Published in Vadose Zone Journal 2:287-296 (2003)
© 2003 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
ORIGINAL RESEARCH PAPER
On Leakage and Seepage from Geologic Carbon Sequestration Sites
Unsaturated Zone Attenuation
Curtis M. Oldenburg* and
André J. A. Unger
Earth Sciences Division 90-1116, Lawrence Berkeley National Laboratory, University of California, Berkeley, CA 94720
* Corresponding author (cmoldenburg{at}lbl.gov).
Received 25 February 2003.
 |
ABSTRACT
|
|---|
Geologic carbon sequestration is the direct injection of CO2 into deep geologic formations for permanent disposal. Although numerous trapping mechanisms exist in the subsurface, it is possible that CO2 will leak from the primary sequestration target and seep out of the ground. The unsaturated zone has the potential to attenuate leaking CO2 and decrease seepage and near-surface CO2 concentrations. Attenuation processes include permeability trapping, ponding as dense CO2 spreads out on the water table, solubility trapping by infiltrating or residual water, and dilution through mixing with ambient soil gas. Numerical simulations of CO2 flowing upward through a thick model unsaturated zone were performed to investigate the sensitivity of various unsaturated zone properties on CO2 seepage flux and near-surface CO2 gas concentrations. These two quantities are considered drivers for health and environmental risk due to exposure to CO2. For the conceptual model considered, seepage flux and near-surface CO2 gas concentrations are most strongly controlled by the leakage rate at the water table, followed by the source zone radius. Permeability and permeability anisotropy, as well as porosity and infiltration rate are also important, although to a lesser degree. Barometric pumping causes local maxima in seepage flux and near-surface CO2 concentrations, but has negligible effect in a time-averaged sense. When the leakage source is turned off, the CO2 plume attentuates through dissolution into infiltrating water. For the case of a constant leakage rate, the unsaturated zone can attenuate low leakage fluxes but should not be expected to attenuate large CO2 leakage fluxes.
|
INTRODUCTION
|
|---|
GEOLOGIC CARBON SEQUESTRATION is one strategy for reducing the rate of increase of global atmospheric CO2 concentrations (Table 6 in International Energy Agency, 1997; Reichle et al., 1999). As used here, the term geologic carbon sequestration refers to the direct injection of liquid or supercritical CO2 deep into subsurface geologic formations. The target formations will typically be either depleted oil and gas reservoirs, or brine-filled permeable formations. The idea is to trap injected CO2 by one or more of the following mechanisms (e.g., Bachu et al., 1994): (i) permeability trapping, for example when buoyant supercritical CO2 rises until trapped by a confining layer or cap rock; (ii) solubility trapping, for example when CO2 dissolves into the aqueous phase in the pore space; or (iii) mineralogic trapping, such as occurs when CO2 reacts to form stable carbonate minerals. When CO2 is trapped in the subsurface by any of these mechanisms, it is effectively sequestered away from the atmosphere where it would otherwise act as a greenhouse gas.
Although the purpose of geologic carbon sequestration is to trap CO2 in the subsurface, there is the risk that injected CO2 will migrate away from the primary target formation (Holloway, 1997). Migration away from the primary target formation is referred to here as leakage. Carbon dioxide that leaks from the primary sequestration target and moves up through subsurface formations is likely to undergo secondary trapping in up-section structural traps and by dissolution processes, resulting in very long transport times (Lindeberg, 1997). This is in contrast to leakage that may occur through a well or other fast-flow path, in which case subsurface gas transport can be very fast (e.g., Allison, 2001). If CO2 reaches the shallow subsurface or migrates out of the ground into the ambient air, health and environmental risks can arise. By analogy to existing processes whereby water, oil, and gas migrate across the subsurface–ground-surface interface, we refer to the migration of CO2 out of the ground as seepage. Seepage of CO2 can lead to locally high CO2 concentrations, especially in topographic depressions because CO2 is a dense gas relative to air. Ambient CO2 concentrations in the atmosphere are approximately 350 ppmv, while CO2 concentrations of 1% or more cause measurable adverse physiological effects, and concentrations above 10% can be deadly (National Institute of Occupational Safety and Health, 1976).
