Assessment of basin-scale hydrologic impacts of CO₂ sequestration, Illinois basin

Abstract

Idealized, basin-scale sharp-interface models of CO₂ injection were constructed for the Illinois basin. Porosity and permeability were decreased with depth within the Mount Simon Formation. Eau Claire confining unit porosity and permeability were kept fixed. We used 726 injection wells located near 42 power plants to deliver 80 million metric tons of CO₂/year. After 100 years of continuous injection, deviatoric fluid pressures varied between 5.6 and 18 MPa across central and southern part of the Illinois basin. Maximum deviatoric pressure reached about 50% of lithostatic levels to the south. The pressure disturbance (>0.03 MPa) propagated 10–25 km away from the injection wells resulting in significant well–well pressure interference. These findings are consistent with single-phase analytical solutions of injection. The radial footprint of the CO₂ plume at each well was only 0.5–2 km after 100 years of injection. Net lateral brine displacement was insignificant due to increasing radial distance from injection well and leakage across the Eau Claire confining unit. On geologic time scales CO₂ would migrate northward at a rate of about 6 m/1000 years. Because of paleo-seismic events in this region (M5.5–M7.5), care should be taken to avoid high pore pressures in the southern Illinois basin.

Introduction

The Illinois basin is host to dozens of coal-fired power plants which generate about 300 million metric tons (MT) of CO₂ annually (squares in Fig. 1A). As part of the Department of Energy's Geological Sequestration Regional Partnership program, the Midwest Geological Sequestration Consortium (MGSC), a partnership that includes the three state geological surveys in the Illinois basin, is investigating the technical constraints on the injection of CO₂ into deep (>1.5 km) reservoirs under supercritical conditions (Midwest Geological Sequestration Consortium, 2005). The Mount Simon Sandstone, a Cambrian siliciclastic unit that is present throughout the basin, is a saline water filled aquifer that is a candidate reservoir for geological sequestration (Fig. 1B). At these depths, the Mount Simon pore fluids are often in excess of 100,000 mg/L (chloride concentrations of about 1700 mM; Hanor, 1987, Hanor and McIntosh, 2006). Simple back of the envelope calculation suggests that there is adequate pore space to accommodate the amount of annual CO₂ production. Assuming that CO₂ is emplaced as a supercritical fluid having a density of 500 kg/m³ into a 500 m thick Mount Simon Sandstone having an average porosity of 10%, 80 million metric tons of CO₂ (about 25% of the CO₂ produced across the basin) could displace all of the brines from an area surrounding a well field and would have an approximate aerial footprint of about 1.8 km × 1.8 km each year. Assuming a simple geometry for the Mount Simon and Illinois basin (500 km × 600 km × 500 m) and a piston-type displacement, of this volume of CO₂ would result in the displacement of brines out of the center of the Illinois basin towards its outer margins at a rate of about 0.003 m/year. This calculation does not account for more rapid solute migration along permeable pathways (e.g. Liu et al., 2004) and must be considered as a minimum estimate.

It is critical to ensure that fluid pressures near the CO₂ injection wells do not exceed 66% of lithostatic levels in order to minimize the potential of hydrofracturing of the overlying Eau Claire confining unit or inducing earthquakes (e.g. Hsieh and Bredehoeft, 1981, Nicholson and Wesson, 1990, Zoback and Harjes, 1997). More site-specific studies are required to delineate maximum pressures of hydrofracturing. In case of pre-existing fractures or faults, this value would be much smaller. This is particularly important for the southern Illinois basin which hosts active seismogenic zones including the Wabash Valley and New Madrid Seismic Zones (Fig. 2; Eager et al., 2006, Cramer et al., 1992). Carbon-14 dating of sand dikes and other liquefaction features near fault zones suggests that a series of high magnitude (M5.5–M7.5) earthquakes occurred during the Holocene across the southern Illinois basin (Wheeler and Cramer, 2002, Obermeier et al., 1992, Munson et al., 1992; Fig. 2). The largest historic earthquake on North America is associated with the Reelfoot Rift system (New Madrid Seismic Zone) whose northern margin extends into the southern Illinois basin. Eager et al. (2006) suggest that the tight clustering of recent micro-earthquakes (∼M1.1) along the Wabash Valley Seismic zone may be due to secondary injection of water associated with water floods within the Illinois basin. Several seismic events (M4.3–M2.8) are believed to be the result of 11 MPa of induced pore pressure associated with pumping of 164 m³/day of hazardous waste into a Class-I injection site in the basal Paleozoic reservoir (not the Mount Simon Formation) in Ohio (Nicholson and Wesson, 1990, Nicholson et al., 1988, Seeber et al., 2004). These authors report that these earthquakes occurred 2 km below Mount Simon formation along a previously unmapped Precambrian fault zone. What is troubling is that the induced pore pressures were far below lithostatic levels and that some of the faults within the southern Illinois basin are sub-parallel to the maximum horizontal compressive stress direction (Zoback and Hickman, 1982, Zoback and Zoback, 1989a, Zoback and Zoback, 1989b, Zoback and Zoback, 1989c).

