Grantee Research Project Results
2011 Progress Report: Understanding and Managing Risks Posed by Brines Containing Dissolved Carbon Dioxide
EPA Grant Number: R834383Title: Understanding and Managing Risks Posed by Brines Containing Dissolved Carbon Dioxide
Investigators: Falta, Ronald W. , Benson, Sally M. , Murdoch, Lawrence C.
Current Investigators: Falta, Ronald W. , Murdoch, Lawrence C. , Benson, Sally M.
Institution: Clemson University , Stanford University
EPA Project Officer: Aja, Hayley
Project Period: November 1, 2009 through October 31, 2012 (Extended to October 31, 2014)
Project Period Covered by this Report: November 1, 2010 through November 1,2011
Project Amount: $891,342
RFA: Integrated Design, Modeling, and Monitoring of Geologic Sequestration of Anthropogenic Carbon Dioxide to Safeguard Sources of Drinking Water (2009) RFA Text | Recipients Lists
Research Category: Targeted Research , Water
Objective:
Background:
Geologic disposal of supercritical carbon dioxide in saline aquifers and depleted oil and gas fields will cause large volumes of brine to become saturated with dissolved CO2 at concentrations of 50 g/l or more. As CO2 dissolves in brine, the brine density increases slightly. This property favors the long-term storage security of the CO2 because the denser brine is less likely to move upwards towards shallower depths. In fact, one proposed strategy for reducing risk from CO2 injection activities involves pre-dissolving the CO2 into brine at the surface, and injecting this brine into the disposal formation. While dissolved phase CO2 poses less of a threat to the security of shallower drinking water supplies, the risk is not zero. There are plausible mechanisms by which the CO2 laden brine could be transported to a shallower depth, where the CO2 would come out of solution (exsolve), forming a mobile CO2 gas phase. This significant mechanism for drinking water contamination has received little attention, and there are basic science and reservoir engineering questions that need to be addressed in order to reduce risks to underground drinking water supplies.
Research Approach: Six main activities were identified in the research proposal.
- Laboratory Experiments. Laboratory core experiments that flood cores with CO2 saturated brines at reservoir pressure and temperature. These cores are then gradually depressurized, and imaged using a medical CT scanner to study the CO2 phase evolution and movement.
- Pore-Scale and Core-Scale Modeling. Pore-scale multiphysics and multiphase continuum modeling of these experiments are used to develop a fundamental understanding of the exsolution and CO2 bubble coalescence phenomena as the CO2 starts to form a mobile phase.
- CO2 Phase Relative Permeability Functions for Multiphase Flow Models. Core-scale multiphase continuum modeling to upscale the experimental results with a focus on developing effective relative permeability functions for use in field scale modeling.
- Regional-Scale Variable Density Groundwater Modeling. Simulations of regional scale behavior using a variable density groundwater flow model. The simulations are designed to evaluate the effects of upward hydraulic gradients, upward pressure driven brine flow (for CO2 brine injection projects) and most importantly, the effects of groundwater pumping from shallower aquifers.
- Multiphase Flow Simulations of Field Scale CO2 Injection. Local field scale (hundreds to thousands of meters) multiphase simulations of the likely failure modes using realistic hydrogeologic and geologic conditions that are representative of CO2 storage locations.
- Remediation Designs. These models will then be used to study remediation strategies and alternative storage methods for each CO2 release scenario.
Progress Summary:
During our second year on this project, we have made very good progress on Activities (1), (3), (5) and (6). We have begun work on Activities (2), (4), and (6). So far, we have published one journal paper and three MS theses on this project. We have 4 journal manuscripts that are nearly ready for submission, and are preparing one additional manuscript for submission this summer. A detailed summary of progress by activity area is listed below. Where appropriate, the relevant journal paper or portions of the MS thesis has been included.
Task 1. Laboratory Experiments
This work is nearly complete, and we have already published a journal paper on the work. We plan to conduct some additional experiments this summer to evaluate the effect of initial CO2 saturation on the trapped residual saturation. This paper was published in the January, 2012 issue of the journal Transport in Porous Media.
Task 2. Pore-Scale and Core-Scale Modeling
We are currently performing studies of CO2 exsolution in small micromodels. These models include individual pore structures, and by using a microscope, we are able to visualize the exsolution process as water saturated with CO2 is depressurized. This work is ongoing, and will be reported in detail in next years report.
Task 3. CO2 Phase Relative Permeability Functions for Multiphase Flow Models
Our work in this area so far has focused on the development and testing of new hysteretic relative permeability and capillary pressure functions for cases where supercritical or gas phase CO2 is moving through brine-filled porous media. We have completed a MS thesis on this work, and we are preparing a journal manuscript from the thesis. The relevant parts of the thesis are included below.
