear stress, causing bed deformation or failure near production wells or offshore drilling rigs [6,7].The fault zones, caused by bed deformation, can provide pathways along the hydratebearing sediment that function as conduits for liberated gas moving to the production well [5,6]. This can improve the efficiency of gas production from the hydrate reservoir [5,6]. The pore pressure affects the safety of drilling through the hydrate-bearing sediments and the mechanical reliability of the casing [7-9]. For these reasons the pore gas pressure is of fundamental importance to exploration of methane hydrate systems and gas production from hydratebearing sediments. Many theoretical models have been developed to estimate the pore gas pressure, flow pathways and fluid fluxes related to the methane hydrate dissociation process [5,10]. These models rely on measured porosity, and seismic data, neglecting the dynamic equilibrium between water, gas and hydrate phases as this equilibrium changes with time during the dissociation process [5,11-12]. Models based on porosity can not employ the dynamic feedback between sediment compaction and pore pressure in their estimations [5,13]. Seismic data, obtained from seismic reflection profiles, does not generally discriminate between the capillary-trapped gas and free gas [14]. This can decrease the accuracy of seismic based models. The models developed have not been verified with experimental data because direct pore gas pressure measurements are challenging, if possible at all [15-19].

6

Magnetic Resonance (MR) and Magnetic Resonance Imaging (MRI) are nondestructive techniques that have proven advantageous in previous laboratory studies of hydrate formation and dissociation processes [20,21]. MR/MRI techniques, in principle, can directly measure the pore gas pressure and monitor related phenomena [21]. Well-designed MR and MRI methods can also permit direct measurement of gas and water saturations, fluid dynamics and fluid environment, all space and time resolved [22]. MR relaxation time measurements, T1 (spin-lattice relaxation time constant) and T2 (spin-spin relaxation time constant), can be employed to monitor gas and water phase transitions of hydrate dissociation processes [21,23]. The physical properties of fluids and their environmental conditions influence the MR relaxation time constants [24]. An increase in pressure increases the T1 lifetime of methane gas as a spin-½ gas due to the spin rotation relaxation mechanism [25]. But the water T1 depends on the pore surface/fluid interaction [24]. In comparison to traditional T1 or T2 measurements, a T1-T2 relaxation correlation measurement can improve one’s ability to discriminate complex fluid species and environments produced during methane hydrate dissociation process because of employing two contrasts rather than one [26]. In the present study, an elevated pore gas pressure of 8500 psi was estimated with bulk T1-T2 MR measurements during hydrate dissociation. The presence of capillary-trapped residual gas, known to cause elevated pore gas pressure, was experimentally identified. The capillary-trapped residual gas saturations were quantified with bulk T1-T2 MR measurements. Bulk T1-T2 MR measurements permitted different molecular species and fluid environments to be observed

7

during hydrate dissociation. In this work water and gas phase transitions occurring during hydrate dissociation were measured with MR/MRI. The results show that the majority of gas is initially confined as capillary-trapped gas upon dissociation. A variety of phenomena associated with an elevated pore gas pressure including water migration, sand pack failure, crack growth, and the spatial pattern of hydrate dissociation were observed spatially and temporally resolved. The interdependence between these phenomena and the pore gas pressure is discussed with the interpretation of 1D MRI measurements, 3D MRI measurements and bulk T1-T2 MR measurements. These experimental results may assist engineers and researchers to better understand methane hydrate dissociation processes. In particular, estimation of the pore gas pressure and identification of the capillary-trapped residual gas can enable safer drilling procedures and more efficient gas production. The pore gas pressure estimated can also be employed to modify and calibrate existing theoretical models. The methane hydrate dissociation process, presented in this paper, was studied employing a new style MR/MRI compatible metallic core holder at a low magnetic field of 0.2 T with a suite of advanced MR/MRI methods. Low field MRI instruments are more compatible with equipment/components required for high pressure petroleum studies [27].

8

2. Experimental Section 2.1. Apparatus A custom-built metallic core holder was employed for 1H MRI measurements of hydrate dissociation at 8.52 MHz [28,29]. The core holder, illustrated in Fig. 1, was placed vertically in the bore of a Maran DRX-HF magnet operating at 0.2 T (Oxford Instruments Ltd., Oxford, UK). This 0.2 Tesla MRI instrument, with a vertical wide bore magnet, readily accommodated the core holder, heat exchange jacket and inlet/outlet tubing. The metallic core holder benefitted from an RF probe inside the metal core holder and an exterior heat exchange jacket [28, 29]. The core holder was fabricated from Inconel 718 and built for operation with a sustained working pressure of 4000 psi, between -28 and 80 ˚C. The core holder was built with a safety factor of 4. An open frame RF probe [29] was installed inside the core holder Fig. 1. The RF probe design minimizes MR background signal. The RF probe was immersed in Fluorinert FC-40 (3M Electronic Liquid, Saint Paul, MN, US), employed as the confining fluid. The confining fluid, with no 1H MR signal, exerts force on the core holder interior and functions as a thermal and dielectric bath [29]. A heat exchange jacket enclosing the core holder exterior regulates the confining fluid temperature, as shown in Fig. 1. A non-magnetic Type-T thermocouple, and pressure gauges, were attached to the core holder. Pressure gauges monitored the sample vessel and the confining fluid pressures during measurements. The thermocouple head was submerged in the

9

confining fluid to ensure accurate temperature measurements. The thermocouple was disconnected from the data acquisition system during MR/MRI measurements to preclude the introduction of external RF noise.

Fig. 1. Cross-sectional diagram of the MR/MRI compatible metallic core holder: (a) sample vessel, (b) hydrate-bearing sand pack, (c) 290 psi gas head and (d) metal closure. The outlet/inlet tube (e) was blocked. The temperature regulating fluid (f) was passed through the space between (g) the heat exchange jacket and (h) the metal vessel. The open frame RF probe (i) was submerged in the high pressure confining fluid (j). Fluorinated oil was utilized as the confining fluid filling the space between the RF probe and the sample vessel. Solid coax cable (k) was soldered to the RF coil. A thermocouple (l) employed in the closure monitored the temperature of the confining fluid.

