Prediction of pressure drop during evaporation of R407C in horizontal tubes using artificial neural networks

Abstract

This work presents a discussion on pressure drop of R-407C during two-phase flows, and the application of artificial neural network (ANN) to predict these pressure drops in a smooth copper tube, for 4.5 mm and 8.0 mm inner diameter. The ANN was trained using data from 127 experiments encountered in the literature. Diameter, mass flux, saturation pressure and local vapor quality were used as inputs, whereas the pressure drop was considered as output. The number of neurons and hidden layers were determined based on the accuracy of results. The trained ANN was able to estimate the experimental data with a MAPE (Mean Absolute Percentage Error) of 6.11%, and a correlation coefficient (R) of 0.999 for all data, using a configuration with 14 neurons in the hidden layer. The obtained results were within ±10% for 90% of all data, and ±30% for 99% of all data. Compared to the well established literature correlations for pressure drop, the ANN demonstrates how important this tool is to predict pressure drop accurately.