Task 4. Regional-Scale Variable Density Groundwater Modeling
We are using the Army Corps of Engineers Ground Water Modeling Software (GMS) package to construct detailed three-dimensional MODFLOW flow models of potential CO2 storage sites. Our initial effort is focused on the Atlantic Coastal Plain sediments in Georgia, South Carolina, and North Carolina. This work will be reported on in detail in next years report.
Task 5. Multiphase Flow Simulation of Field Scale CO2 Injection
Our efforts on this task have been directed towards evaluation of dissolved CO2/brine transport from storage formations to drinking water formations through open abandoned wells, as we (and others) view this to be the most likely CO2 release scenario. We have published a MS thesis on this, and have prepared a journal manuscript.
Task 6. Remediation Designs and Alternative Injection Schemes to Reduce Risk
Our work so far in this area has focused on a comparison of dissolved CO2 injection compared to the more conventional supercritical CO2 injection. Dissolved CO2 poses a lower escape risk compared to supercritical CO2 because it is not upwardly buoyant. However, in order to be practical, the areal footprint occupied by the dissolved CO2 should be comparable to supercritical CO2 injection.
Comparison of Sweep Efficiency Between Supercritical and Dissolved CO2 Injection
Understanding the sweep of CO2 injected into storage formations leads to the optimization of injection strategies and reduction of possible risks to drinking water aquifers. Of several proposed methods mentioned in the Year 1 Annual Report, supercritical CO2 injection and CO2 saturated brine injection have been chosen for comparison in the current work.
Findings from the previous years work have been used to construct improved numerical models. The extent of all models has been increased, while more finely refining the area around the injection wells. Hysteresis effects on relative permeability and capillary pressure functions are included in cases of supercritical CO2 injection. The areal footprint and total storage volume required for supercritical versus dissolved CO2 injection are compared. Short term and long term mobility is considered in both cases.
Methods
Using Lawrence Berkeley National Labs TOUGH2-ECO2N, local and regional scale multiphase flow models are being created to simulate CO2 injection into deep saline aquifers. The first of these flow models compares CO2 sweep efficiency following injection into a homogeneous formation. The second set of models compares CO2 sweep efficiency following injection into a stratified heterogeneous formation. The third set of models incorporates heterogeneous, random, spatially correlated permeability fields generated from Lawrence Berkeley National Labs iTOUGH2-GSLIB.
Model parameters and hydrogeologic characteristics are held constant between comparisons. The same mass of CO2 is injected in each case; therefore, a much greater total mass is injected for the dissolved CO2. To accommodate for this pressure increase, injection schemes are being designed so as not to exceed the overburden pressure of the formation. Injection is modeled for 20 years and is followed by an 80 year monitoring period.
Injection into a homogenous storage formation
A 200 meter thick homogenous storage formation is represented using a radial grid extending out from the well 200 kilometers. Pressure and temperature gradients are established between the bottom of the storage formation at 20 MPa and 50ºC and the top at 18 MPa and 46.1oC. All models are run isothermally. The formation is uniformly given a salinity of 50,000 mg/L. The van Genuchten-Mualem Model was chosen to represent relative permeabilities and the van Genuchten function was used for capillary pressure curves. In both functions, effects of hysteresis are quantified through the calculation of residual CO2 saturation as a linear function of the maximum CO2 phase saturation in each grid-block.
Relevant hydrogeologic properties of the storage formation are given in Table 1. Parameter values were determined from Birkholzer et al. [2009] and Barnes et al. [2009], which represent studies of similar natures. Hysteretic relative permeability (Figure 1a) and capillary pressure curves (Figure 1b) have been fit to match non-hysteretic saturation curve parameters given by Zhou et al. [2010]. These values are characteristic of the deep saline aquifers under consideration for carbon capture and storage activities.
Properties | Values |
---|---|
Density, ρ | 2300 kg/m3 |
Porosity, φ | 0.1289 |
Horizontal Permeabilities, kx, ky | 1.417e-13m2 |
Vertical Permeability, kz | 1.1417e-14m2 |
Residual water saturation, Slr | 0.3 |
Residual CO2 saturation, Sgr |
Approximately 200,000 tons of CO2 per year is injected along a 200 meter well screen for 20 years. The rate of CO2 injected was chosen as the maximum possible amount injected from any one well in the dissolved phase without overcoming the fracture pressure of the storage formation, set at 3.17 MPa. In the case of CO2 dissolved in brine, the brine is assumed to have the same composition as the resident fluid. In this case, a volume of brine 20 times that of the CO2 is needed for full dissolution. Results are shown in cross sections spanning the distance from the injection well at X=0 and at depth of 2 km from surface, where Z=0 is the bottom of storage formation.
