Bearing Damage on VFD motor 9

[everest]

We have had the bearings on three centifugal pump motors on three different jobs fail. The pumps have been in service ranging 1 year to three months. We had the bearigs on one motor analyzed, they said the races were fluted (I think thats the right word) due to electric discharge. I know this can be a problem with VFDs but was told it was rarely a problem.

Is there any thing in a particular motor design that can increase the potential for this to occur

 

[BigMax]

Hello everest,

I suspect the bearing failures you describe have been caused by common mode high frequency current capacitively coupled from the motor stator - to rotor - to ground via the bearings.

This is an unfortunate and occasional consequence when VSDs are used. At 50/60Hz, these capacitive paths within a motor can be considered to be insulators with negligable current flow. As you may be aware, the output of a VSD is not simply a smooth 50/60Hz sine waveform, but also includes some high frequency (Mhz) content. This can result in current flow through motor bearings to ground.

Solutions include the use of VSD output reactors and filters to reduce the high frequency content.

The above is a very simplistic description, see

http://www.reliance.com/pdf/white_papers/Inverter_Driven_Indct_Mtrs_Shft_Brngs.pdf

for a more detailed technical description (no connection to me or affiliation)

Hope this helps.

Cheers!

BigMax

 

[electricpete]

also insulated ball bearings are becoming more widely available in smaller sizes.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[Skogsgurra]

everest,

If you have fluting, then you almost surely have EDM in the bearings. I have done quite a lot of measurement on motors with this kind of problems and it is quite easy to see what is going on. Do this:

Use a portable (digital) oscilloscope with 50+ MHz BW. Set the time-base to 1 microsecond/division and vertical sensitivity to 2 or 5 V/div, Auto-trigger.

Connect GND to the motor frame and touch the shaft with a carbon brush (even a screwdriver will work) connected to the Y input of the scope. (Do not hurt yourself! Watch out for keyway and bolts on the coupling!).

The scope screen will show a lot of noisy signals that are the result of inductively and capacitively coupled voltages. Most of the time, the voltage is between 5 and 15 V and sometimes you will see very sharp voltage drops going from typically 10 V down to 0 V and with a fall time around 50 nanoseconds. Switch to Single Sweep and try to catch one of these fast voltage drops.

This sharp voltage drop is when the rotor capacitance is discharged through the bearing and the discharge causes Electrical Discharge Machining (EDM), which erodes the bearing surfaces and eventually causes fluting (looks like the wash-board pattern on a dirt road).

I am right now developing a circuit that triggers on this signal and counts the number of PPS (Pusles Per Second) with respect to the V/microsecond level. There are some subtle problems with this (do not count sharp edges that do not go to zero, etc) but there is a working prototype on my bench at this moment. I have found that less than 100 PPS at 200 V/us and <1000 PPS @ 100 V/us usually doesn't cause any damage while >1000 PPS @ 200 V/us does so within a year. More than 4000 PPS @ >200 V/us will ruin a 6514 bearing in less than 6 month.

I have quite a few reports on this and I will be happy to send a copy if you e-mail me (EDM at gke.org replace "at" with @)

 

[BJC]

everest
Try searching this form for shaft grounding. There was a thread on the subject within the last year.
This system seems to work very well for grounding the shaft of motors.

http://www.shaftgrounding.com/index_1.html

Their kits may seems expensive but they work very well.

The paper on "Bearing Damage Due to Electrical Damage" is very good.
Allen Bradley has a couple of papers on the subject. You can find them by snooping around on their website. IF I run across them I'll post them.

 

[Skogsgurra]

BJC,

SGS are good for small motors like ventilation in clean rooms etc where the coupling is predominantly capacitive. I have used them (they were in Seatle at that time) and they worked very well.

They will not work well on larger motors (200+ kW) where the coupling is mostly inductive. In fact, putting a shaft grounding "brush" on the ND end usually worsens things for the D end bearing. An insulated D end bearing does not always help in such a situation since the HF current easily passes through the Al2O3 (yes, Alumina, same as used in electrolytic capacitors) insulation.

A common mode filter (ferrite or amorphous rings around phase leads) is a simple and effective means of coping with most EDM problems in larger machines.

 

[peebee]

Regarding electricpete's suggestion -- ceramic bearings are often specified for this reason with VFD motors -- BUT -- beware that the ground currents may now try to pass through your pump bearings rather than the motor bearings -- you will have protected the motor but may have inadvertantly shifted the problem to another location.

I'd recommend a shaft grounding device rather than trying to insulate the shaft.

There was a great article on bearing fluting in the IEEE IAS magazine within the last 2 years or so.

