re: Acceptable NPSH margin 5

[TurbineBlade]

Hi,

I would like to ask what is an acceptable NPSH margin in the industry. I have a pump in an open system that is for accident mitigation purpose. Therefore, should I specify a higher NPSH margin like 50% above the NPSH required.

What will happen if the pump runs below the NPSH required? Will the pump still deliver the flow but at a much lower flow rate?

thanks

Turbineblade

 

[MacMech]

Hi TurbineBlade,

NPSH available should be at least 10% greater than NPSH required...usually I use a range of
NPSHa=(1.15 to 1.35)*NPSHAr. The reason to use greater NPSHa from NPSHr is to avoid incipient cavitation.

Regards

 

[TurbineBlade]

Hi,

Thanks.

This pump is required to work for at least 3 months following an accident. In the first day the water temperature goes to 100C, and its vapour pressure increased substantially. It is this period where the NPSH margin is the lowest.

For that 1 day period, I wonder if a pump can survive even if there is no enough NPSH or during the cavitation? After the 1 day, can a cavitated pump still delivery some flow, say 50% of the design flow?

 

[MacMech]

Hi,

my suggestion is not to play your luck...provide in any case the appropriate NSPSHa to avoid any incovenient situation..even for 1 day.

Regards

 

[Artisi]

If the water is at 100C then the water source needs to be above the pump centreline - no ifs, buts, or maybes.

To offer you better advice more detail regarding the installation, pump type/configuration, flow and head etc would be helpful.

 

[djack77494]

I'd really recommend more specifics. You are refering to your pump as if it operates at a single point; in fact, you can reduce the throughput below the design value if there is less than expected NPSHa. Also, pump sensitivity to the NPSH margin can vary significantly. As Artisi points out, don't fool with water near its boiling point. It is EXTREMELY unforgiving. A subcooled hydrocarbon will behave much better. There are many variables, so it's hard to make generalizations.
Doug

 

[rmw]

Don't believe your pump vendor's NPSHr curve(s). They tend to test the pump under laboratory conditions and publish curves that are overly optimistic. Allow a sufficient margin above the maximum point on the curve where you expect the pump to ever operate. Lots of pumps are picked for optimum conditions well back on the curve and then when they run out on the curve they destroy themselves when they don't have the NPSH required.

rmw

 

[ccfowler]

TurbineBlade,

Your application is a very severe duty with potentially spectacularly unpleasant consequences from a failure. If you check around a bit more, you will find that NPSHa vs. NPSHr is a very troublesome issue, and based on some very unpleasant experiences, failure to provide really adequate NPSHa can have painful consequences that can be nearly impossible to mitigate.

At the risk of repeating myself (and perhaps ranting a bit), the published NPSHr curve for a pump is usually more a bad joke than useful data unless the implications are fully understood. I personally consider the term NPSHr to be inherently misleading since, depending on the standard being applied, it represents conditions where the pump's performance has already been degraded by either 1% or 3% due to cavitation. The true cavitation-free NPSHr is likely to be somewhere in the range of from 4 to more than 20 times the published NPSHr. There has been considerable discussion of this, and the science of this matter is still evolving.

Pumps can function with less than cavitation free NPSHa, and indeed, most pumps do operate in this mode. The simple reality is that these pumps will experience shortened service life, higher maintenance costs, etc. This is not necessarily all bad since in many cases, physical and economic considerations may not justify providing cavitation free NPSHa. Unfortunately, the decision to sacrifice potential pump life is not usually made in a deliberative fashion, instead, it comes as an unpleasant reality when failures or seemingly excessive wear is discovered after a period of service. At this point, it is usually determined that correcting the problem is not economically or physically practical, and less than ideal operating and maintenance costs just become "the new reality."

For an application such as yours where "failure is not an option," it is important to truly undertand the actual needs of the pumps under consideration under the full range of reasonably expected operating conditions. Simply applying some nominal margin in the NPSHa vs. NPSHr will almost certainly amount to an invitation to disaster.

Your needs may be much better served by entering into forthright discussions with the potential pump manufacturers including providing them with complete information on the full range of operating conditions expected for the pump (flow, temperature, head) to determine both the proper pump selection and the necessary setting and piping.

 

[Artisi]

ccfowler

From my point I wouldn't have too much to disagree with in our analysis, however, I think the statement regarding NPSHr being somewhere in the range of 4 to 20 times the published curve is a bit over the top as a general statement. Sure you need to clearly understand what you are doing, know the equipment selected for the application and get very warm and cuddly with the manufacturer, but for standard mass produced pumps of which most pump applications consist of, the guaranteed curves that most manufacturers can supply are usually sufficient provided you as the end user give the correct information on which to base any pump selection.

