Good Morning everyone,

It is great to see some open discussion regarding this important topic. Unfortunately, we ( in North America ) live in a very litigious society. Being right or wrong comes secondary to who has the money to purchase the best lawyers.

Having said that I am sure that a lawyer could look at my posted observations and somehow equate that to unsubstantiated slander.

At the time I posted my observations, I was really hoping that a pile of ALL-Test users ( or other low-voltage testers for that matter ) would come out of the woodwork and state emphatically that my observations were nothing more than a bunch of bad coincidences. I was hoping to see a lot of positive feedback of how the technology works wonderfully.

Remaining anonymous would allow me to then return to the vendor without too much embarrassment to begin negotiations.
I really was looking for the holy grail of motor testers. One or two pieces of equipment to be used in the field to reasonably be able to identify failing machines without such a huge price tag that the return on investment becomes something in the order of 15 years. I realize now that it is not going to happen. Motor test equipment is just too expensive to be used on motors in the sub-100hp range.

Dan Steinwender - 18 years crawling around industrial controls, closed-loop servo systems, and sub-100 hp motors in general
EE Eng. Technologist
Industrial Electrician 442A
Master Electrician ( pending )
ITC Lvl 1 Infrared Thermographer

Good morning,
DanS, I was reading your posting this morning regarding testing on sub-100HP motors. I would urge you to reconsider testing of this type of motor based on the criticality and cost of down time. I will share a story from just a couple of weeks ago, a plant not far from me had low voltage off-line testing done on what they determined where there critical motors. The definition they decided on was 150 HP and above. I recieved a call from them because they had failures on three 60 HP motors in the same application within three days and felt they were getting bad motors. My response was if there have been three failures that quick then more than likely it is not the motors. I went over to look at the situation for them and found out that this 60 HP motor shuts the entire plant down if it fails. Knowing that they had had a vendor in to do testing just a week earlier I asked to see the data on this application and that is when I was told that it wasn't tested because of HP rating. What I found to be the problem was a simple loose connection in a local disconnect that would have easily been seen had they performed the low voltage testing when they did their other so called critical motors. Yes the equipment is expensive but not nearly as much as shutting a plant down. Just to let you know my back ground, I have spent 35 years in the motor repair industry and now represent several predictive maintenance eqipment lines including PdMA. I have several case studies that I will share here in the comming days and weeks as I can get the time to post. But for now I have a 2 hour drive in front of me on icy roads.

When I referred to sub-100 hp motors not being suitable for PdM I guess I should have specified non-mission critical motors. Most motors that I have come across in the sub-100 hp range are workhorses being used on a stand-alone piece of machinery. If they fail then the machine fails without immediate impact on the rest of the plant.

A typical customer would have 40 pieces of machinery each having a dozen important motors. I say important in that these motors if left to fail would result in the loss of the use of ONE piece of machinery. Quite honestly there might be one or two motors at most at a given facility that would force a complete shutdown in the event of failure. Shop air compressors come to mind. Now to confidently be able to offer a program with a reasonable chance of eliminating MOST motor failures would require the purchase of at the very least an IR camera ( for motors you can see ), Vibration Analyzer ( for motors you can not see ), a Surge Tester ( for motors you want to stress test ) and a low-voltage tester ( for motors that you do not want to stress ). The latter two winding testers depending on what side of the fence you sit will still miss some faults as practical experience has shown. A power quality analyzer is an additional helpful tool.

Now that I have invested something in the order of $75k I can still not really ensure my customer that I will find all his motor faults, but hey I can find most of them. This limits me offering a performance guarantee. Okay so now I have about 500 motors to test on a regular basis. Each motor if it fails can cause a loss of something in the order of about $2000 lost revenue per 8 hour shift. To use these test instruments requires that I shut down production and spend likely 4 hours per motor from start to finish doing these comprehensive tests and generate reports. Assuming I ask $100 per hour to perform this service ( hell I get $85 per hour if I just let the darn thing break and then fix it ) then I earn a bonus $15 per hour towards my $75,000 investment in diagnostic tools. I would require 2000 hours to do the entire survey and would be asking $200,000 to do the job. The customer would also incur at least 1000 of those 2000 hours as pure downtime. I would quickly be shown the door.

