Study of Performance Evaluation and Design Techniques for a Multiphase Pump

다국어 초록 (Multilingual Abstract)

The crude oil produced from well contains a mixture of oil, gas and water. When developing an oil well, the pump system for the pressure rising is applied to increase the production rate. The existing pump system that uses a single phase pump requires a separator to separate the crude oil. Changing from a single phase pump to a multiphase pump significantly reduces costs because a multiphase pump does not require a separator. Therefore, most wells currently being developed apply the multiphase pump system, and related studies are rapidly increasing. Studies related to the multiphase pumps must consider the characteristics of multiphase flow. Consequently, these studies are much more difficult compared to studies of general fluid machinery.
This study presents the performance evaluation and the design techniques for a Multiphase Pump. To perform the study, the design specifications were selected considering the existing multiphase pump model, and a base model was designed using design theories for fluid machinery. The single phase and multiphase performances of the base model were evaluated through numerical analyses by applying a variety of numerical method. In addition, the performances of the base model were evaluated through experiments to verify the reliability of the numerical analysis, and each performance was compared and analyzed. Based on comparison results, the numerical analysis schemes suitable for the performance evaluation of a multiphase pump was established.
To improve the performance of the base model, design optimization was carried out using techniques that combine design of experiment (DOE) with proven numerical analysis method. For design optimization, the design variables and variable ranges were selected for the impeller and the diffuser. 2k factorial experiment conducted to select final design variables that had major effects on the pump performance. Experiment sets for optimization were generated based on final design variables and performances of the generated sets were evaluated using the numerical analysis. The results of the performance evaluation were used to select an optimum model that could satisfy the objective function through the analysis of the regression equation of the response surface method (RSM). Improvement of the performance of the optimum model was confirmed through the numerical analysis and the experiment. The reasons for the performance improvement were analyzed through comparisons of the internal flow fields of the base model and the optimum model.