Lanlan Jiang*ab,
Minghao Yub,
Bohao Wub,
Tetsuya Suekanec,
Weizhong Lib and
Yongchen Song*b
aResearch Institute of Innovative Technology for the Earth, Kizugawa City, Kyoto 619-0292, Japan. E-mail: lanlan@dlut.edu.cn
bKey Laboratory of Ocean Energy Utilization and Energy Conservation of Ministry of Education, Dalian University of Technology, Dalian, Liaoning 116024, China. E-mail: songyc@dlut.edu.cn
cDepartment of Energy Sciences, Tokyo Institute Technology, Nagatsuta, Midori-ku, Yokohama 226-8502, Japan
First published on 24th November 2016
A clear understanding of the mass transfer properties during fluid injection into porous media is of importance to the safety of CO2 storage. In this study, several experiments were conducted to elucidate the displacement and dissolution processes of gaseous and supercritical CO2 in Berea sandstones using the X-ray CT technology. The initial CO2 distribution before brine injection was related to the pore structure of the sandstones. Transient images during brine injection at different flow rates showed a transformation from displacement to dissolution, and the dissolution fronts were affected by the core heterogeneity and flow rates. Then, the CO2 saturation was determined by imaging analysis. Both supercritical and gaseous CO2 saturation decreased sharply, meaning that dissolution dominated the flow process. The dissolution time could be correlated in terms of the flow rate, initial gas saturation and heterogeneity of the sandstone. The relationships between CO2 volume content and specific surface area were verified to qualitatively predict the influence of heterogeneity. The dynamic concentration and mass transfer coefficient were obtained, which gave the information for the mass transfer rate during CO2 storage.
• the rate at which free-phase CO2 is immobilized by capillary gas trapping,
• dissolution in the reservoir fluids (brine) or subsequent geochemical reactions, or
• the likelihood that free-phase CO2 will leak out of the storage formation.
The need for an enhancement of our understanding of the fate and transport processes is becoming more evident.
CO2 and brine are immiscible fluids and hence subject to immiscible displacement.3,5–8 CO2 distribution is controlled by many factors, i.e., physical properties of the aquifer rock (pore structure, pore size and heterogeneity), flow properties (fluid properties, flow patterns, injection parameters) and so on. Fluid saturations alone are inadequate characterizations of the system state during fluid flow.9–11 Information regarding the distribution of the fluid phases is also required because it is the distribution of fluids at the pore scale that determines both the interfacial areas and the relative permeability of the medium to each fluid. Also, CO2 distribution depends on the surface tension caused by the combined forces arising at the interface separating two or more phases,12 thus including fluid–fluid13,14 as well as fluid–solid interactions.15 Heterogeneities result in CO2 clusters distributed non-uniformly throughout the porous rock at high brine saturations; this also affects the relationship between CO2 saturation and porosity distribution. Heterogeneity in the porosity of the sub-core significantly impacts the CO2 migration at low injection rates; the influence of the porosity heterogeneity on the mean CO2 saturation becomes gradually diminished as the injection rate is increased.16,17 While CO2 and brine are mutually soluble, leading to a mass transfer between these fluid phases, with a potential impact on the displacement process in those areas where the two fluid phases are not yet equilibrated, such as near the injection point and close to the flood front. Brine injection is substantially more effective in accelerating dissolution in the confined geometry because buoyancy forces acting on the CO2 bring it into closer contact with the injected brine. Characterization of dissolution process is essential for CO2 distribution related to the security of CO2 storage.
Experimental studies have also yielded new insights into how pore-scale processes affect liquid–liquid phase dissolution in uniform porous media.18,19 For liquid–liquid phase dissolution, the mass transfer is constant over time for dissolving spheres and decreases for pendular rings as the liquid phase is trapped as spheres and pendular rings in a single layer of glass beads; while the wall affects the dissolution and the time period required for doublets to break apart into singlets. Also the pore-scale variations of the flow field were found to significantly affect the dissolution rate of individual liquid phase blobs.20 Although dissolved gas transport plays an important role during CO2 storage, only limited research has been performed on the dynamic dissolution characteristics of gas/liquid in porous media. The displacement stability of the CO2/brine drainage process has been investigated by numerical modelling, and aspects of mass transfer during drainage and imbibition that are relevant for non-equilibrated fluid phases were studied by core flooding with unsaturated CO2 and brine phases: after 1 PV of brine injection, a decrease in CO2 saturation is visible with a substantial pressure difference which can be interpreted as the onset of CO2 dissolution.21 Then, a non-equilibrium dissolution at the core scale was observed using the concentration of total dissolved CO2 in the outflowing brine. The non-equilibrium dissolution possibly results from many factors, which need to be studied experimentally.