This strategy clearly does not work. PdM needs to be at the first stage completely unobtrusive to production. IR scanning, Vibration analysis, and Power Quality analyzers in conjunction with traditional techniques need to be used to identify flags which subsequently get verified with intrusive ( shut down the machine ) methods such as winding testers. This will still acrue downtime but it would be kept to a minimum. Labour performing the testing is still very significant.

Before you say it should never take 4 hours to test a motor I would like to add. To do an IR inspection of an entire machine with guarding all around it requires a lot of preparation before and reassembly afterwards. Motors are often under floorplates or up in the air buried in the machine itself. It is not unheard of to take an entire day to do an IR inspection of a dozen motors on a machine and this does not even include the time spent afterwards reviewing images and preparing a report.

I just do not see where most industries can absorb the cost of a predictive maintenance program utilizing this expensive equipment. I service equipment now and owning and using a dozen standard test instruments ( multimeters, meggers, oscilloscopes, tachometers, etc. ) is just part of my repertoire of tools. PdM test equipment is priced so high that the return on investment is unrealistic for most potential users. Everyone loves the idea of being proactive with their machine maintenance until you tell them that they need to spend $400 on each and every motor on the shop floor. Then you drop the bomb on them and tell them that they should probably do it quarterly for the first year.....

I suppose I have officially hi-jacked my own thread since this clearly is not talking about Low-Voltage Testers anymore. I will move this comment to a new thread to be discussed there.

Thanks,

DanS

quote:

Originally posted by Noah:
A better explanation would have been to reference the increase in flux density resulting from the loss of opposing flux from the open bar causing a change in the permeability of the surrounding area not just the air gap.

