Cavitation is an essential problem that occurs in any pump. It highly contributes to deteriorating the performance of the pump. In industrial applications, it is important to detect and decrease the effect of the cavitation in pumps. In this work, detecting and diagnosing the cavitation phenomenon within centrifugal pumps using vibration technique was investigated. The results obtained for vibration signal in both time and frequency domains were analysed in order to gain a better understanding about the detection of cavitation in the pumps in question. The effect of different operating conditions including various flow rates and pump rotational speeds related to the cavitation were measured. Different statistical features in time domain analysis (TDA) and also the Fast Fourier Transform (FFT) technique for frequency domain analysis (FDA) were applied.

In centrifugal pumps, it is important to select appropriate turbulence model for the numerical simulation in order to obtain reliable and accurate results. In this work, ten turbulence models in 3-D transient simulation for the centrifugal pump are chosen and compared. The pump performance is validated with experimental results. The numerical results reveal that the SST turbulence model was closer to the experimental results in predicting head. In addition, the pressure variation trend for the ten models is very similar which increases and then decreases from the inlet to outlet of the pump along the streamline. The SST k- ω model predicts the performance of the pump was more accurately than other turbulent models. Furthermore, the results also found that the error is the least at design operation condition 300(l/min), which is around 1.98 % for the SST model and 2.14 % and 2.38 % for the LES and transition omega model. Within 7.61 %, the errors at higher flow rate 350(l/min) for SST. The error for SST model is smaller as compared to different turbulent models. For the Realizable k- ɛ model, the errors fluctuate were more high than other models.