Methane Production from Gas Hydrate Deposits through Injection of Supercritical CO₂

Abstract

The recovery of natural gas from CH₄-hydrate deposits in sub-marine and sub-permafrost environments through injection of CO₂ is considered a suitable strategy towards emission-neutral energy production. This study shows that the injection of hot, supercritical CO₂ is particularly promising. The addition of heat triggers the dissociation of CH₄-hydrate while the CO₂, once thermally equilibrated, reacts with the pore water and is retained in the reservoir as immobile CO₂-hydrate. Furthermore, optimal reservoir conditions of pressure and temperature are constrained. Experiments were conducted in a high-pressure flow-through reactor at different sediment temperatures (2 °C, 8 °C, 10 °C) and hydrostatic pressures (8 MPa, 13 MPa). The efficiency of both, CH₄ production and CO₂ retention is best at 8 °C, 13 MPa. Here, both CO₂- and CH₄-hydrate as well as mixed hydrates can form. At 2 °C, the production process was less effective due to congestion of transport pathways through the sediment by rapidly forming CO₂-hydrate. In contrast, at 10 °C CH₄ production suffered from local increases in permeability and fast breakthrough of the injection fluid, thereby confining the accessibility to the CH₄ pool to only the most prominent fluid channels. Mass and volume balancing of the collected gas and fluid stream identified gas mobilization as equally important process parameter in addition to the rates of methane hydrate dissociation and hydrate conversion. Thus, the combination of heat supply and CO₂ injection in one supercritical phase helps to overcome the mass transfer limitations usually observed in experiments with cold liquid or gaseous CO₂.

1. Introduction

Large amounts of natural gas, predominantly methane, are stored in gas hydrates in sediments below the seafloor and the permafrost [1]. Current estimates of the global methane hydrate inventory range between 1000 and 10,000 Gt of carbon [2,3,4,5]. Motivated by these results, gas hydrate research activities worldwide center around the exploitation of this potential new energy resource. The methods that are currently discussed to produce the methane from gas hydrates are generally derived from standard techniques used in conventional oil and gas business, i.e., reduction of the pressure in the reservoir and thermal stimulation, as well as injection of hydrate inhibitors, such as salt, to induce dissociation of the gas hydrates [6]. In addition, the substitution of CH₄ by CO₂ as guest molecule in the gas hydrate structure has been proposed as a more elegant production technology with respect to greenhouse gas policies [7,8].

All of these methods have been studied in laboratory experiments to validate their feasibility as well as in numerical simulations to gain first ideas about their applicability on reservoir scale. First production tests were carried out in the permafrost reservoir of Mallik in northern Canada in 2002 [9] as well as in 2008 [10]. Gas hydrates were successfully destabilized by injection of hot water and by depressurization, respectively, producing limited amounts of CH₄ gas over a few days. Further field trials in 2012 will test the chemical exchange of methane in gas hydrates by injection of CO₂ below the permafrost of the Alaska North Slope [11] and the depressurization technique in the first offshore test in the Nankai Trough [12,13].

Overall, the conclusions drawn from those studies are that thermal stimulation by injecting hot water is slow and inefficient, whereas depressurization seems to be the more promising strategy [6]. However, due to the endothermic nature of gas hydrate dissociation, in the long run, the reservoir will cool down, re-establishing stable conditions for gas hydrates and consequently, methane production rates are expected to cease after some time [14,15].

Thus, being able to achieve stable and economic methane production rates will require a combination of depressurization and methods (re)activating the methane hydrate reservoir.

One elegant way to activate the methane hydrate reservoir is the injection of CO₂. Since CO₂ hydrate is thermodynamically more stable than CH₄-hydrate and both form structure-I, the exchange reaction will proceed exothermically [16], adding heat to the system. Besides its attractiveness in combining energy production with CO₂ storage as a measure to mitigate further increases in greenhouse gas emissions to the atmosphere, a technological advantage is that it sustains the integrity and geomechanical stability of the sediments, thus reducing the potential risk of slope failures.

Several laboratory-based studies have shown the feasibility of the hydrate conversion reaction on pure or sediment-dispersed gas hydrates and from molecular scale to volumes of a few liters (an overview is given in Discussion). However, the overall reaction rate turns out to be quite sluggish and the conversion is often incomplete. This results primarily from a shell of CO₂-hydrate that forms around the methane hydrate grain. Any further mass transport of CO₂ into and CH₄ out of the inner core is drastically slowed down or even completely inhibited as analogous experiments with C₂H₆ as attacking guest molecules have shown [17]. Moreover, excess water, as usually encountered in the pore space of hydrate deposits, generally induces immediate CO₂-hydrate formation blocking permeable pathways for the gas exchange.

Common to all those studies is that the chosen pressure-temperature conditions were either within the stability fields of CO₂- and CH₄-hydrate or only CO₂-hydrate. Hence, any attempt towards economic CH₄ production rates based on the CH₄-to-CO₂-hydrate conversion requires a different approach.

In the Ignik Sikumi Gas Hydrate Field Test (January-April 2012) [11] on the Alaska North Slope a 2-stage Huff’n’Puff-like procedure was, thus, followed. First a CO₂/N₂ gas mixture (~6000 m³ over 13 days) was injected into the CH₄-hydrate bearing sandstone. Subsequently, gas was produced by depressurization for 6 weeks with CH₄ contents dominating the produced gas mixture already after 2 days [18].

Here, we present a different strategy: in our experiments we have injected hot (95 °C), supercritical CO₂ into CH₄-hydrate bearing sand at typical marine p-/T-conditions. The hot CO₂ was injected in several discrete intervals, actively dissociating the CH₄-hydrate. Each injection was followed by a period of thermal and chemical equilibration of the system. The results are analyzed and presented in terms of CH₄ production rate and overall CH₄ yield, hydrate conversion efficiency, degree of CO₂ and H₂O retention in the system, as well as efficiency of the energy input for CH₄-hydrate dissociation. Partitioning of the components CO₂ and CH₄ into the different liquid, gaseous and hydrate phases is estimated and the results are compared in light of previous experiments which used cold CO₂. Finally, we conclude with perspectives for this hydrate exploitation strategy and outline the steps that are necessary to develop a successive exploitation strategy for marine gas.

2. Experimental Section

2.1. Preparation of Artificial Sediment and Seawater Medium

The sediment samples were prepared at −20 °C from a homogeneous mixture of quartz sand (grain size 0.1–0.6 mm, G20TEAS, Schlingmeier, Schwülper, Germany) and fine grained ice particles (grain size fraction 0.3–1.0 mm, deionized water). Individual sample compositions before and after CH₄-hydrate formation are summarized in Table 1. The sample mixture of quartz sand and ice was filled into a sample bag made of PTFE cloth and was placed inside a stainless steel pressure vessel which was cooled to −7 °C. The pressure vessel was then pressurized with CH₄ gas to 13 MPa. To accelerate hydrate formation, water availability was increased by freezing-thawing cycles from −7 °C to +2 °C in a procedure similar to [19]. The formation process was continuously monitored by logging the CH₄ gas pressure. Within 3–7 days, the conversion of water to hydrate was completed. Subsequently, the sample was brought to −5 °C before being de-pressurized to atmospheric pressure. System re-pressurization and water saturation of pore spaces was achieved by instant filling and re-pressurization with saltwater medium (S ≈ 31). The volumes of saltwater injected into the pressure vessel accounted for the total pore/cavity volumes and are listed in Table 1. Hydrate dissociation during the brief period of depressurization was minimized by taking advantage of the anomalous self-preservation effect, which reaches an optimum close to the chosen temperature [20]. Subsequently, the sample temperature was changed to match the individual requirement of the experiments. Saltwater medium was prepared according to [21] and was close to seawater composition.

Table 1. Sediment preparation and CH₄-hydrate formation. List of initial parameters for sediment and hydrate preparation.

Experiment		Quartz sand/g	Ice/g	Void volume/mL	CH4-hydrate produced from ice/mol
No.	Experimental conditions
1	2 °C, 13 MPa	1449	314	1112	3.03
2	8 °C, 13 MPa	1338	297	1172	2.87
3	10 °C, 13 MPa	1514	305	1097	2.95
4	8 °C, 7.5 MPa	1603	315	1053	3.04

Table 1. Sediment preparation and CH₄-hydrate formation. List of initial parameters for sediment and hydrate preparation.

Experiment		Quartz sand/g	Ice/g	Void volume/mL	CH4-hydrate produced from ice/mol
No.	Experimental conditions
1	2 °C, 13 MPa	1449	314	1112	3.03
2	8 °C, 13 MPa	1338	297	1172	2.87
3	10 °C, 13 MPa	1514	305	1097	2.95
4	8 °C, 7.5 MPa	1603	315	1053	3.04

2.2. Experimental Setup and Procedure

Experiments were carried out in a custom-made high pressure apparatus (NESSI, Natural Environment Simulator for Sub-seafloor Interactions). All wetted parts of the set-up are made of stainless steel. Saltwater medium was supplied from reservoir bottles (DURAN, Wertheim, Germany) using a HPLC pump S1122 (SYKAM, Fürstenfeldbruck, Germany), CO₂ was supplied with a piston pump (Teledyne ISCO, Lincoln NE, USA) and heated to 95 °C inside a temperature controlled conditioning chamber prior to injection into the sample vessel. Experiments were carried out in upflow mode with injection of CO₂ at the bottom of the sample vessel. Pressure, salinity and temperature were monitored in the influent and the effluent fluid streams. Flow control was achieved by the high-pressure pumps that supplied the fluids. Pressure was adjusted with a back-pressure regulator valve (TESCOM Europe, Selmsdorf, Germany) in line with a fine-regulating valve for compensation of pressure spikes (TESCOM Europe). All experiments were carried out at constant temperature conditions. Temperature control was achieved with a thermostat system (Huber, Offenburg, Germany). The experimental scheme is shown in Figure 1.

