Gas-liquid two-phase flow measurement using coriolis flowmeters incorporating neural networks

摘要

Coriolis flowmeters are commonly used to measure single phase flow. In recent years attempts are being made to apply Coriolis flowmeters to measure two-phase flows. This paper presents a neural network based approach that has been applied to Coriolis flowmeters to measure both the liquid flow rate and the gas volume fraction of a two-phase flow. Experimental tests were conducted on a purpose-built two-phase flow test rig on both horizontal and vertical pipelines. The mass flow rate ranges from 700 kg/h to 14500 kg/h whilst the gas volume fraction is between 0 and 30%. A set of variables, including observed density, apparent mass flow, differential pressure across the Coriolis flowmeter and signals to maintain flow tube oscillation, are considered as inputs to a neural network. Two neural networks are established through training with experimental data obtained from the flow rig on horizontal and vertical pipelines, respectively. The performance of both neural networks is assessed in comparison with the reference readings. Experimental results suggest that the relative errors of the corrected mass flow rate of liquid for the vertical and horizontal installations are no greater than ±1.5% and ±2.5%, respectively. The gas volume fraction is predicted with relative errors of less than ±10% and ±20%, respectively, for vertical and horizontal installations in most cases.