Predicting naturally fractured reservoirs for subsurface characterization using machine learning and the EDX database

Characterizing naturally fractured reservoirs is essential for subsurface energy applications, including hydrocarbon exploration, geothermal development, and carbon storage evaluation. However, predicting fracture properties (density, aperture, permeability, etc.) is difficult, especially in areas where there are few existing wells. There are three general approaches for characterizing naturally fractured reservoirs: Geological, Geophysical and Petrophysical. Individually, these methods fail to provide universal fracture predictions with high levels of confidence in areas where there are few well calibrations. This Phase I project aims to prove its concept of combining all three natural fracture characterization approaches using ML techniques to create a workflow that accurately predicts fractured reservoirs before wells are drilled. The subsurface EDX dataset will be used as training data for the machine learning model. Seismic data will be used to generate seismic attributes; horizon and fault interpretations will be used to generate structural restorations, strain calculations, and other structural attributes; petrophysical information from wells, outcrop analogs, and seismic inversions will be used as petrophysical attributes. Combined, these attributes will constitute the feature data. The machine learning model will be trained on these feature data using label data from wells. There is an urgent need to bring on additional sources of carbon-free energy sources if the United States is going to meet its carbon reduction targets. Predicting these fractures is the primary subsurface technical challenge needed to discover more geothermal resources that then can be developed into power plants. Unlocking fractured carbon storage and geothermal resources will also create more green energy jobs, particularly for geoscientists and the recently displaced oil and gas workforce. Specific groups in the carbon storage sector that can benefit from such a predictive tool include those looking at developing deep saline, oil and gas, or geothermal reservoirs for carbon storage sites. These include the oil and gas majors (e.g., ExxonMobil, Chevron and Shell), all of whom are currently partnered on developing one of the largest carbon capture and storage sites in the world.