Torque vs Speed Pump Curve

[GaTechTheron]

Question:

I have noticed pump companies provide speed vs torque curves for their pumps. These curves are % Max Torque vs % Max Speed. This implies that this curve can be applied to any pump so long as motor rpm and torque (at synchronous operating speed) is known.

My question is 2 fold:

1. How can you make this assumption? How can you assume the torque is proportionally the same in two completely different types of pumps? It seems long multistage pumps (Reda or Woodgroup Style) would have far more friction than a simple stage OH2, for example. Maybe I am making an underlying assumption somewhere?

2. How is this curve developed, and why does it start higher (around 15% torque at 0 rpm),then decrease to 5% torque at 12% Speed? Since this is not a transient curve, and (I think) rpm is assumed to change slowly, it seems % torque should always increase, starting from zero, and inertia effects should not show up.

I hope this is worded clearly. Thoughts anyone?

Thanks,
Theron

 

[GaTechTheron]

Really what I am looking for is a proof of how this curve is developed. That would help me understand completely.

 

[vpl]

If you want a "proof of how this curve is developed," your best place to start is with the manufacturer who developed it.

Patricia Lougheed

******

Please see FAQ731-376: Eng-Tips.com Forum Policies for tips on how to make the best use of the Eng-Tips Forums.

 

[GaTechTheron]

VPL-

I have seen this this same thing from multiple different manufacturers. Thus, my question.

 

[NGLENGR]

Well, I'd probably start with the hydraulic calculation to determine where the energy comes from. Then, you take the energy and determine the forces on the impellers. Then the forces become torque.

There's also friction (static and kinematic) to account for.

You'll be lucky to find out what the actual proof looks like.

 

[electricpete]

why does it start higher (around 15% torque at 0 rpm),then decrease to 5% torque at 12% Speed?

That represents the effects of breakaway-type friction.

Most pump curves I have seen are geared to specific pump or small group of pumps.

However in absence of pump curve, sometimes it is assumed for purposes of motor starting calculations that torque is proportional to speed squared (in which case you just need one point to determine proportionality constant... not a whole curve). I believe that is related to pump laws under certain idealized conditions.

=====================================
(2B)+(2B)' ?

 

[GaTechTheron]

Example pump curve is attached.

 

[BigInch]

The torque equation for centrifugal pumps,

T = 3300 * SG * dH_ft * Q_gpm / (3960 * eff * 2 * pi * rpm)

can be plotted for any speed, remembering that for a given point on a pump curve at one speed, the head and the flow will change as speed reduces, but efficiency is assumed to remain equal for all speeds, which is "nearly" true.

The flow varies directly with the ratios of speed, Q₂ = Q₁ * RPM₂/RPM₁

and head varies by the ratio of the speeds squared,
H₂ = H₁ * RPM₂²/RPM₁²

If you looked at the BEP full speed point and saw an efficiency of 0.76, it would be assumed to be equal at all speeds when sketching up a theoretical Torque vs Speed curve and the curve begins at 0,0 and increases with speed squared.

In a real curve the efficiency, being much lower than 0.76 at very low speeds, dropping to around 20% eff at 10% of maximum speed, due to break away effects of bearing and stuffing friction EPete mentions plus internal flows not being fully developed at low speeds, causing the Brake Power and Torque to rise above the theoretical curve for all speeds lower than around 10% of maximum speed. That is pretty much true for all centrifugal pumps, as are the affinity laws for flow and head, so the curves for one point on a pump curve modified for a speed change will all follow the same tracks as speed decreases and all will have the same shape, even though the actual shape of the pump curves may be vastly different between different types of pumps.

The Torque vs Speed curve you are seeing only represents one point on the pump curve as it is modified by % speed. Since one point on the pump curve does not make a full curve, what you see is not dependent on the shape of any particular pump curve and the result is the same for all pumps.

 

[Artisi]

to you point 2 question - if torque at 0 rpm was zero the pump would remain stationary, the 15% is "breakaway torque" needed to get the pump away from 0 rpm, once rotation is achieved torque reduces to suit the hydraulic load and then gradually increases as the pump comes up to full speed.

 

[BigInch]

Yes, but how does it happen. Motor input torque is high when starting because slip is maximum and probably torque delivered by the motor is around 150% of rated at very low rpm, but none of it can be converted into useful hydraulic torque to develop any power to deliver to the fluid, because the pump is overcoming static friction and fluid inertia as it starts spinning up. There is also no other force from frictional flow that requires any further torque until the fluid is moving fast enough to start developing friction, but as it finally starts getting to the point where it starts developing friction, torque and power increase with its velocity squared. Note that much of fluid friction is actually converted into useful work and power within the fluid, whereas stuffing and bearing friction is not. What you're seeing is first a higher static friction coefficient to a mostly lower linear dynamic friction coefficient, but still totally parasitic force, where none of that frictional force is converted into any usefull hydraulic work and hydraulic efficiency remains very poor to a region where the frictional forces can be used by the fluid. When the rapid addition of fluid friction effects quickly become more prominant over the linear friction effects, because fluid friction increases, not linearly, but with the square of the velocity (shaft speed), plus most of that friction within the fluid itself is actually developed into the forces used to perform the useful work needed within the fluid to increase pressure and flow of the fluid, (useful work), efficiency quickly becomes quite high. Its the transition between mostly static/dynamic and parasitic friction to the useful fluid friction that casues the dip in the torque curve.

