Testing a pump dead-headed

[impeller1]

Good day Folks,

Im sorry if this is a dumb question, but I have 6 vertical pumps. I'd like to test them to determine where they're operating on their performance curves. However, I do not have the system curves and I'd like to test the pump alone minus the system. These pumps are VFD's so I can vary the speed to vary my flows and heads. They're high head, rated - 388 feet @ 4422gpm. Is it acceptable to close the discharge valve on the pump and quickly take the pump through about 5 different speeds to obtain flow and head values? I know the water would recirculate in the pump and may increase vibration and temperature on the bearings and pump itself, but if it's done for only a short time just for me to get my readings, is it plausible? And would I then be able to ignore them system curve as the discharge valve is several feet from the pump discharge.

Thanks

 

[MikeHalloran]

I don't understand how knowing head values for zero flow at some commanded RPM does you any good. I don't think it does the pumps any good. So, don't close the valve.

You do have the pump curves, right? So if you know the pressure difference and the RPM, you can estimate the flow, which lies on the system curve.

Change the RPM, get another pressure difference, estimate the flow using the pump curves, and you can map some of the sytem curve. Or just log the DP and RPM as the pumps do their thing.

Or extrapolate from the one point. If nothing else changes, the entire system curve should follow a square law curve, of the form Q{gpm}=Cv*sqrt(deltaP{psid})


Mike Halloran
Pembroke Pines, FL, USA

 

[impeller1]

It's a strap on ultrasonic flow meter we're planning to use. So the plan was to strap it on to the discharge and measure the flow and then measure the pressure from an existing gauge that we have.

 

[impeller1]

But then again..when I think about it..the recirculation may not allow you to get an accurate value of flow..???

 

[Artisi]

Correct - and why risk running these pumps at CV for no good reason as the information you will get is near enough to useless.

 

[Artisi]

forgot to add - you will be better served to "strap" the ultrasonic flowmeter onto the pipe work well away from any turbulence and as pointed out by Mike, once you have an accurate flow and total discharge pressure - knowing the static head you can calculate a system curve.

 

[BigInch]

Just get a point at minimum flow and, if you really think you need the shutoff head for something, just extrapolate the curve backwards. Practicly speaking, you really don't have to know those value below minimum stable flow for any good reason I can think of right now, unless there is an outside chance that you have some startup problem with a rising head in that region that traps the system. Hard to believe that. A freehand value should do OK for most purposes, don't you think?

**********************
"The problem isn't working out the equation,
its finding the answer to the real question." BigInch

http://virtualpipeline.spaces.live.com/

 

[JJPellin]

These are pretty big pumps to be running dead-headed. You didn't mention how many stages they have. But, given that head, I assume that they are multi-stage. I would be concerned about thrust running at no flow. The set of curves I was just looking at for a big multi-stage vertical turbine pump included thrust curves for down-thrust per stage. For the particular pump I was looking at, the maximum down thrust was at zero flow and depending on the number of stages, this could be a risk of motor thrust bearing failure.

I agree with the other replies above. If you can run the pump under normal flow conditions at a couple of different points, you have three variables that you can measure to determine where you are at on the curve: Discharge pressure, motor amps and flow rate. Depending on the piping configuration, the clamp-on flow meter may be the least accurate of the three. But, between discharge pressure and amps, you should be able to confirm or deny the flow measurement result. We do this with our largest cooling tower pumps since it is impossible for us to get any flow measurement for an individual pump.


Johnny Pellin

 

[BigInch]

BTW, with a closed valve, you don't have a system curve. If the valve's not closed, you can never ignore the system curve.

**********************
"The problem isn't working out the equation,
its finding the answer to the real question." BigInch

http://virtualpipeline.spaces.live.com/

 

[impeller1]

Thank You all for your replies and insight

 

[Artisi]

It's good to see a 'thank-you' for a change.

 

[BigInch]

Ya. Impeller1's a nice guy. That's probably why he gets all the good answers too.

**********************
"The problem isn't working out the equation,
its finding the answer to the real question." BigInch

http://virtualpipeline.spaces.live.com/

 

[dabluffrat]

Shut off head is important to know when considering impeller design and modification.

On the mfg test bed, radial flow impellers in vertical pumps are taken back to shut off almost all of the time.

Axial flow impellers in vertical pumps, not so much as the HP rise to shut off and down thrust generated are often too high for the motor.

I've done many field tests as well and agree, pumps should not be intentionally taken back to shut off when installed. MCSF is close enough.




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