Keywords: cavitation, spectrum, centrifugal pump, rotational speed, blade passing frequency, broadband frequency, time domain, frequency domain.

The use of the vibration method for the condition monitoring of rotating machinery has been well established over recent decades. Many studies have shown the positive results of using the vibration method for monitoring the condition of centrifugal pumps. This paper reviews the sources of vibration in a centrifugal pump during normal operation and explains how the pump vibration level is excited by the onset of cavitation.

Vibration signals at seven different flow rates (200, 250, 300, 325, 350, 380 and 400) were collected. These signals were acquired using an accelerometer. The vibration signal from the accelerometer was filtered at the input of the linear amplifier, which raised the vibration signal to a level suitable for the data acquisition system. The collected data was sampled at 62.5 kHz with 842176 data points. The acquired vibration signals were averaged using the data segment averaging method. The segment averaging method was done by dividing the length of the collected data (842176) by the nfft (131072) required for the fast Fourier transform (FFT) calculations. This procedure will divide the acquired signal into six equal segments, giving a frequency resolution of 4.78 Hz. The spectrum of each segment was calculated by applying the FFT algorithm. The final spectra were obtained by averaging the six spectra. It is important to know that the accelerometer has a linear response up to 20 KHz. The vibration signals were analysed in the time and frequency domains.

In this work vibration signals were acquired, processed and analysed for cavitation monitoring of a centrifugal pump. The acquired vibration signals were analysed in the time and frequency domains. PDF, RMS, SD are used for analysing vibration data in the time domain. The RMS and the SD for vibration signal analysis agrees with the mechanism generating vibration in pumps, but these parameters can be used only for detecting cavitation not for diagnosis. In the frequency domain, the spectrum analysis particularly the mean values of the vibration spectra can be a reliable method for cavitation detection and diagnosis. In this study in addition to broadband frequency, a low frequency range (20Hz - 20 KHz) was found to be a useful method for detecting and diagnosis cavitation in a centrifugal pump. Using a low frequency response sensor reduces the cost of the monitoring system.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £145 +P&P)