(352e) Can Fault Activation Cause Long-Term Caprock Integrity Issues? Findings from a Self-Sealing Field Experiment

The widespread pressure increases expected from large-scale deployment of CO₂ storage in the deep subsurface may cause unwanted geomechanical effects such as caprock failure per fault activation. As previously reported (e.g., Guglielmi et al., 2021; Cappa et al., 2022), we conducted a series tens-of-meter scale experiments of fault activation within the Main Fault intersecting the Opalinus Clay formation in the Mont Terri underground research laboratory in Switzerland. These tests involved short-term fluid injections into the fault to explore how aseismic and seismic events could jeopardize the integrity of a sealing caprock overlying a CO₂ sequestration reservoir. The experiments demonstrated that significant fluid leakage can occur along the initially low-permeability shale faults when rupture is activated. However, an apparent mechanical closure of the fault was observed as soon as injection ceased and fluid pressure dropped. Unfortunately, it was not clear at the time whether the apparent mechanical closing would lead to complete hydraulic sealing of the fault. There were some indications of a small residual opening remaining that could allow for small long-term leakage.

Here we report on findings from long-term monitoring of the time-dependent sealing of the activated patch of the Mont Terri fault. The monitoring study started in 2020 right after the highly permeable flow path of about 10 m length had been created by injecting water inside the fault zone. The hydraulic transmissivity of the flow path increased by several orders of magnitude from initially very small (10^-12 m²/s) to about 3.2 10^-7 m²/s. In order to observe whether and/or when the fault transmissivity would return to its initial low transmissive state, pressure pulse tests were repeated every two weeks during the year following the fault activation. Local fault transmissivity was estimated by matching the analytical Cooper-Papadopoulos solution to the observed pulse pressure decay vs time. Fault transmissivity variations were then compared to fault movements monitored during and after activation by a network of distributed optical fibers (Distributed Strain Sensing, DSS), local three-dimensional borehole displacement sensors, and changes in p-waves velocities measured with a Continuous Active Seismic Source Monitoring system (CASSM).

At the injection point, the strong transmissivity increase observed during the high-pressure fault activation was initially associated with dilatant slip caused by normal faulting, with a strike slip component in good accordance with the local state of stress, and subsequent further opening due to a decrease in the effective normal stress (Cappa et al. 2022). The peak dilation value of about 400 micro-strains measured during the injection experiment decreased with time until a permanent irreversible fault zone deformation of about 200 micro-strains was established about 70 to 200 days after activation, mainly related to irreversible fault shear displacement. During the first 10 days after activation, the pressure pulse tests showed a fast drop in transmissivity from 3.2 10^-7 m²/s during the high-pressure activation phase to 1.8 10^-9 m²/s (Figure 1). This period was dominated by normal closure of the fault due to the injection pressure drop and subsequent effective stress increase. After 10 days and until 70 days, there was a slower transmissivity decrease related to an approximately ~1.6 10^-11 s^-1 compaction creep of the fault zone. After 70 days, fault deformations progressively followed an extensional regime while transmissivities showed a slightly accelerated decrease. We hypothesize that factors such as clay mineral swelling may have influenced this long-term behavior.

We discuss this complex evolution of fault self-sealing processes using different long-term fault transmissivity relationships by drawing parallels with generic creep laws used to describe long term deformation of clay rocks at the laboratory scale (Figure 1). Due to the multiple competing processes involved, describing the fault sealing process with a single relationship proves challenging. By extrapolating these models, we estimate complete fault sealing may take at least 50 years. However, it is important to note that our estimations are based on experiments conducted at shallow stress levels, involving a limited volume of fluid leakage and a brief fault activation duration. While our data align reasonably well with laboratory-scale studies under similar stress and rock conditions, the variations in clay content within faults spanning multiple mudrock layers and the nature of fluids may influence fault sealing timescales differently than observed in our experiment. This underscores the necessity for additional observations at relevant basin-scale depths to enhance our understanding of these processes.

Figure Caption

Figure 1. Evolution of predicted fault transmissivity over time compared to experimental data estimated from the series of pulse tests (blue dots with error bars taken at 20% to take into account possible uncertainty of measurements). Gray lines are obtained for 100 inversions with a first transmissivity law, while the orange lines correspond to 100 inversions with a second transmissivity law. The red dashed line represents the mean solution among the 100 best-fit solutions for each law.

References

Guglielmi, Y., Nussbaum, C. Cappa, F., DeBarros, L., Rutqvist, J. and J. Birkholzer [2021] Field-scale Fault Reactivation Experiments Highlight Aseismic Leakage in Caprock Analogs: Implications for CO₂ Sequestration, International Journal of Greenhouse Gas Control, 111(103471).

Cappa, F., Guglielmi, Y., Nussbaum, C., De Barros, L. and J. Birkholzer [2022] Fluid Migration in Low-Permeability Faults Driven by Decoupling of Fault Slip and Opening. Nature Geosci 15(9), pp 1-5.

Shadoan, T.A., Ajo‐Franklin, J.B., Guglielmi, Y., Wood, T., Robertson, M., Cook, P., Soom, F., Daley, T.M., Hopp, C., Tribaldos, V.R. and P. Marchesini [2023] Active‐Source Seismic Imaging of Fault Re‐Activation and Leakage: An Injection Experiment at the Mt Terri Rock Laboratory, Switzerland. Geophysical Research Letters, 50(23), p.e2023GL104080.