<p>Failure along rock discontinuities can result in economic losses as well as loss of life. It is essential to develop methods that monitor the response of these discontinuities to shear loading to enable prediction of failure. Laboratory experiments are performed to investigate geophysical techniques to monitor shear failure of a pre-existing discontinuity to detect signatures of impending failure. Previous studies have detected precursors to shear failure in the form of maxima of transmitted waves across a discontinuity under shear. However, those experiments focused on well-matched discontinuities. However, in nature, rock discontinuities are not always perfectly matched because the asperities may be weathered by chemical, physical or mechanical processes. Further, the specific shear mechanism of mismatched discontinuities is still poorly understood. In this thesis, the ability to detect seismic precursors to shear failure for various discontinuity conditions—well-matched (rough and saw-tooth), mismatched (rough), and nonplanar (discontinuity profile with a half-cycle sine wave (HCS))—was assessed. The investigation was carried out through a coupled geophysical and mechanical experimental program that integrated detailed laboratory observations at the micro- and meso-scales. Shear experiments on gypsum discontinuities were conducted to observe changes in compressional (P) and shear (S) waves transmitted across the discontinuity. Digital Image Correlation (DIC) was used to quantify the vertical and horizontal displacements along the discontinuity during shearing to relate the location and magnitude of slip with the measured wave amplitudes. </p> <p>Results from the experiments conducted on planar, well-matched rough discontinuities (grit 36 sandpaper roughness) showed that seismic precursors to failure took the form of peaks in the normalized transmitted amplitude prior to the peak shear stress. Seismic wave transmission detected non-uniform dilation and closure of the discontinuity at a normal stress of 1 MPa. The results showed that large-scale roughness (presence of a HCS) could mask the generation of precursors, as it can cause non-uniform closure/dilation along the fracture plane at low normal stress. </p> <p>The experiments on idealized saw-toothed gypsum discontinuities showed that seismic precursors to failure appeared as maxima in the transmitted wave amplitude and conversely as minima in the reflected amplitudes. Converted waves (S to P & P to S) were also detected, and their amplitudes reached a maximum prior to shear failure. DIC results showed that slip occurred first at the top of the specimen, where the load was applied, and then progressed along the joint as the shear stress increased. This process was consistent with the order of emergence of precursors, i.e., precursors were first recorded near the top and later at the center, and finally at the bottom of the specimen. </p> <p>Direct shear experiments conducted on specimens with a mismatched discontinuity did not show any precursors (in the transmitted amplitude) to failure at low normal stresses (2 MPa), while those precursors appeared at higher normal stresses (5 MPa). The interplay between wave transmission, the degree of mismatch, and the discontinuity’s micro-physical, -chemical and -mechanical properties was assessed through: (1) 3D CT in-situ Xray scans to quantify the degree of mismatch at various normal stresses; (2) micro-indentation testing, to measure the micro-strength of the asperities; and (3) Scanning Electron Microscopy (SEM) and Electron Xray Diffraction (EDX), to study the micro-structure and chemical composition of the discontinuity. The X-ray results showed that contact between asperities increased with normal stress, even when the discontinuity was mismatched. The results indicated that: (1) at 2 MPa, the void aperture was large, so significant shear displacement was needed to interlock and damage the asperities; and (2) the micro-hardness of the asperities of the mismatched discontinuity was larger than that of the well-matched discontinuity, which points to inducing less damage for the same shear displacement. Both mechanisms contribute to the need for larger shear displacements to the mismatched discontinuity asperities to cause damage, which is consistent with the inability to detect seismic precursors to failure. The experimental results suggest that monitoring changes in transmitted wave amplitude across a discontinuity is a promising method for predicting impending failure for well-matched rock discontinuities. Precursor monitoring for mismatched rock discontinuities seems only possible when there is sufficient contact between the two rock surfaces, which occurs at large normal stresses. </p>

