Spatial modeling of complex sandstone bodies to maximize reservoir contact for wells drilled in clastic formations. 9th Middle East Geosciences Conference, GEO 2010.

The gigantic clastic reservoirs in Saudi Arabia contain thick, prolific and continuous sandstone members; however, incremental development may include numerous laterally discontinuous prolific oil-bearing sandstone bodies intercalated with non-reservoir rocks, in the so-called stringers. The optimization of hydrocarbon production requires advanced modeling workflows to identify and predict the spatial distribution of clastic discontinuous rock bodies. This study proposes cross-validation of 3-D models with new well bores to improve future predictions. The modeling approaches include sequence-stratigraphic interpretations and identification of the depositional environment. Object-modeling and sequential indicator simulation techniques were used to produce multiple realizations of 3-D geocellular facies models that predict the geometry and location of sandstone bodies. New wells were planned and drilled based on the most probable predictions. Once a well was completed, the real data collected at the wellbore was compared to multiple geocellular realizations to evaluate an average error at each location. That error was later used to modify the facies model and workflows. The ultimate goal was to reduce uncertainty and optimize new wells planning. The proposed optimization approach, for drilling new wells, was tested in the Cretaceous Safaniya stringers member of the Wasia Formation. Upward increasing gamma-ray logging values, and upward decreasing grain size from core descriptions were interpreted to indicate fining upward sequences associated with sandstone channels. Localized crevasse splays show coarsening upward and blocky shapes on the gamma-ray. Other bodies identified are bays and mouth bars. These bodies and sequence boundaries were incorporated into an initial 3-D geocellular facies model. Object modeling was used to populate the 3-D model, with objects drawn with realistic shapes and sizes. The models were cross-validated with new drilling. Each new well provides new logging data values, which were compared to predictions from various realizations of the 3-D geocellular model, and the average error was plotted against petrophysical properties and gamma-ray derivatives. Results are summarized to recommend corrections in the geological interpretation and modeling approaches. It was concluded that a hybrid approach - combining both object and sequential indicator modeling techniques - is the optimum way to predict rock bodies with current technology.