This paper presented a methodology of artificial neural network (ANN) for the prediction of flow patterns in two-phase air-water flow along upward inclined pipes. In the built ANN model, superficial velocity of air, superficial velocity of water, and inclined angle were set as inputs while the quantified flow patterns were defined as the output. In total, 1952 experimental data points that were reported in the literature were trained and tested by the designed network structure. The predicting accuracies of stratified smooth, stratified wavy, annular, intermittent, bubble flow are all above 90%, with the exception of dispersed bubble flow.
Keywords artificial neural network (ANN); two-phase flow; flow patterns.