Can Pumps Operate Below Shutoff Head? 1

[vpl]

I've got a situation where I'm reviewing some calculations and test results on a centrifugal multi-stage pump. The vendor gives a "best efficiency" point of 800 gallons per minute (gpm)and a "shutoff head" value of 70 gpm. The system has a minimum flow line that has both an upper and lower limit (too little flow hurts the pump, too much means the system can't remove the heat that it needs to.) This line has two check valves, a normally open isolation valve, an orifice and a permanently installed flow meter in it. If the flow exceeds the upper band, then the procedurally stated expectation is that the orifice be replaced.

What I'm worrried about is that the calculations sized this minimum flow line to have a lower value of 50 gpm and an upper one of 72 gpm. And, under certain situations, the pump might have to operate on minimum flow for extended times (over 100 hours). A test was done recently to confirm the design of the minimum flow line. The test was run for four hours and fluid temperature rise across the pump and vibration data were taken. This test resulted in a flow across the orifice of 70 gpm, temperatures that went down at first, then started rising, and vibration data that had a "zig-zag" pattern, but was well below the acceptance criteria. This was a one time test meant to show that the pump could safely operate for the extended time without degradation. Normally the pump is tested at a "full flow" condition and the flow through the minimum flow line is recorded but the pump has the discharge valve open so its not being challenged.

What I'm (mainly) having problems with is the lower acceptance criteria of 50 gpm. I can come up with scenarios where I can get partial flow blockages due to water quality such that the flow through the line could drop below the current tested value. The normal test won't verify that the pump can survive on this lower flow but the procedure will be marked off as "satisfactory" due to it being in the acceptance band. My position, right now, is that a pump can't operate at 20 gpm below its shutoff head (even for short periods of time) and either we need to devise a test to show that it can or we need to CHANGE THE ACCEPTANCE CRITERIA. Management does not want to change the acceptance criteria because that doesn't give them any operating margin.

So with the above convoluted explanation, I'll get to my title question - does anyone think a centrifugal pump can operate at well below its shutoff heat - for an extended period of time? Am I out to lunch here?

Patricia Lougheed

Please see FAQ731-376 for tips on how to make the best use of the Eng-Tips Forums.

 

[quark]

When you say shut off head, you should have zero flow. This is actually what a pressure gauge placed between the pump and the isolation valve reads when the isolation valve is closed.

Generally pump curve is not graphically prolonged to cut the vertical axis. It stops at a certain distance from the vertical axis and the point where it stops reads your safe minimum flow. As the curve becomes somewhat flat at this region even for steep curve pumps, it is difficult to obtain a graphical solution. However, it is not good to run a pump (particularly a multistage pump) below the safe minimum flow and avoid at all times to run it at shutoff condition.

I saw one single stage chilled water pump(50m head) becoming a boiler within 30 minutes of operation at shut off condition.

When you say "too little flow hurts the pump, too much means the system can't remove the heat that it needs to." I understood the first part. What do you mean by the second part?

I hope you enjoyed your lunch

Regards,

 

[vpl]

quark

The upper limit is there because the pump has to supply a minimum amount of flow so that process conditions are met. In this case, the minimum flow line is always open (there's no control valve). So if too much flow goes through the orifice then the process is starved. This pump is at a nuclear plant and does not normally operate. However, when it HAS to operate, it has to provide a certain flowrate. However, there are times when the safety actuation system starts the pump, but the pressure is too high for it to inject. Worse case for that scenario is that it has to operate for 4 or 5 days on only minimum flow. 70 gpm is where the pump manufacturer stopped the pump curve and the curve is pretty much a straight horizontal line at that point. I'm being asked to accept that the pump will continue to operate successfully at an even lower flow rate based on test data done right at the ragged edge of the pump curve. I'm outside my comfort zone, but have to justify why it isn't ok.

And since I just got back from lunch when I read your message, yeah, I did have a good lunch. Actually took it today for a change!



Patricia Lougheed

Please see FAQ731-376 for tips on how to make the best use of the Eng-Tips Forums.

 

[PUMPDESIGNER]

Hello Patricia,
One thing that I just love about certain centrifugal pumps is that they are so very flexible and gentle, without harsh reactions to things like you are doing and often without requiring expensive equipment to help and protect them.