10

2.2. Hydrate Formation and Dissociation Procedures A sand pack (sand, mesh -50+70, Sigma-Aldrich, Oakville, CA), with length 38 mm, diameter 33 mm and a porosity of 44% was employed as the host porous media. The dry sand pack had a weight of 45.10 gram. Based on the Kozeny-Carman model [30], the sand pack, with 44% porosity and an average particle size of 0.25 mm, had a water permeability of 95 Darcy. The sand pack resembles sediment layers hosting natural methane hydrate deposits [31]. The sand pack was saturated with 14.7 grams of distilled water solution containing 500 ppm sodium dodecyl sulfate (SDS) (Fisher, Toronto, CA). Although, SDS surfactant may alter the methane hydrate morphology, it was added to reduce the induction period and to speed the rate of hydrate reaction [32,33]. Distilled water was employed rather than brine to ensure more complete hydrate formation [34]. A sample vessel (2.5 mm wall thickness, length 51 mm) fabricated from PEEK enclosed the sand pack. The sample vessel was placed in the center of the core holder, as illustrated in Fig. 1. The heat exchange jacket stabilized the confining fluid temperature to 4°C and methane gas (Praxair Canada Inc, Fredericton, CA) was introduced into the sample vessel at 1500 psi during the hydrate formation process, as shown in Table 1. A Teledyne ISCO 100DX pump (Teledyne ISCO, Lincoln, NE, US) set the confining pressure at 1650 psi. Most of the water saturating the sand pack was converted to methane hydrate within ten days. The hydrate-bearing sand pack had a 2.8% irreducible water saturation, preventing further hydrate formation [21]. These

11

experimental procedures have previously been reported as part of a related hydrate formation study [21]. The gas head pressure was decreased from 1500 psi to 290 psi over approximately 30 mins to initiate dissociation. The methane hydrate was dissociated at 290 psi and 4 °C, as shown in Table 1. The confining pressure was held at 500 psi during hydrate dissociation. A back-pressure regulator (EB1HP1, Equilibar, Fletcher, NC, US) set the pressure of the sample vessel at 290 psi during hydrate dissociation. The hydrate bearing-sand pack, enclosed in the sample vessel, was the only source of gas and water during the hydrate dissociation process. Table 1. Pressure and temperature conditions employed during methane hydrate formation and dissociation studies Process

Pressure (Psi)

Temperature (℃)

Formation

1500

4

Dissociation

290

4

2.3. MR/MRI Methodology MR/MRI measurements were undertaken in a continuous cycle during hydrate dissociation. These measurements include 1D dhk SPRITE MRI, bulk Free Induction Decay, bulk CPMG, bulk T1-T2 and 3D π-EPI MRI measurements. These five measurements were continued for the first 24 hours. 1D dhk SPRITE MRI is ideal for imaging fluid content in the hydrate bearing sand pack [35]. With short encoding times and sufficiently long relaxation delays this method yields fluid saturation images [35]. 1D dhk SPRITE profiles were acquired with 64 k-space points in 4 mins with 16 signal averages. The RF pulse was 9° with

12

duration of 1.5 µs at 100% RF power. The encoding time tp was 150 µs with a repetition time TR of 2 ms. A delay time of 5 s, equal to 5× the water saturated sand T1, was employed between each of the two halves of data acquisition. π-EPI is a fast 3D frequency encoding MRI method [36] employed in this work to monitor the spatial distribution of recovered water during hydrate dissociation. The π-EPI method produces high quality 3D images with relatively short acquisition times [36]. Each 3D π-EPI image was acquired in 25 mins with 16 signal averages. Individual π-EPI images were not smoothed and had a nominal resolution of 1.4×1.4×1.3 mm3. A delay time of 5.7 s was employed between each of the 16 interleaved trajectories in k-space. The echo time was 3 ms. A short echo time and the long T2s of different species in the system resulted in 3D fluid saturation π-EPI images [36]. The bulk T1-T2 measurement, as a 2D MR relaxation correlation method, identified different fluid phases and environments in the dissociating hydrate system by measuring 1H density as a function of T1 and T2 relaxation time constants. Bulk T1T2 measurements [26,37] required 22 mins with 4 signal averages with a delay time of 14.9 s. The T1 of peak P1 in Fig. 2 was longer than anticipated based on the choice of timing parameters for the T1-T2 measurement but was still sufficiently measured. Bulk CPMG, and free induction decay measurements [38] were undertaken throughout the hydrate dissociation process. 3. Results and Discussion 3.1. Discriminating water and gas species in the system One major objective of this study was to estimate the pressure of methane gas, trapped in the pore space during methane hydrate dissociation. This gas is known

13

as capillary-trapped residual gas [5,10]. Capillary-trapped residual gas saturation results in an extra pore gas pressure during hydrate dissociation [5,10]. The signal from recovered water and low pressure gas saturation must be discriminated from capillary trapped gas. Bulk T1-T2 measurement characterized molecular species present in the hydrate system [21]. Bulk T1-T2 measurements were acquired at regular intervals throughout the 22 hours of hydrate dissociation. Five representative bulk T1-T2 plots are reported in Fig. 2. The bulk T1-T2 measurements, Fig. 2a-e, have signal to noise ratios of 46, 100, 376, 530 and 436 in the time domain. Variation of the regularization parameter did not change the major characteristics of the peaks [26]. The four distinctive peaks, or clusters of peaks, shown in Fig. 2, may be ascribed as follows: peak P1 is high-pressure capillary-trapped gas, P2 is water recovered in the apparatus and in the pore space, P3 is 290-psi methane in the gas head, while P4 is 290-psi liberated gas in the pore space.

14

Fig. 2. Bulk T1-T2 of the hydrate-bearing sand pack as a function of time during hydrate dissociation. In each sub figure, the recovered water saturation is reported at top right: (a) 0.2 hours (b) 2.4 hours, (c) 5 hours, (d) 7.3 hours, and (e) 21.9 hours. The peak labeled P1 is capillary-trapped residual gas saturation, P2 is bulk-like water in the apparatus and water in the pore space, P3 is 290-psi bulk gas in the gas head, while P4 is 290-psi gas saturation in the pore space. The display scale employed maps color to signal intensity of the peaks observed in each sub figure. Signal is normalized to the highest intensity of each sub figure.

