Experimental identification of flow regimes in gas–liquid two phase flow in a vertical pipe

Abstract

In this experimental study a new flow regime map for gas–liquid two-phase flow regime detection is presented based on interpretation of transient flow pressure signals and utilizing neural networks. In this regard, the pressure signals of an air–water two phase flow at different location on a vertical pipe with an inside diameter of 50 mm and a total length of 6 m are measured. Averaged pressure and pressure fluctuation data were analyzed in different flow regimes along the upriser pipe to recognize the effects of flow regimes on pressure response. A set of pressure parameters are introduced based on averaged pressure data to present meaningful criteria for flow regime identification. Pressure fluctuations also used for flow regime recognition by applying two common signal processing methods. The processed signals are used as the inputs of a feed forward back propagation neural network to predict the flow regime. The neural network outputs are then used to present a flow regime map. The proposed neural network flow map is compared with the experimental data by photographic flow visualization technique. This comparison shows an acceptable agreement between the neural network outputs and experimental results.