I have seen graphics that show the arc flash boundary existing only in front of the gear and another showing it in front and extending around the sides.

If you were to mark the arc flash boundary with tape, would you only put it in front (semicircle), or all the way around the gear (circle), assuming the gear is situated in the middle of the electrical room?

If you are modelling it as panels, switchgear, etc. (not open air) then usually you are assuming semicircle and increasing the afb. if you are assuming open air then the opposite applies.

If you can access the inside from the front and the back (covers on the back), then I guess a circle is what you need. Else, since the heat is mostly radiated (I'm not talking about the other dangers associated), a semicircle is more appropriate in my view.

So, if you were to do a semicircle (assuming no back access) would you take the more conservative approach and measure the AFB from the front face of the gear, even though the exposed, energized conductor or circuit part is within the gear?

A little confusing. Conductor exposure has nothing to do with arc flash results except that it can increase chances if causing an arc depending on the task. Although presence of doors, arc chutes, etc. does influence the incident energy, there is currently not enough detailed experimental data to account for this publicly available. The assumption right now is doors do not matter but enclosure sides and back do. The standard working distances assume front doors with bus bars in back. So back access mostly should require reduced working distance. The next 1584 edition is supposed to cover enclosure size and equipment configuration as more years have been run considering these details.

I was referring to a couple of graphics on google images in relation to the "exposed, energized conductor or circuit part". The forum doesn't allow me to post them quite yet.

If the definition of arc flash boundary is "an approach limit that is the distance from the exposed live parts..." then you would measure from the point where the arc flash would occur. However, that point can vary within the equipment, so it would be more conservative to just measure from the front of the gear, correct?

No. The working distance table in IEEE 1584 measures from the power feeds at the back of the gear to the face/chest if standing in front of the gear. The afb just extrapolates this to 1.2 cal/cm^2. Front of the gear would increase distances by between 18 and 36 inches depending on gear type.

The reason is that arcing fault energy is assumed to be highest at that point. It turns out that this part of the construction of the equipment is very similar among different manufacturers. Although theoretically an arcing fault can occur anywhere, practically the most likely areas are at connections (stabs, splices, terminal lugs) or less commonly especially in highly contaminated gear on bare conductors such as bus bars especially at medium voltage over 2-3kv where creepage is a factor. Load side connections are lower in incident energy due to presence of additional overcurrent devices. Although the longer leads within the enclosure contribute to lower current thus usually higher incident energy, the variability is tiny relative to other contributions such as asymmetrical fault current with inductive loads.

Yes, I understand that it would increase the distance by 18-36 inches. In doing so, the tape on the ground would result in a value less than 1.2 cal/cm^2, which would be more conservative.

However, you are saying that the measurement should be from where the stabs and vertical bus bars are located (for an MCC, the vertical bus bars are roughly 9-10.5" from the front) because that is what IEEE 1584 references for the working distance. Am I correct in saying this?

The IEEE 1584 calculation is based on the distance from an arc to the face/chest area. Assuming worst case is at the bus bars then standing with a screwdriver in front of an MCC bucket, places the person about 18" away or at least thats what 1584 is based on. So with a 10" depth to the busbars puts the distance about 8" from the bucket face. Obviously in the back, things would be closer.