Burned GFCI Question

[tjincbfl]

I was recently called to look at a GFCI in the men's room of one of our buildings. It was apparently the cause of a fire alarm pull when someone observed "flames shooting 4-5 inches out of it..." It's obviously burned but hard to say exactly what it looked like at the time.

It was located in a poor location, directly beneath the sink edge and I believe water and/or soap running down the wall was a significant contributing factor. (Sink edge lines up with the H side of outlet.)

The top lugs of the GFCI were not hooked up, so no other outlets in the area were protected. The in/out wires were hooked together (solid core 12GA) and then stranded wire was jumpered to the H and N (also 12GA). No grounding wire was present, but the outlet was bolted to the box, tied to the conduit.

It was on a 20A circuit breaker that did NOT trip.

Since only the outlet was really 'protected', the GFCI didn't really play any role, other than having more components/material to burn. I'm thinking that water created a short between the H and case. All conductors and insulation beyond the GFCI box (into the conduit) show no signs of damage or heating.

Is it possible to melt through a 20GA stranded wire without tripping a breaker?

And while I don't think it's what happened, does a GFCI care about a H-N short?

 

[tjincbfl]

Location of outlet

 

[Buck Parrish]

I was recently called to look at a GFCI in the men's room of one of our buildings. It was apparently the cause of a fire alarm pull when someone observed "flames shooting 4-5 inches out of it..." It's obviously burned but hard to say exactly what it looked like at the time.

It was located in a poor location, directly beneath the sink edge and I believe water and/or soap running down the wall was a significant contributing factor. (Sink edge lines up with the H side of outlet.)

The top lugs of the GFCI were not hooked up, so no other outlets in the area were protected. The in/out wires were hooked together (solid core 12GA) and then stranded wire was jumpered to the H and N (also 12GA). No grounding wire was present, but the outlet was bolted to the box, tied to the conduit.

It was on a 20A circuit breaker that did NOT trip.

Since only the outlet was really 'protected', the GFCI didn't really play any role, other than having more components/material to burn. I'm thinking that water created a short between the H and case. All conductors and insulation beyond the GFCI box (into the conduit) show no signs of damage or heating.

1. Is it possible to melt through a 20GA stranded wire without tripping a breaker? Yes. The breaker works mostly off of a ground fault or overload.

And while I don't think it's what happened, does a GFCI care about a H-N short?

Yes, but if the problem was on the line side of the gfci and not some thing plugged in to it. The breaker would not neccessarilly trip.
It would not have mattered if the gfci tripped.
Just like the coils in a heater will get red hot and not trip a beaker, unless a ground fault occurs or it becomes overloaded.

 

[cowboyjwc]

My guess would be that the problem was in the GFCI it's self. It blew the OC device didn't see anything, and once it blew it was enough to cause the small fire.

Remember a fault has to occur in order for anything to trip, but once it has occured, even if the OC device had tripped, it could have still caused a fire.

I know that's a little simplified so don't read to much into it.

 

[tjincbfl]

Yes, but if the problem was on the line side of the gfci and not some thing plugged in to it. The breaker would not neccessarilly trip.
It would not have mattered if the gfci tripped.
Just like the coils in a heater will get red hot and not trip a beaker, unless a ground fault occurs or it becomes overloaded.

Correct, there was nothing plugged into the GFCI at the time, so I don't think that circuitry mattered. But if the water shorted the line side to the box/conduit, wouldn't that be seen as a ground fault? Or would that only be the case if there was a dedicated ground line going back to the breaker panel?

From my 2008 NEC, a 12GA wire has ampacity of 25A...

 

[Buck Parrish]

Correct, there was nothing plugged into the GFCI at the time, so I don't think that circuitry mattered. But if the water shorted the line side to the box/conduit, wouldn't that be seen as a ground fault? Or would that only be the case if there was a dedicated ground line going back to the breaker panel?

From my 2008 NEC, a 12GA wire has ampacity of 25A...

Their may have been enough minererals or 'gook' etc... in the water to cause resistance (heat) but not enough to cause a fault.
It could have been inside the gfci as well.

 

[LarryFine]

If the water conducted line to line, the current would be seen as 'normal' by the GFCI.

If the line and load terminals are jumped, the GFCI circuit has literally been bypassed.

