Article pubs.acs.org/est
Multiwell CO2 Injectivity: Impact of Boundary Conditions and Brine Extraction on Geologic CO2 Storage Efficiency and Pressure Buildup Jason E. Heath,†,* Sean A. McKenna,‡,∥ Thomas A. Dewers,† Jesse D. Roach,§ and Peter H. Kobos§ †
Department of Geomechanics, Sandia National Laboratories, Albuquerque, New Mexico 87815-0750, United States Department of Geoscience Research and Applications, Sandia National Laboratories, Albuquerque, New Mexico 87815-0735, United States § Department of Earth Systems Analysis, Sandia National Laboratories, Albuquerque, New Mexico 87815-1137, United States ‡
S Supporting Information *
ABSTRACT: CO2 storage efficiency is a metric that expresses the portion of the pore space of a subsurface geologic formation that is available to store CO2. Estimates of storage efficiency for large-scale geologic CO2 storage depend on a variety of factors including geologic properties and operational design. These factors govern estimates on CO2 storage resources, the longevity of storage sites, and potential pressure buildup in storage reservoirs. This study employs numerical modeling to quantify CO2 injection well numbers, well spacing, and storage efficiency as a function of geologic formation properties, open-versus-closed boundary conditions, and injection with or without brine extraction. The set of modeling runs is important as it allows the comparison of controlling factors on CO2 storage efficiency. Brine extraction in closed domains can result in storage efficiencies that are similar to those of injection in open-boundary domains. Geomechanical constraints on downhole pressure at both injection and extraction wells lower CO2 storage efficiency as compared to the idealized scenario in which the same volumes of CO2 and brine are injected and extracted, respectively. Geomechanical constraints should be taken into account to avoid potential damage to the storage site.
■
but nonzero permeabilities,1 which can lessen pressure buildup and provide for a larger accessible fraction of porosity as opposed to closed-boundary systems. Zhou and Birkholzer9 argue that pressure buildup in a CO2 injection scenario will not be as high as that demonstrated by Ehlig-Economides and Economides8 due to leakage into adjacent formations and migration of brine from semiopen reservoirs. Pressure management can thus directly influence efficiency of CO2 storage. Buscheck et al.6 examine CO2 injection and simultaneous brine extraction into a single well in flat and sloping reservoirs. Their results indicate marked benefits in terms of controlling the size and location of CO2 plumes and mitigation of pressure buildup. Bergmo et al.10 examine “passive” brine extraction (i.e., extraction of brine driven only by CO2 injection and not by active pumping) and conclude that, to utilize more than 1−2% of the pore space for CO2 storage, it is necessary to produce “massive” amounts of brine from the formation. Kobos et al.11 investigated “active” brine extraction while storing CO2 from a 1848 MW coal-based power plant and found that approximately 6% of the power plant’s cooling water requirements could be met from the extracted and treated saline waters. Birkholzer et al.12 demonstrate use of brine extraction to minimize pressure
INTRODUCTION Research on large-scale geologic CO2 storage has generated opposing viewpoints on the assumptions that govern estimates on the size of CO2 storage resources, the longevity of storage sites, and potential buildup of reservoir pressure.1−4 The storage estimates depend on many factors, including geologic formation properties, pressure interference of multiple wells in the same formation, well placement patterns, and whether the formation is open or closed.1,5−7 Open or closed boundaries is a key point of interest, which refers to behavior of the vertical and lateral boundaries of the storage reservoir with regard to continuity of pressure and fluid flow. Ehlig-Economides and Economides8 suggest that CO2 injection in a closed-boundary system will not permit storage of significant amounts of CO2 due to rapid pressure buildup. The validity of this suggestion may depend on the size and fluid-flow properties (e.g., absolute and relative permeability) of the storage reservoir relative to the project lifetime. Furthermore, in an open-boundary scenario, multiple injection wells may impose pressure interference on interior wells and could possibly create an effectively closedboundary system for these wells;3 thus, careful site planning and operations are needed to avoid overly close well spacing and concomitant pressure interference. Cavanagh et al.1 emphasize that boundaries of large-scale, regionally extensive CO2 storage reservoirs should be treated as open. The concept is that no-flow boundaries are more suited to small-scale hydrocarbon applications. Regional flow typically involves shale or other rocks that surround the reservoir that may have low © 2013 American Chemical Society
Received: Revised: Accepted: Published: 1067
April 18, 2013 July 27, 2013 August 23, 2013 August 23, 2013 dx.doi.org/10.1021/es4017014 | Environ. Sci. Technol. 2014, 48, 1067−1074
