1. Got to remember, an AC sine wave is 1/60th of a second, or 16 ms. Granted we can talk about the positive/negative lobes which makes it 1/2 of this, going on over up to 3 phases. So right away, you can forget about AC control signals. It's a DC-only world, and with 24 VDC systems requiring a lot of current usually (but not always) the relaying ends up operating on 125 VDC which is kind of a substation standard voltage.

2. Assuming you can DETECT an arc, in the case of most circuit breaker designs (vacuum in particular), they usually rely on the current going through a zero to actually quench an arc. For the most part, only fuses with their electrochemical reactions actually interrupt during the peak voltage. Fuses operate at 4 ms. Vacuum interrupters operate at 2 cycles. There are some small MCCB's with air interrupters and even a few SF6 units that move fast enough to open faster than this but I don't believe I've ever seen a breaker with a spec less than 1-1.5 cycles, or 16-25 ms.

3. If however you goal is simply to QUENCH an arc, remember that the arc has some impedance (though low). So all arc "quenchers" operate on one of two principles. Some of them have a resistor in series with the bus and a high speed bypass contactor. The contactor opens and transfers the load onto the resistor, lowering (but not extinguishing) the available arc energy to a very small amount. No arc interrupting is happening so the contactor merely has to open fast enough to make the contacts a high impedance path, something that can be done in less than a cycle with a high speed magnetic actuator.

The other, more common, mechanism purposely closes a short piece of bus bar onto the bus, effectively creating a 3 phase bolted fault. Since this path is lower impedance than the arc, the arc is instantly extinguished at any point in the cycle. Again, closing can happen at high speed so these operate in less than a cycle. The challenge in this case is that the system is subjected to a bolted fault condition so any deviation from AIC ratings will result in the gear coming apart.

In either case, the resistor or contactor has an obviously limited thermal life of a few milliseconds. It buys you time until the main breaker interrupts the load. Otherwise, the arc quencher will be destroyed. All of the ones I have seen are not rated to open under load.

The major difficulty with these devices is that to catch all situations, they must be installed between the main breaker and the bus being protected. The devices themselves are not large but unless the equipment is purpose built there is usually not enough room to add something like this in.

I would also throw a third device here into the mix though it's not strictly speaking a "quencher". Power distribution systems more and more are making use of a device called a recloser. In distribution lines, generally over 90% of faults will self-clear and the speed of clearing is less than a second. Reclosers take advantage of this by opening like a conventional breaker, and then reclosing back onto the line automatically. When they close again, they generally change settings so that if a fault is detected a second or subsequent time, they reopen immediately and then stay open. This allows the fault a chance to self-clear and restore power immediately. The downside is that the breaker has to be heavy enough to close onto a fault, and that the equipment is stressed every time that this occurs and there really is a fault.

4. The statement about holes in the switchgear and concerns about pressure developed in an enclosure is not important in my mind. Even if you have arc resistant gear where arc blast pressures are controlled when the doors are closed and latched, if it is necessary to open the doors for test or maintenance work of any kind, the protection is compromised. These are also the same workers that most commonly appear in the news for arc flash injuries. It is well recognized that even "standard" doors offer some protection but there is no guidance currently available to quantify it. I would argue that the best designs are those that do NOT rely on doors for protection at all. If it can sit out on the floor with no covers (or even within a box) and still offer adequate protection, it will protect against arc flash/blast for all workers in all tasks. That is something that arc resistant gear cannot do.

5. The question about CT speed is a nonissue. The arc quencher people won't say it but the key is to use a Rogowski coil. These work under different principles than your standard bushing CT and can actually be used to measure current signals up into the gigahertz range. They are the same sensing coils used for partial discharge detectors. Measuring rapid dv/dt changes are really only limited by the speed of the measuring circuit itself, something that can be done in microseconds these days.

6. As to arc clearing time minimums, I have not seen any data published publicly indicating how long it takes for an arc blast pressure wave to develop or to subside. As I understand it from numerous sources mentioning it there is a maximum that seems to be something of a constant as well. However, I haven't seen any published data especially in a peer reviewed journal, only comments here and there. This gets into the physics of arc flashes. Right now there is very little information about this. The most commonly used model (IEEE 1584) is an empirical model which is curve fitted and treats time as a linear factor only. Arc blast is not taken into account.

