CO₂ storage and potential fault instability in the St. Lawrence Lowlands sedimentary basin (Quebec, Canada): Insights from coupled reservoir-geomechanical modeling

Highlights

• CO₂ injection rates control fluid pressure build-up and fault stability.

• A high permeability of reservoir units reduces the risk of fault reactivation.

• Fault permeability affects timing, localization, rate, length of shear slip.

• Sealing fault behavior causes asymmetric fluid pressure build-up and CO₂ plume shape.

• Tensile fracturing postdates shear failure along fault, localizes below caprock units.

Abstract

Coupled reservoir-geomechanical (TOUGH-FLAC) modeling is applied for the first time to the St. Lawrence Lowlands region to evaluate the potential for shear failure along pre-existing high-angle normal faults, as well as the potential for tensile failure in the caprock units (Utica Shale and Lorraine Group). This activity is part of a general assessment of the potential for safe CO₂ injection into a sandstone reservoir (the Covey Hill Formation) within an Early Paleozoic sedimentary basin. Field and subsurface data are used to estimate the sealing properties of two reservoir-bounding faults (Yamaska and Champlain faults). The spatial variations in fluid pressure, effective minimum horizontal stress, and shear strain are calculated for different injection rates, using a simplified 2D geological model of the Becancour area, located ∼110 km southwest of Quebec City. The simulation results show that initial fault permeability affects the timing, localization, rate, and length of fault shear slip. Contrary to the conventional view, our results suggest that shear failure may start earlier for a permeable fault than for a sealing fault, depending on the site-specific geologic setting. In simulations of a permeable fault, shear slip is nucleated along a 60 m long fault segment in a thin and brittle caprock unit (Utica Shale) trapped below a thicker and more ductile caprock unit (Lorraine Group) – and then subsequently progresses up to the surface. In the case of a sealing fault, shear failure occurs later in time and is localized along a fault segment (300 m) below the caprock units. The presence of the inclined low-permeable Yamaska Fault close to the injection well causes asymmetric fluid-pressure buildup and lateral migration of the CO₂ plume away from the fault, reducing the overall risk of CO₂ leakage along faults. Fluid-pressure-induced tensile fracturing occurs only under extremely high injection rates and is localized below the caprock units, which remain intact, preventing upward CO₂ migration.

Introduction

Changes in fluid pressure and shear-stress accumulation may trigger reactivation of faults, which are optimally oriented relative to in situ stress fields (Davies et al., 2013, Miller et al., 2004, Sibson, 1992, Streit and Cox, 2001). The high fluid-injection rates and low permeability of reservoir formations may contribute to the risk of induced seismicity (Berry and Hasegawa, 1979, McClain, 1970). Incidents of fault reactivation and induced earthquakes related to large-volume fluid injection or gas extraction are well known in such sites globally, e.g., Snipe Lake and Strachan (Alberta), Wilmington (California), Rangely and Denver (Colorado), the midcontinent region of the United States, and at In Salah (Algeria) (Baranova et al., 1999, Ellsworth, 2013, Healy et al., 1968, Hsieh and Bredehoeft, 1981, Mathieson et al., 2010, Mathieson et al., 2011, Milne, 1970, Nicol et al., 2011, Raleigh et al., 1976, Shemeta et al., 2012, Suckale, 2010, Wyss and Molnar, 1972).

Even if no harmful induced seismicity were associated with global carbon capture and storage (CCS) demonstration projects as of February 2011 (Committee, 2012, NETL, 2013), continuous CO₂ injection – at high rates under high pressures for very long periods of time – may lead to an increase in fluid pressures within storage reservoirs, and thus to potential fault reactivation. The CO₂ volumes stored in deep saline aquifers in industrial projects to date vary from about 0.7–1 Mt/yr. (e.g., SnØhvit and Sleipner, Norway; In Salah, Algeria; Illinois, U.S.; Quest, Alberta) up to 3–4 Mt/yr. (Gorgon, Western Australia) (GCCSI, 2012). The maximum period of injection time in operating CCS projects to date varies from 17 years in deep saline aquifers (Sleipner, Norway) to 40 years in enhanced oil recovery (EOR) projects (Val Verde, Texas) (GCCSI, 2012). And in any case, the risks of induced seismicity associated with CO₂ storage may be minimized through careful site characterization and numerical modeling.

