Harnessing the Power of AI: a Data-Driven Well Reservoir Virtual Flow Metering Tool Accessible to All

Abstract Data-driven Virtual Flow Metering (VFM) by Machine Learning is an increasingly attractive method to perform multiphase flow metering in Oil and Gas production systems. A VFM is a complement to physical meters, using the continuous available field measurements such as tubing pressures, temperatures, and other relevant parameters (choke opening, activation parameters, etc.) as inputs to provide flow rate estimates at any given point of the production and injection network. Good knowledge of Flow Metering is required for production optimization, flow assurance, reservoir management and allocation activities. In the allocation context, production and injection figures are typically computed monthly using available information at the time to answer budget and operation management immediate needs. As such, information could be biased and the data-driven VFM models are used to provide alternative allocations at each wellhead level to improve the confidence in reservoir history matching. The paper presents successful use cases of such data-driven VFM built within the tool TDASH developed by the Company as part of its Digital Transformation Roadmap providing reservoir and production engineers friendly access to Artificial Intelligence means through a no code/low code approach.