Prediction of Pressure Gradient in Two and Three-phase Flows in Horizontal Pipes Using an Artificial Neural Network Model

Abstract

Concurrent flow of gas with a mixture of oil and water in production equipment is common necessitating the need for additional investigations to gain more insight and development of more accurate correlations for prediction of flow characteristics including pressure drop. In this study, an experimental study was conducted using air-water and air-water-oil mixtures in a 0.075-m diameter pipe. Superficial gas and liquid velocities ranged from 0.03 to 0.13 m/s and 1.26 to 41.58 m/s respectively. Slug flow was the main flow pattern observed. In addition, transition to annular and annular flow were also observed. Due to the homogeneous nature of the oil-water-air mixture, the three-phase flow was evaluated as a pseudo-two-phase mixture. An Artificial Neural Network (ANN) model developed for the prediction of two- and three- phase pressure drop performed better than all models considered during the evaluation. Generally, it is found that the accuracies for pressure drop were considered adequate.