The hydromechanical behavior of faults has recently been studied to assess the risks of induced seismicity and fluid leakage related to CO₂ storage (Chiaramonte et al., 2008, Hawkes et al., 2005, Lucier et al., 2006, Murphy et al., 2013, Rutqvist, 2012, Streit and Hillis, 2004, Vidal-Gilbert et al., 2010, Zhang et al., 2013, Zoback and Gorelick, 2012). Coupled reservoir-geomechanical numerical modeling has been shown to be an effective tool for testing fault instability and potential shear failure (Cappa and Rutqvist, 2011b, Rutqvist et al., 2008). The potential for shear failure, and the type and orientation of failure, is to a large extent controlled by the three-dimensional initial stress regime. An extensional stress regime is shown to be favorable for shear failure along high-angle faults (60°) that may cut through overburden rock above the pressurized storage zone (Rutqvist et al., 2008). Fault shear rupture and dilation may induce or enhance fault permeability, which in turn facilitates the rupture propagation across the overlying caprock (Cappa and Rutqvist, 2011b, Rinaldi and Rutqvist, 2013). The effect of initial permeability on co-seismic (sudden) fault slip has been investigated by Cappa and Rutqvist, 2011a, Cappa and Rutqvist, 2012 and Mazzoldi et al. (2012), whose findings show the relatively minor impact of initial permeability. Additionally, it was found that a fault with high permeability was likely reactivated later in time than a fault with low permeability, due to easier fluid-pressure dissipation along the fault (Cappa and Rutqvist, 2011a). However, the effect of initial fault permeability on timing, localization, and rate of fault shear slip, as well as total aseismic fault slip, has not yet been fully investigated.

High-angle faults located close to a CO₂ injection area are usually considered as probable pathways for CO₂ or brine migration and leakage due to fault permeability, either initial or induced, triggered by shear reactivation (Chang and Bryant, 2008, Barton, 2011, Hannis et al., 2013, Jordan et al., 2013). Here, we investigate if there are any other possible effects, from the presence of an inclined fault near the injection zone, on fluid pressure buildup and buoyancy-driven CO₂ plume-migration paths, using the real geological setting of the St. Lawrence Lowlands region.

The deep saline aquifers within the Early Paleozoic sedimentary basin of the St. Lawrence Lowlands (about 200 km × 40 km) are recognized as the best target for geological storage of CO₂ in the Province of Quebec-based on both geologic and practical criteria (Malo and Bédard, 2012). The Cambrian-Lower Ordovician sandstones of the Potsdam Group (Fig. 1) form 200–600 m thick reservoir units in the potential storage area (depths < 4 km). The mean total effective capacity for CO₂ storage in the sandstones of the Potsdam Group is 3.18 Gt at the basin scale (Malo and Bédard, 2012). The Utica Shale and siliciclastic rocks of the Lorraine Group form a 0.8 km to 3.5 km thick caprock system (Fig. 1).

The high-angle SW–NE normal faults dipping to the SE affect both the sedimentary succession and the Grenvillian metamorphic basement (Fig. 1). Faults are oriented subparallel to oblique (10–36°) to the S_Hmax stress orientations and are likely near critically stressed for shear slip, with slip tendency (shear-over-normal-stress ratio) ranging from 0.34 to 0.58 (Konstantinovskaya et al., 2012). Thus, the initial shear-over-normal-stress ratio is less than the range (0.6–1.0) that would substantially increase the likelihood of shear reactivation. The optimally oriented high-angle normal faults in the area might become unstable under the present-day stress field if fluid pressure during CO₂ injection exceeded the critical threshold of 18–20 MPa for a depth of 1 km, thus increasing the risk of induced seismicity and CO₂ leakage (Konstantinovskaya et al., 2012).

In the present study, coupled reservoir-geomechanical numerical modeling is applied to estimate stress changes related to migration of injected CO₂ and the associated increase in fluid pressure (P_f), to evaluate the risk for reactivation of high-angle normal faults in the St. Lawrence sedimentary basin. The coupled reservoir-geomechanical simulator TOUGH-FLAC applied in this case is described in detail by Rutqvist (2011).

Coupled TOUGH-FLAC modeling is performed for the Becancour area site (Fig. 2) using a simplified 2D geological model (Fig. 3). The Becancour industrial park area, located about 110 km southwest of Quebec City, is characterized by relatively high CO₂ emissions (between 0.5 and 1 Mt/yr. in 2009) (Malo and Bédard, 2012). The site is easily accessible and has a well-developed infrastructure, and thus represents one of the best potential sites for CO₂ storage in the St. Lawrence Lowlands (Malo and Bédard, 2012). A high density of hydrocarbon exploration wells and seismic lines in the Becancour area have made it possible to characterize deep saline aquifers (Tran Ngoc et al., 2012, Tran Ngoc et al., 2014) and to create a 3D geological model of the site (Claprood et al., 2012).

The model parameters (boundary conditions, CO₂ injection rate, initial fault permeability, and permeability of reservoir units) were varied to study their influence on fluid pressure buildup, shear failure of faults, and tensile fracturing, in order to estimate the risk of CO₂ leakage. The permeability of the SW–NE high-angle normal faults that affect the St. Lawrence Lowlands sedimentary succession (Fig. 1) is not well constrained. Therefore, in this study, to evaluate the possible range of fault permeabilities that may be applied in the numerical simulations, we conducted field observations of fault-zone rocks and analyzed sample descriptions from fault-crossing wells. We also carried out first-time-ever fault-seal-capacity estimates of the Champlain and the Yamaska faults (Fig. 3).