Geothermal energy generation relies on a permeable subsurface that allows for the effective release of heat when cold fluids are introduced into the rock. By establishing a connection with microearthquakes, which are detected on the surface by seismometers, researchers have identified optimal times for energy transfer. The team's findings were published in Nature Communications.
Supported by the U.S. Department of Energy (DOE) and leveraging data from the EGS Collab and Utah FORGE projects, the team employed machine learning to clarify data "noise" that previously masked this link. Through a technique known as transfer learning, they developed a model at one site and successfully applied it to another, suggesting a universal principle applicable across all geothermal sites.
"Pengliang Yu, a postdoctoral scholar at Penn State and the study's lead author, stated, "The success of transfer learning underlines our model's broad applicability, indicating seismic monitoring could significantly improve geothermal energy efficiency at various locations."
Yu emphasized the importance of rock permeability not just for geothermal energy, but for fossil fuel recovery and renewable energies like hydrogen production. Hydrofracturing, which enhances rock permeability by injecting cold fluids, generates microearthquakes. These microearthquakes increase rock permeability, facilitating the extraction of heat and hydrocarbons.
Their algorithm established a direct correlation between seismic activity and increased rock permeability. This knowledge allows for optimized energy extraction while ensuring microearthquakes remain undetectable to the public and cause no damage.
"Machine learning was pivotal in discovering the link between seismic activity and rock permeability," said Parisa Shokouhi, engineering science and mechanics professor. The algorithm not only identified crucial seismic data attributes but also confirmed a previously unrecognized physical connection.
The researchers believe that this advancement could reduce reliance on fossil fuels. Moreover, understanding the relationship between rock permeability and microearthquakes could aid in carbon sequestration and subsurface hydrogen production and storage.
This research is part of a larger DOE initiative aiming to lower geothermal energy costs and enhance production, with machine learning as a tool to better understand and predict microearthquakes.
Chris Marone, geosciences professor at Penn State, noted, "Yu's work advances our understanding of geothermal energy and earthquake prediction through machine learning. Our lab's findings on the pre-earthquake evolution of elastic properties are now echoed in natural settings."
Research Report:Crustal permeability generated through microearthquakes is constrained by seismic moment
Related Links
Penn State
Bringing Order To A World Of Disasters
When the Earth Quakes
A world of storm and tempest
| Subscribe Free To Our Daily Newsletters |
| Subscribe Free To Our Daily Newsletters |