Four experiments were carried out at different pressure-temperature (p-/T-) conditions: 2 °C, 13 MPa; 8 °C, 13 MPa; 10 °C, 13 MPa and 8 °C, 8 MPa. The phase diagram in Figure 2 visualizes the experimental conditions with respect to the thermodynamic stability regimes of CH₄- and CO₂-hydrate.

At the beginning of each CO₂ injection experiment, the water-saturated sediment-hydrate sample was continuously percolated with saltwater medium at a flow rate of 1.0 mL∙min⁻¹. This initial water injection interval was performed to verify that the sample body was permeable and homogeneously pressurized. Injection of CO_2,sc was realized stepwise, in 4 to 6 injection intervals, simulating a sequential injection strategy as opposed to a continuous injection. During each injection interval, CO_2,sc was injected at constant flow rates of 2.5 to 5 mL∙min⁻¹.

In between injection intervals, the system pressure was maintained by injection of small amounts of CO₂ at constant pressure. This was necessary to compensate for volume changes due to cooling of the injected CO₂ and to phase changes. The time intervals between injection steps are referred to as equilibration intervals during which no effluent fluid was produced. The number of injection intervals was chosen to ensure at least matching ratios between the injected CO₂ volume and the volume potentially accessible to fluid within the pressure vessel. In most experiments, this was achieved after five injections. However, at 2 °C, 13 MPa, the experiment had to be aborted after the fourth injection due to substantial loss of hydraulic conductivity.

Figure 1. Experimental scheme. (a) Experiments were carried out in a high-pressure experimental system suitable for flow-through experimentation. Water and CO₂ were supplied from reservoir containers using suitable high-pressure delivery systems. The hydrate-sediment sample was prepared in a 2 L stainless steel pressure vessel inside an internal pouch made of PTFE cloth. Before entering the pressure vessel, the pressurized CO₂ was heated to 95 °C. The sample-temperature was controlled by circulating cooling fluid through a cooling jacket around the pressure vessel. Pressure, temperature, and conductivity (PTS) were measured discontinuously in the influent and effluent fluids. Bulk effluent fluids were de-pressurized and collected downstream in a 100 L gas sampling bag after de-pressurization; (b) Cross-section of the sample pressure vessel.

Figure 1. Experimental scheme. (a) Experiments were carried out in a high-pressure experimental system suitable for flow-through experimentation. Water and CO₂ were supplied from reservoir containers using suitable high-pressure delivery systems. The hydrate-sediment sample was prepared in a 2 L stainless steel pressure vessel inside an internal pouch made of PTFE cloth. Before entering the pressure vessel, the pressurized CO₂ was heated to 95 °C. The sample-temperature was controlled by circulating cooling fluid through a cooling jacket around the pressure vessel. Pressure, temperature, and conductivity (PTS) were measured discontinuously in the influent and effluent fluids. Bulk effluent fluids were de-pressurized and collected downstream in a 100 L gas sampling bag after de-pressurization; (b) Cross-section of the sample pressure vessel.

Figure 2. Experimental conditions, stability of CH₄- and CO₂-hydrates. Phase diagrams of the CH₄-H₂O (red) and the CO₂-H₂O (blue) system at S = 31 and pressure-temperature (p-/T-) conditions of the respective experiments. Phase diagrams were calculated according to ([22,23,24]). Experiment 1 was carried out under low-temperature conditions at 2 °C, 13 MPa, inside the stability region of both CH₄-hydrate and CO₂-hydrate. Experiment 2 was carried out at an increased sediment temperature of 8 °C and 13 MPa. Experiment 3 was carried out at 10 °C, 13 MPa inside the stability zone of pure CH₄-hydrate but outside of the stability region of pure CO₂-hydrate. Experiment 4 was carried out at the same temperature conditions as experiment 2 but at a reduced pressure of 8 MPa, close to the phase boundary of pure CH₄-hydrate.

Figure 2. Experimental conditions, stability of CH₄- and CO₂-hydrates. Phase diagrams of the CH₄-H₂O (red) and the CO₂-H₂O (blue) system at S = 31 and pressure-temperature (p-/T-) conditions of the respective experiments. Phase diagrams were calculated according to ([22,23,24]). Experiment 1 was carried out under low-temperature conditions at 2 °C, 13 MPa, inside the stability region of both CH₄-hydrate and CO₂-hydrate. Experiment 2 was carried out at an increased sediment temperature of 8 °C and 13 MPa. Experiment 3 was carried out at 10 °C, 13 MPa inside the stability zone of pure CH₄-hydrate but outside of the stability region of pure CO₂-hydrate. Experiment 4 was carried out at the same temperature conditions as experiment 2 but at a reduced pressure of 8 MPa, close to the phase boundary of pure CH₄-hydrate.

2.3. Fluid Sampling and Gas Analysis

Sub-samples for fluid composition monitoring were taken discontinuously from the influent and effluent streams using gas-tight glass syringes. Gas volumes were measured after expansion at atmospheric pressure. Bulk effluent fluids were collected inside 100 L gas tight TEDLAR sampling bags (CEL Scientific, Santa Fe Springs, CA, USA). The sampling bags were mounted inside water filled sampling containers. After expansion of the effluent fluids at atmospheric pressure, their volume was measured as volume of displaced water from the containers. Effluent gas analysis was carried out with gas chromatography.

2.4. Terms and Definitions

2.4.1. Components and Phases

At the beginning of each experiment, the system consists of three components CH₄, H₂O and quartz sand. All CH₄ is bound in a single phase as CH₄-hydrate (CH_4,hyd). H₂O is present as liquid salt water or as solid phase hydrate water. The quartz sand is assumed to remain chemically unchanged as an inert solid phase. Injection of supercritical CO₂ (CO_2,sc) introduces one additional component and three additional phases, CO₂-hydrate (CO_2,hyd), liquid CO₂ (CO_2,liq) and gaseous CH_4, (CH_4,g). Because of mutual solubilities of components, phase composition is somewhat more complicated, which is, however, only of minor importance for this study.

The formation of mixed CO₂-CH₄-hydrates of variable composition in our experiments is likely. However, in this study we were not able to determine the final composition of the gas hydrates at the end of the experiments. Moreover, the exact composition of the gas hydrates does not affect the calculated mass and volume balances and hence the interpretation of our results. Thus, for simplicity reasons, hereafter the term “CO₂-hydrate" is used for gas hydrates containing CO₂, regardless of the fact that they will likely also contain methane in variable amount. Mass flow analysis is exclusively based on component inventories, whereas phase composition is calculated based on volume balancing of influent and effluent fluids.

2.4.2. Calculation of Mass Balances

Mass balances were calculated from measurements of influent (subscript infl) and effluent (subscript effl) substance amounts n_CO2, n_CH4, and masses m_H2O. Mass balances are exclusively based on overall amounts of components, but do not consider phase distribution. Experimental results are presented as bulk sample inventories of components i.e., amounts of components inside the pressure vessel. Data evaluation is based on calculation of efficiency (E_CH4), yield (Y_CH4), and rate (r_CH4) of CH₄ production from CH₄-hydrate bearing sediments at different sediment temperatures and pressures, and efficiency (E_CO2, E_H2O) and amount of CO₂ and H₂O retention (S_CO2, S_H2O).

The usage of the term efficiency, (E) always relates to the cumulative amount of CO₂ injected (n_CO2,inj), the subscript init refers to the initial amount or mass of component:

$$E_{{\text{CH}}_4}=\frac{Y_{{\text{CH}}_{\text{4}}}}{{\displaystyle \sum n_{{\text{CO}}_{\text{2,infl}}}}}$$

(1)

$$E_{{\text{CO}}_{\text{2}}}=\frac{S_{{\text{CO}}_{\text{2}}}}{{\displaystyle \sum n_{{\text{CO}}_{\text{2,infl}}}}}$$

(2)

Yield, (Y) refers to the cumulative amount of produced components:

$$Y_{{\text{CH}}_{\text{4}}}={\displaystyle \sum n_{{\text{CH}}_{\text{4,effl}}}}$$

(3)

Retention, (S) is related to the amount of substances that remain in the sample vessel:

$$S_{{\text{CO}}_{\text{2}}}={\displaystyle \sum \left(n_{{\text{CO}}_{\text{2infl}}}-n_{{\text{CO}}_{\text{2,effll}}}\right)}$$

(4)

$$S_{H_{2}O}={\displaystyle \sum \left(m{H}_2{O}_{init}-m{H}_2{O}_{effl}\right)}$$

(5)

We use the term CO₂ retention rather than CO₂ storage to describe the time-dependent accumulation of CO₂ in the sample matrix. This takes into account that on the time scale of the experiments the conversion to CO₂-hydrate is not complete and CO₂ remains partly mobile.

