Fault orientation: how it affects the stick-slip behaviour of saturated sandstones at the laboratory scale

In this work we experimentally explore stick-slip behaviour of saturated sandstone specimens with prefabricated saw-cuts representing smooth faults of different orientation. A surface-ground saw-cut was machined across cylindrical samples with a diameter of 50 mm and a length of 99 to 103 mm. Two different orientations were tested: saw-cuts inclined at 40° and 45° to the sample long axis. A 2 mm diameter borehole, leading directly to the fault plane, was drilled at the base of each specimen.
Triaxial compression experiments were performed using a servo-hydraulic loading frame from Material Testing Systems (MTS). Two Quizix pore pressure pumps controlled the applied pore pressure and measured the injected fluid volume. Initially, the specimens were loaded under hydrostatic pressure up to 55 MPa, while applying vacuum. Then, the samples were saturated in situ applying a constant inlet pore pressure of 5 MPa. Distilled water, injected through the borehole, was used as pore fluid. Finally, specimens were loaded up to almost 1% axial strain at a constant displacement rate of 20µm/min. Confining pressure and pore pressure were held constant during this loading stage.
P-wave velocities and Acoustic Emission (AE) activity were recorded during the experiments. Sixteen P-wave sensors were glued to the surface of the specimens and sealed in a neoprene jacket with two-component epoxy. Two P-wave sensors were embedded in two metallic spacers placed at the bottom and top ends of the specimens. P-wave velocity measurements were made along 65 different transmitter-receiver traces every 30 seconds. During the in-between time intervals all sensors were recording AEs. Full AE waveforms and ultrasonic signals were stored continuously in a 16-channel transient recording system and they were automatically discriminated after the experiments. An automatic picking algorithm was used to pick the P-wave onset times. AE hypocentre locations were calculated by minimising travel time residuals using the downhill simplex algorithm, considering time-dependent variations in P-wave velocities and employing an anisotropic heterogeneous ultrasonic velocity model, consisting of five horizontal layers. Time dependent anisotropic velocities inside each layer were periodically updated using ultrasonic transmission measurements. First motion amplitudes were picked. An average first motion polarity for each AE event was calculated as the mean of all sensor recordings. AE events were defined as T-, S- and C-type sources (tensile, shear and pore collapse, respectively).
Stick-slip behaviour is observed in samples with 40° and the 45° saw-cuts during the triaxial compression experiments. More stick-slip events are detected in the 40° than in the 45° fault, at similar total shear strain. Peak differential stress at failure and stress drop increase for consecutive slip events for both samples. Stick slip events are initiated at lower stress for the 40° saw-cut. For the 40° saw-cut sample, stress drops and displacements along the fault plane are smaller than those measured in the 45° saw-cut sample. AE locations reveal that during slip events crack damage is strongly localised along the faults, while during loading between slip events damage propagates progressively into the wall rock. AE activity leading to unstable slip events is considerably higher for 45° saw-cut sample compared to the 40° fault. Tensile and shear events dominate stick-slip cycles in 40° saw-cut samples, whereas grain crushing and pore collapse dominate AE activity in sample with 45° saw cut.