It was blamed on bonding, loose connections and even the transition type of the ATS.
Those are statements. A statement is either true or false. That would be a good place to start, with the statement and asking is that true or false. Is there evidence proving or supporting the statements. It reads like no one knows if it's true or false.
The UPS front end very well could be 3 wire no neutral but the UPS front end filters, caps and spd's, will be Y connected and grounded, to the equipment ground. With a widely spread out distant layout like that, there will be substantial current flow over the EGC, which bonds back to the GEC. If you go right to the point were the EGC bonds to the GEC common busbar and Amp clamp it, it would not surprise me to see hundreds of Amps on the main bonding jumper and tens or more of Amps on the EGC's and GEC's at their common grounding busbar.
Not to say this is bad, just that it is an essential system performing and essential duty, more than normal because it is a USP of that size. For sure the grounding system is noisy. Handled properly that could be normal.
The Y services would have to be solidly grounded (the UPS is looking for a solid ground Y at the input) and the neutral is probably carried to the first service disconnect, .but the loads can all be 3 wire and nothing needs to be connected to the neutral as a load. For a dual source bus I would say this can be preferable, no neutral connected loads at the dual source bus.
Actually if you ask me, the type of noise on the UPS front end should be on a supply neutral, but they build them as 3 wire as I have seen and ground front end components with substantial current and balancing power flow.
To start with it needs a good survey with Amp clamps and IR scans to see what it is really doing, and someone with a good eye for it with their eye on it. You want a ghostbuster to look at it.
What modules were burning out, line or load side. The ATS, line or load side. The ATS might be a static switch with quarter cycle open transition, unless you have paralleling gear, which it sounds like ( is in question ).
I have seen beyond my comprehension, but people in a position and duty to know who do know and see it (hundreds of people in one place it was so big), they do leave things for whatever reason but the good ones probably fear job loss. Likely there could be something to find, even something that was seen and not escalated.
Examine all of the main and system bonding jumpers first, line and load side. make sure they are physically there by eye and with an Amp clamp. Also look for the secondary main and system bonding jumpers, the ones that may be there but are not supposed to be. Improper main or system bonding jumper could cause your trouble. The the usual, line side Voltage and current balance, load side Voltage and current balance.
A recording instrument, a data logger on the system at the time of switching transitions would be one of the things I would want to see. What it shows could be surprising.
