I have personally seen the results of having _separate_ ground rods in a structure. This is not directly applicable to the current discussion, since the grounding electrodes in question should have been bonded under the NEC:
At a rural location, perhaps 150 feet from a transmission right of way, some friends have an old building. Telephone enters from the northwest corner of the building from overhead wires. On the outside of the building at about 6 feet above the ground there is an old school lightning arrestor box, from there the phone wire enters the building. A (at a guess) 4 or 6ga aluminium conductor goes from the lightning arrestor box down the outside wall to a ground rod.
Electrical is supplied by a single phase overhead line to a transformer on a pole about 50 feet away from the building, and then underground to a meter on the northeast corner of the building. I was unable to locate the grounding electrodes when I was looking at this (jes visitin' friends, not working, not much electric motor research to do in an old church in the middle of upstate New York), although there appeared to be a GEC in the panel.
When I was going up to visit, they mentioned that the phone line was not working, and that a local ranger who sometimes looks in on the property said that there had been a pretty impressive thunderstorm recently.
I'd found that the phone wire _inside_ the building, starting at floor level directly below where the wire had entered the building, and extending for 3-6 feet was burnt up in a most impressive fashion. Every couple of inches, there was a shard of copper sticking out through the insulation, looking just like corona discharge solidified. The wood floor and baseboard around the wire had a number of small round dark patches, with roughly the same spacing. I can only guess that I saw the after effects of enough current and voltage to melt the copper and physically move it around.
Interestingly, there was absolutely _no_ insulation damage on the wire at the location of the lightning arrestor. This even though, in addition to being electrically connected to the appropriate terminals, the insulated wire was randomly wrapped around the ground electrode terminal. If the wire were at elevated voltage relative to the ground electrode, then there should have been insulation damage at this point.
My _suspicion_ is that a lightning strike to the transmission line cause lots of local earth current. The telco ground rod was placed at an elevated potential relative to the rest of the building, and this ground rod actually _energized_ the telephone wire rather than dissipating the surge. The wire was placed at elevated potential relative to surrounding materials, thus the discharge. The building is stone and wood, with metal plates that bolt various beams together, so not much 'building steel' to speak of. However there is _old_ RMC throughout the building, enameled black heavy threaded pipe, possibly behind/below the area where the phone wire 'discharged'.
Now the NEC would _require_ the telco ground to be bonded back to the electrical system ground, so this _shouldn't_ have been an issue.
But it gets me to think that if you plant a grounding electrode in a building, for any reason, this grounding electrode _should_ be joined to the building grounding electrode system.
I understand that 250.54 permits 'supplementary' grounding electrodes to sidestep the bonding requirements of 250.54...and if a piece of equipment doesn't even have a circuit feeding it you would not be _required_ to bond back to the building ground system. But my hunch is that doing so is a _good_ design choice.
-Jon
