Induced voltage causing GFCI breaker to trip

[cdcengineer]

So I have a project where we have some long branch circuit runs. I don't recall the specifics, but let's assume that there are two 50A/2P, 208V receptacle circuits served from a 208Y/120V, 3-Phase, 4-wire service. Both of these branch circuits share a common underground PVC conduit and the conductors were sized per 310.15(B)(3)(a), re: more than (3) ccc's. For discussion purposes, let's say that the longer of the two is 475' and the shorter is 425'. I believe they might be #2 AWG CU.

The local AHJ required that the circuits be protected by GFCI breakers. The circuits serve 50A receptacles that serve occasional event power via 50A spiders. I argued with the AHJ that the spiders have (6) 20A GFCI receptacles and that the GFI protection is redundant.

Anyway, the contractor tried to energize these circuits and when trying to heat both up, one would trip. I believe he said that if #1 was energized, than when he fired up #2 one would trip. Same goes for if he had #2 on and fired up #1. I'm not sure if it was one recep or the other. I will verify this tomorrow as it now seems to be a critical component as I write this.

He questioned whether there could be an induced voltage from the first circuit which was causing the GFI to trip. He stated that when they broke apart one of the circuits into individual conductors, they were able to read a voltage to ground from the hots (between 8 and 14 V). I don't believe they were reading any voltage when referencing the neutral (grounded) conductor. My first question was, "what type of meter are they using". A high impedance DMM might read ghost/phantom voltages. Whereas a low impedance tester like the T+ pro or Fluke 117 might not see any voltage. This should clear up quite a bit. My understanding is that with a low Z meter, if we see voltage, than there could be a back-fed or bleed voltage from another conductor. Considering that all the wires were broken apart, I would think that they can rule out back-feed. But because the wires are underground and it's not unusual to find water in conduits here at 10k feet above sea level, than perhaps compromised insulation is causing the voltage they are reading.

I don't believe unduced voltage is the culprit in this case. But perhaps the members can provide some of their own experiences on this topic. I initially thought that redundant GFI protection could be the culprit. However, after talking with a manufacturer, they said that that hasn't been a problem with newer generations of GFI for a long while.

I did some research on induced voltages, and I would not doubt that there could be an induced voltage at such long lengths and I'm betting that the wires are a twisted mess within the conduits. However I don't see that as possible with no loads plugged into the circuit. My thought is no current = no chance for induction. Although perhaps the length of the conductor (ie resistance) is acting as a load. However this should not cause a GFI brkr to trip. The breaker is looking for heat, overcurrent and current out = current in within 3 to 5mA.

Is it possible that an induced voltage with such long conductors (resulting in high resistance Z) is adding current to the circuit which when seen by the GFI is causing this?

I suggested they get the low Z meter and start there. If that shows no voltage than I would say the next simplest solution would be to look at the possibility of a bad breaker (or two). This scenario is one of at least a half dozen that were used at the same event, and there were no other problems. Therefore I though they might try swapping breakers with a known good GFI. If for some reason they see a voltage with the low Z meter, than it's gotta be a case of bad resistance. This could be found with a MEGGER.

As of right now, I am just trying to provide some helpful tips. And if anyone has some first hand experience with a situation similar to this, please feel free to provide insight.

As always, thanks.

 

[starbolin]

GFCIs don't sense voltage, they sense current. Replace the GFCI with non-GFCI breakers and put an amp-clamp around all the CCs in the branch circuit. Randomly twisted CCs can lead to current induction in adjacent circuits however, the currents should remain balanced within a circuit leaving the GFCI unaffected. A ground fault on one of the CCs would then upset this balance and could cause a GFCI to trip.

Likely culprits for circuit to circuit coupling are circuit to circuit leakage and poorly laid conductors coupled with a ground fault.

 

[gar]

140918-2213 EDT

If by induced voltage you mean a magnetically generated voltage or current, then almost certainly it is not.

Capacitive coupling might be a possibility. 400 ft times 80 pfd per foot would be about 32000 pfd or about 0.03 ufd. This is equivalent to about 80,000 ohms at 60 Hz. At 208 V (line to line) this is possibly insufficient to trip a GFCI, and also you actually have a capacitive voltage divider. Is my capacitance per foot estimate low? I really don't expect so.

On the load side of the GFCI between the output hot and the output neutral of one breaker place a 1 ufd 600 V polypropylene capacitor and see if any difference.

I assume you have separate neutrals for the output side of each GFCI breaker. If not you are in trouble.

.

 

[kwired]

GFCI is for people protection, whether or not it is required in your application may be questionable, but if it is required it would only be required for the receptacle outlet, not the entire circuit.

If you were to move the GFCI closer to the outlet instead of having several hundred feet of conductor to possibly have capacitive current leakage causing the tripping, you only have limited conductor to have losses and likely will have little or no such trouble.

