The basic question is: Should ground fault protection be taken into account when performing an arc flash study?

Most (MOST, not all) arcing faults will propagate to 3 phase to ground faults, and will probably be taken out with the ground fault detection (if it exists). So, should this be taken into account when calculating trip times?

Currently, we do not look at this at our location, as I have yet to see anything on this, but it would make sense to use the ground fault clearing time.

I use IEEE 1584 method without having deeply analyzed the reasoning behind this standard.

I believe the standard does NOT include ground fault sensing as that might not represent the worst case exposure. In some cases a medium voltage fault will go phase to phase prior to tripping on ground fault pickup so that might be the rationale.

Right now the IEEE 1584 Calculations are based on 3 phase faults. They do not consider zero sequence impedance etc. required to estimate the line-ground currents. Maybe sometime in the future this might be considered but the current standard was based on a limited budget from several years ago so three phase is all we have.

Yes, the concern was that a single phase arc could turn into a three phase event very quickly. I could see a ground fault device tripping if it was quick enough and trips before the fault goes to three phase but most ground fault devices have a minimum time delay of 6 cycles (0.1 seconds) so most people take the conservative route. Similar logic is used in suggesting that a local main might not be the limiting factor for an arc flash since the plasma could reach the line side terminals of the main and continue even if the main trips.

Some day new tests and new interpretations might develop but for now, using three phase faults and phase devices seems to be the standard interpretation.