Bearing damages from ultrasonic vibrations?

[Comaqc]

Our company manufactures roll heating systems using 25 kHz induction heaters. Because we have measured that our induction roll heating equipment generates ultrasonic vibrations in the roll where induction is applied, we are asking ourselves whether the presence of ultrasonic vibrations on the roll could cause an eventual bearing failure. Some of our customers are experiencing premature roll bearing failures which can be linked to the installation of induction heating being used to raise roll temperature. The presence of 25 kHz electrical current through the bearings has been confirmed, and bearings analysis by manufacturers has established that electrical current caused the failures. However, I remain sceptical of the root cause of the problem for the following reasons.

The current version of our induction roll heating systems has been in operation on 40 systems beginning in 2002. No bearing problems have manifested themselves until the recent installation of twelve identical induction systems on paper machine calenders. Because the bearing problems are evidently linked to the installation of induction, we went back and measured the levels of bearing currents at sites which have been running for as many as 5 years without problems. We found the level of electrical current to be of the same magnitude on systems installed earlier as on newer systems linked to bearing failures. Our measurements indicated that the levels of bearing currents are the same regardless whether bearing problems are present.

In our search for an explanation for this phenomenon, we measured the level of mechanical vibrations on the roll, the bearing housings and the beam supporting our induction heating workcoils and discovered the presence of the same frequencies that are used for induction heating on a pilot roll in our lab. Thus the roll vibrates at 25 kHz, in sync with the electrical current flowing through the workcoils. Because the systems causing bearing failures all used the same workcoil mounting to their supporting beam and to the machine frame, we confirmed that the systems with different workcoil to machine mounting showed a level of ultrasonic vibrations about 10 times lower.

My question is this: could the presence of ultrasonic vibrations on the roll shaft initiate mechanical damage leading up to an eventual bearing destruction by the presence of electrical current?

 

[MikeHalloran]

Electrical current alone can kill a bearing, no ultrasonics needed.

As for ultrasonic vibrations, they alone could also kill a bearing, no electrical current needed.

Given proper installation and care, rolling element bearings eventually fail from fatigue, at a time that's statistically predictable for a given population of bearings.. basically a fixed number of ball passes over a given point at a given load. But the statistics represent decades of experience with machines that don't intentionally vibrate at high frequencies. Your machines induce a fatigue cycle with every cycle of the ultrasonics, which happens much faster than the likely ball pass frequency.

In the older installation, you increased the number of ball passes beyond what a non- ultrasonic installation would see, and got away with it. In the newer installation, you've also increased the load applied (by the higher vibration level), and didn't get away with it.

In the fine print in the thick catalogue, e.g. SKF gives you a mathematical way to predict the life of a bearing that's subject to varying loads. You might be able to extend that to approximate the loading situation with ultrasonics, or they may well have done it already somewhere.

In the meantime, it seems pretty clear that you need to reduce the vibration level or squeeze in a bearing that will take more abuse.



Mike Halloran
Pembroke Pines, FL, USA

 

[Comaqc]

Mike,

Gees! Thanks a million for the clear explanations. You are telling me what I have been desperately waiting to hear from bearing experts. We have been swimming upstream during the last 6 months, unsuccessfully trying to get the attention of the end user, the system integrator and the bearing manufacturer regarding what we believed was an high frequency vibration problem. It was only when we traveled to other sites where no bearing problems had occurred since the installation of our systems and measured the same level of electrical current flowing through the bearings but no history of bearing failures for the last 4 years that we focused less on bearing current and more more on vibrations induced by our induction system.

In spite of the evidence that induction did not cause problems when ultrasonic vibrations were at a low level on the roll, our customers and partners have been relying only on the bearing manufacturers report and suggestions that bearing currents are the root cause of the bearing problems.

Do you know of an organization or individual we could hire to study our present bearing problem and propose solutions. We need to explain technically what is causing the premature failure of paper machine calender roll bearings.

 

[MikeHalloran]

I'm guessing the bearing manufacturer's laboratory was originally given failed bearings to examine, with little or no supplemental information about the bearings' environment. They would certainly find evidence of electrical currents having been present. Said evidence would also be found to a similar degree on bearings that had survived for the same time in older machines of different design, but the lab would have had no reason to examine bearings that hadn't failed, or even to suspect their existence and ask for them. So they might logically conclude that the first possible cause they found was the most likely cause.

