Effect of Fluid Viscosity on Fault Reactivation and Coseismic Weakening

Abstract

High-viscosity fluids are often used during hydraulic fracking operations in georeservoirs. Here we performed dedicated experiments to study the influence of fluid viscosity on fault reactivation and associated induced earthquakes. Experiments were conducted in the rotary-shear machine Slow to HIgh Velocity Apparatus on experimental fault of Westerly granite saturated by fluids with increasing viscosity (at room temperature) from 0.1 mPa s (water) to 1.2 Pa s (99% glycerol). Fault reactivation was triggered at constant effective normal stress by increasing the shear stress acting on the fault. Our results showed that independent of the viscosity, fault reactivation followed a Coulomb-failure criterion. Instead, fluid viscosity affected the fault weakening mechanism: flash heating was the dominant weakening mechanism in room humidity and water-saturated conditions, whereas the presence of more viscous fluids favored the activation of elasto-hydrodynamic lubrication. Independent of the weakening mechanism, the breakdown work W b dissipated during seismic faulting increased with slip U following a power law (W b ∝ U 1.25 ) in agreement with seismological estimates of natural and induced earthquakes.Plain Language Summary One of the most alarming recent findings in solid earth sciences is the worldwide exponential increase of human-induced seismicity. This is due to engineering operations in deep reservoirs for hydrocarbon production, CO 2 storage, wastewater storage, and exploitation of geothermal resources which result in the reactivation of faults hosted in the reservoirs. While the reactivation of faults due to fluid pressure has been extensively studied, the influence of fluid properties including its viscosity has been overlooked, even if the viscosity of injected fluids spans from the one of water to that of honey. In this study, we discuss the influence of stress perturbations on the reactivation of fluid-permeated experimental faults and on induced earthquakes. Our experimental observations suggest that the viscosity of the fluid does not influence the onset of fault reactivation. Instead, the viscosity of the fluid controls the type of deformation mechanism activated during induced earthquake rupture.

Thermal Weakening Friction During Seismic Slip: Experiments and Models With Heat Sources and Sinks

“…In a different context (no melting), Cornelio et al. (2019, 2020) have shown how viscous fluids permeating natural rock samples affect frictional weakening at high slip velocity by activating elastohydrodynamic lubrication.…”

“…Though the heating is not localized at the asperity contacts, it is still localized and concentrated within a thin shear layer provided that slip is brief enough (earthquake-like duration, typically seconds) that heat diffusion away from the PSZ is reduced (close to adiabatic conditions). In a different context (no melting), Cornelio et al (2019Cornelio et al ( , 2020 have shown how viscous fluids permeating natural rock samples affect frictional weakening at high slip velocity by activating elastohydrodynamic lubrication.…”

Parametric analysis of the elastohydrodynamic lubrication efficiency on induced seismicity

“…density and viscosity) can influence the rock-fluid interactions (Acosta et al 2018) and, in general, the fault strength. In particular, Cornelio et al (2019Cornelio et al ( , 2020 have recently shown that high fluid viscosity might also play a role in decreasing the dynamic fault strength and, consequently, promoting the propagation of earthquakes. They experimentally showed that the onset of reactivation is independent of the on-fault fluid viscosity (i.e.…”

Investigating the Groningen gas reservoir: From passive seismic monitoring to experiments on effects of pore pressure on fault slip

“…The source of these signals was found to be the m b 5.4 earthquake in the Fiji Islands region (2013-12-01, 10:01:31.81, -17.574 ○ N -178.261 ○ E, at 535 km depth; International Seismological Centre (2020)), which is at a distance of 144.102 ○ from SDM-1. The travel time of the PKP phase is 1113.34 s (18 min + 33.34 s), as calculated with the travel time toolkit TauP (Crotwell et al, 1999). Thus, the expected arrival time is 10:20:05.15 which fits the onset time of our recorded signal at ∼10:20:05.4 very well (Fig.…”