Physics-data fusion for digital metering of gas-liquid two-phase flow in oil nozzles

Oil well production metering is an important basis for assessing resource reserves and production capacity. Existing metering methods require high investment and involve significant implementation difficulties and high computational costs. A new method of multiphase flow metering method was proposed by combining the fluctuation of pressure drop and temperature difference generated after throttling through the oil nozzle. The throttling equation coupling differential pressure, flow rate and gas fraction was derived. By regulating the gas and liquid superficial velocity, we carried out the experimental test for true nozzle aperture of 10 mm in the loop system, covering various flow patterns. In addition, the deep neural network was constructed for inversion prediction of mass gas fraction. The proposed flow metering equation has an adaptable gas mass fraction range of 0 to 100%, unaffected by the variations in flow patterns, gas-liquid velocities, and system pressure changes. The metering errors of gas mass fraction and mass flow rate is below ±10%. The proposed metering method only relies on the existing temperature and pressure measurement system in the gathering and transportation system, and no additional measuring devices are required. It can be widely used in oil filed production to realize the virtual metering.