Virtual Mass Multiphase Flow Meter (vMPFM) – A Digital Future Enabled by AI/ML & PINN

Abstract As the energy industry transitions toward digitalization and environmental accountability, real-time and accurate flow measurement in multiphase environments has become essential for production optimization, emissions management and leak detection. This paper introduces the Virtual Multiphase Mass Flow Meter (vMPFM), an AI/ML and PINN-enabled multiphase flow metering system that eliminates the need for complex hardware installations by utilizing existing field sensors, and demonstrates how the integration of PINNs, trained and validated using CFD-simulated environments, can enable accurate, scalable, and real-time mass flow measurement together with leak detection and emissions measurement across field-wide deployments, including remote wellheads and processing facilities. The proposed vMPFM architecture leverages existing field sensor data available at the wellhead or processing facilities to infer real-time multiphase mass flowrates. Using a hybrid modelling approach, CFD is conducted to model diverse flow regimes and sensor responses, forming the primary training and validation dataset for a PINN. Physical constraints - such as the Navier-Stokes equations, interfacial closure laws and Euler-Lagrange Model are embedded directly into the training process, enabling the model to learn flow behavior consistent with governing fluid dynamics. Calibration incorporates historical flow test data, flow loop benchmarks, and where feasible, field-deployed measurements. This approach ensures robustness and adaptability to a wide range of operating conditions. The trained model is deployed as a vMPFM for field wide monitoring, either as a standalone digital solution or embedded in minimal edge computing devices (e.g. compact hybrid MPFM units) for remote operations. Furthermore, the applicability to use vMPFM as a leak detection tool was evaluated against traditional numerical methods. Validation of the vMPFM across diversely simulated operational regimes demonstrates its robustness and high accuracy. Benchmarking against traditional MPFMs and flow loop shows that the vMPFM system consistently achieves mass flow measurement accuracy within 5%, even under varying flow conditions from low to high water cut, low to high Gas Volume Fraction (GVF), and even fugitive emissions events. The vMPFM adapts dynamically to changing flow profiles without the need for frequent recalibration, significantly lowering maintenance overhead and measurement uncertainty. By leveraging only conventional sensor inputs, it delivers continuous real-time mass flow data with minimal hardware intervention. These capabilities enable proactive production surveillance, early leak detection, and precise emissions mass quantification. Furthermore, its scalable digital architecture extends measurement capacities to locations previously deemed impractical or cost-prohibitive to monitor, reducing both capital and operating expenditures while supporting broader field-wide digitalization, decarbonization and sustainable development goals. This paper presents a novel framework for digitized multiphase mass flow measurement that unifies physical modeling and data-driven learning. By removing the dependency on costly hardware and enabling scalable deployment through existing field sensor, the vMPFM redefines how multiphase flows are measured, monitored and managed including potential applications of emissions monitoring and leak detection—bridging the gap between conventional metering and intelligent field development.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.