3 Condensate pumps not pumping equally

[terminator4]

We have 3 condensate pumps that pump directly out of a condensor into the deareator tank. I have tested the 3 motors and pumps recently and all are working within 10 to 15% of each other. They have all been rebuilt within the last year. They are sitting next to each other at a distance of 6 ft centerline to centerline. All three discharge into the same discharge pipe. If we run any 2 combinations of pumps they appear to be pumping the same. I am looking at the amps that the motors are drawing and they are the same. When we turn on all three, the end pump ( call it the east pump) does very little work (amps drop to 25% of rated work amps. The middle pump drops amps slightly and the west pump ends up exceeding the motor rated amps. To run all three pumps the operators throttle the discharge valve on the west pump until the amps fall into the normal range. I have looked to see if maybe the condensor was configured in a way that might cause one pump to work harder, but I can find nothing that makes any sense. Has anyone dealt with this kind of problem before? Any help would be greatly appreciated.

 

[SeanB]

Have you considered that your system could be hydraulically limited to combined flow from only two pumps? When you operate the third pump perhaps the increased flowrate creates excessive backpressure which in turn then backs out the flow from one of the other pumps?

 

[terminator4]

Yes I did think of that, but after investigating, the system was designed to carry more that what the combined total of all three pumps could deliver.

 

[BigInch]

What's the control system?

Offhand, this doesn't sound like a hydraulically balanced configuration maybe aggrevated by pumps with relatively flat curves.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[katmar]

If the west pump is drawing excess current then it must be pumping more than expected, and unless you have a very unusual pump curve a centrifugal pump can only pump more if the differential pressure is less. So the question becomes "what can cause a decrease in the differential head across the west pump?"

I can imagine only two causes. Either the east pump is running backwards, or the combined delivery from the east and middle pumps is somehow causing a venturi effect as it flows past the branch from the west pump delivery.

Do you have pressure gauges on the pump deliveries and on the combined discharge manifold? Watching these gauges as you run the pumps individually, then in pairs and then all three together might give you some clues. Do the pumps have NRV's in the delivery piping?

Katmar Software
Engineering & Risk Analysis Software

http://katmarsoftware.com

 

[rzrbk]

Not sure what would cause what you are describing.
Two guesses are:
-Check the suction of the two low amp pumps for cavitation. If NPSHa is borderline, flow surge when all pumps are switched on could cause mid and East pumps to lose suction and not recover. This would leave West pump to take up the slack.
-Are the pump curves continuously rising, or do they fall as flow approaches zero.

 

[ccfowler]

terminator4,

Can you describe your piping system arrangement (both suction and discharge) more completely? It seems most likely that the piping configuration is a major controlling factor. Seemingly minor issues can be very important. I would want to pay very close attention to the arrangement of manifold branches, elbows, pipe sizes, size transitions, tees, etc.

As mentioned above, paralleling pumps with relatively flat head vs. flow curves is inviting lots of trouble. Unless you are dealing with unusually carefully matched pumps, operation within about 10% - 15% of each other is not unusual. How are you measuring the balancing of the flows from the pumps? If you are relying on motor amps, you would do well to find a more accurate method. (Site and cost constraints will probably dictate the choices available.)

Motor amp draw is a tolerable indicator, but it is not nearly as good a basis as actual true power measurement. It may be helpful to monitor the condensate temperature leaving each pump with very sensitive sensors. You may find significant heating from churning at the unloaded pump.

From what you have described, it seems likely that there may be suction flow problems at the unloaded pump.

It would be helpful to know more about your pumps, too.

Since the problems appear to be associated with higher flow rates, it seems likely that there may be an effect from variations in deaerator pressure and condenser temperature as the flow (or system load) varies. This couldf shift the operation of the pumps to a less favorable region of their head vs. flow curve while making NPSHa vs. NPSHr less favorable.

 

[electricpete]

Even though you say you tested them ... how did you test them exactly? Within 15% what under what conditions?

It sure sounds like the pumps do not have equal performance. Perhaps a different number of stages, different types or trim of impellers, or severe wear ring wear on the weak pump.

Possibly even the weak pump is flowing backwards when in parallel with the strongest pump.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[terminator4]

Thank you all for your valuable input. Let me see if I can answer some of the questions that were posted above. The pumps are all identical 3 stage vertical centrifugal type pumps. They each have a capacity of 1050 GPM, 555 ft total head, 74% eff. The pump curve is relatively flat which just exacerbates the problem. The pumps were all rebuilt within the last year and had new impellers, wear rings etc. When they were placed back in service, I ran them until the amps started to redline and then was able to determine what the flow was. I know using motor amps is not the best, but I am unable to place any flow measuring devices in the piping due to the unit being on line and will not come off for at least a year.
As far as the layout, all three pumps get their suction out of the center of the condenser water box ( the center pump is in line with the centerline of the waterbox and the other pumps are 6 ft on center to it). They all pump into a header pipe with the closed end next to the east pump. 8 ft after the west pump the header makes a 90 degree turn up and starts it's ascent back to the boiler. I had the mechanics inspect all of the inlet strainers, expansion joints, discharge valves, etc. when the unit was off last time. They found nothing of real significance.

 

[BigInch]

You should be taking flow from a symetrically loaded header, meaning header outlet from the at the center, not the end. This is especially important for flat curve pumps.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[BeerIsGood33]

I was thinking the exact same thing as BigInch.

terminator4 you said the end pump was doing the least work and the first pump was doing the most.

The first pump has plenty of liquid to pump, the most in fact, then the second pump has less but more than the last pump, which is sand bagging it.

The inlet manifold has to be symmetrical to get even flow.