Figures 1c and 1d show the sweep of the saturation of supercritical CO2 at the end of injection at 20 years and at end of a 20 year monitoring period. These cross sections span the distance from the injection well at X=0 and at depth of 2 km depth from surface, where Z=0 is the bottom of storage formation. Considering a CO2 mass fraction of 0.005 the cutoff, the farthest the supercritical CO2 travels from the well during the injection period is 750 meters. This results in a volumetric sweep efficiency of 0.10. Monitoring the CO2 mobility after injection, buoyancy and viscous forces drive the supercritical CO2 to the top of the formation. Within 20 years post injection, the CO2 travels 950 meters from the well. This results in a volumetric sweep efficiency of 0.06. At this point, there is still significant mobility in the supercritical phase.
Figures 1e and 1f show the sweep of mass fraction of dissolved CO2 at the end of injection at 20 years and at end of an 80 year monitoring period. The dissolved CO2 travels 1,280 meters at farthest and achieves a volumetric sweep efficiency of 0.63. The dissolved CO2 is denser than resident brine and sinks to travel 1,290 meters from well along the bottom of the aquifer. The sweep efficiency is reduced to 0.62. Compared to the supercritical CO2, the dissolved CO2 is essentially immobile within 80 years post-injection.
Figures 1e. and 1f. CO2 dissolved mass fraction after 20 years injection and 80 year monitoring period.
In comparison with simulations included in the Year 1 Annual Review, updated characteristic curve parameters and inclusion of hysteresis has notably changed the shape and extent of the supercritical plume.
Injection into a stratified formation
It is anticipated that vertical heterogeneity in the storage formation may improve the sweep efficiency during supercritical CO2 injection. In order to simulate this type of geology, a system of stacked aquifers and aquitards 240 meter thick of radial extent 200 kilometers is modeled. Each aquifer is 65 meters thick, intermittent aquitards are 5 meters thick, and a confining unit 40 meter thick caps the system. An injection well is screened from the base of the confining unit through all three aquifers. The base of the storage formation and below is considered to be impermeable bedrock. A radial cross section of the area surrounding the injection well is shown in Figure 2a.
Figure 2a. Model set up for injection into a stratified storage formation
Pressure and temperature gradients are established between the bottom of formation at 20 MPa and 50oC and the top at 17.6 Mpa and 45.4oC. Models are run isothermally. The formation is uniformly given a salinity of 50,000 mg/L. Aquifer properties were held the same as the case of injection into a homogenous storage formation. Relevant hydrogeologic properties of the aquitards are given in Table 2. Again, parameter values were determined from Birkholzer et al. and Barnes et al. and are characteristic of the hydrogeology around deep saline aquifers.
Hysteretic relative permeability (Figure 2b) and capillary pressure curves (Figure 2c) have been fit to match non-hysteretic saturation curve parameters given by Zhou et al. [2010]. These values are characteristic of the deep aquitards in regions under consideration for carbon capture and storage activities.
Properties | Values |
---|---|
Density, ρ | 2600 kg/m3 |
Porosity, φ | 0.059 |
Horizontal Permeabilities, kx, ky | 6.0246e-18 m2 |
Vertical Permeability, kz | 6.0456e-19 m2 |
Residual water saturation, Slr | 5.97E-02 |
Residual CO2 saturation, Sgr | 0.2 |
Figure 2b and 2c. Hysteretic characteristic curves of squitards. Fitted to Zhou et al [2010]
Approximately 200,000 tons of CO2 per year is injected along the well screen for 20 years. Figures 2d and 2d show the sweep of the saturation of supercritical CO2 at the end of injection at 20 years and at end of an 80 year monitoring period. Considering a CO2 mass fraction of 0.005 the cutoff, the farthest the supercritical CO2 travels from the well during the injection period is 700 meters. This results in a volumetric sweep efficiency of 0.12. The CO2 front has traveled 1,000 meters from the well 20 years after injection. This results in a volumetric sweep efficiency of 0.06, matching that of the homogeneous injection.
Figures 2f and 2g show the sweep of the mass fraction of dissolved CO2 at the end of injection at 20 years and at end of an 80 year monitoring period. The dissolved CO2 travels 1,280 meters at farthest and achieves a volumetric sweep efficiency of 0.63. The dissolved CO2 is denser than resident brine and sinks to travel 1,290 meters from well along the bottom of the aquifer. The sweep efficiency is reduced to 0.62.
Figure 2d and 2e. Figures 1c and 1d. CO2 saturation after 20 years injection and 80 year monitoring period.
Figure 2f and 2g. CO2 dissolved mass fraction after 20 years injection and to year monitoring period.