Environmental Science & Technology
Article
closed-boundary case at the end of a 30-year injection period (see the Supporting Information for volumetric injectionextraction information). This 1:1 injection-extraction ratio is a base case that has been previously used;6 however, we revisit this 1:1 assumption from the perspective of wellbore damage in the Results and Discussion. Both infinite-acting open and closed reservoirs are considered by allowing the sides of the domain to be held at hydrostatic pressure or no flow, respectively. Gravity override, heterogeneity, and salinity are neglected. The other assumptions on boundary conditions and multiphase flow are the focus of this study. Mesh refinement (e.g., the approximate minimum size of a mesh cell) is restricted to the size of a large wellbore diameter of approximately 1 m as opposed to realistic wellbore sizes (e.g., 0.0254 m or 6 in.), as this affects only initial very early time results of the simulations and has a negligible impact on the overall results. Table 1 presents geologic properties, domain dimensions, and multiphase flow parameters used in the simulations, where k is permeability; ϕ is porosity; Slr is residual brine saturation; and Sgr is residual gas saturation. For each permeability and porosity pair, four simulation runs were completed for open and closed domain boundaries, and with or without brine extraction, totaling 36 runs for the 1:1 volumetric injectionextraction ratio. All of the simulation runs have the domain size of 256 km2 (i.e., 16 km on a side). Particular combinations of permeability and porosity were chosen to result in CO2 plume sizes that would be near to the brine extraction wells after the 30-year project lifetime for a given number of wells. The goal was to obtain large CO2 plumes without allowing the CO2 to reach the brine extractors. The permeability and porosity pairs were systematically and iteratively chosen using simple analytical equations and then testing the pairs and the resultant plume sizes in preliminary numerical simulations (for details, see the Supporting Information). These permeability-porosity pairs were developed for the case of closed-boundaries and CO2 injection with simultaneous brine extraction and then applied to the cases of injection-only and open boundaries conditions for comparison of storage efficiency factors. Capillary pressure is implemented using van Genuchten‘s20 formulation, fitted to experimental characteristic curves for the Mount Simon Sandstone (Illinois Basin) as given by Heath et al.21 Relative permeability follows Corey22 functions, based on a residual brine value Slr of the capillary pressure fitting (see Heath et al.21). For different combinations of permeability and porosity (see Table 1), we employ Leverett scaling for capillary pressure through varying only the Slr parameter. Efficiency Calculations. There are a large number of working definitions of storage efficiency.23−27 One widely used approach for storage in deep saline (i.e., brine-containing) reservoirs is based on the following equation:23
buildup at specific locations, such as near a fault. Brine extraction, therefore, can dramatically affect cost estimates and performance efficiency of large-scale CO2 storage.13,14 Using numerical simulation, this study investigates storage efficiency of multiwell CO2 injection. The objective is a quantitative comparison of the impact of boundary conditions and formation properties on the storage efficiency. The set of simulations is broad to incorporate: open versus closed boundaries; CO2 injection with or without brine extraction; a volumetric injection-extraction ratio of one; and a volumetric injection-extraction ratio affected by geomechanical constraints on downhole brine extraction pressure. Geomechanical considerations for prevention of wellbore damage have typically been addressed for CO2 injection, but have been neglected for brine extraction.6,15
■
MATERIALS AND METHODS General Approach. We investigate quantitative relationships between CO2 storage efficiency, CO2 injection rates, absolute and relative permeability, porosity, capillary pressure, well spacing, well numbers, the position of an injection well in the multiwell pattern, and domain size (i.e., storage reservoir size) using numerical simulations with TOUGH2-ECO2N.16,17 The conceptual model comprises a two-dimensional square, tabular, homogeneous, isotropic reservoir with over and underlying impermeable strata. CO2 injector and brine extractor wells are placed in an inverted five-spot (i.e., four extractor wells at the corners of a square area with an injector well in the center; see Figure 1). Numerical mesh generation
Figure 1. Map view of well patterns and meshes for numerical simulation of CO2 injection and brine extraction. From left to right, the well patterns are 3 × 3, 5 × 5, and 7 × 7. Triangles are CO2 injectors, whereas circles are brine extractors. The dotted-gray lines on the top row highlight inverted five-spot well placement. The same meshes are used for CO2-injection-only cases or those in which no brine is extracted.
