Three Window Air Conditioners Catch on Fire

Three identical failures of compressor failure, which then causes capacitor failure, and then the oil ignites, does not appear that the design was validated by any underwriter/safety compliance organization.

Did someone mess with the design and bypass the motor protection current tripping device? To me the only common mode possibility is the maintenance personnel.

Arc Fault Circuit Interrupt detects arcing, I bet the current flow though a failed capacitor looks like a low resistance behavior current path, and does not trigger the breaker.

motor cirucit protection (branch circuit breaker) is installed, anything internal to the window unit would be intact before the fires, our maintenance stafff is unlikely to be active or creative enough to take apart window units and bypass what is internal to them

so what i need is a 1-piece thermal trip device, I cut the input power, wire in this device, power is immediately cut when temperature exceeds 60 degrees C, for example.

Sounds like a line voltage problem.

C'mon, three suspicious fires in the same time period? Either you have a severe persistent overvoltage (but why didn't more equipment fail), or a pyromaniac on the loose.

Usually it is low line voltage that leads to failures (electrical over heat) of high wattage 120VAC or 220VAC (EU) appliances. Apparently, there is a power gen problem (local grid overload due to high heat) in your sandbox. Check with them, and request voltage records (open records in the U.S.) if they allow that where you live.

You probably had all three of the AC units on separate breakers, and these were probably breakers that were too large for the load to the AC units. I.E.- if the normal voltage supply would operate the full load of the AC at 15 amps, then use a 20 amp breaker circuit, not a 30 amp one. The undervoltage stayed there long enough to overheat components, cause high current draw, extreme phase angle.

I'd fix your line voltage first

Where all 3 the same brand? What was the time between failures?

they were all LG brand 2 ton units. model W24. I will research line voltage if that was a problem. Elec dept says 2 out of 3 had AFCIs installed and breakers were UL and to code, and breakers did trip, but too late to prevent a fire. As ignator said, maintenance dept had a role in this, but mainly that maintenance was virtually non- existent except for cleaning filters, and units were 8+ years old in a very dusty mid-east desert. Yes, I should write LG, and i will, with supervisor's permission of course. My overall opinion is even if breakers were to code, they did not prevent 3 fires, so "something" should have been done differently. And SolarEagle is right, I dont want to do unauthorized alterations to a manufacturer's design, and share blame for failures when i wasn't on the job........but otherwise a thermal relay (IEEE # 49) or a failure to start relay seems like a good idea ???

Be sure to monitor line voltage at maximum worst case load.

So the breakers did trip, but only after the capacitor started burning. Seems like initial failure drew enough current that the breakers did not see this as arcing current.

Would be interesting to see a photo of the burned capacitors. I'm guessing once they burned to where air exposure allowed arcing, is when the breakers tripped.

I'm assuming this is a permanent split capacitor motor, so this capacitor is a start/run capacitor, and always in the circuit.

Seems like Andy German's solution to replace all the capacitors in the existing units would be prudent. 3 identical failures is showing a statistical trend.

Capacitors are most likely failing in voltage breakdown. You could go to the next higher voltage rating if you have room for this larger volume capacitor. But the problem, as others commented, is quality of dielectric failing over time.

GA Andy.

"....or its resistive value is very low,"

You mean its parallel resistance is too low or its series resistance is too high. This cannot be tested with simple digital capacitance meters that are common in handheld multimeters today. It requires a capacitance meter that also measures conductance.

I meant that if you used a high voltage Ohmmeter, you would probably read a relatively low value, not good..

Usually, a good cap of any type has a very high DC resistance, at least megohms...though depending upon construction, some have a higher resistance than others....

But I am not an expert on high voltage un-polarised motor start caps, rather mostly the reverse (pun intended)....I know enough to be dangerous!! There are people here that understand them far better than I....

Being nosy, I usually check them with a cap meter and then simply go out any buy a new one no matter what the meter says..... I find they are often a long way away from the marked value....

Funnily enough, it often fixes the problem....

Some things make a single phase motor run hot (I am sure there are more!) :-

1) Too low a voltage definitely or even too high, but that seldom happens in my experience!

2) Centrifugal switch is always shorted

3) Start cap is too low in value - dried out, motor fails to turn. Probably a run cap too if fitted, but I have never had that problem.....thats just a wild guess.

4) Bad bearings - mechanical problems.

Some of the problems are affected by the actual mechanical load that the motor physically drives.....even if the motor itself turns freely.

GA!

It's amazing how they often make cheap capacitors recently.

Apple computer had a major problem with cheap Chinese capacitors in their 2003-2004 G5 series Mac computers, which have virtually all died by now. Fortunately, I'm not aware of them causing any fires, and they haven't had the problem in more recent models.

My Carrier brand A/C is 48 years old, and still operating fine on the original caps. I did replace the fan motor a few years back, but the compressor is original.

This last year I helped a friend repair a 70 year old table saw motor. The insulation on some of the wiring became so brittle that it fell off and shorted out. I fully expected to have to replace the flat-pack capacitor, but to my amazement it was and is fine!

