Pump test

[mottto]

Hi,
I don’t know if anybody has seen this before, but recently as i was testing a pump, I found out something strange, and would like you guys input. Here is what s going on. When i test a pump and go from shut off all the way to the max flow, i get a set of data, that is different than one i will get starting backward, meaning i will start from the end of the curve and start closing my valve until the shut off point.
@ Shut off i got the same head, but at every other point my flow was different by a bout 400 GPM. I m using a magnetic flow meter, and it shouldn't matter if I m going backward or forward, it should be reading the same correct?
Anybody has any idea?
Thank you

 

[BigInch]

Strange.

Don't you have the same data point at the end of the curve too? Seems like if you started at shut-off, opening the valve going to max flow, reached max flow and started closing the valve, you would get the same data points at beginning at shut-off and also at max flow at the right end of the curve.

I guess that you are also mesuring differential head rather than total discharge head, right?

The test fluid is water, correct?

What pump type, power, flowrates, pressures and pipe diameters are we talking about here?

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[FACS]

Is there a pressure relief valve on the unit?

I know while designing relief valves, I had trouble getting past the crack pressure issues. If you start at a dead head, the relief valve is open, and then closes at a rate different than if you were bringing up the pressure from full flow. This could give you a different curve.

Charlie

http://www.facsco.com

 

[mottto]

BigInch , you are right. I should be getting the same data, but im not. The head we are measuring is the Total head.
Pumping Clean water on a vertical pump. Bowl Size is 24"
Piping thru a 12" discharge.400 HP , about 10000 GPM @ 85' of head.

 

[BigInch]

Do you want to print the curve data you have?

You should be measuring pump differential head to discount the following possibility,

If the suction head is varying when going in one direction vs the other, possible curve differences could be expected, as in a case where increasing pump flow empty a supply tank which reduces suction head, then reaching the end of the curve, you start cutting back on pump flow, thereby beginninbg to fill the supply tank and increase NPSHa.



BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[mottto]

BigInch ,
Thanks for the reply , i still dont think it is an NPSH problem , since the pit is 90' deep and about 30'X 50' long.
Im posting a layout of the lab. Im suspecting they might be some turbulance messing up the flowmeter readings.

http://img211.imageshack.us/my.php?image=testttlw7.jpg

 

[BigInch]

Is it a new test arangement? Offhand I would think 10 diameters off a pump discharge is very close to get a good accurate and consistant flow measurement. That is the minimum recommended distance between any type of fitting and a pump, not a flow meter and a pump. I would at least put 10 D + straightening vanes then another 10D before a meter.

Have you noticed this with other pump tests done with the same configuration and pump type, or just this one?

I have recently seen an article about a type of prerotation that is sometimes set up in pumps that persists until flow is radically changed, which just might account for a variation of this type. Just a guess.


BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[FACS]

I'll just say something stupid here, cause I'm good at it... I don't suppose there could be a siphon effect on that set-up...?

Charlie

http://www.facsco.com

 

[BigInch]

No problem... we're all good at it from time to time.

Possibly, There may be some induced backpressure from the turns there when the pipe is flowing full, but probably you would see the same effects when either increasing or decreasing flowrate. And IMO putting backpressure on the pump by using the gate valve where it is will tend to negate any downstream effects. Wonder if the gate is affecting the meter somehow too. That gate should be at least 5D downstream of the meter too.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com