 

[Artisi]

ccfowler
please read .... in your analysis,

 

[ccfowler]

Hi Artisi,

Your comments are most welcome, and I believe that we are probably more in agreement than most would think. Your comment on the need for true flooded-suction conditions is particularly important and prudent.

In thankful response to your comments, I think that I should provide additional clarification on my thoughts and comments.

There was a time when I would have agreed with you on the apparent absurdity of the large NPSHa margins that I have recommended to be considered, but my experiences with several major, critical pumps have lead me to think that margins of such proportions may well be wise. Indeed, I do agree with you to the extent that it is often practical and prudent to install less critical, cost sensitive pumps with less (sometimes much less) than optimum NPSHa. For an application as critical as the suject of this discussion, knowingly installing a pump with anything but very generous NPSHa margins would be beyond all possible reason.

In any case, compromised durability or performance should be a should be a considered choice and not an unpleasant surprise. After all, engineering work always involves optimizing compromises to achieve suitable results.

The experiences to which I refer involved pumps that had undergone individual witnessed performance tests at full rated speed and with numerous test data points over the entire flow range. In all cases, these pumps were installed and operated with NPSHa in the range from about 1.5 to over 3 times the demonstrated 3% NPSHr, and all suffered from substantial cavitation that resulted in excessive wear and correspondingly elevated maintenance and repair costs. The pumps all did perform adequately, and the observed operating power consumption and performance characteristics were in reasonable agreement with the test stand performance data. In all cases, the costs and practical physical considerations of the installations precluded doing anything to substantially improve the NPSHa. The use of some hard surfacing materials was about the limit of practical efforts to mitigate the effects of the cavitation. Once their rates of deterioration were recognized, the pumps could always be depended upon to run reliably between scheduled major repairs, but the service life between of these major repairs was always far less than it should have been had the NPSHa been truly adequate. In the contemporary context of the original system design and pump selection for these pumps, there was nothing remotely irresponsible about either the selection or application of the pumps, but the results were far from ideal. (One could, perhaps, observe that innocent ignorance does not always result in bliss.)

When I was in engineering school and early in my career, I never once encountered a recommendation by anyone that it would really be a good idea to provide a substantial NPSHa vs. NPSHr margin. Professors, older engineers, and pump manufacturers' representatives with whom I dealt "way back then" always insisted that NPSHa = NPSHr was entirely sufficient. Somewhat similarly, there was a common presumption that centrifugal pumps were "variable flow" devices that could be casually operated well away from BEP without any significant concern presuming that the drive motor and control valve were adequate to their tasks. Wasted energy and installation costs were matters of economic choice for determining whether multiple pumps should be used to cover especially wide flow ranges.

Experience and study over many years has revealed to me the absurdity of the presumptions mentioned in the above paragraph. Pumps are fascinating and potentially rugged machines, but the older I get, the more I recognize the importance of suitable pump selection and application within the context of the specific application. The current fad of seemingly using VFD's to compensate for poor, ill-informed, or ill-defined pump application has served to further emphasize to me the necessity of prudent pump selection and application.

I have often thought that it may be wise for the pump industry to adopt a standard notation for NPSHr in a form such as NPSHr(1%) or NPSHr(3%) to clearly indicate that the NPSHr values represent conditions where the pump's performance is already compromised by 1% or 3% due to cavitation. This would help to serve as a warning to non-engineers and non-pump-specialist engineers that no "warm, fuzzy comfort" should be attached to the NPSHr values and that NPSHa really does need to be substantially greater than NPSHr.

 

[BigInch]

I seem to recall that cavitation process can begin in many applications at 20% above NPSHr curve values. I can't find that article now, but while looking for it I rediscovered this text, which discusses in detail many factors of bubble cavitation.

Download "Cavitation and Bubble Dynamics" here,

http://caltechbook.library.caltech.edu/1/04/content.htm

http://virtualpipeline.spaces.msn.com

"What gets us into trouble is not what we don't know, its what we know for sure" - Mark Twain

 

[TenPenny]

Remember that under HI rules, the NPSH shown on pump curves represents the NPSHr for a 3% head drop due to cavitation. In other words, when NPSHa = NPSHcurve, you are already experiencing cavitation resulting in a 3% head loss.