The objective of this work was to examine the potential of the unsaturated zone to attenuate CO2 leaking from a geologic sequestration site. This is in contrast to prior studies that have focused on ambient soil processes (e.g., Amundsen and Davidson, 1990; Kabwe et al., 2002). Attenuation processes include CO2 gas ponding on the water table due to its high density, permeability trapping, solubility trapping, and simple dilution by mixing with ambient soil gas. We are not considering mineralogic trapping, which generally occurs on a time scale much longer than gas transport in the vadose zone. Direct interaction with the atmosphere is considered in terms of barometric pumping, although the effects of winds on shallow soils are beyond the scope of this study. With an aim toward contributing to analyses of potential health and environmental risk due to CO2 seepage, we present simulation results in terms of seepage flux and near-surface CO2 concentrations, both of which are drivers of exposure risk for humans and other living things in the biosphere. Rather than defining a detailed or site-specific unsaturated zone, we have adopted the approach of a sensitivity analysis. In this approach, the effects of various properties of a model system, such as leakage rate, permeability, radius of leakage zone, porosity, and infiltration rate, can be simulated. This approach determines the trends in seepage flux and near-surface CO2 concentrations for natural systems with various combinations of properties. The sensitivity analysis is based on a conceptual model in which CO2 is discharged at the water table at a constant rate (e.g., through a high-permeability zone with direct connection to the target sequestration formation) and migrates upward through a thick unsaturated zone. Finally, we also consider a single case in which the leak is turned off and the CO2 plume is controlled by density and infiltration effects.
|
METHODS
|
|---|
The approach we used to investigate leakage and seepage of CO2 in the subsurface was numerical simulation. Simulations presented in this work were performed using TOUGH2 (Pruess et al., 1999) with a special module called EOS7CA applicable to flow and transport of CO2 and air in subsurface systems. TOUGH2/EOS7CA models the subsurface flow and transport of aqueous and gas phases containing five components (H2O, brine, CO2, gas tracer, and air) under isothermal or nonisothermal conditions. TOUGH2/EOS7CA uses real gas mixture properties calculated using the Peng–Robinson equation of state model. Air is approximated in EOS7CA as a mixture of 79% N and 21% O2 by volume. Solubility of CO2 in the aqueous phase is modeled by Henry's Law, with Henry's coefficients calculated from Cramer (1982). Viscosity is estimated using the method of Chung et al. (1988), as described by Poling et al. (2001).
Gas Properties
The physical properties of CO2 relative to air and liquid water have a significant impact on the ability of the subsurface to attenuate leaking CO2. Carbon dioxide is a colorless and odorless gas with critical pressure (Pc) equal to 7.38 MPa and critical temperature (Tc) equal to 31°C. We present in Fig. 1 the phase diagram for CO2 showing the gaseous, liquid, solid, and supercritical regions along with an approximate curve representing a P–T path in the subsurface assuming hydrostatic pressure and 25°C km-1 geothermal gradient. As shown in Fig. 1, the geothermal gradient ensures that CO2 will be supercritical in the subsurface at depths greater than approximately 800 m and gaseous at shallower depths.
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 1. Phase diagram for CO2 with approximate P, T path in the subsurface assuming hydrostatic pressure and geothermal gradient of 25°C km-1.
|
|
The density of CO2 increases drastically as it changes from subcritical to supercritical conditions; however, the geothermal gradient ensures a sufficiently high temperature that CO2 is buoyant relative to water in the saturated zone. Shown in Fig. 2 is the density of CO2 as a function of pressure at three different temperatures, where the symbols show results from TOUGH2/EOS7CA and the lines are from the NIST14 database (National Institute of Science and Technology, 1992). Because the focus in this paper is the unsaturated zone, we present in Fig. 3 the density of CO2 and air mixtures at 0.1 MPa. As shown, pure CO2 at 0.1 MPa and 20°C has a density of approximately 1.8 kg m-3, while pure air has a density of 1.2 kg m-3. The effects of water vapor in the air will further decrease ambient soil gas density. Thus, the high density of CO2 relative to soil gas creates the potential for CO2 to pond on the water table and resist moving upwards to the ground surface. In summary, Fig. 2 and 3 show that while CO2 will tend to rise upward in the saturated zone, it may tend to accumulate in the unsaturated zone and in topographic depressions due to its high density.
Other transport properties of CO2 are also relevant to leakage attenuation. In particular, the solubility of CO2 in water is relatively high, approximately 50 times that of air at 0.1 MPa, 20°C. The large solubility of CO2 has the potential to attenuate CO2 by solubility trapping in water in the unsaturated zone. At 0.1 MPa and 20°C, CO2 is slightly less viscous and therefore more mobile than air (µCO2 = 1.5 x 10-5 Pa s, µair = 1.8 x 10-5 Pa s) (National Institute of Science and Technology, 1992). At ambient atmospheric conditions of 0.1 MPa and 10°C, the free gas molecular diffusivity of CO2 in air is typical of other gases and is of order 10-5 m2 s-1 (Vargaftik, 1975). Simulation results presented here use the Fickian advective–diffusive model for gas transport, a good approximation for high-permeability systems (Webb, 1998; Oldenburg et al., 2003).
