A numerical study on flow patterns inside an electrical submersible pump (ESP) and comparison with visualization experiments

This paper presents a numerical study of flow pattern recognition inside the rotating impeller of an electrical submersible pump (ESP) using the transient multiphase Computational Fluid Dynamics (CFD) simulations. The calculation domain is constructed based on the previous experimental facility for visualizing flow patterns in an ESP (Barrios (2007)). The high-quality structured mesh comprising hexahedral grids is generated using multi-block technique in ANSYS ICEM. For CFD simulations, the realizable k-epsilon turbulence model with volume of fluid (VOF) and Eulerian-Eulerian multiphase models is successfully implemented in ANSYS Fluent solver. The sliding mesh technique is applied to interfaces where rotating and stationary parts interact. By incorporating the same boundary conditions as Barrios experimental study, three flow cases with constant gas flow rates and varying liquid flow rates are selected to conduct numerical simulations. The comparison of simulation results with Barrios' observations shows that the Eulerian-Eulerian multiphase model is superior to VOF model for simulating gas-liquid two-phase flow in a rotating ESP. The single-phase simulation results match catalog curves, which validates the numerical methodology. For two-phase simulations, the simulated flow patterns using Eulerian-Eulerian model agree well with visualization experiments. Different flow patterns prevailing inside the rotating ESP impeller are captured. At low gas flow rate, bubbles are dispersed in liquid phase, and the flow pattern is categorized as dispersed bubble flow. As gas flow rate increases, bubbles can accumulate and coalesce, causing large gas-pocket formation leading to intermittent/slug flow. Transient multiphase CFD simulation is an efficient and reliable tool to predict flow patterns inside ESPs.