The Influence And Identification Methods Of Unsteady Flow On Blade Vibration

In terms of centrifugal compressor, impeller as the weakest part suffers centrifugal force, fluid flow effect, acoustic action and complex coupling interactions at the same time, as a result, the impeller fault diagnosis of fatigue failure is the most difficult problems. Due to irregular geometric shapes and complex unsteady flow inside the compressor, flow-induced excitation phenomena are widespread throughout the compressor, therefore it is need to investigate the expression form of these flow-induced excitation phenomena and the influence of unsteady flow, so as to grasp the mechanism of unsteady flow on blade vibration, from the source to avoid blade vibration and to prevent blade fatigue failure. This paper mainly from three aspects of acoustic resonance, surging and rotating stall analyzed the unsteady flow to get three cases unsteady flow characteristic, meanwhile, studied the extraction method of weak blade cracking fault feature.Firstly, acoustic mode simulation analysis was investigated on three different diffusers including full-height vaned diffuser, half-height vaned diffuser and vaneless diffuser for the same centrifugal compressor, then acoustic resonance experiment was designed. The pressure pulsation level would increase typically when the blade passing frequency (BPF) was close to the acoustic modal frequency for full-height vaned diffuser. In order to avoid the acoustic resonance conditions, by changing the diffuser types could obviously reduce the pressure pulsation level. It illustrated the blade damage mechanism of unstable flow caused by acoustic resonance and the identification methods of acoustic resonance.Secondly, the characteristic of pressure pulsation at diffuser inlet was tested under different flow rate, results showed that mild surging would occur at the minimum flow rate, and as the flow rate decreased again, deep surge would occur. Through the strain test, it was found that the blade vibration frequency and pressure pulsation main frequency was consistent when deep surging occurred. It was the first time to verify the effect of compressor surging on pressure pulsation and blade vibration by experiment, to fundamentally explained the mechanism of surging on the blade damage.Thirdly, the effect of inlet guide vane opening on the pressure pulsation was investigated since when the guide vane opening was reduced, compressor performance curve would move to the upper right, that is to say, rotating stall and surging were more likely to occur. The results showed that the smaller inlet guide vane opening angle, the larger amplitude of pressure pulsation generated by the low frequency which was the frequency of rotating stall, that is, rotating stall was occurred in the compressor and caused unsteady flow. Meanwhile, the effect of guide vane opening on pressure pulsation RMS was tested under three kinds of flow rate, and the results revealed that the smaller the flow rate, the greater overall level of pressure pulsation RMS.Finally, the extraction method of weak blade cracking fault feature was investigated. Through the simulation of the blade cracking, pressure pulsation signal was used to monitor the blade crack information. As the signal of blade crack failure was belong to low frequency weak fault, that is to say, the fault frequency was often modulated to the blade passing frequency. Especially the crack was hard to identify in the early stage, and later quickly fracture. A method to blade crack extraction was present by wavelet filtering at the blade passing frequency and envelope spectrum analysis was used. The method of WSG stochastic resonance was applied to obtain the characteristic frequency by double frequency method, namely blade crack fault frequency. Then the strain experiment was carried out to verify the blade crack fault frequency. The results showed that WSGSR model was an effective and reliable tool for blade cracking early fault detection on centrifugal compressor by simulation and experiment analysis.