Interpretation of pump flow from performance curve

[tbedford]

I have been an air balancing technician for years and have thought of centrifugal fans similar to centrifugal pumps in that they both create a pressure difference. This pressure difference is what starts the fluid moving.

I am working more and more on various sizes of closed loop hot/chilled water hydronic systems. When it comes time to determine pump flows from the supplied performance curve, I usually find that the pump flows are "off the chart"; above the pump curve, so to speak. I have always thought of this as an inefficient place for a pump to be; that is to say energy is being wasted.I have just calculated very high pump flows but using about 60% of the rated amps on the motor. My assumption is that the pump is likley to be over-delivering. Is this a correct assumption and what is the pump doing if it is not on the manufacturer's plotted curve? Is there a reference for reading on this subject?

 

[MikeHalloran]

The pump flow cannot be 'off the chart'. You must be doing something less than perfectly right.

The Operating Point for a pump lies at the intersection of the flow/head curve for the pump, and the flow/head curve for the system.

How are you estimating the system curve?












Mike Halloran
Pembroke Pines, FL, USA

 

[Artisi]

What does "off the chart" mean -- more than likely pumping at a head lower than end of curve and at a flow rate greater than shown at the end of curve - in a closed loop this is very possible.

What size pump unit are you talking about.

Phitsanulok
Thailand

 

[tbedford]

I am calculating the difference between pump discharge and suction (psig x 2.307/sg = ft head). I am usually measuring water so sg is usually one. On the performance curve, I track the calculated feet head parallel to the flow line (abcissa)until this line meets the pump curve. I read perpendicular down to the pump flow. My confusion with this method is that very often my calculated head (ordinate) is often higher than the highest performance for a given pump horsepower. I am confident of my pressure readings. Most contractor's are cheap and install the smallest gages possible, so I have built a manifold with large gages and I calibrate them against a dead weight for accuracy.

So if my readings are accurate, why is my head higher than my performance curve? I would think the pump should not be operating but the building has plenty of heat. In some cases I can even hear high water velocity in pipes. When I try to increase system resistance by closing off various valves ( gate valves or balancing valves), velocity noise doesn't decrease.

Mike...I don't estimate system curve. I am going to start but I must confess my ignorance about where to begin. Any reference material?

Phitsanulok...pumps are usually small...in the order of 3 to 10 horsepower.

Thank you for the questions,

T

 

[BigInch]

Its a differential pump head on the curve, so you are subtracting the suction head or pressure equivanent, right?

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[whieber]

Replying to the first post by tbedford. My experience with closed sysytems is limited but I have run across them a few times and here is what I have learned. It's been my experience that people who design closed loop systems often choose a pump that is over sized especially considering the head because they have to select a pump that will also be able to fill the system while it is open. So you may find a fluid circuit that requires 75 ft of head for filling purposes but only requires 20 for normal closed circuit operation. In the event that the system has to be drained and then refilled a pump that provides only 30 ft of head at dead head isn't going to do much good when it requires 50 ft. The key is to select a pump that can do both and still perform efficiently. Some of the best pumps that I have seen to do this are by MTH pumps:

http://www.mthpumps.com/

Hope I might have been helpful.

Walter Hieber
PIB, INC. -

http://www.thepumpbox.com

 

[BigInch]

Seems like that could be a good maintenance feature, but of course could be quite wasteful when operating. Is there a recommended practice for that?

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[tbedford]

Big Inch, correct, I am subtracting the measured suction pressure (or its equivalent) from the measured discharge. This differential pressure (or feet head) is the basis for determining pump flows.
Wieber...we typically use city pressure to fill the systems not the pump. What you say makes sense.

T

 

[whieber]

As far as operating efficiently lately I have seen a real big push towards using VFD's to control systems with a big benefit in energy efficiency. If you aren't familiar with VFD's I'll give you a quick summary: VFD= Variable Frequency Drive. In essence its an electrical device placed between the raw power source and the pump that allows you to control many features on the motor such as tourque, RPM, ramp up speed, etc. There are several of these that have been programmed specifically for pumps so when they are installed they simply require you to set some basic motor data and some basic system data and then you can control the rest with a keypad similarly to a TV remote control with a read out. They aren't really necessary in all cases but in cases where you find that you are wasting a lot of watts due to ineffcient system matching a VFD could pay itself off in a years time.