 

[gar]

100701-1332 EST

tjincbfl:

Some questions you should ask:
1. How much power is dissipated in a resistive load of 120 V at 20 A?
2. How much temperature rise will occur with this power dissipated in a 0.5" x 0.5" x 0.5" cube?
3. Does a short imply 0 resistance?
4. What components in a GFCI might fail and why?
5. What is the resistance from one face of a 1 CM cube of water to an opposite face?

I do not understand this paragraph. What are you trying to say?
What are top lugs? What are in/out wires?

The top lugs of the GFCI were not hooked up, so no other outlets in the area were protected. The in/out wires were hooked together (solid core 12GA) and then stranded wire was jumpered to the H and N (also 12GA). No grounding wire was present, but the outlet was bolted to the box, tied to the conduit.

.

 

[John120/240]

I think he's trying to say "down stream recptacles", nothing else was protected

 

[gar]

100701-1406 EST

John:

It was semi-clear there were no downstream loads. But in/out wires connected together sounds like the input hot was connected to the output hot by a solid #12 and the same for the neutral. But this would not cause overheating. Just a moderately non-functional GFCI.

If these were criss-crossed then lots of heat would be generated and fail internal wires or conductors in the GFCI and not trip the supply breaker.

The input MOV or other components in the GFCI might have shorted and burned up.

Both moderate shorts between hot and EGC or neutral could produce lots of heat.

.

 

[tjincbfl]

100701-1332 EST

tjincbfl:

Some questions you should ask:
1. How much power is dissipated in a resistive load of 120 V at 20 A?
2. How much temperature rise will occur with this power dissipated in a 0.5" x 0.5" x 0.5" cube?
3. Does a short imply 0 resistance?
4. What components in a GFCI might fail and why?
5. What is the resistance from one face of a 1 CM cube of water to an opposite face?

I do not understand this paragraph. What are you trying to say?
What are top lugs? What are in/out wires?


.

1. 240 Watts P=VI
2. P=(t1-t2)/[(1.27/(4.186*K*1.61)] Where K is thermal conductivity of material, which you didn't say. Assuming copper K=.401 W/cmK My guess is just over 500 degrees
3. Sometimes, but not always. Real world applications would be no.
4. Too many to list, but I think a better question is not why, but how.
5. Depends on the water.. Tap water is different than DI water, different from distilled water...

On the GFCI there are four screw terminals. I was calling those 'lugs'. The top terminals had no wires connected to them. The hot wire coming up the conduit was tied directly to the hot wire going to the next receptacle. Similar condition for the neutral. I was calling those "in" and "out" meaning in and out of the box. A stranded wire was used to connect the bottom two terminals of the GFCI to the in/out wires in the box.

 

[gar]

100702-1421 EST

tjincbfl:

With you description now it appears that internal components in the GFCI failed. Without a very large ion concentration in water there will not be enough conductivity to cause a large current and therefore large power dissipation. However, the leakage current from water over the printed circuit board could cause the SCR to stay on. Another possibility is a large voltage transient with lots of energy failed the input MOV.

Open the GFCI and see if there is a unique problem area.

I have a heat treat oven that with less than 1800 W can heat components to over 2200 F. The interior is about a 6" cube.

.

 

[mxslick]

I see no need to overthink this, it looks like water and soap got into the GFCI circuits and caused it to fail. The replacement would get a weatherproof cover if it were me doing the repair.

It took a few reads for me to understand what the wiring was but it is simply stranded pigtails tied to the solid conductors to make installing the gfci easier.

Since the wires inside the box got toasted I hope you pulled in replacements or if there was enough slack you pulled down to undamaged. :grin:

I had an outdoor GFCI do the same thing despite the weatherproof cover...it was pretty old and I think the combination of years of dust and dirt buildup inside plus a few days of very high humidity finally started a small arc failure that escalated. Luckily I was home at the time and killed power before it caused any serious damage.

 

[kwired]

1. 240 Watts P=VI
2. P=(t1-t2)/[(1.27/(4.186*K*1.61)] Where K is thermal conductivity of material, which you didn't say. Assuming copper K=.401 W/cmK My guess is just over 500 degrees
3. Sometimes, but not always. Real world applications would be no.
4. Too many to list, but I think a better question is not why, but how.
5. Depends on the water.. Tap water is different than DI water, different from distilled water...