used Voronoi tessellation with refinement near wells (Figure 1). Increasing numbers of wells are added in 3 × 3, 5 × 5, and 7 × 7 patterns (Figure 1). We specify a constant pressure at CO2 injectors of 90% of a fracture gradient, which is the maximum recommended by the Interstate Oil and Gas Compact Commission.18 Fracture pressure at the injection depth is obtained by assuming a fracture gradient of 0.0147 MPa/m (0.65 psi/ft), which is relevant for the Mount Simon Sandstone in the Illinois Basin19 and the depth of the top of the reservoir (2134 m). The initial reservoir pressure is set to hydrostatic pressure at the depth of the vertical midpoint of the reservoir (i.e., 21.74 MPa). Using trial and error, the pressure at the brine extractors is set to 11.2 MPa in order to achieve approximately 1:1 volumetric CO2 injection and brine extraction for the
GCO2 = A t hg ϕtotρEsaline
(1)
where GCO2 is the CO2 storage resource mass estimate; At is the total gross area of the formation; hg is the gross formation thickness; ϕtot is total porosity; ρ is CO2 density; and Esaline is the storage efficiency factor, which is the fraction of the total pore volume that will be occupied with the injected free-phase CO2. The Esaline parameter is a function of many geologic factors, which may include: the net-to-total area or fraction of the total regional area available for CO2 storage; the net-togross ratio of the formation thickness that can accept the CO2; 1068
dx.doi.org/10.1021/es4017014 | Environ. Sci. Technol. 2014, 48, 1067−1074
Environmental Science & Technology
Article
Table 1. Parameters Used in Numerical Simulations of CO2 Injectiona simulation cases for specific perm. and porosity pairs
side of square domain (km)
well spacing (m)
number of CO2 wells
k (mD)
1 2 3 4 5 6 7 8 9
16.0 16.0 16.0 16.0 16.0 16.0 16.0 16.0 16.0
4000.0 2666.7 2000.0 4000.0 2666.7 2000.0 4000.0 2666.7 2000.0
5 13 25 5 13 25 5 13 25
95.1 39.0 20.9 158.5 65.0 34.8 31.7 13.0 7.0
k (m2) 9.4 3.9 2.1 1.6 6.4 3.4 3.1 1.3 6.9
× × × × × × × × ×
10−14 10−14 10−14 10−13 10−14 10−14 10−14 10−14 10−15
ϕ
Slr
Sgr
0.15 0.15 0.15 0.25 0.25 0.25 0.05 0.05 0.05
0.31 0.33 0.35 0.31 0.33 0.35 0.31 0.33 0.35
0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25
a
For each of the nine simulation cases of specific permeability and porosity combinations, four runs each were performed for open and closed domain boundaries, and with or without brine extraction. Other capillary pressure parameters that are the same for each simulation include: λ = 0.41; 1/P0 = 3.33 × 10−5 Pa−1; and Sls = 1 [see Heath et al.21]. The temperature for each run is 58°C. The reservoir thickness is 60 m. The reservoir top depth is 2134 m.
