Applying Subsurface DNA Diagnostics and Data Science in the Delaware Basin

Abstract The application of DNA sequencing in the oil industry provides a non-invasive, economical, and high-resolution data source for elucidating hydrocarbon fluid movement. DNA Diagnostics could be used to understand drainage height and inform the vertical spacing of wells, as well as provide the ability to derive/predict depositional environments based on biomarker signatures. To date, DNA sequencing has been applied in over 200 wells across 8 basins in North America, including the Permian, Eagle Ford, and Bakken. Subsurface DNA Diagnostics are used to guide well spacing, evaluate well interference (e.g. frac hits), determine oil potential, and understand production profiling. The effective application of DNA sequencing in the oilfield requires the use of advanced data science techniques and machine learning. To date, the industry's application of machine learning has been focused on production and completion parameters that provide a statistical view of the subsurface. In oilfield DNA sequencing, millions of data points per well are generated; and to apply these direct measurements to the subsurface requires novel machine learning techniques which are presented here. In this analysis, we present the results of DNA sequencing from a 33-well study in the Delaware Basin. Well cuttings and produced fluid samples were obtained in a non-invasive manner to generate stratigraphically unique signals for hydrocarbon fluid movement. Various data science techniques are presented to interpret the data with initial observations providing a novel view into production across formations. Additionally, we will provide an overview of future work and how we plan to augment the dataset with other more conventional techniques in order to fine-tune the uncertainties associated with a new data source and methodology.