The equipment under consideration is a molded case circuit breaker feeding the ac input to a battery charger. All of the data that I have is from load tests where a resistor bank is hooked up to the dc side of the charger (the same load in all cases... corresponds to approx 70 amps ac through this breaker).

In August 2005, we found a hotspot on phases B and C at the line side of the breaker. A phase was 54C above C-phase, B phase was 25C above C phase. (load side connection temperatures appeared balanced) The connections were found tight. They were disassembled and inspected, no anomalies found. Lugs were not replaced. Was reconnected and tested in the same manner – the problem was gone.

In October 2005, the breaker was replaced for reasons unrelated to any thermography (plant modification). I’m not sure if the lugs were replaced at that time.

On May 31, 2007, we began a load test at 06:00. At 07:30 we performed thermography and found A and B phase hotspots which looked almost identical to those found in August 2005. Specifically, we had A phase 52C above C phase, and B phase 17C above C phase. Load side was examined and qualitatively noted that A was hotter than the other two, but not nearly as much as on the line side (we didn’t save any image of that).

Approx 2.5 hours later at 10:00, the hotspots were just about gone. We now had only approx 4C rise of A and B above C phase on the line side. Load side still balanced. Currents were checked using rms clampon at 12:00 noon and found balanced. The thermal pattern remained like this for the duration of the 8-hour test which ended at 2pm.

We inspected the terminations around 4pm and found them tight, no anomalies. Lugs visually inspected sat. We did not relug (I’m not sure why). We reconnected and retested and the results were again much more balanced (only 4C difference A and B phase above C phase on line-side).

We never did replace the lugs, although they looked good based on visual inspection. Since we replaced the breaker before, the lugs are the very logical thing to follow up on and we certainly will replace them.

Some other miscellaneous facts that may or may not have relevance:
1 - The load side leads were significantly larger than the line side leads (see photo in powerpoint which I will attach to my next post).
2 – The load is a charger which may draw significant current harmonics on its input.

I have a brief summary attached to this post. Miscellaneous supporting slides will be
attached to my next post.

My questions: If the lugs were the problem, how could it be that both of them “cured” themselves at the same time back in 2005 and once again they both cured themselves at the same time between 07:30 and 10:00 am on 5/31/07? I can imagine PERHAPS one connection acting somewhat randomly, but two connections both changing at the same time together seems to suggest something outside the connections... maybe a system effect such as an unbalance. We ruled out unbalance by current measurement at 12:00 noon.... but could change over time? Well if that’s the case we should have seen it in August 2005 when we caught the line-side hotspots but load side temperatures appeared fairly balanced.

Any thoughts on what could explain these symptoms?

Summary.doc (518 Kb, 79 downloads)

A few more details attached

pws_ir1_forpost.ppt (3,760 Kb, 88 downloads)

I know this probably sounds crazy, but could the sun in some way be a factor?

EPete:

Is this the type of charger that performs a rapid charge and then a trickle charge? (or a standard charge then trickle charge?)

Is it possible that the heating occurs based upon how the charger is operating? For instance: Significant harmonics during the initial charging then lesser THD during the trickle phase? Possibly even an issue with the SCR's (assuming that it is using SCR's) in one state versus the other?

Vibeguy2004. Interesting point, but the heating seems to appear from two different angles.

Howard

quote:

Originally posted by electricpete:
A few more details attached

I have two questions about your testing:
a) was the charger charging the same battery in 2005 and 2007?
b) was an amp reading taken while the hot-spot was in evidence?

This message has been edited. Last edited by: Metalworker Mike,

quote:

Originally posted by Metalworker Mike:
I'm probably just missing something obvious, here <snip>

And I probably still am, but I just realized that my questions (and possible Howard's, as well) are answered by the 2005 picture. It shows the line and load at the same time during a hot-spot, and there is no corresponding heat rise in the load side. Unless the wires are different sizes or something is going on that I just don't see (and it wouldn't be the first time), the hot spot can't be caused by a change in demand on the load side. Not without a corresponding increase in the heats A and B phases on the load side. So since the breaker has been replaced then I would have to say that either it is the lugs, as you suggested, or something upstream of the lugs. Do you have a shot of the wires a bit upstream of the breaker? Is it possible that they're even hotter upstream? If so, is it possible that there is some kind of a very weak short between the A and B phases? Something that might be heat or vibration sensitive? Maybe in the disconnect switch... if the throw is not perfectly reliable? Maybe it doesn't 'make' the B connection as well as C, and A is even worse (linear)?
If this were my problem to solve then I would try to get a hot-spot to reappear, then check the temperatures of those wires upstream of the breaker. If they're cooler upstream then one would assume that it is, indeed, the lugs.
Remember! I am not an electrician! I'm just a millwright with delusions of competency, so keep your dopeler filter on.

