Assessment of an identification strategy for the prediction of the dynamics of two-phase flows

The present study aims at assessing a flexible and light modeling methodology specifically oriented to the short-term prediction of the dynamics of the void fraction of experimental two-phase flows. The proposed strategy consists in the assessment and optimisation of a generalised NARMAX model for the input-output identification of the dynamics of the experimental time series of the void fraction, detected through a high resolution resistive probe during an extensive experimental campaign. Such a model has been implemented by means of Multilayer Perceptron artificial neural networks, trained using input-output data detected during experiments expressing different flow patterns. Reported results show that a satisfactory agreement is reached between simulated and experimental data, showing that the model is able to predict two-phase flow dynamics.