The seismic signature of fractures and faults: Low-frequency shear anelasticity measurements toward determining stress and frictional conditions of reservoir fractures in situ

摘要

Stress conditions, asperity geometry, and frictional properties affect the propagation of high amplitude seismic waves across rock fractures and faults. Laboratory evidence suggests that these properties can be measured in active seismic surveys, potentially offering a route to characterizing these important fracture properties in situ. To improve geophysical constraints on the mechanical state of fractures, we modified a unique seismic-frequency (∼0.1 - 100 Hz) apparatus to also include key environmental controls (high strain amplitude and low effective stress). By measuring the modulus and attenuation of numerous rocks before and after tensile fracturing under a range of normal stresses (from 0.25 to 15 MPa), we were able to estimate the seismic detectability of fractures under equivalent effective stress conditions, and show that even well-mated fractures could be monitored with repeatable seismic sources. Utilizing a novel pressure sensitive film technique, we mapped the normal stress distribution on the fracture surface asperities, providing the ability to characterize the contact geometry and estimate the real contact area for each applied normal stress. This is crucial information for understanding and modeling the shear mechanics, which is influenced by surface geometry effects. Connecting past work from the rock physics, nonlinear elasticity, and fault friction communities; we suggest seismic field measurements of strain-dependent softening and frictional attenuation across fractures or faults can constrain their in situ frictional properties. By measuring how the rock fractures' moduli decrease and attenuation increase with strain amplitude, we inverted for the effective asperity area (consistent with real contact area measurements) and friction coefficient using a simple partial slip model. This model can also be extrapolated to fully sliding asperities, predicting the conditions for failure or the critical slip distance in rate-and-state friction. Since these measurements are made using oscillating shear stresses over a range of frequencies, they also contain information about the dynamic, strain-rate dependent frictional response. In fact, measured strain oscillations show strong nonlinearity due to the transition from static to dynamic friction on the sliding regions of the fracture surface. Although rate-and-state friction is commonly used to model fault dynamic friction, this formulation is not capable of capturing the reversal of slip observed over the strain oscillations. In order to overcome this difficulty, we used another frictional parameterization, with a similar state variable structure, based on the Dahl model, common in the engineering community. We inverted the model to fit the time-series and frequency dependent behavior, showing that high amplitude shear wave data could provide information on velocity strengthening or weakening behavior of faults and fractures in situ. This is particularly exciting because these properties determine whether a fault is expected to slide stably or fail seismically in an earthquake. Our lab measurements show the promise of a field technique to map spatial and temporal heterogeneity in the conditions of failure and the frictional properties determining the behavior during these events, vital information for informing fault modeling and seismic hazard assessment.