This message has been edited. Last edited by: Metalworker Mike,

quote:

Originally posted by electricpete:
A few more details attached

I keep looking at this, and I just noticed something else. This might be a trick of the picture, but in the very first picture it _looks_ like L1 is aluminum wire, but L3 is copper. If L1 and L2 are aluminum and L3 is copper (I know... I know... why on earth would they be different?) then it could explain a lot. Highly unlikely, I know.. probably just a bit of highlight on the image, but I had to ask.

Thanks for all the comments.

Vibeguy – I hadn’t thought about reflections beofre, so it is worth discussing. I think I can rule it out based on three items:
1 – As Howard says, the 2005 image and the 5/31/07 07:30 image are viewed from two different locations, but look the same. Assuming they were both reflecting from the same source, we’d expect to see a difference.
2 - The temperatures are fairly uniform accross the entire conductor (with slight tailing visible on A phase). I would not expect a reflection to make the whole conductor look hot... only a portion.
3 – A good thermographer would move around and look for changes as a double check for reflections. I didn’t take these images myself, but our thermography technicians are well trained in this aspect.

Howard – all of these images were taken during load testing of the charger (that’s the only time this breaker sees much load).. A resistive load bank is hooked up to the dc side of the charger. Load doesn’t change during the 8-hour load test and was the same in all cases. I believe the charger is not connected to the battery during this particular test (we have two chargers feeding this dc system).

Mike – we had no ac amp readings taken while the hotspot was present. I will go back and see what readings they do take during the test (voltage? dc current?.... dc current is supposed to be fairly contstant during this test as far as I understand).

I don’t think this comes from upstream. If you look at slide 9 where I have adjusted the temperature scale to “zoom in” on A phase, you can see it is slightly hotter where it enters the breaker.

Interesting comment about the material of the conductor. I can see what you’re talking about in slide 1, conductor L1.... looks like either aluminum or tinned copper. I wouldn’t expect aluminum... I might expect tinned copper, but at any rate there’s no reason for them to be different. I'll ask our electricians about that one.

Someone had asked why the breaker looks different in the first two slides than the others. The reason is that there is a cover and operating handle that fits over the breaker’s own cover/handle when installed into the MCC so you can easily operate it without openining the MCC cubicle door. For the infrared images, we opened the cubicle door but did not remove those covers. For the powerpoint slides 1 and 2, those covers are removed for easier access when we lifted the leads for inspection and retightening.

Sitting here thinking about it, the only beginnings of a remotely plausible wild ideas I can come up with:

1 – Both lugs have some undetected looseness in the crimp. When sits for awhile, it corrodes or accumulates some contaminant from the atmospher (this is a clean area though). When energized it’s initially hot, but then the heat cures the problem (burns off the contaminant or oxidation?) over the course of a few hours for some reason. (I know, typically heat increases oxidation, but it's gotta be something).

2 – Perhaps there is an unbalance in harmonics that changes over time. The different size conductors on line and load side respond differently to the harmonics. That explains why the charger-induced harmonic unbalance might show up on one side of the breaker but not the other. Actually if these were solid wires, the larger wires would have a more dramatic increase in resistance due to skin effect. But perhaps there is something about the difference in cable strand configurations that allows a harmonic unbalance to show on the line side and not the load. I know this one is far out, but it’s the only way I can see that a system-induced unbalance would not affect the load side.

Good comments. Keep them coming. Any other suggestions or even wild ideas? We will be re-evaluating this on Monday.

quote:

Howard – all of these images were taken during load testing of the charger (that’s the only time this breaker sees much load).. A resistive load bank is hooked up to the dc side of the charger.

That answers my first question, too, and puts the final nail in the load-side coffin.

quote:

Mike – we had no ac amp readings taken while the hotspot was present. I will go back and see what readings they do take during the test (voltage? dc current?.... dc current is supposed to be fairly contstant during this test as far as I understand).