 

[BeerIsGood33]

Is there a way to edit post on this site?

I meant to say "The first pump has plenty of liquid to pump, the most in fact, then the second pump has less but more than the last pump, which is sand bagging it because it has no liquid to pump.

 

[BigInch]

Multiple pumps can maintain equal flows in a parallel configuration if it is hydraulic balance AND they can each individually readjust to any slight change in their inlet and discharge heads with a corresponding change in their individual flow rates, or visa versa. Pumps can readjust themselves hydraulically only by following their curves. Steep curve pumps will get a large change in head with a small change in flowrate or v-a-v, so flow is quickly forced back to the initial value. The other pumps can help, since any deviations in the first pump's flow is initially forced to the other pumps, so all act to keep flows and heads nicely balanced.

A flat curve pump system will get a very tiny change in head for a large change in flow, thus can experience large deviations from an initial upset of one variable without having much of an effect on the other. The other pumps also get large deviations before they can help redistribute flows and equalize heads too, which only excerbates any initial upsetting condition, so all pumps tend to diverge and each one will run-away along the curves as far as they can go. Upset conditions come and go in some form or another all the time, but a header that is not hydraulically balanced assures that upsetting conditions are continuously present.

The operating point for pump & pipe system is the intersection of both the pump H-Q curve (or the combined curve of all pumps) and the pipe H-Q curve. The intersection point is hard to find both for our eyes and Mother Nature's eyes when those curves are only flat parallel lines. Steeper pump and pipe system curves add more hydraulic stability to the system.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[katmar]

Hi BigInch, I agree that it is difficult to get flat-curved pumps to run perfectly matched. But I struggle to envisage a situation where two pumps are running acceptably matched and then bringing in a third pump causes one of the original two to increase its flow to the point where it exceeds its rated current. By my thinking, bringing the third pump online can only increase the system head, and this extra head should cause the original two to back off slightly, even if not perfectly equally.

That was why I said earlier that the only way I can see the flow on the west pump increasing is if its head has decreased and I suggested that maybe the flow from the other two pumps was causing a bit of a venturi effect where the west pump discharge joined the main header. Terminator4 has confirmed that the configuration is such that the flow from the east and mid pumps does flow past the west pump's tie-in point, so it is at least theoretically possible. If the pump curves are flat, the pressure at the west pump discharge only has to be reduced slightly to cause it to go into an over-capacity situation. Do you think this is feasible? The operators' remedy of throttling back on the discharge valve to increase the head the pump "sees" seems to validate this theory.

Terminator4 - is there any way you can let us see a photograph of the discharge header? And perhaps give an indication of the pipe sizes in the discharge header and branches. Do you have pressure gauges on the pump discharges and on the header?

regards
Harvey

 

[JJPellin]

I have a set of pumps that sound similar to yours. We have two events with this set of pump when one was not pumping well in parallel. We have three vertical turbine pumps pumping off the hot well of a surface condenser. The pumps are four stage vertical turbine pumps in cans. We cannot measure the flow from each pump or the total flow. The suction and discharge are set up as symmetrical manifolds. The combined flow from all pumps is controlled with a control valve to hold constant level in the hot well. Even though we cannot measure flow, we know the rate of steam flow coming into the condenser. So, with a little math, we can calculate the total flow of water off the bottom.

The first time we had a problem getting balanced flow from the pumps was right after an overhaul. We ended up finding a leak between the pump and the can. Since it is under vacuum, it was leaking air into the pump suction at a rate sufficient to keep that pump from pumping.

The second time we had a problem was shortly after a turnaround on the condenser. We ended up finding a piece of gasket that the pipe fitters had dropped into the condenser lodged in the pump suction line.

The main difference between our pumps and yours is the size. Ours are only rated for about 200 gpm each at much lower head. We only pump the water to a DA and have another set of pumps that pump it back to the boiler. Your pumps must pump all the way to the boiler.

Rather than focusing on the piping configuration, I would suggest that you first rule out all of the usual suspects. Verify that all pumps are identical in terms of number of stages, impeller diameters and shaft rotation. If your system is under vacuum, check for air leaks. Depending on the configuration of your pumps, even a leaky mechanical seal can be a source of air. If you can, calculate the total flow based on steam load and verify that the flow is adequate to satisfy the minimum flow requirements of three pumps running.

I believe that minimum flow is your problem. If two pumps in any combination pump well, then most of the things I just suggested can’t be the problem. But if the total flow is too low for three pumps to run above minimum flow, then it matches your description. With three pumps in parallel and no measurement of flow, all three could be running below minimum. The two weaker pumps are driven back to shut-off where they gas pump and loose flow completely. The single pump that is still pumping takes the full load and runs out to high amps.

Why do you need three pumps running? If the head required to get into the DA is too high for two pumps, can you try dropping the DA pressure? If your pumps are vertical turbine configuration, you may need to add stages to get the extra head with only two pumps running.


Johnny Pellin

 

[BigInch]

I do think piping configuration and flat curve pumps without individually fitted controls tend to make balanced ops difficult, but these other concerns make a lot of sense too.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[EdStainless]

When we ran multiple pumps in parallel we usually added some restriction to the outlet of each pump. This minimized how much the pumps would 'talk' to each other.
This really does sound like a header flow dynamics issue.

= = = = = = = = = = = = = = = = = = = =
Rust never sleeps
Neither should your protection

http://www.trent-tube.com/contact/Tech_Assist.cfm

 

[BigInch]

It can be surprizing how an unbalanced header can affect flat curved pumps. Simulate it with just using a simple Excel spreadsheet with the 3 pumps, give one a bit of a differential flow and see if the others pick up the change and give any significant compensation.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com