While the short term volumetric sweep efficiency is increased in supercritical injection, the results suggest that vertical heterogeneity will not greatly affect the long-term areal footprint of CO2 on the sealing formation.
Injection into a storage formation with heterogeneous, random, spatially correlated permeability fields
In reality, aquifers are likely to be highly heterogeneous in all directions. To increase the hydrogeologic complexity, Lawrence Berkeley National Labs iTOUGH2-GSLIB is being used to generate heterogeneous, random, spatially correlated permeability fields for these storage sites. Highly heterogeneous permeability fields promote viscous fingering and may lead to poor sweep efficiency.
A 200 meter thick heterogeneous storage formation is represented using a three dimensional Voronoi grid covering an area of 7.5 km by 7.5 km. Pressure and temperature gradients are established between the bottom of the storage formation at 20 MPa and 50oC and the top at 18 MPa and 46.1oC. All models are run isothermally. The formation is uniformly given a salinity of 50,000 mg/L.
Using parameters from Barnes et al., the intrinsic permeability is assumed to be log- normally distributed with a variance of 2.06 about an average permeability of 1.1417e-13 m2. Sequential Gaussian Simulation is used to create a spherical variogram model with a vertical range of 1 m for log permeability. A very long horizontal to vertical ratio of 10,000 : 1 is assumed for the variogram length. Figure 3a shows logarithmic isosurfaces of the permeability modifiers applied to the storage formation.
Figure 3a. Permeability modifiers applied to homogeneous storage formation.
To improve efficiency in simulating supercritical CO2 injection, the relative permeability curve is calculated by a cubic function of saturation described by Fatt and Klikoff [1959]. Matched to parameters given by Zhou et al., Figure 3b. shows the relative permeability curve for this simulation. Residual CO2 saturation is again calculated as a linear function of the maximum CO2 phase saturation in each grid-block. The capillary pressure curve expressed by the van Genuchten function shown in Figure 1b is used again.
Figure 3b. Hysteretic cubic relative to permeability function.
Figures 3c and 3d show the three-dimensional sweep of the saturation of CO2 using cross sections and isosurfaces. After 20 years of injection, the farthest the supercritical CO2 travels from the well is 1,160 meters. This results in volumetric sweep efficiency 0.04. After an 80 year monitoring period, the CO2 travels 1,400 m from the well at farthest. This results in volumetric sweep efficiency 0.03.
Figure 3c. CO2 saturation after 20 years injection.
Figure 3d. CO2 saturation after 80 year monitoring period.
Figure 3e. CO2 dissolved mass fraction after 20 years injection.
Figure 3f. CO2 dissolved mass fraction after 80 year monitoring period.
It is clear from these simulations that heterogeneity in aquifer permeability creates preferential flow paths promoting CO2 migration much farther away from wells than anticipated in simpler models. It is also notable that in cases of dissolved CO2, there is no appreciable CO2 mobility post injection.
Conclusions
To reduce the risk of CO2 leakage to drinking water aquifers, CO2 storage should be permanent sequestration and immobilization. In all cases, the dissolved CO2 has a larger radial extent but is also more evenly distributed throughout the formation. As seen in Table 3, sweep efficiency comparisons quantify the ratio of stored CO2 to available storage volume. The denser CO2 saturated brine acts with a slight downward force and quickly becomes essentially immobile. This uniform distribution and quick immobilization should be taken into account when designing CO2 injection schemes.
Homogenous | Stratified | Heterogeneous | ||||
---|---|---|---|---|---|---|
Supercritical | Dissolved | Supercritical | Dissolved | Supercritical | Dissolved | |
20 Years | 0.10 | 0.63 | 0.12 | 0.63 | 0.04 | 0.28 |
100 Years | 0.06 | 0.62 | 0.06 | 0.62 | 0.03 | 0.28 |
Future Activities:
To increase the accuracy of these sweep efficiency models, future simulations will include highly resolved grids and will be run using a parallel version of TOUGH2-ECO2N. The case of injection into a storage formation with heterogeneous, random, spatially correlated permeability fields will be run on a stratified regional scale. To reduce numerical error, the grid orientation effect of Voronoi grids versus rectangular grids will be compared in the case of supercritical CO2 injection.
Journal Articles on this Report : 1 Displayed | Download in RIS Format
Other project views: | All 16 publications | 5 publications in selected types | All 5 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Zuo L, Krevor S, Falta RW, Benson SM. An experimental study of CO2 exsolution and relative permeability measurements during CO2 saturated water depressurization. Transport in Porous Media 2012;91(2):459-478. |
R834383 (2011) R834383 (2012) R834383 (2013) R834383 (Final) |
Exit Exit |
Supplemental Keywords:
dissolved CO2, CO2 exsolution, wellbore leakageProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.