Figure 2. CO2 saturations at 30 years for simulation cases using 3 × 3, 5 × 5, and 7 × 7 well patterns, open or closed boundaries, and with or without brine extraction at a constant domain size of 16 km2. These correspond with the parameter values of the first three rows of Table 1, which have increasingly lower permeability for higher numbers of wells, and a constant porosity of 0.15.
domains used in this study to result in the global and local storage efficiencies are, respectively: (1) the global domain equal to the entire numerical simulation domain; and (2) the local domain of the five-spot defined as the square domain around a CO2 injection well where the four closest brine extraction wells (or those locations for CO2-injection-only cases) are the corners of the square (see Figure 1). These efficiency calculations derive from simple volumetric calculations and do not include displacement processes due to heterogeneity or density and mobility differences, which are typically termed areal, vertical (geologic layering), and gravity displacement efficiency;23,28 however, these calculations include residual brine and gas saturations (i.e., the microscopic displacement efficiency). 23 Consistent application of these efficiency calculations to all cases studied herein facilitates direct comparison of the results with regard to boundary conditions and the impact of brine extraction and well placement.
the effective-to-total porosity; and the areal, vertical, and microscopic displacement.23 Herein we use two different way of calculating the storage efficiency factor, which depend on global and local domains. The basis of these efficiency calculations is the sum of CO2 volume contained within the pore volume, Vnorm, of the normalizing domain at each output time: N
E (t ) =
∑ i=1
SCO2(i , t ) × V (i) × ϕ(i) Vnorm
(2)
where N is the number of grid cells within the normalizing pore volume (i.e., the pore volume of the domain of interest); SCO2 is the fractional CO2 saturation at output time t; and V and ϕ are the volume and porosity, respectively, of the ith cell within the normalizing pore volume. E(t) represents the same quantity as Esaline of eq 1, namely the fraction of the total pore volume of a domain that is filled with CO2, but reflects our specific methods and emphasizes the time dependence. The two normalizing 1069
dx.doi.org/10.1021/es4017014 | Environ. Sci. Technol. 2014, 48, 1067−1074
Environmental Science & Technology
Article
Table 2. Global Storage Efficiencies at the 30-Year-Time Limit for the Different Well Patterns and Scenariosa global storage efficiency factor injection only simulation cases for specific porosity and perm. pairs 1 2 3 4 5 6 7 8 9 a
k (m2) 9.4 3.9 2.1 1.6 6.4 3.4 3.1 1.3 6.9
× × × × × × × × ×
10−14 10−14 10−14 10−13 10−14 10−14 10−14 10−14 10−15
ϕ 0.15 0.15 0.15 0.25 0.25 0.25 0.05 0.05 0.05
injection and extraction
well pattern
closed boundaries
open boundaries
closed boundaries
open boundaries
× × × × × × × × ×
0.0046 0.0045 0.0045 0.0046 0.0045 0.0045 0.0046 0.0045 0.0045
0.21 0.15 0.11 0.21 0.15 0.11 0.21 0.15 0.11
0.17 0.21 0.23 0.17 0.21 0.23 0.17 0.21 0.23
0.30 0.29 0.29 0.30 0.29 0.29 0.30 0.29 0.29
3 5 7 3 5 7 3 5 7
3 5 7 3 5 7 3 5 7
Some columns of Table 1 are repeated for easy reference of the run information for each global storage efficiency factor.