15

Bulk T1-T2 measurements, undertaken during the hydrate formation process, identified a peak with a long T1 signal component [21]. This peak was assigned to the residual gas saturation trapped in the pore space. Peak P1 in Fig. 2 has a very long T1 and is thus likely to be high pressure gas. Capillary-trapped residual gas is known to yield high pore pressures, up to tens of 1000 psi, during hydrate dissociation [5,24]. An increase in pressure increases the T1 lifetimes of spin-½ gases due to the spin rotation relaxation mechanism [25,39]. Peak P1 can not be assigned to capillary trapped residual gas through a control experiment. We must examine alternate explanations. We now consider these alternate assignments for peak P1 then separately examine and reject each hypothesis. There are three possible explanations for peak P1 (1) water closely associated with the hydrate network, (2) methane gas dissolved in the water phase, and (3) methane gas nanobubbles. The first hypothesis seems unlikely. The integrated signal ratio of peaks P1/P2 is approximately 3.5 at 0.2 hours of dissociation, Fig. 2a. In a water wet system, it is difficult to assume that water closely associated with the hydrate network would not appear as part of peak P2. Although only 5% of the initial water saturation was recovered at 0.2 hours, the T1 and T2 lifetimes of peak P2 are consistent with the lifetime of water in the pore space, somewhat more than 1 s [40]. The quantity of material associated with peak P1 is important in considering the second hypothesis. The quantity of material is significantly larger than that suggested from a 0.1% methane solubility in water [41-43]. Although SDS is a surfactant and may be expected to enhance hydrocarbon solubility, it has been rejected as a promoter of methane solubility in previous studies [42]. Control

16

measurements showed that methane solubility in SDS solution was negligible at 290 psi. Methane nanobubble formation [44,45] is an unlikely explanation of the large signal in P1, when considering (i) methane pressure in the nanobubbles, (ii) long nanobubble lifetimes and (iii) low quantity of methane gas in nanobubbles. The methane pressure in nanobubbles has been reported to be roughly 870 psi during hydrate dissociation [44,45]. Control measurements, including bulk T1-T2 and conventional MR measurements, revealed that the T1 lifetimes of methane gas saturating sand and bulk methane gas at 870 psi were 468 ms and 1.2 s respectively. Both T1 lifetimes are dramatically less than the T1 of peak P1. The lifetime of methane nanobubbles after hydrate dissociation is reported to be as long as two weeks [44,45]. Peak P1 decreased in amplitude from 0.2 hours to 5 hours, then disappeared at 7.3 hours.

A five-hour lifetime of peak P1 is

insignificant compared to the anticipated lifetime of methane nanobubbles. An absolute total of 0.003 moles of methane gas could be stored in nanobubbles given the initial water saturation (14.7 g) based on literature reports of nanobubble saturation. 600 cm3 of methane gas can be stored in nanobubble solution per 1 dm3 water at 25 °C [45]. This quantity of methane, 0.003 mole, is much less than the 0.077 moles of methane gas calculated from the amplitude of peak P1 at 0.2 hours. None of the alternate hypotheses discussed above explains the unusual T1 and T2 values of peak P1. A very similar peak appears in analogous hydrate formation experiments [21]. A viable hypothesis for such a long lifetime peak should explain the methane behaviour during both formation and dissociation processes.

17

This results in an examination of capillary-trapped residual gas saturation as the fourth hypothesis. Capillary-trapped residual gas refers to the gas saturation that was released, then trapped in the pore space, causing high local pore fluid pressure due to volume expansion associated with hydrate dissociation [5,10]. The volume occupied by liberated gas and recovered water is dramatically larger than the initial volume of hydrate. A low dissociation pressure leads to significant volume expansion. If hydrate dissociates at a pressure of 360 psi a volume expansion of up to 500% can be anticipated [5]. Pore gas pressure is also proportional to initial hydrate saturation and sediment bulk stiffness [5,10]. High initial hydrate saturation results in a larger bulk stiffness, generating higher pore pressures [10,46]. Initial hydrate saturations greater than 50% can result in hydrate-bearing sediments with a bulk stiffness between 1450 ksi to 14500 ksi [46-48]. For example, dissociation of a hydratebearing sediment with only 20% initial hydrate saturation and 1450 ksi bulk stiffness can result in a pore pressure of 8700 psi [10]. It is reasonable to assume that a small fraction of the liberated gas dissolves in the recovered water phase; however, the remainder remains in the initially hydrate-filled pore space before displacing water and escaping from the sand pack [10,49]. As mentioned above, a hydrate-bearing sand pack with 96% initial hydrate saturation was employed in this study [21]. Such a high hydrate saturation is relevant to field studies. The Nanaki methane hydrate deposit showed a hydrate saturation of 80%, based on field test measurements [50]. The 96% hydrate saturation will result in a bulk stiffness, up to 14500 ksi, causing significant pore pressures during hydrate dissociation [5,46]. Based on a modified

18

Kozeny-Carman model [51], the hydrate bearing sand pack with 96% hydrate saturation resulted in a permeability of 0.7 mD. This is a dramatic reduction in the primary permeability of the sand pack, 95 Darcy, and suggests that fluid flow through the sand pack will be greatly hindered during the initial stages of hydrate dissociation. This phenomenon can significantly elevate the pressure of the gas saturations, trapped in the pore space, up to thousands of psi [5]. Significant MR evidence in support of the capillary-trapped residual gas hypothesis is the bifurcation in the T1 and T2 relaxation times of peak P1. Methane gas at pressures less than 1500 psi has equivalent T1 and T2, with lifetimes determined by spin rotation [25,52]. At pressures above 1500 psi, molecular dipolar interactions dominate MR relaxation processes [25,52] resulting in increased relaxation lifetimes that bifurcate in T1 and T2 with T2s substantially less than T1s. Methane gas at elevated pressure in microporous solids is also reported to bifurcate in T1 and T2; with T2 much less than T1 [52]. Kausik [52] investigated this phenomenon in detail in a model porous vycor glass. The decrease in T2 is due to interaction with the pore surface causing slower motion and a greater sensitivity to relaxation [52]. The capillary-trapped residual gas hypothesis requires that the pore pressure increases as gas is released, with gas trapped in the pore space while expansive hydrate dissociation occurs [10,53]. During hydrate dissociation, the T1 of peak P1 remains on the order of 10 s but decreases in amplitude. This would be expected if the gas pressure in the pore space is quasi constant. The fact that T1 is constant suggests maintenance of a high pressure while a decrease in amplitude suggests gas is released from the capillary trapped environment.