I don't have a lot of time today, so I apologize for the non-systematic presentation of my ideas, just sort of random.

The lower the specific speed, the better behaved the pumps will be at low flow rates. Low Ns pumps at very low flows down to shut off basically have only disc friction for the most part. Higher Ns pumps have loads of turbulence at low to no flow.

Diffusers help a lot by maintaining a hydraulically balanced environment around the impeller, preventing unbalance. Velocity energy is changed to pressure energy, fluid movement becomes diffusive, slow and smooth.

If the heat from friction is removed, and vibration is no problem, then you should be fine except for one thing, how vulnerable is the pump going to be to recirculation cavitation, which can eat up the impeller and wear ring. I do not have enough time or information from you to look at that, and the manufacturer may be the correct source for that info anyway, this is highly specific to each pump brand. One quick thing is to consider material resistance to cavitation. Iron is the worst, bronze/brass is a little better, stainless much better, and on up to the exotics.

I did not have time to try to understand everything you were saying and implying, so what I missed I may come back later and read.

PUMPDESIGNER

 

[JohnGP]

Patricia,

One thing that raised alarm bells in my mind was that with the low flow test at 70gpm, the temperatures "went down at first, then started rising". Test duration 4 hours, possible operating condition 100 hours plus, design flow down to 50gpm..........hmmmmm. I guess what I am pondering is whether the test proved that heat dissipation would be acceptable during extended operation - it certainly didn't for minimum acceptable flow.

You may have considered and rejected already, but to get 50gpm in the minimum flow line, can you obtain enough regulation with your normally open isolation valve, sufficient to reduce flow in the line to 50gpm for an extended test? I believe test needs to proceed until temperature has plateaued.

John

 

[ccfowler]

Patricia,

The test where the temperature first dropped and the rose suggests that the source flow passed (and perhaps passes) through piping in a lower temperature region. Since the test flow is so far below the design flow rate for the pipe, it could take quite a while to warm up. Even if everything is indoors, there could be seasonal issues for the temperature of the piping. Seasonal effects could also influence heat dissipation effectiveness of the piping system during the long periods of low flow operation.

I would not be surprised to learn that, for the test, the pump outlet temperature was monitored but the inlet temperature was simply assumed to be either stable or the same as some distant monitored point.

I would expect the amount of power going to heat the fluid in the pump would be roughly equal (for estimation purposes) at both 50 gpm and 70 gpm, so the temperature rise should be significantly greater as the actual net flow rate is reduced. As the pump's condition deteriorates from the abuse of the low flow rate operation, the temperature rise through the pump will probably increase.

 

[vgarzani]

Your pump sounds like the 900 hp big brother of the 600 hp variety that we have. Q: Is the miniflow routed directly back to the pump suction? Also, how sensitive do you think vibration is as an indicator of poor hydraulic conditions in this pump? I'm not really sure, even though our pumps are fairly well instrumented.

 

[vpl]

vgarzani

Actually in this case, the pump recirculation line goes back to an outside tank. It doesn't reconnect with the pump other than in the most general sense that the pump suction is from the tank.

And in regard to vibration, I don't know - we had one case where we saw routine vibration monitoring readings on a (different) pump and then, without warning, the pump shaft broke; so there was something going on that the vibration monitoring didn't pick up. But I'm not a vibration expert by any stretch of the imagination..

Patricia Lougheed

Please see FAQ731-376 for tips on how to make the best use of the Eng-Tips Forums.

 

[tstead]

Patricia,

The pump vendor will furnish the minimum flow. There are a few things that go into this determination. One is limiting the heat rise of the fluid, another is maintaining seal integrity, and another is preventing pump damage (e.g. shaft damage). The minimum flow is essentially the highest of these "minimums."

Problem is, you'd have an easier time pulling teeth then getting this basis from the vendor. All they'll give you is the minimum flow without regard to which of the "minimums" is the limiting one.

If the pump is intended to maintain it's mechanical seal integrity over the 100 hours then you should ask to see the shaft deflection calcs at both 50 and 70 gpm. Both should be less than 2 mils at the seal face, which is the limiting design criteria for most mechanical seals.

What it all boils down to is that the pump vendor sets the minimum flow to protect the pump. In many cases, the flow should never drop below 50% of this value. Ever. Brief periods of operation below this min flow should not damage a pump (pumps are not delicate machines). So, what is considered brief? I've seen many people define this as a few hours per month.

