Numerical Observations of a Stall Phenomenon in the NASA CC3 Compressor

Abstract

Numerical analysis of the transonic NASA CC3 centrifugal compressor stage with vaned diffuser is undertaken and the mechanism for the onset of stall investigated. The analysis methodology employs standard practice numerical methods, the mixing-plane approximation at the rotor-stator interface and periodic boundary conditions within a given row. Simulation results were then compared against full annulus time accurate simulations and the methodologies evaluated for estimation for the onset of stall. In the study, steady state and full annulus time accurate simulations were conducted at the 100% design speed, from choke to stall. Many time-resolved simulations “stalled” at pressure ratios where steady state mixing-plane solutions converged. One observation from the time-resolved simulations is that the number of revolutions required for the flow to stall was over 5 revolutions and increased to 9 revolutions closer to the last stable point. Integrations of the time-resolved solutions at the onset of stall reveal that the flow field of both the impellers and diffusers are not circumferentially uniform, and a stall “cell” is rotating in the opposite direction of the rotor. However, the “cell” is stationary in the stationary frame of reference thus, it is not a conventional rotating stall. There is a notable difference in choke flow between experiment and CFD results, which has been seen by others for this configuration. The reason for the miss is not a focus of this paper, and the numerically predicted unsteady stall point should not be viewed as the exact mechanism for stall within the NASA CC3 experiment, instead this work is to provide a review of standard practice numerical approximations and how cost effective full wheel unsteady analysis can be used to improve transonic compressor design.