Integrating Field Performance and History Matching Uncertainties Using Fuzzy Logic to Minimize Well-Placement Risks

Abstract

To a great degree, the success or failure of a field development plan depends on the choice of well locations, which together with areal and vertical sweep drives hydrocarbon recovery from the field. Choosing the optimum well locations is, however, non-trivial and requires a thorough understanding of hydrocarbon reservoir performance and properties. Even after the locations are identified there is a need to further rank the wells in terms of confidence as part of the risk assessment. Many methods currently used to place wells rely heavily on the final history matched model, which may be misleading because history matching is a non-unique solution and different history matched models may give different prediction results. This paper presents a workflow using the fuzzy logic approach that combines history matching uncertainties with field performance data to help generate a confidence index for well placement. In contrast to neural networks, which take training data and generate opaque models, fuzzy logic allows a geoscientist to build models on top of experience from the experts. All the expert opinions and reservoir understanding can be dictated as a simple set of rules and logic to define the well placement confidence index, which could then be used to propose field development plans with a much higher level of confidence. This, in turn, helps with better decision making and reduces the risk involved in drilling additional infill wells.