Conceptual Model
The conceptual model for attenuation of leakage by the unsaturated zone is based on a prototypical geological sequestration site containing a mass of 4 x 109 kg of CO2. The hydrogeological properties of the system for the base case are listed on Table 1. These properties are typical of poorly sorted and unconsolidated sediments such as can be found in California's Central Valley. Figure 4 depicts the conceptual model and grid used to simulate the transport of CO2 through the unsaturated zone. The model is cylindrical, with the Cartesian axis located in the z-direction. The model unsaturated zone is 30 m thick, with a saturated zone 5 m in thickness. The mesh used for simulations contains 20 by 120 nodes, with a minimum and maximum radial size of 5 and 30 m, respectively. The vertical discretization is uniformly 1.75 m. Temperature is assumed to be constant at 15°C throughout the domain.
For the base case, CO2 is introduced into the model system across a radial distance of 100 m, which corresponds to an area of 3 x 104 m2 across which leakage of CO2 from the reservoir arrives at the water table. This area is small relative to the expected geologic sequestration site footprint and represents a focused leak as might occur along intersecting subnormal faults. The leakage rates used were 4 x 104, 4 x 105, and 4 x 106 kg yr-1, representing leakage rates of 0.001, 0.01, and 0.1% yr-1 of the initial mass of a 4 x 109 kg CO2 sequestration reservoir. For reference, an overall leakage rate of 0.01% yr-1 or less would still meet atmospheric stabilization targets, and 0.1% yr-1 is effective for some energy and population scenarios (see Hepple and Benson, 2002).
The bottom boundary is held at hydrostatic pressure, while the top boundary is used to enforce an atmospheric pressure equal to 0.1 MPa. In addition, the gas-phase CO2 concentration at the top of the system is held at 350 ppmv. Recharge water enters the top of the domain uniformly in equilibrium with the 350 ppmv atmospheric CO2 concentration as controlled by Henry's Law. The right-hand side boundary is constant pressure (hydrostatic below the water table, and gas-static in the unsaturated zone). The left-hand side boundary is no-flow appropriate for symmetry about the z-axis.
Sensitivity Analysis Method
To focus the analysis on the ability of the unsaturated zone to attenuate leakage of CO2, we adopted a worst-case scenario where we assumed that CO2 from the leaky sequestration target reservoir discharges directly at the water table. Such a release could occur as the result of a high-permeability conduit that allows fast flow through the saturated zone, circumventing probable attenuating influences in the saturated zone such as secondary solubility and permeability trapping. The objective of this analysis was to determine the influence of hydrogeological properties of the unsaturated zone and leakage characteristics on the maximum seepage flux of CO2 as well as the maximum near-surface mole fraction of CO2 relative to the base scenario. Specifically, we varied six parameters, including the source zone leakage rate, permeability, permeability anisotropy, source zone radius, porosity, and infiltration rate. For this analysis, the unsaturated zone was relatively thick at 30 m. Suffice it to say that attenuation by the unsaturated zone will be generally smaller for cases where the water table is nearer the surface. Permeabilities in the range 10-12 to 10-9 m2 were chosen to represent a permeable unsaturated zone, and extrapolation to lower or higher permeabilities can be made from the sensitivity study presented below. Hydrogeological properties for the base case are presented in Table 1. The potential exposure risk to CO2 at the ground surface will be put into perspective by comparison to both a typical ecological CO2 efflux taken as 4.4 x 10-7 kg s-1 m-2, or 10 µmol s-1 m-2 (Baldocchi and Wilson, 2001), as well as the mole fraction of CO2 in the gas phase in soil at which tree mortality has been observed, taken as 0.3 (e.g., Farrar et al., 1995, 1999).
|
RESULTS
|
|---|
Leakage Rate
Results are shown in Fig. 5 as vertical cross sections showing mass fraction of CO2 in the gas phase, water saturation, and gas phase pore velocity vectors for leakage rates of (Fig. 5a) 4 x 104 kg yr-1, (Fig. 5b) 4 x 105 kg yr-1, and (Fig. 5c) 4 x 106 kg yr-1 after relatively steady seepage rates are obtained after 100 yr. These figures show that although the CO2 mass fraction is nearly unity above the source, and therefore forms a dense gas phase relative to ambient soil gas, there is very little lateral spreading of CO2 on the water table. In fact, the CO2 plume spreads a maximum of 120 m beyond the radius of the source zone for the highest leakage rate. Instead, the CO2 plume reaches the ground surface for all injection rates. However, it should be kept in mind that the radial geometry involved ensures that there is a significant mass of CO2 contained within the region where spreading has occurred between 100 and 120 m from the axis of the system.