Note that in either case, the bearing manufacturer didn't screw up, and their lab has no particular incentive to investigate beyond that finding.

All I can suggest is that you give all the field measurements of current and vibration and demonstrated service life by site to the bearing manufacturer, and ask if they'd like to amend their opinion, or acknowledge the possibility of a cause other than the one they first postulated, given the additional data.

At this point, it may not matter anyway. People are already pointing fingers, and lawyers will soon be there to follow them. What's at stake now is not the technical issue of what caused the problem, but the more important issue of who is going to pay to fix it.

This would be a good time to engage someone to design an affordable and retrofittable modification to reduce the vibration level seen at the bearing, even if you wouldn't ordinarily get involved in design of that part of the machine.

Even if you can prove that you're right on paper, you may have to fix one machine at your expense, and bet (and pray) that it doesn't suffer a failed bearing for some negotiated time after the retrofit. Then you and the machinery builder get to share the expense of the remaining retrofits, in proportions based on the outcome of the 'test', and on how intimidating each of your lawyers can be...



Mike Halloran
Pembroke Pines, FL, USA

 

[Skogsgurra]

What is important is how you measured the bearing currents. I have dealt with this kind of problem for about fifteen years and the most efficient way of detecting and measuring bearing "currents" are to measure voltage across the bearings instead.

Then, it is important to use enough bandwidth. Voltmeters shall not be used. An oscilloscope having at least 20 MHz, preferably more, shall be used. Look for edges going from 5 - 15 V to zero in 20 - 100 nanoseconds. A short explanation can be found at

http://www.gke.org/beppe/files/Bearing currents on one page 1.pdf

Sorry for the commercial at the end. It could help, actually.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[unclesyd]

I don't have all the details and can't get to the report as the site's library has closed down.

We had a similar problem on some high speed rolls that were induction heated. We had some rolls that had been running several years and some that were failing within 6 a month period. I don't have type or size of these bearings, but they were on the order 2x1/2". After a long investigation and a lot of name calling the problem was traced to a change in the lubricant. The lubricant was changed due to concern over price based on the advice of a salesman that had one equal to or better than the original lubricant, dumb.

I have to echo the post by Skogsgurra about the use of an oscilloscope to measure voltage across a bearing. In our case it was assumed that there was no current across the bearings until someone use an oscilloscope.

 

[Comaqc]

Since bearing problems first occurred about 1 1/2 year ago, we have been using Rogowski coils with oscilloscope to measure shaft currents. The Rogowski coil wraps around the roll shaft and accurately measures the current circulating inside its loop. We have found it not to be affected by external magnetic fields. Rogowski coils come in various length to accommodate shaft diameters. An example of the current measurements is attached. The peak value of current is around 80 amps. Current waveforms reflect the current which circulates in the workcoils, which are maintained 3-6 mm away from the roll face.

We have not found a reliable method to measure shaft voltage. Because induction is only applied when the machine has reached full operating speed, we have been using a broom stick with a # 8 AWG stranded wire cable attached to it. We brought the broomstick in contact with the rotating roll face. Measured voltages were of the order of 4 volts peak to peak, but the waveforms were noisy.

As mentioned in my earlier post, the current levels were the same on systems associated with bearing failures and the ones running since 2002 without evidence of bearing failures.

 

[Tmoose]

The various brg manufacturers have failure analysis guides that show the various damages and their causes.

FAG has a good one. It's probably still on their website someplace.
I just downloaded a copy here.

http://rs17.rapidshare.com/files/21561688/fag_brg_fail_greatYET.pdf

If the bearing is rotating it may be possible to establish EHD lubrication, which has terrific load capacity and prevents metal to metal contact.
I'd check to be sure that the lube operating conditions are Kappa > 2. The viscosity of the oil (base oil if grease) and temp may be the only things you can control.

 

[Skogsgurra]

If you are measuring 80 A pk, you should ask yourself where that current is going.

It cannot circulate throgh air. Not at 25 kHz. And since the shaft is rotating, it has to be running in bearings. The only conclusion I can draw is that the current must be going through the bearings. Or that the measurements are wrong.

If we exclude the latter, there is a current with around 50 A RMS going through the bearings. That is a lot more than is good for the bearings.



Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Comaqc]

Gunnar,

We believe that our bearing current measurements are reliable. We have calibrated our Rogowski coils with a stationary wire inside the loop at 25 kHz.

