Dual-driving of data and knowledge to reduce uncertainty in lithofacies interpolation

Abstract

Abstract The reservoir in focus has braided river delta front deposition, with multiple periods of submerged distributary channels within the reservoir. It also displays frequent cutting and stacking with local-connecting characteristics. Forecasting the sand distribution characteristics between wells in this type of reservoir brings a significant challenge for modeling. The data- and knowledge-driven modeling method proposed is applied to the Sangtamu Oilfield as an example. Obtain channel-scale information from the geological knowledge database. Geological expertise is applied to interpret the characteristics of channel distribution. These results are employed as conditional data in the process of geological modeling. By combining the expertise of experts with multi-sources of geoscientific data, this method can obtain accurate and reliable spatial information about the channel. This information is crucial for stochastic simulation between wells and enables to minimize uncertainty in predicting results. The method is useful for 3D modeling of similar sedimentary bodies or well-sparse areas.

Keywords

• Dual-driving modeling
• Interpretation of connected well profile
• Conditional data
• Lithofacies interpolation