 

[zbang]

The local AHJ required that the circuits be protected by GFCI breakers. The circuits serve 50A receptacles that serve occasional event power via 50A spiders. I argued with the AHJ that the spiders have (6) 20A GFCI receptacles and that the GFI protection is redundant.

Does the AHJ have a citation for his requirement? I'd love to see it. Art 525 addresses a potential use, but 525.23(B) sez:

Where GFCI Protection Is Not Required. Receptacles that only facilitate quick disconnecting and reconnecting of electrical equipment shall not be required to be provided with GFCI.

He might be looking at 590.6, but that only has it on 15/20/30a receptacles.

 

[kwired]

Does the AHJ have a citation for his requirement? I'd love to see it. Art 525 addresses a potential use, but 525.23(B) sez:


He might be looking at 590.6, but that only has it on 15/20/30a receptacles.

590.6(A)(1) does mention 15/20/30 amp receptacles. But also specifies 125 volts. But read on into part (B) and it requires either GFCI protection for all other temporary outlets for construction purposes or a assured equipment grounding conductor program, this would include the 50 amp receptacle in OP, with the exception that the OP's isn't necessarily being used for construction purposes which should disqualify it from applying.

 

[kingpb]

I'm confused. You have 2 x 50A 2P, 208V circuits sharing a conduit, and they both feed 50A receps that are intended to serve spiders that each have 20A GFI receptacles. Where is the neutral coming from for the single pole loads?

Sounds like an induced current issue, common when two separate GFCI circuits run in parallel, and in close proximity, i.e. same conduit.

I can see why the AHJ wants it on a GFCI breaker, because there is nothing that is keeping something else from being plugged into the remote 50A recep that is not GFCI protected at the utilization point. Yes, when you plug the spiders in, it becomes redundant, but what guarantees that is the only thing that the recep will be used for.

 

[kwired]

I'm confused. You have 2 x 50A 2P, 208V circuits sharing a conduit, and they both feed 50A receps that are intended to serve spiders that each have 20A GFI receptacles. Where is the neutral coming from for the single pole loads?

Sounds like an induced current issue, common when two separate GFCI circuits run in parallel, and in close proximity, i.e. same conduit.

I can see why the AHJ wants it on a GFCI breaker, because there is nothing that is keeping something else from being plugged into the remote 50A recep that is not GFCI protected at the utilization point. Yes, when you plug the spiders in, it becomes redundant, but what guarantees that is the only thing that the recep will be used for.

590.6 only requires the 50 amp receptacle to have GFCI protection if it is used for "supply temporary power to equipment used by personnel during construction, remodeling, maintenance, repair, or demolition of buildings, structures, equipment, or similar activities".

That is not the application in the OP. If anything we start in art 525 for the OP's application, I see no GFCI requirements in 525 at all. If this is in a fixed structure then 525.3(B) tells us to see 518 and 520 and utilize them if they apply, but I didn't really see any mention of GFCI protection scanning through those sections either.

 

[cdcengineer]

I'm confused. You have 2 x 50A 2P, 208V circuits sharing a conduit, and they both feed 50A receps that are intended to serve spiders that each have 20A GFI receptacles. Where is the neutral coming from for the single pole loads?

Sounds like an induced current issue, common when two separate GFCI circuits run in parallel, and in close proximity, i.e. same conduit.

I can see why the AHJ wants it on a GFCI breaker, because there is nothing that is keeping something else from being plugged into the remote 50A recep that is not GFCI protected at the utilization point. Yes, when you plug the spiders in, it becomes redundant, but what guarantees that is the only thing that the recep will be used for.

Each 50A circuit has it's own neutral. As for citing an article for "why" the AHJ wants GFI protection - they feel as if they don't have to provide a specific reason. And I can agree that it might be possible to plug something into one of the 50A receps in other installations. But here, the town owns these receps and the town staff goes out and energizes the circuits and plugs in the spiders for specific events. If for some reason a vendor had a camper style trailer and wanted to plug into one of the 50A (hypothetically), than I would suspect that the camper would have GFI protection where it was required.

Induced current? I always thought it was induced voltage when current was present and flowing. I'll need to do some more reading. Anyway, am I correct that a low Z tester will not read an induced voltage? Is there a way to eliminate the induction (barring shielded cable)?

Thanks for responses.

 

[kingpb]

590.6 only requires the 50 amp receptacle to have GFCI protection if it is used for "supply temporary power to equipment used by personnel during construction, remodeling, maintenance, repair, or demolition of buildings, structures, equipment, or similar activities".

That is not the application in the OP. If anything we start in art 525 for the OP's application, I see no GFCI requirements in 525 at all. If this is in a fixed structure then 525.3(B) tells us to see 518 and 520 and utilize them if they apply, but I didn't really see any mention of GFCI protection scanning through those sections either.

Notice I said why I thought the AHJ would want it, did not say I thought it was required or he was justified for asking for it.

 

[gar]

140920-0007 EDT

cdcengineer:

An induced voltage will generally mean a voltage generated in a conduct from a change of magnetic flux lines linking the conductor.

