Pump flow from kW?

[birkATO]

I have a bunch of kW data (hour by hour)used by some pumps that are used for pumping water from one area to another. They are rather high volume pumps...around 3000 gpm each. What I want to do is take all the kW data and use that to estimate a flow rate for each pump during each hour to get a total of how much water has flowed. I have never done one of these types of calculations. I have some pump curves. Would the following steps be a correct way to go about it?
1. Determine total head, which is 236 ft.
a. I have estimated the friction head to be around 2 ft using a friction head calculator on the internet (10,000 ft of 30" ID steel pipe)
b. Water level is raised by 234 ft.

2. Using the bowl head vs. flow curve, find the flow rate.

I don't know where to go from here though and it seems as though I am missing something since from what I have assumed, the flow rate does not depend on power input. I presume I will have to multiply the kW number I have by some efficiency, but with the efficiency vs. flow curve, I need a flow to find efficiency.

I also have a curve for BHP vs flow. Can someone point me in the right direction?

 

[pmover]

for centrifugal pumps:

bhp = gpm x head (in feet) x SG / [3960 x efficiency]

where
bhp - brake horsepower
gpm - flow rate in gallons per minute
head - total pump differential head in feet of liquid
SG - specific gravity of pumpage
efficiency - pump efficiency.

since you have the pump curves, you should be able to readily determine flow knowing rpm, impeller diameter, and pump total head.

good luck!
-pmover

 

[birkATO]

I do know the total head, but I don't know the efficiency.

 

[cvg]

you say 2 feet headloss, this seems low. with just one pump operating at 3,000 gpm you will have closer to 3 feet of head loss. With two, your headloss increases quite a bit to about 9 or 10 feet of loss. To get a better handle on the loss, you need to estimate the loss coefficient for your pipe and use the total flow through the pipe to calculate the loss.

 

[EdStainless]

Look at your pump curves. It should have eff and power curves on it.

= = = = = = = = = = = = = = = = = = = =
Corrosion, every where, all the time.
Manage it or it will manage you.

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

 

[JohnGP]

birkATO,
To know total head you need to know the total flow in the system, as friction head is proportional to square of the velocity, so I'm not sure how you figured 2 ft friction head.

You have BHP vs flow for each pump. The BHP is pump BHP, so you need to convert the kW readings to HP for each pump(divide by .746), multiply that by motor efficiency, then read what flow that gives from the BHP/flow curve.

For the overall system you can develop a System Curve taking account of the static suction and discharge heads, and the friction head as it increases with flow. Against this can be plotted the combined curve for all the pumps operating in parallel to give the operating point, and hence an indication of the total flow in the system. Use this to back check whether you are in the right ball park with the individual pump flow estimates.

Things can be a little bit out if the pumps are not behaving exactly as their curve predicts, but you may get a reasonable idea of flows.

Cheers,
John

 

[Artisi]

For what you are asking and the way in which you are trying to establish the flow rate it is impossible to get an accurate estimate as there are far too many unknows.

Firstly, you would be a certified test curve for each pump unit and assume that the pumps are still in exactly in the same condition as when tested.
Secondly, you will need to accurately measure total head.
Thirdly, you will need to measure the pump speed at the operating duty.
Fourthly, you will need motor test data to assentain the motor efficiency and speed at which it was operating on test and which needs to be speed corrected to the actual duty operating speed.
If you had this information, ie test curve, total head, and motor efficiency and speed, you could then extrapolate this information onto the "known" pump performance curve to give you a fairly accurate flow rate.

However, it is much easier if you measure the flow rate directly which also gives you an accurate answer.

You can measure the discharge trajectory at the discharge point which can give a reasonably accurate flow rate.

Or get an "expert" to look at the discharge point - "yeah -that looks like about 3000 gpm" -- about as accurate as guessing without all the necessary information.


Naresuan University
Phitsanulok
Thailand

 

[JohnGP]

Artisi, measuring the trajectory at the discharge point, or having your "expert" have his say, will only provide an indication of the total flow in the system, provided there is only a single outlet and has an open ended discharge. But then that may be good enough anyway - if I understand the original post correctly, there are a number of pumps operating in parallel and it is desired to extrapolate the kW readings for each to obtain a total flow. That was the basis of my earlier response anyway, although I was suggesting how to estimate single pump performance as well.

Of course, if the pumps are old then efficiencies may have deteriorated and the published curves may no longer tell the truth, but it all depends on how accurate an estimate one requires. I'll stop rambling now.

Cheers,
John

 

[Artisi]

JohnGP, The comment on "Expert" to measure the flow by looking at the discharge point was of course tongue-in-check. You must have run across these "experts" at various times, these the same people who can measure the temperature of overheating bearings / electric motors by hand and can also measure vertical and horizontal distances by eye to an accuracy of +- 1 - 2%.

Also, the accuracy you need dictates the method to use.

Naresuan University
Phitsanulok
Thailand

 

[JohnGP]

No problem, Artisi, I knew exactly what you meant. I lost count of the number of times that even seasoned operators would say to me "Gee that motor is running hot" after having felt it by hand.

Without knowing all the ins and outs of the above problem it is difficult to provide precise advice. I agree that the accuracy required will dictate the method to use, and conversely, the method you use will dictate the accuracy of the result.

Cheers,
John