Probabilistic Determination of Two-Phase Flow Regimes Utilizing an Automated Image Recognition Technique

Probabilistic two-phase flow map models are used in the literature to predict pressure drop and void fraction in multi-port microchannels. In the present study probabilistic two-phase flow maps are experimentally developed for R134a at 25, 35, and 50C, R410A at 25C, mass fluxes from 100 to 600kg/m-s, qualities from 0 to1 in 8mm, 5.4mm, 3.9mm, 1.7mm I.D. smooth adiabatic tubes in order to extend the probabilistic two-phase flow map modeling technique to single tubes. A new web camera based flow visualization technique utilizing an illuminated diffuse stripped background was utilized to enhance images, detect fine films, and aid in the automated image recognition process. The developed automated image recognition software determines the flow regime time fraction in approximately 900 images per flow condition (~1 million images total). The maximum and average error in the automated flow regime time fraction determination is found to be 4% and 0.6% respectively.