Singular spectrum analysis as a tool for early detection of centrifugal compressor flow instability

Centrifugal compressor machinery is subject to a potentially damaging phenomenon called surge at low mass flow rates. This effect may be preceded by phenomena known as inlet recirculation — reversal upstream of the impeller. A methodology isolate characteristic feature for monitoring centrifugal instability presented in this study. The based on nonparametric time series analysis technique singular spectrum (SSA). SSA decomposes signal into number Reconstructive Components (RCs), from which data trends and oscillatory components extracted. frequency spectra each RC their relative contributions reconstruction original were examined comparisons made with spectral maps existing literature. Individual independent RCs chosen construct compressor’s system. Additionally, performance was determined Window length parameter. modification parameter also studied, various viable choices component basis diagnosis considered. implemented pressure dynamical signals measured an experimental rig. High measurements taken conditions locations within compressor. results demonstrated potential identify extract information about local eventually successfully monitor instability.