View larger version (54K):
[in this window]
[in a new window]
|
Fig. 5. Shading indicates the mass fraction of CO2 in the gas phase, labeled contour lines indicate the water saturation, and vectors indicate the pore velocity of the gas phase for the base case at steady-state seepage rates with a leakage rate of (a) 4 x 104, (b) 4 x 105, and (c) 4 x 106 kg yr-1. The maximum vector size represents a value of approximately (a) 0.057, (b) 0.53, and (c) 3.6 m d-1.
|
|
The temporal evolution of the total seepage and maximum near-surface CO2 concentrations are presented in Fig. 6 for three different leakage rates. The maximum CO2 concentration at the top always occurs at the center (left-hand side) of the model system. With time beginning when the leakage reaches the water table, the results show that the unsaturated zone retards seepage, but the retardation time depends strongly on the leakage rate. For the largest leakage rate, the unsaturated zone retards CO2 seepage by only several days, while for the lowest leakage rate, it retards CO2 by nearly a year. As shown, the total seepage and maximum CO2 concentrations at the surface are nearly steady after approximately 50 yr for the three leakage rates tested.
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 6. Total flow rate of CO2 across the top boundary and maximum mole fraction of CO2 in the gas phase at the top of the system as a function of time after leakage enters the unsaturated zone.
|
|
Permeability and Permeability Anisotropy
Additional example simulation results are shown in Fig. 7 for comparison with the base case shown in Fig. 5. Figures 7a and 7b show vertical cross sections of the mass fraction of CO2 in the gas phase, water saturation, and gas-phase pore velocity vectors for a leakage rate of 4 x 105 kg yr-1 for a permeability (k) of 1 x 10-9 m2 and an anisotropy (kr/kz) of 1000:1, respectively. By comparing Fig. 5b with Fig. 7a, we observe that as both the horizontal and vertical permeabilities are increased from 1 x 10-12 to 1 x 10-9 m2, horizontal spreading of the plume increases dramatically, while vertical transport is reduced. Comparison of Fig. 5b and 7b reveals that as the anisotropy is increased from 1:1 to 1000:1, the same trend of increased horizontal spreading and decreased vertical transport occurs. This trend is more prominent for the anisotropy case because the vertical permeability, kz, remains fixed at the lowest value of 1 x 10-12 m2, forcing the CO2 to be preferentially transported horizontally.
View larger version (63K):
[in this window]
[in a new window]
|
Fig. 7. Shading indicates the mass fraction of CO2 in the gas phase, labeled contour lines indicate the water saturation, and vectors indicate the pore velocity of the gas phase for a leakage rate of 4 x 105 kg yr-1 and at steady-state seepage rates with (a) a permeability of 1 x 10-9 m2, (b) an anisotropy of 1000:1, (c) a source radius of 10 m, and (d) a source radius of 1000 m. The maximum vector size represents a value of approximately (a) 1.0, (b) 8.4, (c) 17, and (d) 0.0048 m d-1.
|
|
Source Zone Radius
The radius of the source zone where leakage occurs strongly controls unsaturated zone leakage attenuation. Consideration of a very small source radius corresponds to the assumption that the migration pathway that CO2 has followed to the water table is confined to a small zone, such as a borehole, while a large radius is consistent with a model in which CO2 migration occurs through multiple fracture zones or widespread permeable formations. As part of a sensitivity analysis, we adjusted the source radius from the base-case value of 100 m to a maximum value of 1000 m and a minimum value of 10 m. All hydrogeological parameters shown on Table 1 were held constant. The leakage rate was also varied from 4 x 106 to 4 x 105 and 4 x 104 kg yr-1.
Figures 5b, 7c, and 7d show vertical cross-section results for a leakage rate of 4 x 105 kg yr-1 with a source radius of 100, 10, and 1000 m, respectively. As the source radius is decreased by an order of magnitude to 10 m, the gas-phase pressure increases significantly around the source zone perturbing the water table. The width of the CO2 plume emanating from the 10-m source zone is only slightly smaller than that of the base case (Fig. 5b). This indicates that for a leakage rate of 4 x 105 kg yr-1, the CO2 plume extends out a minimum radial distance of 130 m from the origin and is not simply confined to a radius of the source zone as might be inferred from the base case. As the source radius is increased by an order of magnitude to 1000 m, the flux of CO2 decreases dramatically, yielding significantly lower mole fractions of CO2 in the gas phase above the source zone.