2.4.3. Calculation of Volume Balances

Evidence of phase changes, i.e., dissociation of CH_4,hyd, with release of CH_4,g and formation of CH_4,hyd or CO_2,hyd was derived from a comparison of the influent and effluent fluid volumes, which were converted to account for the thermodynamic conditions inside the sample vessel (subscript PT).

The release of CH_4,g from CH_4,hyd is the only process which can account for substantial positive fluid volume differences (V_diff,rel) with larger bulk effluent fluid volumes compared to influent CO₂ volumes. CH₄ release summarizes the volume of produced CH₄ (V_CH4,effl) and the volume of CH_4,g retained in the pressure vessel according to Equation 6. V_CO2 and V_H2O are symbols for volumes of CO₂ and H₂O:

$$V_{\text{diff,rel}}=V_{{\text{CO}}_{\text{2,}}\text{effl,PT}}+V_{{\text{CH}}_{\text{4,}}\text{effl,PT}}+V_{{\text{H}}_{\text{2}}\text{O,effl}}-V_{{\text{CO}}_{\text{2,}}\text{infl,PT}}$$

(6)

Similarly, negative volume differences (V_diff,for) in the converted influent and effluent volumes were interpreted as volume consumption of CO_2,liq or CH_4,g due to hydrate formation. Hydrate formation capacity is calculated based on Equation (7) rather than on Equation (6), which takes into account that CH₄ production from CH_4,hyd does not induce a relevant volume difference inside the pressure vessel:

$$V_{\text{diff,for}}=V_{{\text{CO}}_{\text{2,}}\text{effl,PT}}+V_{{\text{H}}_{\text{2}}\text{O,effl}}-V_{{\text{CO}}_{\text{2,}}\text{infl,PT}}$$

(7)

Volume differences caused by gas dissolution were of minor importance for this study. However, since particularly dissolution of CH₄ in CO_2,liq and dissolution of CO₂ in H₂O could become relevant in short term experiments, the maximum error related to our results is estimated in the Results section. Quantitative evaluations of CH₄ release and hydrate formation are always in accordance with inventory limits of components.

2.4.4. Calculation of Energy Efficiency

We use a simplified energy ratio to estimate energy efficiency, E_∆H. The energy efficiency relates the theoretical energy input for CH₄-hydrate dissociation to the thermal energy input from injection of hot CO_2,sc:

$$E_{\Delta \text{H}}=\frac{m_{{\text{CH}}_{\text{4,hyd}}}\cdot \left(C_{p{\text{,CH}}_{\text{4}}-\text{H}}(T_{\text{diss}}-T)+\Delta H_{\text{diss}}\right)}{m_{{\text{CO}}_{\text{2,infl}}}\cdot \left(\Delta H_{\text{f,T,p}}-\Delta H_{\text{f,95}°\text{C,p}}\right)}\cdot 100$$

(8)

We again carefully distinguish between CH₄ production and CH₄ release and calculate energy efficiencies for produced CH₄ (E_∆H,prod, based on mass balance analysis to calculate the mass of hydrate to match produced CH₄) and released CH₄ (E_∆H,rel, based on analysis of fluid volume differences to calculate the mass of hydrate to match released CH₄). The meaning of further symbols in Equation 8 is as follows: C_p,CH4-hyd: Specific isobaric heat capacity [25], T_diss: dissociation temperature of CH₄-hydrate, T: experimental temperature, ∆H_diss: enthalpy of CH₄-hydrate dissociation, m_CO2,infl: mass of injected CO₂, ∆H_f,T,p: enthalpy of formation at experimental pressure and temperature, ∆H_{f,95 °C,p}: enthalpy of formation at 95 °C and experimental pressure.

3. Results

3.1. Analysis of Mass Balances

3.1.1. CH₄ Inventory and CO₂ Inflow during Injection and Equilibration Periods

Mass balances were calculated from measured compositions of influent and effluent fluid streams. Figure 3 shows the time-dependent CH₄ inventory and the cumulative amount of injected CO₂ during the whole experimental period. The bulk sample composition was altered considerably by production of CH₄ and retention of CO₂ during injection/production intervals, but was only slightly changed during equilibration intervals as a result of pressure-controlled slow injection of CO₂.

CH₄ production was clearly dependent on sediment temperature and hydrostatic pressure. Between 2% and 35% of the initial amount of CH₄ fixed in hydrate was produced after stepwise injection of 22 mol and 21 mol CO₂ (Figure 3a at 2 °C, 13 MPa and 3d at 8 °C, 8 MPa, respectively). From the graphs of effluent fluid pressure in Figure 3, it is apparent that pressure is not constant during injection periods. Pressure oscillations are partly due to the backpressure regulation mechanism where pressure is discontinuously released by a valve, or due to accidental freezing of regulators after cooling from adiabatic gas expansion. However, exceptionally large pressure spikes were observed in Experiment 1 at 2 °C, 13 MPa. These pressure spikes occurred after sample permeability decreased drastically, suggesting rapid CO₂-hydrate formation. Permeability could only be restored after sample fracturing with very high differential pressures >15 MPa. Sample blocking became irreversible by experimental means at the end of Experiment 1.

Figure 3. Inventory of CH₄, cumulative amount of injected CO₂ and sample effluent pressure (p_effl) in each of the experimental runs. (a) 2 °C, 13 MPa; (b) 8 °C, 13 MPa; (c) 10 °C, 13 MPa; (d) 8 °C, 8 MPa. It should be noted, that the axis scaling for p_effl is different for every experiment. CO₂ was injected stepwise as hot supercritical CO_2,sc and at a constant flow rate (steep gradients in dashed lines). Injection of CO₂ at constant pressure and low flow rate alternated with equilibration periods. Fluids including CH₄ could only exit the system during injection/production periods. CH₄ production is shown as a decrease of the CH₄ inventory. Effluent fluid pressure (p_effl) is constant during equilibration periods. Strong aberrations indicate technical problems, which could be solved without destabilization of the hydrate sample. However, p_effl is unstable during injection periods. Massive pressure spikes were observed in Experiment 1 (2 °C, 13 MPa) after sample permeability changed due to CO₂-hydrate formation. Sample permeability could be restored temporarily by fracturing at high differential pressures (>15 MPa). CO₂-hydrate formation eventually led to complete congestion of the fluid pathways at the end of the experiment. Sample blocking was not observed in other experiments at higher temperatures.

Figure 3. Inventory of CH₄, cumulative amount of injected CO₂ and sample effluent pressure (p_effl) in each of the experimental runs. (a) 2 °C, 13 MPa; (b) 8 °C, 13 MPa; (c) 10 °C, 13 MPa; (d) 8 °C, 8 MPa. It should be noted, that the axis scaling for p_effl is different for every experiment. CO₂ was injected stepwise as hot supercritical CO_2,sc and at a constant flow rate (steep gradients in dashed lines). Injection of CO₂ at constant pressure and low flow rate alternated with equilibration periods. Fluids including CH₄ could only exit the system during injection/production periods. CH₄ production is shown as a decrease of the CH₄ inventory. Effluent fluid pressure (p_effl) is constant during equilibration periods. Strong aberrations indicate technical problems, which could be solved without destabilization of the hydrate sample. However, p_effl is unstable during injection periods. Massive pressure spikes were observed in Experiment 1 (2 °C, 13 MPa) after sample permeability changed due to CO₂-hydrate formation. Sample permeability could be restored temporarily by fracturing at high differential pressures (>15 MPa). CO₂-hydrate formation eventually led to complete congestion of the fluid pathways at the end of the experiment. Sample blocking was not observed in other experiments at higher temperatures.

3.1.2. CH₄ Production

Generally, data in this study are presented as amounts of substance or masses of the components CH₄, CO₂ and H₂O which were retained inside the pressure vessel, i.e., the development of the inventory of the respective components in the bulk sample, as a function of the amount of injected CO₂. This approach allows for easy comparison of CH₄ production efficiencies as shown in Figure 4. Changes in inventory amounts are calculated by balancing substance amounts of influent and effluent fluids. Results in Table 2 indicate that CH₄ concentration in the overall effluent gas was comparatively high during early injection steps. However, only small amounts of bulk effluent gas consisting of both CH₄ and CO₂ were produced. During later injection steps, the produced fluid volume is much larger with high amounts of CO₂ contributing to the effluent gas. Therefore, the absolute amounts of produced CH₄ are increased while the relative amounts of CH₄ are decreased. We found that CH₄ production from CO₂ injection is most efficient at 8 °C, while the gas recovery was apparently independent of pressure. CH₄ recovery at 2 °C and 10 °C was both impaired compared to experiments at 8 °C. The experiment at 8 °C, 13 MPa was terminated after the injection of 27.5 moles of CO₂. At this point, 31% of the available hydrate reservoir was depleted and produced as effluent CH₄. Particularly at 8 °C, CH₄ production showed a linear trend with respect to CO₂ injection, which indicates that CH₄ from the CH₄ inventory can be produced with constant efficiency using CO_2,sc. The experiments were stopped before reaching a maximum production yield after the volume of injected CO₂ was approximately equal to the sample void volume.