I am not sure if the explanation would be any better.
From the “not just the airgap” in the quote, it seems, that you are still not convinced that the permeability of the airgap is always equal to 1. However for the sake of scientific accuracy, we have to insist on that. I have said that the science of the low voltage testing is my main concern, and I mean it.
There is an excellent way how to simulate a broken bar. You can just imagine that there is another bar in the same place (where we want to see the broken bar), and it carries an opposing current. The field caused by one lonely bar can be figured out very simply and superimposed on the original “unbroken” field. In case of the most trivial 2-pole motor, the field induced by this fictional bar crosses the airgap in one direction (for example ROTOR TO STATOR) on one side of the bar and in the direction STATOR TO ROTOR on the other side of the bar. Half and half around the circumference of the rotor. For some reason this step change is called swirl as I have learned from the Kliman’s paper. So if we go back to your quote, there cannot be only “increase of the flux due to loss of the opposing flux”, because there must be also “decrease of the flux due to increase of the opposing flux” on the other side of the broken bar.
Correct me if I am wrong, but your explanation seems to follow PdMA paper by McKinnon and Smolleck. (It seems to be an official view lately in explaining the RIC test). When you wrote your paper in 1998 there was no mention of increasing or decreasing flux, no mention of the residual magnetism:
http://www.pdma.com/PDF/Articles/Influence_of_Residual_...nt_of_Inductance.pdf
There are same serious drawbacks to this paper as well. The first one that comes to mind is, that the result of the analysis would depend on the moment when the power was shut down. If the broken bar were at that moment in position that it would not have carried any current anyways, no anomaly in the residual flux would have been established.
Let’s look on the amplitude of this variation. In a 2-pole motor with 40 bars one broken bar would mean variation of the main residual flux of 5 Gauss probably in the neighborhood of 1/20 (0.25 Gauss – just a guess) at the most, mainly much lower (depending on the moment of shut down). Even your LOW VOLTAGE MEASUREMENT will distort those values great deal as is shown in the paper (it changed the residual magnetism from –0.26 Gauss to +0.18 Gauss).
The idea of the anomaly in residual magnetism introduced by a broken bar is truly an iffy one.
The paper continues with equation #1. A puzzling equation, because it takes the impedance, measured with AC current, (it means it includes both the rotor and the stator) and from this inductance it subtracts the stator resistance only. The rotor resistance is ignored. Yet the rotor resistance is just as important as the stator resistance. I do realize that the resistances are a small part, but why this inconsistency?
Then there is equation #2. And there is a problem again. It is not a formula for calculation of the motor inductance, but (similarly as in your paper from 1998), a formula for calculating the stator inductance when the rotor is an OPEN CIRCUIT.
If we know how to calculate the stator inductance calculating the rotor inductance L2 and the mutual inductance M should not be much harder. The motor inductance will be then:
Lm= L1-(M^2)/L2
Both rotor and the stator inductance have to be known, plus the mutual inductance.
I don’t really know; shall I ignore those inaccuracies again because of the “general message of the paper”?
The paper continues saying that the inductive imbalance is 7 to 12%. If the rewind mechanic in our rewind shop finds, that his rewound motor has inductance unbalance of 7 or 12%, he will rewind it once more. No question asked. He would know that he screwed up. There are no motors with such an inductive unbalance. It is only when you measure with LOW VOLTAGE that you find inductive unbalance of such a magnitude. Otherwise it virtually does not exist. And we can go further: Low voltage test may find an unacceptable unbalance on concentric winding. The concentric winding is an EXACT equivalent of the lap winding in normal operation conditions. There is absolutely no difference in normal operation. Yet the low voltage testing may find a substantial inductive unbalance (according to PdMA). Who is at fault? Certainly not guy who put the concentric winding in the slots, but the guy who chooses a bad method of measurement. The same applies to a rotor where the bars are skewed in certain way. It may result in 15% inductive unbalance with low voltage testing (according to PdMA) but a zero unbalance in real life.
How the low voltage test really works is clearly shown in:
http://www.pdma.com/PDF/CS0402.pdf
1) It is a 3500 hp motor, so one would not expect cheep material.
2) It is a manufactured cage, yet the rotor is not “LIR”.
3) The average inductive imbalance between the 3 phases is a virtual zero.
4) The RIC test does not show any anomaly. (However the text says that there is anomaly. Where is it?)
5) There were 22 broken or “cracked” bars. How come it did not deform the sine waves? It did not cause any residual magnetism anomaly?
6) The locked rotor current calculated from the inductances measured by PdMA gives only 170% of the nameplate current instead of something like 600%.
7) This is despite the fact that the inductance was measured with 1200 Hz. High frequency will always decrease the results of inductance measurement. It means it should have been probably only 2/3 of the inductance at 60 Hz (in other words something like 900%).

No amount of manipulation of the residual magnetism and no abbreviations (see “LIR”) can explain the above drawbacks.
Yet the explanation is as simple as the fact that the ROTOR SLOTS WERE CLOSED! It is really mind boggling that this crucial difference is constantly overlooked. The impact of the residual magnetism is just a tiny fraction of this important influence. (And it only shows on low voltage testing results).
I am not talking on behalf of thousands of non-satisfied customers as some people on this board talk on behalf of thousands of satisfied customers. I am also not after the PdMA; I am talking about the LOW VOLTAGE TESTING. I have no doubt that PdMA is the leader in this field. It published the most and it also means it automatically exposed itself to criticism. (Anybody can take their science apart). The others just follow and try to imitate (MY OPINION!).
Aren’t the above items interesting? I believe they are. In other words, a good reason for discussion.
jank

The R.I.C. test in the 3500hp 2-pole case study doesn't really look sinusoidal to me. It looks more like a sawtooth wave with rounded tops/bottoms. What does that pattern tell us?

This message has been edited. Last edited by: electricpete,

I have got an interesting E-mail from a friend of mine today, who is familiar in my interest in the motor and particularly the rotor testing. It is a white paper from General Electric about rotor testing with high frequency and low voltage. I was little surprised when I saw the date of this paper:1997. So read what was known about the low voltage rotor testing in 1997.
Link: http://www.geindustrial.com/publibrary/checkout/White%2...7CGET-8065%7Cgeneric
jank

Wow. Thanks for posting that Jan. That link provides quite an interesting new addition to the dialogue about rotor bar testing.

