Dr. Toti Larson, Ph.D., PI
Mudrock Systems Research Laboratory
Bureau of Economic Geology
The University of Texas at Austin
This study applies machine learning tools to characterize Wolfcamp XY and Third Bone Spring ‘sand’ chemofacies using high resolution X-ray fluorescence core-based measurements from the Delaware Basin in west Texas. Supervised chemofacies training datasets that integrate wireline log response curves and core plug data are explored with XGBoost Decision Tree machine learning to predict and inform electrofacies from wireline logs across the basin. Three chemofacies are organic matter-rich mudstones with good to excellent oil generation potential, and two chemofacies are siltstones with high porosity and median pore throat radius indicating good storage and permeability potential. Two dolomitic chemofacies have the lowest measured porosity and median pore throat radius, indicating low storage and suggesting low permeability. The distribution and abundance of the siltstone and organic matter-rich mudstone facies are consistent with the Wolfcamp XY submarine fan complex developed in Loving and Reeves County, and can be used to differentiate stacked reservoirs that have excellent potential to store migrated oil vs. good unconventional shale-oil targets. This study demonstrates the importance of informing basin models using core-based data and represents an important breakthrough in upscaling core-based data to wireline logs.