Porosity
The porosity of the unsaturated zone has the potential to influence the horizontal and vertical transport of the CO2 plume by changing the pore volume available to the gas phase CO2 plume. A decrease in porosity should increase the gas velocity and plume size, while an increase in porosity should decrease gas velocity and plume size. As part of a sensitivity analysis, we doubled the porosity from the base-case value of 0.2 to 0.4 and alternatively halved it to 0.1. All other hydrogeological parameters shown on Table 1 were held constant. The leakage rate was also varied from 4 x 106 to 4 x 105 and 4 x 104 kg yr-1. Results are summarized in Fig. 8 as described below.
View larger version (48K):
[in this window]
[in a new window]
|
Fig. 8. The maximum seepage flux of CO2 and the maximum near-surface mole fraction of CO2 as a function of leakage rate at steady-state seepage conditions.
|
|
Infiltration Rate
Water infiltrating through the unsaturated zone is in equilibrium with atmospheric concentrations of CO2 that are orders of magnitude lower than the values above the source zone in the base case. Therefore, this water has the capacity to attenuate the upward migration of CO2 through the unsaturated zone as it continually dissolves CO2 from the gas phase. As part of a sensitivity analysis, we increased the infiltration rate from the base-case value of 0.1 m yr-1 to 0.5 m yr-1, and decreased it to 0.02 and 0.0 m yr-1. All hydrogeological parameters shown on Table 1 were held constant. The leakage rate was also varied from 4 x 106 to 4 x 105 and 4 x 104 kg yr-1. Results of this sensitivity study are presented in Fig. 8, and discussed below.
Summary of Sensitivity Analysis
Results from a multitude of simulations are presented in Fig. 8 as a comprehensive summary of the sensitivity analyses we have performed. Figures 8a and 8b show leakage rate on the x axis vs. seepage rate and maximum near-surface CO2 mole fraction in the gas, respectively, on the y axis. The first conclusion to be drawn from Fig. 8 is that the main parameter controlling the seepage and concentration of CO2 at the ground surface is the leakage rate from the reservoir. For our base-case scenario, the maximum leakage rate that could be simulated without significantly perturbing the water table was 4 x 106 kg yr-1, corresponding to 0.1% yr-1 leaking from the reservoir. As the leakage rate decreases from 4 x 106 kg yr-1, the maximum seepage flux of CO2, which always occurs at the center of the radial system, drops below that of the ecological flux. Although the maximum seepage flux of CO2 appears to be quite small, the corresponding maximum near-surface mole fraction of CO2 may pose a significant health and environmental risk. This is indicated by values above the 0.3-mol fraction tree mortality level as shown in Fig. 8b.
Figure 8 shows that the smallest source radius causes the greatest seepage of CO2 for a given leakage rate. As the source zone radius increases from 10 to 100 and 1000 m, the seepage of CO2 drops dramatically for all three leakage rates. Specifically, the maximum seepage flux of CO2 is significantly larger than the typical ecological flux except at the lowest leakage rate. For comparison, the seepage for a leakage rate of 4 x 106 kg yr-1 approaches the maximum values measured around the Horseshoe Lake tree-kill area at Mammoth Mountain, CA (Sorey et al., 1999). Not surprisingly, the maximum near-surface mole fraction of CO2 also exceeds the tree-mortality limit. As the source radius is increased to 1000 m, the CO2 seepage and near-surface concentration are below the ecological flux and tree mortality limits for all three leakage rates.
After the leakage rate and leakage area, the next most sensitive parameters controlling the seepage and near-surface concentration of CO2 are the permeability and anisotropy of the unsaturated zone. As part of a sensitivity analysis, we increased the radial and vertical permeabilities, kr and kz, from the base-case value of 1 x 10-12 to 1 x 10-11, 1 x 10-10, and 1 x 10-9 m2. Similarly, we also increased the anisotropy of kr/kz from 1:1 in the base case to 10:1, 100:1, and 1000:1. All other hydrogeological parameters shown on Table 1 were held constant. The sensitivity of the seepage and near-surface CO2 concentrations to an increase in the permeability and the anisotropy is also shown on Fig. 8. For a leakage rate of 4 x 106 kg yr-1, the maximum seepage flux of CO2 is relatively insensitive to an increase in permeability but is very sensitive to an increase in anisotropy. Specifically, the maximum seepage flux of CO2 is greater than the typical ecological flux for the full range of permeabilities used in the sensitivity analysis, whereas only an anisotropy ratio of 1:1 and 10:1 are greater than the ecological flux. As the leakage rate decreases to 4 x 105 and 4 x 104 kg yr-1, the seepage flux is always less than the ecological flux independent of variations in permeability and porosity. For a leakage rate of 4 x 106 kg yr-1, the maximum near-surface mole fraction of CO2 exceeds the tree mortality value of 0.3 for all ranges of permeability and all values of anisotropy from 1:1 to 100:1. As the leakage rate decreases to 4 x 105 kg yr-1, both a permeability of 1 x 10-11 m2 and an anisotropy of 10:1 are close to the tree mortality limit, with all other values of permeability and anisotropy below this threshold.
