Reduced-order description of phase fraction data measured using wire-mesh sensor in two-phase pipe flows

Proper Orthogonal Decomposition (POD) technique has been applied to the three dimensional (cross-sectional and time) void fraction data acquired by the application of wire-mesh sensor (WMS) measurement technique. The measurements are performed in a 76.2 mm ID horizontal pipe with air and water as fluids, and with 16 x 16 WMS located at a distance of 17m from the inlet (L/D = 223). In order to perform the detailed analysis using the POD technique, flow patterns covering stratified-wavy, slug, pseudo-slug, and annular low liquid loading are investigated in the paper. Thanks to the time-resolved data obtained using the WMS, classical POD method is applied in spectral space after converting time dependence to frequency. The analysis showed that for all the cases considered, substantial part of the original data can be recovered using a couple POD modes and a few frequency modes. The most dominant two phase flow features are captured at near perfect accuracy. The global flow parameters like space- and time-averaged liquid holdup values are compared between the original and reconstructed fields and the agreement is found to be very good. The phase portraits obtained from the random coefficients associated with different POD mode diagrams correlate with the characteristics of the flow pattern, such that it is possible to observe two distinct orbits in slug and pseudo slug flow cases, and single orbit indicating more random behaviour for the stratified wavy and annular flow cases. (C) 2018 Elsevier Ltd. All rights reserved.