Specifically, it provides info on the subject of comparing two types of test checking inductance variation vs rotor position as an indicator of rotor bar defects:

• Test 1 – "single phase test" = "open bar impedance test" – conducted at power frequency and a voltage approx 25% of nameplate
  
• Test 2 – "low voltage tests" – "rotor influence check" (ric) - conducted at higher frequency and much lower voltage.

For simplicity, I'll call them "single phase test" (test 1) and "low voltage test" (test 2). Note I have lumped a lot of different tests together under LV test... there can be variation in frequency used for one thing. If it is not fair to lump them all together (for example the r.i.c. test and the LCR meter test by GE discussed further below), then I hope someone will explain the differences.

The single phase test has been used for decades by motor repair shops. It is included in EASA documents and EPRI documents. For many years (before the advent of lv testers), it was considered without question the most effective way to check for rotor defects that we can achieve without running the motor under load for on-line test or opening the motor for more direct inspections.

The low voltage test has some advantages. You can put it into a portable testbox combined with other useful motor tests and carry it to the field for testing installed motors. A special purpose motor test box can easily check all three phase pairs at once for a more detailed pattern analysis than we get with the single phase test. It does not carry the concern for overheating the motor as exists with single phase test if we are not careful.

But the big question: How does the low voltage test compare to the single phase test in terms of effectiveness at detecting rotor bar defects (sensitivity to actual faults and lack of false alarms).

Jan has long shared his test results and discussion suggesting that the LV test is not as effective, particularly for rotors with closed slots.

Now we have a published document from GE. If I can paraphrase some pieces of the article:

• 1 - Motor #1 had 1.3% variation in inductance with the LV test. It was known to have rotor defects created by drilling four holes. Conclusion: Missed the call.
  
• 2 - Motor #2 had 16% variation in inductance with the LV test. But passed all other tests, the rotor was inspected and considered good by the OEM. Conclusion: false alarm.
  
• 3 - The article attributes the false alarm (2) to residual magnetism
  
• 4 – The article attributes lack of sensitivity (1) to the closed rotor slot design

On bullets 1 and 2, I really would have liked to see a single-phase test of these same two motors. It would have been very logical to do that test to support the conclusions, but I can't tell whether it was done. Does anyone see it discussed in the paper or know the authors to get clarification?

Bullet 4 is something Jan has told us many times and has shown us data demonstrating the difference in behavior of open and closed rotor slot designs.

Now the third bullet mentions residual magnetism. I think everyone involved acknowledges the lv test is affected by residual magnetism. The new twist of the McKinnon/Smolleck 2004 paper is that it is claimed that high residual magnetism (which is known to cause high inductance variation during lv rotor test) can be indicative of a broken bar.

I think the GE article to a certain extent accepts the possibility that a broken bar can affect residual magnetism as in the following quote:

quote:

GE publication
"It is conceivable that a complete open bar or several open bars adjacent to one another will establish a pattern of residual flux which will show up as a distortion of the normal sine wave variation. In rder for this to occur, the establishment of the residual flux must be done when the open bars at the time of opening the circuit will not be carrying any current. Therefore, the effect of the open bars will not result in any field distortion. Thus lack of variance of the inductance does not prove the rotor to lack dissymmetry."

In addition to discussing the possibility that an open bar can affect residual magnetism, they rightly (IMO) point out that residual magnetism from a broken rotor bar (if it in fact is an expected result of a broken bar) must be variable in a random manner based on the relative position of the max current compared to the broken rotor bar at the moment of deenergization. That doesn't seem to lend itself very well to trending. And of course, we are still left with the documented example 16% deviation on known healthy rotor and 1.3% deviation on known bad rotor.

To me, the arguments begin to seem more compelling that the lv test is not as effective at the single phase test (it is less sensitive and more prone to false alarms). But I am still trying to keep an open mind.

We have the benefit and pleasure to have a lot of proponents of the LV test method participating here on these forums. So far imo we have heard not much detailed response over the years to concerns raised by JanK about the lv test for rotor bars. But now, it is not just JanK. We have an official publication from GE, which is a motor OEM and is also widely respected authority on motor rotor testing (by virtue of corporate association with people like Kliman). I would really be interested in hearing some form of response.