Results for maximum seepage flux and near-surface mole fraction of CO2 for infiltration rates of 0.1 m yr-1 (base case), and 0.0, 0.5, and 0.02 m yr-1 are also shown in Fig. 8a and 8b. These results show that the CO2 seepage and near-surface concentrations do not deviate significantly from the base case as the infiltration rate changes. Figure 8 also shows the maximum surface flux and near-surface mole fraction of CO2 for porosities of 0.2 (base case), 0.4, and 0.1. These results also do not deviate significantly from the base case because of variability in porosity for all three leakage rates.
We have also investigated the effects of barometric pumping on the seepage and near-surface mole fraction of CO2. We specified a time-varying top pressure boundary condition corresponding to an actual pressure variation measured in the Central Valley of California for the year 1997 and performed a simulation with properties of the base case. This pressure profile, shown in Fig. 9, was assumed to repeat annually in our simulations. Results of the maximum seepage flux and near-surface mole fraction are shown as a function of time (log scale) in Fig. 10. As shown, the time-averaged effect of barometric pumping on seepage and shallow CO2 concentration is negligible, even though locally higher and lower fluxes and concentrations can arise from day to day variations in pressure. Furthermore, the small leakage rate cases show larger amplitude variation as barometric pumping occurs because the air flow driven by atmospheric pressure variation dominates the CO2 leakage flux for small CO2 fluxes. The lack of importance of barometric pumping observed in these simulations arises from the cyclic nature of barometric pressure variation. Wind forcings that act more consistently in one direction are expected to influence the shallowest soil layers and will tend to enhance CO2 seepage and dilute shallow soil gas CO2 concentrations but are beyond the scope of this study.
View larger version (24K):
[in this window]
[in a new window]
|
Fig. 10. Temporal evolution of (a) total surface flow of CO2 and (b) maximum near-surface mole fraction of CO2 for the base case with variable-pressure top boundary condition.
|
|
Another scenario relevant to the unsaturated zone is the case where the CO2 leakage is eventually detected and the leak is effectively stopped, for example by means of operations to lower the pressure in the leaking CO2 sequestration reservoir. We considered a single case of this scenario by starting from the steady-state solution (t = 100 yr) of the 4 x 106 kg yr-1 leakage rate case and then turning off the CO2 source. In this case, the driving force for the CO2 plume will evolve to depend on density effects and dissolution as infiltration passes through the plume. Presented in Fig. 11 are cross sections of the simulated results after 6 mo, and 1, 5, and 10 yr. Note that between 6 mo and 1 yr, a small amount of lateral spreading occurs, and the near-surface CO2 concentrations rapidly decrease as the plume dissolves in the infiltrating water. Large-scale effects of lateral density-driven flow are not observed because the pressurization arising from the prior injection at the water table dominates over the buoyancy effects for the entire time before the plume dissipates through dissolution into infiltrating water. However, gas-phase velocities rapidly diminish after the source is turned off. For comparison, gas-phase velocity vectors are approximately 100 times smaller after 6 mo, and 1000 times smaller after 10 yr than the steady-state initial condition shown in Fig. 5c. As shown, infiltration in the unsaturated zone is very effective at attenuating CO2 plumes that are not being continuously replenished by leakage from below for this particular conceptual model.
View larger version (68K):
[in this window]
[in a new window]
|
Fig. 11. Shading indicates the mass fraction of CO2 in the gas phase and labeled contour lines indicate the water saturation for the case of zero leakage and an initial CO2 plume present in the unsaturated zone. (a) t = 6 mo, (b) t = 1 yr, (c) t = 5 yr, and (d) t = 10 yr. The maximum vector size represents a value of approximately (a) 2.8 x 10-2, (b) 2.0 x 10-2, (c) 7.5 x 10-3, and (d) 3.9 x 10-3 m d-1.
