KTRman wrote:

thank you, you did answer my question. We do the same as you describe.

I have a situation where the place is huge and would take years to finish, let alone would we find where every panelboard is fed from. We were thinking of a methodology where we model most of the equipment (MV systems, 480 volt feeders, for example) and then at some point just use the table method for lower amperage devices (say 480 volts 150 amps and below). Otherwise, it will be a few years to complete.

This question has sort of been answered before, particularly in a couple IEEE articles about "boundary" methods. To begin with I put together an article (in the articles section) pulling together most of the information on <300 VAC systems. It should be clear that arc flash hazards start SOMEWHERE in the neighborhood of 200-300 VDC/AC. So at that point we have to do SOMETHING. Below that point, I can't find a single case of a hazard.

In the standards, IEEE 1584 gives some sort of vague hand waving wording towards 208 VAC and fed by "125 kVA or less" which covers a lot of ground on what could be considered <1.2 cal/cm2. Other research (which my article points to) lowers this limit somewhat and it is my understanding through some vague information that most likely the next version of IEEE 1584 will lower this limit down to somewhere in the neighborhood of say 60-70 kVA if it's on the kVA scale. NESC starts with "PPE Level 1" as the minimum PPE level due to pretty much a mandate by OSHA that all utility workers wear PPE Level 1 as a bare minimum. In that environment based on a lot of real world testing they clearly delineate everything except a few isolated types of equipment that is 300 VAC or less as PPE Level 1 or lower (4 cal/cm2). In a rather large plnat that took quite a while to do a survey for (about 3,000 busses), this sufficed in getting rid of a whole lot of lighting panels and other light duty/commercial type 208/240 VAC panels altogether with some rather simple limits.

That brings us to true boundary methods. There have been a smattering of articles published over time about the subject but for most common trip curves, when you are in a fixed trip time region such as "instantaneous", incident energy is at a maximum at the threshold trip current. It DECREASES from there as you'd expect. Time isn't changing and incident energy is a function of trip current alone if we don't change the equipment type. In fact this even applies to cases where the breaker "never" trips or never trips for a very long time because we fix the incident energy calculation at a maximum of 2 seconds.

If however you are dealing with an inverse time curve, then the incident energy in general INCREASES as the fault current decreases. Some of the very, very mild ANSI trip curves will flip this relationship and the incident energy will DECREASE as fault current decreases but that's not the typical case, especially with every molded or insulated case breaker on the marker. This really only happens with very low time dial settings on very mildly inverse curves in protection relays.

This causes a very interesting pattern to develop when you think about it. Every time we go through a transformer, everything kind of "resets" and we have to pay attention to the incident energy downstream of the transformer up to the first breaker. This breaker then sets up a "zone" downstream of it where for all practical purposes the incident energy is either fixed (and low) for a long distance if we are in the instantaneous trip "zone" or gradually increasing if we are in the inverse time zone, until we hit either another breaker (which MAY change this) or the 2 second rule which means after that point unless we hit a transformer, incident energy is fixed from that point outward. These various "zones" are predictable in terms of their location and if we also consider that we're not issuing PPE with a continuous rating but rather discrete PPE "levels" (which probably don't match the ones used in the 70E tables), this also makes the "zones" even more discrete...we only care when we change PPE "levels".

Once you recognize this, we really only have to look at say an infinite bus, a transformer, a long length of cable (pick something reasonable such as 1,000 feet), and the most inductive load that the breaker can handle (typically a motor load) without causing nuisance tripping (look closely at the TCC). You can actually calculate this without the transformer and only the breaker details but I'm describing something practical you can do in SKM/ETAP/Easypower. Now the boundaries we have to be concerned with are the instantaneous/short term transition, the short term/long term transition, and the long term/2 second rule transition. Whatever these fault currents are determines the various extremes that are possible in terms of incident energy from the circuit breaker and if we look at the maximum of each of these "points", these are the true MAXIMUM incident energy values that would ever be possible through a given breaker, regardless of what the upstream feed is. Or you can resolve these other ways such as by picking large steel conduit lengths and using a rule such as "less than 1,000 feet of conduit"...whatever you use to set boundaries. And if you also look at maximums across different breaker sizes, manufacturers, etc., you can develop a rule for say 50 A molded case breakers (generic) with any load and up to 1,000 feet of cable. Then you can quickly ignore modelling huge parts of the plant.

This is actually done in utilities. With hundreds of miles of overhead lines, it is physically impractical/impossible to model every single transformer or pole in the system. Utilities don't even try this...they have maps of their service area marked out in various PPE "zones" so that say within 1 mile of a particular substation is one PPE category but it is a different one for miles 2-10 and another rating for anything over 10 miles away.

I was working towards some of the same kind of thing for a mine. They had about 150 substations that are basically constantly moving (every substation moves roughly every 6 months) in the mine area as the mine moved at roughly 25 acres per month. And that's the slow stuff..at the active face the equipment moved about every 3-4 days. Keeping/updating arc flash models in this environment is a joke. So it's simply not possible to model in the same way as recommended in say IEEE 1584.1. Instead you have to model general regions, zones, areas, and equipment types, making everything discrete according to PPE levels.