Previous experimental studies gave insufficient information about the mass transfer coefficient, because it is hardly to calculate rather than to experimentally obtain.18,22 Numerical studies focused on the pore network model were used to simulate the dissolution experiments, and showed dissolution fronts shorter than the length of the network model; the geometries significantly affected the mass transfer coefficient.23 Although a limitation of a pore network model is that the ideal spherical pore bodies cannot represent the natural pore network.24 Zhao and Ioannidis used a pore network model found from the natural porous media with pore throats of square cross section, and explored the complex relationship between dissolution rate and interfacial area, saturation and Pe.25
According to Fick's law, the mass transfer from a gaseous CO2 phase into a liquid phase depends on the interfacial area, concentration gradient and solubility.26 Thus, a complex interdependency exists between the dissolution coefficient, Peclet number (Pe), saturation and interfacial area.27,28 Among the related parameters, the interfacial area between fluid phases is one of the crucial parameters in several flow and transport processes in porous media.29–31 The interfacial area depends not only on saturation, but also on the direction of the saturation change (drainage and imbibition).32 Among the inherent difficulties for mass transfer measurement is the interfacial area, especially a direct measurement of the interfacial area of the heterogeneous porous media at the pore scale, because of the complex size, shape, and spatial distribution of the gas phase. To simplify this complexity, much recent effort has been devoted to obtaining experimental values of the gas/liquid interfacial area. With a gas chromatograph and an UV detector, the mass of liquid and surfactant were obtained during gravity drainage and imbibition in two-three phase systems.28 The data were used in the Gibbs adsorption equation and in a surfactant mass balance equation to obtain the interfacial area over a range of brine saturation. With the recent developments in X-ray CT technology, the interfacial area can be measured accurately.
Unfortunately, there are few experimental data in the literature involving the mass transfer between brine and CO2 under the conditions of a natural reservoir. In this study, several experiments were conducted to elucidate the displacement and dissolution mechanisms of CO2 in the Berea sandstones using the X-ray CT technology. Brine was injected upward into CO2 filled sandstones with different flow rates. The CO2 distribution and saturation were visualized during dynamic displacement and dissolution, and the dissolution properties were investigated during brine injection.
CO2 and brine were used in the flow experiments. For the sake of enhancement in the contrast in CT images, chemicals containing iodide as a component were used for the wetting phase. The brine contained 7.5 wt% of sodium iodide (NaI). To simulate the reservoir conditions, the temperature and pressure of the flow experiments were 313 K/8.9 MPa and 298 K/0.1 MPa, respectively. The properties of CO2 and brine are shown in Table 1.
Experimental conditions | σ (mN m−1) | CO2 viscosity (pa s) | Brine viscosity (pa s) | CO2 density (kg m−3) | Brine density (kg m−3) |
---|---|---|---|---|---|
6 MPa/313 K | 33.2 | 1.79 × 10−5 | 6.55 × 10−4 | 149.51 | 1015.87 |
8 MPa/313 K | 28.5 | 2.24 × 10−5 | 6.56 × 10−4 | 279.49 | 1016.68 |
The core samples were scanned with the X-ray CT scanner to take 610 slice images of 608 × 608 pixels with a resolution of 13.11 μm per pixel. The brightness of the images was adjusted so that the brightness of the Teflon heat shrinking tube and surrounding air remained same for all the scans. In these experiments, scanning was performed at 90 kV and 78 μA.
Before the brine and CO2 injection, the core samples were scanned. A two-dimensional CT image of sandstone (V1) is shown in Fig. 2 to visualize the internal structure of the Berea sandstone. The solid matrix is shown in grey, and pores are shown in black. The white parts denote the solid matrix containing impurities with a high X-ray attenuation and low porosity and permeability.