I certainly hope none of my comments are viewed in any way negative toward any individual or viewpoint. I think respectful discussion, debate, disagreement are healthy and an essential prerequisite to learning. I am looking forward to learning some more.

This message has been edited. Last edited by: electricpete,

April 2008 "Electrical Apparatus" Magazine has an article from the Engineering Editor Richard Nailen, PE on the new standard IEEE1415 for testing motors that we have talked about on the forum.

He is generally very complimentary of the new standard. The one area that he seems to have some questions on is some of the new low voltage test methods as addressed in the sidebar on page 30. Rather than paraphrasing his comments (which might introduce my own bias), I would like to quote a portion of that sidebar:

quote:

Excerpt from Electrical Apparatus April 2008, page 30:

Questions left unanswered by the IEEE document on testing
Some of the diagnostic methods recommended for motors by IEEE 1415 include "phase angle" and "phase balance tests," involving such procedures as "timed measurement of voltage and current angle in degrees," and a "variable frequency" test that purports to compare "complex impedances of the phases at different frequencies." The details of those procedures are said to be found in IEEE Standards 388 and 389.

However, IEEE 388 is not a test procedure as such, but is a "standard for transformers and inductors in electronic power conversion equipment." How it might apply to a rotating machine has not been established. IEEE 389 is a Recommended Practice for Testing Electronics Transformers and Inductors", and again its relevance to the evaluation of motor winding condition is unclear.

Those tests purport to reveal certain winding defects. Unfortunately, peer-reviewed technical literature describing controlled experiments proving the relationship of specific test outcomes to particular failure modes and incipient breakdowns does not seem readily available. For example, although the text states that "less than 5% impedance unbalance or 10% inductive unbalance calculated from the rotor test or phase comparison methods" are "commonly used thresholds", no such tests calculations or thresholds are explained in any industry standard for rotating machinery testing.

Furthermore, the explanations given are inadequate. "Low voltage ac signals... are to be applied phase to phase" in the stator. What sort of "signals"? How is "phase to phase" interpreted for a delta connected winding? "Standard inductance measurements apply" - yet no such standards exist for an induction motor. .....

In my opinion, it appears from the above quote that he is skeptical of some of the standard's discussion related to the new low-voltage tests. I also think in fairness it should be pointed out that he may not be totally familiar with some of those new low voltage tests.

This message has been edited. Last edited by: electricpete,

Thanks Electricpete for the update. Every once in a while a piece of literature comes by with a little enlightenment on the LV testers. Try as I may, I can not find any really successful unbiased ( ok even mostly unbiased ) articles on LV testing. We all want it to work, I do anyway, but it doesnt really.

If those LV test companies were smart then someone would put out some models dirt cheap. Market them as auxiliary tools to add to our bag of tricks. I would pick one up to play with. In the meantime, forums would pop up with many folk discussing what can and can not be done as well as how to do it. The technology would then have a chance to evolve.

Sorry LV testers but I am not going to drop 5 figures on a shiny diviner's rod but someone will. I think it was once said that " a buyer is born every minute " or something like that.

quote:

Originally posted by DanS:
If those LV test companies were smart then someone would put out some models dirt cheap.

I don’t think it is going to happen. The low voltage companies have to charge a lot to make living. If you want something dirt-cheap, you have to make it yourself. I have one low voltage tester in the basement that did not cost me more than probably 20 bucks. (I have lots of electronics in the basement despite the fact that I am NOT exactly an electronic wizard).
Basically you are looking for a sine-wave inverter. There are some, very easy to make with excellent waveform shape (such as Mapham inverter).
However if you add up your time, this approach is not cheap either.
jank

Too much of nasty stuff being written about LV testing! I think the main points about LV testing are being ignored, viz. quick assessment of the rotor bars & the high frequency impedance measurement can pick up shorts missed at power frequencies. The phase angle & phase balance stuff is not of much relevance.

The AT-31 costs $ 1500, hardly expensive. I've heard of Chinese knock-offs available that sell for much less.

Jank seems to have a point about closed rotor slots. However; most of the MV motors I have encountered are open slots, a few have semi-closed designs. The rotor test works well enough & can be correlated with the ESA data. In fact, most of the times when it did not work was in smaller die-cast aluminium rotors.