Create two coils of 1 turn each and the same size. Connect one coil to a sinewave AC current source. To the second coil connect an AC voltmeter.

Place the second coil exactly in contact and in perfect alignment with the excitation coil. The coils are assumed to be insulated. There is near perfect flux coupling between the two coils. This results in the maximum possible induced voltage in the second coil.

Now place an aluminum foil sheet between the two coils and electrucally connected to one end of the excitation coil. There will be little change in the induced voltage at low frequencies. To substantially reduce the induced voltage you would need a magnetic shield between the coils.

Remove any shield. Gradually move one coil away from the other. As the coils are separated the induced voltage is reduced because fewer flux lines are linking the coils.

Go back to the original configuration. Modify the second coil so that it becomes two coils each covering half of the excitation coil, but with one of the two coils flipped upside down. Now the induced voltage is zero.

.

 

[hurk27]

I have seen some instruction sheets with GFCI's that have a max circuit length on the load side of the GFCI for this very problem, we have a city park where they used to hold a jamboree/festival every year, like you we installed GFCI protected circuit's with very long runs to 20 and 30 amp receipts through out the park for vendors and camper trailers as it was a whole weekend event, and we had the same problem, many of them we just put in GFCI receipt's, but they didn't like it because they would get stolen, so we found that twisting each set of load side circuit conductors would stop the coupling between the circuits, we would cut all the wires to the length needed then use a drill and twist them till we had about 3 turns an inch, this would put an 180? out of phase field between the conductors every other twist and cancel the coupling out just like they do with cat3,5, and 6 network cable never had a problem since, of course the festival was canceled a couple years later when they figured out it was costing the city more then they were making:blink:

 

[gar]

140920-0912 EDT

The word "conduct" in the first sentence of my post numbered 11 should be "conductor".

A long loop consisting of two parallel wires in a cable and an open circuit end, and adjacent to another similar loop with the second loop closed and a moderate current of 10 to 100 A at 60 Hz will have only a small voltage magnetically induced in the open circuit loop. In turn virtually no current induced.

Short the load end of the previously open circuit loop, and a small current will flow. Or at least use a low impedance load.

If the secondary loop is an open circuit, then coupling into the loop has to be resistive or capacitive.

With twisted wires on the output side of a GFCI it is likely that there will be some cancelation of capacitively coupled current.

.

 

[PetrosA]

Wayne is correct. Many GFCI products come with max. circuit lengths specified in the instructions.

Square D states with regard to QO series:

Do not connect to more than
250 feet of load conductor
for the total one-way run to
prevent nuisance tripping.

I wasn't able to find the information from other manufacturers online, but I'm pretty sure you'll find a max length in the instructions for the breaker you have.

 

[don_resqcapt19]

Wayne is correct. Many GFCI products come with max. circuit lengths specified in the instructions.

Square D states with regard to QO series:

Do not connect to more than
250 feet of load conductor
for the total one-way run to
prevent nuisance tripping.

I wasn't able to find the information from other manufacturers online, but I'm pretty sure you'll find a max length in the instructions for the breaker you have.

My understanding that the maximum length issue is based on conductor leakage current and not on and induced or coupled voltage from other circuits in the same raceway or cable.

 

[jim dungar]

My understanding that the maximum length issue is based on conductor leakage current and not on and induced or coupled voltage from other circuits in the same raceway or cable.

I agree, or else there would be some type of application note, from some manufacturer, that discussed proximity to other conductors.

 

[PetrosA]

My understanding that the maximum length issue is based on conductor leakage current and not on and induced or coupled voltage from other circuits in the same raceway or cable.

I agree, or else there would be some type of application note, from some manufacturer, that discussed proximity to other conductors.

Yeah, because manufacturers are known for telling us all we need to know

Why would conductor leakage current not apply to separate circuits? The only thing separate is the hots (maybe) whereas the neutrals are common, and monitored for each pair of hots. If you've got leakage current from circuit A leaking onto the neutral or ground of circuit B, what's to stop one from tripping? The fact that this happens when the second breaker is closed leads me to think that it happens when the field develops around the wires as capacitance builds up when it's energized. Wayne's workaround to twist the wires would probably solve the problem.

 

[gar]

140920-114 EDT

My guess is still capacitive coupling from one hot phase to the other hot phase.

In the original post it is stated that for either breaker, if only one breaker is turned on, then there is no tripping.

Tripping only occurs when the second breaker is turned on. I believe it is not known which of the two, or if both breakers trip, or how random is the tripping.

.

 

[kwired]

My understanding that the maximum length issue is based on conductor leakage current and not on and induced or coupled voltage from other circuits in the same raceway or cable.

My understanding as well. That leakage could easily be resistive or capacitive in nature just by simply having two conductors in a raceway or cable, inductive loss needs a core/coil type of condition.

 

[don_resqcapt19]

... Why would conductor leakage current not apply to separate circuits? ...

It does apply to separate circuits, the notes that talk about long lengths don't say anything about multiple circuits.