AAPG Annual Convention and Exhibition

Datapages, Inc.Print this page

Representing Fluvial Channel Belt Heterogeneity in Reservoir Models


Sediment body geometry, connectivity and heterogeneity vary across the spectrum of fluvial channel belt depositional styles. The differences in fluid flow behavior associated with this variability have not been demonstrated for diverse fluid types and recovery mechanisms. As such, the appropriate scale and heterogeneity of fluvial channel belts required for reservoir simulation models is not well established. Moreover, detailed representation of the three-dimensional architecture and spatial distribution of fluvial reservoir properties remains a challenge using conventional modeling approaches. We present a reservoir modeling workflow that uses a hierarchy of surfaces to mimic the evolution of deposits internal to fluvial barforms and abandoned channels. The volume bounded by each surface records a specific stratigraphic interval, with properties distributed according to rules derived from the spatial and temporal position of each surface. In this manner, the conceptual understanding of the type of fluvial deposition is imposed directly on to reservoir architecture. The hierarchical association of surfaces allows the consistent addition or removal of orders of scale (effective down- and up-scaling) described within the reservoir model, as required to modify the predicted fluid flow response during hydrocarbon production. In addition, surfaces can be conditioned to well data, and permit all features of the reservoir model to be represented without the prior constraints of grid cell resolution. Application of the modeling workflow is demonstrated using examples drawn from a classification of fluvial channel belt reservoir styles. The reliability of the method at reproducing the known reservoir style is compared against subsurface data, observations of modern systems, and physics-based numerical models. Consideration is given to both what level of detail is required of reservoir simulation models (relative to benchmark cases), and how the modeling technique needs to be varied depending upon the geologic scenario. For example, in the end-member scenarios described by our fluvial classification, a meandering channel system may be effectively described through the direct application of the modeling workflow described herein, but a braided channel system may be more efficiently modeled using a hybrid approach that first describes the bar and channel fill using surfaces, and then populates a suite of these unique ‘objects’ using traditional techniques.