19

Bulk T1-T2 measurements, Fig. 2a, reveal that the absolute quantity of methane gas associated with peak P1 is 0.077 moles 0.2 hours after commencement of dissociation. Five hours after commencement of dissociation, capillary-trapped residual gas decreased from 0.077 moles to 0.004 moles. This resulted in a gas recovery efficiency of 95% after 5 hours of gas production. The gas recovery efficiency is hypothesized to be proportional to the trapped residual gas saturation [49]. Peak P1 disappeared from T1-T2 plots, Fig. 2, 5 hours after commencement of dissociation. The disappearance of peak P1 was concurrent with mechanical failure of the sand pack, as observed in the MRI images illustrated in 3.2 section. This reinforces the hypothesis suggested by others that release of capillary trapped gas is related to bed failure [5,10]. The experimental results from this study show that the majority of liberated gas was initially capillary trapped. We now consider quantification of the pore pressure based on the pressure dependence of the spin-lattice relaxation time constant T1 of methane gas. An increase in methane pressure increases T1 [25,52]. Fig. 3 plots T1 of the methane gas saturating an identical dry sand pack as a function of pressure. The T1 values were obtained from MR control measurements of methane gas saturating a dry sand pack at 8.52 MHz. Extrapolating the linear fit permits one to predict the pressure of the capillary-trapped methane gas, peak P1, with a T1 of 10 s. The estimated pressure was 8500 psi. Independent estimation based on the ideal gas law, with approximate volumes, yielded similarly high pressures. Such a high pore gas pressure has been reported in the literature [10]. This is also in agreement with the low permeability of 0.7 mD reported earlier in this section. An elevated

20

pore pressure of 8500 psi, resulting from a dissociation pressure of 290 psi, led to an excess pore pressure of 8200 psi. An elevated pore gas pressure, 8500 psi, confirms that dissociation of methane hydrate occurred in confined pore spaces [5,10]. In offshore fields, such a high pore gas pressure may be accumulated below a horizontally large methane hydrate layer, which seals the dissociation zone [5,10]. A pore pressure of 8500 psi is large enough to overcome the strength of overlying sediment. For example, the regions close to the bottom-simulating reflector, BSR, at the Blake ridge collapse had an overburden pressure of 4000 psi to 6000 psi [5,10]. The BSR represents the methane hydrate and free gas phase boundary in methane hydrate deposits. A dissociation of only 1% methane hydrate saturation in confined pore spaces would result in a 1500 psi pressure increase, and a pore gas pressure that may overcome, the overburden pressure [5,10]. A pore gas pressure as high as of 8500 psi can cause the growth of cracks, which may develop an unstable region within the host sediment [5,10]. This region, weakened by the pore gas pressure, may undergo sudden mechanical failure [10]. 3D π-EPI measurements, section 3.2, observed destabilization of the sand pack as mentioned above. This supports the hypothesis suggested by others that a pore gas pressure of thousands of psi can lead to mechanical failure of the host sediment [5,10]. An elevated pore gas pressure of thousands of psi resulting from methane hydrate dissociation in nature can cause landsides, sea level fall or tectonic uplift [5,10]. The magnitude of this sediment relocation can be calculated employing

21

the pore gas pressure [5,10]. 3D π-EPI measurements, section 3.2, showed movement of the sand pack front during the release of pore pressure. The pore gas pressure of 8500 psi, estimated in this study, can be also employed in models to quantify the volume expansion associated with methane hydrate dissociation [5,10]. The dissociation of methane hydrates close to the base of the hydrate stability zone requires a finite amount of heat. The measured pore gas pressure can permit determination of the required heat and predict the geological behaviour of the host sediment [5,10]. Detailed analysis and calculations of these phenomena, associated with the pore gas pressure, are beyond the scope of this study but should be pursued in future work. The MR methods employed in this study directly measured the gas trapped in the pore space, and its pressure, during methane hydrate dissociation. The MR/MRI methods simultaneously acquire many different data sets. Processing and interpreting all of the data acquired permitted study of many hydrate dissociation phenomena, concurrent with monitoring the pore pressure. The MR technique is non-destructive, cost-effective and operable in laboratory environments. This makes it possible to study the influence of different parameters on pore gas pressure in controlled experiments. Non-destructive study of pore pressure using conventional laboratory tests would be challenging, if possible at all. Laboratory measurements of elevated pore gas pressure, induced by methane hydrate dissociation, have not been addressed in the literature. Limited studies has been undertaken with subsea pore gas measurements in pilot field applications [15-19]. No existing methods directly measure the pore gas pressure.

22

They also do not provide any insight relevant to other phenomena occurring during methane hydrate dissociation [17,18].

Fig. 3. T1 of methane gas saturating a dry sand pack ( ) plotted as a function of pressure. The data show a straight line relationship. These T1 control measurements were undertaken at 25 ◦C with the same apparatus as for dissociation measurements. Having extensively analysed peak P1 we now return to the behaviour of the other peaks in the T1-T2 measurements of Fig. 2. The cluster of peaks labeled P2 are assigned to water in the pore space and in the apparatus. This peak changes slightly in T1-T2 coordinates, shifting from 1 s to 2 s during time 0.2 hours to 2.4 hours, as seen in Fig. 2a-b. The shift in T1-T2 coordinates of peak P2 is concurrent with the water migration observed in the 3D π-EPI measurements. The migrated water had T1 and T2 values similar to bulk SDS solution with a T1 and T2 of roughly 2 s. The T1-T2 coordinates are consistent with bulk like water, undergoing minimal surface relaxation. This was confirmed by control measurement.

23

Peak P2 increases in amplitude as hydrate dissociation proceeds, as revealed in Fig. 2a-c. The recovered water saturation, determined from peak P2, reached 90% of the initial water content of the sand pack, before initial hydrate formation. Peak P2 decreased slightly in amplitude from 7.3 hours to 21.9 hours. Such a water loss was confirmed spatially by 1D dhk SPRITE and 3D π-EPI measurements. It is assumed that water was displaced from the measurement volume into the external volume of the apparatus. Peak P3 is liberated methane gas that accumulated in the reservoir head at a pressure of 290 psi. This peak has consistent integrated signal and invariable T1-T2 coordinates during the first six hours, as observed in Fig. 2a-c. The T1 and T2 of P3 are roughly equivalent suggesting high mobility and a 1H species in the extreme narrowing regime [24]. This was confirmed by control measurement. The integrated signal from peak P3 decreased by 50% at time 5 hours when the sand pack top moved and decreased the available head space. Pressure gauges connected to the sample vessel verified a constant pressure of 290 psi throughout hydrate dissociation. The methane gas cylinder was disconnected from the sample vessel during dissociation and the hydrate bearing sand pack was the only source of methane gas during hydrate dissociation. Peak P4 is the 290-psi liberated gas saturating the sand pack during hydrate dissociation. This peak does not change in T1-T2 coordinates, nor in amplitude, during the first 6 hours, but it does increase in T2 coordinate and amplitude as the sand pack top moved upwards at time 5 hours. At a constant pressure of 290 psi, an increase in gas-occupied pore volume, due to the bed expansion, increases the T2 of the gas saturating the pores. This was confirmed by control measurement.

