The prediction of cavitation in high speed centrifugal pumps.

Abstract

The primary focus of this research was to analyse the flow through a high speed centrifugal turbopump and validate the original meanline design. This research forms part of a larger project which is to develop a kerosene turbopump capable of launching a hypothetical vehicle into a sun synchronous orbit. The original design work was done by Smyth (2013) which dealt with the design of the impeller. The next step, dealt with by Philogene (2014), was concerned with an experimental test rig capable of testing a scaled version of the original impeller. Unfortunately the test rig has not yet provided experimental data for the impeller leaving the meanline design as the only source of comparison for this project. Before undertaking the computational analysis of the high speed impellers, a review was conducted into the parameters that should be used in the computational fluid dynamic models. The optimum mesh configuration was determined to be unstructured polyhedral cells away from the walls and prismatic layers near the walls to accurately capture the boundary layer flow. The SST K-Omega turbulence model was deemed sufficient to model the turbulent eddies within the flow. The steady state models were run using a rotating reference frame which was applied to the impeller region of the model. The steady state results for the scaled impeller deviated from the meanline analysis by 12% at the operating point. This deviation increased as the flow rate was increased. The full scale model provided excellent steady state results with regards to the agreement with the meanline design. At the operating point of the full scale rotor, the difference between the results was only 1.31%. At 130% of the operating flow rate the difference between the two sets of results was only 8.5%, a remarkable improvement from the scaled impeller. In order to model the cavitation performance of each impeller an unsteady analysis was conducted. A multiphase flow model, the Volume of Fluid model, was employed. This model recognises that the flow has two separate fluid phases present. The formulation of cavitation bubbles was then controlled by the Rayleigh Plesset Equation which determines the changing bubble radius. The results of the cavitation modelling showed no significant change in the scaled impeller performance, which exhibited less cavitation. The low percentage of cavitating flow also meant that the flow was stable in terms of the pressure outlet fluctuations. However, the full scale model contained a much larger cavitating percentage of flow. This drastically affected the performance of the impeller, dropping the outlet pressure by 11 b at the operating point. Although both impellers were designed to exhibit similar levels of cavitation, the numerical results show differences which could be due to the complex process of cavitation scaling.