From the grey image, the CT values along the sandstone, excluding the partial regions, were similar which means a relatively homogeneous core on the macroscale; however, the pore distribution was non-uniform, reflecting the anisotropy on the microscale. And the CT values at the ends of the sandstone were small because of artefacts. At about 4.92 mm (375 pixel) and 7.01 mm (535 pixel), CT values are larger than in other regions with a higher density and lower porosity of the solid matrix. Heterogeneity can be found in all of the sandstone samples. It is very difficult to measure the spatially related parameters with conventional methods. These measurements show that CT can provide not only the bulk properties but also the local distribution of pore properties, which is useful for investigation of the effect of heterogeneity on fluids distribution.
![]() | ||
Fig. 4 CT images for the XY plane during brine injection of 0.3 mL min−1 in the gCO2 filled sandstone (V1). |
The images at 0 PV show the original status of CO2 before brine injection, corresponding to the first row of the figure. As seen from the images, the original distribution of ScCO2 varies with the slices; CO2 preferred to fill the parts with large pores, reflected as the dark points in the images. However, the original distribution of gCO2 was similar for each slice, which agreed with the pore structure. In theory, CO2 may fully fill into the slices close to the inlet of injection with sufficient pressure. However, the results show that gCO2 at the positions of 7.35 and 7.57 mm concentrated at the centre of these slices, because the solid matrix containing the impurities resulted in a low porosity and permeability zone, shown in Fig. 2. The CO2 distribution in H2 was homogeneous and different to that in V1, which is caused by heterogeneity and the initial gas saturation before brine injection.
Taking the image at 1.24 mm as an example in Fig. 3, most of the ScCO2 concentrated on the centre and the left side; and little ScCO2 diffused into the regions close to the solid matrix containing impurities with a high X-ray attenuation (reflecting a low porosity and permeability zone) around the right side. The results demonstrated that the ScCO2 distribution was also affected by the pore structure. For the locations at around 2.98 and 4.73 mm, ScCO2 barely diffused to the area close to the solid matrix. However, the ScCO2 diffused to the right side at 5.61, 7.35 and 7.57 mm because of the decreasing area of the solid matrix containing impurities.
With brine continuously injected, the dark points disappear, reflecting CO2 lost. There are two main mechanism for CO2 lost: one is brine displacement and the other is dissolution. One important problem is the transfer between the two main mechanisms. In previous similar studies,26 CO2 dissolution changed to the dominant transport process during brine imbibition coupled with a pressure difference decrease, even after 0.1 PV of brine injection. Also, the CO2 saturation calculated from the density value was larger than the total value after brine imbibition, meaning a substantial amount of CO2 dissolved and subsequently miscible displacement in the brine phase. While lots of studies also verified that the brine displacement continued for less than 10 PV till the CO2 saturation never changed,21,34 in this study, we did not monitor the pressure change. But with similar sandstone and fluids, the displacement reached a steady state at around 10 PV.
Seen in Fig. 3, ScCO2 disappeared continuously because the brine injected got into sufficient contact with ScCO2. From the images at 5.4 PV. In some regions (i.e., inner part of the black circle) at the location of 2.98 mm, ScCO2 decreased sharply at 5.4 PV because brine preferred to go through the larger pores; and then increased a lot because of ScCO2 movement from the relatively small pores before 11.57 PV. After 11.57 PV, ScCO2 in the black circle continuously decreased and finally disappeared. Similar phenomena were found at the location of 5.61 and 7.35 mm. Before that, displacement dominated the flow process, and then dissolution dominated. However, the transformation time for displacement and dissolution changed with the slices. After 17.75 PV, dissolution dominated the transport through the whole sandstone. At 25.43 PV, most of remaining ScCO2 dissolved into brine. As seen from Fig. 4, the second row corresponded to 20 PV. At that time, gCO2 continued to disappear from the pores because of dissolution.
The onset of dissolution depended on the initial gas saturation and distribution. The larger the concentration of CO2 is, the earlier CO2 dissolution exists. When dissolution dominated the flow of brine imbibition, CO2 remaining in the pores after displacement got contacted with brine sufficiently. From the CT images, the dark points disappeared during the dissolution process because of the greater density of brine saturated CO2 and lighter CO2.
