Flow pattern mapping of horizontal evaporating refrigerant flow based on capacitive void fraction measurements

In daily life, air-conditioning devices are all around us. To optimize these devices, the prediction of two-phase flow heat transfer of evaporating refrigerants has to be improved. This heat transfer is strongly related to the two-phase flow patterns. Improving the current two-phase flow models should therfore focus on incorporating more flow phenomena into the models. In this work, two-phase flow patterns of horizontal evaporating refrigerant flow in smooth round macro-scale tubes are investigated. A macro-scale test facility for two-phase flow and heat transfer studies of HFCs was designed and constructed. Secondly, a capacitive void fraction sensor was developed to study the dynamic behavior of two-phase flows. The capacitance transducer can measure dynamic signals of fluids with the capacitance of the liquid smaller than 1pF and still achieve a SNR greater than 300. Capacitance signals of HFCs can thus be investigated. The dynamic capacitance signals in combination with high speed camera images were used to characterize horizontal two-phase flows in an 7.91mm ID smooth tube. Two datasets were gathered using R410A and R134a respectively. A detailed description of the two-phase flow phenomena was presented together wit a sensor signal charaterization using the time signals as well as PDFs and PSDs. All major two-phase flow phenomena were clearly represented in the sensor signals. To investigate the objectivity in the current flow pattern maps, several statistical parameters were analyzed in combination with the use of the fuzzy c-means clustering algorithm. The clustering in the selected feature space, groups the data points in clearly separable areas in a flow pattern map. Applying the technique to the HFC datasets, the slug flows could be easily separated from non-slug flows by using the variance of the sensor signal. The AVG and the F95 parameter were found most suitable for separating intermittent flows from annular flows. From the output of the soft-clustering algorithm a probabilistic flow pattern map was presentd for the HFC data. These maps clearly quantify the width of the transition zones and can be applied for probabilistic heat transfer and/or pressure drop modeling