New techniques from various disciplines can be used to improve our ability to characterize hydrocarbon reservoirs. Some common problems in building geologic models to be used in reservoir modeling are addressed by techniques from sequence stratigraphy, artificial neural networks, and production decline type curve analyses. Integrating results from these applications provides a useful model of reservoir architecture and plumbing systems. These techniques are illustrated with data from Red Oak Gas Field in the Arkoma Basin of Oklahoma. Production comes from deep marine Red Oak formation sandstones of highly variable reservoir quality.

Correlations based on stratigraphic sequences form the basis for determining flow units and predicting lateral sandstone connectivity. Permeability is predicted from well logs (gamma ray, induction, and density curves) using neural networks. These predictions are compared with kh from production decline type curve analyses; resulting values are displayed with the sand body correlations to illustrate permeability connectivity. Cross sections can now be constructed that illustrate permeability connectivity, flow units, appropriate model layers, and stratigraphic barriers. Maps and, ultimately, 3D models show the distribution of reservoir and non-reservoir lithologies as well as lateral reservoir connectivity.

AAPG Search and Discovery Article #90937©1998 AAPG Annual Convention and Exhibition, Salt Lake City, Utah