Experimental analysis of flow instability detection in a centrifugal compressor using variational mode decomposition

Abstract

AbstractIn centrifugal compressors, the identification of flow instability signals from experiments is a difficult problem owing to the nonlinear and non-stationary characteristics. Otherwise, the complicated asymmetric structure of the volute brings a huge challenge to the evolution and circumferential nonuniformity characteristics of the flow instabilities. This paper presents experimental and numerical investigations on internal flow field to understand the flow instability characteristics in a centrifugal compressor. Considering nonlinear and non-stationary signals, a method based on Fourier-transform and variational mode decomposition was introduced to analyse the flow instability characteristics. The Fourier spectrum results show that at 0.21kg/s of 80krpm, the pressure signal has a noticeable high-frequency fluctuation, which indicates that the compressor enters the flow instability state. The variational mode decomposition results show that before a surge, the compressor experiences different flow instability stages: the RI stage, the coexistence stage of RI and stall, and the stall stage. Moreover, obvious circumferential nonuniformity characteristics of flow instabilities were observed during the throttling process. RI first occurred at the 180° circumferential position and then the stall first appeared in the circumferential range of 60° to 240°. The simulation results that it is because that the asymmetric volute causes the adverse pressure gradient inside the impeller passage and a high-pressure region (120°–240°) at the upstream of the impeller inlet. Under this combined action of the two, the effect region of tip leakage vortex expands the upstream of the impeller inlet. Meanwhile, the tip leakage vortex core migrates to a lower span of blades. This study demonstrates the ability to analyse nonlinear and non-stationary signals from a centrifugal compressor via variational mode decomposition, and provides a useful guidance for the identification of flow instability signals.