Extraction of AE events to estimate their b values under a triaxial compressive condition Examination using continuous broadband records

Seismic b values estimated in previous laboratory compressive tests had been utilized for natural earthquake studies. Randomly sampled enough number of events over a wide magnitude range are essential for accurate b value estimation. In former triaxial tests, PZTs had sensitivity only in a narrow frequency range. In addition, the recording system could not extract all signals because of mask times or threshold setting. Recently, Yoshimitsu et al. (2014) enabled to use broadband transducers under triaxial conditions and achieved to acquire waveforms continuously in several hours. With such a system, they estimated the seismic moment of AE at very small magnitude scale. We expected that their continuous broadband recording system made it possible to record much more AE with a wider magnitude range for credible b value estimation in a laboratory. In this study, we performed a compressive test under a higher confining pressure as an updated experiment of Yoshimitsu et al. (2014) and extracted an enough amount of AE. We prepared an intact cylindrical Westerly Granite sample, 100 mm long by 50 mm in diameter. We conducted a triaxial compressive test under a confining pressure of 50 MPa, at a room temperature with drying conditions. Seven broadband transducers (sensitive range; 100 kHz - 1,000 kHz) were located in different height, respectively. Besides, a PZT was mounted to transmit elastic waves for velocity estimation during the experiment. At first, we increased the confining pressure and then started the loading. We switched the load control method from the axial load control to the circumferential displacement one. After exceeding the peak stress, compressive stress was unloaded with a high speed and the sample was recovered. A potential fault was observed on the recovered sample surface. Waveform recording was continued throughout the test for more than 200 minutes. The result of extracting signals by an STA/LTA ratio method for the waveforms recorded by each transducer, we detected about 2,170,000 signals at the most and about 450,000 at the minimum. Recorded waveforms may also include the elastic waves from the PZT and electrical noises. To find the combination of the signals derived from the same event, we used the largest differences in travel times for all transducer pairs. Finally, we obtained about 450,000 combinations.