Effect of Vibration and Noise Measuring Points Distribution on the Sensitivity of Pump Cavitation Diagnosis

Abstract

Cavitation is an essential factor in the deterioration of the hydraulic performance of centrifugal pumps. The study of cavitation fault diagnosis can help prevent or reduce the damage it causes. The vibration and noise analysis method can predict the incipient cavitation more accurately. In order to improve the accuracy of cavitation fault diagnosis, this paper studied the sensitivity of measuring points distribution for centrifugal pump cavitation diagnosis. The research object is a centrifugal pump with an inducer and splitter blades. Vibration acceleration sensors and hydrophones were used to collect structural vibration and liquid-borne noise signals at different positions of the pump unit. Root-mean-square (RMS) and fast Fourier transform (FFT) methods were used to construct spectrums of vibration and noise signals with different NPSHa and compare the sensitivity of different measuring points to the inception and development of cavitation. In addition, the SST k-ω turbulence model and Zwart cavitation model were used to study the cavitation volume distribution in the pump under different cavitation stages. By setting monitoring points at the impeller outlet, the frequency domain signal distribution of pressure pulsation was studied. The results show that the vibration measuring points at the inlet flange and pump axial position (increased by about 0.6 % at NPSHr) and liquid-borne noise measuring point at the inlet position (reduced by about 14 % at NPSHr) are more sensitive to the diagnosis of cavitation fault. Motor current is also the basis for judging the inception of cavitation. When severe cavitation occurs, the current drops sharply by approximately 12 %. Moreover, the pressure pulsation intensity at the inlet decreases by 66.3 %and by increases 13.9 %at the outlet, respectively, with a 3 %drop in head. As the cavitation intensifies, the dominant frequency of the pressure pulsation in the pump is partially shifted. The presented results indicate the distribution of measuring points with good sensitivity, providing a reference for improving the accuracy and efficiency of cavitation predictions for centrifugal pumps.