Thanks for the replies.
It seems like the PV GFDI requirement was created mostly with grid-tied systems in mind, where the PV array is really the extent of the DC system. In this way, I can see why it would make sense. If you have a PV array mounted on a wooden roof in a battery-less grid-tied system, you really have no good way of protecting against fire from ground faults in the PV system unless you have GFDI. Then at least maybe someone will notice the problem quickly...
I wouldn't agree with all of that - I think the code making panel's concern was simply to prevent PV conductors from shorting to each other and causing fires (see next comment). This is why there's an exception if the DC conductors don't enter a building, because then the fire isn't on a structure with people inside. However I would agree that the code has never very adequately addressed how to integrate PV GFDI with a directly paralleled battery system that was traditionally grounded, and that since most systems are grid-tied there probably hasn't been any real pressure on them to do so.
... I guess with a strictly PV based DC system this is a moot point anyway, since the overcurent protection devices for the PV array are sized larger than the maximum current output the array can generate. You also can’t “de-energize” a PV array. Am I on the right track here?
Yes, you are getting it. If the grounded PV conductor faults to bonded metal, and that is not detected and fixed for years, and then the ungrounded PV conductor also faults to bonded metal, then you will have uncontrolled arcing that no overcurrent protection will stop. i.e. fire.
Again, where this becomes a concern for me is that the PV array is not the only source of power in the DC system I am working with. In fact, it is the least concerning part. You couldn’t get this PV array to catch fire with a can of gasoline and a blow-torch. It’s mounted to metal racking which is sitting on a steel pole mount 10 feet in the air in the middle of dirt/concrete clearing. The conductors run through metallic conduit strapped to concrete blocks.
Again, there's an exception to the GFDI requirement for that sort of install if the DC conductors don't enter the building. But if you're DC conductors enter the building they could still short to each other (or via a double fault) inside the building and cause a fire there. The concern isn't the PV array catching on fire. The concern is the PV power source conductor ssetting something else on fire because there is no feasible way to set up overcurrent protection to stop that.
What concerns me is the large battery bank which is sitting in a metallic rack connected to a metallic inverter by metallic conduit, all of which are readily accessible/touchable inside the building.
So, setting aside the code requirement and speaking purely theoretically, does it really make sense to unground the battery rack just so that the GFDI on the charge controller will work? Honestly, no one besides me will ever notice the GFDI indicator. Or is the battery rack less of an issue then I am making it out to be?
No it doesn't make sense to unground the battery source. That's just robbing Peter (battery safety) to pay Paul (PV safety). I'm sure arguments could be made about which danger is more serious but the proper approach would be to address both.
I'm inclined to say that if you really care about the safety issues then you need to address your PV output equipment. You could, for example, install
this device or a similar one on the PV circuit, and get a new charge controller without GFP that is compatible. Unless the charge controller you have has an option to not use the GFP device.
FWIW, I think the direction technology is going is towards ungrounded systems and fancier ground fault protection than the fuse method can provide. You've got a relatively old-school setup.