The pressure fluctuation caused by unsteady flow in runner is one of the main reasons of vibration for a large Francis hydraulic turbine. It directly affects the steady operation of the hydraulic turbine unit. The existing research of the pressure fluctuation in hydraulic turbine mainly focuses on the unsteady flow in draft tube. Accurate distribution of pressure fluctuations inside a runner is not very clear. In this paper, the numerical method for predicting the pressure fluctuations in runner is investigated and the numerical simulation is performed for a large Francis hydraulic turbine. It is proved that the combination of shear-stress transport（SST） k-o） turbulence model and pressure-implicit with splitting of operators（PISO） algorithm could give more reliable prediction of pressure fluctuations in runner. The frequencies of pressure fluctuations in runner are affected by the flow in guide vane and the flow in draft tube The first dominant frequency is significantly determined by the flow in draft tube, especially at part load condition. This frequency is approximately equal to one-third of the runner rotating frequency. The evident second dominant frequency is exactly equal to the guide vane passing frequency. The peak-to-peak amplitudes of pressure fluctuations in runner at small guide vane open angle are larger than that at large open angle at the same operating head. The amplitudes at points on blade pressure surface are generally greater than that on suction surface. The research results could be used to direct the hydraulic design and operation stability improvement of a large Francis hydraulic turbine.
The current research of large eddy simulation （LES） of turbulent flow in pumps mainly concentrates in applying conventional subgrid-scale （SGS） model to simulate turbulent flow, which aims at obtaining the flow field in pump. The selection of SGS model is usually not considered seriously, so the accuracy and efficiency of the simulation cannot be ensured. Three SGS models including Smagorinsky-Lilly model, dynamic Smagorinsky model and dynamic mixed model are comparably studied by using the commercial CFD code Fluent combined with its user define function. The simulations are performed for the turbulent flow in a centrifugal pump impeller. The simulation results indicate that the mean flows predicted by the three SGS models agree well with the experimental data obtained from the test that detailed measurements of the flow inside the rotating passages of a six-bladed shrouded centrifugal pump impeller performed using particle image velocimetry （PIV） and laser Doppler velocimetry （LDV）. The comparable results show that dynamic mixed model gives the most accurate results for mean flow in the centrifugal pump impeller. The SGS stress of dynamic mixed model is decompose into the scale similar part and the eddy viscous part. The scale similar part of SGS stress plays a significant role in high curvature regions, such as the leading edge and training edge of pump blade. It is also found that the dynamic mixed model is more adaptive to compute turbulence in the pump impeller. The research results presented is useful to improve the computational accuracy and efficiency of LES for centrifugal pumps, and provide important reference for carrying out simulation in similar fluid machineries.