 

[thinker]

This is the link to publications on bearings problem:

http://www.ab.com/drives/techpapers/ieee/ieee.html

 

[DickDV]

In my experience and that includes some motors over 1000 hp, insulated bearings are the first and easiest solution. In the case of coupled shafts to pumps, the coupling may also have to be insulated. I have never seen an insulated bearing not cure this problem but, as others have suggested, there may easily be exceptions that I have not seen.

The other reliable solution is a grounding brush on the shaft. On larger motors, this is more often the method of choice due simply to cost. Insulated bearings or insulated bushings for standard bearings get pretty expensive in the larger sizes.

I have never seen any reliable way of predicting when shaft current problems will occur. It is more common in larger motors but, occasionally, with no apparent explanation, it pops up in smaller motors, say, under 50hp.

I have tried motor lead reactors in a couple of cases and have had no success with them.

 

[corrosivo]

Everest

I agree with DickDV; this type of arrangement have been used for decades in the gas and steam turbines multimegawatt generators.
You need a grounding brush riding on the shaft to give the shaft induced voltages safe path to ground, what is happening is that an enough high potential develops in the rotor and this voltage buildup is breaking down the oil film in the bearing. The damage is done when the spark discharges this potential to ground in the races of the bearings, with the time it will damage the highly polished surface of the race.
Generally only one bearing needs to be insulated, this is done to prevent induced shaft current generated by a magnetized shaft to have a closed loop thu both bearings.
Both solutions will take care of most types induced voltages problems.
The reason for these voltages develop in the shaft can be a magnetized shaft, static electricity buildup, high frequency inductive or capacitive coupling etc.

This site: sohreturbo.com have lots of pictures about this type of damage on the bearings and thechnical info. Regards

 

[ScottyUK]

While agreeing with DickDV and corrosivo's comments about the use of shaft grounding brushes on large machines, for the technique to be effective on a machines exhibiting problems due to HF currents orginating from a VFD it is important that the grounding brush has a low-impedance return to the earth used by the drive, otherwise the HF current will follow an easier path which may be through the bearing assembly.

In my understanding - incomplete as it often is - the shaft grounding brush on very large machines is primarily designed to prevent windage-induced static voltage buildup, and the insulated bearing or bearing pedestal is to prevent circulation of LF shaft currents induced by imperfections in the machine's magnetic circuit. The considerations are slightly different if the same principals are applied to minimising HF currents because shunt capacitance and series inductance become more significant at HF.




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

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

 

[Skogsgurra]

Yes Scotty,

One must differentiate between the "classic" bearing current problems and the new ones. The classic mechanism was/is that magnetic asymetries produced a net flux change that induced a voltage in the shaft. Problems of this kind seldom occurred in small and low speed machines, but were more or less the rule in large turbo generators running at 3000 and 3600 RPM. The induced voltage was always low frequency - very often only fundamental - and simple insulation schemes worked well. The voltage could be measured with an ordinary moving iron voltmeter (moving coil with rectifier has a threshold at around 0,5 V which makes it less suitable) and the recommended shaft voltage was <1 V.

The new bearing current problems are caused by the HF components in the PWM voltage and since there are four main failure mechanisms, there is not one remedy - but several, that work in different ways. (This will be rather lengthy, I'm afraid):

1 Induced shaft voltage. Shaft ends move electrically in opposite directions due to asymetric currents in the stator winding. The rotor voltage is "buffered" by the rotor/stator capacitance so that one insulated bearing has no effect on the voltage at the other end. Damages will still occur in the non-insulated bearing. The reason is that we do not have a circulating current in this case. This kind of bearing current is common in machines with shaft heights >270 mm. Sometimes even smaller.

2 Capacitively coupled voltage. The whole rotor moves "common mode". I.e. both shaft ends move electrically in the same direction. Discharge happens at about 10 V and produces EDM. This kind of damage is often seen in small machines and the reason is that the ratio between capacitance from stator winding to rotor and rotor to stator frame is larger in small machines than in large machines, so the voltage division ratio can be as high as 5% in real small motors while is is a fraction of a percent in bigger ones.

3 Frame voltage. This is when there is a long motor cable and the motor is not tightly connected to the driven machinery. A paper machine drive on a concrete base and a well grounded paper machine frame is a typical case. The HF currents that are coupled from stator winding to stator frame try to find a path to ground and since the motor cable ground lead presents a rather high inductivity (if not co-axial) there will be a resultant frame voltage. The currents find other routes to ground - often through bearing and motor shaft to the grounded machine frame.

4 "Non-electric" cases. Usually when the motor shaft is charged by a rubber conveyor band, oil in a pump and several other charging mechanisms. Even a fast moving fan blade can accumulate enough charge for damages to occur.