<p dir="ltr">A key challenge in the assessment of the stability of fractures in rock is the identification of precursory geophysical signatures of shear failure. Accurate estimation and prediction of shear failure along rock discontinuities is crucial to prevent failure of geotechnical structures and potential natural hazards, such as landslides and earthquakes. Active seismic monitoring, such as compressional (P) and shear (S) waves, has been used to monitor the evolution of contact area and contact stress along rock discontinuities. Past laboratory experiments determined that changes in the amplitude of the transmitted, reflected, and converted P- and S-waves can be used to assess local changes in contact area and fracture specific stiffness, and to identify precursory events to shear failure of rock fractures. Those studies have identified the peaks (maxima or minima) in wave amplitudes as the seismic precursors to shear failure. Past studies were performed on dry artificial rock discontinuities with homogeneous and well-matched contact surfaces. However, in nature, rock discontinuities are not always homogeneous and well-matched, and are often found below the water table. In addition, at large depths, e.g. in enhanced geothermal systems (EGS), fractures are subjected to high temperatures.</p><p dir="ltr">The objectives of this research are to: (1) characterize the geophysical response of rock fractures during shear for dry and saturated conditions at room temperature, and saturated conditions at elevated temperatures; and (2) detect and identify seismic signatures of shear failure/slip for each of the three conditions. To achieve the goal of the research, a novel shear test apparatus was designed and built to test saturated jointed rock specimens under normal and shear loading, with a back pressure and at elevated temperatures, while also being capable of housing seismic transducers to monitor simultaneously the mechanical and geophysical response of the rock joints during shear. The system consisted of a sealed and heated pressure chamber and a biaxial compression frame. The pressure chamber was also used to perform B-value tests on cylindrical rock specimens to determine the minimum magnitude of back pressure required for fluid saturation.</p><p dir="ltr">Laboratory direct shear tests were performed on tension-induced fractures in Indiana limestone and Sierra White granite specimens with non-homogeneous rough contact surfaces. The contact surfaces were created by axial splitting of prismatic rock blocks. Shear tests were conducted on the rock fractures at a constant displacement rate in the pressure chamber, which enabled control of effective normal stress, pore water pressure, and temperature. During the tests, transmitted and converted P- and S-waves propagated across rock fractures and their changes in wave amplitude were monitored to assess the evolution of local contact areas during shear and detect precursory changes in wave amplitudes prior to shear failure.</p><p dir="ltr">Seismic precursors were observed in the wave amplitude data from all tests conducted under the three conditions. Precursors were most identifiable in the transmitted S-wave data. For all three conditions, the transmitted S-wave showed the same form of a seismic precursor; a peak (maximum) in wave amplitude was observed prior to the peak shear strength, as local contact surfaces interlocked and failed before macroscopic shear failure. However, the transmitted P-wave and converted waves (P-to-S and S-to-P) exhibited different behavior compared to the transmitted S-wave and depended on the test conditions. While, for dry conditions, the transmitted P-wave and converted waves still exhibited seismic precursors as peaks in their wave amplitudes, they did not display an observable peak for saturated fractures at room temperature, but rather either a very slight increase or a continuous reduction in amplitude. Instead of observable peaks, an abrupt change in the rate of reduction in the transmitted P-wave and converted amplitudes was observed that either coincided or occurred close to the peak in the transmitted S-wave amplitude. Thus, an onset of dramatic change in the reduction rate can be also taken as a seismic precursor to shear failure. This phenomenon can be explained by the large stiffness of the highly incompressible fluid, water, which leads to a decrease in P-wave sensitivity to changes in the normal fracture stiffness that arise from rock asperities under saturated conditions.</p><p dir="ltr">Even though the seismic wave amplitude generally contains a seismic precursor to shear failure, some exceptions exist: the wave amplitudes also depend on the local characteristics of the frictional area. No peak or seismic precursor in wave amplitude is observed prior to failure when the contact area between the fractures surfaces decreases because of dilation/opening. In addition, a delay peak in amplitude after shear failure may be observed when the fracture surfaces contain an initial large void or aperture in the region probed by the sensor. These exceptions may occur at a relatively low effective normal stress (2 MPa) and may disappear when a better contact has been established between the fracture surfaces by increasing the effective stress. Direct shear tests under an effective stress of 6 MPa, but at 50^oC, showed that both the transmitted P-waves and converted waves exhibited peaks in their amplitudes prior to the failure. However, these exceptions still require further exploration for the systematic identification and detection of seismic precursors.</p><p dir="ltr">The research shows that seismic monitoring is an effective tool to monitor the shear behavior of discontinuities, to provide an assessment of the local behavior of the frictional surface under the transducer, and to predict failure of the discontinuity. It can be used for dry, saturated discontinuities and for a wide range of pore pressures and temperatures. Other potential applications include fault monitoring, and even possibly earthquake prediction with additional research.</p>