E-Pete: I am not sure if a variation in inductance is the right way of assessing rotor bar damage. Whenever I get strong PPF sidebands in ESA & cross-check with the rotor test, the variation seems to be erratic, in other cases it is repeatable. The software quantifies this in some way, I do not know how that works.

Regarding inter-turn shorts: I have attached three case studies, as under:

11-KBX-203: Inter-turn shorts detected in rotor field pole coils of 5.1 MW synchronous motor using both surge comparison & motor circuit analysis.

201-RM-01A: 4500 KW induction motor, tested routinely with MCA, Tan Delta, Surge, DD, etc. No problem detected. Motor failed within a minute of start-up, R & B phase line coils failed. Copper melted from both coils, insulation found burnt.

401-FN-01: 2250 KW induction motor, tested routinely with MCA, Tan Delta, Surge, DD, etc. No problem detected. Motor failed within a minute of start-up, Y phase line coil failed, insulation found burnt.

Were the failures due to switching surges or incipient faults? No idea, but nothing showed up in the disturbance recorder.

Point - Every now & then, I see a motor/generator failing that tested fine earlier by LV, HV & on-line testing. Doesn't mean that the technology is rubbish, just that there are other factors involved.

Regards,

Aditya

P.S. Would really appreciate your views on why these two motors failed.

This message has been edited. Last edited by: Aditya,

11-KBX-203,_Jan_2007.pdf (257 Kb, 26 downloads)

Data of 201-RM-01A

201-RM-01A.doc (58 Kb, 20 downloads)

Data of 401-FN-01

401-FN-01.doc (62 Kb, 11 downloads)

Aditya,

If an outside service contractor bills hundreds or even thousands of dollars to inspect motors as part of a PdM program and subsequently the motor fails shortly after FOR ANY REASON then that contractor can kiss any future business goodbye.

I often wondered why I never ran into other competitors offering PdM services using any of these 'wonderful' technologies. These tools all offer the potential to find problems but because they can all miss issues means that only customers willing to spend hundreds / thousands to get a 50% chance at saving tens of thousands or more are going to proceed with the service.

I realize now more than ever that PdM is really for operations where machine downtime runs in the thousands per hour or more.

You guys are out of my league. I will just stay on the porch and watch the big dogs.

Dan,

Most of the work I get is for large motors & generators in the MW range & often not having standby or spare motor options. Also, for petrochemicals & nuclear plants where motor problems can lead to safety issues.

Once in a way, a problem does get missed but most are detected correctly & rectified. I do get bad-mouthed by the clients in case of the miss but most also keep in mind the times I've saved their skin. Your 50 % error figure is way off, I would put it around 10-15 %. That is around the same as what is touted for vibration analysis & no one is discarding vibration as a technology.

I started off testing in 1994 with a megger & surge tester & have been collecting some tools on the way. Some have proved to be much more effective, thats all. I'm not planning on giving up any of them yet. At the same time, I would never risk giving a report to a client based only on low voltage testing.

Regards,

Aditya

I think it continues to be a good discussion.

I agree strongly with Aditya on the point there is not a single right or wrong answer such as LV testing is good or LV testing is bad. LV testing covers a bunch of stuff. And there are a variety of users with a variety of unique situations and needs. So what's right for one may not be right for another. It is more productive to discuss and what we see as the strengths/limitations and advantages/disadvantages of different approaches.

I have mentioned before I have very little experience in the LV test methods discussed (certainly not as much as Aditya, Jan and others), so I watch here to learn. And there is a lot to be learned right here in this thread. Any opinions, thoughts, analysis, articles, test results, case studies are all welcomed and add to the pool of shared knowledge so we can all make our form our own opinions and make our own decisions. Likewise if a claim is made that sounds wrong, it should be open for respectful questioning/discussion.

