The giant Pinedale gas field in the Green River Basin of Wyoming produces from a 5500 to 6000 ft (1700–1800 m) interval of Upper Cretaceous and lowermost Tertiary sediments. The reservoir comprises discontinuous, lenticular fluvial sands intercalated with overbank sand, silt, and mud. Average porosity in reservoir sandstone is <10% with permeability in the micro-Darcy range. A typical well may have 50 channel sand packages, bundled into 15 to 20 frac stages and commingled. Modeling to date has focused on the interaction of complex fluvial sand geometry with hydraulic fractures, increasing pore pressure with depth, and variable water saturation. Although natural fractures have been recognized, their demonstrable impact to production is localized.

Despite significant compaction and cementation, we can demonstrate good correspondence of core and log petrophysical properties to facies. Because of this, it is desirable to use facies to populate reservoir models. A multi-step approach was used to populate small (approximately one square mile [2.6 sq km]) “sector” models of different parts of the field. Logs were used to determine facies via neural nets and petrophysical cutoffs. Facies were distributed via object modeling, and then petrophysical properties were distributed within facies using sequential Gaussian simulation.

Gross channel ribbons and bar objects were placed first, guided by interpolated V-shale, which is a proxy for sand correlation. Detailed facies bodies were then distributed within those elements. Because net/gross, sand thickness, sand correlation, and overbank character change throughout the section, different zones were modeled using different body dimensions in consideration of analogs.

In dynamic reservoir simulation, acceptable history matches were attained despite the architectural complexity using production data, bottom-hole pressure, production logging tools, and distributed permanent pressure gauges. These models were used to help assess incremental recovery related to increased well density.