I tend to agree with Bob's (petersonra's) comments in post 18. Money spent on this 'repair' is money that won't be spend on other repairs. Even if the management decision is made to do this particular repair, I would suggest that you list all of the 'hazards' found in the plant electrical system, rank them in order of most dangerous to least dangerous, and start spending the money on the most dangerous ones first.
IMHO your current install does not meet code (and thus is a liability issue), but is also not particularly dangerous. Also, the current install is nearly 40 years old, and you might want to replace it as part of a PM cycle...but on a reasoned out schedule rather than 'OMH hazard replace it yesterday!!!'
The only part of the grounding that is inadequate is _in the conduits_. A large enough EGC has been run to the equipment at the end of the conduit, and that EGC roughly parallels the path of the supply conductors. As wawireguy mentions in post 15, if _all_ of the phase conductors were in a single large conduit, then the single 3/0 EGC would be sufficient (of course derating would kill the phase conductor ampacity, but...) Since the installation was run in PVC, there will be only small issues with 'magnetic loop area' caused by the spacing between EGC and conductors (though it might pay to check for rebar loops around the individual conduit).
Now it is true that having 3 parallel 3/0 EGCs would be a better EGC than the single 3/0, and furthermore the smaller loop area would mean lower EGC impedance, but I bet that if you actually measure the EGC impedance you will find that it is more than up to the task of proper fault clearing.
_Inside_ the conduit the lack of an EGC means that the only fault clearance path is to another phase or to neutral. Sit down and evaluate what a fault inside the conduit will do. IMHO the lack of the EGC doesn't change the risks much in either direction. If something causes a cable to cable fault inside the conduit, I don't think that the EGC would have a big chance of preventing a phase-phase or phase-neutral fault. It is possible that the fault could hit the EGC first and cause the breakers to trip, but IMHO it is just as likely at anything that causes the fault would just damage a couple of conductors, and likely exclude the EGC entirely.
Finally there is the issue of conductor ampacity. What is the ambient temperature of the duct bank? Is it possible that the actual ampacity of the conductors, as adjusted for temperature, totals 1200A? Could you get the ampacity of the conductors evaluated under 310.15(C) engineering supervision? As others have asked, what is the actual load? Can you change OCPD to a lower value rather than increasing the conductor size to match the current OCPD? Could you add junction boxes and splice in short lengths of 600kcmil conductors, using 90C rated splices (assuming that the conductors in the conduit had 90C ratings.)
I would strongly consider the approach suggested by dana1028; if you can make 3x500kcmil conductors meet your ampacity requirements, then you could use one of the 500kcmil conductors already in the conduit as the EGC, and make all your changes in the conduit currently housing the EGC.
Good Luck! Let us know what transpires!
Jon
