Arc Fault Protection

[nosparks1]

Two part question,,, How is PV Arc-Fault protection on D/C circuits according to NEC 690.11 being accomplished and if it is built into the inverter the way Ground Fault Protection is, how does it provide protection since the D/C wiring is ahead of the inverter?

 

[ggunn]

Two part question,,, How is PV Arc-Fault protection on D/C circuits according to NEC 690.11 being accomplished and if it is built into the inverter the way Ground Fault Protection is, how does it provide protection since the D/C wiring is ahead of the inverter?

At SPI I asked a rep of an inverter company which will soon introducing integral arc fault protection into their designs how they will detect an arc. He told me that their detection circuitry will "listen" for a noise signature of an arc on the DC inputs. One company I spoke to is coming out with combiner boxes with contactors which will open to de-energize the PV output circuits when the inverter tells them to.

 

[jaggedben]

I think most AHJs are not enforcing this yet, either because they haven't fully adopted the 2011 code cycle, or because the third paragraph of 90.4 comes into play. As ggunn's post shows, the latter will start to change.

if it is built into the inverter, how does it provide protection since the D/C wiring is ahead of the inverter?

Note 690.11(2)(a). The system is allowed to interrupt the circuit at the inverter or charge controller.

Why is it not necessary to interrupt the circuit closer to the panels? Because in a PV system, an arc under load is a much, much more dangerous arc than an arcing condition that is not under load. It's also more likely to happen. I heard Bill Brooks explain this in a seminar. Under load, an inverter's MPPT system will actually respond to the arc in a way that increases and maintains the stability of the arc. Remember, the MPPT is trying to make the PV panels operate at maximum efficiency. So when there's an arc, it will find the point at which the arc operates at maximum "efficiency." The most likely point of this happening is at connectors between modules. So shutting down the inverter will open the circuit and actually cause the arc to go away.

To put it another way, the main purpose of 690.11 is to interrupt an arc within one conductor when the system is under load. The chance of an arc between the + and - conductors is much less, especially since the ground-fault detection will likely kick in before it happens.

It is also possible to interrupt the circuit at each panel using at least one brand of power optimizer, which is somewhat more safe. This is the method in 690.11(2)(b), but you can use either method.

 

[ggunn]

I think most AHJs are not enforcing this yet, either because they haven't fully adopted the 2011 code cycle, or because the third paragraph of 90.4 comes into play. As ggunn's post shows, the latter will start to change.



Note 690.11(2)(a). The system is allowed to interrupt the circuit at the inverter or charge controller.

Why is it not necessary to interrupt the circuit closer to the panels? Because in a PV system, an arc under load is a much, much more dangerous arc than an arcing condition that is not under load. It's also more likely to happen. I heard Bill Brooks explain this in a seminar. Under load, an inverter's MPPT system will actually respond to the arc in a way that increases and maintains the stability of the arc. Remember, the MPPT is trying to make the PV panels operate at maximum efficiency. So when there's an arc, it will find the point at which the arc operates at maximum "efficiency." The most likely point of this happening is at connectors between modules. So shutting down the inverter will open the circuit and actually cause the arc to go away.

To put it another way, the main purpose of 690.11 is to interrupt an arc within one conductor when the system is under load. The chance of an arc between the + and - conductors is much less, especially since the ground-fault detection will likely kick in before it happens.

It is also possible to interrupt the circuit at each panel using at least one brand of power optimizer, which is somewhat more safe. This is the method in 690.11(2)(b), but you can use either method.

There is also the question of whether it's a series or parallel arc. A series arc occurs when a connection opens under load and the current continues to flow through the air gap; shutting down the inverter will quench it. A parallel arc is harder to detect and may persist after the inverter is shut off. GFP shutdown will quench a + to ground arc but not a + to - arc. The Bakersfield fire was the result of a + to - parallel arc, I believe.

 

[jaggedben]

There is also the question of whether it's a series or parallel arc. A series arc occurs when a connection opens under load and the current continues to flow through the air gap; shutting down the inverter will quench it. A parallel arc is harder to detect and may persist after the inverter is shut off.

