Causes of UPS Fire 3

[cslater]

A facility manager I work with recently had a small APC UPS catch fire and burn up. A week before, they had an SCR on a large UPS fail.

We suspect the cause to be some sort of power quality issue on the distribution system. The issues have come up since introducing some large, non-constant DC power supplies to the system.

My first thought was harmonics, but some reading I've done indicates that it might also be voltage fluctuations. I understand why harmonics could cause failures, but I'm not clear on why voltage fluctuations might.

Two questions then:
1) Is it possible for voltage fluctuations to cause this kind of damage?
2) Are there other power quality issues that he should be looking at - like Surge, Sag, etc?

Thanks very much!

 

[wayne440]

I can't answer your question, but some APC models are noted for overcharging batteries; rarely they catch fire, more often roast the batteries until they swell and make a terrible odor. The case will be hot enough burn your hand, which will immediately make the odor almost insignificant.

 

[cslater]

I looked inside this unit and the batteries were fine, but the board and coils immediately connected to the incoming and outgoing power were completely cooked. It looks like the fire started somewhere in that area.

 

[Skogsgurra]

Except for someone putting things on the UPS and blocking cooling, there's the situation where incoming voltage is low but still not out.

If the UPS is heavily loaded and input voltage goes down, then the UPS will compensate the low voltage by drawing more current (needs to get same power from grid) and that does it to the input circuitry.

If this happens often, then check if load is high or if the low voltage limit is set too low (some UPS don't even have a low voltage limit).

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[Boof87]

Check the load power factor. I have seen and heard of certain units that can't handle a leading power factor and it has caused the output chokes (and output isolating Tx if the unit had one) to catch fire. It rarely happens on the newer units as they are designed to run at a leading power factor up to to 0.9 leading where the older ones were not.
How old is the unit and was there regular maintenance performed on it, and to what extent?

 

[bayoubare]

a cause of the fire could have been an out of sync transfer.

 

[LionelHutz]

Some UPS models do line voltage regulation and filtering but are not on-line all the time. So, they must have some sort of passive filtering to accomplish this regulation and filtering. Typically, most passive filtering schemes have limits on the type and level of electrical noise they are capable of filtering. I would suspect there is excess noise on the power grid if this UPS was so equipped.

 

[sibeen]

I suspect you may be chasing a gremlin where no gremlin exists.

UPS systems fail and the small UPS systems fail quite a lot. I used to work for a manufacturer of these units and the failure rate for small units (warranty claims) was calculated to be about 3% in the first year. Most of the failures will be quite benign and the UPS will shut down but in rare cases you'll end up with smoke and flame.

The units are built cheap, especially the magnetics within them so it's not surprising that they fall over quite regularly.

As for the large UPS you state that an SCR failed. What part of the circuit? Was it the rectifier, the bypass (reserve) static switch or the inverter output static switch? Each one could have a different cause and therefore a different set of theories in to what caused the failure or it just may be that the thyristor failed because it was its time, or a manufacturing defect or hit by an energetic cosmic ray.

Two UPSs failing on one site within a week of each other, especially when one is a cheap and nasty, wouldn't send me into a spasm of investigation.

 

[WoodrowJWeen]

Gotta agree with Sibeen on this one, especially the cosmic ray thing. Personally I like "sunspots". Chasing gremlins or ghosts will eat up a lot of time and effort and probably get you nowhere. And yes, small UPSs fail all the time, what do you want for 40 bucks.

 

[DanEE]

One point being overlooked so far in the discussion is the design of the UPS itself..

The corporation I worked for had very stringent design standards in that significant smoke/fire emission under any failure mode circumstance was a no-no.

Not all companies have developed the FMEA (failure mode effects analysis) skills/expertise for design and test verification of circuit boards and power assemblies to not cause fire/smoke under any failure mode. But it has been done.

A simple example is the failure mode carbon resistors in a power circuit can exhibit. Carbon resistors as one might image, in overload conditions can glow red hot for sustained periods and become a source of ignition. Metal film resistors tend to behave more like fuses and safely open the circuit. Base circuit board composition is another factor that can determine the degree of escalation that will occur. Primary fusing alone may not be sufficient depending upon power supply topology and additional internal fusing points may also be needed. Ferro-resonant transformer designs are notorious for this due to ferro fold back on overload.

The core design premise used is that any fault condition capable of producing fire/smoke must "clear" (e.g. design must operate like a circuit breaker) or energy levels be low enough at the point in question to not generate fire/smoke.

The design principle is no different than circuit breaker/fuse trip/clear curves must be consistent with acceptable temperature rise/insulation limits of connected downstream circuits.

I think I would consider using a different UPS supplier.

 

[electricpete]

We had two UPS's of the same design fail. The first one was a magnetic choke. I don't know al the details but a new firing card was developed by the vendor which was supposed to reduce the heating of the choke in light load conditions if memory serves me right (they got some money out of that deal). A few years later, another UPS of same design failed. This time it was a dry transformer inside the UPS. We had no testing program for thse tiny transformers, and a little bit of research showed they were operating above their temperature rise limit, in part due to the whole cubicle being hot. In retrospect, the heat may have contributed to the first failure as well. Our corrective actions based on consulation were the vendor for the second failure were improved ventilation (added some forced ventilation via muffin fans to a previously natural-circ design). Additionally we added periodic thermography survey.


I don't work too closely with UPS's. My general impression is they are complicated devices with lots of design variations and many possible failure modes.

Suggestions for investigation:
1 - if you study the fusing and protection of the device perhaps you could rule out some failure modes.
2 - do a thermography survey of other similar units to look for anything unusual that might indicate a design vulnerability
3 - my very limited experience (2 failures) tends to support sibeen's generalization that magnetic components are more likely suspects.
4 - get the vendor involved. Not only are the familiar with the design, but chances are they have been called by other customers to consult on failures and have some history of weaknesses in your design.


=====================================
(2B)+(2B)' ?

 

[electricpete]

Clarification:
We had two UPS's of the same design fail
should have been:
We had two UPS's of the same design catch fire
(There are a whole lot of ways for a UPS to fail and very few of them should result in fire)

=====================================
(2B)+(2B)' ?