24

3.2. Spatial and Temporal Monitoring of Hydrate Dissociation The 1D dhk SPRITE MRI measurements were undertaken to monitor the longitudinal hydrate dissociation process. 1D dhk SPRITE profiles, Fig. 4, spatially resolved bulk gas in the reservoir head and recovered water in the reminder of the vessel during hydrate dissociation. Measurement parameters were set to yield fluid saturation profiles for liberated methane gas, at 290 psi, and water recovered in the pore space due to hydrate dissociation. The 1D dhk SPRITE profiles are dominated by the spatial variation of water content in the pore space. SPRITE profiles did not resolve capillary-trapped residual gas due to relaxation time contrast. This was verified by control measurements. The longitudinal hydrate dissociation process was monitored with the 1D dhk SPRITE measurements. The 1D longitudinal profiles, illustrated in Fig. 4b, report the spatial variation of water content in the pore space. Methane hydrate dissociation commenced at the sand pack/gas head interface, as shown in Fig. 4b. The MRI signal observed from the gas head fluctuated in amplitude during the first three hours after commencement of dissociation, as seen in Fig. 4b. The migration of recovered water into the gas head caused this variation in signal amplitude. Such a water migration was anticipated based on the literature [54]. 3D π-EPI images also verified that a fraction of the recovered water was displaced from the sand pack to the gas head during the first three hours. The hydrate dissociation process was completed in 22 hours with two distinct stages. A two stage dissociation was also anticipated based on the literature [55]. In the first stage, 8 hours in length, the water content in the pore space increased to a maximum saturation of 75%. In the second stage, 14 additional hours, the

25

water saturation gradually decreased by 7%, as observed in Fig. 4b-c. This water signal loss is ascribed to water migration [9]. The recovered water saturation was determined based on the signal integrated from the sand pack area in 1D dhk SPRITE profiles, and referenced to the initial water saturated sand pack.

Fig. 4. Space- and time-resolved 1H fluid content profiles, Y direction, of hydrate dissociation. 1H profiles before hydrate formation (+) and after hydrate dissociation (○) are shown in (a). The gas head is visible at left, 24 mm to 30 mm. Observed signal intensity is normalized by the maximum water signal in the zerotime profile before the introduction of gas. Early-time 1H fluid content profiles at t=0.5 (•), 2 (▪) and 3 (◊) hours of dissociation are shown in (b). Late-time 1H fluid content profiles at t=5 (▪), 8 (•), 12 (+) and 22 (○) hours of dissociation are shown (c). (d) Schematic of the sample vessel. The gas head (orange region), connected to a back pressure regulator, is located on top of the hydrate-bearing sand pack (turquoise region).

26

The initial hydrate dissociation was longitudinally inhomogeneous but the recovered water saturation became more uniform in space as hydrate dissociation proceeded, as seen in Fig. 4b-c. 1D profiles represent the water distribution in the sand pack longitudinally and the profiles may be misleading if there is lateral structure in the pattern of hydrate dissociation. Rapid π-EPI measurements were thus employed to monitor the hydrate dissociation process in 3D. π-EPI measurements, Fig. 5, revealed that hydrate dissociation commenced from the gas head/sand pack interface at time 0.7 hours. 2D slices extracted from the 3D π-EPI measurements at times 0.7 hours to 5.2 hours demonstrate substantial recovered water present in the periphery of the sand pack and much less near the center.

Fig. 5. 2D slices from 3D π-EPI images of 1H content in the hydrate system during dissociation. At time 0.7 hours, the image intensity increased at the interface of the sand pack and the gas head. At time 1.8 hours, the liberated gas moved some recovered water into the reservoir head. From 2.9 hours to 5.2 hours, hydrate

27

dissociation progressed from the core periphery to its interior as revealed by the pattern of water produced. From 4.5 hours to 5.2 hours, the sand pack top moved upwards into the gas head reservoir region. From time 6.2 hours to 21.25 hours, the recovered water was redistributed along the sand pack. Signal intensity (a.u.) is mapped linearly to the display range at right of the figure. Progression of the dissociation front resembles a shrinking core behaviour [56] from 2.9 hours to 5.2 hours. Hydrate dissociation is endothermic [57,58], and will decrease the local temperature, while the sample vessel wall exterior was in direct contact with the confining fluid at 4°C. It is likely that hydrate dissociation is more favorable near the wall due to heat transfer effects [54,59]. Fig. 6 plots position of the hydrate dissociation front as a function of time, in the sand pack interior, between 1.8 hours and 4 hours. The position data is extracted from the images of Fig. 5. Movement of the hydrate dissociation front is radially symmetric about the Y axis. We may, in the first instance, consider the shrinking core behaviour to be due to heat transfer limiting the dissociation process considered as a phase change. Carslaw [56] showed that for a one dimensional sample a moving boundary phase change of ice to water moves as the square root of time with heat supplied to the sample exterior. The plots of Fig. 6 do not show a square root of time dependence of front movement. Rather a parabolic dependence on time was observed. This may be due to the 3D nature of the problem as was observed in related work [60]. The trend of data points in Fig. 6 is illustrated by a spline curve fit to the data. The early time derivative with respect to time shows the rate of front movement was approximately 2.5 mm/hour at 2 hours.

