Improvement of TOPFLOW void fraction data using potential field simulations of the wire-mesh sensor response

At the TOPFLOW test facility of the HZDR, gas-liquid two-phase flow experiments were conducted in a vertical DN200 pipe aimed at the creation of a high-quality database. A quality check of the measurement results obtained for 48 different combinations of superficial air and water velocities (water: 0.04 -1.6 m/s, air: 0.0025 - 3.2 m/s) has shown that an overestimation of the cross-section averaged void fraction by the wire-mesh sensor is evident. This finding is supported by the analysis of drift fluxes calculated from the measured void fraction and the injected flow rates of liquid and gas, which are too low, often even negative, to be physically plausible. Another indication of the overestimation is the miss of the reproduction of the gas flow rate from a multiplication of void fraction and gas velocity profiles obtained from signals of a pair of successive sensors. Reconstructed superficial gas velocities tend to exceed significantly the values known from the injected gas flow rates in the bubbly flow region. Prasser & Häfeli (2018) showed that a linear dependency between conductance and liquid holdup at a crossing point of wires delivers too high void fraction values. The application of Maxwell’s equation for the conductivity of an emulsion was instead proposed to improve results. Furthermore, the nature of frequently observed overshoots of the conductance above the calibration values was found to correspond to a real physical phenomenon. During the previous evaluation of the TOPFLOW data, these overshoots were eliminated by setting the gas fraction to zero when they occur. The paper presents the results of a reevaluation of the TOPFLOW data using an approach based on Maxwell’s equation and a correction of the overshoots instead of truncating them. It is shown that the problem of negative drift velocities is now eliminated; the match of reconstructed and injected gas velocities considerably improved.