Currently, relatively large errors are found in numerical results in some low-specific-speed centrifugal pumps with unshrouded impeller because the effect of clearances and holes are not accurately modeled. Establishing an accurate analytical model to improve performance prediction accuracy is therefore necessary. In this paper, a three-dimensional numerical simulation is conducted to predict the performance of a low-specific-speed centrifugal pump, and the modeling, numerical scheme, and turbulent selection methods are discussed. The pump performance is tested in a model pump test bench, and flow rate, head, power and efficiency of the pump are obtained. The effect of taking into consideration the back-out vane passage, clearance, and balance holes is analyzed by comparing it with experimental results, and the performance prediction methods are validated by experiments. The analysis results show that the pump performance can be accurately predicted by the improved method. Ignoring the back-out vane passage in the calculation model of unshrouded impeller is found to generate better numerical results. Further, the calculation model with the clearances and balance holes can obviously enhance the numerical accuracy. The application of disconnect interface can reduce meshing difficulty but increase the calculation error at the off-design operating point at the same time. Compared with the standard k-ε, renormalization group k-ε, and Spalart-Allmars models, the Realizable k-ε model demonstrates the fastest convergent speed and the highest precision for the unshrouded impeller flow simulation. The proposed modeling and numerical simulation methods can improve the performance prediction accuracy of the low-specific-speed centrifugal pumps, and the modeling method is especially suitable for the centrifugal pump with unshrouded impeller.
Currently, the flow field of annular seals disturbed by the circular whirl motion of rotors is usually solved using computational fluid dynamics(CFD) to evaluate the five rotordynamic coefficients. The simulations are based on the traditional quasi-steady method. In this work, an improved quasi-steady method along with the transient method was presented to compute the rotordynamic coefficients of a long seal. By comparisons with experimental data, the shortcomings of quasi-steady methods have been identified. Then, the effects of non-uniform incoming flow on seal dynamic coefficients were studied by transient simulations. Results indicate that the long seal has large cross stiffness k and direct mass M which are not good for rotor stability, while the transient method is more suitable for the long seal for its excellent performance in predicting M. When the incoming flow is non-uniform, the stiffness coefficients vary with the eccentric directions. Based on the rotordynamic coefficients under uniform incoming flow, the linearized fluid force formulas, which can consider the effects of non-uniform incoming flow, have been presented and can well explain the varying-stiffness phenomenon.
