Did some more digging. The key is to look at the extremely long final rule text which shows all the justifications. TVA proposed using 18" based on the same anthropomorphic data that I would assume IEEE 1584 used and the same data I've looked at myself before. OSHA disagreed and the fundamental difference comes down to this. Everyone can agree more or less that if you are working in enclosed gear, there will be a screwdriver or finger tips involved. In overhead line work, the employee is grasping the cable directly in some cases, so OSHA felt that the distance between the finger tips and the crotch of the hand between the thumb and fingers (2.4") should be included which reduces the distance to 15". For enclosed equipment, it stays with the IEEE 1584 recommended distance of 18". Since OSHA 1910.269 is primarily focused on outdoor, overhead equipment, they chose to use 15" as the basis for comparison in all of their tests.
See:
http://www.dol.gov/find/20140401/2013-29579.pdfStarting around page 608.
Note that you can still use CYME (NESC) data but you'd have to scale it for the shorter distance. This is easily done because the exponent will be 2 for open wire work. And also note that when using a hot stick, OSHA agrees that the 15" rule goes out the window. All my calculations with hot sticks drop the incident energy way down to almost a non-issue.
As far as model acceptance goes, Table 12, page 616, is a summary of what's in the text. Suffice to say that for 601-15 kV, OSHA accepts Lee, IEEE 1584, AND ArcPro for single phase arcs although obviously ArcPro is going to produce the lowest value. For 3 phase arcs, they only accept IEEE 1584 or Lee and IEEE 1584 will produce the lowest results. Above 15 kV, its ArcPro all the way.