Besides the inadequate NPSH ratio, cavitation may occur in a centrifugal pump because of the internal flow recirculation of the fluid. Recirculation can result in vortices, where the local pressure of the fluid is decreased below the vaporisation limit, resulting in cavitation. Flow recirculation may occur if the pump is driven at a partial flow rate. Typically, flow recirculation occurs in the operating region where the pump is already operating with a notably decreased efficiency. Hence, the detection of occurring flow recirculation provides an opportunity to avoid inefficient use of a centrifugal pump.
Since it seems possible to detect NPSH-related occurrence of a fully developed cavitation by a frequency converter, and flow recirculation may cause cavitation phenomenon in a centrifugal pump, the presented method may also be applicable to the detection of flow recirculation. In order to test the frequency converter’s applicability to detect flow recirculation especially in the case of radial flow centrifugal pumps, measurements were carried out with the Sulzer pumping system.
For the evaluation, a measurement sequence was performed at a 1450 rpm rotational speed by changing the operating point location with a control valve on the discharge side. With this method, the operating point location, shaft torque, rotational speed, radial vibration, acoustic emission and estimates of the frequency converter were measured and stored at 14 different flow rates ranging from 0 to 143 % of the SulzerQBEP. Each measurement was performed when the pump was operating in the steady state. The NPSHA/NPSHR ratios were calculated for each operating point on the basis of the measured water temperature, pump flow rate and suction pressure values (see (4.4)). However, the NPSH ratio values have not been calculated for the flow rates below 7.5 l/s (27 % of the QBEP), since there is no given NPSHR curve for this operating region. Since the pump manufacturer has given the allowable operating region with
the NPSHR curve, the pump may be susceptible to the flow recirculation and other harmful phenomena at these partial flow rates. The obtained values for the NPSH ratios are shown in Fig. 4.41, where a limit value of 1.35 has been applied as a threshold for the occurrence of NPSH-related, head-decreasing cavitation.
0 10 20 30 40 50
0 4 8 12
Flow rate (l/s) NPSH A/NPSH R
Risk of flow recirculat ion
T hreshold value for cavit at ion occurrence
Fig. 4.41: NPSHA/NPSHR ratios as a function of flow rate in the measurement sequence with a constant rotational speed. In this measurement sequence, the NPSH ratios are sufficient to avoid the cavitation occurrence. However, the pump may be susceptible to flow recirculation and other harmful events at reduced flow rates.
For comparison, the results of the AE measurements are shown in Fig. 4.42. They support the assumption of flow recirculation occurring in the pump at flow rates below 7.5 l/s, as in this region the highest values are attained for the AE activity. Since the AE values increase with a decreasing flow rate below 23 l/s, an incipient occurrence of flow recirculation is possible also in the flow rate region of 8.9–14.4 l/s, as the AE values are there larger than at the maximum flow rate of 40 l/s.
0 10 20 30 40 50
0 10 20 30 40 50 60
Flow rate (l/s)
AE (dB)
Fig. 4.42: Measured acoustic emission as a function of flow rate. The AE reaches its maximum value when the pump flow rate is at its minimum, and there is an apparent risk of flow recirculation occurrence in the pump.
Pre-filtered estimates of the rotational speed and shaft torque were applied to determine nRMS/nRMS,N andTRMS/TRMS,N as a function of flow rate. Normal valuesnRMS,N andTRMS,N were calculated from the RMS values of the rotational speed and shaft torque, which have a sufficient flow rate (Q > 7.5 l/s) to avoid the occurrence of flow recirculation. The results are shown in Fig. 4.43 for both estimates. There is a notable increase in the RMS values, when the flow rate is below 10 l/s and the risk of flow recirculation is apparent. This supports the hypothesis that also the flow recirculation of a centrifugal pump can be detected by a frequency converter.
0 10 20 30 40 50
0 1 2 4 5
Flow rate (l/s) n RMS/n RMS,N
0 10 20 30 40 50
0 1 2 4 5
Flow rate (l/s) T RMS/T RMS,N
Fig. 4.43: nRMS/nRMS,N andTRMS/TRMS,N ratios as a function of flow rate. The estimated values are highest at partial flow rates, especially below 10 l/s, where the risk of flow recirculation is also more apparent.
The shaft torque measurements were analysed to see if the flow recirculation has similar effects as the NPSH-related cavitation. The frequency content of the measured shaft torque is illustrated in Fig. 4.44. As previously, the energy of the measured shaft torque is concentrated on the frequencies of 84 Hz and 96 Hz, which is the blade passing frequency of the pump. The energy content at these frequencies increases when the pump is operating away from its best efficiency point. In addition, there is a low frequency component (0–10 Hz) in the shaft torque at flow rates below 10 l/s. It can be concluded that the occurrence of flow recirculation has similar effects on the shaft torque as the occurrence of cavitation.
Flow rate (l/s)
Fig. 4.44: Welch power spectral density of the measured shaft torque as a function of flow rate. The spectral density values are highest at partial flow rates, especially below 10 l/s, where the risk of flow recirculation is apparent. At these flow rates, a low-frequency variation (0–10 Hz) is visible in the shaft torque.
Also the unfiltered estimates for the rotational speed and shaft torque were analysed. As assumed, the occurrence of flow recirculation is visible in the low-frequency region (0–10 Hz) of their spectrograms. In Fig. 4.45, a spectrogram for the unfiltered rotational speed estimate is presented. Besides the low-frequency component (0–10 Hz) that is apparent at flow rates below 10 l/s, the rotational speed and the blade pass frequency component are visible in the spectrogram. However, there is no visible frequency component around 80 Hz as in the spectrogram for the measured shaft torque. Thus, also the occurrence of flow recirculation seems possible to detect by monitoring the low-frequency variation of the rotational speed and the shaft torque at least in the case of the Sulzer laboratory pumping system. Naturally, these results may not hold true for all other pumping systems and for other pump types. However, it can be assumed that the corresponding phenomenon could be visible with similar pumping systems having a radial flow centrifugal pump.
Flow rate (l/s)
Fig. 4.45: Welch power spectral density of the estimated rotational speed as a function of flow rate. The most notable component is in the low-frequency region (0–10 Hz) when the flow rate is below 10 l/s.