In a Converter cabinet I have both, AC components or circuits and DC components or circuits. I created the model, perform the AC and DC Arc Flash study now, the question is: which labels should I use on the cabinets, the DC, the AC or both?

What is your opinion?

You've only done a hazard analysis. Now roll back and do the risk analysis...what tasks are going to be done in the cabinet and what might the worker be exposed to? That determines what label to apply. And if both risks exist then you might have more than one label. Or you might want to put on whichever one has the worst case hazard...that's the problem with a "one label" solution...the real solution is far more complex. In fact you could just put on a very generic "Arc flash hazard present" label and meet all the requirements of NEC which is the only truly regulatory requirement with regards to labels.

Paul thx a lot for the information.

I have done quite a few calculations for cabinets that have both AC and DC components. I always calculate both and use the highest incident energy. In almost every case so far, it has been the AC component with the highest incident energy. The only exception to date is a battery enclosure.

I too would recommend just placing the Largest of either the AC or DC Arc Flash Energy Level on the equipment. Putting two labels on could cause confusion. Usually the DC side is lower. When two labels are attached, the technician who is only working on the DC would probably wear the lower rated PPE. Depending upon the design of the equipment, he/she could cause an DC arc flash that could jump over to the AC side. This would then be a much higher energy fault.

So a fault occurs, plasma is generated, and the other source now has a conducting path as well. Think of lightning striking a power line. Suggest posting the summation of the two calculated values.

That's an interesting analogy precisely because it's what doesn't usually happen with power lines. In pole lines almost always during a lightning storm, one phase trips phase-to-ground. Occasionally two or even all 3 phases are involved but the vast majority of trips are single phase. The second interesting thing about this type of fault is that it usually stays rooted right to the spot where it occurred whereas a phase-to-phase fault will be propelled magnetically down the power line and rapidly move away from the point where it initiated (at a few hundred feet per second).

In an enclosure, things work a little different. First, the heated air (it is not really plasma) quickly expands and fills the enclosure. According to tests conducted by Mersen, E-Hazard, and several others, nearby phases start arcing within about 1 cycle so we quickly evolve into a 3 phase acring fault. This appears to happen even if there are phase barriers (according to Mersen's results) or even if the bus is insulated. Due to the second phenomena below, I've also seen even metal clad switchgear progress into a 3 phase arcing fault almost every time in practice...it seems to me at least that metal clad gear doesn't really do what it is purported to do (limit arcing to phase-ground only). In busbars (such as MCC's and switchgear) the arc will be propelled just like on power lines until it reaches a point where the arc is blocked from moving either by reaching the end of the bus bars or if there is a substantial insulating barriers. It is a practice according to the NFPA standard on fire investigations to look for "arc tracks" which are tiny (sometimes microscopic) alligator looking marks where an arc moves from the original initiating location to the point where it stops moving.

Going further with this though, and this is where it gets tricky trying to quantify when and how far the arc propagation happens...I can state with almost certainty that it doesn't happen on pole lines because the spacing is just too large. Similar results happen with some widely spaced equipment like in throat sections of network or substation-style transformer air terminations, even if it's medium voltage. Even the much feared "jumping across a breaker to the line side" effect only happens rarely. I've never seen this happen outside of a lab even in some of those ridiculously high current I-line panels out there. but I've heard others report cases where it happens. So if this theory held water in every case, we'd find almost every panel utterly destroyed by all arcing faults but that's not what happens in practice.

So I'd have to say that this arc propagation theory would have to depend partly on the arc power, partly on circuit/phase spacing, and partly on how "enclosed" the panel is. I can't point to any documentation on how much is too much other than to say that I'm a bit dubious of the various arc propagation claims out there that seem to suggest that an arcing fault could start in a light switch some place in a plant that uses conduit exclusively and eventually grow to the point where it burns up everything halfway down the utility's incoming poles before a distance relay finally shus it down.

I disagree. The lightning is a DC source that establishes the conductive path for the AC to ride along on. The resulting arc energy contains both contributions.

I wasn't aware that heated air was conductive.

