I've seen some reports that anecdotally mention that they do NOT see any differences compared to the IEEE 1584 calculation. It might be simply a numerical anomaly or not. The calculation itself is purely an empirical curve fit so there is no theoretical explanation as to any physical meanings behind the IEEE 1584 empirical equation because by nature, there isn't one.
The article below is actual testing that tested the ungrounded/resistance grounded concept and found that you get a 3 phase arcing fault but that the A-C fault is actually more like A-N-B where N is the surrounding housing. Given that in a solidly grounded system it can just as form A-G or C-G as the third arc and there is no need for a double line-to-ground fault (LGL), the reduction in the amount of arcing columns might provide some explanation here. Obviously impedance (through the grounded connections) plays a role for the third and fourth arcs. This is purely a guess on my part. I have no documentation to back it up.
http://www.neiengineering.com/wp-conten ... esting.pdf