Bearing life calculation for a wind-turbine generator?

[AshleyK]

Hi, I'm after a bit of assistance with a calculation for a wind turbine generator bearing. I have performed the calculation by hand for other plant before, but obviously the wind turbine operates at a constantly varying RPM, and despite contacting SKF and searching every handbook and engineering reference in my library I still can't find a guideline on where to start with anything other than a calc for constant RPM. Even SKF say they just plug the numbers into various 'bins' on their system and a life is provided.

Any experience with working this out by hand? Any references online worth checking out?

Many thanks in advance,

Ashley.

 

[Skogsgurra]

You have that problem in many cases. A paper machine that produces over a wide speed range, a smoke stack fan, a propulsion system - just about any system with variable speed. Calculate for worst case and regard any extra life as a bonus.

But... and that's a Big But (no pun), a generator for a wind turbine that is able to run at different speeds is usually double fed. That means that a variable frequency is fed to the rotor winding and that the tight capacitive coupling between winding and rotor iron causes a heavy high-frequency current to flow through the bearings to generator frame, which destroys the bearings - see EDM.

The biggest challenge is to provide protection for that destructive current. Insulated bearings do not work well because of the high capacitance in the Al2O3 coating and grounding brushes needs to be kept clean from grease. They also need regular maintenance. The most effective way is to make sure that capacitive currents are kept to a minimum. There are several techniques for this and the suitability is dependent on many variables. It is not possible to give any general rules other than make the capacitive stray currents low.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[AshleyK]

Cheers Skogsgurra.

I'm aware of the insulation issues for the bearings, and we're currently using some SKF hybrid bearings with silicon nitride rollers.

Thanks for the calculation tips, to clarify: If I'm 'calculating for worst case', should I just use the highest speed (1,500 RPM) the bearings will run at?

 

[MikeHalloran]

Just using a high design speed may be unnecessarily conservative.

In the back of the thickest SKF catalog I have is a procedure for calculating bearing life at variable speed, or rather at multiple speeds. It must be online somewhere..

Basically, you factor in the %of the device's year spent within a range of speeds, e.g. 20% at 0..100 rpm, 5% at 100.200 rpm, etc, which is what SKF means by 'binning'.

In my last case, my machine was run by a stepping motor and computer, so the exact duty cycle was known. In your case, you need to gather historical data from extant turbines in the vicinity of yours, or gather a wind history in some detail, and work from there.

You might include some small % spent at an rpm above the design speed, depending on how many overspeed trip/brake mechanisms you have, and how much confidence you have in them.



Mike Halloran
Pembroke Pines, FL, USA

 

[Tmoose]

Is the bearing in a position such that load increases when speed increases?

Where is the "equivalent load" compared to the bearing's published fatigue or endurance load limit?

Sometimes grease life is more important than calculated bearing life, or the sealing is inadequate, and they are responsible for the bearings' demise.

 

[Skogsgurra]

OK, those SiN rollers are safe. But very expensive.

You also have the problem that the capacitive currents will find their way to other bearings if you don't put an insulated coupling (not easy to do in a MW+ plant) in-between.

And, if you do that, you will have lethal voltage on the shaft. So, a couple of grounding brushes are needed. The heavy peak currents through the brushes (peak is often ten to twenty times higher than what you measure with an RMS current clamp) usually destroy slip-rings and eat brushes.

Been there. Done that.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[AshleyK]

MikeHalloran: I have had a good look online and I still can't find any examples of the calculation, but in terms of historical information we do have turbine logs which will be able to tell us how much time is spent running in star, and how much in delta, which will give us two values to find an average of I suppose. It won't be anywhere near as accurate as the SKF binning, though.

Tmoose: The load doesn't increase on the bearing with the speed, as far as I know. The DE floats within a housing and it's coupled with a set of composite discs and a composite coupling to stop any thrust going to/from the generator/gearbox.

Skogsgurra: The DE of the generator has grounding brushes and an insulated coupling.

 

[Skogsgurra]

Looks like youv'e done all that is needed. Have you noticed any sparking under the brushes? Or any tendency to uneven wear on the slip rings? Look for "lobes" when you do a roundness test.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[geesamand]

Use Miner's Rule to combine the effect of the load cases on total life. That's how the SKF "bin" calculation works.

I have used this to work through the effects of varying load magnitude, varying RPM, and varying load directions.

If you're trying to calculate L10 as a close estimate for real-world life, I suggest including life adjustment factors for cleanliness and lubricant viscosity.

 

[electricpete]

The load doesn't increase on the bearing with the speed, as far as I know

I'd think you could express your L10 life in revolutions...
Then integrate your rpm vs time projection to come up with accumulated revolutions vs time....
then look for where that crosses L10 revolutions

Did I miss something?


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

 

[Tmoose]

Where is the "equivalent load" compared to the bearing's published fatigue or endurance load limit?