Tomographic segmentation in multiphase flow measurement

Measurement of multiphase pipe flow of gas, oil and water is not at all trivial and in spite of considerable achievements over the past two decades, important challenges remain. These are related to reducing measurement uncertainties arising from variations in the flow regime and the fluid properties, improving long term stability and developing new means for calibration, adjustment and verification of the multiphase flow meters. In this work the pipe flow is split into temporal segments using multiple gamma-ray measurements. One 241Am source with principal emission at 59.5 keV was used because this relatively low energy enables efficient collimation and thereby shaping of the beams, as well as use of compact detectors. One detector is placed diametrically opposite the source whereas the second and eventually the third are positioned to the sides so that these beams are close to the pipe wall. The principle is then straight forward, that is to compare the measured intensities of these detectors, and through those identify the instantaneous cross sectional gas–liquid distribution, i.e. the instantaneous flow pattern. By counting the intensity in short time slots of <100 ms, experiments verify that rapid variations exist. The water salinity is one of the fluid properties which challenge most multiphase flow meters because its variations affects component volume fraction calculations based on gamma-ray, electrical conductance and other measurements methods. At the University of Bergen a dual modality method has been developed using simultaneous measurements of transmitted and scattered gamma-rays from a 241Am source. This allows the gas volume fraction to be determined independent of changes in the water salinity, provided that the fluid is fairly homogeneously mixed. Tomographic flow segmentation allows selection of low gas fraction segments where the salinity, in combination with running averaging methods, can be calculated with higher accuracy. - Highlights: • The feasibility of flow pattern recognition using two low energy gamma-ray transmission measurements has been experimentally proven. • The feasibility of flow segmentation, i.e. splitting the flow into sub-second temporal segments, has been experimentally proven. • Multiphase flow pattern recognition and segmentation has been implemented in an industrial prototype and successfully tested in a flow loop facility