Bob,
Thanks for double checking those code references. 250.122(F) describes parallel conductors in raceways _or_ cables, and I didn't notice that 250.122(F)(2) further restricted to multiconductor cables only.
Regarding 240.8, I see how one could argue that this might or might not apply to the situation of using limiting lugs on the individual cables in a parallel set. Current flow in the conductor set clearly is in parallel to the final load, but each limiting lug is in series with a different conductor; so the limiting lugs are _not_ directly in parallel.
Ryan,
What happens if you have the undersized EGC in a metallic raceway, and you have a phase to insulated EGC fault that doesn't involve the raceway? It seems to me that the problem that 250.122(F) tries to address is all of the fault current flowing in a single EGC from the parallel set. I would suggest that if a variance were permitted, it would be better to stretch 250.122(F)(2) to include conductors in raceways, then to try to allow an undersized EGC on the basis of the _potentially_ parallel raceway.
Hmm, which leads to another silly 'what happens when you stretch the code' questions on this topic, relating 250.122(A) to conductors in parallel. Does 250.122(F) every force the EGC to be larger than the phase conductors, or does 250.122(A) prevent this? If a 600A service were run as 5 1/0s Al in parallel, how large should the EGC be
-Jon