Since the load-side wires are consistent in temperature it seems unlikely to me that the amperage _would_ be different, but stranger things have happened.

quote:

I don’t think this comes from upstream. If you look at slide 9 where I have adjusted the temperature scale to “zoom in” on A phase, you can see it is slightly hotter where it enters the breaker.

It was a long shot, for sure. I thought it was possible that the wire might 'look' cooler as it bends away from the camera.

quote:

Someone had asked why the breaker looks different in the first two slides than the others.

That was me. I figured it out and edited my reply so that I wouldn't need a credibility transplant. 8-)

This is an odd problem, alright. I'd like to see a shot of the upstream wiring in the same box during the load test while it's got the hot-spot... at least a shot of the fuses that feed the breaker (I assume they're in the same box). I'd also like to see amperage values on both lines and loads during and after the hot-spot.

As far as wild-ass guesses go, how about this: there's a ventilation fan for that box that works on a thermostat, and that disguises the problem once it kicks on? Or the box door is closed then they open the door, take the first shot showing the problem, then leave the door open for two hours then take the next shot showing that the problem has mysteriously disappeared?
The load cell is faulty and it has a different resistance when cold than when saturated hot (should show up in amp readings during and after hot-spot) but the larger load wiring hides it. (the A phase load wire actually could be a bit hotter than the others, looking critically at the pictures again)

I think that's it for me for the moment.

Pete:

I am going to pass this by someone who may have a better idea than I of this specific situation when I am at GM next week. I will be working on some EMD stuff with their corporate QNPM infrared people. The battery chargers they primarily work with are for forklifts, but may provide some insight.

Howard

For good measure, we replaced the breaker and relugged the line side leads.

During PMT, problem still shows with the same pattern Line side has A hottest, B middle, C coolest. Load side similar. Currents measured as balanced.

Scratching our heads. Think maybe the problem is heat being transferred along the leads from behind (even though the image doesn't particularly look that way, what else could it be?). Those leads have lugs on the other end which are terminated to terminal of stab connection. The same terminals also have another load paralleled (a constant voltage transformer). Coincidentally (?), this other load is paralleled on phases A and B.

Plan is to replace those leads and stabs. Hope that will fix it.

Any comments?

Electricpete,
Did you ever get the current readings of the phases during hotspots?
Did you check the conductor material of L1? Aluminum vs copper?

I didn't see these questions answered. The answers might help point this problem some direction.

If these are 3 phase rectifier sets, could some of the rectifiers be open on the other phases?

J-

Yes, we have numerous current readings... always indicate balanced among the three phases within a percent or two. One thing I realized today - the clamp-on ammeters we have been using have not been true r.m.s. ammeters. Next time (tomorrow in fact) we will use a true r.m.s. ammeter just in case there might be some unbalance in the harmonic content that we are not fully seeing with our clamp-ons.

The lugs were saved with the portion of leads that were snipped off. I could clearly see they are tinned copper. (Copper color in the middle, silver/tin color on the outside).

This has gone from weird to weirder. We replaced the line side leads on phases A and B all the way back to the bucket stabs in back. We also replaced the bucket stabs.

The problem remains (hottest on A phase).

We checked with true rms instrument, still appeared balanced.

An update of the long history of this item is attached.

================
9/21 6pm - Replaced attachment Summary2Post.doc with Summary3post.doc that corrects the glitch which hid the dates from view.

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

Summary3Post.doc (1,098 Kb, 22 downloads)

Ahh the weird, you got to love it... Just a few things to ponder.
1. The reference point chosen on 'C' phase may not be in the best location. Definitely on the first image taken in '05 and possibly the one in '07 the spot was placed on the lug, which is nowhere near .95 emmissivity. There still is a problem just not as severe as these numbers indicate.
2. Check the spaces above, behind, and to the left. Transformers can radiate heat in very strange patterns and conduct through localized spots in the buckets. It appears that phase A and B insulation are in mechanical contact with the top of the cabinet which could conduct heat to them, with the picture provided we can't be sure where 'C' phase ends up when landed.
3. The top of this bucket also does not appear to have any paint on it, which typically would not emit very much heat, however if the door is closed a considerable amount of heat can build. Then when the door is open for any length of time the heat will dissipate. Put a piece of tape on the topside of the cabinet and measure the temp.
4. You should monitor the current for one complete cycle or at least until the problem appears. Then you may want to set the IR camera on a tri-pod and take a series of images until the wires cool to ambient. This may help to isolate the problem.
5. Someone had mentioned in an earlier post the possibility of reflection, in fact the absence of a reflection off the back of the bucket seems odd.
6. With the size of the load wires being larger there could be a substantial load imbalance, which may or may not show thermally, if the line side is not properly sized, a slight load imbalance will appear...ahh well weird.

