摘要：For pump–turbines, most of the instabilities couple with high-level pressure fluctuations, which are harmful to pump–turbines, even the whole units. In order to understand the causes of pressure fluctuations and reduce their amplitudes, proper numerical methods should be chosen to obtain the accurate results. The method of large eddy simulation with wall-adapting local eddy-viscosity model was chosen to predict the pressure fluctuations in pump mode of a pump–turbine compared with the method of unsteady Reynolds-averaged Navier–Stokes with two-equation turbulence model shear stress transport k–ω. Partial load operating point (0.91QBEP) under 15-mm guide vane opening was selected to make a comparison of performance and frequency characteristics between large eddy simulation and unsteady Reynolds-averaged Navier–Stokes based on the experimental validation. Good agreement indicates that the method of large eddy simulation could be applied in the simulation of pump–turbines. Then, a detailed comparison of variation for peak-to-peak value in the whole passage was presented. Both the methods show that the highest level pressure fluctuations occur in the vaneless space. In addition, the propagation of amplitudes of blade pass frequency, 2 times of blade pass frequency, and 3 times of blade pass frequency in the circumferential and flow directions was investigated. Although the difference exists between large eddy simulation and unsteady Reynolds-averaged Navier–Stokes, the trend of variation in different parts is almost the same. Based on the analysis, using the same mesh (8 million), large eddy simulation underestimates pressure characteristics and shows a better result compared with the experiments, while unsteady Reynolds-averaged Navier–Stokes overestimates them.
