Aseismic Motions Drive a Sparse Seismicity During Fluid Injections Into a Fractured Zone in a Carbonate Reservoir

Abstract

An increase in fluid pressure in faults can trigger seismicity and large aseismic motions. Understanding how fluid and faults interact is an essential goal for seismic hazard and reservoir monitoring, but this key relation remains unclear. We developed an in situ experiment of fluid injections at a 10 meter scale. Water was injected at high pressure in different geological structures inside a fault damaged zone, in limestone at 280 m depth in the Low Noise Underground Laboratory (France). Induced seismicity, as well as strains, pressure, and flow rate, was continuously monitored during the injections. Although nonreversible deformations related to fracture reactivations were observed for all injections, only a few tests generated seismicity. Events are characterized by a 0.5‐to‐4 kHz content and a small magnitude (approximately −3.5). They are located within 1.5 m accuracy between 1 and 12 m from the injections. Comparing strain measurements and seismicity shows that more than 96% of the deformation is aseismic. The seismic moment is also small compared to the one expected from the injected volume. Moreover, a dual seismic behavior is observed as (1) the spatiotemporal distribution of some cluster of events is clearly independent from the fluid diffusion (2) while a diffusion‐type pattern can be observed for some others clusters. The seismicity might therefore appear as an indirect effect to the fluid pressure, driven by aseismic motion and related stress perturbation transferred through failure.

Illuminating the Rupturing of Microseismic Sources in an Injection‐Induced Earthquake Experiment

“…We estimate corner frequencies of 10 pairs of highly similar events (eight master earthquakes) that have Brune‐type spectral ratios (Figure S1). The new corner frequencies are comparable to the values obtained by Duboeuf et al (). The largest difference between the estimates for a certain event is about a factor of 2 (Table S1).…”

“…The magnitudes were computed by Duboeuf et al () by fitting a Brune's model in acceleration, following Boore (). They used a quality factor Q = 50 for the attenuation to obtain a flat plateau at high frequency and averaged the results from all sensors and components to remove radiation pattern effects (Daniel, ).…”

“…The largest difference between the estimates for a certain event is about a factor of 2 (Table S1). To calculate the source dimension from corner frequency estimates (Figures 3 and 4a), we assume a circular crack model (Eshelby, 1957) and an S wave velocity (v s ) of 2,750 m/s (Duboeuf et al, 2017):…”

“…The Brune source parameter results in larger rupture patches and overlap of events, whereas the Madariaga source parameter leads to the lower bound of rupture dimensions (Figures S2 and S3). The magnitudes were computed by Duboeuf et al (2017) by fitting a Brune's model in acceleration, following Boore (1983). They used a quality factor Q = 50 for the attenuation to obtain a flat plateau at high frequency and averaged the results from all sensors and components to remove radiation pattern effects (Daniel, 2014).…”

Injection‐Induced Moment Release Can Also Be Aseismic

“…Indeed, Figure 4 of McGarr and Barbour () suggests that if injection leads to a substantial earthquake response, then the cumulative moment scales linearly with total injected volume. Part of the motivation for our study comes from recently published reports (e.g., Duboeuf et al, ; Galis et al, ; van der Elst et al, ) suggesting that maximum moment does not scale linearly with injected volume.…”