Skip to main content

Replies sorted oldest to newest


A very good question which requires participation from maximum maintenance engineers disclosing the various techniques used by them for predictive maintenance of motors.

Following online techniques are used in our plant for determining the motor condition

1.Motor current measurement and logging every
15 days
2.Motor Temperature measurement and logging
every 15 days

The above logged data are compared with the previous ones and also with the rated specs.If found beyond limits,motor is taken for overhauling/replacement on available opportunity.

Here's my take: I'll start a list. Others can add.

* Vibration
* Lube oil analysis (may or may not be an on-line sample depending on oil reservoir configuratino).
* Current signature analysis - Use in conjunction with vibration analysis (some people find or confirm misalignment, bearing problems, system oscillations). For us we use it primarily on large motors to monitor for rotor problems.
* Large Critical motors should have winding temperatures monitored by plant computer and sleeve sliding bearing temperatures monitored by plant computer.
* Consider thermography is you don't have temperature monitoring - depends on your preference.
* Current - check magnitude and balance.
* Read winding winding RTD's individually. Leave the highest-reading RTD connected to plant computer input. Investigate unusual difference among RTD's.

* Insulation resistance test / polarization index - gross check for insulation condition and dryness. (IEEE43)
* Winding resistance check - gross test for loose connections or shorted turns (normally shorted turns would
* DC Step voltage test - potentially destructive test which is more sensitive to insulation anomalies than simple insulation resistance / P.I. (IEEE 95.)
* Surge test - potentially destructive test which is more sensitive to weak turn insulation than winding resistance test. Requires expertise to perform. (IEEE 522).
* Insulation power factor = tan delta - another check of insulation condition and dryness. More trendable and sensitive than insulation resistance / P.I. NOT potentially destructive. Equipment is relatively expensive but often is purchased/leased to support transformer testing.
* Check space heaters are on and carrying expected current.

We don't use any of the text boxes from PDMA, BJM, etc. These have many additional tests.

I would add audibile sound analysis to the list. I use sound pressure spectrum at fixed locations (repeat surveys or parametric tests) and pressure gradient measurements (source location) for diagnostic tests. In both cases it is just a different sensor connected to a portable vibration analyzer or multi-channel DAQ system.
Audio observations can be very important, and measurements can offer unique or complimentary machine condition information.


Attached is from my PdM-2006 workshop on motor testing. These are all of the technologies identified in the IEEE P1415 following 11 years of development. The standard has been voted and accepted and should be issued next year.

The presentation (PPT) lists each of the test methods outlined within IEEE P1415 and whether or not they are 'trendable.' We decided, during the standard's development, to outline the test methods instead of the individual collection used by test instrument manufacturers. For instance, a Baker AWA would include Surge PD, resistance testing, polarization index and dielectric absorption, insulation to ground and a hi-potential test. An ALL-TEST IV PRO would include the resistance, impedance and inductive balance, phase angle, current/frequency response and insulation to ground testing.

Draft 13 of the P1415 is available on the IEEE website for a fee (I can't control that).



Let me declare a blatant self-interest up-front as a PD test equipment vendor before adding another test. On-line partial discharge testing of MV & higher motors (> 3 kV) is a good 'condition-based' maintenance test. 'Condition-based' is more appropriate than 'predictive' maintenance as one cannot predict from PD tests when the motor insulation will fail. It does enable you, however, to base your maintenance & off-line testing decisions on the insulation condition rather than on time, operating hrs, # of starts, etc.

Vince Green, p.Eng.
ADWEL International
We have partial discharge monitoring capability for thirty motors ranging in size from 2500 hp to 9000 hp, all 13.8kv voltage. We use partial discharge bus coupler capacitors supplied by a different vendor. For those just getting familiar with the technology, my opinion is that it's not worth it to dwell on the differences among partial discharge monitoring systems from different vendors at this point, but to focus on the strengths and weaknesses of partial discharge monitoring technology as a whole (others are welcome to address differences as they see fit).

We have had the monitoring equipment installed for 8 - 10 years on these machines. We test between quarterly and semi-annually while the motor is running. We trend the 10pps cutoff level "Qm" and evaluate the phase pattern in this trends up The test is an on-line test and we find it very useful.

We found one item by partial discharge which we elected not to correct and it progressed to failure within two months (we identified the failure but plant was not prepared to remove the motor); we found one item and inspected and found severely oil-degraded endwinding jumpers; and we had one motor fail to ground that never showed any PD symptoms prior to the failure. I guess you could call the 2 out of 3 caught by partial discharge technology. Also we have been monitoring for a few years two motors that have tracking two that have surface partial discharge on the interface between semi-con and grading treatments. This seems to be a fairly benign partial discharge and we gain confidence from the fact that we can see changes in behavior.

