Simulação numérica do escoamento bifásico líquido-gás em uma bomba centrífuga radial

Abstract

Centrifugal pumps are commonly used for transportation or circulation of fluid mixtures. Electrical Submersible Pumping became the second most used oil and natural gas artificial lift method. On the other hand, cooling cycles are often subject to fluid evaporation, resulting in a two-phase flow especially hazardous to the nuclear industry. The presence of a gaseous phase can lead to severe performance degradation, in a phenomenon called surging. The gas starts to accumulate and occupy most part of the impeller channels, being capable of shutting down the whole operation. Under this context, the present work aims to employ the Ansys® CFX® computational fluid dynamics tool to numerically reproduce the performance and flow dynamics of a pump operating liquid and gas. Specifically, reproduce experimentally obtained images and data that came from a centrifugal pump of same geometry. Water and air were used to obtain performance curves with rotating speed between 300 and 600 rpm, normalized liquid volumetric flow rate between 0,2 and 1,5 and constant air mass flow rate. Both the Euler-Euler Two-fluid and the polydisperse MUSIG models were employed, the latter considers breakage and coalescence mechanisms to model the dispersed gaseous phase. The numerical model was capable of predicting the pump’s performance drop, but overestimated its operational window, indicating a degradation to lower liquid volumetric flow rates, when compared to the experiment. Flow patterns were also successfully reproduced, bubble breakage and coalescence effects inside the impeller have shown to be crucial in the modelling of the phenomenon. The present work contributes with the literature by comparing with detail the pressure increment and flow dynamics obtained numerically with experimental data and by demonstrating the robustness of the tool and models used. It can be used as starting point to further studies involving other operating conditions and models’ combinations.