Figure 4. CH₄ production. CH₄ sample inventory during injection of CO₂. Low CH₄ production efficiency and yield was observed at 2 °C, 13 MPa ($▲$) and 10 °C, 13 MPa ($♦$). CH₄ production efficiency and yield were highest at 8 °C (both runs).

Figure 4. CH₄ production. CH₄ sample inventory during injection of CO₂. Low CH₄ production efficiency and yield was observed at 2 °C, 13 MPa ($▲$) and 10 °C, 13 MPa ($♦$). CH₄ production efficiency and yield were highest at 8 °C (both runs).

During the initial saltwater percolation period prior to CO₂ injection (data not shown), only minor amounts of CH₄ were produced with the effluent water. The continuously injected saltwater is undersaturated with respect to CH₄ and therefore stimulates the dissociation of CH₄-hydrates until saturation is reached. The occurrence of this effect indicates a satisfactory surface contact between the percolating liquids and the hydrate. However, the production of free CH_4,g was not observed during the saltwater percolation period.

Table 2. Effluent amounts of CH₄ and CO₂. Amounts of CH₄ and CO₂ recovered in the effluent fluid for each injection step. The third column for each injection step shows the CH₄ content in the effluent gas consisting of only CH₄ and CO₂. Although the amounts of effluent CH₄ increase, the CH₄ content is low because of large amounts of effluent CO₂

Experimental Conditions	Effluent gas composition
	Inj. Step 1			Inj. Step 2			Inj. Step 3			Inj. Step 4			Inj. Step 5
	n (mol)		CH4 /% (v/v)	n (mol)		CH4 /% (v/v)	n (mol)		CH4 /% (v/v)	n (mol)		CH4 /% (v/v)	n (mol)		CH4 /% (v/v)
	CH4	CO2		CH4	CO2		CH4	CO2		CH4	CO2		CH4	CO2
2 °C, 13 MPa	0.004	0.02	17.3	0.022	1.16	1.8	0.011	1.44	0.7	0.034	2.21	1.5	/	/	/
8 °C, 13 MPa	0.009	0.02	29.3	0.019	0.01	65.2	0.317	3.33	8.7	0.237	4.09	5.5	0.255	4.05	5.9
10 °C, 13 MPa	0.142	1.45	8.9	0.175	2.81	5.9	0.079	2.71	2.8	0.044	2.86	1.5	0.039	2.55	1.5
8 °C, 8 MPa	0.070	0.05	56.2	0.231	1.50	13.3	0.162	2.3	6.5	0.221	2.4	8.4	0.227	2.7	7.7

Table 2. Effluent amounts of CH₄ and CO₂. Amounts of CH₄ and CO₂ recovered in the effluent fluid for each injection step. The third column for each injection step shows the CH₄ content in the effluent gas consisting of only CH₄ and CO₂. Although the amounts of effluent CH₄ increase, the CH₄ content is low because of large amounts of effluent CO₂

Experimental Conditions	Effluent gas composition
	Inj. Step 1			Inj. Step 2			Inj. Step 3			Inj. Step 4			Inj. Step 5
	n (mol)		CH4 /% (v/v)	n (mol)		CH4 /% (v/v)	n (mol)		CH4 /% (v/v)	n (mol)		CH4 /% (v/v)	n (mol)		CH4 /% (v/v)
	CH4	CO2		CH4	CO2		CH4	CO2		CH4	CO2		CH4	CO2
2 °C, 13 MPa	0.004	0.02	17.3	0.022	1.16	1.8	0.011	1.44	0.7	0.034	2.21	1.5	/	/	/
8 °C, 13 MPa	0.009	0.02	29.3	0.019	0.01	65.2	0.317	3.33	8.7	0.237	4.09	5.5	0.255	4.05	5.9
10 °C, 13 MPa	0.142	1.45	8.9	0.175	2.81	5.9	0.079	2.71	2.8	0.044	2.86	1.5	0.039	2.55	1.5
8 °C, 8 MPa	0.070	0.05	56.2	0.231	1.50	13.3	0.162	2.3	6.5	0.221	2.4	8.4	0.227	2.7	7.7

3.1.3. CO₂ and H₂O Retention

In addition to CH₄ production, the efficient retention of CO₂ and H₂O is important for the overall process. As shown in Figure 5, CO₂ retention was most efficient at 2 °C, 13 MPa, where the thermodynamic conditions were most favorable for CO₂-hydrate formation. Increasing amounts of CO₂ were retained over the entire experimental period with a linear increase in the CO₂ inventory until the experiment was stopped after loss of hydraulic conductivity. Also in experiments at higher temperatures the maximum CO₂ retention capacity was not reached until the end of the experiments. However, CO₂ retention efficiency was lower than at 2 °C. Similar to CH₄ production, differences in CO₂ retention were minor in experiments at 8 °C at different pressures. While the retention during the first injection steps was comparable to the retention at 2 °C, it was less efficient during later injection steps (n_CO2,inj > 10 mol) where substantial amounts of the injected CO₂ were produced in the effluent gas stream. At 10 °C, 13 MPa, i.e., slightly above the CO₂-hydrate dissociation temperature, CO₂ breakthrough occurred almost immediately after injection.

In contrast to CH₄ production and CO₂ retention, the efficiency of water retention was similar in all experiments. Approximately 50% of the initial water was retained in the pressure vessel at all temperatures and pressures (Figure 6).

Figure 5. CO₂-retention. CO₂ sample inventory during injection of CO₂. CO₂ retention efficiency was highest at 2 °C, 13 MPa ($▲$). Rapid CO₂ breakthrough was observed at 10 °C, 13 MPa ($♦$). CO₂ retention efficiency at 8 °C (■, 13 MPa and ○, 8 MPa) was dependent on pressure conditions.

Figure 5. CO₂-retention. CO₂ sample inventory during injection of CO₂. CO₂ retention efficiency was highest at 2 °C, 13 MPa ($▲$). Rapid CO₂ breakthrough was observed at 10 °C, 13 MPa ($♦$). CO₂ retention efficiency at 8 °C (■, 13 MPa and ○, 8 MPa) was dependent on pressure conditions.

Figure 6. Water retention. H₂O sample inventory during injection of CO₂. Retention of pore water was similar for all experiments with approximately 50% of initial water content (pore water plus hydrate water) being removed.

Figure 6. Water retention. H₂O sample inventory during injection of CO₂. Retention of pore water was similar for all experiments with approximately 50% of initial water content (pore water plus hydrate water) being removed.

Typical conductivity values from discontinuous measurements of bulk effluent water samples were 10.3 ± 3.5% below influent saltwater conductivities, indicating partial freshening of the pore water by hydrate water. A small number of samples showed exceptionally low conductivity values compared to influent saltwater medium values (injection step 2 at 10 °C, 13 MPa: 16% of saltwater conductivity, injection step 4 at 8 °C, 8 MPa: 67% of saltwater conductivity). These substantial differences could be due to errors during sampling. However, in accordance to volume balance calculations in the next paragraph they are probably due to distinct events of discharge of low salinity hydrate water released from hydrate dissociation. Conductivity measurements were not used to quantify hydrate dissociation inside the reactor since it was not possible to adequately calculate effects of mixing and transport of H₂O in the pressure reactor. Additionally, amounts of discharged H₂O were relatively small and could be easily contaminated by leakage of water from the sampling container into the sampling bag. Because of such a leak in the sampling bag, water inventory values for the two last injection steps at 8 °C, 13 MPa could not be calculated.

3.1.4. Investigation of Sampling and Analytical Error

Error in results was investigated from mass balancing after release of fluids from the high-pressure vessel at the end of Experiments 3 and 4. However, mass balance analysis was not carried out in Experiments 1 and 2, since we have attempted to sample undisturbed hydrates at the end of each experiment instead. While fluid release for mass balancing had to be performed slowly, hydrate sampling has to be performed fast and demanded rapid release of remaining pressurized fluids.

However, depressurization and bulk effluent sampling after completion of the experiment by stepwise releasing large amounts of CO_2,liq via the manually adjusted backpressure regulator was very problematic. The instant expansion of the CO₂, especially after lowering the backpressure, led to pressure spikes in the sampling lines and bag due to flow restrictions in the tubing which had a length of several meters. This caused substantial leakage and extensive loss of effluent gas, which could be visually observed. Pressure spikes were even more distinct and severe due to frequent freezing of the valves from adiabatic gas expansion and subsequent expulsion of CO₂ after required additional lowering of the backpressure settings and warming of the frozen valve and tubing. However, this effluent gas loss during final depressurization was of minor importance during the experiment because of much lower flow rates and no adjustment of the backpressure regulator being necessary.

Because of these problems with depressurization, final recovery of CO₂ and CH₄ showed some deviation with respect to amounts of injected CO₂ or initial CH₄ amounts. The error in CO₂ recovery was −6.5 mol CO₂ (30% from total injected CO₂) in experiment 3, and −3.4 mol CO₂ (16% from total injected CO₂, in experiment 4). Mass balance deviations for CH₄ were −1.2 mol (40% from initial CH₄) in experiment 3 and 0.4 mol (14% from initial CH₄) in experiment 4. However, water recovery was excellent in both experiments with a cumulative relative error of < 2%.

