← Back to Sessions

Clustering Analysis of Microseisms Generated During Hydraulic Fracturing Recorded by Downhole Geophones

Description: 

Statistical properties of microseismic signals induced during hydraulic fracturing could help us better understand the correlation between seismic recordings and injection activities. Clustering-based machine learning algorithms demonstrate great potential to identify changes in fracturing processes by searching for systematic differences in spectral properties of microseisms (e.g., Holtzman et al., 2018). In this study, we apply a Hierarchical clustering algorithm to power spectral density (PSD) of microseismic waveforms generated during hydraulic fracturing simulations. For each event, we average PSD calculated over three components and nearby downhole geophones to increase the signal-to-noise ratio. Then, the attenuation effect is corrected from the PSD using a best-fitting quality factor Q estimated from the linear trend of frequency spectra. The Wasserstein distance is used to quantify the difference in PSD between events. Applications to data collected from the Montney Formation and the April 2022 simulation at the FORGE site in Utah indicate event clusters with clear differences in spectral properties. The grouping results could potentially provide information on important topics, such as the characterization of fracture complexity, degree of fracture connectivity, and monitoring of fracture growth.

Session: De-risking Deep Geothermal Projects: Geophysical Monitoring and Forecast Modeling Advances

Type: Oral

Date: 4/18/2023

Presentation Time: 10:30 AM (local time)

Presenting Author: Hongrui Qiu

Student Presenter: No

Invited Presentation: 

Authors