![]() | ||
Fig. 5 CT images for the coronation direction plane during brine injection with different flow rates: (a) 0.15 mL min−1 for gCO2, (b) 0.3 mL min−1 for gCO2 and (c) 0.02 mL min−1 for ScCO2. |
For all the sandstones, CO2 dissolution occurred along certain paths, which correlated to the pore structures. The fronts for dissolution can be visualized in different samples, which were defined as the position of the interface separating the continuous brine phase from the CO2 phase. From the images, the dissolution front flows along the large pores with a large CO2 concentration, corresponding to the space with a larger porosity. In horizontal sandstones, the dissolution front looked like a finger: the front flow along the vertical space with the larger pores up to the top side; the breakthrough occurred and then the front flowed along other layers. In part, the reservoir heterogeneity in the vertical case might negatively affect flooding in different ways: channelling and fingering of CO2 through the brine bank, which leads to bypassing brine zones and a disproportionate flow of flood CO2 through different strata, resulting in an early breakthrough in the high permeability stratum, while the low permeability one remains largely unwept. In practice, the flow rates influence the stability of the dissolution front. A small flow rate resulted in enough residual time for the brine in the sandstone and enough contact for the brine and CO2. Once the dissolution had completed, brine displaced the brine saturated CO2 along the large pore paths, which dominated the mass transfer process over time. While the ScCO2 dissolution front uniformly migrate upward compared to the gCO2 in the horizontal samples. Of course, the uniform front was also attributed to the small flow rate.
3D images of CO2 trapped at different times (Pore Volume, PV), corresponding to Fig. 5 as brine flushed through the core sample, are shown in Fig. 6. The sandstone matrix and brine are invisible as transparent background. The green colour means CO2 existing in the pore spaces. The interface between gas and brine was clearly visualized and differed with the sandstones, which agreed with the description with 2D images. It seems that all the gaseous CO2 dissolved into brine except the top side, which was caused by the end effect. However, little supercritical CO2 dissolved into the injection brine, meaning a lot of CO2 remained in the top of sandstone. Actually, this was caused by the small flow rate and small initial CO2 saturation.
Based on the binary images, profiles of saturation along the sample during brine injection are shown in Fig. 7. The initial saturation is close to 0.8–1.0 under normal pressure because there is sufficient time for the CO2 flush; meanwhile, it is about 0.5 under the supercritical pressure.
![]() | ||
Fig. 7 CO2 saturation profile during brine injection: (a) 0.15 mL min−1 for gCO2, (b) 0.3 mL min−1 for gCO2 and (c) 0.02 mL min−1 for ScCO2. |
With brine injection at a constant flow rate, the saturation at the bottom declined; but it kept stable at the top. A clear decrease in CO2 saturation at the inlet of the brine injection is observed, caused by a larger concentration difference. The moment when the saturation decreased sharply was defined as the transformation between displacement and dissolution. In fact, dissolution occurred after about 45 PV, 20 PV and 11.07 PV for cases I, II and III.
The dissolution front moved along the sample with a constant flow rate, so the saturation gradually declined along the sample. CO2 saturation for cases I and III was uniform along the sample because of the stable dissolution front, caused by the pore structure and lower flow rates. While in the case of 0.3 mL min−1, the CO2 saturation decreased randomly, especially in the first part of the dissolution. For 40–60 PV, the CO2 saturation declined slowly. But the dissolution completed at 140 PV, shorter than the 933 PV for the 0.15 mL min−1 flow. For the ScCO2, the saturation decreased sharply once dissolution dominated the flow after 17.57 PV; it continued only for 25 PV, even with 0.02 mL min−1. The most important reason was the low initial gas saturation at about 0.5. A similar phenomenon was that the saturation decreased along the flow direction linearly and gradually, reflecting the dissolution at a constant rate at that period.
Comparison of the dissolution process showed a dependence of the gas initial saturation and heterogeneity. The brine channelling or dissolution front formed in a different pattern. But in our studies, the most impacted factor is the heterogeneity. The heterogeneity resulted in different initial saturations, which are an important factor in the dissolution process. The high connectivity between the large pore regions and the inlet face of the core resulted in a high CO2 saturation. Therefore, some thin channels were established, and the brine ran through the channels vertically in a short period, so the channelling or fingering phenomena were more obvious on the paths. In conclusion, the initial gas saturation, flow rate of the brine injection and heterogeneity of the porous media influenced the dissolution front. To demonstrate the influence, the concentration and mass transfer coefficient will be discussed.
