Prediction of the pressure drop in water-high viscosity oil flows using artificial neural network
This study investigated the ability of the improved artificial neural network to calculate the pressure drop in two-phase flows (water, high-viscosity oil) in the horizontal tube. After setting the experimental system, first the experimental data related to two-phase flows in slow and turbulent regimes was collected. Flow patterns at different fluid velocities were recorded by a high-speed camera, and pressure drop was measured by two pressure transmitters mounted on the tube. The friction coefficient was calculated by the previous correlations. The desired neural network was coded in C++ environment, then 243 data were used to train the network. Also, 81 and 82 data were used for validation and network testing, respectively. In this research, diameter, roughness, density, viscosity, water and oil superficial velocities and friction coefficient were considered as inputs. Finally, a model for predicting the pressure drop was presented. Next, the developed model was compared with previous correlations.