28

Fig.6. Position of the hydrate desaturation front plotted along the Z ( ) and Y ( ) directions as a function of time. Smoothing splines were fit to the data. Front position in the two directions was determined from discrete lines bisecting the shrinking core at each experimental time. The position front data were extracted from the 2D YZ slices shown in Fig. 3 between 1.8 hours to 5.2 hours. Position of the front was referenced to the interior surface of the sample vessel. 2D slices from 3D π-EPI images, Fig. 5, revealed that the sand pack expanded at 5.2 hours after commencement of dissociation. The longitudinal expansion of the sand pack was also revealed in 1D dhk SPRITE profiles where the sand pack top moved upwards, into the gas head, during hydrate dissociation, as shown in Fig. 2a. Such an expansion was anticipated through previous work in the literature [7,61]. Several cracks appeared along the sand pack during this longitudinal expansion and were presumably filled with liberated gas at 290 psi. The dark regions seen in Fig. 5 between 5.2 hours and 21.2 hours are representative of these cracks. The 2D slices shown in Fig. 5 only display the water recovered in the apparatus and the pore space due to low 1H density of 290-psi methane gas compared to water. This was confirmed by control measurement.

29

Liberated gas displaced a portion of the recovered water from the sand pack to the gas head between 0.7 hours and 5.2 hours. A portion of this water was further displaced to the reservoir outlet between 5 hours and 21 hours, after commencement of dissociation, as revealed by Fig.5. This effect has also been represented in the literature [9]. Between 6.2 hours and 21.2 hours, liberated gas redistributed a significant amount of the recovered water from the sand pack bottom to the top. This phenomenon has been observed by other researchers [9,62]. 3D π-EPI measurements reveal that the quantity of water in the sand pack top is almost twice that of the sand pack bottom. 4. Conclusions In this study elevated pore gas pressure and related phenomena occurring during hydrate dissociation were studied employing a methane hydrate bearing sand pack with 2.8% residual water at 290 psi and 4 °C. The elevated pore gas pressure was estimated to be 8500 psi. Significant capillary-trapped residual gas saturation was identified during the first 5 hours of hydrate dissociation. Results showed that the majority of liberated methane gas is initially capillary trapped. The capillary-trapped gas saturation and recovered water were quantified during hydrate dissociation with bulk T1-T2 measurements. A variety of phenomena associated with the pore gas pressure, such as water migration, bed failure, and crack growth were observed. The sand pack failure occurred concurrent with the release of the capillary-trapped gas saturation from the pore space.

30

Methane hydrate dissociation commenced at the interface of the sand pack and gas head. It was observed that hydrate dissociation was more significant close to the sand pack walls, with dissociation spreading in the sand pack in a core shrinking pattern. The recovered water saturation was 82% after 22 hours. Water content in the sand pack top was higher compared to the sand pack bottom as liberated gas flow displaced water. This study verified that the gas recovery process from the hydrate bearing sediments can be divided into five major steps: (1) the gas leaves the hydrate structure, (2) the liberated gas is temporarily trapped in the pore space, causing high pore gas pressure, (3) the capillary-trapped gas saturation then escapes from the pore space, (4) the pore gas, released from the pore, causes cracks in the sand pack and (5) finally deforms the sand pack structure.

31

Acknowledgments B.J.B thanks NSERC of Canada for a Discovery grant and the Canada Chairs program for a Research Chair in MRI of Materials. The authors also thank Green Imaging Technologies, ConocoPhillips, Saudi Aramco, the Atlantic Innovation Fund and the New Brunswick Innovation Fund for financial support. M.S. thanks Jim Merrill and Brian Titus for machining.

Glossary MRI: Magnetic resonances imaging MR: Magnetic resonance T1: Spin–lattice relaxation time constant T2: Spin–spin relaxation time constant RF: Radio frequency SPRITE: Single point ramped imaging with T1 enhancement CPMG: Carr-Purcell-Meiboom-Gill π-EPI: π echo-planar imaging SE-SPI: Spin echo single point imaging

32

References 1. E.D. Sloan, C. A. Koh, Clathrate Hydrates of Natural Gases Third edition, CRC Press Taylor & Francis Group, New York, 2008, pp. 16–193. 2. X. S. Li, C. G. Xu, Y. Zhang, X. K. Ruan, G. Li, Y. Wang, Investigation into Gas Production from Natural Gas Hydrate, Appl. Energy, 172 (2016) 286-322. 3. J. Zhao, Z. Zhu, Y. Song, W. Liu, Y. Zhang and D. Wang, Analyzing the Process of Gas Production for Natural Gas Hydrate using Depressurization, Appl. Energy, 142 (2015) 125-134. 4. X. H. Zhang, D. S. Luo, X. B. Lu, L. L. Liu and C. L. Liu, Mechanical Properties of Gas Hydrate-Bearing Sediments during Hydrate Dissociation, Acta. Mech. Sin., 34 (2018) 266-274. 5. W. Xu, Excess Pore Pressure Resulting from Methane Hydrate Dissociation in Marine Sediments: A Theoretical Approach, J. Geophys. Res., 111 (2006) 1-12. 6. N. Sultan “Excess Pore Pressure and Slope Failure Resulting from Gas-Hydrate Dissociation and Dissolution”, Offshore Technology Conference, Houston, US, (2007). 7. M. F. Nixon and J. LH. Grozic, Submarine Slope Failure due to Gas Hydrate Dissociation: a Preliminary Quantification, Can. Geotech. J., 44 (3) (2007) 314325. https://doi.org/10.1139/t06-121. 8. J. M. Strout, NGI and T. I. Tjelta, Excess Pore Pressure Measurement and Monitoring for Offshore Instability Problems, Offshore Technology Conference, Houston, US, (2007). 9. J. Lu, Y. Xiong, D. Li, X. Shen, Q. Wu and D. Liang, Experimental Investigation of Characteristics of Sand Production in Wellbore during Hydrate Exploitation by the Depressurization Method, Energies, 11(7) (2018), 1673. https://doi.org/10.3390/en11071673.

