While the only one I have ever experienced was a transformer 480 primary with a 240 grounded phase (intentionally) secondary, the problem was, who ever hooked up the secondary bonded the X1/X3 connection instead of the X2/X4 and when the 480 volt system went to ground on H4 side of the primary, it caused this transformer to behave like an auto transformer, giving an additive voltage between the H1 on the 480 side and X2/X4 on the secondary side which gave us about 720 volts between the two systems, luckily nothing was effected since no loads sourced both systems, and was caught during a shutdown time, we checked all other control transformers and found five more with the X1/X3 bonded instead of the X2/X4, these were just small control transformers,
but what I can't see is how a inductive reactive circuit would rise the whole system voltage when you still have the transformer windings across the phases which would limit the voltage through its impedance? to me this would have to be in a very local loop isolated to where the inductive fault was placed, its been a while since I done major industrial work, but I cant for the life of me ever remember having a problem like this (other then the one we found above) this sounds like having an inductive kick back that is occurring continuously at 60 hz, which would make sense, but as it says this would be a extremely rare occurrence? Kind of like one of those What if's that code can't be enforced on?
As to the OP it sounds like a plant that has a bottling production line, which could meet the requirements of having an ungrounded system, sounds also like it was wired back before the requirements of ground monitors, yes it would be beneficial to install ground monitors because it will tell them when a phase goes to ground, and allow for a safe proper shutdown, and repair,
but even with the added info from don, this would still not effect a circuit phase going to ground, whether it is a feeder or a branch circuit, a OCPD will not open and everything will keep operating till a second phase were to go to ground, this would be nothing more that having a grounded phase system at this point, which to me would pose no additional threat, but could cause an event of un-necessary shutdown if a second phase went to ground before they repaired the first fault, so yes a ground monitor would benefit the production staying up and running, but to me nothing else. oh and ground/phase monitors don't trip anything, thats what a GFP does, which would not work on a ungrounded system.