Thus the only reason these caps are failing is poor construction and/or selecting values with inadequate tolerance for handling higher temperature, voltage surges, etc.

Wow, so old and still fine!

Like me!!

The modern capacitor crap is simply awful!!

GA for Andy Germany, made me think to replace the existing wet caps, "filled with vegetable oil" with dry film caps of the same value, like this one here

http://datasheets.globalspec.com/ds/992/VossLighting/6488609C-7BA4-4A1C-A85C-91313F65C06F

which is almost exactly perfect, I need 60 μF, this one is 55 μF

Over many years I have seen many designs in electrics and electronics fail just through the point you made in your post.

GA.

Even 600 volt SCRs and TRIACs will fail regularly here, as we found out, when used on European mains, instead of US mains, where they worked fine.

I once had a massive job of de-soldering dozens (eventually hundreds really!!) of them and replacing them with 1000 volt devices, that simply worked!!!

May 1981, my first few days on the job in Germany.....you don't need a local language to solder!!!

Here is the NERC study on FIDVR (fault-induced delayed voltage recovery):

http://www.nerc.com/docs/pc/tis/FIDV_R_Tech_Ref_V1-1_PC_Approved.pdf

Especially pay close attention to figure 1, that reveals the voltage-time response in a system relying on capacitor banks to correct phase lag, and undervoltage after a substation trip, etc. Note the longer-term behavior where the voltage is considerably under-voltage. This leads to things like refrigerators, AC units, and televisions burning out.

To catch any UL-approved equipment on fire, one must "try really hard" - meaning that undervoltage must exist due to local supply overload and undervoltage that persists for hours.

Unfortunately, that link brings up a solid black page, and the download won't open in any of my PDF viewers.

OK, I just checked link, and it worked fine for me. You may be using the Cavalier version of browser (Windoos -2.6).

Well that was strange! I commonly use Firefox (on a Mac) for CR4, because it has the toolbar, while Safari does not. I just tried your link in Safari, and it works fine! Just in case, I did go back and try Firefox again, and it still does not. In fact Firefox does present a message saying "This PDF may not display correctly", so Firefox can detect something awry.

From first glance, that's a really interesting article; I'll have to study it more...

Thanks!

Great article! A key sentence from it:

"The mechanical torque of the compressor for these units tends to remain relatively constant for the first few seconds which makes the motors more prone to stalling."

"these units" to mean single phase a/c units. The exception is the scroll type compressor.

window acs are not rated for 24x7 use. have appropriate additional capacity with cyclic timer to avoid continuous working

AC compressor motors typically have an overload within the circuit. 'Klixon' is a common name in these parts.A simple bimetal disc that warps (or unwarps) upon reaching a designed temperature opening the circuit.When cooled to 'normal' temps the disc closes the circuit.

Maybe the 'Klixon' contacts became welded closed (high V & or I)

Any recent electrical work @ the facility? Maybe someone decided a 480 leg to ground would suffice for a 240 pair?? Or perhaps grabbed the 'stinger' ???

Someone's gonna get killed if this isn't immediately addressed-& I don't mean by posing a question to CR4.

Your posting here has declared your awareness of this situation.

Tada

From what I could find, the cap may be a, " 370v 45 ". Not enough info provided to be accurate.

You seem hyper focused on the "failure" of the AFCI breakers. Note that AFCI does not mean All Fire circuit interrupters, they mean Arc Fault Circuit interrupter.If your motor overloaded or the caps failed WITHOUT causing an arc, the AFCI would not trip. It's not designed to.It is still a Thermal Magnetic Circuit Breakers, so if there is no direct short circuit, it is only the thermal trips that take it off line. If your capacitors failed, that's what I would expect to happen. The fact that they a) failed and b) caught on fire before a CB tripped is a serious problem WITH THE AC UNITS, but the circuit devices ahead of it were likely just doing their job.

If you are interested in pursuing this further, do a search on the term "capacitor plague", a well known and documented problem that took place a few years ago in which an Asian capacitor mfr stole the "secret sauce" formula for the electrolyte from another, only to discover later that it was a fake. Unfortunately, the fake electrolyte made it into literally billions of capacitors and out into the marketplace for years before it was discovered, long after recalls could be effectively accomplished. The legacy continues to this day.

Followup:

as part of the annual or semi-annnual PMs, the HVAC dept will now be doing two new things: 1. changing out the caps, and 2. megging out the compressor motors.

Great moves, well done getting the implementation passed by your company.

Maybe you should also install a line voltage reference, and record line voltage just to be smart.

Also please measure presence of harmonics in the electrical line to avoid caps failure/burning.

Manindra.

Andy G, your right, the most amazing part is that HVAC cooperates with this solution, even tho it involves extra work for them

the latest: another one overheated, the room resident smelled something (almost) burning, and turned it off before there was an actual fire, so we will take it apart tomorrrow and i will post results here.

Its getting too hot in here to think of anything but line voltage or conductor capacity. I'd use an exclamation point, but somebody used them all up.