Machine learning seen as a means to enhance geothermal exploration and production
By using machine learning algorithms, a pair of researchers from the Pennsylvania State University are hoping to reduce the inherent risks of geothermal exploration and production.
Two researchers from the Pennsylvania State University are examining the idea of using machine learning algorithms to manage the risks of induced seismicity during geothermal exploration and production. The research, entitled “Machine learning approaches for safe geothermal exploration”, has won Jing Yang and Chris Marone the 2019 Penn State Multidisciplinary Seed Grant.
The pair are hoping that machine learning algorithms can be used to predict seismic events such as microearthquakes when conducting fracture formation via hydraulic stimulation. So far, successful runs have been done in the lab, but the researchers are working to replicate this success in the field scale.
The second part of their research aims to help determine the amount of fluid that needs to be injected into the ground to achieve high geothermal energy production without causing structural damages to the site.
To meet these objectives, Yang and Marone will develop a safe reinforcement learning framework with scalable algorithms that can be transferred from lab to field and can handle unknown conditions. With success, the researchers are hoping that the work can be applied to other areas outside of geothermal energy production.
For more details on the study, you may check the link below.