Centrifugal Pump in Series or Flooded Suction Operation

Large NPSH is good. That's one less thing to worry about.

As long as NPSH (a) (NPSH available from the system) is greater than NPSH (r) (NPSH required by the pump) by a reasonable safety margin, then you are OK. You appear to have so much in hand you don't need to tell the pump vendor. If it had been very close, or NPSHa less then NPSHr, then the vendor would need to know, to see if a different pump selection would meet the duty.

I do not altogether agree. Even though the margin between the required and the available NPSH is of no consequence, the NPSH available should always be made known to the vendor/designer.

This allows them to calculate the Nss (suction specific speed) which assists in choosing/designing an impeller and diffuser (volute) combination best suited to the specific duty specified.

If you are just buying some bog standard off the shelf pump then this is probably academic, but if you wan the best solution, include NPSHa to the vendor.

Does it mean that if a pump has been selected considering NPSH(A) = 53 m, flow rate 22 m3/hr @33 mWC can be safely operated for any positive (flooded)suction condition ? And what so ever be the new suction condition NPSH(A),duty point will change accordingly.

The pump is not selected considering a specific NPSH (except for specials; see Kaisan's remarks above), but is selected only on the consideration that NPSHa exceeds NPSHr. If NPSHr is 5m, it does not matter whether NPSHa is 53m or 10m, and it will not have any influence on pump performance.

The duty point will not change when you change to flooded suction conditions, however the discharge pressure will. The 33mWC you mention is the TDH (Total Dynamic Head) that the pump will produce, irrespective (within limits) of the suction conditions. (Head is used in centrifs rather than pressure because it remains constant for a given impeller type/size/speed, whereas pressure can change according to the SG of the fluid).

Therefore, if you have the pump in series, and assuming a suction pressure of approx 5 bara, you will get a discharge pressure of approx 8 bara (7 barg, and assuming we're talking water @ SG 1.0). If you change to flooded suction, assuming a suction pressure of 1 bara, you will have a discharge pressure of approx 4 bara (3 barg). But as far as the pump is concerned, the TDH is always the same.

NPSHa will only affect pump performance if it is less than, or close to NPSHr, as this could lead to cavitation which causes loss of performance, a lot of noise, and eventual destruction of the pump.

"If you change to flooded suction, assuming a suction pressure of 1 bara, you will have a discharge pressure of approx 4 bara (3 barg)"

Another analysis of same case : Let system remain constant, i.e pump is taking suction from a reservoir where water EL. is 310 and pumpis installed at EL.265 to give water supply to filling another tank at same EL.343 @ atmoshpheric pressure.

This is the first case i.e Approx NPSH(A) is 53 m, flow is 22 m3/hr @ 33mwc and filling another tank at EL.343.0.

2nd case : Assuming change in reswervoir water level to EL.267,that will give new approx. NPSH(A) is 10 m and end user remain same.Here although pump suction is flooded because installation is remain same @ 265, pump will not be able to fill the same tank and probably run out condition.

so, pump manufacturer has very little to do with NPSH(A), rather system designer has more to do with NPSH(A).Here system designer wiill check run out point and suitability for variable suction condition from 310 to 267m water elevation.

In anoyher word NPSH(A) more related to system designing (assuming flooded suction) rather than pump manufacturing.

You are confusing NPSH with Suction Head. NPSH has always been very easily misunderstood and misused, and is very difficult to explain clearly. NPSHa only ever has any meaning when it is related to the NPSHr of the pump. NPSH involves pressure, but is more than that (it also considers the vapour pressure of the fluid), and could perhaps be defined as the ability of the 'system' to supply fluid fast enough to the pump suction to avoid cavitation. If you're interested, I'm sure wikipedia has something to say about NPSH, or check out: http://www.engineeringtoolbox.com/npsh-net-positive-suction-head-d_634.html

NPSH is entirely related to pump manufacturing/design, since that is what it is calculated for, and is only a consideration in pump selection when NPSHa and NPSHr are close, where, for example. a larger pump with a lower NPSHr might have to be selected.

When you are sizing a pump for correct flow and pressure, then Suction Head, Discharge Head and TDH are the terms that should be used, and the pump requirements come from the system design.

"be defined as the ability of the 'system' to supply fluid fast enough to the pump suction to avoid cavitation."

Good definition for the masses.

I took a job as EM at a PCB fab equipment company long ago. Mostly due to space limitations, we sometimes used cavitation to set pump flow limits. I rigged a crude accelerometer to the pump and had a portable spectrum analyzer to read the vibs vs frequency.

Worked ok. Oh, yes, the pumps were 5HP and less.

Lynlynch, did you use this method for both high and low flow limits? If so, do you have any data or can you remember the frequencies that occurred at suction condition induced (low NPSHa) cavitation and low and high flow recirculation induced cavitation, or were you just looking for change in overall levels?

I have seen so many contradictory examples of spectra for these that it would be great to have some data based on actual experience and testing.

A clever and logical way to set pump flow limits.

This is very interesting; probably for different reasons for Kaisan and myself. As a former purveyor of the finest centrifugal chemical process pumps to all and sundry, my only concern was to avoid cavitation by ensuring NPSHa exceeded NPSHr by a suitable margin. I had no interest in the nature of cavitation itself.

Having long since turned to the Dark Side (PD pumps), cavitation has usually been less of a concern. However, I have noticed some intriguing differences between centrifs and gear pumps, namely:

- Cavitation appears to be far less destructive on gear pumps.

- Gear pumps can actually get quieter with the onset of cavitation, before getting noisier.

- There are some applications where cavitation is almost inevitable. Noise could be greater, and pump life much less, but considering the alternatives (none!), these become acceptable.

