OT: Motor Predictive Maintenance

I was recently tasked with researching a predictive maintenance program for motors, and I'm starting to be convinced that the ol' megg it out and make sure it's above 1.48 Mohms is not really very effective (especially when we can't trend the data as most motors don't ever cool down to ambient temperatures when tested). It seems like a good multi-tester that can preform other tests beyond high voltage ground fault tests, like surge testing is needed for to really be able to predict electrical failures.

Anyway, I started researching some of the equipment available for this application and have been somewhat put off by bulky equipment and high prices. The only vendor that looks promising I've found so far is All-Test Pro ( http://alltestpro.com ).

You fellas always have some good input, so I was hoping one of you might have some recommendations for other vendors of this type of equipment.

The package that we are using (I can't remember the name of it) includes vibration testing of bearings and driven loads at the take-off.

The information is recorded digitally and displayed in a nice graphic form. You can maintain a log of periodic tests and then make over-lay comparisons. The displayed recordings can be subjected to frequency filtering so that you can isolate various aspects.

That information is quite useful in detecting growing problems while they are quite small. If you see a trend developing you can schedule repairs/replacements at a more convenient time.

It also helps if you have one of those spendy Infrared cameras for looking at the mags and distribution panels.

What size motors are you looking at? What type of environment are the motors in? What is the importance level of these motors? All of these are relevant questions. I would not spend much time on 30HP and less motors unless they are vital to the process. If your tester requires you to take the motor off line and disconnect the leads and then run the test, I would have to look at that amount of time versus just replacing the motor on a regular basis or waiting for a failure.Preventative and predictive maintenace on motors is still not an exact science from all that I've looked at. I have seen all kinds of companies that sell all kinds of systems. I haven't been impressed with many of them. At one point a company came in (when I worked at a gov site) and tested all of our 4160 volt motors. When I looked at the "engineer" who was running the test and their test equipment, I noticed that it was only a 250 VDC meggar. He hooked the leads to the motor at the main breaker (no starter) and had them on for only 1 minute. From that he gave all kinds of stuff like DAR, PI, waveform, etc. The only problem was that to get some of these numbers (like a Polarization Index - PI) the convention is the 10 minute meggar reading divided by the 1 minute meggar reading. When I question them on this, first they said that the computer model extrapolated the results then the changed their mind and basically said that their PI is not the same as other PI. Also when I question how effective a 250 v meggar is on 5 kv insulation, I was told that their computer model extrapolated the results. Unfortunately, I threw out the BS flag on that one. It turned into a big argument and I was labeled a troublemaker from that point on. They made lots of money off the taxpayers, recommended that 25% of the motors be tested again in a year and left. I was pretty sure we had just be sold snake oil.

I guess my extremely long rant on this is to let you know that some of what you see out there is really no better that putting on a meggar and looking for 1 megohm per kv plus 1 megohm (hence the 1.48 on a 480 volt motor). Look at vibration and infrared (even though it is rarely cost effective to actually purchase this equipment since many companies offer these services extremely reasonably priced). Also look at how you are actually using the motors. A few minutes spent on a good design goes a long way (like on VFD motors and cable length)

Also, higher voltage motors lend themselves to much more testing than 480 volt motors (ie doble PF testing and dc hipot testing). I have had much better luck on these than the 480 volt motors.

I know that I really didn't answer your question, but if you give a little more information and I will speak up on my past experiences (even though they may not be the best).

We do vibration and infrared on motors larger than 25 H.P., we use an outside service and the pricing is reasonable.

Downtime cost money, so we do not do any off line testing. We basically run all motors until they fail.
If testing indicates failure is imminent we will try to swap out the motor before hand, but can't always get the line to shutdown.

By infared I am assuming people are talking about thermography. This is the best way I have found to predict motor failure, if something is going to fail in a motor chances are it will get hot first.

Ir Thermography is one thing we are going to implement on our motors and control panels. It's come in quite handy so far for finding loose terminals in some of the poorly built control panels We have.

