It is currently difficult to detect low-magnitude, unidentified seismic events and locations used to discover hidden fault or fracture systems in the earth’s crust with great accuracy. Discovery of these unknown or hidden fault systems can lead to predictions of the magnitudes, rates and locations of potential earthquakes when monitored properly. Sandia’s researchers look to create new methods for sensing of rock failures by inducing seismic activity into 3D printed samples and identifying the signatures of these events.
Sandia is using integrated experimental, numerical, and data analysis in order to link mechanical discontinuities, fracture mechanics, and pore pressure or stress to the geophysical signatures identified in both laboratory and field scenarios. We then analyze the geophysical and mechanical data using various machine learning techniques including convolutional neural network (CNN) for event detection, waveform similarity-based event detection, and template matching (Eqcorrscan).
As a result of this research, additive manufacturing applications will be set to be used for more broad themes in the future. This research will set the groundwork for characterizing seismic waveforms by using Multiphysics and Machine Learning approaches and improve the detection of low-magnitude seismic events leading to the discovery of hidden fault/fracture systems.