Parallel arc fault prevention for PV

[MWh_Pro]

I've just gotten done reading an article from NREL on arc fault detection and mitigation for PV, and I've been thinking: what's a practical strategy for reducing parallel arc faults?

For your reference, here's the article to which I am referring: "Low Cost Arc Fault Detection and Protection for PV Systems" by Scott McCalmont / Tigo

The article mentions that a major difference between series arc faults and parallel arc faults is that the former is driven by the connected load (e.g. inverter), whereas the latter is driven by the source (PV modules). This makes extinguishing parallel arc faults inherently difficult, because removing the load will actually provide more energy to the fault. Proposed solutions include safely shorting the PV source to ground (difficult), or module level electronics that disconnect the PV source at the junction box or before the string wiring (probably less difficult). One major connection that Scott makes is that parallel arc faults can be the end result of undetected ground faults. In that sense the best defense is a good offense, and that in lieu of a Type 2 (parallel) AFD/AFCI, it is likely that most parallel arc faults could be prevented by more sensitive ground fault protection. What are your thoughts on this? Do you see a distinct advantage in preventing parallel arc faults using "floating"/ungrounded PV arrays over the more common grounded ones (due to the greater sensitivity of GFI's in such inverters)?

Until there are listed parallel AFD devices, what do you think is an acceptable plan for parallel arc fault protection? Below are some ideas:
- Module level optimizer with Type 1 AFCI built in. Any combiner boxes are basic and without detection hardware. The noise spectra for Type 1 and Type 2 faults is essentially identical, so this probably provides adequate protection for both types.
- String inverter with Type 1 AFCI and standard GFCI included
- Type 1 AFCI enabled combiner boxes and standard GFCI @ inverter
- Type 1 AFCI combiner box w/ built in GFCI

I know there's a lot here, and I appreciate any input you may have. I've been studying this stuff a bit and there aren't many people who I know personally who are well-versed on the subject. I figured I'd try my luck at the Mike Holt Forum, as the quality of responses tends to be quite good =)

Thanks in advance,
Dave

 

[jaggedben]

Some thoughts...

1) The primary practical means of preventing parallel arc faults is GFDI, combined with attention to system performance and maintenance. In my estimation parallel arc faults are mostly likely to be the result of a ground fault on one side of the circuit, followed by a second ground fault on the other side at a later time. (See next thought.) Addressing and correcting a GF error promptly after seeing it on an inverter is probably the single most reliable way of preventing parallel arc faults from occurring in the first place.

2) The GFDI 'blind spot' revealed by the Bakersfield fire still needs to be addressed in the code. In my opinion ungrounded systems with sensitive electronic GFDI (i.e. not a 5amp fuse) are the way to go on this, as a practical matter. So yes, I see an advantage in 'floating'/ungrounded arrays. (I also think they are for installation crews.)

3) As far as your list of options, I think all of them have their appropriate uses. The first two are appropriate for residential. The second two are good options for any systems large enough to require combiners. An ancillary advantage of combiner boxes with GFDI would be to aid in finding and fixing ground faults, as well as preventing whole-system shutdown. But there would still have to be adequate alarm raised to ensure someone would go out and fix ground faults.

4) As far as actually shutting down the source in case of parallel arc faults, module level electronics may be the only real answer.

 

[MWh_Pro]

Hi Jaggedben, that was to the point, and incredibly helpful.

Thanks,
Dave

 

[GoldDigger]

4) As far as actually shutting down the source in case of parallel arc faults, module level electronics may be the only real answer.

It was a little hard at first to get my head around the concept that one way to clear a parallel arc fault is to provide a deliberate short of lower impedance, but that is certainly an option, rather than providing a module level disconnect contact.

But since emergency shutdown is also an upcoming issue, that may determine the way that the module level circuits evolve for both purposes.