Instantaneous and objective flow regime identification method for the vertical upward and downward co-current two-phase flow

An instantaneous and objective flow regime identification method for the two-phase flow is represented in the paper. The previous methods have been evolved to be an objective by replacing the heuristic determination using the sensor signals in terms of the statistical indexes. However, the flow pattern in the rapid transient or the inherently unstable flow such as the flow in the microgravity cannot be identified because of the observation time for the statistical meaning. The design of the neural network fed by the preprocessed impedance signals of the cross-sectional void fraction is proposed here to satisfy the requirement of both objective and an instantaneous identification. For the preprocessing, the both feed forward neural network and the self-organized neural network as an objective reasoning engine were tested using the experimental data for both upward and downward two-phase flow in the pipes with the inner diameter of 25.4 mm and 50.8 mm. It was found that the proposed flow regime identifier could successfully identify the flow regime using the short term observation data within 1 s. Furthermore, the obtained flow regimes were in a good agreement with the Mishima–Ishii criteria for the upward two-phase flow. However, for the downward flow, it was found that the current flow regimes are in reasonable agreement with the Usui criteria for the slug flow region, only. Other flow regimes have strong dependency on the pipe diameter and some phenomena related to the kinematic wave propagation which was not considered reasonably in the previous criteria. Therefore, theoretical studies to build up the transition criteria for the co-current downward two-phase flow are recommended.