 

[GaTechTheron]

Big Inch Strikes again.

Thank you for the thorough reply. I may assume the curve the manufacturers provide is basically a general curve, and by the definition above these are not exactly the same but only the same in characteristic.

Thanks!
-Theron

 

[rmw]

Pumps vary one to another due to variances in castings - shape and surface features, impeller features, and other general manufacturing imperfections inherent with things made from castings. A pump manufacturer will take a pump or some pumps which are generally as nearly textbook perfect as possible and test it (them) to develop curves and then publish them as the curves for that series of pump. Some will test one size pump and then extrapolate the curves developed for the "family" of pumps.

They also base their curves on water at a specific temperature so any variation you have from that has to be taken into consideration.

They also prevaricate sometimes so if you want to be absolutely certain about the curves, test the pumps yourself. I worked for a company once that did just that in a test lab and we knew which pump curves we could trust and those that had to be taken with a grain of salt.

On the other hand, pump physics is a pretty mature science so they can't be too far off or their competitors would nail them in the market place. I have no problem using pump curves as published when sizing and selecting pumps. Unless I'm dealing with NPSHr curves, and then I add my grain of salt.

rmw

 

[ccfowler]

rmw, I very much agree with your observations regarding published pump curves--especially the reliability (or excessive optimism) of the NPSHr curves.

GaTechTheron, I've never encountered a pump manufacturer that would not discuss their pump's actual characteristics forthrightly when the questions were posed in good faith, intelligently, and with adequate information being provided regarding the actual pump installation and its connected system.

Unless a very large or important (translation--very expensive) pump is involved (where pump-specific, witnessed test bed derived performance curves are included in the purchase price of the pump), it is important to recognize that the generally published data is only representative of the performance that can be expected within about 10%. Although there may be some optimism in the published data, the variations that a specific pump may display (when compared to the published data) are primarily the result of reasonable manufacturing variations and tolerances.

Just for the record, I have never worked for or had a financial interest in any pump manufacturer, distributor, ....

Valuable advice from a professor many years ago: First, design for graceful failure. Everything we build will eventually fail, so we must strive to avoid injuries or secondary damage when that failure occurs. Only then can practicality and economics be properly considered.

 

[Artisi]

What must be born in mind with published curves especially NPSH data is that the testing has been or should have been be under controlled conditions and corrected to standard temp, altitude, atmospheric pressure using air free water under ideal inlet flow conditions etc etc.
These conditions are rarely if ever found with site testing, hence the continual cry from end users "the curves are wrong, the pump doesn't perform as we expect" etc etc.

It's always easier to blame the pump / manufacturer for all site / installation short-comings than accept faultly installation, suspect measuring equipment or under conditions other than what the pump was selected to achieve.

Just for the record I've work for manufacturers, suppliers and for end users so have seen all side of the problems and can say from experience that only "sometimes" is it actually the pump at fault.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[ccfowler]

Artisi,

Great quotation! What excellent advice for all to observe!

Valuable advice from a professor many years ago: First, design for graceful failure. Everything we build will eventually fail, so we must strive to avoid injuries or secondary damage when that failure occurs. Only then can practicality and economics be properly considered.

 

[BigInch]

Arn't we getting away from the original question. Pump curves usually infer Head vs Flow and Torque vs Speed are quite different. HQ curves are entirely dependent on the specific pump. TS curves are drawn for only one point on a HQ curve. Once the efficiency is known for any given point on any pump's HQ curve, the TS curve for that point, or any other pump capable of reaching the same point at the same efficiency, is defined and does not change shape between pumps.

 

[Artisi]

BigInch - correct as usual but not the first thread to get turned on its head.
If you search the net there is a number of references to this subject, this is one of then.

http://centrifugalpump.org/speed_torque_curve.html

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[BigInch]

Right. You do have to get used to reading upside down at times.

 

[rmw]

My post was to try to address and amplify on the assumption made by the OP in the preceeding post which related back to question 2 of the OP although in more about curve development in general than the speed torque in particular.

rmw

 

[BigInch]

OK. It looked to me that he was still talking about the torque curve in question 2 also. No harm done=No worries.