I agree with you that 80 Amps sounds awfully high. The RMS value is closer to 15 amps because our induction power supplies use PDM mode switching to control output power.
I agree that this current has to flow through the bearing. We also reasoned that the only way for such a high current level to flow through the bearing is if there is continuous metal to metal contact.

The dilemma that we have when agreeing that elevated 25 kHz current causes bearing failures is that we have systems in operation since 2002, (about 80 bearings) where this current level has been measured and yet no failure are recorded. Thus our original question regarding whether ultrasonic vibrations could at least initiate the bearing destruction.

 

[Skogsgurra]

Yes Comaqc. That is very possible.

I have worked with calanders and also had bearing failures in the stack. But not from induction heating (I assume you control moisture profile or thickness profile with your equipment - or do you do all of the heating with 25 kHz?). I had static electricity which discharged through the bearings. Fitted one single grounding brush and got rid of it.

If you have an induced current, you will either need two brushes, one at each end. Or better, an insulated bearing. The SKF Insocoat comes in standard sizes and will work well at 25 kHz. But do not trust them at higher frequencies. The capacitance of a large DGBB's insulation is up in the tens of nanofarads region. A clean short circuit for MHz signals.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Comaqc]

Gunnar,

You have hit it right on with our induction heating application.
We have installed two induction heaters per supercalender on 6 supercalenders, for a total of 12 induction beams on 12 rolls (24 bearings). Roll #3 at the bottom of the stack uses 120 sixty mm wide zones at 4 kW each. Thus this roll has 480 kW of 25 kHz induction heating power installed on it. Another induction beam is installed on roll # 8 higher up the calender stack. This one uses 120 mm wide zones at 6 kW; same installed heating capacity as on the bottom roll. Two rolls are heated per supercalender to heat both sides of the paper web. Induction is used to primarily raise roll temperature to increase paper gloss. Ten percent of the power is reserved for caliper control.

A pair of SKF Insocoat bearings have been installed on Roll #8 last June. Bearing fluting started 2 weeks afterward. In a different mill, four rolls were fitted with aluminum oxide insulated bearing housings. Those roll also developed bearing fluting shortly after the insulated housings installation. On all the rolls fitted with Insulated bearing or insulated housings, bearing current was measured at less than 0.5 amps compared to 80 amps before. Again, this made us suspicious that bearing current by itself is causing all the problems because insulation practically removed all current flow, yet bearings keep failing.

Grounding brushes were installed initially on all rolls fitted with induction, but because of the initial high current level, current reduction was less than 20% with it. It appears that the bearings lasting for the longest time are on two super calenders in yet another mill where the lubricant grease has been replaced with conductive grease. Bearings have been in operation for 5 months on those rolls without problems so far.

So, I am back to the ultrasonic vibrations hypothesis.

Best regards,

René Larivé

 

[unclesyd]

Do you have access to an Ultrasonic Bearing testing instrument?

It would be intersting to get the signature of each bearing to compare the good and the ugly.

 

[Skogsgurra]

I am surprised that the grounding brushes didn't help better than that. It is probably because of the low bearing voltage. I assume that you did measure after the brush and before the bearing, counting in "energy flow direction"

A side note: Half a megawatt of ultrasonic energy is bound to produce some side effects. That's for sure. I assume that dogs and cats wouldn't like staying close to that building. Even if the human ear cannot "hear" 25 kHz, I wonder if the sheer power of it could have any effects on hearing. I think it is something worth considering.

Re the effects on the bearings. All calenders I know have either every second roll soft or clad with a soft material. That would effectively absorb any high-frequency vibrations. Especially radial ones. If you have measured the vibrations, have you noticed any difference in vibration amplitude as a function of stack pressure?

My gut feeling is that bearing currents are not entirely ruled out. On the other hand, the insulation capacitance in the Insocoat bearing represents something like 100 - 500 ohms impedance at 25 kHz. Pushing 15 A RMS of 25 kHz current through that impedance would need more than 1000 volts. So, that can also be ruled out since you got fluting also in insulated bearings.

Two weeks before fluting is rather extreme. That, I have only seen in large smoke stack motors running off 690 V VFDs with high switching frequency. But never in rolls in a calender stack. A question pops up: Are the rolls by any chance driven by separate motors? I know, I shouldn't ask that kind of question. But, who knows? Stranger things have been seen...

