Recently an electrician at one of our facilities brought up a concern about arc flash potential while changing brushes on a generator. There's a great deal to consider with that question (I previously modeled the arc-flash hazard for a generator field breaker on another machine, but that's a story for another day...), but I was wondering if anyone else has come across this question?

Good question and I would look at performing a risk assessment of the task to determine if an arc flash event is reasonable and if so, utilize the DC table in NFPA 70E-2015

Well, that's my thought. I guess there are a couple scenarios to analyze:

1- Generator running with a field ground, person provides 2nd current path

2 - Accidental short between collector rings (I know, very unlikely)

The complication comes in that there is more than just DC fault current available. In fact, I'd suggest the NFPA table is much too simplistic to accurately asses. I'm just wondering if anyone else has heard this question asked?

I've worked with true DC generators. The tables are waaay off. As in totally unrealistic. The maximum power transfer equation in the annex is also waaay off compared to test work. How far off? Voltage dependent but a factor of 3-8 from tests. Ammermon's equation is better but still off by multiples. And as to AC+DC, the big difference is DC does not self extinguish (pass through zero). The maximum power transfer equation is literally Lee's equation in DC. Ammermon's equation accounts for how a DC arc responds to differences in current but still models it as 100% energy conversion to heat (not true). So you can just add AC and DC and use those equations.

I would say it depends on the type of generator - if we are talking AC generator with DC field brushes, I think the tables are probably conservative enough. Some, especially modern excitation systems, have controller limiters on field current. If you know what those are that would be a reasonable value to pick as the maximum DC current. Otherwise the maximum current would be the maximum field voltage divided by the field resistance. For static excitation systems this would be 1.35*Vac, where Vac is the incoming AC voltage for the rectifier.

I noticed this too. Why is this?

It's all in the input parameters. And it is probably the tail wagging the dog.

If we were developing a table from scratch then we have a couple choices for nominal voltages. Motors and generators are typically 150, 300, and 600 V. Batteries are typically 60, 125, and 250 V. These are nominal voltages.

Now to encompass the greatest number of potential loads we want to pick a "ridiculous" kA rating, such as 20 kA+. We might even optimize this one by plugging in different kA valeus and observing the resulting incident energy, "optimizing" the table value to encompass the greatest number of loads.

70E has thrown out "PPE 0" so the tables all now start at 4 cal/cm^2 or PPE is not even needed. So with this as a target we can mix and match the above inputs to derive the fewest number of entries that covers the greatest number of loads.

But there's an obvious fallacy here. DC equipment that doesn't even reach 1.2 cal/cm^2 is the greatest amount of equipment. And there's no longer a "PPE 0" in the table. In fact the overall approach that has pretty much always existed in 70E, which is a huge fallacy, is that for any given task and any given equipment, there are 4 possible states:
1. Low risk, low hazard
2. Low risk, high hazard
3. High risk, low hazard
4. High risk, high hazard

Note that all versions of 70E from at least the 2000 edition address #4 with a PPE requirement and as of 2012 (and more strongly in the 2015 edition) point to the ANSI Z10 hierarchy of controls as the correct approach while supplying specific PPE requirements as a last resort. In the 2012 edition and earlier item #3 (1.2 cal/cm^2 or less incident energy) was addressed as either "H/RC 0" in the tables or simply ignored (ala arc flash hazard boundary). Items #1 and #2 were "addressed" extremely poorly by the definition of an arc flash hazard only. As of the 2015 edition with regards to tables items #1 and #2 are now explicitly addressed while with the expunging of H/RC 0 there is no longer any room in the tables for item #3 (no hazard). The net effect is that when using the tables case #3 no longer exists and only cases 1, 2, and 4 are addressed adequately. Effectively the tables start at 4 cal/cm^2. The same condition exists in the tables in NESC but arises out of the fact that OSHA is adamant that utility workers must wear arc flash PPE for effectively all work almost without regard for the magnitude of the hazard because the concern is over hazards other than arc flash (with copious hand waving going on).