In any event 100 hours certainly does not seem to imply "brief."

The shaft deflection WILL be higher at 50 gpm as opposed to 70 gpm. Did they take this into account when it was determined that 50 gpm is o.k.?

Another factor is the shape of the curve in this region. Is it sloping back down? Operating that pump in the region you state MUST (I cannot emphasize it enough) have a continuosuly rising head curve towards shutoff (zero flow). If it is not, and the curve starts to slope back down, then operating the pump in that region can have devastating effects on it. Basically, there would be two flows at which the TDH is the same and the pump can hunt between the two. Not good.

In short, the min flow is there for a reason. The manufacturer determined it based on engineering analysis, tests, experience, or a combination of all three. If the plant changed the minimum flow from 70 to 50 gpm, then they should have performed an analysis commensurate with the original manufacturer's determination. (Crit. III)

 

[25362]

One note on drooping curves.
Although it is true that a drooping curve, as tstead rightly says, may cause flow instabilities, it appears that as long as the shut-off head of the pump is higher than the system's head (including friction) at discharge, the system's head curves would cut the pump's characteristic only at one point without hunting and fluctuations.

 

[abeltio]

i have seen a similar problem before in the refrigeration circuit of a print.
the contractor that designed and built pumping system (with the recirc line back to the suction tank - long recirc) messed it up completely...

with the recirculation line open there was not enough flow to the refrigeration circuit... (pump just barely big enough)
with the main service line closed we did not have enough flow to keep the pump cool...

instead of embarking in a costly and lenghty process to get a vendor recovery... because what we needed was the system up and running what we did was:
1. add a recirc line with an orifice from disch to suction of the pump (short recirc)
2. add fluid operated valves (NO) in both recirc lines (the long and the short one)with the tie-in after the discharge shut-off valve.

In normal operation the discharge pressure of the pump would keep the recirc lines closed.
During running stand-by the recirc lines would open and the total flow was big enough to keep the pump cool (calculated the orifices so that the short recirc was 50% of the required min flow) the long recirc was providing the "cool fluid" to keep the temp down, the short recirc was used to contribute the required flow.

After a few trials (we prepared different size orifices for both lines) we got a combination that worked just fine.

This was a very crude and "field type" solution... but fixed the problem

After typing for about 15min i realize that never answered your original question... anyway... your problem sounded really familiar... in any case:
happy 2004




saludos.
a.

 

[vanstoja]

In nuclear powerplant pumping systems (as yours seems to be)with multiple pumps operating in parallel (hence your check valves), it is customary design practice to provide both isolation valves and check valves with seat bypass orifices in order to maintain pumped fluid temperatures in temporarily isolated loops close to the that in the operating loops so as to avoid thermal shock damage to piping and/or components during startup of pumps in isolated loops. This usually causes some flow to circulate through the bypass orifices when the valves are fully seated. Hence one has a non-zero "pump (isolation valve) shutoff flow". The bypass orifice is usually sized to provide enough flow to maintain idle loop temperature within some non-damaging margin to operating loop flow, such as 50F, for example.
If this minimum flowrate is specified to the pump designer (not necessarily the pump supplier), then he may advertise it as a "shutoff flow" limit which may or may not have anything to do with requirements for safe operation of the pump, particularly if it is higher than the calculated minimum flow to keep pumped fluid temperature rise within safe limits. The other parameter of concern at very low flowrates is the stability of radial and/or axial hydraulic thrust loading on the pump impeller which can only be determined by testing at least one pump of a particular design down to flows approaching zero. This is mostly a concern for pumps/drivemotors employing fluid film rather than rolling element bearings.
Since the isolation valve seat bypass orifices may become increasing constricted with operating time (say by high velocity flow-induced electrokinetic corrosion deposit buildup at the orifice ID), then a range of non-zero bypass flowrates may need to be specified by the user. Unless the pump designer can tell you that temperature rise, cavitation or radial/axial hyraulic thrust instability actually dictate his indicated non-zero "shutoff flow", then I would not be concerned about operating the pump anywhere within a deliberately selected and specified shutoff flow range that is related to temperature control seat bypass orifices in isolation/check valves.