11:00 AM - 11:15 AM
[SSS05-17] Giant rock friction experiments for simulating complex faulting process and its close observation
★Invited Papers
Keywords:Rock friction experiment, Large-scale experiment, Fault rupture
Numerous rock friction experiments have been conducted to understand the mechanism of faulting, which is the essence of earthquakes, by simulating it in the laboratory. While they provided us with many important findings such as nucleation process, foreshock activities accompanying it, and the effects of fault surface heterogeneity on earthquake preparation process (e.g. Dieterich, 1978; Ohnaka et al., 1986; McLaskey and Kilgore, 2013; Yamashita et al., 2021; Xu et al., 2023), further research is required due to the complexity of actual faulting in nature. The complexity of faulting should be caused by the heterogeneity of the frictional properties and/or stress distribution on the fault. To properly introduce the heterogeneity into laboratory experiments and investigate its effects in detail, experiments on a larger spatial scale are necessary. For this reason, we have developed an apparatus that enables us to conduct the rock friction experiments on a larger scale than existing ones (Yamashita et al., 2023), and we are carrying out the experimental studies using it.
In the experiments, we stack a rock specimen with a length of 6.0 m, a width of 0.5 m, and a height of 0.75 m on top of another rock specimen with the same width and height but a length of 7.5 m, and then give relative displacement in the longitudinal direction. The rock type is metagabbro collected in India. The simulated fault has a contacting area with dimensions of 6.0 m in length and 0.5 m in width, resulting in a total area of 3 m2. A normal load is applied to the top surface of the upper specimen using six hydraulic jacks, and then a shear load is applied to the side surface of the lower specimen on a low-friction roller using another hydraulic jack. Since the maximum load of each jack for the normal load is 2 MN, an average pressure on the simulated fault is 4 MPa at maximum. The velocity of the shear load jack can be fully controlled between 0.01 mm/s and 1 mm/s, and the jack can extend up to 1000 mm. In addition to the measurements of macroscopic loads, we installed many sensors in the vicinity of the simulated fault to closely observe local phenomena during the faulting process; 64 piezoelectric sensors (Olympus V103-RM) were installed on the side surface of the lower specimen to measure the ground motion. 44 each of uniaxial, biaxial, and triaxial semiconductor strain gauges were installed on the side surface of the upper specimen to measure local strain (Kyowa KSN-2-120-E4-11, KSN-2-120-F3-11, and SKS-30282, respectively). 44 fiber optic FBG (Fiber Bragg Gratings) strain sensors were also installed. Local displacements of the upper and the lower specimens are measured with laser displacement transducers (Keyence IL-S025) at eight locations, and the local relative displacements of the simulated fault are estimated from the relative values.
We conducted the experiments under the normal stress of 1 MPa or 2 MPa, the loading velocity of 0.01 mm/s, and the shear load jack displacement of 10 mm, with varying initial location of the shear load jack (Dini, hereafter). Stick-slip events were consistently observed under those conditions. The stiffness of the experimental system ksys was estimated from the ratio of the shear load drop to the displacement at the load point during the stick-slip event as 3.8±0.2 GN/m at Dini = 0 mm. However, we observed that ksys decreased as Dini increased and it reached 0.4±0.0 GN/m at Dini = 990 mm. The reason for this reduction in ksys could be due to the increase in volume of compliant oil within the shear load jack as it extends. The strain gauge array clearly showed the unilateral fault ruptures during the stick-slip events in most cases, though they involved some complicated behaviors. We will further introduce heterogeneous normal stress distribution by setting different loads for each normal load jack and will investigate its effects on the faulting process.
In the experiments, we stack a rock specimen with a length of 6.0 m, a width of 0.5 m, and a height of 0.75 m on top of another rock specimen with the same width and height but a length of 7.5 m, and then give relative displacement in the longitudinal direction. The rock type is metagabbro collected in India. The simulated fault has a contacting area with dimensions of 6.0 m in length and 0.5 m in width, resulting in a total area of 3 m2. A normal load is applied to the top surface of the upper specimen using six hydraulic jacks, and then a shear load is applied to the side surface of the lower specimen on a low-friction roller using another hydraulic jack. Since the maximum load of each jack for the normal load is 2 MN, an average pressure on the simulated fault is 4 MPa at maximum. The velocity of the shear load jack can be fully controlled between 0.01 mm/s and 1 mm/s, and the jack can extend up to 1000 mm. In addition to the measurements of macroscopic loads, we installed many sensors in the vicinity of the simulated fault to closely observe local phenomena during the faulting process; 64 piezoelectric sensors (Olympus V103-RM) were installed on the side surface of the lower specimen to measure the ground motion. 44 each of uniaxial, biaxial, and triaxial semiconductor strain gauges were installed on the side surface of the upper specimen to measure local strain (Kyowa KSN-2-120-E4-11, KSN-2-120-F3-11, and SKS-30282, respectively). 44 fiber optic FBG (Fiber Bragg Gratings) strain sensors were also installed. Local displacements of the upper and the lower specimens are measured with laser displacement transducers (Keyence IL-S025) at eight locations, and the local relative displacements of the simulated fault are estimated from the relative values.
We conducted the experiments under the normal stress of 1 MPa or 2 MPa, the loading velocity of 0.01 mm/s, and the shear load jack displacement of 10 mm, with varying initial location of the shear load jack (Dini, hereafter). Stick-slip events were consistently observed under those conditions. The stiffness of the experimental system ksys was estimated from the ratio of the shear load drop to the displacement at the load point during the stick-slip event as 3.8±0.2 GN/m at Dini = 0 mm. However, we observed that ksys decreased as Dini increased and it reached 0.4±0.0 GN/m at Dini = 990 mm. The reason for this reduction in ksys could be due to the increase in volume of compliant oil within the shear load jack as it extends. The strain gauge array clearly showed the unilateral fault ruptures during the stick-slip events in most cases, though they involved some complicated behaviors. We will further introduce heterogeneous normal stress distribution by setting different loads for each normal load jack and will investigate its effects on the faulting process.