Characteristics of 2-D Rupture Propagation of Stick-slip Events during Meter-sized Biaxial Friction Experiments

Abstract:

Using centimeter-sized rock specimens, friction experiments have been carried out. Ohnaka and Kuwahara (1990) showed a stably propagating shear strain drop (nucleation phase) followed by its accelerated propagation. Such process was reproduced by numerical simulations considering heterogeneous fault strength and the weakening (e.g., Dieterich, 1992; Matsu'ura et al., 1992). However, in previous studies, the rupture propagation was measured only with 1-D array of strain gauges along in-plane direction. In addition, very few nucleation phenomena were observed in natural earthquakes. We performed friction experiments using a pair of meter-sized Indian gabbro specimens. The fault plane was 1.5 m long and 0.5 m wide. The specimens were loaded at a speed as low as 0.0025 m/s. We continuously observed shear strains and elastic waves with 2-D array of piezoelectric sensors and strain gauges to improve our knowledge about rupture nucleation process. Sensors and gauges were installed at 24 sites located 60 mm below the fault plane at intervals of 150 mm and 75 mm for in-plane and anti-plane directions, respectively. The stick-slip events were extracted using wave amplitudes and the ratio of share and normal stresses. Two typical stable phases (nucleation phases) of decrease in shear strain were recognized followed by the dynamic rupture. The first decrease was initiated around the edge in anti-plane direction (northern side). The decreasing rate was lower than 10^-6 s^-1, and the decrease preferably propagated in anti-plane direction at a speed slower than 100 m/s. The second decrease was initiated about when the first one reached the opposite end (southern side). The decreasing rate was higher (~10^-4 s^-1) than that of the first one, and the decrease preferably propagated in anti-plane direction at a higher speed (an order of 100 m/s). About when the second phase got back to the northern end, the rapid decrease (~10^-3 s^-1) was initiated at the western edge, and two-dimensionally propagated faster than 1 km/s, which corresponded to dynamic rupture. Then, the rupture nucleation and dynamic rupture propagations in laboratory were suggested to be controlled by the sample edge.