Mitigation techniques include shaft grounding, insulated bearings, hybrid (ceramic balls) bearings, common mode (ferrite rings) filters, shaft grounding through coupling and load (centrifugal pumps can sometimes function as shaft grounding devices). Isolated shafts or couplings are effective in some cases (3 above), but detrimental in other cases. An inverter with a sine-wave output is a good solution and sine filters also do a good job.

I have been working with this class of problems in steel works, paper mills and power generation plants for several years and I can very well understand the confusion that exists when one tries to apply "One size fits all" solution to these problems. They have to be carefully analysed before an antidote is applied.

 

[Skogsgurra]

Sorry, I forgot.

This is a link to a presentation I made a little more than a year ago. It is in Swedish - so most of you will have some difficulty understanding the text - but I think that the pictures explain the main failure mechanisms.

http://www.gke.org/nfo/Sunnepresentation EDM 2003 a4.pdf

 

[DickDV]

skogsqurra, that's a more thorough explanation of shaft currents and bearing problems than I have ever seen. While I have not seen many of the problems you mention, they seem plausible under the right circumstances.

Thanks for the perspective. Now I've got even more to worry about than before!

 

[Skogsgurra]

Sorry about that, Dick!

 

[corrosivo]

Everst
You have a mechanical background, I do not know how familiar you might be with these solutions we are suggesting to you, but here are some additional comments.
I agree with skogsgurra explanation, some type of insulation must be provided to the bearing (Ceramic, Insulated pedestal etc.)to prevent the spraks in the bearing races and also provide a safe bypass path to the motor frame, the brush riding on the shaft will provide a low impedance path to ground, use a high quality brush. This will cure your problem if it falls in any of the 3 categories(classic)explained by skogsgurra.
For the HF case, I suggest you cure the problem at the source not in the motor, if the cause of the bearing failures are High Frequecy harmonics voltages induced in the shaft.
Do not let the HF harmonics leave the VFD cabinet. Use shielded reactor filters in the phase leads and a bypass resistor/capacitors to ground in each lead of the power circuit outgoing and incomming the cabinet. Keep all leads short and use stranded wire, do not use the cabinet as ground path to the HF currents to its source. Make sure the drive have a low impedance to the ground bus. Better yet, ask the OEM for suggestions as these additions to the VFD are technically complex. Or change the drive type as suggested. Regards

 

[everest]

Thanks for the great responses. We are presently working on a project that has around 150 pumps on these drives ranging in Hp from 5 to 200. I became very alarmed when we began to have so many failures on other jobs in the last few months. These pumps have a very critical application.

Thanks again, all of you have been a great help.
everest

 

[ozmosis]

One more weblink to have a look at.

http://www.gambica.org.uk/newsite/pdfs/vsd2.pdf

GAMBICA is the UK trade group for all things under instrumentation, control, automation and laboratory technology. Before departing the UK to Australia I got involved on a regular basis and can recommend the info as factual, non-company biased and uptodate. This particular article was written by my old boss in Siemens in collaboration with ABB guys and Control Technique guys.

 

[Skogsgurra]

sed2developer,

It was some time ago that I reviewed the Gambica papers. I am somewhat surprised that the new paper does not mention capacitive coupling in small motors with floating shaft (like fans etcetera). That problem has popped up during the last years and is now a real headache in many hospitals, clean rooms and also office buildings.

What is your experience with these small drives/motors and ball bearings?

BTW: I am surprised to see that the D.F. (dU/dt filter) cost is so close to that of O.I. (output inductors); 34% and 32% of motor cost. The low cost of C.M.F. is in close agreement with my experience, though.

 

[ozmosis]

skogsgurra
From my experience. The industries you mention (effectively the HVAC domain)seem to be the lowest cost base in devices such as motors and so finding out if a motor has failed due to EDM or simply the motor winding has given up is often very difficult. For small power motors they simply replace them rather than repair them and so close inspection is rarely done.
Certain pump motors I have got involved in that are VSD driven have come under close scrutiny when failing and the direct culprit seems to be the fact the motors are mounted on rubber anti-vibration blocks and also the motor couplings are often plastic/rubber so the bonding of the motor is often suspect. Putting a braided or flat strap from the motor to a good earth does improve matters.
I have also found certain drive manufacturers worse than others. Certain flip-flop technologies in swtiching have caused specific problems in the past. I think this is now in the past but it pays to observe that not all VSD's(VFD's) are the same.
I'm not too sure but the cost of output devices was based on the average. Some dv/dt filters rely on the capacitance effect of the motor cables rather than capacitors in the device and so the cost comes down a fair way. Maybe this is the reason.