■
RESULTS AND DISCUSSION Figure 2 presents CO2 saturation after 30 years of injection for the 3 × 3, 5 × 5, and 7 × 7 well patterns. The choice of permeability has resulted in CO2 plumes that are near to the brine extractors for closed-boundary extraction (see Figure 1 for well locations). This was a goal of the permeability-porosity pairs that were chosen, as discussed in the Materials and Methods. The CO2 saturation maps indicate that differences in CO2 saturation between open and closed boundaries become negligible under the case of brine extraction as the number of wells increases (see the third and fourth columns of Figure 2). These results are corroborated in Table 2 by the global storage efficiency values taken at 30 years. These results show that adding brine extraction is essentially equivalent to injection into an open-boundary system. Global efficiencies vary by only a factor of 2 or less between the open boundary injection-only case and the closed boundary injection and extraction case, where the volumetric CO2 injection-to-brine extraction ratio is approximately 1:1. In contrast, for closed boundary cases, global efficiencies increase by a factor of 37−51 with the addition of brine extraction. The open boundary, injection-only case (column two of Figure 2) illustrates improved CO2 injectivity, where CO2 saturation increases from the absolutecenter well to the corner wells; thus, with no brine extraction, a greater number of wells leads to pressure build-up for the interior wells and thus lower injection rates. Global efficiencies calculated here range from less than one percent for closed boundaries and no extraction to between 11−30% for cases with open boundaries and/or brine extraction (Table 2). Global storage efficiencies increase as a function of well spacing for the injection-only open-boundary cases. At the largest well spacing with the fewest number of wells (3 × 3 pattern), the impact of pressure interference is the least. As the well spacing decreases and the number of wells increase, only the wells near the open boundaries maintain relatively high CO2 injection rates (see Figure 2, column 2). The storage efficiencies decrease for injection-extraction-closedboundary cases as a function of well spacing (Figure 3). This reflects the dissection of the domain with greater numbers of wells and hence lower storage efficiencies. The injection-only closed-boundary case displays constant storage efficiency with a change in well spacing. The pressure interference with no pressure sink (i.e., no open boundaries nor extractors) results in the same storage efficiency regardless of the well spacing. Reported efficiencies in open boundary systems (or systems
Figure 3. Global storage efficiency factor as a function of well spacing for the nine base cases of Figures 2 and 3.
assumed to be managed to behave as open boundary systems23) typically range from less than one percent up to four percent29−32 as summarized by Cavanagh et al.1 and more recently estimated by Goodman et al.23 The difference between these reported values and the study herein is due to our implementation of E(t) (eq 2). Our approach maximizes storage efficiency for the brine extraction case. The permeability-porosity pairs were chosen to result in fully developed plumes (i.e., plumes near to the brine extractors) after the 30-year injection period. This maximized the storage efficiency. Our approach also does not incorporate net-to-gross thickness or gravity override of the CO2 plume over the aqueous phase. Our storage efficiency estimates thus represent an upper bound on the highest achievable storage efficiencies. Reservoir pressure behavior explains differences in global and local storage efficiencies. At 30 years, the CO2-injection-only cases with open boundaries, or closed-boundary cases with brine extraction, exhibit pressure gradients that correspond to continued CO2 injection (Figures 4 and 5). However, pressure buildup for the CO2-injection-only cases occurs relatively quickly and reaches the pressure of the injectors (Figure 5a), at which time the global storage efficiency becomes constant (Figure 5b). Either open boundary conditions or simultaneous brine extraction or the combination of the two is able to maintain average reservoir pressures after 30 years at levels well below the fracture gradient (Figure 5a). CO2 injection into a closed system with no brine extraction will increase average pressures to the fracture gradient for chosen simulation volumes. 1070
dx.doi.org/10.1021/es4017014 | Environ. Sci. Technol. 2014, 48, 1067−1074
Environmental Science & Technology
Article
Figure 4. Reservoir pressure at 30 years for simulation cases using 3 × 3, 5 × 5, and 7 × 7, open or closed boundaries, and with or without extraction. These results correspond to the same simulations of Figure 1. Hydrostatic pressure for the 2D simulations is 21.7 MPa.
