Mapping of horizontal refrigerant two-phase flow patterns based on clustering of capacitive sensor signals

Abstract

A capacitive void fraction sensor was developed to study the objectivity in flow pattern mapping of horizontal refrigerant two-phase flow in macroscale tubes. Sensor signals were gathered with R410A and R134a in a smooth tube with an inner diameter of 8 mm at a saturation temperature of 15 °C in the mass velocity range of 200–500 kg/m 2 s and vapour quality range from 0 to 1 in steps of 0.025. A visual classification based on high speed camera images is made for comparison reasons. A statistical analysis of the sensor signals shows that the average, the variance and a high frequency contribution parameter are suitable for flow regime classification into slug flow, intermittent flow and annular flow by using the fuzzy c-means clustering algorithm. This soft-clustering algorithm predicts the slug/intermittent flow transition very well compared to our visual observations. The intermittent/annular flow transition is found at slightly higher vapour qualities for R410A compared to the prediction of Barbieri et al. (2008) [20]. An excellent agreement was obtained with R134a. This intermittent/annular flow transition is very gradual. A probability approach can therefore better describe such a transition. The membership grades of the cluster algorithm can be interpreted as flow regime probabilities. Probabilistic flow pattern maps are presented for R410A and R134a in an 8 mm ID tube.