![]() | (1) |
Because of the complex size, shape, and spatial distribution of the entrapped gas blobs, it is difficult to measure directly; C is the concentration of CO2 in the liquid phase, mol L−1; Ce is the equilibrium concentration of CO2 in the liquid phase mol L−1; KL is the liquid side mass transfer coefficient, m s−1. Based on the empirical equation, CO2 concentrations were calculated from the porosity and saturation at any time and position:
![]() | (2) |
In this study, of most interest is to quantitate the specific surface area. Imaging results show that the specific surface area is linearly related to the volumetric content (Fig. 8). The volume content of CO2 concentration had a uniform distribution from 0.01–0.11 for 0.15 mL min−1; from 0.01–0.05 for 0.02 mL min−1; from 0.02–0.06 for 0.3 mL min−1 (not shown in the figure). And the specific surface area decreased from 500 to 100 mm−1 for 0.15 mL min−1; from 6–1.5 mm−1 for 0.02 mL min−1; from 15–5 mm−1 for 0.3 mL min−1 (not shown in the figure). The volume content of for 0.15 mL min−1 is larger than that for 0.3 mL min−1, even at a similar initial gas saturation. The constant of proportionality between these parameters is determined by the gas size and geometry distribution. The specific surface area decreased with flow rate because of the gas bubble size and heterogeneity, with the smaller one an order magnitude of 102 for 0.15 mL min−1. However, the specific surface area for ScCO2 with 0.02 mL min−1 was smaller than that for gCO2 with 0.3 mL min−1. The reason was that the initial gas saturation was low.
![]() | ||
Fig. 8 The specific surface area as a function of volume content of CO2: (a) 0.15 mL min−1 for gCO2 and (b) 0.02 mL min−1 for ScCO2. |
The concentrations profiles along the porous media are shown in Fig. 9. The CO2 concentration in brine increased along the porous media with brine injection. With slow flow rates, the CO2 has sufficient residential time to travel through the porous media and to have local equilibrium dissolution within the pores of the flow paths. While with relatively fast flow rates, the CO2 does not have sufficient residential time for equilibrium. The ScCO2 concentration in brine was similar for 5.71 PV and 11.57 PV. And the concentration at the bottom side decreased a little because displacement surpassed dissolution and dominated the process. After that, ScCO2 concentration gradually increased. Although the gCO2 saturation decrease was stable for case I, the gCO2 concentration increased disorderly because of the relative flow rate and high interfacial tension compared to the supercritical state.
![]() | ||
Fig. 9 The concentration profiles along the porous media: (a) 0.15 mL min−1 for gCO2 and (b) 0.02 mL min−1 for ScCO2. |
Based on the concentration and specific surface area, the dynamic mass transfer coefficients are shown in Fig. 10. Similar to the saturation profile, the mass transfer coefficient for 0.15 mL min−1 at the initial time was at the peak value of 0.8 × 10−4 m s−1. Then with the concentration difference decreasing, the mass transfer coefficient decreased along the brine flow direction. As seen from the figure, the mass transfer became stable along the sample after 185 PV. It was very interesting that the mass transfer coefficient at the top side increased finally, which may be caused by dissolution and the displacement of brine and brine saturated CO2. In contrast, the mass transfer coefficient for ScCO2 gradually and steadily decreased with time along the sample. The largest coefficient for ScCO2 was nearly 0.08 m s−1. The bulk mass transfer coefficients for gCO2 and ScCO2 were different: 1.49 × 10−5 and 0. 54 × 10−2 m s−1. Most of the ScCO2 dissolved into water during imbibition and little dissolved during dissolution, reflecting a high dissolution rate and a high storage security.
![]() | ||
Fig. 10 Mass transfer coefficient during brine injection: (a) 0.15 mL min−1 for gCO2 and (b) 0.02 mL min−1 for ScCO2. |
The original CO2 distribution before brine injection depended on the pore distribution of the pore structure. CO2 preferred to fill into large pores with a large porosity, and disappeared with brine injection because of displacement and dissolution. During the displacement process, CO2 moved through the paths and the distribution in the slices changed. As dissolution dominated, a lot of CO2 disappeared from the images.
The CO2 saturation profile during brine injection was obtained. The saturation decreased sharply, reflecting the transformation from displacement and dissolution. The onset of dissolution depended on the initial gas saturation and the heterogeneity of the core. And CO2 dissolution occurred along certain paths correlated to the pore structures.
The specific surface areas during dissolution were quantified relative to the volumetric fraction. The dynamic concentration profiles and mass transfer along the porous media were investigated. The bulk mass transfer coefficient for ScCO2 was larger by orders of magnitude than that for gCO2, meaning a high dissolution rate and high security storage of ScCO2.
This journal is © The Royal Society of Chemistry 2016 |