NEC2017 - AFCI exception for non-building systems

[Carultch]

NEC2017's 690.11 section now has an exception to not require AFCI on ground-mount systems, provided certain conditions are met.

1. The system & its wiring are not on or in a building
2. The wiring is either
   1. Direct buried
   2. In a metal raceway
   3. In an enclosed metal cable tray

This exception seems to systematically exclude what I'd do by default on a ground mount system, which is wiring underground in PVC conduit. Possibly unintentionally, it just excludes this by omission. Or this may be intentional for reasons I don't understand. Yet it includes direct burial, which is less safe than wire in a PVC conduit. Direct burial is subject to physical damage from underground rock movement, while PVC provides a physical barrier against this, as well as the ability to future-proof and replace wire without re-digging.

I'd like to ask if wire underground in PVC is part of the condition of "direct buried" for this exception. And if it isn't, I'd like to ask if PVC underground was intentionally or unintentionally excluded from this exception, and why.

 

[SolarPro]

According to Bill Brooks, who sits on CMP4, buried conductors in conduit should also qualify for the exception.

The rationale is: 1.) that buried output circuit conductors can't start a fire; 2.) ground-fault protection will detect any arcing faults where output conductors are fully enclosed in metal raceways.

 

[Carultch]

According to Bill Brooks, who sits on CMP4, buried conductors in conduit should also qualify for the exception.

The rationale is: 1.) that buried output circuit conductors can't start a fire; 2.) ground-fault protection will detect any arcing faults where output conductors are fully enclosed in metal raceways.

Ok, thanks for the clarification. Any links where I can see anything to back up that interpretation?

A possible reason PVC underground was not included, is that a wire-to-raceway fault with metal conduit, or a wire-to-earth fault with direct buried wire, will trip the OCPD. By contrast, a wire-to-PVC raceway "fault" will go undetected, as the current hits a dead end where the wire comes in to contact with the PVC wall. You would still need a simultaneous fault on both polarities in the same neighborhood to cause an arc. And though the chance is low, PVC can burn, unlike metal.

The reason I ask, is that I'm not confident in direct burial as a basis of design that I'd recommend. And with the exception of a few serious applications, metal conduit underground is significantly more physical damage protection than you really need. So I'd prefer to stick with my existing underground wiring method of choice, if possible.

 

[SolarPro]

No links. But we're about to publish an article about the practical implications of NEC 2017. The authors were concerned about the lack of clarity w/ regards to buried PVC and reach out to Bill. This is the email report I got back:

Bill B. has just clarified that "direct buried circuits" may include circuits direct buried in conduit, therefore buried PVC may be included in the output circuit exemption for DC AFCI underground. Since the Code reference is to "circuit" not raceway or conduit, Bill says PVC conduit can be considered to be a direct buried circuit.

That's the best I got right now, other than explaining the intention behind the exception.

Remember that dc arc-fault protection is all about fire prevention. There's even an allowance to in 691.10 to forego 690.11 arc-fault protection if an engineer designs and alternative fire mitigation plan. Article 691, of course, applies to large-scale PV plants (>5 MWac).

 

[jaggedben]

Ok, thanks for the clarification. Any links where I can see anything to back up that interpretation?

A possible reason PVC underground was not included, is that a wire-to-raceway fault with metal conduit, or a wire-to-earth fault with direct buried wire, will trip the OCPD. By contrast, a wire-to-PVC raceway "fault" will go undetected, as the current hits a dead end where the wire comes in to contact with the PVC wall. You would still need a simultaneous fault on both polarities in the same neighborhood to cause an arc. And though the chance is low, PVC can burn, unlike metal.

A fault to metal might or might not trip any OCPD depending if it's a source or output circuit. It would more likely trip the GFDI. And anyway, that's GFDI, not AFCI, and neither of those will stop the arc from a sudden positive-to-negative fault.

I'm pretty sure that series arcs are the point of this section. A few years ago I attended a talk by Brooks and he showed a video of old MC3 connectors that had pulled just slightly apart, the contacts arced inside the connectors, and the connectors caught fire. I believe it's the main reason why connectors are now required to be of a latching or locking type. Nothing to do with faulting to other conductors or parts.

I'd tend to agree that the language should simply be changed to 'underground'. I think what we're trying to do here is prevent grassfires. I don't see fires starting underground, and there really shouldn't be any connectors or such there. We're worried about connectors and terminals in boxes and such.

At any rate, you still have conductors on the array that need AFCI protection, no? I guess it comes down to whether you can put that in a combiner and not have it be required between combiner and inverter.

 

[SolarPro]

That's the issue. UL has a dc arc fault standard, but it only goes up to 40 A. So we basically have dc arc-fault devices for source circuits and string inverter input circuits, but not for larger PV output circuits associated with central inverters.

Some AHJs may have ignored this issue under NEC 2014. You can't really get around it under NEC 2017, which basically makes it impossible to use a central inverter in a rooftop application. This is not just because of rapid shutdown, but also due to 690.11.

The exception under 690.11 carves out some ways to use central inverters in ground-mount applications, even though you don't have dc arc-fault on the PV output circuits leaving the combiners.

 

[Carultch]

At any rate, you still have conductors on the array that need AFCI protection, no? I guess it comes down to whether you can put that in a combiner and not have it be required between combiner and inverter.

Good point. Inevitably you'll have some of the system wiring outside of conduit, because few (if any) modern PV modules are built for raceways to enclose 100% of the wiring.

Does this mean that even when this particular exception applies for a ground-mount array, that you still need an AFCI combiner (or other AFCI equipment) to open upon arc faults originating on the string wiring upstream of the combiner?

 

[SolarPro]

Correct. The exception only applies to PV output circuits, and not to PV source circuits.

If the project qualifies for Article 691, then you can look at other fire mitigation options, provided a PE designs and documents the plan.

 

[PVfarmer]

I guess it comes down to whether you can put that in a combiner and not have it be required between combiner and inverter.

Good point. Inevitably you'll have some of the system wiring outside of conduit, because few (if any) modern PV modules are built for raceways to enclose 100% of the wiring.

Does this mean that even when this particular exception applies for a ground-mount array

Correct. The exception only applies to PV output circuits

Here's one option- don't use any PV output circuits. (DC AFCI compliant to UL 1699B)

http://www.smacore1.com/wp-content/uploads/2016/06/SMA-STP-CORE1-Brochure-US.pdf

And here's another, "enclosed wire management"-
http://solardock.com/

 

[Carultch]

Here's one option- don't use any PV output circuits. (DC AFCI compliant to UL 1699B)

http://www.smacore1.com/wp-content/uploads/2016/06/SMA-STP-CORE1-Brochure-US.pdf

And here's another, "enclosed wire management"-
http://solardock.com/

I'm not asking about proposing another solution, I'm asking to understand completeness of the rule. It seems like the exception doesn't really except anything from what we were already doing in 2014. It just acknowledges that 2014's rule couldn't be satisfied completely, and now allows you to not satisfy the blindspot in certain conditions.

And even if something other than a raceway or cable tray is enclosing the wiring, such as part of a racking system, does that count?

 

[jaggedben]

And even if something other than a raceway or cable tray is enclosing the wiring, such as part of a racking system, does that count?

If the concern is a fire starting then I gather it has to be completely surrounded by very fire resistant material. So for example a racking system that just encloses conductors between the rack and the flammable back-film of the module isn't really sufficient. I think maybe the reason that PVC wasn't specified is that it can actually burst into flame, unlike metal under atmospheric conditions. Of course if the PVC is underground then the flames aren't likely to get so far.