The Permian Basin is currently the most active unconventional resource play in North America. The combination of high quality Petrophysical and Geomechanical characteristics together with advances in horizontal drilling, and completion innovations in hydraulic fracturing has allowed the successful development of several stacked reservoir targets within this basin.
This paper focuses on horizontal wells targeting different benches, from Wolfcamp to the First Bone Spring formation. It presents the use of reservoir flow facies defined from the guidance of the geological and petrophysical facies to predict best potential landing targets. These flow-quality facies were created using machine learning techniques. Geological and petrophysical facies were initially defined using 756 petrophysical wells, 9 facies training wells and 64 "high tier wells" with NMR, Sonic and Minerology logs. 34 Core-calibrated petrophysical models were also incorporated. Rock mechanical facies were defined from sonic and geological data integrated with closure pressure gradients, net pressure and end-of-job shut-in pressure matching for hundreds of fracture treatment stages.
Through an integrated multidomain workflow combined with experience from neighboring areas, 14 Production Quality Faces were defined from the combination of 9 Geo-Facies and 11 Rock Quality Facies This facies definition evolved into a 3D geo-model, where sector models were cut across multiple areas of interest where engineering datasets (micro-seismic, DFIT, core data, etc) existed. Finally, several poro-perm relations and facies-based relative permeability curves were defined through the history-matching of production data.
Using the presented workflow, different potential landing targets in the Delaware Basin were evaluated for optimal development strategies, from Avalon to Wolfcamp A.
A series of property-specific machine learning based facies models were created using a set of training wells spread across the Permian Basin and extending from the Upper Bone Springs through Wolfcamp-A formations. The model is underpinned by the wireline logs and is extended to three coupled discipline-centric facies sets. The first of these is the Reservoir Quality Facies (RQF), which discriminates the porosity, saturation and kerogen properties of the reservoir independent of geologic characteristics. Next is the Geofacies (GF), which does the reverse. It discriminates the mineralogic properties of the formation independent of the pore system. Third is the Geomechanical Facies (GMF), which discriminates the mechanical properties of the rock independent of the other influences. Each of these coupled facies sets allows for independent analysis but can be combined to produce additional facies sets that can be used for a 3D reservoir model. For this purpose, the GF and RQF were cross-tabulated to produce Production Quality Facies, PQF. This interplay of RQF and GF shows how the mineralogy (independent porosity/saturation) and porosity/saturation (independent mineralogy) relate. It is the combination of these two fundamental properties sets that describes expected flow behavior but by providing the fundamental inputs (porosity and mineralogy) separately, we can also evaluate them independently. This is a benefit of having discrete coupled sets of property facies.