I didn't perceive that there was a lot of "nasty" stuff being written about LV testing. I viewed the comments primarily as discussion of specific aspects. On this particular page we spent a lot of time talking about the rotor bar testing using the single phase test vs the LV/RIC test. Aditya's experience sounds in agreement that you have seen the LV off-line test such as RIC may work for open slot rotors but does not seem to work for closed-slot (cast) rotor motors typically used in small/medium size motor applications (perhaps 600vac and below). I think that's the same as what Jan said. If we are all in agreement, that would be great. But, in the literature and case studies from the LV test manufacturers and distributors, I don't think you will find anyone suggesting that the test is not appropriate for those smaller motors. This is more than just a minor detail IMO and I would love to hear some form of response from one of the OEM's or distributors so we can either put some closure on this particular conclusion or else hear if there is some other side of this particular subject that we might be missing.

By the way, one of our motor shop guru's tells me that even the single phase test (at power frequency approx 25% voltage) becomes more and more sensitive as the voltage increases (although it gets trickier because you don't want to cause overheating). And he works on large motors (generally open slot). So if we could manage to agree that the lv rotor test has very limited usefulness on the small motors, it is still a question in my mind for the large motors. In a shop enviroment where both were available, I would certainly be more inclined to go for the single phase test even for large motors (does anyone disagree?)

quote:

E-Pete: I am not sure if a variation in inductance is the right way of assessing rotor bar damage. Whenever I get strong PPF sidebands in ESA & cross-check with the rotor test, the variation seems to be erratic, in other cases it is repeatable..

I think what you are saying is that on-line current signature analysis testing is more sensitive/accurate than off-line testing for detecting rotor bar problems and I agree 100% (we all know about thermal effects). I believe you may have misinterpretted my post dated 29 March 2008 12:01 PM where I made the statement: "The single phase test has been used for decades by motor repair shops. It is included in EASA documents and EPRI documents. For many years (before the advent of lv testers), it was considered without question the most effective way to check for rotor defects that we can achieve without running the motor under load for on-line test or opening the motor for more direct inspections". You can see I have bolded the last part of that sentence because it was important to the meaning of the sentence. That entire post was devoted to comparing two test methods performed under similar conditions (off-line testing by single-phase test and by LV RIC test) and had nothing to do with on-line testing. And although I have more confidence in the on-line test, when I pull a motor and send it to the shop based on on-line testing, the first thing I want them to do before motor disasesmbly is to perform an off-line test to gather more info. So I would like to know which of these tests will give better results and under what conditions.

Another point that I am in 100% agreement with Aditya is that motor testing is by no means an exact science. We can use our tools and make our best guesses, but there will always be machines that "outsmart" us.


Speaking of advantages/disadvantages of LV testing, I think one of the big advantages of some LV testers comes in convenience of having several tests bundled in one box which you can carry into the field. You hook up your leads and it sequences through several tests and handles the data recording. It seems like somewhat of a waste of manpower that at our plant we send people to the field to do an insulation resistance test with one test set and a winding resistance test with another test set and record the data into hardcopy workpackages, when they could gather more data using a computerized LV tester in a smaller amount of time and have the data captured directly into a trending database. And we have would extra info like trend of insulation resistance vs time over the course of the 10 minute PI test (perhaps can reveal contamination by spiking down of resistance), along with inductive imbalance and a few other items.


Now the subject of turn faults. I'll share what I think I know about turn faults (I am sure many of you are already familiar with this.... just wanted to share for those who are not... and of course if what I say sounds wrong I'd appreciate you letting me know.... I've been known to be wrong many times before).

There is one big difference between ac circuits (such as induction motor, or sync motor stator) and dc circuits (such as dc motor or sync motor rotor) in terms of how they react to a hard turn-to-turn short. In a dc circuit, the machine can continue to operate with a hard turn fault. There is only a slight increase in current through the coil (related to the fractional decrease in resistance), and machines can continue to operate for long periods with these shorts (we had some in our large syncronous turbine generator rotors). But in an ac circuit, there will be an autotransformer effect which will dramatically increase the current at the location of the fault many times above the remainder of the winding and almost always result in cascading failure and motor trip within a very short time. So detecting a hard turn fault in ac machines is not much of a consideration... it will come and find you.

