Application of Expert Systems to Hydrocarbon Reservoir Characterization--Conceptual Approach

P. G. Sutterlin

The behavior of fluids in hydrocarbon reservoirs, predicted using conventional computer-based simulation models, often is not borne out during production. The quantitative data required as input to the models typically are derived from geophysical logs, drill-stem tests, cuttings, in some instances detailed core analyses, and pressure-decline curves. These data reflect reservoir characteristics that occur within individual boreholes and are regarded for modeling purposes as representative of the reservoir as a whole.

However, these data provide little insight into geological characteristics that occur between wells or field wide. These characteristics can produce reservoir inhomogeneity, thereby significantly affecting recovery efficiency. Unfortunately, many of these latter characteristics, such as sedimentary structures, textural changes, and mineralogical differences, are not easily quantified and therefore cannot be incorporated into existing models.

The application of expert systems technology provides the ability to introduce the element of judgment into the predictive processes by taking into account the effect of those geological characteristics that are known, through experience, to influence hydrocarbon reservoir behavior.

AAPG Search and Discovery Article #91022©1989 AAPG Annual Convention, April 23-26, 1989, San Antonio, Texas.