On the GFCI there are four screw terminals. I was calling those 'lugs'. The top terminals had no wires connected to them. The hot wire coming up the conduit was tied directly to the hot wire going to the next receptacle. Similar condition for the neutral. I was calling those "in" and "out" meaning in and out of the box. A stranded wire was used to connect the bottom two terminals of the GFCI to the in/out wires in the box.

check #1 again it should be 2400

 

[danickstr]

I like the theory by Larry that if the unit is not properly grounded (or the water somehow avoided the ground) and the GFI saw "normal" as the water shorted the rec. UC to GC it just melted away not seeing a problem.

 

[tjincbfl]

check #1 again it should be 2400

Ooops... Yep. Bad math day.


Unfortunately it apparently burned for a while, it's tough to tell what's what with it. It's not going to 'open up' nicely. From what I can tell using my 'x-ray vision', the MOV is still intact. I don't know if it's fused, but it didn't rupture like others I've seen. That makes me think it wasn't a spike on the line.

I'd also guess that the GFCI is actually tripped. Or let me rephrase that to 'The reset button is no longer latched at the inside.' Might have seen some damage that caused the catch to move, but I was thinking the tension on it would have held it in place if it wasn't 'triggered'.

I'm still trying to get a hold of the same GFCI for comparison so I can tell what I'm looking at.


I'm not sure what what this means... (from danickstr)... "as the water shorted the rec. UC to GC" What's UC and GC? ???? and Ground Circuit?

 

[tjincbfl]

I see no need to overthink this, it looks like water and soap got into the GFCI circuits and caused it to fail. The replacement would get a weatherproof cover if it were me doing the repair.

The receptacle was in a mens room, the fire department got called and the building was evacuated. The fire department just wants some solid proof as to why this happened. Agreed that it was a poor location and it will probably be simply eliminated from the bathroom. They just want to know if it was a problem with that GFCI for some reason and if they have to worry about others in service.

 

[mxslick]

The receptacle was in a mens room, the fire department got called and the building was evacuated. The fire department just wants some solid proof as to why this happened. Agreed that it was a poor location and it will probably be simply eliminated from the bathroom. They just want to know if it was a problem with that GFCI for some reason and if they have to worry about others in service.

So they want you to do their investigative work for them? :roll:

It doesn't take a rocket scientist to see that it was in a bad location, and the combination of water and soap runoff caused the electronics to short out.

It was a problem strictly with location in relation to the sink. Replace it, put a w/p cover on it and move on. Get W/p covers on any others in the building in similar locations as a preventative.

If the FD wants any more info or legwork done they should be paying you for it.

I don't mean to sound as if I'm in a bad mood, but it bugs me when the guys responsible for investigations try to pass off their responsibility to someone else....and BTW are you ready/insured for the liability if they use your input into investigations or later claims?

As for figuring out how the GFCI itself failed, don't mess with it, send it to the manufacturer (or better yet, turn it over to the FD and tell THEM to send it in) as the manufacturer can tell exactly what failed, in what order and from what.

Edited to add:

I'm not sure what what this means... (from danickstr)... "as the water shorted the rec. UC to GC" What's UC and GC? ???? and Ground Circuit?

He means ungrounded conductor and grounded conductor i.e. hot and neutral.

 

[LarryFine]

The receptacle was in a mens room, the fire department got called and the building was evacuated.

At first glance, I thought I saw " . . . in a mens room at the fire dpartment . . . " :roll:

The fire department just wants some solid proof as to why this happened. . . . They just want to know if it was a problem with that GFCI for some reason and if they have to worry about others in service.

I'd offer the manufacturer's name and suggest they call.

So they want you to do their investigative work for them? :roll:

Exactly.

 

[tjincbfl]

So they want you to do their investigative work for them?

Yes.

It doesn't take a rocket scientist to see that it was in a bad location

Funny, but apparently it does...

If the FD wants any more info or legwork done they should be paying you for it.

They do...

...and BTW are you ready/insured for the liability if they use your input into investigations or later claims?

Yes

As for figuring out how the GFCI itself failed, don't mess with it, send it to the manufacturer (or better yet, turn it over to the FD and tell THEM to send it in) as the manufacturer can tell exactly what failed, in what order and from what.

Sometimes. A manufacturer is not always a good unbiased opinion. Lots of times things go back and you never hear from them again. They are more focused on the next million units they can sell than the 10+ year old items they no longer carry. Sometimes they are VERY helpful, but not always.