There are reasons to be concerned that pore pressures could approach the lithostatic pressure (fracture pressure) as a result of large-scale CO₂ injection within the Cambrian Mount Simon Sandstone. First, the average core permeability of the Mount Simon reservoir is relatively low (less than 100 mD) and (as discussed below) core permeability values decrease with depth. Second, 20th century water withdrawals from the Illinois basin Cambro-Ordovican aquifer system (which includes the Mount Simon Formation; Walton, 1962) resulted in large regions (50 km by 50 km) of significant sub-hydrostatic fluid pressures around Chicago (Olcott, 1992, Visocky, 1982; Fig. 1C) due to freshwater withdrawals (Visocky, 1997; Fig. 1D). The annual withdrawals from municipal water supply wells from deep siliciclastic aquifers beneath cities of Chicago (e.g. 235 million metric tons of H₂O; Fig. 1D) are of the same order of magnitude as the mass of CO₂ generated across the Illinois basin by coal-fired power plants. Fresh water withdraws have modified regional groundwater flow patterns and resulted in a decrease of as much as 45 m (0.45 MPa) in the Mount Simon Sandstone potentiometric surface in a limited area around the Twin Cities of Minneapolis/St. Paul, and by over 182 m (1.8 MPa) in the Chicago area. Would CO₂ injection from more regionally distributed power plants result in comparable increases in heads and modification in groundwater flow directions? While CO₂ injection would presumably occur over a much more distributed area, the Mount Simon is probably less permeable to the south of Chicago where many coal-fired power plants are located.

The purpose of this study is to assess some of the hydrologic and potential seismogenic consequences of CO₂ injection into the Mount Simon Formation at the sedimentary basin-scale. While a number of studies have quantified the effects of CO₂ injection into subsurface reservoirs around the world (e.g. Chadwick et al., 2003, Hovorka et al., 2001, Lu and Lichtner, 2007), these studies have focused on localized regions (up to about 10 km by 10 km by 100 m in the x-, y-, and z-directions, respectively) around injection wells and have neglected the impact of pore pressure increase on the regional flow field. With exception of (Birkholzer et al., 2008, Birkholzer et al., 2009a, Birkholzer et al., 2009b, Birkholzer and Zhou, 2009, Nicot, 2008, Yamamoto et al., 2009), we are aware of no quantitative models that have been developed to consider the consequences of large-scale CO₂ injection at the sedimentary basin-scale. A secondary objective of this study is to demonstrate the efficacy of numerical sharp-interface representations of CO₂ injection. We propose to address several specific questions regarding basin-scale CO₂ injection in this study: Will large-scale CO₂ injection result in the generation of near lithostatic pressures in relatively low permeability regions of the Mount Simon Sandstone reservoir or will pressure anomalies be mitigated by leakage across the overlying Eau Claire confining unit? What is the likely radius of pressure disturbance away from CO₂ injection wells? Will there be significant well–well interference within- or between-injection well centers? Does CO₂ injection result in significant displacement of brines towards the margins of the basin to the North where the Mount Simon Sandstone is exploited as a water resource? Will CO₂ injection result in significant vertical brine displacement across the Eau Claire confining unit? Are there optimal regions of the basin that would likely minimize pore pressure generation and the potential for induced seismic activity?

As noted by the ground breaking studies of Birkholzer et al. (2008) and Birkholzer et al., 2009a, Birkholzer et al., 2009b some of these questions need to be addressed at the sedimentary basin-scale to ensure that unforeseen well–well interactions do not occur while others can be addressed with more localized analysis. We argue below that there are alternatives to using computationally expensive, multi-phase simulators to address these issues at the sedimentary basin-scale. Recently, Nordbotten et al., 2004, Nordbotten et al., 2005 argued that simple, analytical models based on sharp-interface theory are of great value in providing order of magnitude estimates of the hydrologic and environmental consequences of CO₂ injection. By using sharp-interface theory, these authors were able to capture much of the relevant physics of the CO₂ injection problem without relying on numerical methods. In this study, we addressed the questions presented above by developing sharp-interface models of CO₂ injection across the Illinois basin. Our calculations are compared to multi-phase models including TOUGH2 (Birkholzer et al., 2008). Our study does not consider the effects of CO₂ leakage along abandoned petroleum wells (Nordbotten et al., 2004, Nordbotten et al., 2005). We also compare our model results to single-phase analytical solutions of Hantush and Jacob (1955).