NPSH required AT THE END OF OPERATING CURVE

[roker]

Dear all,

When evaluating a pump proposal is it realistic to check if the NPSH AVAILABLE is higher than the required by the proposed pump? if yes, what are the reasons and does it have any "written" rules like tha API or other codes.

Regards,
roker

 

[JohnGP]

It is not just realistic, but is essential to check that NPSHa is greater than NPSHr by the pump over the full operating range. There has been recent discussion on this subject in the forum - see thread407-232673.

Why do we need to ensure NPSHa is greater - basically to ensure reliable operation. Insufficient NPSHa leads to cavitation, which in turn can cause damage to impellers and casings, reduce flow, and can lead to accelerated bearing and seal failures.

Cheers,
John

 

[rmw]

Usually the first place my eyes go on a new pump curve is the NPSHr curve, and especially out at the high flow region.

If I see something I don't like about the NPSH characteristics, I don't even bother evaluating the rest of the curve. Why would you?

rmw

 

[JohnGP]

True, Ok, I take back the "over the full operating range" part of my post. The main point though is that NPSHa must be greater than NPSHr for the stated reasons.

 

[JJPellin]

The tricky parts is not just making sure that the NPSHa is greater than the NPSHr, but it is defining the conditions under which this must be true and the reference in terms of flow, product temperature, level in the vessel, etc. I believe API specifies a minimum NPSH margin of 3 feet for hydrocarbon and 5 feet for water. But I am not sure what API document that may be contained in. Our internal specifications require these margins with the vessel level at bottom tangent or bottom nozzle, whichever is lower. I would warn against using too many extreme requirements for this margin. In other words, if you require a 3 foot margin with the vessel empty, with the suction pressure at the lowest expected value and the product temperature at the highest possible value, then you may paint your self in a corner where no pump can meet the service. Or, you could be forced to buy a very high suction specific speed pump to try and get the NPSHr as low as possible. I would use what you believe to be realistic values for pressure, temperature and flow rate, but still keep the requirement that the pump should be capable of pumping the vessel down to minimum level. And for high energy pumps, a 3 foot margin is not good enough. For a high speed Sundyne, for example, I would specify that the NPSHr is defined for 1% head loss rather than 3% and raise the minimum margin up to 5 feet.

Johnny Pellin

 

[Artisi]

Johhny - good info, however you are talking about a positive suction head in this example whereas it is not clear if the OP has a positive or negative inlet condition.

 

[JohnGP]

There is a danger here that the OP will be drowned in specifics - may I suggest a Google search on key words such as "NPSH margin available required centrifugal pump" (if it is centrifugal type), to obtain an understanding of the basics.

 

[JJPellin]

Artisi,

I was referring only to Net Positive Suction Head in units of absolute pressure. This can only be positive. I understand that some industries use terms like suction head or suction lift and denote these in units of gauge pressure rather than absolute pressure. But, in my industry (refining) I am only accustomed to NPSH in absolute units.

Johnny Pellin

 

[JohnGP]

I think whatever the industry, NPSH is defined in terms of absolute units. Suction head or suction lift though is measured in terms of actual liquid head, or gauge pressure converted to units of liquid head.

 

[Artisi]

To keep it simple I like the following as being the simplist way to show the calculation for NPSHa.

For "suction lift"
NPSHa = Ha - Hvpa - Hst - Hfs

For "flooded suction"
NPSHa = Ha - Hvpa + Hst - Hfs

To this you can include your own margins, fudge factors etc.

 

[MartinSr00]

Just an observation:

Many people associate NPSH requirements with cavitation free operation. This is incorrect.

To perform an NPSHR test at a flow of X gpm, suction pressure is set at some "high" level, and the test loop control valve is set for X gpm. Slowly, the suction pressure is reduced, while monitoring the pump differential pressure. At each reduction, the control valve may need to be readjusted to maintain X gpm flow. When the pump differential begins to fall by an agreed upon amount (usually 1% or 3%), the suction pressure value in absolute units is recorded and NPSHR computed therefrom (considering suction flow velocity, fluid vapor pressure).

This differential head falloff is caused by the cavitation bubbles becoming so numerous that they block the flow passages. Cavitation is by this time quite advanced.

A margin of NPSHA over NPSHR is (1) to make sure that you have some margin over incorrect assumptions and minor upsets, and (2) to make sure that cavitation hasn't become so advanced that significant damage to pump flow passages could occur.

Cavitation damage is related to the heat of vaporization of the fluid involved. The damage is caused when the water boils locally creating a vapor bubble, and the vapor collapses as it's pressure is increased in the impeller. As the fluid changes from vapor to liquid, the heat of vaporization is released as the bubble becomes microsopic. A high speed micro-jet of fluid emmanates from the collapsing bubble. The pressure can be in the high 10's of thousands of psi. Over time, this fatigues the metal surfaces -- kind of like hitting the surfaces lightly with a center-punch.

Since the heat of vaporization is the big factor in damage, petroleum, for example, causes less damage than, say, water.

Just remember, healthy NPSH margins mean less chance of shortened life due to cavitation. Even with healthy margins, almost all pumps in routine commercial service have some incipient cavitation present. They perform thus for years without sifnificant problems.

In some cavitation tests I performed in the past, where there were viewports in the suction eye of the impeller, it took 3 times NPSHR to suppress all cavitation. Trying to do this is excessive and would burn your bank account into oblivion. Earlier posts have suggested some values. Consider those.

 

[ccfowler]

The previous posts provide excellent guidance. An additional consideration is to look at the NPSHr curve vs. the head vs. flow curve. The NPSHr curve rises for flow rates above and below the BEP flow rate. Commonly, the NPSHr curve does not cover the full range of the head vs. flow curve. This is for the very good reason that the NPSHr rises progressively more steeply for flow rates farther from the BEP flow rate, and operation more than momentarily beyond the range covered by the NPSHr curve is usually a bad idea. At the lower flow rates, NPSHa usually increases providing some mitigation to the cavitation problem. At the higher flow rates, NPSHa usually drops substantially thereby augmenting the cavitation problem.

Summing up the situation, unless there are some very compelling reasons, operation well out on the head vs. flow curve should be avoided.

 

[Artisi]

Good point ccfowler, I was just mulling over the title of this posting and realised the significance of "--- AT THE END OF OPERATING CURVE" , why would you greatly concern yourself regarding NPSHa / NPSHr at end of curve - you wouldn't under normal or even abnormal operating conditions expect the pump to operate in this region as other problems just as important as NPSH could be of greater concern.

 

[dabluffrat]

roker,

based on the tone of your question i'd guess someone bought a pump with out checking NPSHr vs NPSHa "over the full operating range..."

all above postings are good, but from an applications standpoint API-610 defines the end of the curve at 120% of BEP flow, and defines NPSH margin as "sufficient for all flows"

API-610 10th ed (5.1.10) The Purchaser should consider an appropriate margin...that is sufficient at all flows (from MCSF to max operating flow)

they way interpret this there is some wiggle room if rated flow is back on the curve and it's not practical to size a motor and NPSH all they way to 120% of BEP... you need to determine what flow is considered AT THE END OF OPERATING CURVE

Did you know that 76.4% of all statistics are made up...