That's all correct. The series arc is the one that will be prevented by arc-fault interruption at the inverter. To prevent a parallel arc one would need electronics on the panels. That said, the series arc is the one that is more likely to occur and less likely to be caught ahead of time by the GFP, because one way a parallel arc can occur is with both + and -faults to ground.

GFP shutdown will quench a + to ground arc but not a + to - arc. The Bakersfield fire was the result of a + to - parallel arc, I believe.

Regarding the Bakersfield fire, you're right, that was a parallel arc (or fault). However, it was the result of both + and - faults to ground. Bill Brooks has pointed out that the issue there was really the inadequacy of the ground-fault protection. If the GFP had tripped on the first fault then it probably would have been fixed before the second fault occurred. It was the second fault that created the parallel arc and fire.

 

[nosparks1]

Thanks for the explanations.

 

[ggunn]

That's all correct. The series arc is the one that will be prevented by arc-fault interruption at the inverter. To prevent a parallel arc one would need electronics on the panels. That said, the series arc is the one that is more likely to occur and less likely to be caught ahead of time by the GFP, because one way a parallel arc can occur is with both + and -faults to ground.



Regarding the Bakersfield fire, you're right, that was a parallel arc (or fault). However, it was the result of both + and - faults to ground. Bill Brooks has pointed out that the issue there was really the inadequacy of the ground-fault protection. If the GFP had tripped on the first fault then it probably would have been fixed before the second fault occurred. It was the second fault that created the parallel arc and fire.

One small quibble - the problem wasn't the faults to ground, per se, that blew things up, but the positive and negative conductor faults to each other. It's been a while since I looked at the post mortem on that fire, but as I remember the root cause was an unsecured conduit joint that pulled out and successive ambient heating and cooling caused the metal edge of the conduit end to saw through the positive conductor which shorted to ground, and then the ensuing arc melted the insulation on the negative conductor and it was off to the races. It may be true that adequate GFI would have stopped the original arc, but if the conduit end had cut through both conductors' insulation, then GFI couldn't have stopped it. It's also possible that the negative conductor was the first one breached; ground faults on the negative conductor are much more difficult to detect. If that were the case, then when the conduit finally cut through to the positive conductor as well, GFI wouldn't have stopped anything.

One result of that fire has been that it has everyone paying a lot more attention to conduit couplers and expansion joints. Ounce of prevention, pound of cure.

 

[jaggedben]

One small quibble - the problem wasn't the faults to ground, per se, that blew things up, but the positive and negative conductor faults to each other. ...

Not according to Bill Brooks.
Read: http://solarprofessional.com/article/?file=SP4_2_pg62_Brooks

According to him when the fire started only the + conductor was damaged at the expansion joint. It was a previous ground fault at a different location that completed the circuit.

If the first fault had tripped the GFI and been repaired when it happened, then when the fault at the broken expansion joint tripped the GFI it could possibly have been repaired before both conductors were damaged and a fire causing circuit completed.

Brooks concludes that the need for different GFP methods is the "most important" lesson from that fire.

 

[ggunn]

Not according to Bill Brooks.
Read: http://solarprofessional.com/article/?file=SP4_2_pg62_Brooks

According to him when the fire started only the + conductor was damaged at the expansion joint. It was a previous ground fault at a different location that completed the circuit.

If the first fault had tripped the GFI and been repaired when it happened, then when the fault at the broken expansion joint tripped the GFI it could possibly have been repaired before both conductors were damaged and a fire causing circuit completed.

Brooks concludes that the need for different GFP methods is the "most important" lesson from that fire.

Ah, yes, I see, but it was still the positive to negative connection that started the fire; the ground was just the common node between them; if the conduit had been floating the fire still would have happened. The first fault was on the negative conductor, and a "normal" GFI in a negatively grounded system won't normally catch that. The second was on the positive conductor and the conduit completed the circuit. The inverter caught that one, but couldn't do anything about it. But certainly, effective GFI on the negative conductor would have prevented the fire. TL designs detect faults by differential current on the positive and negative conductors and would have caught the first fault.

 

[DanS26]

Some new products coming out on the market to address this NEC requirement. See here:

http://www.civicsolar.com/resource/arc-fault-detection

Will be interesting to watch how the market responds.

Dan