CFD Analyses of Centrifugal Pumps with Emphasis on Factors Affecting Internal Pressure Pulsations

The operation of centrifugal pumps can generate instabilities and pressure pulsations that may be detrimental to the integrity and performance of the pump. Until recently these pressure pulsations could only be determined experimentally which resulted in a limited understanding of pressure pulsations around the pump. Industrial pump guarantees are limited to pulsation levels measured at the discharge. However, numerical analysis techniques have advanced to such a stage that they can now be used to explore these effects. The multi-block, structured grid CFD code TASCflow was used to investigate the time variation of pressure within a complete centrifugal pump. A parametric study covered four geometric parameters, namely the cutwater gap, vane arrangement, snubber gap and the sidewall clearance. Taguchi methods allowed the number of transient analyses to be limited to a total of twenty seven. Three flow rates were investigated and the pulsations were extracted at fifteen different locations covering important pump regions. The velocity flow patterns from the transient analyses exhibited important features that were in agreement with two independent sources. The transient flow results compared reasonably with the Weir experimental tests and clearly indicated the pump locations experiencing the largest pulsation levels. It was also noted that monitoring pulsations at the top dead centre of the pump volute casing would provide a better indication of internal pump pulsations than monitoring at the discharge. Taguchi post-processing analysis tools were used to rank the relative importance of the four geometric parameters at each location for each flow rate. The cutwater gap and vane arrangement were found to exert the greatest influence across the various monitored locations and the flow range. However the snubber gap had a dominant influence on the pressure differential across the impeller shroud and pulses in the pressure differential were evident at reduced flows. Through a rationalisation process reductions in pressure pulsations aimed at increased component life and reduced noise/vibration have resulted in a single recommended geometric arrangement. Further analyses confirmed that the new arrangement did indeed produce lesser pulsations levels. Multiple steady state simulations were analysed to determine if they were a viable substitute for the transient analyses. However it was demonstrated that the steady state pulsations did not adequately capture the magnitude and phase of the pulsations shown by the transient results. Likewise the steady state analyses were unable to predict trends for two differing pump geometries. In order to identify the implications of the CFD data for mechanical integrity, the pressure differential predicted by the transient analyses was compared with the pressure loadings currently utilised in Weir design guidelines; this resulted in a new recommendation for use in future designs. Also finite element analyses were conducted using four pressure loadings taken from the numerical results and a centrifugal loading. These supported the recommendation for an increased loading to be used in the design guidelines. The stress levels at the impeller outlet were found to be extremely sensitive to the snubber gap. The completion of this project has allowed a useful set of recommendations to be made regarding the design of high head double entry pumps.