The amount of energy that would be released during an arc flash event depends mostly on two things: the available fault current and the amount of time it would take the upstream OCPD to terminate the event when that much current is flowing. The intent of this NEC section is to reduce the hazard to a worker by one of several approved means.
But let’s back up a step. It turns out that reducing arc flash energy and providing for selective coordinate are contradictory goals. If you want a breaker close to the fault point to trip before your main breaker trips, then you slow down the trip mechanism on the main breaker, so as to give the downstream breaker time to trip first. However, that will mean that if the main panel is open for inspection, and if an arc flash event begins, it will take the main breaker a long time before it reacts. That will allow for a high value of arc flash energy to be released.
One way to mitigate this possibility, and this is what I think the engineer is proposing, is to install a “maintenance switch.” The worker will turn this switch from “normal mode” to “maintenance mode” before opening the switchboard. What this does is to reset the main breaker’s trip characteristics, so that it would trip faster in the event of a fault. Thus, a lower amount of arc flash energy would be released. This switch will have the side effect of disabling the selective coordination arrangement, since on a fault further downstream it is possible that the main breaker will trip before the feeder or branch circuit breaker. That is the risk that you accept, in order to gain the benefit of a lower risk to the worker during an arc flash event. For this reason, the code requires that the maintenance switch be accompanied with an indicating light, so that the system is not inadvertently left in a condition that has selective coordination disabled.