Cheers,

Lee

I agree with the vibration monitoring route. We first used this technique about 20 years ago and the results were astonishing.

Prior to that the test was a hand on the motor and a screw driver to the ear.

With the vibration trends we were able to take motors out of service and repair them before catastrophic failure, even though they appeared fine at the time.

I would not worry about vendors etc, what you need to do first is learn what is involved if you decide to go with PdM (predictive maintenance).

This link offers some reasons for motor failure:
http://www.reliabilityweb.com/art04/motor_pdm.htm

That site in general offers an abundance of information and tutorials: http://www.reliabilityweb.com/tutorials.htm

Another thing to consider is CMMS; maintenance management system. The problem with these is that they have a high failure rate for various reasons. The primary reason is that it requires full time talent just for the system.
http://www.cmmscity.com/

The problem with PdM is that many companies are not willing to have someone inhouse that just tests equipment, maintains inventory, and does paperwork. It is that reason that sometimes it is easier to outsource motor analysis.

There is an abundance of work and time involved with developing any form of predictive maintenance.

Any predictive maintenance is going to have to be cost justified. We have CMMS, and Ron is correct, it does require a fulltime person to administer, which we have. CMMS does nothing in the way of predictive maintenance though. What the system will do for us is generate PM work orders at specified intervals that we set to remind us when things are due for checkup or replacement. Again, that is determined by us. With some more predictive maintenance monitoring systems, I know that there is the capability to have the PDM generate a work order when it senses bearing going bad, or something along those lines. In a nutshell, you need to determine how much downtime a year is spent replacing motors, and which devices caused this downtime. You then need to estimate how much it will cost to implement a Predictive Maintenance system and run it. Usually, it has been my experience that management will reject anything that does not have a 2 year payoff, but you never know. Good luck.


Russ

Thanks for all the tips guys. We do have a preventative maintenance program in place and schedule all PM's through a maintenance management system. We're just real light when it comes to motors. Especially when it comes to any type of predictive maintenance. The motors are mostly <5HP, but downtime hurts none the less...

For 5hp and below, you're almost better off to determine the mean time between failure, derive the worst case and replace them before then! Simplistic, I know, but I'm guessing these are low end induction motors?

Mean Time Betwen Failures is not, by any means, deterministic... it is only a wild a$$ guess as to how long something might, heavy-stress on "might", last.

Only real-time (current) monitoring can give real-time (current) conditions.

And it is only by comparing previous readings to a current reading that one has even the slightest chance of predicticting an emminent failure.

HISTORY IS EVERYTHING!

All else is... Smoke & Mirrors!

However... there is a concept of "Smoke & Mirrors" that is History based...

Terry,

I stated that the MTBF solution was simplistic in my post. The only reason I suggested it was cost and only after it was stated that these are 5hp motors. MTBF is not so much a wag if you have years of supporting data (motor replacement records) on a particular production line. How much money do you want to spend on these little motors? Especially if it involves disconnecting the T leads at the peckerhead.. at some point the time spent by the maintenance tech will pay for a new motor. On some production lines, you are way ahead to replace these little motors on a schedule than to have unscheduled downtime. That's all.

At 5hp and less, I would implement a good on-site spares program and forget the predictive program except for vibration analysis. I would contract out for the vibration analysis.

That will end up costing a lot less than a predictive program in my opinion.

I keep coming back to this guy:

http://alltestpro.com/pdf/AT PRO 31_.pdf

I haven't called on a price yet, but they say it's comprable to the price of a good megohmmeter. Even if the tracking software was ridiculous, one critical shutdown avoided during high volume production and the thing would pay for itself. We have volume spikes where we need every last bit of capacity that we can squeeze out, and a good replacement stock don't cut it when even a half an hour of downtime has production managers sweating bullets...

I think I'm going to bite the bullet and test this model and if it's successful, move to the full predictive maintenance kit they offer.