My theories are as follows:

- The distance and time between bubble generation and implosion is considerably shorter for centrifs, thus gear pumps are more 'gentle'.

- Implosion in a gear pump could occur anywhere between where the gears unmesh, and where they mesh, depending on conditions. If it occurs at, or near to, where the gears mesh, could it result in a 'softer' meshing of the gears (pneumatic buffering)?

The nature of the onset of cavitation; frequency vs. amplitude vs. hydraulic flow vs. component geometry is quite interesting.

As a matter of interest, there are a number of companies and research institutions currently looking into 'constructive cavitation', for heat generation, speeding up of hydraulic and chemical processes, hydrogen generation, etc.

What do you think?

This is interesting discussion related to centrifugal pump and for time being not agree with Mr. Holzfeller on following two points :

1) The duty point will not change when you change to flooded suction conditions,

Very recently while analyzing a fluid water circuit on AFT Fathom, where suction head of pump changing drastically ( approx. 30 m, in normal & rainy season due to change of reservoir water level) and that result in changing static head between 50 m to 80 m seasonally, as per Fathom ,pump duty point or say system curve was changing and many flat type pump curve given by different vendor is not supporting software analysis and giving warning of pump run- out condition.

If I do agree with Mr. Holzfeller explanation that duty point will not change it implies that not to bother about pump curve nature (whether it is flat or stiff ) and warning given by software regarding pump run-out condition is not valid.

Generally, if system curve is not intersecting with pump curve is called run-out condition and as per engineering practice such type of pump selection is not recommended, why ?

2) confusing NPSH with Suction Head,

Yes and why not ? Agree for NPSH is more relevant to pump manufacturer rather than system designer. NPSH is calculated with suction head so why isolate both ?

Also, in reply to the original poster, we should always give NPSH(A) to pump vendor may be because static pressure on pump seal required to be accordingly.

1) I should have explained more clearly, my apologies. What I was driving at is that a change of suction pressure does not in itself change the performance of the pump. The pump curve is the pump curve, and if the duty point is 22m3/hr at 33m TDH then that is what it will be whether the suction head is 53m or 5m. I did say that the discharge pressure will change, so of course if the change is made in the same system then you will be operating at a different point on the system curve. However, I made the assumption that these are entirely different systems, based on the OPs statement that the pump was firstly to be used as a booster pump in series with another (hence high suction pressure) to boost pressure to 7bar. Flooded suction then implies to me an entirely separate system, so system curves don't come into it (also assuming the pump is suitably sized for the second application).

Now, drastically changing suction pressures in the same system is a different matter and it warrants serious consideration of pump curves and system curves. But this is where the pump vendor needs to know suction pressure, not NPSH. With standard pumps and flattish curves, the pump vendor might not be able to do anything about this with pump selection alone. This is where control valves or VFDs come into the picture.

2) NPSH is not the same thing as suction head. They are not interchangeable. As I said, NPSH also takes account of the vapour pressure of the fluid (in fact, it used to be common for NPSH curves to give the figures as "xx feet above vapour pressure"). Thus, for example, you could have two situations where the suction pressure is exactly the same, but if one has a fluid temp of 20degC and the other has, say, 60degC, the NPSHa will be entirely different because their vapour pressures are different. The first could work fine whilst the second cavitates, although suction pressure is the same. As I said, NPSH is only used for ensuring that suction conditions are sufficient to avoid cavitation, it does not tell you what the suction pressure is, so cannot be used for pump selection (nor for ensuring that seal or pump pressure ratings are not exceeded).

If NPSHr is 5m, it does not matter whether NPSHa is 53m or 10m, and it will not have any influence on pump performance.

This statement is correct, however, just to clarify for the OP; a pump designed for a low NPSH required and a pump designed for a higher NPSH required will have different performance and stability characteristics, especially as regards to operating window (stability at low flows for example). See chart from Lobanof & Ross below showing the suction velocity triangles to achieve different NPSH required.

The impeller design, including number of vanes, eye diameter, angles etc is different for pump designs for higher or lower NPSH required. What you choose from a "family" of curves is a compromise, albeit in most cases an acceptable one, as long as you know what the limitations are.

Here is another diagram showing the hydraulically stable pumping range of pumps with different NPSH required.

As an example, condensate pumps by their nature have extremely low NPSH available. They need a pump design with a very high Nss due to the low NPSHr design. They have to use large impeller eye, fewer impeller vanes, different inlet vane angles etc. This means lower NPSH required, but also lower head at BEP and lower efficiency with a steep Q-H curve. Though this would satisfy the OP's NPSH requirements it would clearly be a bad choice of pump design for the duty.

As a matter of course we always specify the NPSHa on the pump requisition sheet. However, we only insist on NPSH performance testing if the margin between required and available is below 2 meters or the pump is in a vacuum service. For all other pumps (unless there are other factors) we accept the OEM's experience based NPSHr curves.

No.

Read post #2 again. Depending on a number of fluid conditions, different pumps require different NPSH. The key word is minimum positive head.

NPSHR and NPHRA is very important in selecting a pump especially on a suction lift situation. But since you have a flooded set-up, which i presumed you mean the supply is at the level or above the pump, then npsh can be neglected. Focus on pump net discharge head and volume flow rate.

Something doesn't add up: <...53m...flooded suction...> plus <...33m...differential head...> is greater than <...7 Bar(g)...water...>.

That's a good point. I trust you haven't been pondering this daily the last 15 years? It begs a couple of questions:

1. Where is the high suction pressure coming from? It is unusual for a booster pump duty on such a small water pump. Is it a mistake?

2. Why is he talking about two pumps in series when a single pump would do the duty?