|
|
In Table 2 we summarize the flow rates, attenuation efficiencies, storage rates, and inventories of the unsaturated zone as a function of the three different leakage rates for the base case. The attenuation efficiency is defined as one minus the ratio of CO2 seepage rate to the CO2 leakage rate when the seepage rate reaches an approximate steady state, in this case after 1000 yr. As shown in Table 2, the seepage rate increases nonlinearly with the leakage rate. The reason for this is that smaller leakage rates result in more lateral spreading and CO2 capture by downward-moving infiltration relative to larger leakage rates. The physical process limiting the lateral spreading of CO2 is the downward advection of CO2 dissolved in infiltrating water, and the subsequent discharge of this CO2 through the lower hydrostatic boundary condition. For the lowest leakage rate of 4 x 104 kg yr-1, the attenuation efficiency of the unsaturated zone is 96%, while for the highest leakage rate of 4 x 106 kg yr-1, the attenuation efficiency is 19%. The attenuation efficiency of the unsaturated zone decreases with increasing leakage rate because the higher pressures surrounding the source zone force more vertical flow that directly causes seepage of CO2 above the source zone. We note also from Table 2 that the amount of CO2 crossing the lower boundary increases as the leakage rate increases, but at a fraction of the rate of increase of the seepage. Essentially, the downward infiltrating water cannot transport all of the CO2 being introduced to the system and increasing amounts end up seeping out of the ground. The rate at which CO2 is being stored in the aqueous and gas phases at 1000 yr is defined as the leakage plus infiltration minus the seepage and minus the bottom boundary loss. For small leakage rates, the ratio of seepage to storage is very small, showing the effectiveness of the unsaturated zone in attenuating leakage migration to the surface. For higher leakage rates, the ratio of seepage to storage increases, showing the ineffectiveness of the unsaturated zone in attenuating high leakage fluxes. As for the inventories, note again that the total amounts present in the aqueous and gas phases increase as some fractional power of the leakage rate.
View this table:
[in this window]
[in a new window]
|
Table 2. Carbon dioxide flow rates, attenuation efficiencies, and inventories after 1000 yr of leakage for various leakage rates.
|
|
|
CONCLUSIONS
|
|---|
This work focused on determining the potential of the unsaturated zone to attenuate CO2 from leaking geologic carbon sequestration sites. The conceptual model involved CO2 reaching the water table at a constant rate, for example through a potential high-permeability zone with direct connection to the target formation. We use the maximum seepage flux of CO2 across the ground surface as well as the maximum mole fraction of CO2 just below the ground surface as rough indicators of exposure risk. The potential risk posed was put into perspective by comparing results to both a typical ecological flux of CO2 taken as 4.4 x 10-7 kg m-2 s-1 as well as the mole fraction of CO2 (0.3) in the gas phase in soil at which tree mortality has been shown to occur.
Results from the unsaturated zone conceptual model indicate that the source zone leakage rate combined with source zone radius have the greatest influence on the maximum seepage flux of CO2 as well as the maximum mole fraction of CO2. Once the CO2 reaches the unsaturated zone, it forms a gas phase that is denser than that of the ambient soil gas. Pressure gradients between the source zone at the water table and the ground surface easily overcome this density contrast for all leakage rates tested, causing CO2 to discharge at the ground surface. The pressure driving force is significantly reduced by increasing the source zone radius, but not by adjusting hydrogeological properties of the unsaturated zone, such as the permeability, anisotropy, infiltration rate, and porosity.
The unsaturated zone conceptual model was used to calculate the steady-state seepage rate for a prescribed leakage rate. For the lowest leakage rate of 4 x 104 kg yr-1, the unsaturated zone attenuated 96% of the CO2 after 100 yr. For the highest leakage rate of 4 x 106 kg yr-1, the attenuation efficiency of the unsaturated zone decreased substantially to 19%. The attenuation efficiency of the unsaturated zone decreased with increasing leakage rate because the higher pressures surrounding the source zone caused more vertical migration of the CO2 relative to lateral migration, which is more strongly affected by infiltration.
Barometric pumping has a negligible effect on the time-averaged seepage flux and near-surface CO2 concentration because of the cyclic nature of the pressure-induced flows. For a CO2 plume present in the unsaturated zone with no continuous replenishment, dissolution into infiltrating water causes relatively rapid attenuation.
|
ACKNOWLEDGMENTS
|
|---|
We thank Karsten Pruess and Christine Doughty (LBNL) for constructive review comments, and Sally Benson, Marcelo Lippmann, Robert Hepple, and Preston Jordan (all LBNL) for stimulating discussions that helped focus the study. This work was supported in part by a Cooperative Research and Development Agreement (CRADA) between BP Corporation North America, as part of the CO2 Capture Project (CCP) of the Joint Industry Program (JIP), and the U.S. Department of Energy (DOE) through the National Energy Technologies Laboratory (NETL), and by the Office of Science, U.S. Department of Energy under contract DE-AC03-76SF00098.
|
REFERENCES
|
|---|
- Allison, M.L. 2001. Hutchinson, Kansas: A geologic detective story. Geotimes 46(10):14–19.
- Amundsen, R.G., and E.A. Davidson. 1990. Carbon dioxide and nitrogenous gases in the soil atmosphere. J. Geochem. Explor. 38:13–41.