3.2. Analysis of Volume Balances

While mass balances allow for the calculation of the quantities of different components inside the sample vessel, they do not provide any information on the distribution of the components among the liquid, gas and hydrate phases. Influent and effluent fluid volumes corrected to experimental pressure and temperature were analyzed as a measure to further constrain the amounts of CH_4,g and to evaluate the reservoir size of CH₄- and CO₂-hydrate. Effluent volumes include the cumulative amounts of CH₄, CO₂ and H₂O in the production fluid; the influent volume is exclusively composed of injected CO₂.

CH₄ released in Figure 7 summarizes the amount of CH₄ produced (red areas) and CH_4,g retained in the pressure vessel (pink areas) after hydrate dissociation. Release of CH₄ from dissociating hydrates causes V_diff,rel to positively deviate from the neutral volume balance in Figure 7. However, V_diff,for used for estimating hydrate formation (shown as blue areas in Figure 7), does not consider CH₄ production in the volume balance, thus both hydrate formation and CH_4,g accumulation are shown as simultaneous events in Figure 7. The exclusion of CH_4,effl from the volume balance is motivated from considering that CH₄-hydrate dissociation releases CH_4,g and H₂O. Thus, CH₄ production from hydrates does not induce a relevant volume difference inside the pressure vessel. In contrast, CH₄ production from CH_4,g induces a volume difference, and hydrate formation capacity estimated from fluid volume balancing would be lowered accordingly. Since it is currently not possible to distinguish between the possible sources of produced CH₄, we assume that all produced CH₄ originates from hydrate dissociation.

Figure 7. Fluid volume balancing and evaluation of phase distribution. Deviations from even volume balances [grey dashed lines in (a)–(d)] are shown as colored areas related to the cumulative volume of injected CO₂. The figure emphasizes the difference between CH₄ production (red areas) and CH₄ release (pink areas plus red areas). The volume based hydrate formation potential was derived from negative volume differences (blue areas). Results are presented for single experiments (a) 2 °C, 13 MPa; (b) 8 °C, 13 MPa; (c) 10 °C, 13 MPa; (d) 8 °C, 8 MPa. Volume balances were calculated from inflow amounts of CO₂ and effluent amounts of CO₂, CH₄ and H₂O for each injection step. During early injection steps most positive deviations were observed with largest net deviations for 10 °C, 13 MPa (c). Distinct negative deviations were observed in the experiments at 2 °C and 10 °C during later injection steps. Positive deviations are considered to be entirely due to CH₄ gas release from CH₄-hydrates, whereas negative deviations are considered to be due to hydrate formation of either CO₂-, CH₄- or mixed CO₂-CH₄-hydrates.

Figure 7. Fluid volume balancing and evaluation of phase distribution. Deviations from even volume balances [grey dashed lines in (a)–(d)] are shown as colored areas related to the cumulative volume of injected CO₂. The figure emphasizes the difference between CH₄ production (red areas) and CH₄ release (pink areas plus red areas). The volume based hydrate formation potential was derived from negative volume differences (blue areas). Results are presented for single experiments (a) 2 °C, 13 MPa; (b) 8 °C, 13 MPa; (c) 10 °C, 13 MPa; (d) 8 °C, 8 MPa. Volume balances were calculated from inflow amounts of CO₂ and effluent amounts of CO₂, CH₄ and H₂O for each injection step. During early injection steps most positive deviations were observed with largest net deviations for 10 °C, 13 MPa (c). Distinct negative deviations were observed in the experiments at 2 °C and 10 °C during later injection steps. Positive deviations are considered to be entirely due to CH₄ gas release from CH₄-hydrates, whereas negative deviations are considered to be due to hydrate formation of either CO₂-, CH₄- or mixed CO₂-CH₄-hydrates.

Overall V_diff,rel revealed different tendencies with respect to p-/T-conditions. Only in Experiment 1 at 2 °C, 13 MPa overall V_diff,rel was negative (−115 mL, 12% of volume of injected CO₂). All other experiments revealed positive volume differences. At 8 °C, 8 MPa V_diff,rel was +246 mL (24% of volume of injected CO₂) and at 10 °C, 13 MPa V_diff,rel was +182 mL (18% of volume of injected CO₂). In contrast, the overall fluid volume balance of +31 mL (3% of volume of injected CO₂) was small, but still positive at 8 °C, 13 MPa. Experiments at 2 °C and 10 °C showed a trend to positive volume change values during early injection steps, whereas negative values were observed mainly during the late injection steps. Such a trend was not observed in the experiments at 8 °C. CH₄ release was clearly exceeding CH₄ production in experiment 3, but also in Experiment 1, volume differences indicated transient CH₄ release without CH₄ production. The smallest apparent differences between volumes of produced CH₄ and released CH₄ are observed in experiments at 8 °C.

Volume differences caused by component dissolution were neglected here, however, in all cases of dissolution of a pure component an error is introduced in the volume analysis. Particularly dissolution of CH₄ in CO_2,liq and dissolution of CO₂ in H₂O could be significant. The saturation concentration of CO₂ in an aqueous solution at 10 °C and 13 MPa is approximately 1.6 mol/kg. Dissolution of 1.6 mol of CO₂ of ρ ≈ 942 kg/m³ in 1 kg of saltwater ρ ≈ 1030 kg/m³ produces a brine of ρ ≈ 1046 kg/m³. The volume difference between the resulting brine and the initial two phase liquids amounts to approximately −23 mL that are not accounted for in the volume balance calculations. Only few data are available for estimating the solubility of CH₄ in CO_2,liq and the possible error from this phase transfer cannot easily be quantified. Further, the solubility of H₂O in CO₂ and CH₄ phases is of minor importance for volume balancing. However, please note that all phase transfers resulting from mutual solubilities of components would induce a negative volume difference and would finally result in an underestimation of CH₄ release. CH₄ release estimates in this study are therefore minimum values.

3.3. Analysis of Energy Efficiencies

Energy balancing evaluates the efficiency of thermal stimulation of hydrate dissociation. It complements the results from mass and volume balances carried out in the previous paragraphs. We use a simplified energy ratio Equation 8 to estimate energy efficiency of CH₄ production E_∆H,prod and CH₄ release E_∆H,rel.

Figure 8. Energy efficiency. (a) Cumulative energy efficiency of gas production (E_∆H,prod) was calculated from free enthalpy changes comparing effluent CH₄ gas recovery with theoretical CH₄ release from hydrate dissociation due to injection of supercritical CO₂. Lowest E_∆H,prod was observed at 2 °C, 13 MPa ($▲$). E_∆H,prod at 10 °C, 13 MPa ($♦$) peaked at 11.3%, but decreased with ongoing CO₂ injection. E_∆H,prod was highest at 8 °C, independent of hydrostatic pressure. The energy efficiency calculated from CH₄ production is compared to the energy efficiency calculated from CH₄ gas release (E_∆H,rel); (b, please note the difference in axis scales). Minor differences were observed in experiments at temperatures 8 °C. However, E_∆H,rel at 10 °C is close to optimum ($♦$ in (b)), whereas energy efficiency of CH₄ production is only 11.3% ($♦$ in (a)). Also at 2 °C, E_∆H,rel is considerably increased compared to E_∆H,prod. These results indicate that a significant amount of CH_4,g liberated from dissociating CH₄-hydrates is trapped and might eventually re-crystallize in secondary CH₄- or (CH₄-CO₂)-hydrates within the sediment.

Figure 8. Energy efficiency. (a) Cumulative energy efficiency of gas production (E_∆H,prod) was calculated from free enthalpy changes comparing effluent CH₄ gas recovery with theoretical CH₄ release from hydrate dissociation due to injection of supercritical CO₂. Lowest E_∆H,prod was observed at 2 °C, 13 MPa ($▲$). E_∆H,prod at 10 °C, 13 MPa ($♦$) peaked at 11.3%, but decreased with ongoing CO₂ injection. E_∆H,prod was highest at 8 °C, independent of hydrostatic pressure. The energy efficiency calculated from CH₄ production is compared to the energy efficiency calculated from CH₄ gas release (E_∆H,rel); (b, please note the difference in axis scales). Minor differences were observed in experiments at temperatures 8 °C. However, E_∆H,rel at 10 °C is close to optimum ($♦$ in (b)), whereas energy efficiency of CH₄ production is only 11.3% ($♦$ in (a)). Also at 2 °C, E_∆H,rel is considerably increased compared to E_∆H,prod. These results indicate that a significant amount of CH_4,g liberated from dissociating CH₄-hydrates is trapped and might eventually re-crystallize in secondary CH₄- or (CH₄-CO₂)-hydrates within the sediment.

The importance to distinguish between CH₄ production and CH₄ release becomes obvious from Figure 8. Figure 8a shows the cumulative E_∆H,prod based on CH₄ production yields. While experiments at 8 °C show reasonable energy efficiencies (19% at 13 MPa, and 18% at 8 MPa), energy recovery was apparently poor at 2 °C, 13 MPa (2%) and 10 °C, 13 MPa (8%). In contrast, cumulative E_∆H,rel based on apparent CH₄ release estimated from fluid volume differences was considerably higher, with maximum efficiency values close to 100% for early injection steps at 10 °C. Although the energy efficiency of CH₄ release due to CO₂-injection was close to optimum values for the highest sediment temperature (10 °C), this was not reflected in CH₄ production, which was relatively inefficient compared to experiments at lower temperatures. A similar trend was also observed in Experiment 1 at 2 °C, however, E_∆H,rel was considerably lower compared to the experiment at 10 °C. The increase in cumulative energy efficiency E_∆H,rel between injection steps 1 and 2 at 10 °C can be explained by a very short equilibration time (0.07 d), which was too short for thermal equilibration of the sample.

