Detection of Cavitation in Kaplan Water Turbines (original) (raw)

Cavitation causes excessive pressure pulsations, which damage the surfaces of the runner and channels of a turbine. As a result, the overall operating efficiency of the water turbine decreases and repair costs increase. Traditionally, there have been efforts to detect cavitation using vibration, pressure and acoustic emission measurements. For instance, extensively used vibration velocity measurements are not effective enough to detect all cavitation areas so more sensitive and accurate signal processing methods are still demanded. This study concentrates on vibration measurements in the real operating environment of a Kaplan turbine. Altogether 29 measurement periods were carried out at different power levels from 1.5 to 59.4 MW. The vibration analysis was based on the use of traditional velocity and acceleration signals and novel higher order derivatives: x(3) and x(4). The features used were rms, peak, kurtosis and crest factor. Normal accelerometers could be used and the upper cutoff frequency did not have to be high in order to detect all cavitation areas reliably. The sample length has to be over 30 seconds in order to detect all cavitation areas accurately. The rms value works sufficiently well at high powers, whereas kurtosis and crest factor are effective at low powers only. The feature working well throughout the whole power range is peak value.