Something I have no way to test...

Since arcs usually travel down vertical bus in MCC's and then "arc out", what if we put a " quencher" on the bottom of vertical runs?

Second thinking about fuses, they are sand filled because melting silica is very endothermic. Could the arc quencher be as simple as a "bucket of sand" covering the ends of the vertical bus bars? Even a highly sophisticated 'bucket' would be inexpensive and a lot less than arc resistant gear.

Down the bus? Arcs tend to travel up as they follow the flow of the conductive hot ionized gas, like with a Jacob's ladder. Also not sure why the arc would enter the bucket. It would simply stop moving at the interface while continuing to arc.

My experience and test work that is out there is that arcs do not follow gravity very much, if at all. They are magnetically propelled away from the source of energy, regardless of the "direction", as long as the plasma does not encounter some kind of barrier that impedes travel. You can see this very obviously inside of a contactor where the arc chutes are often mounted on the bottom.

Quenching an arc as far as I know can be done in one of 3 ways:
1. Remove power from it. This can be done by the obvious (interrupt flow) or the less obvious method (short circuit somewhere else on the bus with a lower resistance path).
2. Increase the arc gap. This is what arc chutes do for the most part.
3. Cool the arc down. This is what happens inside low voltage sand filled fuses. It can also be accomplished by various "puffer" mechanisms.

http://www.youtube.com/watch?v=hIkNY5xjy5k Note how the arc continues to rise until it contacts the line above. If the bus were vertical, it would travel up. Source direction matters too. DO all MCCs have the source at the top?

First, we're not talking about the same thing. The arc from that video was constrained to two electrodes and not allowed to move. So the plasma or hot gases rose upwards which lifted the arc over the course of a few seconds. The arc was constrained in this case. Any movement "along" a bus bar would have been outwards along the electrodes of the switch contacts. Since that would entail an increasing resistance (metal is more conductive than plasma), it won't happen.

Arcs on open bus bar are another matter entirely. The arc is constained such that movement away from the bus bars themselves would follow the exact same physics that you described. That is, it would move based on air currents. However that is simply not true in the direction parallel to the bus bars. In that direction, so long as there is not a substantial barrier to the plasma, the arc will move along the bus bar. This has actually been measured at JHU-APL and is roughly a few hundred feet per second. This is much faster than would be captured by a normal camera on a Youtube video. You can often see the effect in some videos as the arc shoots along parallel lines on an overhead power line akin to a Jacob's ladder.

Both arc flash incidents that happened at my current employer that I have on record involving MCC's had the arc initiate at one location, travel down the vertical bus, and erupt out the ends of the vertical section. This paper shows damage from an MCC where the arc travelled vertically DOWN the bus multiple times:

http://www.neiengineering.com/wp-conten ... esting.pdf

Or this one:

http://www2.schneider-electric.com/docu ... ect038.pdf

In fact travelling plasma leave small sawtoothed "arc tracks" on the bus bars where it passes by and is forensic evidence that can be used to determine the origin of an arc. This is mentioned in the NFPA standards for forensic fire investigations.

Hence the reason that we are talking about two entirely different things, the movement of an arc along an unconstrained surface such as parallel bus bars, compared to the movement when it is constrained to a point such as on the ends of high voltage contacts in open air switches or on an arc gap style voltage multiplier ladder.

I have watched an arc flash that initiated at the tie breaker of a double ended substation, and traveled to both ends of the sub. I think that agrees with your predicted movement pattern.

Final "gotcha" with arc travel...

The forces that act on the arc itself are magnetic and generally propel it away from the power source. However this only holds true as long as the hot gases are available to restrike an arc. The gases themselves tend to expand as well as move upwards (hot air rises) although the former is quite fast and the latter is quite slow compared to hundreds of feet per second for "mobile" arcs. If the gases are constrained such as by an insulating barrier, the arc and/or plasma expands outwards from that point. If the plasma/hot gas reaches another phase or conductive surface, another arc forms. This is the reason that even though all arc faults will start with a single arc between two phases, at least within enclosed equipment they rapidly convert into 3 phase arcing faults within a cycle.

But, and this is the final gotcha, if the hot air/plasma reaches from the load to line side of a breaker it can "jump" a breaker as a new arc forms on that side. At this point the breaker no longer matters as far as tripping goes and the incident energy goes up a lot. This is the reason for modelling panelboards using values from the line side of the main breaker (pretend none of the breakers are in the panel) compared to switchgear. Opinions differ on MCC buckets. Some practitioners model the section with the main breaker (if it has one) separate from the other sections. Within a section it is certainly possible (likely) that an arc will migrate within the section but much less likely that it can move from one section to another. This then gives the "main breaker" section one rating and the other sections a different rating.

In practice because of this for new installs, I'm doing one of two schemes:
1. For medium voltage with external protection relay, install a second set of bushing CT's on the transformer. Trigger main breaker on the primary side bsaed on a fault on either side of the transformer.
2. For other cases, install a breaker or fused disconnect on the secondary side of the transformer, frequently a panelboard. Install a main breaker in the MCC. The MCC will have low incident energy ratings on both the main breaker (if it has one) and the individual buckets. Generally the secondary protection of the transformer does not get used for de-energizing purposes at all. Use the MCC main breaker for most cases or if this is the area that needs service, open the disconnect on the primary side of the transformer. In any case, arc flash will be very reasonable.