Preparatory acoustic emission activity of hydraulic fracture in granite with various viscous fluids revealed by deep learning technique

Abstract

Summary To investigate the influence of fluid viscosity on the fracturing process, we conducted hydraulic fracturing experiments on Kurokami-jima granite specimens with resins of various viscosities. We monitored the acoustic emission (AE) activity during fracturing and estimated the moment tensor (MT) solutions for 54 727 AE events using a deep learning technique. We observed the breakdown at 14–22 MPa of borehole pressure, which was dependent on the viscosity, as well as two preparatory phases accompanying the expansion of AE-active regions. The first expansion phase typically began at 10–30 per cent of the breakdown pressure, where AEs occurred three-dimensionally surrounding the wellbore and their active region expanded with time towards the external boundaries of the specimen. The MT solutions of these AEs corresponded to crack-opening (tensile) events in various orientations. The second expansion phase began at 90–99 per cent of the breakdown pressure. During this phase, a new planar AE distribution emerged from the borehole and expanded along the maximum compression axis, and the focal mechanisms of these AEs corresponded to the tensile events on the AE-delineating plane. We interpreted that the first phase was induced by fluid penetration into pre-existing microcracks, such as grain boundaries, and the second phase corresponded to the main fracture formation. Significant dependences on fluid viscosity were observed in the borehole pressure at the time of main fracture initiation and in the speed of the fracture propagation in the second phase. The AE activity observed in the present study was fairly complex compared to that observed in previous experiments conducted on tight shale samples. This difference indicates the importance of the interaction between the fracturing fluid and pre-existing microcracks in the fracturing process.

Microseismic Monitoring and Analysis Using Cutting-Edge Technology: A Key Enabler for Reservoir Characterization

“…They showed that a CNN model was capable of locating microseismic events and reconstructing the velocity model simultaneously in real-time from seismic waveforms. Tanaka et al [49] employed a deep learning model to perform moment tensor inversion of acoustic emissions during a hydraulic fracturing experiment of granite rock and obtained 54,727 solutions.…”

Source parameter estimation of acoustic emissions induced by hydraulic fracturing in the laboratory

“…Polarity analysis of P-wave first motions (Bennour et al 2015;Stoeckhert et al 2015) and moment tensor (MT) analysis (Rodriguez et al 2017;Yamamoto et al 2019;Naoi et al 2020) of such AE events have provided information on the fracture modes of individual AEs, which have been helpful to understand the progression of macroscopic fractures. Tanaka et al (2021) have revealed a detailed preparation process of hydraulic fracturing from many MT solutions (54 727 for 10 experiments) that were obtained by introducing a deep learning technique.…”

“…Hence, the sensitivity should be estimated for each transducer after its installation. Some studies (Yamamoto et al 2019;Naoi et al 2020;Tanaka et al 2021) have overcome this problem by estimating the relative sensitivity of each transducer using data from pulse radiation tests that were conducted immediately before each experiment.…”

“…However, these methods necessitate evaluating the bending points in a spectrum, and the use of traditional narrow-band AE transducers with a strong resonance makes accurate estimation unachievable. A broad-band AE transducer, such as the one used by Naoi et al (2020) and Tanaka et al (2021), is preferable.…”

“…Edwards et al 2010;Naoi et al 2014;Goodfellow et al 2015;Ross et al 2016) assumed events to be shear failures and used the root mean square (RMS) values of the radiation pattern coefficients (Boore & Boatwright 1984) to estimate their M o . However, this assumption is possibly inappropriate for hydraulic fracturing experiments because of the occurrence of many AEs having various types of focal mechanisms (Yamamoto et al 2019;Naoi et al 2020;Tanaka et al 2021). Appropriate corrections to reflect the actual focal mechanisms of AEs are necessary.…”