Haiwen Zhu (Doctor of Philosophy in Petroleum Engineering)
Experiments, CFD Simulation and Modeling of Sand Wear and Performance Degradation in ESPs
Conclusions and Recommendations
A closed-loop experimental facility was designed and constructed to study the sand erosion process in a mixed type ESP, which is referred as MTESP in this dissertation. Four sets of experiments were conducted under different rotation speeds and gas volumetric fraction to investigate erosion effect on ESPs performance, including boosting pressure, efficiency, and vibration. Geometries, including impeller hub outside diameter (OD), impeller hub inside diameter (ID), impeller skirt ring OD, impeller balance ring OD, impeller outside shroud OD, sleeves OD, diffuser skirt slot depth, diffuser outside shroud ID, diffuser skirt slot ID, diffuser balance slot ID and diffuser bore ID were measured after each time-period test to indicate abrasion in ESPs. Stages were painted to obtain erosion patterns, and weight loss was recorded to compare with the predicted erosion rates. Three ESPs, including two mixed type pumps (DNI 750 and MTESP) and one radial type pump (TE2700), were numerically modeled to study the pump type effect on ESP erosion. For each pump, the flow domain of two stages was selected and high-quality structured meshes, comprising 1.2 to 1.8 million hexahedral grids per stage, were generated by ICEM or Turbogrid. Six erosion models were evaluated to obtain the most accurate model in ESP erosion prediction.
Based on the seal failure analysis and wear mechanism, an improved mechanistic model for prediction ofESP performance in terms ofboosting pressure has been developed. The erosion model was chosen based on the comparison among the calculated erosion rate and test results by CFD methods. The method of abrasion prediction is based on Archard (1953) equation. Leakage loss in ESP is calculated with the eroded hub, sleeves, and seals. The discrepancy in model predictions can be improved by adding closure relationships, including seal forces and leakage coefficient correlations.