Figure 5. Time series of (a) average reservoir pressure and (b) global storage efficiency for the 12 base case runs (those of Figures 1 and 2 at constant porosity of 0.15, and the first three permeability and porosity pairs of Table 1). The 3 × 3, 5 × 5, and 7 × 7 cases are shown in black, dark gray and light gray, respectively. Open and/or closed symbols denote open and/or closed boundaries. Solid lines are for injection only and dashed for injection and extraction. The fracture pressure at the top depth of the reservoir is shown by the horizontal dash-dot line. The legend of part b also applies to part a.
Variation in local storage efficiency between well types (i.e., CO2 injection wells at different locations in a well pattern) are greatest under open boundary conditions where corner injection wells have the highest efficiencies. Local efficiency values calculated at the 30-year limit are shown in Figure 6a−b for each scenario for center and corner injectors. For the closed-boundary system with brine extraction, injection storage efficiencies are a function of the relative position of the injection well within the well pattern. At early times, the variation in efficiencies across relative positions is minimal with efficiency variations dominated by variations due to grid cell shapes and sizes. At times longer than 20 years, the relative well positions in this closed-extraction scenario cause large deviations in the resulting efficiencies. The interior wells have the highest efficiencies with edge well efficiencies being slightly lower and corner wells having the lowest efficiencies, nearly
The lowest average pressures occur with open boundaries and brine extraction (see the black, dark gray, and light gray dashed curves with open symbols in Figure 5). The highest average pressures occur for 3 × 3, 5 × 5, and 7 × 7 well patterns with closed boundaries and no extraction, with almost identical pressure buildup profiles. The other cases of only CO2 injection with open boundaries, or CO2 injection in closed boundaries with extraction, exhibit distinct intermediate pressure buildup profiles that never reach the fracture gradient. Pressure history also affects CO2 injection rate as a function of time. For the injection-only closed case, injection rate drops dramatically in response to pressure-buildup. The other cases have similar initial injection rates, but are more sustained and increase in time for open and/or brine extraction cases. This is shown in Figure S1 in the Supporting Information. 1071
dx.doi.org/10.1021/es4017014 | Environ. Sci. Technol. 2014, 48, 1067−1074
Environmental Science & Technology
Article
Figure 6. (a and b). Local storage efficiencies at 30 years for all cases (open and/or closed; extraction and/or no-extraction; and 3 × 3, 5 × 5, and 7 × 7 well patterns) also showing the variations in injection well types (absolute center and corner), for the constant porosity of 0.15 (see rows 1−3 in Table 1). (c and d). Percent change of storage efficiencies for cases in parts a and b, but using a higher extractor pressure of 17.7 MPa to avoid wellbore damage.
below a critical value given by (see the work by Haimson and Herrick;33 and Moos and Zoback,34 their Figure 4a)
10% lower than the interior well values. These results are due to the proximity of the closed boundaries to the corner and edge injection wells. Efficiencies plotted in Figure 6a and b are extremely low and independent of relative injection well location for the case of closed boundaries with no brine extraction. The highest efficiencies obtained under the openboundary injection-only case occur at the corner injection wells where the injected CO2 does not experience as much interference with other CO2 pressure plumes relative to the edge and center injection wells. This result is contrary to edge (not shown) and absolute-center injection wells that achieve higher efficiencies with closed boundaries and brine extraction. The variation in local efficiency between the different well patterns is of secondary importance compared to overall variation between different boundary and extraction scenarios. One potential issue with the 1:1 volumetric CO2 injection-tobrine extraction ratio, given the chosen well placement patterns, is the large under-pressure or “cone-of-depression” that results. For example, in Figure 2 columns three and four, a 1:1: volumetric injection-extraction ratio results in a significant cone-of-depression with a flowing bottom hole pressure of ∼11.2 MPa, which is much lower than hydrostatic pressure of ∼21.7 MPa. One concern with a pressure drop of this magnitude is wellbore damage due to the changing stress conditions. A simple analysis of borehole breakouts in wellbores within a three-dimensional stress field suggests that breakouts and/or well sanding would occur parallel to the direction of the least principal compressive stress if wellbore pressures drop
Pcrit