We don't use on-line partial discharge on our 4kv motors as it is generally believed that detectable partial discharge only occurs right before failure at this voltage level. There is apparently much less warning time because at this lower voltage level, partial discharge does not occur until the insulation is almost totally breached. Therefore to be useful, you would probably need continuous monitoring at this voltage level. And you would have to be prepared to respond on a moments notice to these indications. In contrast, pd on 13.2 KV machines can be trended and sometimes be effectively be used for long-term motor refurbishment/ rewind planning and prioritzation on the scale of years, depending on the severity of the symptoms.
Last edited by Registered Member
I set up predictive maintenance systems on a regular basis and you have arrowed in on the biggest error that most people make when they set these systems up.

Before you pick up an analyser or any sensor you should make an assessment of which failures are going to happen, identifying the symptoms of those failures and only then should you start deciding which readings to take.

On a standard induction motor there are a limited number of failures that can happen. Of those failures, some are more likely to happen in specific operational scenarios. For example - you are more likely to see broken rotor bars on motors with frequent starts.

Once you have identified which failures are likely to affect your motor THEN you decide which readings to take. I appreciate the several comments posted above and they are all valid. My point is that you will get the best "bang for your buck" if you use your head.

I presented POTENTIAL FAILURE ANALYSIS at the Entek National User Group conference (remember Entek) in Cincinnati in 1997. At that time most people saw this as some form of Rocket Science but we have seen continual improvements in the systems we have set up because of this procedure.

Good luck and just remember to use the posts above sensibly.
In considering how likely is a specific motor to experience rotor bar defects, we prioritize our frequency for current signature analysis (which we use primarily for broken rotor bar monitoring, even though others use it for much more) considering the following factors:

* frequency of motor starts (frequent starting is the worst)
* inertia of the load (high inertia is the worst - primarily fan motors or loads with flywheel such as RCP motors and rod drive MG motors)
* motor speed (2-pole is worst)
* horsepower (high horsepower is the worst)
* on-site failure history (we have only had rotor bar failures on one family of large motors so far).

In addition to inertia, other factors which affect the severity of a given start are voltage (low voltage is worse) and torque loading (high load torque during start is worse). You can roughly roll up the effects of inertia, loading, and voltage into starting time. Motors with a longer starting time have a more severe start than motors with a shorter starting time, and deserve more frequent monitoring, all other things being equal.

These comments apply only to direct on-line start motors. VFD have a much easier start.

A final point on rotor bars, I believe it will typically give a long warning before it completely fails. The most likely scenario for rotor bar degradation will be increasing vibration, possibly increased starting time (which are not desirable but generally do not make the motor unavailable). Unless a rotor bar comes completely out of it's slot or starting time degrades to the point that the relays take the motor out during start, you will likely still have use of the motor for a long period after detection. In contrast, a winding failure removes the motor from service immediately.

These are factors specifically related to rotor bars. Of course in deciding any PM or PDM, you have standard considerations of motor criticality, availability of a spare etc etc.
Last edited by Registered Member

Good question. Actually, there are a number of technologies that are not represented but there were no champions for those tests. The complete standard actually outlines how to troubleshoot and evaluate rotating machines, including flow charts, etc. Many of the technologies mentioned in the standard also include pass/fail recommendations where they do not in other standards.

Some interesting things are happening in the IEEE standards groups, not just including alignments with IEC standards. For instance, IEEE 43 was re-issued as IEEE Std 43-2006 with the understanding that the DEIS Dielectric Materials Subcommittee had to review a number of issues. We did this at the IEEE CEIDP conference in Kansas City during the week of October 16. Some wanted to return the insulation resistance minimum standard back to 1 MegOhm plus 1 MegOhm per kV for all insulation systems. This had to do with one motor manufacturer who had problems meeting that standard for sealed insulation systems.

Also, there were issues with some of the diagrams and write-ups that conflicted with IEEE Std 95.


You are absolutely correct. And, of course, the electric motor should not be held as an isolated system. It is part of a larger system that includes the incoming power (and distribution), controls, motor, coupling, load and process. When selecting technology to provide condition-based monitoring (some also refer to it as PdM, but it is actually a different animal), the correct selection for the types of problems that you are looking for is excellent. Especially when the technology can look at other parts of the system.

That is one of the reasons for the application of such tools as RCM. It involves the review of potential issues and the selection of strategies to identify or mitigate the potential failures based upon logic or history.

By the way, Entek is now Rockwell just like CSI is now Emerson.

To give my point of view on the topic, and can said that we have to consider also the environment of the machine, how is use to work is function and all solllicitation require on it. At that time we can dicide set
1- Acoustic analysis
2- Thermography analysis even thermometry analysis
3- Current analysis
4- Vibration analysis.

