Fast classification of two-phase flow regimes based on conductivity signals and artificial neural networks

On-line identification of flow regimes is important in two-phase flow because hydrodynamics and adequate operation of multiphase systems are highly dependent on the flow pattern. This work describes the application of an artificial neural network (ANN) to process the signals measured by a conductivity probe and classify them into their corresponding flow regimes. Experiments were performed in an adiabatic air-water upward two-phase flow rig. Some statistical parameters of the cumulative probability density functions (CPDF) of the bubble chord length were used as the inputs to the ANN. Different ANN configurations were evaluated to optimize the characteristics that best suit the specific ANN application. The results demonstrate good agreement with the visual flow map identification, even for reduced temporal conductivity signals.