DC On Metal Roofs

[sketchy]

I have a job with a standing seam roof. I'm using 3 CPS inverters, 36 kW each, and was going to have them inside of the array to avoid using MCI's. The more I think about it the more I'm worried about a potential fault and how there's no way to stop PV production unless I use MCI's. Even with proper bonding and grounding, there's no way to prevent the energizing of the metal roof. Am I missing something?

 

[jaggedben]

The inverters will have ground fault protection and cease production if one side of the circuit faults to the roof and the roof is sufficiently bonded to the inverter EGC. The DC circuit would then be grounded on one side, unintentionally. Only a second fault on the remaining ungrouded conductor would cause an arc. There should be no serious shock hazard to anyone not working on the wiring. However the long term safety relies on the first fault being addressed in a timely manner. So that's what MCIs could do (besides perhaps being a way to do rapid shutdown). That is, they could reduce the voltage and power available to the second fault or exposed to the personnel working to find the first one.

 

[Carultch]

"the roof is sufficiently bonded to the inverter EGC"

I've wondered for years how you are expected to make sure of that in practice.

Setscrew clamps are designed with rounded tips, so as not to penetrate the coating of the metal roof. This is one of the selling points of S-5 clamps, that they keep the coating intact and don't introduce corrosion or leaks. This is OK for a conductive coatings like galvalume, or galvanized sheet steel, where you'd automatically get your ground through the clamps that mechanically secure the racking.

However, a lot of metal roofs don't necessarily have a metallic coating. Powder-coating or paint would be an impediment to electrical continuity. And for metal roofs like trapezoid fold roofs, where you attach with penetrating screws, are commonly installed with rubber sealing washers, which would also hinder electrical continuity.

Are metal roofs bonded to building steel inside the ridge cap? Or somewhere else that's hidden from view?
Perhaps the clips that secure the roofing to the steel purlins, might bond the metal to the purlin, if the metal roofing is only coated on one side.
Metal roofs may also be installed on wood decking or wood purlins, which wouldn't help you for obvious reasons.

 

[electrofelon]

I've wondered for years how you are expected to make sure of that in practice.

Setscrew clamps are designed with rounded tips, so as not to penetrate the coating of the metal roof. This is one of the selling points of S-5 clamps, that they keep the coating intact and don't introduce corrosion or leaks. This is OK for a conductive coatings like galvalume, or galvanized sheet steel, where you'd automatically get your ground through the clamps that mechanically secure the racking.

However, a lot of metal roofs don't necessarily have a metallic coating. Powder-coating or paint would be an impediment to electrical continuity. And for metal roofs like trapezoid fold roofs, where you attach with penetrating screws, are commonly installed with rubber sealing washers, which would also hinder electrical continuity.

Are metal roofs bonded to building steel inside the ridge cap? Or somewhere else that's hidden from view?
Perhaps the clips that secure the roofing to the steel purlins, might bond the metal to the purlin, if the metal roofing is only coated on one side.
Metal roofs may also be installed on wood decking or wood purlins, which wouldn't help you for obvious reasons.

I have heard of people removing the ridge cap and bolting lugs with star washers to each panel.

I'm skeptical that S5's wouldnt sufficiently bond the roofing, perhaps some testing is in order

 

[pv_n00b]

I don't think you are missing anything OP. You could put all the DC conductor behind some kind of protection in the array mounting to isolate it from contact with the roof panels. Then if there is a fault it is to the grounded array and not to the roof panels.
From what I have seen people kind of hold their nose and look the other way when it comes to metal roof grounding. It is hard to argue that putting modules on a metal roof with unprotected DC conductors under them is not likely to energize the roof in the event of a fault, but installers and most AHJs agree to not worry about it or make some pretty janky arguments about why it's not needed. No one wants to put a ground lug on every roof panel and run an EGC all around the roof.

 

[Zee]

I've wondered for years how you are expected to make sure of that in practice.

Setscrew clamps are designed with rounded tips, so as not to penetrate the coating of the metal roof.

Yeah I went around and round this thought process years ago.
I settled on this reasoning:
Either
the metal roof is so conductive that a lose, live wire would fault to it.... in which case our solar racking (that is clamped and bolted to it) will also de facto be sufficiently bonded to it.
Or
the roof is so super insulating, that even several (hundreds?) setcrews, sheetmetal screws, bolts etc do not bond to it....in which case there is also no worry from any damaged conductor.

I figure either way we are covered.

 

[jaggedben]

Yeah I went around and round this thought process years ago.
I settled on this reasoning:
Either
the metal roof is so conductive that a lose, live wire would fault to it.... in which case our solar racking (that is clamped and bolted to it) will also de facto be sufficiently bonded to it.
Or
the roof is so super insulating, that even several (hundreds?) setcrews, sheetmetal screws, bolts etc do not bond to it....in which case there is also no worry from any damaged conductor.

I figure either way we are covered.

Yeah that is what I was going to say. Either the roof coating is an insulator or it's not. Fwiw I have one data point from direct experience, and it was not.