A Cut Bonding Jumper, No Production Loss: Can It Wait?

Correct answer: B) It has to be restored before the array is left in service: a broken bond can stop the ground-fault protection seeing a fault on the isolated section, so the defect disables a safety function rather than degrading one

Bonding does nothing at all until something else has already gone wrong. Its entire job is to give fault current a low-impedance path back to where the protective device can see it, so that a fault trips something instead of sitting there. That means a year of uneventful operation is not evidence of safety. It is evidence that the protection has not been called on, which is the condition under which a broken bond and an intact one look exactly alike.

What the break actually costs you is worth being specific about. If a conductor faults to the frame of a module on the isolated section, the current has no route back to the inverter's ground-fault detection, so the detector does not operate and the array keeps running. The metalwork on that section is now energised and stays energised, and the next low-impedance path to earth may well be the next person to put a hand on a rail in the rain. The system will not tell you any of this. It will keep producing exactly as before.

The owner's question also contains a false trade. Restoring a bonding jumper between two racking sections is a short job on the roof, and it does not require the array to be taken offline for a production day. Being able to say that clearly, rather than accepting the premise that safety work means lost revenue, is part of doing this job well.

The O&M point is that the damage was done by somebody else's trade. A roofer who lifts and refits membrane around an array has no reason to know what a bonding jumper is, and no obligation to tell anyone they cut one. Nothing in the monitoring data marks the day it happened. Faults introduced by other contractors working on the same roof are a category of defect that only ever surfaces because someone inspected, and they are invisible to every other method you have.

Why the other options are wrong

A) It can wait for the next scheduled outage — a year of normal operation with no production loss is evidence the array is safe as it stands

This treats absence of consequence as evidence of safety, which is the specific reasoning error that protective systems invite. A fuse that has never opened, an alarm that has never sounded and a bond that has never carried fault current all look identical to one that does not work. The year of normal production is real and it is not about this: bonding has no effect on how much energy the array makes, so production data cannot report on it either way.

C) Nothing is actually missing — the two sections are bolted together, so the mechanical connection provides the bond

Two things are wrong here. Aluminium racking is usually anodised, and the anodising is an insulator, so a bolted joint through it is not a reliable electrical path however tight it feels. Some racking genuinely is listed to provide integrated bonding through its own connections, but that is a listing you verify for the product in front of you, not something you infer from the fact that two parts are bolted. And there is a clue in the scenario itself: somebody fitted a bonding jumper across that joint, which tells you the design did not rely on the bolts there.

D) No action is needed on the bond, because the inverter’s ground-fault detection protects the whole array regardless

This gets the relationship backwards. Ground-fault detection does not protect the bonding; the bonding is what allows the detection to work. The detector sees a fault because fault current returns to it along the equipment grounding path, so cutting that path on one section does not leave the detection covering it anyway. It removes the section from the detector's view while leaving everything looking normal from the inverter.

References

• NABCEP OMAT JTA v.2026.5 — Domain 2, Task 3.H: "Grounding and bonding (e.g., system and equipment)"
• NABCEP OMAT JTA v.2026.5 — Domain 3, Task 2.F: "Grounding and bonding methods (e.g., integrated grounding, listed grounding components, star washers)" — the JTA naming integrated grounding and listed components separately is the reason option C is a listing question rather than an assumption. No electrical code article is cited here; the JTA names OSHA and NFPA as the standards in scope and refers to electrical codes only in general terms.