33

10. T. H. Kwon and G. C. Cho, Gas Hydrate Dissociation in Sediments: PressureTemperature Evolution, Geochem. Geophys. Geosysm., 9 (3) (2008) 1-14. https://doi.org/10.1029/2007GC001920. 11. B. S. Hart, Porosity and Pressure: Role of Compaction Disequilibrium in the Development of Geopressures in a Gulf Coast Pleistocene Basin, Geology, 23(1) (1995) 45-48). 12. N. L. B. Bangs, G. K. Westbrook, J. W. Ladd and P. Buhl, Seismic Velocities from the Barbodaos Ridge Complex: Indicators of High Pore Fluid Pressures in an Accretionary Complex, J. Geophys. Res., 95(B6) (1990) 8767-8782. 13. P. B. Flemings, X. Liu and W. J. Winters, Critical Pressure and Multiphase Flow in Blake Ridge Gas Hydrates, Geology, 31 (2003) 1057-1060. 14. U. Tinivella, the Seismic Response to Overpressure versus Gas Hydrate and Free Gas Concentration, J. Seism. Explor., 11 (2002) 283-305. 15. J. M. Strout and T. I. Tjelta, In situ Pore Pressures: What is their Significance and How can They be Reliably Measured?, Mar Pet Geol., 22 (2005) 275-285. 16. E. E. Davis, K. Becker, T. Pettigrew, B. Carson and R. MacDonald, 3. CORK: A Hydrologic Seal and Downhole Observatory for Deep-Ocean Boreholes, Proceedings of the Ocean Drilling Program, Initial Reports, Vol 139 (1992) 4353. 17. K. Becker, A. T. Fisher and E. E. Davis, 19. the CORK Experimental in Hole 949C: Long-Term Observations of Pressure and Temperature in the Barbados Accretionary Prism, Proceedings of the Ocean Drilling Program, Scientific Results, Vol. 156 (1997) 247-252. 18. E. E. Davis, G. C. Horel and R. D. MacDonald, Pore Pressures and Permeabilities Measured in Marine Sediments with a Tethered Probe, J. Geophys. Res., 96 (B4) (1991) 5975-5984.

34

19. A. J. Whitte, T. Sutabutr, J. T. Germaine and A. Varney, Prediction and Interpretation of Pore Pressure Dissipation for Tapered Piezoprobe, Géotechnique., 51(7) 601-617. 20. G. Ersland, J. Husebø, A. Graue, B. A. Baldwin, J. Howard, J. Stevens, Measuring Gas Hydrate Formation and Exchange with CO2 in Bentheim Sandstone using MRI Tomography, Chem. Eng. J., 158 (2010) 25-31. 21. M. Shakerian, A. Afrough, S. Vashaee, F. Marica, Y. Zhao, J. Zhao, Y. Song, B. J. Balcom, "Monitoring Gas Hydrate Formation with Magnetic Resonance Imaging in a Metallic Core Holder", International Symposium of the Society of Core Analysts, Trondheim, SCA 025, Norway (2018). 22. R. L. Kleinberg, C. Flaum, and C. Straley, Seafloor Nuclear Magnetic Resonance Assay of Methane Hydrate in Sediment and Rock, J. Geophys. Res, 108 (B3) (2003) 2137-2150. 23. A. Afrough, M. Shakerian, M. S. Zamiri, B. MacMillan, F. Marcia, B. Newling, L. Romero-Zeron, B. J. Balcom, Magnetic Resonance Imaging of High Pressure Carbon Dioxide Displacement: Fluid/Surface Interaction and Fluid Behavior, SPE Journal, SPE-189458-PA, February 2018, https://doi.org/10.2118/189458PA. 24. G. R. Coates, L. Xiao, and Manfred G. Prammer, NMR Logging Principles & Applications (Halliburton Energy Services, Houston, 1999) pp. 33–89. 25. C. S. Johnson, J. S. Waugh, Nuclear Relaxation in Gases: Mixtures of Methane and Oxygen, J. Chem. Phys., 35 (6) (1961) 2020-2024. 26. S. Vashaee, M. Li, B. Newling, B. MacMillan, F. Marica, H.T. Kwak, J. Gao, A.M. Al-harbi, B.J. Balcom, Local T1-T2 Distribution Measurements in Porous Media, J.Magn. Reson., 287 (2018) 113-122. 27. J. Mitchell, L. F. Gladden, T. C. Chandrasekera and E. J. Fordham, Low-Field Permanent Magnets for Industrial Process and Quality Control, Prog. Nucl. Magn. Reson. Spectrosc., 76 (2014) 1-60.

35

28. M. Shakerian, F. Marica, A. Afrough, F. G. Goora, M. Li, S. Vashaee, B. J. Balcom, A High-Pressure Metallic Core Holder for Magnetic Resonance Based on Hastelloy-C, Rev. Sci. Instrum., 88 (2017) 123703. 29. M. Shakerian, B. J. Balcom, An MR/MRI Compatible Core Holder with the RF Probe Immersed in the Confining Fluid, J. Magn. Reson., 286 (2018) 36-41. 30. K. Koch, A. Revil and K. Holliger, Relating the Permeability of Quartz sands to their Grain Size and Spectral Induced Polarization Characteristics, Geophys. J. Int., 190 (2012) 230-242. 31. P. Linga, C. Haligva, S. Chan Nam, J. A. Ripmeester, P. Englezo, Gas Hydrate Formation in a Variable Volume Bed of Silica Sand Particles, Energy Fuels, 23 (2009) 5496–5507. 32. J. Yoslim, P. Linga, P. Englezos, Enhanced Growth of Methane–Propane Clathrate Hydrate Crystals with Sodium Dodecyl Sulfate, Sodium Tetradecyl Sulfate, and Sodium Hexadecyl Sulfate Surfactants, J. Cryst. Growth, 313 (2010) 68-80. 33. H. Ganji, M. Manteghian, K. Sadaghiani zadeh, M. R. Omidkhah, H. Rahimi Mofrad, Effect of Different Surfactants on Methane Hydrate Formation Rate, Stability and Storage Capacity, Fuel, 86 (2007) 434-441. 34.K. You. T. J. Kneafsey, P. B. Flemings, P. Polito and S. L. Bryant, Salinity Buffered Methane Hydrate Formation and Dissociation in Gas-Rich Systems, J. Geophys. Res. Solid Earth., 120 (2015) 643-661 https://doi.org/10.1002/2014JB011190. 35. M. Halse, D. J. Goodyear, B. MacMillan, P. Szomolanyi, D. Matheson and B. J. Balcom, Centric Scan SPRITE Magnetic Resonance Imaging, J.Magn. Reson., 165 (2003) 219-229. 36. D. Xiao, B. J. Balcom. π Echo-Planar Imaging with Concomitant Field Compensation for Porous Media MRI, J.Magn. Reson., 260 (2015) 38–45.

