dependence between NPSHr and rotational speed

[adula29]

Would someone know that is some kind of dependence between NPSHr and rotational speed?
Needless to say, that NPSHr decrease with decrease of rotation speed, but are there any solutions to estimate NPSHr in lower rotation speed?

 

[GPRnD]

In typical cases the NPSHr of a centrifugal pump will vary by the ratio of the pump speeds raised to the power of 1.6 to 1.7

So:

NPSHr2 = NPSHr1 * (RPM2/RPM1)^1.7


I say "Typically" because some pumps behave a differently to this. We prefer to test a pump at different speeds before we take it to the market so that we are sure of what the relationship actually is.

You will occasionally see people using the power 2 in place of 1.6 to 1.7. This is ok if you are increasing the pump speed as it errs on the conservative side.

Using the power 2 when you are decreasing the pump speed is a recipe for cavitation damage.

 

[25362]

Another source indicates for a pump with a design flow rate of 40 m³/h, a change in speed from 3550 to 1740 rpm resulted in a drop in NPSHR from 2.5 to 1 m, while for a pump with a flow rate of 600 m³/h between 1770 and 1175 rpm, the NPSHR drop was from 4 m to 2.4 m.

Applying the suggested formulas, the resulting exponents are even more "conservative" (lower than 1.3) than those indicated by bradshsi when reducing speeds.

Therefore, assuming all sources are correct, the conclusion should be...

Not being an expert on centrifugal pumps, just a user, my question is: for the same pump why should the exponents differ when increasing or decreasing the rotating speeds ?

 

[GPRnD]

The exponents should differ when you want to err on the side of caution.

Given the general uncertainty in exactly how NPSHr scales with speed on any given pump, it makes sense to use a smaller coefficient when reducing speed and a larger one when increasing speed.

At first glance this padding might seem excessive, but consider that many test standards will not allow a positive tolerance of NPSHr on the test stand.

Hence the pump manufacturer wants to build in some padding since the consequences of not meeting the promised NPSHr are quite expensive.

Gulich has an interesting discussion on NPSH exponents in section 6.2.3 of his book "Centrifugal Pumps". I don't personally agree with the formula he proposes, but he at least attempts to document some reduced speed tests and the exponents produced.

On average the exponent found on test was 1.6, although the scatter was from 1.3 to 1.8