Hermann MBAMI
Responsible of Predictive maintenance cell.
Pob: 54 SMS&Methodes ALUCAM Edea, Cameroun
Tel: +237 346 40 24 poste 4679
Fax: +237 346 49 49
Port: +237 765 17 43
e-mail: hermann.mbami-engo/
there are a number of technologies that are not represented but there were no champions for those tests

I am disappointed that the IEEE commitee would only include technologies because they are lobbyed by instrument vendors. How can "Noise" be simply mentioned without suggesting audio sound measurement? I could get on a soap box on this topic, but time is limited!

Another obvious omission to the "technologies" is shaft torque and torsional vibrations. I would expect that the IEEE commitee should be focused on identifying all available and practical technologies for assessing motor and power train condition rather than a few suggested by the loud voices of a few vendors.

My personal opinion is that what the IEEE discusses are by and large the mainstream tests. There are a wealth of other possible monitoring opportunities. But they have to draw the line somewhere.

I'll mention there is one very low-tech cheap partial discharge test you can attempt with the Draeger tube sampling equipment available from the Industrial Safety department of most industrial facilities: an ozone test. Any ozone above the detectable level (I think it's 100 ppb) in the motor discharge airstream is a bad sign. The likelihood of actually finding or confirming a bad machine using the standard ozone test would be 0 for 460vac machines, not much better 4kv machines, and a lot better for 13.2kv machines imo. I have tried it out during a shop ac hi-pot tests and it did detect ozone at about 10 times minimum detectable during that test. I also checked several of my 13.2kv motors during operation and didn't find anything above minimum detectable yet. I think the tremendous dilution created by the operating air-flow (compared to the stationary hi-potted machine) means that even if I had the same level of pd going on in my operating motor as was going on in that hi-potted machine, it would still be below minimum detectable due to the dilution. But thinking about it now, there is a way to improve the minimum sensitivity of the test simply by increasing the air volume drawn through a single tube. One of these days I'll get out and keep drawing air on one of my bad actor motors (indicated by traditional p.d monitoring) until I get an indication, then back-calculate the concentration. At any rate it's easy, quick and cheap to try. The ozone tubes costs about a buck each and are used up each time you do a test. The Draeger hand sample pump (if you don't already have it) costs maybe $100, but once you buy it you can use it over and over again.

Not everyone was an instrument vendor, but they were pretty dominant. One of the areas in the IEEE SA (Standards Authority) that many of us have an issue with. Unfortunately, the traditional end-users are not in attendance (ie: utility engineers, partially due to the after-effects of de-regulation). Luckily, there are quite a few that still attend, but they are the 'die-hards.' With the vendors, they are willing to invest in having engineers in attendance because of the commercial benefit of having their technology included in a standard.

At the end of the IEEE P1415 development, I was not a vendor, so assisted the chair in prying out any 'commercial' bits and editing, in general.

We did have to draw the line on the most common tests performed, and a few unusual ones. Mainly descriptions of each and a few removed. The primary purpose of the document, however, is to provide a process for troubleshooting and performing analysis on AC induction machines. The first section of the document outlines the tests that are commonly performed, some with pass/fail values, and the remainder focuses on the process for troubleshooting and analysis.

If you need such a document, I think you will find the end result quite interesting.

We also did not include such things as using an analog ammeter to check for broken rotor bars, or using an AM radio to detect partial discharge, etc.


Are you talking specifically - Baker, PdMA, AllTest types of systems - leaving out the vibration / oil / thermography end of the analysis?

If so, I am familiar with the PdMA test equipment and do believe that their offline / online testing does allow PdM for some failures with trend plot capability for the various parameters measured.

In order to adequately monitor a motor, both the online and the offline PdMA tests should be performed. Some early models did only the online tests.
In addition to PdMA's online and offline test, there are also:

ALL-TEST Pro - primarily their ALL-TEST IV PRO 2000 (off-line) and ATPOL (online) units are for predictive maintenance.

Baker Instruments has their AWA and Explorer systems. They have built lighter systems for both, as their full sized systems tend to be quite heavy.

Areva has the Empath online system. The software is pretty straight forward and will also analyze DC machines.

Iris Power Engineering has a primary line of Partial Discharge systems down to about 4,000 Volts (constant monitoring) and released their CT Meter, which is a single phase motor current signature analyzer.

Liberty is still out there, somewhere, after being purchased by Crane Engineering (nuke company). I was able to see their system and their permanently mounted system some years back. However, there is no marketing going on that I have seen.

That is it, other than the current signature systems built into some vibration data collectors, so far as motor circuit analysis (MCA), motor current signature analysis (MCSA) and electrical signature analysis (ESA).

Most of the above vendors can be linked through (Institute of Electrical Motor Diagnostics) as they are members.

Hey, so are you, Terry!


Add Reply

Link copied to your clipboard.