- Bachu, S., W.D. Gunter, and E.H. Perkins. 1994. Aquifer disposal of CO2—Hydrodynamic and mineral trapping. Energy Convers. Manage. 35(4):269–279.
- Baldocchi, D.D., and K.B. Wilson. 2001. Modeling CO2 and water vapor exchange of a temperate broadleaved forest across hourly to decadal time scales. Ecol. Modell. 142:155–184.
- Chung, T.-H., M. Ajlan, L.L. Lee, and K.E. Starling. 1988. Generalized multiparameter correlation for nonpolar and polar fluid transport properties. Ind. Eng. Chem. Res. 27:671.
- Cramer, S.D. 1982. The solubility of methane, carbon dioxide, and oxygen in brines from 0°C to 300°C, U.S. Bureau of Mines: Report No. 8706.
- Farrar, C.D., M.L. Sorey, W.C. Evans, J.F. Howle, B.D. Kerr, B.M. Kennedy, C.-Y. King and J.R. Southon. 1995. Forest-killing diffuse CO2 emissions at Mammoth Mountain as a sign of magmatic unrest. Nature 376:675–678.
- Farrar, C.D., J.M. Neil, and J.F. Howle. 1999. Magmatic carbon dioxide emissions at Mammoth Mountain, California. U.S. Geol. Surv. Open-File Rep. 98-4217. U.S. Geol. Surv., Reston, VA.
- Hepple, R., and S.M. Benson. 2002. Implications of surface seepage on the effectiveness of geologic storage of carbon dioxide as a climate change mitigation strategy. p. 1–4. In Proceedings of the Sixth International Conf. on Greenhouse Gas Technologies. GHGT-6. Kyoto, Japan. October 2002. Elsevier Science, New York.
- Holloway, S. 1997. Safety of the underground disposal of carbon dioxide. Energy Convers. Manage. 38(Suppl.):S241–S245.
- International Energy Agency. 1997. Carbon dioxide utilization. IEA Greenhouse Gas R&D Programme. IEA, Paris.
- Kabwe, L.K., M.J. Hendry, G.W. Wilson, and J.R. Lawrence. 2002. Quantifying CO2 fluxes from soil surfaces to the atmosphere. J. Hydrol. 260:1–14.
- Lindeberg, E. 1997. Escape of CO2 from aquifers. Energy Convers. Manage. 38(Suppl.):S235–S240.
- National Institute of Occupational Safety and Health. 1976., Criteria for a recommended standard. Occupational exposure to carbon dioxide. NIOSH Publ. 76-194. U.S. GPO, Washington, DC.
- National Institute of Science and Technology. 1992. NIST Database 14 Mixture Property Database. Version 9.08. U.S. Department of Commerce, Washington, DC.
- Oldenburg, C.M., S.W. Webb, K. Pruess, and G.J. Moridis. 2003. Mixing of stably stratified gases in subsurface reservoirs: A comparison of diffusion models. Transp. Porous Media (In press).
- Poling, B.E., J.M. Prausnitz, and J.P. O'Connell. 2001. The properties of gases and liquids. 5th ed. McGraw Hill, New York.
- Pruess, K., C. Oldenburg, and G. Moridis. 1999. TOUGH2 user's guide. Version 2.0. Report LBNL-43134. Lawrence Berkeley National Laboratory, Berkeley, CA.
- Reichle, D., et al. 1999. Carbon sequestration research and development. DOE/SC/FE-1. U.S. Department of Energy, Washington, DC.
- Sorey, M., B. Evans, M. Kennedy, J. Rogie, and A. Cook. 1999. Magmatic gas emissions from Mammoth Mountain. Calif. Geol. 52(5):4–16.
- van Genuchten, M.Th. 1980. A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44:892–898.[Abstract/Free Full Text]
- Vargaftik, N.B. 1975. Tables on thermophysical properties of liquids and gases. 2nd ed. Hemisphere Publishing Corp., London.
- Webb, S.W. 1998. Gas-phase diffusion in porous media—Evaluation of an advective–dispersive formulation and the dusty-gas model for binary mixtures. J. Porous Media 1:187–199.
This article has been cited by other articles:
|
|
|
|
 
C. M. Oldenburg and A. J. A. Unger
Coupled Vadose Zone and Atmospheric Surface-Layer Transport of Carbon Dioxide from Geologic Carbon Sequestration Sites
Vadose Zone J.,
August 1, 2004;
3(3):
848 - 857.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
Y. Zhang, C. M. Oldenburg, and S. M. Benson
Vadose Zone Remediation of Carbon Dioxide Leakage from Geologic Carbon Dioxide Sequestration Sites
Vadose Zone J.,
August 1, 2004;
3(3):
858 - 866.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