You did indicate a "steel/steel" running condition (low bearing voltage). That is not good per se and could cause damage to the bearings, admittedly not fluting. I am not quite sure how that could be. Are there high radial forces transmitted through the bearings? I have a very vague picture of how roll pressure is applied, always thought it was on top roll and down through the stack. So, it shouldn't affect all bearings.

Sorry for turning so many stones. But this is really interesting and I am very eager to learn what you will eventually find. It would make a good addition to my set of "case stories".

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Comaqc]

Gunnar,

The ultrasound energy is a very, very minor portion of the energy generated by the induction heating system. The energy applied to the roll is in the form of heat from eddy currents. My old dog, which was probably deaf, would sleep beneath the wiring going from an induction power supply to an experimental workcoil when I came to work to test circuit operation on weekends.

(see attachment)

Above is a photo of an uninstalled roll induction heating system. The induction coils on the front are positioned 3-6 mm away from the roll. The 25 kHz power supplies (1 per coil) are housed in the rear of the beam. The beam ends are attached directly on the bearing housings mounting plates because the beam has to travel with the roll as the super calender nips open up. Thus the weight of this beam, about 5 tons, is totally supported by the bearing housing. It should be noted that the bearings do not support the weight of the roll because the stack of rolls support each others. Below is an example of a typical installation. Only roll #3 is motor driven through a gear reducer.
(see atachement)

The vibration levels and frequencies I mention in earlier messages were measured on a laboratory calender, equipped with a narrow but otherwise identical induction system. We were not allowed to take vibrations measurements on the super for safety reasons. Thus, I cannot comment whether vibration levels are related to calender stack load. We plan to instrument a calender stack with accelerometers on bearing housings and induction beam during the next machine outage in late November. Again, as noted earlier, we have not been able to get the customer attention away from bearing current as the cause of the failures. It’s been a struggle for us to obtain permission to measure vibration levels.

My reference to metal to metal contact is not based on facts. I use it only to explain to myself how such a high level of current can circulate on the roll shaft.

In response to unclesyd, we do not have access to an Ultrasonic Bearing testing instrument. From the sound of the comments I am hearing on this forum, buying or renting one would be a good investment.

René Larivé

 

[Skogsgurra]

A normal calender stack, as I see it.

Hmm...

Et votre chien - est'il mort maintenant?

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[unclesyd]

Skogsgurra,
I just checked with the electrical group at the plant about measuring and dissipating electrical charges on a large, 25,000 Hp, turbine, air compressor, and recovery unit that had a bad problem with arcing bearings. After the installation of a shaft grounding system using brushes of unknown material there was no change in the bearing damage. It wasn't until the brushes were changed to gold did we eliminate the problem. Also it wasn't until installing the new brushes were the able to measure voltage and current.
As you posted that you were surprised a the small effect of change the grounding system made, would changing the type of shaft grounding system help.

 

[unclesyd]

Comaqc,
I definately would try to do a Ultrasonic analysis on the bearings. You could either rent one or get one on a trial basis. There might be an inspection company that has a UT meter and an operator well versed in the art of using it.

Skogsgurra,
I just checked with the electrical group at the plant about measuring and dissipating electrical charges on a large, 25,000 Hp, turbine, air compressor, and recovery unit that had a bad problem with arcing bearings. After the installation of a shaft grounding system using brushes of unknown material there was no change in the bearing damage. It wasn't until the brushes were changed to gold did we eliminate the problem. Also it wasn't until installing the new brushes were the able to measure voltage and current.
As you posted that you were surprised a the small effect of change the grounding system made, would changing the type of shaft grounding system help.

 

[Skogsgurra]

Yes unclesyd, I remember now that we had brushes installed on a large pump and everything looked well. When I got back on site after a few weeks, I noticed that shaft voltage was like before (high). More pressure on the brush didn't help. It was only after cleaning the shaft that we got the voltage down again.

We now have two brushes on that shaft. One rather hard and abrasive that we think cleans the shaft and hinders build-up of an insulating layer plus the ordinary, bronze-carbon brush.

Gold seems very expensive. Is it really Au? Solid? Wow!

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[unclesyd]

Comaqc,

This may seem like a dumb question but here goes?

Has any one cheked the bearings for shaft position after installing the heaters?