Compressed dynamic mode decomposition for the analysis of centrifugal compressor volute

Highlights: • Analyze the unsteady characteristics of the flow in a centrifugal compressor volute under mild surge operating condition. • The dominant flow structures can be extracted exactly by the dynamic mode decomposition. • The compressed dynamic mode decomposition is more effective than the dynamic mode decomposition. • The compression ratio has an important influence on the results of compressed dynamic mode decomposition, including the eigenvalues and energy ratio spectrum. The compressed dynamic mode decomposition (compressed DMD) is used to analysis the unsteady characteristics of a centrifugal compressor. Firstly, to extract the unsteady flow structures of the volute under the mild surge condition, dynamic mode decomposition (DMD) method is applied to the flow field snapshots. The results indicate that the characteristic frequencies relating to the blade passing frequencies (BPFs) and corresponding modes are captured. Besides, the perturbation oscillating with the characteristic frequency (233.5 Hz), which is generated by the non-synchronic unsteady flow, is also obtained accurately. The non-synchronic unsteady flow is vividly shown by the reconstruction of the mild surge mode. Ten, the unsteady flow of the volute is also analyzed by the compressed DMD method, which is performed on the much compressed datasets and thus saves the calculation time. The dominant characteristic frequencies and corresponding modes can also be extracted accurately by the compressed DMD method when adopting an appropriate compression ratio. The compression ratio has an important influence on the results of the compressed DMD method. Considering the dominant frequencies and the energy ratio spectrum, a conservative recommendation of the compression ratio is 1% in this case. The energy ratio scaling method performs better than the amplitude scaling method.