Flow rate is a valuable piece of information for the oil and gas industry. High accuracy on flow rate estimation enhances the operations that control and manage production. Recognized as a cost-efficient solution, the virtual flowmeter (VFM) is a mathematical-based technology designed to estimate the flow rates using available field instrumentation. The VFM approach developed in this work combines black-box simulations with mixed-integer linear programming (MILP) problems to obtain flow rates while dismissing the tuning process. The methodology includes a set of multiphase flow correlations. The MILP estimates the flow rate and designates the best-fit model. The VFM is evaluated against 649 well test data. The methodology presents a 4.1% absolute percentage error (APE) for the 25th percentile and 13.5% APE for the 50th percentile of the data. The methodology dismisses the model tuning process, which would require flow rate data. This makes the methodology valuable, particularly in scenarios where historical data are scarce or unavailable.

Keywords:
reservoir simulation, paper, reservoir surveillance, artificial intelligence, artificial lift system, machine learning, upstream oil & gas, flow metering, production control, drillstem/well testing
Subjects:
Artificial intelligence, Artificial Lift Systems, Downhole and wellsite flow metering, Drillstem/well testing, Formation Evaluation & Management, Information Management and Systems, Reservoir Simulation, Well & Reservoir Surveillance and Monitoring
Copyright © 2023 Society of Petroleum Engineers
You can access this article if you purchase or spend a download.