Attribute analyses of acoustic emissions in hydraulic fracturing

Abstract Hydraulic fracturing (HF) and horizontal drilling are essential to the development of shale gas and oil. Production depends on the stimulation success. During fracture initiation, propagation, and closure, cracks emit acoustic waves; these can be monitored in real time as microseismics in the field and as acoustic emissions (AEs) in the laboratory. AEs are the laboratory equivalent of field-scale microseismics and contain detailed information about HF fracture mechanics. The number of acoustic events correlates with the number of induced fractures and hence the stimulation volume. Three HF protocols under dry conditions were carried out on Tennessee sandstone: (1) a constant injection rate, (2) a precyclic injection, and (3) a variable-rate injection test. All three tests were performed under the same principal stress conditions: vertical stress of 10.3 MPa (1500 psi), minimum horizontal stress of 3.5 MPa (500 psi), and maximum horizontal stress of 20.7 MPa (3000 psi). In total, 16 piezoelectric transducers were mounted around a cylindrical sample to record the AEs. We have performed postsignal processing to extract AE event attributes, including the amplitudes, signal-to-noise ratio, arrival time, event location (with the velocity-anisotropy input), and frequency analyses. The AE events associated with the constant-rate injection test possessed the lowest frequencies (150–270 kHz). The variable-rate test AE events possessed higher frequencies (160–310 kHz), whereas the precyclic injection had events with the highest frequencies, peaking at 330 kHz. Acoustic events before failure had lower amplitudes, but higher frequency compared to those recorded postbreakdown, suggesting different failure modes. Precyclic injection induced the greatest number of locatable events before and after failure.