4. Discussion

4.1. Effect of Reservoir Temperature

Experiments were conducted at three different reactor temperatures (2, 8 and 10 °C) reflecting typical gas hydrate reservoir temperatures found in the natural environment. As anticipated, at 2 °C CO₂-hydrates formed with the excess pore water rapidly reducing the permeability of the sand matrix and thus, blocking available flow pathways. Applying higher differential pressure enabled fracturing of the sample and renewing fluid flow conditions for a few times, as observed by the high pressure spikes in Figure 3. However, after the fourth injection of CO₂, the experiment had to be aborted as further fracturing of the sample was not possible anymore because the maximum pressure limit of NESSI was reached. This may also illustrate the potential risk of clogging up the surrounding of any CO₂ injection well during CH₄-hydrate production operation under similar unfavorable environmental p-/T-conditions.

However, the observations made at higher reactor temperatures, particularly that CH₄ recovery at 8 °C was more efficient than at 10 °C, were unexpected. At an operating pressure of 13 MPa CO₂-hydrate may form at 8 °C, whereas liquid CO₂ and saltwater are the stable phases at 10 °C (Figure 2). Accordingly, the experimental result at 8 °C suggests that formation of CO₂-hydrate in the pore space at comparably moderate rates reduces the permeability locally, thereby diverging the fluid flow to new pathways through the sample. Hence, new regions of the sediment are accessed by the injected CO₂ and new CH₄-hydrates made available for dissociation and CH₄ release. Overall, this leads to a quite evenly distributed migration of the injected CO₂ through the quartz sand and an efficient production of hydrate-bound CH₄.

In contrast, at a reactor temperature of 10 °C the efficiency of the CH₄ gas recovery was much lower, because CO₂-hydrate could not form. As a consequence, permeability increased locally where the injected CO₂ dissociated the CH₄-hydrate, opening up a preferential flow pathway for the injected CO₂. Hence, the injected CO₂ did only affect a small region of the hydrate-bearing sediment and further CH₄-hydrate dissociation is controlled by lateral heat propagation away from the conduit and the respective CH₄ production is in addition also limited by its mass transport towards the channel.

The unexpected similarity of CH₄ production at 8 °C and 8 MPa or 13 MPa (Figure 4) indicated that sediment depressurization only had a minor effect on CH₄ production. However, depressurization and enhanced destabilization of initial CH₄-hydrate in the presence of CO₂ likely resulted in a CO₂-CH₄ gas mixture which could readily induce formation of mixed CO₂-CH₄-hydrates of modified stability relative to pure CO₂- or CH₄-hydrates [26]. Indeed, evidence for re-formation of hydrates with high CH₄ content was found from volume balancing (4.3).

4.2. Supercritical CO₂ as Injection Fluid

The injection of CO_2,sc for production of CH₄ from CH₄-hydrates was tested experimentally for the first time. The technical strategy of CH₄ production from hydrates by injection of CO_2,sc into the host sediment can be considered as a combination of using an injection fluid for chemical activation of the reservoir with additional thermal stimulation. As a result from this combination CH₄ production from hydrates is accelerated initially since CH₄ release is driven rather by thermal stimulation of hydrate dissociation, than by hydrate conversion. The dominance of heat induced hydrate dissociation over hydrate conversion can be clearly seen from constant CH₄ production efficiencies and the absence of apparent mass transfer limitations which would be expected in a diffusion-controlled conversion mechanism. However, the relevance of thermal hydrate dissociation as compared to hydrate conversion after thermal equilibration of the injected CO₂ with the reservoir needs to be studied in more detail. The relative contribution of thermal dissociation and mass transport controlled hydrate conversion will strongly influence the development of the injection strategy with regard to lengths of injection and equilibration periods.

As we have further shown in this study, heat injection and transport are crucial not only to CH₄-hydrate destabilization but also to flow assurance in the hydrate-sediment porous medium. While different heat injection strategies are potentially feasible, the injection of hot CO_2,sc appears to offer some benefits over cold injection of CO₂ combined with independent heat injection strategies, because the combined fluid and heat flow prevents uncontrolled CO₂-hydrate formation in the best possible way.

Calculations of energy efficiency of CH₄ production resulting from injection of hot CO_2,sc seemingly indicated substantial heat loss to secondary processes, which has raised questions towards the energetic feasibility of CO_2,sc injection. However, this production focused energy balance is somewhat misleading, since CH₄ release is apparently exceeding CH₄ production drastically. In that case, CH_4,g mobilization is the actual problem limiting CH₄ production, rather than energy efficiency. This aspect will be discussed in detail in the following paragraph.

While the primary effect of injection fluid temperature i.e., the direct influence on CO₂-hydrate formation, CH₄-release and production is not clear at the moment, this study provides strong evidence that the secondary effect of injection fluid temperature, which is increasing sediment temperature, is very important. Heat transport and sediment warming as a result of injection of CO_2,sc was not evaluated in this study. To stronger emphasize the relevance of this secondary effect of injection fluid temperature and also to better reflect reservoir conditions the experimental setup and procedure needs to be adapted to provide a spatially and temporally resolved temperature profile of the sample.

Although the combination with additional process techniques such as depressurization is promising, the CO₂ injection strategy itself is a central engineering tool for process optimization, and further modifications to the injection fluid might be of interest. An unsolved problem is the early breakthrough of CO₂ as observed in our experiments which results in a dilute CH₄ production fluid. The subsequent processing of the production fluid would be technically and economically demanding. As a further aspect, CH_4,g mobilization after hydrate dissociation, could potentially be improved by influencing the physical or chemical properties of the injection fluid.

4.3. CH₄ Production and Release

This study clearly reveals the importance to differentiate between CH₄ production and CH₄ release. Mass balance analyses indicated that using the tested injection scheme both CH₄ production and CO₂ retention are best at 8 °C. However, the recovered CH₄ gas only is a fraction of the CH₄ gas being released due to CH₄-hydrate dissociation. That this is true to various extents in the different experiments was concluded from fluid volume balancing, i.e., analysis of fluid volume differences. A positive V_diff,rel as calculated by Equation 6 is assumed to be due to dissociation of CH₄-hydrate and release of free gas. In contrast, a negative V_diff,for must be due to formation of CH₄-, CO₂- and mixed CH₄-CO₂-hydrate.

We observed strong differences with respect to fluid volume balances in the different experiments. In the experiment at 2 °C, 13 MPa, indicated by marked negative volume differences during later injection steps, a strong tendency towards hydrate formation was observed. This is clearly in accordance with results from mass flow analysis as well as with experimental observations showing the fatal loss of sample hydraulic conductivity. However, negative volume differences were preceded by positive net volume changes which indicated at substantial release of CH_4,g from hydrate dissociation. Since only minor CH₄ production was measured in the experiment at 2 °C, CH_4,g must have been retained inside the pressure vessel. Hydrate formation in this experiment was hence probably fueled from two reservoirs, CH_4,g and CO_2,liq. The substantial retention of injected CO₂ at 2 °C, CO_2,liq suggests that CO₂-hydrate is dominantly formed, however, the formation of mixed CO₂-CH₄-hydrates as well as the re-formation of CH₄-hydrates cannot be excluded.

In contrast to the experiment at 2 °C, fluid volume differences in the experiment at 10 °C were clearly dominated by CH₄-hydrate dissociation and release of CH_4,g, particularly in the early phase of the experiment. The high efficiency of CH₄ release with respect to thermal stimulation by CO_2,sc was clearly shown from energy efficiency calculations. However, the efficiency of CH₄, release was not reflected in CH₄ production and up to 2.1 mol CH_4,g which was released from hydrates during CO₂ injection apparently remained as immobile gas inside the sample pressure vessel. The fact that significant amounts of CH₄ remain within the sediment sample raises the question if and to what extent CH₄ recovery is limited by gas mobilization rather than by the release of CH₄ from hydrate. This is an aspect which needs to be investigated carefully in further studies since it influences process optimization strategies drastically. Similar to experiment 1 at 2 °C, volume differences at 10 °C reveal a transition towards hydrate formation during later injection steps. Since pure CO₂-hydrates are thermodynamically not stable under these conditions only the re-formation of CH₄-hydrate or the formation of mixed CH₄-CO₂-hydrates could account for these volume differences. The retention of CH_4,g followed by drastic re-formation of CH₄-hydrates is thus the reason for poor CH₄ production efficiencies at 10 °C.

Interestingly, the strong deviations between CH₄ production and CH₄ release were not observed in experiments at 8 °C. It is possible, that under these conditions CH₄-hydrate dissociation and CH_4,g release are well balanced by CO₂-hydrate formation and volume differences induced by the simultaneous processes are hidden. The absence of net fluid volume changes could indicate the simultaneous absence of excess CH₄ release and CO₂-hydrate formation as well as substantial excess CH₄ release balanced by CO₂-hydrate formation. Indeed, for all experiments, hydrate formation is assumed to be a continuous process and not, as suggested from volume balances, only occurring during later injection periods. Of course, if that is true, ongoing hydrate formation would need to be compensated by additional CH_4,g release and would further increase the discrepancy between CH₄ production and CH₄ release. It shall be emphasized that fluid volume balancing as carried out in this study only helps to distinguish between dominant processes of CH₄ release or hydrate formation as ratios, and as such, phase inventory values are not readily accessible.

