Using machine learning algorithms to predict the pressure drop during evaporation of R407C

Abstract

The calculation of the pressure drop for two-phase flow in evaporation and condensation processes is required by a variety of design practices. In recent years, many correlations were developed in order to determine the pressure drop for two-phase flow. This process needs many experimental tests. Hence, in this study, it is proposed to apply machine learning algorithms (MLAs) to forecast the pressure drop for two-phase flow of R407C. Three methods of MLAs are developed with the purpose of pressure drop prediction in a smooth horizontal copper tube, for 4.5 mm and 8 mm inner diameter. These methods are multilayer feed-forward neural network (MLFFNN), support vector regression (SVR), and group method of data handling (GMDH) type neural network. Mass flux, tube diameter, saturation pressure, and vapor quality of the refrigerant are used as input variables of the models and the target is selected to be the pressure drop of evaporation. The results show that although the developed models can successfully predict the pressure drop of two-phase flow, MLFFNN and GMDH models outperform the SVR model in term of the correlation coefficient close to 1.