Data Driven Approach to Infer Inter-well Connectivity among Production Wells in an Oil Synthetic Reservoir

Optimal reservoir management, modeling and production, depend on understanding the connectivity between production wells in an oil reservoir. Using numerical simulation and a reservoir permeability map can clear this feature, however, it is a time-consuming method and the presence of uncertainty in the input data of simulators leads to employ other methods to understand the feature. Having new sensors that are permanently placed in the wellbore, large amounts of production data (production rate and bottom hole pressure) are available. In recent years, intelligent data-driven techniques have been used to work with the large amount of data obtained from production wells. This paper uses a data-driven approach based on detecting important production well events during its production life. Important well events are divided into three categories: (1) increasing, (2) decreasing, and (3) no flow (shut in the well by operator). These events are identified using derivative and slope of the production figures and some limiting factors. Afterwards, the algorithm has to find the related events between the production wells, and based on the importance of the events, the inter-well connectivity among production wells is determined which it is shown as a connectivity map. A synthetic reservoir was first developed in the Eclipse simulation software, and the three high-permeability streaks were considered between three well pairs. Next, its production data were used as the data-driven model’s input, which it is programmed in MATLAB software, in order to identify the three high-permeability streaks via the data-driven model.