It wouldn't surprise me that true cavitation free operation would be at 20% higher NPSH levels.

 

[Artisi]

Once again I would say that in general we would be in agreement with most aspects of your posting - however, I still have a bit of a problem with NPSHr/a in the context presented and raises the question, are we talking about what is NPSHr really is when it comes to the real world application of pumps.
For me, reported NPSHr can only be considered reliable where flow into the impeller is ideal, this however has to be viewed in context with what your goal is as you cannot always get the required duty point to be right on BEP or at the minimum NPSHr, unless it is an engineered pump for the application and then other constraints enter the discussion in real world applications to cause problems such as re-rotation, air entrainment, vortices etc .
Your profile would indicate extensive experience with large CW and ACW many of which would have been engineered horizontal split case units or pit mounted mixed and axial flow vertical units many of which would have been scaled up from models or stepped up or down from known designs. In my own experience with large pumps many of the pump curves supplied following careful testing at various flows indicate what I call the "apparent" NPSHr curve, this is a parabola rising each side of BEP or the minimum NPSHr point and superimposed over the NPSHr curve to give an inlet pressure required to ensure inlet free problems.
Of course this testing can only be undertaken with a test facility capable of running these large pumps at their design speed or very close to it while being able to vary flow rates and being able to increase the inlet pressure at each test point. Without a test rig of this size / configuration it becomes a bit of guess work which in most cases ends up with an NPSHr curve nicely rising from a low value at low flow to a higher value at the high flow end of the curve, this is probably a true NPSHr requirement but does not account for extra head on the inlet side to ensure the flow gets onto the impeller blade which is now not ideal in terms of entry once off BEP. .
To me the standard nicely rising NPSHr curves are in a sense fictitious and the cause of many of the so called “caviation” problems in the large and high energy pumps resulting from the mis-match of flow onto the impeller blades - I don’t consider this an NPSHr problem as NPSHr in the true sense is probably being met - but rather a problem of insufficient head at the inlet to ensure that damage is not being done to the impeller due to mis-matched flow.
I guess from what I am saying is, that if you are operating away from BEP or the minimum NPSHr point and you do not have any hard data as to what the inlet pressure needs to be, then you need to increase the margin – by how much who knows?
This fits in with your thoughts that NPSHr curves can lead to problems for those who do not fully understand centrifugal pumps and rely only on that has been presented on a pump curve, but I still think 20 times is excessive.
I have chosen to ignore the 1- 3% already in the NPSHr consideration as it only confuses the point I’m hopefully trying to make.
Just to close, luckily in millions and millions of applications throughout the world (many of which have been my selection) the published NPSHr curve suffices and most pumps run ok in spite of other than ideal inlet conditions.

 

[Artisi]

My previous post should be addressed - ccfowler -

 

[ccfowler]

Artisi,

Your comments are very much on the mark. All of the troubled pumps spent much of their operating time well away from the BEP flow rate. Most commonly, the flow was well short of BEP.

 

[d23]

All,

I don’t want to be the devil’s advocate, but…..

The first question should be “what is the flow rate” and type pump used??????????

The Hydraulic Institute standard is 3% head-flow loss due to cavitation. 3% cavitation at 1 GPM would cause substantially less problems than 3% at 1000 GPM. At very high flow rates the Hydraulic Institute standard test of 3% head-flow loss usually means catastrophic cavitation problems.

You should also consider the type impeller. If there is a “low NPSH inducer” or axial flow impeller involved the NPSHr curve will look more like a U than a smooth curve. The flow rate then becomes a problem if you are above or below BEP.

If the pump is operating very far outside the operating range re-circulation could be more of a problem than low NPSHr or intake pressure of the pump.

What is the application data?

 

[TenPenny]

Also note that HI standards call for deaerated water for use in the tests. Any entrained gases can cause significant differences in what actually happens at the eye of the impeller.

 

[TurbineBlade]

Hi all,

Thanks for the discussions.

The pump capacity is 167 L/s. It is a horizontal centrifugal pump. The pump draws water from a sump located in a high elevation. The NPSHr curve from the manufacture show when the flow rate is 15% above the BEP, the NPSHr starts to increase substantially.

Thanks

 

[Artisi]

Turbine Blade

" The pump draws water from a sump located in a high elevation."
Does this mean the supply is above the pump or is the installation at a high elevation in relation to sea level?

Irrespective of the above, if the water goes to 100C then you have a problem during this time of operation and without all the necesary information to make an informed judgement you need to give more detail so that we can evaluate this properly.