Igor Karassik (Google him!) once said that if all the technical papers written about NPSH and cavitation were laid end to end, they would not reach a conclusion. In this commentary, Ron Astall focuses on NPSH testing issues.
Testing a pump to determine Net Positive Suction Head Required (NPSHr) can be problematic, particularly if the NPSHr values are low. Read on to discover why the Test Engineer might be tearing their hair in frustration.
Background
Net Positive Suction Head (NPSH) is a key parameter in a pumping system, as it represents the margin above the liquid vapour pressure. It is a measure of how close the liquid is to vaporising (or boiling). If the liquid vaporises in the pump, it causes noisy, damaging cavitation, which must be avoided to ensure a reliable installation.
NPSHa is the available Net Positive Suction Head (NPSH) in the system. NPSHr is the NPSH required by the pump to avoid cavitation. In simple terms, we need the system NPSHa to exceed the pump NPSHr.
NPSHa is typically presented as the available NPSH at the pump inlet nozzle. However, NPSHa can be calculated at any point in a piping system, as shown in Fig. 1.
Wherever the local pressure drops below the vapour pressure, cavitation, or “flashing,” will occur. Cavitation is typically considered a pump issue, but it can also occur in piping systems, valves, and fittings. This will severely disrupt the flow.
In the example above, NPSHa drops alarmingly at the highest point in the system. In this example, if the highest point exceeded 12.66m above the system inlet, vaporisation (boiling) would occur. This example is a static case without flow. With flow in the system, friction losses and the effects of bends and fittings will create additional losses and, potentially, vapour locking at the high points.
Testing to determine Pump NPSHr
Testing is ideally performed in a closed test loop, where the loop pressure (pump inlet pressure) is progressively reduced while maintaining a constant flow rate. For each pump inlet pressure, the NPSHa is calculated from the measured suction pressure, corrected to absolute by adding atmospheric pressure, then deducting the liquid vapour pressure, and finally converting to head units for plotting.
When the value of NPSHa is reduced sufficiently to cause a measurable effect, this benchmark value is noted as the NPSHr point for that specific flow rate. Conventionally, the recommendations of the Hydraulic Institute for a three per cent head drop criterion are followed, as shown in Fig. 2 below. This method is repeated to obtain NPSHr points for flows across the pump curve, creating a complete NPSHr curve versus flow.
This approach is generally quite straightforward when demonstrating moderate and higher NPSHr values. So, when does the poor test engineer become more frustrated than a dog trying to bury a bone on a marble floor?
When the Pump NPSHr is very low
When the value of pump NPSHr to be demonstrated is low, difficulties may arise in the test loop. A typical Test loop suitable for NPSH Testing is shown in Fig. 3.
A low NPSHa value at the pump inlet indicates a low NPSHa throughout the entire test loop, extending beyond the discharge throttling valve. For a typical cold-water test, an NPSHa value of two metres or less at the pump inlet indicates that the entire loop after the throttling valve will be at a vacuum of at least -80 kPa, which is equivalent to an absolute pressure of less than 22 kPa.
In high velocity areas in the test loop, the pressure will be even lower due to velocity head effects (Bernoulli’s Theorem). This can be a particular problem at the throttling valve outlet, where the flow is constricted. Under these circumstances, vaporisation (cavitation) at the throttling valve is a major issue, with severe surging as vapour pockets form and collapse in turn, particularly during high-flow pump tests. Flow control will be challenging, and in some extreme cases, flow can be lost entirely. The associated acceleration/deceleration heads involved with flow surging will reduce the measured pump differential head and efficiency.
Another less obvious problem when the test loop is operating below atmospheric pressure is the potential for air ingress through the mechanical seals. Conventional mechanical seals are not designed to withstand vacuum conditions in the seal chamber. They can experience the opening of the seal faces or even the dislodgement of the outer face under negative pressure. This will allow air to seep into the suction side of the pump, resulting in associated disruptions to hydraulic performance.
For these reasons, it can be difficult to demonstrate low values of pump NPSHr during performance testing. Although frowned on by most testing standards, using a suction throttling valve ahead of the pump can be a method to maintain higher test loop pressures. Using a suction side throttling valve may transfer the surging problem to the suction side.
When the Pump NPSHr is extremely low
In the oil, gas, and petrochemical sectors, vertical canned pumps (Fig. 4) may be necessary to handle applications where the available Net Positive Suction Head (NPSH) is negligible or even zero.
In these circumstances, it will not be practical to demonstrate pump NPSHr in a conventional works test scenario.
Where the margin between the customer’s site NPSHa and the pump’s predicted NPSHr is significant, purchaser specifications normally waive the NPSH testing requirement based on this margin of safety. An alternative is to test the pump without the suction can by using a temporary hard-piped suction connection.
Past test experience with the same first-stage impeller hydraulics in conventional above-ground configurations will also be a valid guide.
Conclusion
NPSH testing can be time-consuming and has associated costs. If you have plenty of NPSHa in your process compared to the NPSHr of the pump, an NPSH test is probably not warranted. If NPSHa is low, NPSH testing is undoubtedly a good idea, but be aware of the difficulties and approach it with practicality.
Ron Astall
FIEAust. CPEng.
NER 210593 RPEV PE0013304, RPEQ 3119
