A neural network to correct mass flow errors caused by two-phase flow in a digital coriolis mass flowmeter

Coriolis mass flow meters provide accurate measurement of single-phase flows, typically to 0.2%. However gas-liquid two-phase flow regimes may cause severe operating difficulties as well as measurement errors in these flow meters. As part of the Sensor Validation (SEVA) research at Oxford University a new fully digital coriolis transmitter has been developed which can operate with highly aerated fluids. This paper describes how a neural network has been used to correct the mass flow measurement for two-phase flow effects, based entirely on internally observed parameters, keeping errors to within 2%. The correction strategy has been successfully implemented on-line in the coriolis transmitter. As required by the SEVA philosophy, the quality of the corrected measurement is indicated by the on-line uncertainty provided with each measurement value