An OEM and an end user had installed a number of large, double-suction pumps for a water pumping station project (Figure 1). Soon after, they began to suspect that the pumps were a significant source of undesirable noise. The pumping station's owner and the OEM needed to better understand if the noise was the result of cavitation, and subsequently needed to assess the potential for the cavitation to significantly reduce pump life in this installation. Of particular concern was a periodic chugging sound that was present when the largest pumps were operating within a flow range of approximately 100 to 130 percent of BEP.

Figure 1. A number of large double-suction pumps had been installed for water supply service

A high-frequency accelerometer technique had been developed for quantifying the amount of damage that occurs in a cavitating pump. This non-intrusive and lower cost test method can be used to diagnose cavitation in pumps, valves and other fluid system components efficiently. In addition, the technique can be used to estimate the severity of the cavitation damage, determine if component modifications are essential for meeting reliability requirements and establish whether or not the cavitation noise is merely a nuisance.

The most valuable indicator of potential cavitation damage was to measure the instantaneous casing acceleration while the pump operated during the suspected cavitation conditions. This technique measured the high frequency, i.e. short time period, spikes in the vibration time signal, which resulted from the cavitation bubbles collapsing on the internal surfaces of the pump. Since the vibration spikes were considerably greater in amplitude when the bubbles collapsed on a surface as opposed to collapsing in the free stream, a quantitative assessment of the potential for cavitation damage to the pumps could be made from the test data.

Moreover, by measuring the instantaneous acceleration spikes, the pump's local structural response to the cavitation events could be measured directly. Other experimental measurement techniques were considerably less efficient. For example, the measurement of the sound spectrum provided less direct measurements because the effects must be transmitted from the pump casing to the air before they could be measured. Dynamic pressure or hydrophone measurements were comparatively difficult to perform, and were less direct because of the distance between the sensor location and the location of the bursting cavitation bubbles.

For each double suction pump, tests were conducted that consisted of measuring the pressure pulsations in the suction pipe, the sound pressure near the impeller inlet, the instantaneous casing acceleration at many surface locations and the sound pressure in the atmospheric air surrounding the pumps. Typically one method would be used to troubleshoot a problem, but the unique circumstances of the project allowed test data to be gathered via three methods.

Past experience demonstrated that peak instantaneous acceleration levels in excess of 100 g were indicative of damaging cavitation, while levels that fell below 15 g indicated that no damaging cavitation was present. In this instance, the peak acceleration levels on the double-suction pumps had been measured as high as 400 g, which was indicative of enough cavitation to erode the impeller and/or the casing of each pump.

Further post-test inspection of the pump internals revealed that cavitation erosion damage to the impeller vanes, the shroud surfaces and the nearby casing surfaces had occurred. In a similar pump investigation during which the identical technique had been used, it was determined that after 1,000 operating hours at 300 g, the vanes of a 316L steel pump impeller had lost half of their thickness.

Figure 2. Depicted simultaneously are the periodic acceleration, the suction sound pressure and the suction pressure, which had been measured while the pump was operating during a condition of cavitation surge. The data represents three different approaches to quantifying the pump cavitation damage, which are based on: a) the instantaneous casing accelerations, b) the dynamic pressure pulsations and c) the sound spectrum in the surrounding atmosphere.

Comparing the various measurement methods showed the good correlation between the suction pressure measurements, the suction sound pressure measurements (hydrophone) and the instantaneous acceleration (Figure 2). The measurement that provided the most ambiguous data was the more typically performed airborne sound pressure measurement.

Discussions with the pump OEM revealed that the pump suction had been "oversized" to meet the stringent NPSH requirements at the minimum suction head condition. The result was a pump that exhibited suction recirculation at flow rates well above the BEP flow rate. In other cases, damage often occurs below BEP. The problem was compounded as the recirculation set up an unsteady flow pattern in the pump suction, which resulted in cavitation surge. The cavitation surge not only damaged the impeller, but also shook the entire pump system.

To minimize this cavitation surge or "chugging" effect, the pump impeller design was modified, and the material was changed. The modifications increased the impeller operating life from less than two years to 15 years for each of four duplicate pumps, and consequently saved $840,000 in total repair costs.