Unlocking subsurface geoenergy and storage potential using machine learning
Unlocking subsurface geoenergy and storage potential using machine learning
Publish Date: 2026-03-20 05:10:00
Source Domain: www.newswise.com
- Geologists from the Institute of Applied Geosciences at KIT, Germany, introduced a novel machine learning regression approach using microscopic assessments of rock slices to derive porosity and permeability.
- The porosity and permeability properties are crucial for geoenergy production and storage in geological reservoirs.
- Machine learning regression was selected to capture non-linear and multivariate relationships in the distribution and content of minerals from thin sections.
- The regression models applied to data from 51 wells in central Europe, over 25 years, show strong predictive performance with R² values of 0.87 for porosity and 0.82 for permeability.
- The approach aims to reduce the cost and time associated with extracting core material from the subsurface for detailed laboratory analyses, thus providing a cost-effective method for reservoir characterization.
- Future extensions of the method may involve the microscopic analysis of cutting fragments, allowing for prediction of reservoir properties on a wider scale and further reducing drilling costs.