Still talking about hard turn-to-turn faults: how effective are the various tests: In theory a dc resistance is proportional to the number of tunrs and an ac inductance of a purely inductive circuit is proportional to the SQUARE of the number of shorted turns. So a single shorted turn in a 100 turn circuit will reduce the resistance to 99% of its original value and will reduce the inductance to 98% of it's original value. AC testing may not represent purely inductive or purely resistive behavior, but as you increase the frequency it acts less like a resistance and more like an inductance. This tends to explain what we see in the variable frequency tests as far as I can tell.

Now the other subject of interest is detecting turn insualtion that is weak but not yet shorted. There have certainly been repeated claims and white papers claiming that the low voltage test can detect degraded but not shorted insulation. I have never ruled it out, but I am certainly very skeptical of it (there is no logical reason I can see that a LV test can detect turn insulation which is degrading but not shorted). I have aksed before and never seen what I consider a reasonable explanation or documented case studies to support that claim.

Surge testing I believe is better equipped for this job of finding degraded but not shorted insualtion. And I don't think I'm alone in that opinion, because surge testing has an entire IEEE standard all its own (IEEE522) which is referenced by the NEMA MG-1 motor purchase specification. It is limited in that it only looks towards the ends of the windings (although these are the windings which tend to see more stresses and failures.... see Aditya's 2nd and 3rd case studies for failure of line end coils). And it is certainly more effective in a shop with the rotor removed than with the rotor in-place as Aditya reminded us on the prevoius page of this thread.


Now Aditya had three case studies of turn shorts – call them Case 1, case 2, case 3. (Thanks for posting those by the way).

Case 1 would seem to be case of the hard short on a dc circuit of a sync motor rotor (where it may live for years without causing motor failure). As far as I could tell, the shorts were detectable in his test data both by dc test or by 60 hz test. The pattern was perhaps a little more easily discernible from the noise when you add the variable frequency element in there (is that how you would characterize it Aditya?).

Case 2 and 3 I think both were apparent turn shorts in the stator winding of wound rotor induction motors which occurred very shortly after machine startup and very shortly after they had been tested and found healthy by both surge testing and LV testing.

It's tough to draw many conclusion from that. Certainly as always we'd like more data (photo of the failure?, what protection tripped?, what did the post-failure tests look like?).

I think there was an implication that perhaps we consider the possibility that these may be cases where there was a big weakness in the turn insulation and the surge test didn't identify it. I don't rule that out, but the fact that the failure occured on the line end coils makes this less plausible to me (after all... why would a winding survive a transient during testing and then fail later).

Some questions that might help sort through that issue: What was the surge voltage used during the test? What sort of voltage might be expected when this motor is switched on (for example, the switching transient is worse if they were switched by vacuum breakers, worse if there is only a short run of cable between motor and breaker, and worse if there are no surge capacitors).

Some more questions/thoughts in general on these failures:

Since you mentioned the motors failed shortly after testing, was it immediately after a prolonged shutdown period? Any maintenance done on the motors during that period?

Since they occurred immediately after startup, do we know if the space heaters had been energized? Was it a wet environment?

Since these are wound rotor motors: do we know if the motor accelerated normally (the startup time).... improper setting of external resistance could lower the motor torque and cause the motor to take too long to accelerate (normally the electrical protection should trip the motor before damage in this case, but you never know).

I would also be curious to know if these machines had dedicated turn insulation. I have seen medium voltage designs from one particular OEM that rely only on enamel strand insulation to perform the turn insulation function. When you ask the OEM, they tell you that the decision to provide dedicated turn insulation depends on the STEADY STATE OPERATING voltage per turn. I may be missing something, but that seems like flawed logic to me since the challenge to turn insulation is not the steady state operating voltage but the transients.

I would like to complement you Aditya on your professional and complete test reports. Those are certainly more than we or our repair shops would do on a single machine. I think your customers are getting a good analysis of their machines. One very small thing I noticed is that the dc step voltage test is plotted as resistance vs voltage. I certainly prefer current vs voltage. Then we look for the non-linear increase in current vs voltage. I don't think that can be as easily recongized on a resistance vs votlage plot.

Once again, I think it's a good discussion. I know I'm learning from it and I hope the wide participation continues.

This message has been edited. Last edited by: electricpete,