EP,
Since none of this makes sense, you might try taking a physical temperature measurement with a contact measuring device (assuming you can do it safely) to confirm the IR or prove it is some anomoly of IR imagery.

Fluke has a nice thermocouple device that plugs into my meter and displays in deg F or C. The thermocouple could be taped onto the wire's insulation. Decide for yourself whether you can do so safely and wear proper PPE.

Since current is balanced and everything was replaced, this doesn't make sense does it? Unless like someone suggested, that there is a heat producing device thermally conducting to A phase in the image.

Just a suggestion.

Thanks LJ and Wally. Those are great suggestions.

LJ

1 - These images were taken by several different people. I agree the reference point on some images was a bad choice. We'll try to pay closer attention to that.

2/3/5 - I'm not sure to what extent we have looked closely from different angles for any clues on the nearby surfaces and to rule out reflection.

4 - I agree we are missing something, and what we are missing might be transient behavior of the current. Would be better to monitor closer to continuously.

6 – I agree. The load side leads are 4/0 on the line side leads are 6AWG. That's a huge difference. If we did have some kind of transient unbalance, since the rise above ambient on load side is so small to begin with, the unbalance might now show up on the load side.

Wally - This has got some plant people questioning our thermography program. I did ask for a contact measurement and our electricians responded that there is not enough room to work safely. We will continue to look at this option.

==========================================

The big picture is that we have replaced the breaker, the line side leads (including lugs at both ends) and the stabs at the far end of the line side leads. Attached powerpoint shows these components.


And the pattern still continues in a very similar pattern. Most recent readings on 9/21:
Time / A / B / C
10am 63 46 44
1030am 74 50 49
11am 80 59 57
1130am 64 46 42
12am 68 47 42

Interesting that it went up and then came back down. We keep the cubicle doors closed except for shooting our images.

By the way, in our case, the companion load (the other breaker in same bucket sharing the line-side same stabs) is voltage regulating transformer supply breaker only connected to phasea A and B. That VRT was under light load during our test. We surveyed the line-side connections of that breaker and they were balanced and cool around 25C. That seems to rule out heat conduction from the stab area since it is roughly the same distance through copper from the stabs to either line side breaker. We have noted that these types of transformers have caused problems with other ac load before because they draw such high harmonic current.

Also note that in our history the A phase was up to 95C or so at the location we could see. But I inspected ALL of the removed components this week (leads, lugs, breaker- disassembled to check contacts, stabs) - and NONE of them had indication of discoloration. If this were conduction from the stab area, the temperatures back at the stab area would have to have been much higher than 95C and would have caused discoloration I'm sure.

====================

So it seems that the next step is to conduct another survey under load (requires setting up a load bank), but to plan it veru carefully to make sure we get all the data that can help us diagnose the problem. This may include:
1 – more frequent monitoring of current (preferably continuous).
2 – Possibly capture current time waveforms on three phases.
3 – Check whether there is any effect from opening the voltage reg transformer breaker during the test.
4 – More frequent monitoring of thermography during the test. Possibly as LJ said watch it before and after energization as well. Check to see the apparent temperature pattern on cubicle walls and the apparent texture of the surface. Check very carefully for reflection. Hold up cardboard on the outside and see if it makes any difference – rule out reflection from nearby flourescent lights.
6 – Consider alternate temperature monitoring – either installing temperature labels in the cubicle or on the leads or finding a way to use contact probe safely.

Did I miss anything? Is there anything else we should do before/during another loaded run in order to gather data to help us diagnose the problem?

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

Bucket2.ppt (1,254 Kb, 19 downloads)

This sort of problem is far beyond my experience, but if I were in your shoes I'd be interested to try some continious monitoring of the problem as LJ suggested.

Try taking an image every hour, on the hour, for a day or two (or severl cycles of the problem). Depending on your camera you should be able to set it up on a tripod and configure it to capture an image over a set time period.

Idealy you'd run this setup with a thermal viewport (health and saftey/reducing effects of changing ambient light&heat/maintaining normal operating conditions), but that might be too expensive or otherwise impractical.

What I'd be looking for in the results is some form of pattern to the cycles that would give a clue to their cause.

Regards,

Stephen Goodfellow.