Effect of VFD on motor efficiency

[jraef]

Believe it or not this question has never come up for me before. When we reduce speed in a motor with a VFD how do we evaluate the efficiency of the motor itself at the reduced speed? Is it based on it being essentially a lower kW motor running at peak efficiency, or does the motor efficiency vs load curve stay the same regardless? To my mind, if we have reduced the speed of 20kW motor to 50%, essentially we have changed the motor to a 10kW motor, so the efficiency would still be somewhere in the higher range. But what I'm not clear on is that when a motor's efficiency percentage drops off with load, isn't it mainly because the percentage of consumption in magnetic losses remains relatively constant regardless of load? If so, do those losses decrease when using a VFD because we are decreasing the voltage as we reduce frequency?
Any thoughts?

Just for argument sake, forget harmonics etc., this is more about how motor efficiency curves are determined.

"If I had eight hours to chop down a tree, I'd spend six sharpening my axe." -- Abraham Lincoln
For the best use of Eng-Tips, please click here -> faq731-376

 

[ScottyUK]

Windage losses could be significant when looking at efficiency calcs: it would make a big difference if you had to factor in an auxiliary blower to the overall efficiency calcs rather than the shaft-mounted fan which will be consuming much less power as speed drops away.

On the magnetic losses: the iron losses aren't caused by the density of magnetic flux per se, and flux density is determined by the voltage. Magnetic saturation is governed by voltage, so the voltage has to reduce with frequency to avoid saturation, but the relationship with iron losses is more subtle. The iron heating is caused by 'friction' - in the loosest imaginable sense of the word, but it's a convenient choice - between the magnetic domains each time the flux reverses direction. If you think of the stator iron heating as being a twice-per-cycle occurence then as the frequency drops the heating effect should proportionally reduce even if we assume a constant flux density (linear V/f relationship).


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[Brad1979]

My 2 cents: In general, assuming a fixed load the efficiency of the motor will increase as the VFD increases the speed of the motor. Stator and rotor I^2R losses, friction and windage losses, and stray load losses will all probably increase slightly. Iron losses will increase a bit due to higher frequency. But because the input power of the motor is increasing, the efficiency of the motor actually goes up because the increase in the losses isn't as great as the increase in input power. I'm basing this on some simulations I've done in the past on small induction motors (5 HP or less). I'm assuming this would hold for larger motors too, but I can't be sure.

As far as testing, you'd want to set up the test to measure voltage/current/power into the VFD, voltage/current/power into the motor and then hook the motor up to a dyno and measure speed as you vary the torque on the dyno. Do this for various frequency settings on your VFD and you should be able to calculate VFD and motor efficiencies at different input frequencies.

 

[flexoprinting]

Regardless of everything else, couldn't you simply compare current to HP.

 

[electricpete]

At the bottom of the page here you can see that someone else has looked into the same question and found it not easy to nail down.

http://findarticles.com/p/articles/mi_m0BPR/is_8_20/ai_107123416/

I tend to agree with Scotty on magnetic losses. They go roughly proportional to flux density times frequency-squared... so should decrease substantially.

The contributors to loss include friction, windage, core, I^2*R, stray. We'd have to think a little bit about each one.

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

 

[electricpete]

Let’s say I start at 50% load and normal voltage/frequency. We can read the efficiency from our chart.

Then decrease voltage and frequency to 50%, while maintaining the same loading (expressed in horsepower).

The magnetic losses go down. The friction and windage goes down. I suspect the stray goes down to the extent it includes magnetic losses in structural components (although it may also just be a bucket for unknown). So far it sounds like we can predict higher efficiency. BUT the I^2*R goes up since current is higher to deliver the same power at lower voltage without dramatically different power factor. Which one wins now?


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

 

[Brad1979]

Just to be clear, when I said "assuming a fixed load" above, I meant "assuming the torque stays constant" while the speed changes. When the horsepower is constant and the speed changes, then things get more complicated. In this case, my instinct tells me that efficiency is going to decrease because I^2R losses generally dominate the total losses. What I don't know is the opposite situation ... where speed and torque decrease (e.g., fan or pump applications). I will have to think on that one.

 

[electricpete]

I assumed we are comparing same horsepower load, since the intention was to compare efficiencies to the published full-frequency efficiency vs load curve. Maybe Jeff can clarify.

Attached are some curves from Andreas' book on Energy Efficient motors. It looks like (for my assumption), the I^2*R losses dominate the loss picture, and providing a given output horsepower at a lower speed (and voltage) means more losses, lower efficiency.

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

 

[electricpete]

I guess you could roll both pieces of information together.

Determine the torque, power and speed for the operation condition of interest. We want to determine efficiency E for that condition.

Let efficiency Ect be the efficiency of motor operating at the same torque but full speed.

Let efficiency Ecp be the efficiency of motor operating at the same horsepower output but full speed.

Ecp < E < Ect

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

 

[electricpete]

Sorry, brain lapse.
Ecp < E < Ect
should've been
Ect < E < Ecp


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

 

[electricpete]

The inequality presumes that the full-speed efficiency curve must be increasing with load (such as a curve that peaks near 100%). That would be the case for a machine dominated by I^2*R losses since it fits with our assumption about Ecp. May not hold for all cases and if no-load losses end up playing a significant roll we may have Ect > Ecp and the inequality makes no sense. So it's not a perfect analysis, (the only perfect analysis is probably "ask the manufacturer"), but I think it is a pretty good starting point.

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

 

[ozmosis]

Jeff
It has been a very topical subject for a few years in our (both) previous company. A lot of debate over which, if any, standard to try and use and where to base any hard fact/calculations.
The advent of PM Motor technology is making this even more of an interesting debate as you can only run a PM motor away from a mains supply. The analysis we have been doing is quite startling (showing the benefits of PMMotor very high compared to Induction when using similar test processes).
An interesting article from a University in Belgium:

http://www.kupfer-institut.de/front_frame/pdf/kul_pres_rb_pvr.ppt

The bottom line is, all efficiency data produced by the manufacturer on nameplates and manuals goes out of the window when connecting a VFD.

 

[potteryshard]

It has always been my understanding that nothing a VFD can do will increase (or even maintain) motor efficiency, and that adding the losses created by the VFD further reduces drive efficiency.

It is only when considering the sum of not only motor and VFD losses, but also process losses of the driven machinery and further downstream that overall system efficiency can be improved. Typically, the throttling losses of fluid transport systems or machinery which usually operates at less than full output can be saved by precisely matching unthrottled flow (or other work) by adjusting motor speed instead.

On the other hand, it has been some time since I've become involved with any VFD work. Have they improved to the point where the efficiency of a fixed speed drive can be improved by installation of a VFD?

 

[jraef]

Excellent information all. Thanks very much.

I guess I'm a little shocked by how much the efficiency drops when changing speed. I had assumed, as I stated in my opening, that if the LOAD HP requirement dropped along with the speed drop, then the motor would remain at a favorable efficiency point. In other words if my 10HP motor was turned down to 50% speed, it is a 5HP motor, and if the load requirement dropped to 5HP as well, it is still "fully loaded" and operating near the peak of the efficiency curve. But from what I can glean from these studies and what you guys are saying, the "other" losses increase faster. I see now that the effects of running on a VFD are worse than I expected in that regard.

Thanks again. This is exactly what I was asking for, albeit not what I was expecting (or wanting) to see.



"If I had eight hours to chop down a tree, I'd spend six sharpening my axe." -- Abraham Lincoln
For the best use of Eng-Tips, please click here -> faq731-376