Understanding the stratigraphic architecture of geologically complex reservoirs, such as the heavy oil deposits of Western Canada, is essential to achieve an efficient hydrocarbon recovery. Borehole and 3-D seismic data were integrated to define the stratigraphic architecture and generate 3-dimensional geological models of the Mannville Group in Saskatchewan. The Mannville is a stratigraphically complex unit formed of fluvial to marine deposits. Two areas in west-central and southern Saskatchewan were examined in this study. In west-central Saskatchewan, the area corresponds to a stratigraphically controlled heavy oil reservoir with production from the undifferentiated Dina-Cummings Members of the Lower Cretaceous Mannville Group. The southern area, although non-prospective for hydrocarbons, shares many similarities with time-equivalent strata in areas of heavy oil production. Seismic sequence stratigraphic principles together with log signatures permitted the subdivision of the Mannville into different packages. An initial geological model was generated integrating seismic and well-log data Multiattribute analysis and neural networks were used to generate a pseudo-lithology or gamma-ray volume. The incorporation of borehole core data to the model and the subsequent integration with the lithological prediction were crucial to capture the distribution of reservoir and non-reservoir deposits in the study area. The ability to visualize the 3-D seismic data in a variety of ways, including arbitrary lines and stratal or horizon slicing techniques helped the definition of stratigraphic features such as channels and scroll bars that affect fluid flow in hydrocarbon producing areas. Small-scale heterogeneities in the reservoir were not resolved due to the resolution of the seismic data. Although not undertaken in this study, the resulting stratigraphic framework could be used to help construct a static reservoir model. Because of the small size of the 3-D seismic surveys, horizontal slices through the data volume generally imaged only small portions of the paleogeomorphologic features thought to be present in this area. As such, it was only through the integration of datasets that the geological models were established.

No description available.