Flow regime Identification is an integral aspect of modeling two phase flows as most pressure drop and heat transfer correlations rely on a priori knowledge of the flow regime for accurate system predictions. In the current research, two phase R-134a flow is studied in a 7mm adiabatic horizontal tube over a mass flux range of 100–400 kg/m2s between 550–750 kPa. Electric Capacitance Tomography results for 196 test points were analyzed using statistical methods and neural networks. This data provided repeatable normalized permittivity ratio signatures based on the flow distributions. The first four temporal moments from the mean scaled permittivity data were utilized as input variables. Results showed that only 80 percent of flow regimes could be correctly identified using seven flow regime classifications. However reducing to five more commonly used regimes resulted in an improvement to 99 percent of the flow regimes correctly identified. Both methods of neural network training resulted in errors that were off by mostly one flow regime classification. Further analysis shows that transition cases can oscillate between two separate flow regimes at the same time.