36

37. Y. Q. Song, L. Venkataramanan, M. D. Hurlimann, M. Flaum, P. Frulla, C. Straley, T1-T2 Correlation Spectra Obtained Using a Fast Two-Dimensional Laplace Inversion, J. Magn. Reson., 154 (2) (2002) 261-268. 38. E. M. Haacke, R. W. Brown, M. R. Thompson and R. Venkatesan, Magnetic Resonance Imaging Physical Principles and Sequence Design, John Wiley&Sons Inc., New York, 1999, pp. 1-134. 39. C. Straley, An Experimental Investigation of Methane in Rock Materials, SPWLA 38th Annual Logging Symposium, Houston, Texas (1997). 40. R. L. Kleinberg, W. E. Kenyon, P. P. Mitra, Mechanism of NMR Relaxation of Fluids in Rock, J. Magn. Reson. SerA., 108 (1994) 206-214. 41. P. Servio, P. Englezos, Measurement of Dissolved Methane in Water in Equilibrium with its Hydrate, J. Chem. Eng. Data, 47 (2002) 87-90. 42. J. Verrett, D. Posteraro, P. Servio, Surfactant Effects on Methane Solubility and Mole Fraction during Hydrate Growth, Chem. Eng. Sci., 84 (2012) 80-84. 43. I. N. Tsimpanogiannis, P. C. Lichtner, Methane Solubility in Water under Hydrate Equilibrium Conditions: Single Pore and Pore Network Studies, Proceedings of the 7th International Conference on Gas Hydrates, Edinburgh, Scotland (2011). 44. S. A. Bagherzadeh, S. Alavi, J. A. Ripmeester, P. Englezos, Evolution of Methane during Gas Hydrate Dissociation, Fluid. Phase. Equilib., 358 (25) (2013) 114-120. 45. K. Ohgaki, N. Q. Khanh, Y. Joden, A. Tsuji and T. Nakagawa, Physicochemical Approach to Nanobubble Solutions, Chem. Eng. Sci., 65 (3) (2010) 1296-1300. 46. W. F. Waite, J. C. Santamarina, D. D. Cortes, B. Dugan, D. N. Espinoza, J. Germaine, J. Jang, J. W. Jung, T. J. Kneafsey, H. Shin, K. Soga, W. J. Winters and T.-S. Yun, Physical Properties of Hydrate-Bearing Sediments, Rev. Geophys., 47, RG4003 (2009) 1-38.

37

47. J. Y. Lee, F. M. Francisca, J. C. Santamarina and C. Ruppel, Parametric Study of the Physical Propeties of Hydrate-Bearing Sand, Slit, and Clay Sediments: 2. Small-Strain Mechanical Properties, J. Geophys. Res., 115 (2010) B11105. https://doi.org/10.1029/2009JB006669. 48. A. Sultaniya, J. A. Priest and C. R. I. Clayton, Impact of Formation and Dissociation Conditions on Stiffness of a Hydrate-Bearing Sand, Can. Geotech. J., 55 (2018) 988-998. 49. J. Jang and J. C. Santamarina, Recoverable Gas from Hydrate-Bearing Sediments: Pore Network Model Simulation and Macroscale Analyses, J. Geophys. Res., 116 (20111) B08202, https://doi.org/10.1029/2010JB007841. 50. Y. Konno, T. Fujii, A. Sato, K. Akamine, M. Naiki, Y. Masuda, K. Yamamoto and Ji. Nagao, Key Findings of the World’s First Offshore Methane Hydrate Production Test off the Coast of Japan: Toward Future Commercial Production, Energy Fuels., 31 (2017) 2607−2616. 51. S. Dai and Y. Seol, Water Permeability in Hydrate-Bearing Sediments: a PoreScale Study, Geophys. Res. Lett., 41 (2014) 4176–4184. 52. A. Papaioannou, R. Kausik, Methane Storage in Nanoporous Media as Observed via High Field NMR Relaxometry, Phys. Rev. Applied, 4 (2015) 024018. 53. J. Y. Lee, J. C. Santamarina, C. Ruppel, Volume Change Associated with Formation and Dissociation of Hydrate in Sediment, Geochem. Geophys. Geosysm., 11 (3) (2010) 1-13, https://doi.org/10.1029/2009GC002667. 54. M. Yang, Z. Fu, Y. Zhao, L. Jiang, J. Zhao, Y. Song, Effect of Depressurization Pressure on Methane Recovery from Hydrate–Gas–Water Bearing Sediments, FUEL, 166 (2016) 419-426. 55. V. P. Melnikov, A. N. Nesterov and A. M. Reshetnikov, Kinetics of Hydrate Dissociation at a Pressure of 0.1 MPa, Swets&Zeitlinger, Lisse, 2003, pp. 753757.

38

56. H. S. Carslaw, J. C. Jaeger, Condition of Heat in Solids Second Edition, Oxford University Press, Oxford, 1959, pp. 282–296. 57. Y. Song, S. Wang, M. Yang, W. Liu, J. Zhao, S. Wangong, MRI Measurements of CO2–CH4 Hydrate Formation and Dissociation in Porous Media, FUEL, 140 (2015) 126-135.

58. P. Wang, M. Yang, B. Chen, Y. Zhao, J. Zhao, Y. Song, Methane Hydrate Reformation in Porous Media with Methane Migration, Chem., Eng. Sci., 168 (2017) 344-351. 59. P. Wang, S. Wang, Y. Song, M. Yan, Dynamic Measurements of Methane Hydrate Formation/Dissociation in Different Gas Flow Direction, Appl. Energy, (In Press), https://doi.org/10.1016/j.apenergy.2017.08.056. 60. C. Lamason, B. MacMillan, B. Balcom, B. Leblon and Z. Pirouz, Examination of Water Phase Transitions in Black Spruce by Magnetic Resonance and Magnetic Resonance Imaging, Wood. Fiber. Sci., 46 (2014) 1-14. 61. M. Hyodo, J. Yoneda, N. Yoshimoto,Y. Nakata, Mechanical and Dissociation Properties of Methane Hydrate-Bearing Sand in Deep Seabed, Soil Found, 53 (2) (2013) 299-314. 62. I. N. Tsimpanogiannis and P. C. Lichtner, Pore-Network Study of Methane Hydrate Dissociation, Phys. Rev. E., 74 (2006) 056303-13.

Highlights 1. Elevated pore gas pressure was directly measured with MR measurements. 2. Capillary-trapped residual gas saturation was identified and monitored. 3. Elevated pore gas pressure was estimated to be 8500 psi. 4. Water migration/sand pack failure were observed spatially and temporally resolved. 5. Different molecular species and fluid environments were discriminated.

'Declarations of interest: none'. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Thanks,