Researchers at Penn State have utilized machine learning to discover a connection between low-magnitude microearthquakes and the permeability of subsurface rocks. This finding has significant implications for improving the efficiency of geothermal energy transfer, as generating geothermal energy requires a permeable subsurface to release heat effectively when cold fluids are injected into the rock.
The research, published in Nature Communications, highlights the optimal times for efficient energy transfer by linking microearthquakes to the permeability of subsurface rocks. By monitoring these microearthquakes on the surface using seismometers, researchers were able to extract the “noise” from the data that obscured the connection, utilizing two datasets from the EGS Collab and Utah FORGE demonstration projects funded by the U.S. Department of Energy (DOE). They then used machine learning to create a model from one site and successfully applied it to another through transfer learning.
Pengliang Yu, lead author of the study and a postdoctoral scholar at Penn State, emphasized the generalizability of the models, indicating that seismic monitoring could potentially enhance geothermal energy transfer efficiencies across various sites. The team’s algorithm confirmed a direct link between seismic activity and rock permeability, revealing that rock permeability was highest during periods of strong seismic activity.
The ability to increase rock permeability is crucial for various energy extraction methods, impacting both traditional fossil fuel recovery and renewable energies such as hydrogen production. By introducing cold fluids through porous rock, hydrofracturing methods can break the rock and increase permeability, allowing for easier access of heat and hydrocarbons to the surface. Identifying the connection between seismic activity and rock permeability enables more efficient energy extraction while ensuring that microearthquakes remain below damaging thresholds.
Co-author Parisa Shokouhi, a professor of engineering science and mechanics at Penn State, emphasized the role of machine learning in uncovering the relationship between seismic activity and rock permeability. The algorithm identified essential attributes of seismic data for predicting rock permeability evolution, ultimately revealing an previously unknown physical link between seismic data and rock permeability. This discovery could have widespread applications in monitoring gas movement for carbon sequestration and the production and storage of subsurface hydrogen.
The research is part of a larger DOE-funded project aimed at decreasing the cost and increasing the production of geothermal energy using machine learning to understand and predict earthquakes, including microquakes. Co-author Chris Marone, a professor of geosciences at Penn State, highlighted the connections between the evolution of elastic properties and earthquakes in lab studies, indicating similar relationships in nature. The collaborative efforts of researchers from various disciplines at Penn State and the University of South Florida contributed to this groundbreaking research, showcasing the potential of machine learning in advancing geothermal exploration and energy production.