That being the case if I am dealing with off-the-shelf VCB technology for instance, and I'm limited to either 2 cycle or 3 cycle breakers, even most stock relays with AC coils are tripping in 1 cycle. If I'm at a fraction of the breaker speed itself, reducing tripping time further has very little impact. If I'm down to a 1 cycle trip time (quencher speeds), relays operating again less than around 10-25% of the contactor speed have minimal impact, assuming IEEE 1584 is valid at these speeds. Again, arc physics may say differently but we're not there yet.

Aside from this, yes, there are major reasons for claiming to be faster. As everyone knows, faster is always better and every manufacturer in this field is going to do everything they can to say that they have the fastest, most reliable, lowest cost system out there. Generally speaking, you get to pick 2 of the 3 categories. Claiming 2 ms vs. 10 ms is all about marketing. IF you can overcome the lack of information and show that arc blast takes 10 ms to develop and educate the buyer/specifier about this then you can end the millisecond race. Otherwise, faster will always be better as long as it does not significantly compromise reliability or cost.

Sorry to nitpick here, but they use regular CTs (VAMP 221 uses 1A or 5A current sensing inputs, can't do that with a Rogowski coil). Your point remains, a CT will not add a measurable time delay as it's frequency response is pretty much flat up to a few 100 kHz.

One thing I'd like to add to the discussion is that line side arcs are not always caught by these systems. Depending on the position of the CTs an arc in the incoming panel could be detected and acted on, but upstream protections may have a time delay that's larger than the device's withstand time. For LV distribution the largest hazard may not be covered.

Positioning the CTs at the cubicle's cable entry could fix the detecting problem if you're willing to accept that it reliably catches an arc flash inside the cubicle. The upside of a proper bolted fault is that it helps lowering upstream relay settings for just this case (i.e. instantaneous or very low time delay segment), keeping coordination intact for downstream faults.

Thank you.

Pressure peak is reached in approx 10 ms? Is pressure from an arc flash fully understood? What papers describe the pressure from an arc flash?

ABB has a good description.
Try this link.
http://www05.abb.com/global/scot/scot235.nsf/veritydisplay/7313a71f4406af5cc125792d003e1905/$file/PRS_UFES%20Standard%20S3%20EN%204zu3_V110527.pdf

ABB and other high speed "Earthing" switches have been touting this unattributed anywhere that I know of.

I've seen similar data from circuit breaker calculations and testing. Without going too far out on a limb though I think we can safely see where this is coming from. A full cycle is 16 ms, and assuming typical worst case scenarios, X/R is going to be a relatively high value so we have typical industrial inductive loads. Thus, current lags voltage. Any attempt to interrupt an arc where we cut into the first half wave of the current wave form will naturally result in a failure to fully develop the initial pressure wave. So we are talking somewhere around 8 ms plus a little more depending on how much the current waveform lags the voltage waveform.

IEEE 1584 of course ignores this because it treats it as a bulk energy transfer problem where time is simply a linear parameter. When you are tripping at less than 16 ms, it becomes necessary to use a time domain based incident energy model because what happens during the cycle becomes important down to modeling the characteristic "ringing square wave" shape of the arcing current and taking into account the various X/R contributions from system reactance, motors, etc.

My argument is on purely theoretical grounds. I've seen data from arcing effects of circuit breaker contact tip opening models but not from arc flash studies.

Well I am not working for ABB but I like their dokumentation and believe it is relevant to answer the question about pressure from an arc flash. I have tested a MV arc eliminator in KEMA Philaelphia in a typical USA switchgear with very simple doors and lockings which never should pass the 40kA test witout an arc eliminator. The test was for 0.5s and the arceliminator took over the current after 5 ms.
I know that ABB can do the same. Already their previous product which I know better could eliminate arcs 10 years ago.
Here is a report from a real case.
http://www05.abb.com/global/scot/scot235.nsf/veritydisplay/a3abf87f67887becc1256cbb004ee672/$file/FaultcaseAEeng.pdf
Jankar

Thanks for all the answers. I am about to write my thesis about arc flash soon, and need to gather all information possible.
I attended a test with the ABB Arc Guard TVOC 2 last week. It appeared to me that this is not suited for personell protection, it just does not remove the arc fast enough. Even 40-50 ms was more energy than I would recommend any workers to be close to. At best, this system protects nearby personell passing through, but will not protect people working on live switchgear.

The Arcon system creates a short circuit that takes the energy away from the arc, but is it good to be that close to a 3-phase short circuit?
The ABB system looks interesting. I did not know of the UFES before the test and need to go back to ABB to ask them more about this system.

Again, thanks for all information.

Btw, ABB does not recommend current sensors with their system, it adds some ms to the clearing time. If possible only the optical sensors should be installed.