I did not say that lightning creates a conductive path. That much is obvious or we wouldn't get lightning-induced faults on power lines. Anyone that lives out in a rural area knows this even if they aren't electrical experts. What I said was that phase-to-phase faults don't happen very often in practice, and we certainly don't get 3-phase arcing faults MOST of the time with lightning. There are two reasons for this occurring. The first is that the basic flow of electrons in a lightning strike is between the Earth and the energized cloud, so most of the flow is cloud-to-ground. Electrically phase-to-phase is essentially neutral considering the voltages involved. There's simply nothing driving arcing from one phase to another unless leaders touch two lines at the same time. It happens but not very often. The second reason is because the lines are physically spread very far apart and with typical construction techniques, the BIL from line to line is much greater than line to ground. Not only that but as I said this is from experience, using a power line system for a large mine with about 70 miles of power lines located in the Southeast over 90% of trips were phase-to-ground, and most of them are lightning related. And that's with a design that provides roughly 99.9% shielding with static wires from direct strikes, although it was vulnerable to back-flash (lightning going from the ground wire down the pole across the cross arm and insulator to the conductor). Most literature on power line faults also indicates similar activity with distribution lines. It changes somewhat for transmission lines where the insulation coordination is so high that lightning ceases to be a serious problem. Sometimes we would get more than one phase-to-ground fault and I think once over a period of years there was a phase-to-phase fault along with phase-to-ground faults but I kind of attributed these to exceptionally large lightning strikes with multiple leaders or where the lightning voltage was so high that it basically blew right through everything. Statistically this is bound to happen once in a while.

Without getting into the point where we start seeing plasma (at about 4000 K or higher) which occurs when basically the temperature gets high enough that the outer layer of electrons becomes almost free and the air really becomes sort of a sea of electrons model similar to metals,

Vbreakdown=24.22X_6.08*sqrt(X)
X=293PL/(760T)
...using the temperature dependent version of Paschen's law.
V is in kilovolts
P=pressure in Torr's (mm Hg)
L=gap length (cm)
T=temperature in Kelvins

So as pressure or gap length increases, the breakdown voltage increases as expected but as temperature increases, breakdown voltage falls. Note that also as temperature increases, either volume or pressure is going to change as well (PV=nRT or the generalized version) so even though locally we'd expect pressure to at least be elevated (within reason) once the doors are removed (by force...), pressure stays fixed and thus volume is increasing. Thus temperature can increase substantially without a compensating increase in pressure...volume is increasing instead, and thus the breakdown voltage drops rapidly. You can also indirectly see this in arc voltage/current data which shows initially a fairly high restriking voltage that drops as the arc continues until the arcing waveform is almost sinusoidal again for self-sustaining arcs. If this were not the case then arc blast (such as it is) wouldn't happen and the measured pressurs developed by arcs in open air wouldn't exist either since air pressure would increase linearly or close to linearly with temperature close to the arc.

There is quite a bit of detail in several articles published about this by Lowke and others such as "Simple Theory of Free Burning Arcs" which has been referenced several times in the arc flash literature. Lowke's information is really good and gives relatively simple equations for detailing what happens at the arc core. For instance this is the source of the crazy claims that ARC FLASH reaches "20,000 K+" but it is somewhat false because this is true in the core of the arc but NOT true for the surrounding air. The entire quote is:

"The radiation emission coefficient U increases rapidly with temperature,
so that for high currents the term in U dominates the conduction term in equation (1).
For temperatures of the order of 20 000 K encountered in high current arcs in air or
nitrogen about 90 % of this radiation is in the ultraviolet region of the spectrum (Hermann
and Schade 1972, Hermann et al 1974). When this radiation encounters the gas
surrounding the arc core it is reabsorbed and so never leaves the arc column. As a con-
sequence the integrated energy balance equation (2) can generally be used for high
current arcs, without inclusion of a radiation loss term from the arc column. A similar
approximation for arcs in forced flow gives agreement with experimental results (Lowke
and Ludwig 1975)."

So what happens in reality is that energy is emitted from the plasma but then almost entirely reabsorbed by the surrounding air which in turn emits energy again which is absorbed...in a cycle. We currently lack a theory to predict how much thermal energy is emitted via this chain reaction to say a nearby patch of skin so unfortunately even if the core of the arc is well understood, the macroscopic effects so far are less well understood. It is simply incorrect to claim that the air/radiation affecting a person in terms of an "arc flash" are 20,000 K and with such a pathetically small amount of plasma which is also magnetically contained within the arc core, calling the heated air that comes out of an arcing fault "plasma" is simply not correct either. It's just hot gas.

In any case, radiative heat transfer is proportional to the 4th power of the temperature delta, so very little temperature increase is needed to rapidly heat up anything and everything in the area. Not only that but what I see in most of the arc flash videos is a lot of smoke and soot. Even flames are really mostly invisible. I believe that the "plasma" is actually simply luminous smoke particles. See for instance:
https://en.wikipedia.org/wiki/Luminous_flame

And the reason that we are seeing elevated incident energy in those experiments is because of the 4th power effect of radiation. There's really no reason for "plasma" to exist and it would be magnetically confined to the arc itself, not just free floating as it appears to be doing. So my best guess is that this is actually simply smoke that is glowing and visible on camera, and that the increased damage (incident energy) is increased radiative heating.

I brought up lightning as an a example of an arc containing DC and AC components. I'm not sure how delving into the physics of lightning contributes anything more to the discussion. My point is the conservative approach to the OPs question is to add the two IE contributions.