I know that many of the motor manufacturers are producing them and I am a little familiar with some produced by Goulds called the pumpsmart, and the aquavar. The PumpSmart devices are amazing but in many cases overkill because they have enough features to make a full time hobby out of using them. Luckily they have come down in price since they first showed up and are actually not a bad idea at all if you want to save money in energy, extend pump life, control based upon process changes, and have the pump log system changes and failures. Its really pretty cool if you like that kind of thing. The aquavar is a lot simpler but also does a really good job protecting the pump and will allow you to run efficiently and to change pressure and flow.

I only bring these up because I recently installed one on a 125HP vertical turbine and based upon the improved efficiency it was estimated that year would save about $6,000.00 in energy cost not to mention repair cost and valve replacements.

Walter Hieber
PIB, INC. -

http://www.thepumpbox.com

 

[MikeHalloran]

The pump performance curve that you should use to estimate flow Q from head dH is the one that corresponds to the actual impeller RPM. You would have to interpolate the curve if a VFD is present. The formulas for doing that are called 'affinity laws'. The most concise and useful treatment I've found for that and myriad other subjects is:

"Handyman In- Your- Pocket", Young/Glover, Sequoia Publishing, Littleton CO, 2003 ... A small, thick, bright yellow paperback, which I bought at a Cracker Barrel restaurant for US$12.95, cheap at ten times the price.

The "Handyman" has some useful rules of thumb for estimating pipe resistance, etc., but I personally hate doing stuff like that on paper. Poke around on the Internet and you can find Excel programs that will calculate head loss for a given flow, given length and type of straight tube, number and type of elbows, stuff like that. The trick is to 'model' the loop you've got, figure out the pressure drop for an arbitray flow, or the flow to produce an arbitrary pressure drop, then compute a Cv for the system, which has units of

gpm
Cv = ------
sqrt(psid)

If you feel like screwing around with Excel, you can manually input a pump curve as a series of Q-dP points, and use the computed Cv to produce the system curve, graph them both, and see how they overlay. As always in Excel, the first time is the hardest.



Mike Halloran
Pembroke Pines, FL, USA

 

[tbedford]

In my current problem the pump is 3HP and serves 4 floors approx 3000 ft. of pipe with 51 reheat coils and various valves. The size of pump likely does not justify a VFD although speed reduction is a favourable option.
Curious also, pump curves are given at 1800 rpm and motors are 1725/1755 range...the idea being that a small difference is not critical.
The two boiler loops have several circulators to add pressure drop.
I am going to search for the pocketbook.

T

 

[sailoday28]

While this may only be a minor effect, the head, h, on the vedors H-Q curve should include in addition to the difference in pressure head

a) change in kinetic energy V^2/2 and
b) change in elevation
between the suction and discharge flanges oof the pump.

Regards

 

[Artisi]

Suggest you go to google and type in Grundfos - search their site for information on recirculating pumps, should be a gold mine of info there.

Phitsanulok
Thailand

 

[ccfowler]

tbedford,

How are you measuring the actual flow rate? How accurately are you measuring the flow rate? Are you taking water temperature into consideration? Is the density of the water being affected by additives for corrosion control, dissolved contaminants, etc.?

If you are using orifice meters, have you verified the actual dimensions of the orifice and pipe, and have you verified that the orifice plate is properly installed by direct, personal inspection?

If you are using venturi meters or flow nozzles, have you verified their actual dimensions and the actual pipe diameter?

Are the flow meters (of whatever type) installed in sufficiently long straight pipe runs and far enough from any disturbing features (such as valves or fittings) to assure optimum meter accuracy?

Have you verified that all pressure tap holes are clean and free of burrs?

Are the pumps located with long straight pipe runs at both the suction and discharge ports? If not, your suction and discharge pressure readings may be less accurate than you may like. Are you taking elevation of the pressure taps relative to pump suction and discharge into account in your measurement of pump head?

You may want to consider using a differential pressure measuring device (such as a manometer) rather than two gauge preaaure devices for measurement of pump head. This can reduce the number of potential instrument errors being an additive complication for accuracy.

Although these may seem to be fussy questions, flow metering can be very troublesome and misleading unless all of the little details are thoroughly verified and cleaned up as necessary.

Since you are dealing with relatively small, inexpensive pumps (compared to likes of multi-thousand horsepower boiler feed pumps), it is unlikely that you are using unique test-stand based pump curves for each individual pump. As such, the actual pump performance can resaonably be expected to vary within about +/- 10% of the published generic pump curves for the particular pump model.