1. Lawrence Berkeley National Laboratory Earth Sciences Division 1 Cyclotron Road, MS 90-1116 Berkeley, CA 94720 510-486-6455 3D modeling of fault reactivation during CO 2 injection Antonio P. Rinaldi Victor Vilarrasa, Jonny Rutqvist, Frédéric Cappa H11K-05

2. 2 GCS considered as feasible solution but… Overpressures due to large-scale fluid injection may induce seismic events Previous 2D model: CO 2 injection can cause seismicity (depending on injection rate and initial fault permeability) Reactivation may increase CO 2 leakage (but not necessarily) Fault and site architecture play a role (e.g. seismicity and leakage affected by size of caprock) Low potential for structural damage What will change if we account a full 3D model?

3. Importance of Minor versus Major Faults 3 Large offset major faults can be detected at the ground surface and by seismic surveys Sequestration site might be designed to stay away from such faults and because they are known they can be closely monitored, modeled and analyzed in terms of avoiding reactivation The regional injection-induced (slow) crustal straining and minor (hidden) faults might be of greatest concern at future large scale injection sites (minor faults of unknown location and orientation) Both analytical and modeling results show max event magnitude ~3.6 for 1000 m fault. (Mazzoldi et al., IJGGC, 2012)(Cappa and Rutqvist, GRL, 2011a)

4. Numerical modeling TOUGH-FLAC/ECO2N Fully hydro-mechanical coupling 100 m storage aquifer, bounded by 150 m caprock Pre-existing normal fault with dip 80˚ CO 2 injection at -1500 m, 500 m from the fault Isothermal with gradient 25˚C/km Initial hydrostatic linear gradient Constant pressure and stress boundary Extensional stress regime  H =  h = 0.7  V Damage zone as high permeability zone and Fault core with Ubiquitous-joint model Oriented weak plane in a Mohr- Coulomb solid Strain-softening model: friction as function of plastic shear strain 4

5. Geomechanics and fluid flow coupling 5 Damage zone : 10 -15 m 2 porosity as function of mean effective stress (σ’ M ), permeability depends on porosity changes (Davies and Davies, 2001) Fault core : 10 -17 m 2 Anisotropic coupling. Hydraulic parameters depend on anisotropic elasto-plastic properties. Porosity as function of plastic tensile (e ftp ) and shear strain (e fsp ), and dilation (ψ). Permeability as function of normal effective stress (σ’ n ) and porosity changes (Hsiung et al., 2005). a and c empirical constants for normal-closure hyperbola (Bandis et al.,1983)

6. 2D vs 3D: time evolution 6 2D model: Injection rate for pressure evolution (with assumed initial fault permeability and accounting for scaling and symmetry): 0.05 kg/s/m  0.05*1000*2  100 kg/s Pressure decreases at about 500 days Reactivation at about 100 days with magnitude 3.23 (assuming circular rupture) Slip at reactivation ~6 cm (z=-1500 m), then increase up to 10 cm RUNNING TIME: ~4 hours 3D model: Injection rate for pressure evolution (with assumed initial fault permeability and accounting for symmetry): 30 kg/s  30*4  120 kg/s Pressure decreases at about 300 days Reactivation at about 200 days with magnitude 3.57 Slip at reactivation ~ 7cm (z=-1500 m, y=0 m), increases only to 8 cm at later times RUNNING TIME: ~13 hours

7. 2D vs 3D: spatial distribution 7 Rinaldi et al., (2014a,b) 2D model: 3D model: Injection rate: 0.05 kg/s  100 kg/s Injection rate: 30 kg/s  120 kg/s

8. 2D vs 3D: CO 2 leakage at shallow depth 8 Rinaldi et al., (2014a,b) 2D model: Injection rate: 100 kg/s Total injected mass: ~15 Mtons Leakage in upper aquifer: ~1.4 % 3D model: Injection rate: 120 kg/s Total injected mass: ~18.6 Mtons Leakage in upper aquifer: ~2 % STILL HIGH IF COMPARED TO 0.1%/year LIMIT (Hepple and Benson, 2005), BUT 2D MODELS SLIGHTLY UNDERESTIMATE LEAKAGE

9. Horizontal vs Vertical injection well (case 1) 1 MPa difference at peak Different time of reactivation Small difference in slip and event magnitude Critical pressure extent: 700 m for vertical and 1000 m for horizontal Larger rupture area for horizontal well injection

10. Horizontal vs Vertical injection well (case 2) 1 MPa difference at peak, but overall smaller compared to Case 1 (faster reactivation, faster permeability changes) Increased difference in event magnitude Critical pressure extent: 300 m for vertical and 700 m for horizontal Larger rupture area and slip for horizontal well injection

11. Horizontal vs Vertical: permeability changes 11 Flow mostly on damage zone, whose permeability increases up to 25 times the initial value, and at 5 years reduces to about 10 times. Both cases present a similar permeability increase, with the largest changes in the central aquifer, and with changes extending for more than 2000 m along the fault strike and up to shallow depths along the dip. The permeability is higher around the well region (up to 60% larger) for the vertical well case For the horizontal case, the permeability is slightly higher along the fault strike at greater distance from origin. Both cases have similar permeability changes at the end of the 5-year injection, but the horizontal well case presents a slightly higher permeability in the upper caprock.

12. Horizontal vs Vertical: CO 2 leakage 12 Total amount of leakage similar for both cases (~ 2% of total injected mass). For vertical well, larger leakage occurs in the near-well region. For horizontal well, CO 2 is leaking more along fault strike at greater distance.

13. Discussion & Conclusion 13 Overall good agreement between 2D and 3D model In 3D model simulations, the pressure distributes over a larger space than in a 2D model, and a higher injection rate is required to achieve the same pressure increase. 2D model shows a percentage of leakage of about 1.4%, which increases to 2% in a 3D model, although with a larger the injection rate (120 kg/s for the 3D model compared to 100 kg/s for the 2D model). Differences in temporal evolution because of permeability changes within the fault zone (a single line for 2D, a surface for 3D) Horizontal vs Vertical injection well: The faster the pressurization, the smaller the seismic event Vertical well: more localized but faster and larger increase in pressure, then less slip on a smaller rupture area. Horizontal well: pressure distributes over a larger space, and a longer time is required to reach the critical pressurization, then larger slip on larger area. Similar permeability changes and leakage: for vertical well results slightly higher permeability in the near-well region, for horizontal well larger permeability changes along the fault strike, and then leakage varies accordingly.

14. Thank You! This work was supported by the Assistant Secretary for Fossil Energy, Office of Natural Gas and Petroleum Technology, through the National Energy Technology Laboratory, under the U.S. Department of Energy Contract No. DE-AC02-05CH11231. Acknowledgment 14