The reason for discrepancies between CH₄ production and apparent CH₄ release are not clear yet. Interestingly, we observed similar trends for the experiment at 2 °C, which is putatively dominated by substantial CO₂-hydrate formation and the experiment at 10 °C, which excludes the formation of pure CO₂-hydrate. It appears possible, that both the fast passage of the CO₂ as observed at 10 °C hinders the efficient CH₄ production as well as slow passage at 2 °C in low permeability sediments dominated by rapid hydrate formation. At 10 °C, injected and rapidly flowing CO_2,sc cools via heat conduction and distant melting of CH₄-hydrates, thus producing inaccessible CH_4,g inclusions. At 2 °C this effect is possibly modulated and amplified by additional flow restrictions from CO₂-hydrate formation which could shield CH_4,g inclusions effectively.

4.4. Comparison of Hydrate Conversion Experiments

In our experiments injection times and subsequent equilibration periods were arbitrarily chosen and varied for several technical and practical reasons within and between experiment runs (Figure 3). The primary goal of our study was to prove the feasibility of our supercritical CO₂ technique under representative p-/T- conditions, but it was not intended to optimize its performance and efficiency—this must be left for future work. Therefore, calculating rates of hydrate conversion or methane production will give somewhat arbitrary numbers. However, since these are the first hydrate conversion experiments using supercritical CO₂, it seems necessary and appropriate to compare our results to those of previous studies from other groups where CH₄-hydrates were exposed to cold liquid or gaseous CO₂ (see Table 3).

In our experiments the highest CH₄ yields were achieved at 8 ºC, where 31%–36% of the hydrate-bound CH₄ was produced (Table 3) before those two experiments were, deliberately, aborted after 44 h and 73 h, respectively. At this point a relatively constant methane production had been established (Figure 3 and Figure 4). In contrast, the 10 ºC experiment achieved only a methane yield of 12.5% and shows a considerable decrease in CH₄ production after the second injection cycle. At 2 °C, the methane is quite constantly produced (Figure 3 and Figure 4), but with very low overall yield of only 2.4%.

Table 3. Hydrate conversion experiments.

CO2 phase	CH4 yield (%)	Time (h)	Volume (mL)	Sediment	Pressure (MPa)	Temperature (°C)	Hydrate stability	Reactor type	Reference
gas	100	4	90	no	3.5	3	CO2	batch	[27]
gas	27	280	130	no	3.25	0/(−2/+2)	CH4 + CO2	batch + flow − through	[28,29,30]
gas (N2/CO2)	85	20	no info	no	12	1	CH4 + CO2	batch	[31]
gas	100	150	laser spot	no	3.0	5	CO2	batch	[32]
gas	3	120	420	no	3.3	−2/0/+2	CH4 + CO2	batch	[33]
liquid	15	700	3200	no	3.9–4.5	1–3	CH4 + CO2	batch	[7]
liquid	50	5	no info	no	6	−3	CH4 + CO2	batch	[34]
liquid	35	280	130	no	5.4, 6.0	0	CH4 + CO2	batch	[30]
liquid	17	100	200	sand	5	8	CO2	batch	[35]
liquid	50	300	100	sandstone	8.3	4	CH4 + CO2	batch	[36,37,38]
emulsion with H2O	25	100	200	sand	5	8	CO2	batch	[39]
supercritical	40.7/37.5	44/76	2000	sand	13/7.5	8	CH4 + CO2	flow-through	This study
supercritical	10.7	73	2000	sand	13	10	CH4	flow-through	This study
supercritical	3.4	46	2000	sand	13	2	CH4 + CO2	flow-through	This study

Note: In the listed studies pure water as well as saline waters were used. Please refer to the references for more details.

Table 3. Hydrate conversion experiments.

CO2 phase	CH4 yield (%)	Time (h)	Volume (mL)	Sediment	Pressure (MPa)	Temperature (°C)	Hydrate stability	Reactor type	Reference
gas	100	4	90	no	3.5	3	CO2	batch	[27]
gas	27	280	130	no	3.25	0/(−2/+2)	CH4 + CO2	batch + flow − through	[28,29,30]
gas (N2/CO2)	85	20	no info	no	12	1	CH4 + CO2	batch	[31]
gas	100	150	laser spot	no	3.0	5	CO2	batch	[32]
gas	3	120	420	no	3.3	−2/0/+2	CH4 + CO2	batch	[33]
liquid	15	700	3200	no	3.9–4.5	1–3	CH4 + CO2	batch	[7]
liquid	50	5	no info	no	6	−3	CH4 + CO2	batch	[34]
liquid	35	280	130	no	5.4, 6.0	0	CH4 + CO2	batch	[30]
liquid	17	100	200	sand	5	8	CO2	batch	[35]
liquid	50	300	100	sandstone	8.3	4	CH4 + CO2	batch	[36,37,38]
emulsion with H2O	25	100	200	sand	5	8	CO2	batch	[39]
supercritical	40.7/37.5	44/76	2000	sand	13/7.5	8	CH4 + CO2	flow-through	This study
supercritical	10.7	73	2000	sand	13	10	CH4	flow-through	This study
supercritical	3.4	46	2000	sand	13	2	CH4 + CO2	flow-through	This study

Note: In the listed studies pure water as well as saline waters were used. Please refer to the references for more details.

As discussed before, this is due to immediate CO₂-hydrate formation with excess water in the pore space. A blocking of pore space by CO₂-hydrate formation resulting in low CH₄ yields was also encountered by Zunzhao et al. [33] and Hirohama et al. [7]. For similar reasons, Zhou et al. [39] could only achieve ~25% of hydrate conversion when applying a CO₂-H₂O emulsion.

Most of the other studies, summarized in Table 3, avoided excess water in their experiments. Nevertheless, the achieved methane yield varies over a wide range from similarly low values of 17%–27% [28,29,30,35] to almost 100% [27,31,32]. The main reason for this scatter is founded in very different experimental conditions, and as a consequence it is difficult to deduce an overall trend. Many of the experiments were conducted at p-/T- conditions close to the CH₄-water phase boundary and here it seems preferential if the injected CO₂ is gaseous [27,31,32] rather than liquid [35,39]. In two of those studies [27,32] the temperature was actually raised when introducing the CO₂ gas such that the pure CH₄-hydrates were no longer stable during the conversion reaction. Hence, the CH₄-hydrate was actively destabilized by adding heat to the system, i.e., in some way a similar strategy as in our experiments.

Experiments that were conducted at p-/T- conditions well within the stability fields of both gas hydrates [28,29,30,34,36,37,38] usually achieved only moderate hydrate conversion (Table 3). Two processes may be most likely to explain these findings: (1) Formation of mixed CH₄-CO₂-hydrates of varying composition and (2) a relatively quick gas exchange at the surface of the CH₄-hydrate grain forming a shell out of CO₂-hydrate and thereby inhibiting the further gas exchange, i.e., mass transport, into and out of the inner core of the hydrate grain. The latter process has been documented for the CH₄-C₂H₈ gas replacement [17] and seems very likely to occur in the above batch-type experiments. Since the overall process, particularly in a sediment/sand matrix, is quite complex, some other factors, such as local limitations in heat transfer and hydrate grain size, may also play a role (see Conclusion section for more details).

Thus, the major advantage of injecting supercritical CO₂ is the destruction of the original CH₄-hydrate excluding mass transfer limitations at hydrate grains. As a consequence, it is possible to establish a continuous production of CH₄ gas. Furthermore, immediate formation of the more stable CO₂-hydrates or mixed CO₂-CH₄-hydrates can be avoided at suitable reservoir temperatures potentially because the cooling of the CO₂ takes some time, and the liberated CH₄ gas warms up and migrates towards the outlet in the equilibration periods of our experiment.

5. Conclusions

In this study we presented the first experimental results on CH₄ production from CH₄-hydrates by injection of hot, supercritical CO₂. Based on the inventories of the components CO₂, CH₄, and H₂O (Table 4) the overall process performance was analyzed with respect to CH₄ production, CO₂ and H₂O retention as well as energy efficiency (Table 5). The central finding of this study is that continuous CH₄ production can be achieved, if some initial heat is introduced together with the CO₂. The hot, supercritical CO₂ efficiently activates the CH₄-hydrate reservoir and thereby overcomes mass transport limitations typically observed with cold CO₂. However, the reservoir temperature also plays a crucial role and definitely needs consideration when developing an exploitation strategy. At cold temperatures, rapid cooling of the injected CO₂ can induce formation of CO₂-hydrate with excess water causing congestions in the fluid pathways that may lead to complete process failure. We could further show that both, CH₄ production and CO₂ retention is improved under conditions of slow CO₂-hydrate formation. Constant CH₄ production efficiencies at all temperatures indicate that production is driven by fast thermal destabilization of CH₄-hydrates rather than by slow hydrate conversion. Surprisingly, the combination of mass flow analysis, volume and energy balancing suggests that CH₄ production is limited by gas mobilization rather than by CH₄-hydrate dissociation. Overall, it seems to be necessary to consider secondary processes for the overall CH₄ production process.

