The main objective of a production well test is to assist in the identification of reservoir and well parameters needed for regulatory accounting, well surveillance, and asset management purposes. Interpreted information is used to drive decisions on production enhancement, operations optimization, and field-development plans. However, uncertain results may occur when wells are produced from multiple reservoirs. Currently, the industry approach is to allocate well and reservoir parameters based on known petrophysical data, offset well information, and zonal well tests. When possible, testing by difference is commonly performed to control one or more zones; however, this process may result in significant production losses with poor concluding results, especially when zonal interference is vital to a well's operating point (e.g., intelligent wells in a waterflood field).
A methodology was developed to consistently identify reservoir and well-performance parameters from wells produced under commingle conditions from multiple reservoir zones by leveraging available real-time data. This methodology was successfully applied in a field located in offshore West Africa, a waterflooded field with nine intelligent wells. The methodology integrates surface well-test rates, pressures, and downhole triple-gauge data. Collected data is validated via a rigorous history calibration process of an integrated production model consisting of an analytical reservoir, well, downhole and surface chokes, and pipeline models. The calculated parameters (e.g., zonal rates, productivity index, reservoir pressure, gasoil ratio, and water cut) are the result of an error minimization between calculated variables and measured field data.
This paper presents applications of this methodology for two production tests of a single dry-tree well with individually controlled reservoir zones. Benefits of the above application include a 90% reduction of the time required to perform a similar analysis, reduced uncertainty in rate allocation and reservoir parameters, and better understanding of the likely production from every reservoir zone. Because well and reservoir parameters are allocated to individual layers, the resulting rate allocation satisfies all sensor data and physical models, and therefore the uncertainty of the allocation is reduced. In addition, the application provides the basis for rate allocation to multiple zones in real time when well tests are not available.