James M. Borer1, Michael H. Gardner1
(1) Colorado School of Mines, Golden, CO
ABSTRACT: Boolean Rule-Sets for Stochastic Modeling of Submarine Channel Reservoirs
Reservoir characterization studies demonstrate stratigraphic controls on fine-scale heterogeneity. Geostatistical techniques are used to represent this heterogeneity in reservoir models. Sparse geologic information is converted into a dense cube of geologic and petrophysical attributes that honor subsurface data. Variogram-based techniques distribute properties best in reservoirs with smooth facies transitions, such as carbonate shelf, clastic shoreface, and perhaps basin-floor sheet deposits. Limitations of this approach for modeling channel bodies led to Boolean-object techniques. Object models distribute geometric bodies (architectural elements) using rules to define spatial distributions and geometric attributes. Hybrid techniques populate objects across smooth variogram-based backgrounds or fill objects with variogram-based facies or properties.
Since deep-water deposits are best described in terms of architectural elements, rule-sets for Boolean objects represent a viable method to preserve micro-heterogeneity. A deterministic 3-D geologic model of the Brushy Canyon Formation is being used to verify rules for distributing deep-water sediment bodies in stochastic reservoir models. Outcrop statistics are compiled to facilitate the development of rules that capture a hierarchy of stratigraphy, architectural elements and facies. Our object-based approach is based on modules that (1) identify, characterize, and quantify the fundamental objects (architectural elements), (2) provide preconditioning rules for the architectural attributes, such as lithology, topography and sediment body trends, (3) define rules for cross cutting and object linkage, (4) populate objects with facies and properties, (5) define stratigraphic modulation and distribution functions, and (6) provide conditional input and relaxation hierarchy. The accurate placement of objects requires a high-resolution, hierarchical stratigraphic framework to establish the proper stochastic domains for modulating object distribution functions.
AAPG Search and Discovery Article #90906©2001 AAPG Annual Convention, Denver, Colorado