Table 4. Component inventories and phase distribution. Initial and final component and phase inventories are shown. Final amounts of components n_final were calculated from mass balance analysis. Amounts of gaseous CH₄ in the pressure vessel were estimated from volume differences. However, n_CH4,g is putatively underestimated in cases of rapid CO₂-hydrate formation, and overestimated if CH₄-hydrate reformation occurs. Final hydrate inventories n_CH4,hyd and n_CO2,hyd are calculated from component inventories and show maximum values.

No.	T, p	CH4				CO2				H2O
		nInitial	nFinal	nCH4,hyd	nCH4,g	nInitial	nFinal	nCO2,hyd	nCO2,inj	nInitial	nFinal
		/mol	/mol	/mol	/mol	/mol	/mol	/mol	/mol	/g	/g
1	2 °C, 13 MPa	2.9	2.8	2.1–2.8	<0.7	0	17.0	<7.7	21.9	1452	804
2	8 °C, 13 MPa	2.7	1.6	1.4–1.6	< 0.2	0	16.0	<8.3	27.5	1495	860
3	10 °C, 13 MPa	2.8	2.5	0.4–2.5	< 2.1	0	9.1	<8.4	21.6	1429	871
4	8 °C, 8 MPa	2.4	1.5	1.4–1.5	< 0.1	0	12.5	<7.5	21.5	1395	779

Table 4. Component inventories and phase distribution. Initial and final component and phase inventories are shown. Final amounts of components n_final were calculated from mass balance analysis. Amounts of gaseous CH₄ in the pressure vessel were estimated from volume differences. However, n_CH4,g is putatively underestimated in cases of rapid CO₂-hydrate formation, and overestimated if CH₄-hydrate reformation occurs. Final hydrate inventories n_CH4,hyd and n_CO2,hyd are calculated from component inventories and show maximum values.

No.	T, p	CH4				CO2				H2O
		nInitial	nFinal	nCH4,hyd	nCH4,g	nInitial	nFinal	nCO2,hyd	nCO2,inj	nInitial	nFinal
		/mol	/mol	/mol	/mol	/mol	/mol	/mol	/mol	/g	/g
1	2 °C, 13 MPa	2.9	2.8	2.1–2.8	<0.7	0	17.0	<7.7	21.9	1452	804
2	8 °C, 13 MPa	2.7	1.6	1.4–1.6	< 0.2	0	16.0	<8.3	27.5	1495	860
3	10 °C, 13 MPa	2.8	2.5	0.4–2.5	< 2.1	0	9.1	<8.4	21.6	1429	871
4	8 °C, 8 MPa	2.4	1.5	1.4–1.5	< 0.1	0	12.5	<7.5	21.5	1395	779

Table 5. Summary of results: Mass and energy based process efficiencies and yields.

No.	T, p	CH4			CO2			H2O		Energy
		YCH4		ECH4	SCO2		ECO2	SH2O		E∆H,prod	E∆H,rel
		/mol	/mol-%		/mol	/mol-%		/g	/w-%	/J-%	/J-%
1	2 °C, 13 MPa	0.1	2.4	0.003	17.1	77.9	0.78	804	55	2.4	18.3
2	8 °C, 13 MPa	0.8	31.2	0.030	16.0	58.0	0.58	860	58	19.4	23.2
3	10 °C, 13 MPa	0.4	12.5	0.016	9.1	42.5	0.43	871	61	8.2	57.4
4	8 °C, 8 MPa	0.8	35.5	0.039	12.5	58.0	0.58	779	56	17.6	21.0

Table 5. Summary of results: Mass and energy based process efficiencies and yields.

No.	T, p	CH4			CO2			H2O		Energy
		YCH4		ECH4	SCO2		ECO2	SH2O		E∆H,prod	E∆H,rel
		/mol	/mol-%		/mol	/mol-%		/g	/w-%	/J-%	/J-%
1	2 °C, 13 MPa	0.1	2.4	0.003	17.1	77.9	0.78	804	55	2.4	18.3
2	8 °C, 13 MPa	0.8	31.2	0.030	16.0	58.0	0.58	860	58	19.4	23.2
3	10 °C, 13 MPa	0.4	12.5	0.016	9.1	42.5	0.43	871	61	8.2	57.4
4	8 °C, 8 MPa	0.8	35.5	0.039	12.5	58.0	0.58	779	56	17.6	21.0

The combination of CH₄ extraction from gas hydrates and storage of CO₂ results in a complex multi-component, multiphase transport-reaction scheme, as illustrated in Figure 9. Highly dynamic processes on different time scales are putatively important for CH₄-hydrate dissociation, hydrate conversion and pure or mixed hydrate reformation, all of them altering sediment parameters, such as bulk thermal and geomechanical behavior. Due to the spatial heterogeneity and concurrency of multiple processes, it is very likely that optimization efforts will be most successful when considering processes down to the pore scale. This underlines the technical relevance of former studies of CH₄-CO₂-hydrate conversion (Table 3), hydrate dissociation and formation ([40] and references therein), which were strongly focused on gaining mechanistic insights on grain size and molecular scale.

Figure 9. Process dynamics scheme. Relevant transport and reaction processes during injection of hot CO_2,sc into CH₄-hydrate bearing sediment. (1) Injection of hot, mobile CO_2,sc; (2) Fast dissociation of CH₄-hydrate due to thermal stimulation from hot CO_2,sc injection; (3) Conductive heat transport via sediment particles or non-mobile pore water followed by thermal stimulation and dissociation of distant CH₄-hydrate particles; (4) Mixing of CO₂ and CH₄; (5) CO₂-hydrate formation with pore water; (6) Formation of mixed CO₂-CH₄-hydrates from CO₂-CH₄ gas mixtures; (7) CH₄-CO₂-hydrate exchange limited by diffusive transport of CH₄ and CO₂; (8) CO_2,sc cools and rapidly transforms to CO_2,liq. CO₂-CH₄ mixtures remain supercritical at high CH₄ content; (9) CO₂-hydrate formation with excess pore water might result in pore space clogging followed by substantial change of porosity and permeability. Complete loss of hydraulic conductivity is possible; (10) Production of fluid containing CH₄, water and CO₂ in various states and different and changing proportions; (11) Dissociation of load-bearing hydrate particles can cause settling of sediment and change of sediment integrity and geomechanical stability; (12) CH₄ gas might be released into hydraulically isolated pores as non-producible gas.

Figure 9. Process dynamics scheme. Relevant transport and reaction processes during injection of hot CO_2,sc into CH₄-hydrate bearing sediment. (1) Injection of hot, mobile CO_2,sc; (2) Fast dissociation of CH₄-hydrate due to thermal stimulation from hot CO_2,sc injection; (3) Conductive heat transport via sediment particles or non-mobile pore water followed by thermal stimulation and dissociation of distant CH₄-hydrate particles; (4) Mixing of CO₂ and CH₄; (5) CO₂-hydrate formation with pore water; (6) Formation of mixed CO₂-CH₄-hydrates from CO₂-CH₄ gas mixtures; (7) CH₄-CO₂-hydrate exchange limited by diffusive transport of CH₄ and CO₂; (8) CO_2,sc cools and rapidly transforms to CO_2,liq. CO₂-CH₄ mixtures remain supercritical at high CH₄ content; (9) CO₂-hydrate formation with excess pore water might result in pore space clogging followed by substantial change of porosity and permeability. Complete loss of hydraulic conductivity is possible; (10) Production of fluid containing CH₄, water and CO₂ in various states and different and changing proportions; (11) Dissociation of load-bearing hydrate particles can cause settling of sediment and change of sediment integrity and geomechanical stability; (12) CH₄ gas might be released into hydraulically isolated pores as non-producible gas.

6. Outlook

Our results suggest that both efficiency and rate of CH₄ production can be optimized in numerous ways.

The additional combination with depressurization is likely to improve the production process, since CH₄-hydrates could initially be destabilized to facilitate injection of the supercritical CO₂. In turn, the heat of the CO₂ can then compensate the cooling of the reservoir induced by the endothermic hydrate dissociation from depressurization. As a consequence, substantial production rates of CH₄ gas could be maintained over extended periods of time.

Different modifications of the injection fluid with respect to its composition and injection temperature as well as of the injection strategy itself might avoid early CO₂ breakthrough and also improve gas mobilization, which was identified as a possible obstacle in the production process (i.e., in the 10 °C experiment).

According to our results, reservoir conditions including temperature, pressure and permeability have a major influence on the production process and the choice of appropriate reservoir sites and technical means to influence the reservoir properties will be of paramount importance. So far, our experiments lack from the impossibility to directly monitor the processes and phase distribution inside the pressure vessel. Instead, this information has to be obtained indirectly from the analysis of effluent fluids. Hence, future studies would benefit significantly from the application of online monitoring including imaging techniques such as MRI or CT.

Because of the complexity of the numerous physical and chemical processes contributing to the overall CH₄ production we emphasize the importance of understanding the behavior of the system from the reservoir scale down to the pore scale. Only such knowledge can lead to progress in production simulations and laboratory experiments, which are prerequisites for the development of optimized production techniques.