Arc Flash Mitigation

[timm333]

I have a question about NEC 240.87(B). The upstream breaker would clear the fault anyway; so how do ‘zone selective interlocking’ and ‘bus differential protection’ reduce the fault clearing time?

 

[big john]

A fault will generally actuate "ground fault" or "short time" protection first, but those almost always have time delays built in to reduce nuisance tripping and preserve coordination. Istantaneous protectin will catch a fault quickly but must be set high to also avoid nuisance trips. This means normally a fault would be really big or last a relatively long time before it gets cleared, creating a serious hazard.

ZSI and 87 schemes have the advantage of identifying the zone where the fault occurred and allowing local protection to operate both instantaneously and at very low pickup without waiting for GF or STPU. This drastically reduces both the magnitude of the fault current and its duration, which knocks the incident energy way down.

And because the zones can be identified, this also prserves the coordination for everything outside the faulted area, so only local protection will operate immediately, everything upstream waits for normal trip delays

 

[timm333]

Thanks. I think ZSI usually means that the upstream breaker has a time delay so that the local breaker in the faulted zone is allowed to operate first. Does in this case the upstream breaker operate instantaneously? Also I think the magnitude of the fault current should not be an issue because the instantaneous trip should operate no matter whether the fault current is big or very small.

 

[big john]

Thanks. I think ZSI usually means that the upstream breaker has a time delay so that the local breaker in the faulted zone is allowed to operate first.

. Right. If all ZSI did was enable Instantaneous on all breakers then there would be no effective coordination.

Does in this case the upstream breaker operate instantaneously?

Only the one closest to the fault.

Also I think the magnitude of the fault current should not be an issue because the instantaneous trip should operate no matter whether the fault current is big or very small.

Instantaneous trip has a pickup just like any other protectipn band. If you set it low to detect a low magnitude fault then it trips during normal equipment operation and it also makes coordination very difficult. So traditionally it is set high or removed completely, but this creates more fash hazard.

87 protection will distinguish between a load inrush current and a ground fault that should clear instantly. ZSI has the advantage of turning low GF and STPUs into instantaneous trips without creating a race-condition with the breaker uppstream.

 

[paulengr]

ZSI and 87 schemes have the advantage of identifying the zone where the fault occurred and allowing local protection to operate both instantaneously and at very low pickup without waiting for GF or STPU. This drastically reduces both the magnitude of the fault current and its duration, which knocks the incident energy way down.

One big nit pick here. It will reduce the duration of the fault but you can't say anything about the magnitude at all. The magnitude over the short term is entirely dependent on the amount of reactive load on the system. So we will generally see a transient over the short term (a few cycles) which reduces over time. This is typically what happens in an industrial system. If there is significant capacitive load then the opposite happens. Arcing itself occurs in nanosecond or picoseconds so for all intents and purposes, the striking and quenching of the arc in an arcing fault is instantaneous. Although as air temperature increases the arc initiation voltage does decrease, within 1-2 cycles typically two additional physical changes occur. The first is that the arc spreads to encompass nearby phases, converting what started out as a single phase fault (L-L or L-N) into a three phase fault (L-L-L although photograph evidence shows that when the center phase passes through a zero crossing, the arc momentarily becomes L-N-L). The second effect is that air pressure is normally enough to remove the doors whether the equipment is designed for pressure relief (arc resistant) or not. And although very small breakers (<100 A) can indeed open in under 1 cycle, the norm for power breakers these days is around 3 cycles. So we really only need concern ourselves with the magnitude of the arc in the first 3-5 cycles but generally it isn't going to change significantly after the first cycle. Thus most modelling either assumes that the arc power is constant or at best attempts to model the fault current contribution from large inductive loads (transformers, motors).

The solitary exception that I've seen with this is that a recloser from S&C is a typical vacuum recloser but it has individual phase breaker control and issues a very fast close/open sequence timed so that it occurs near a zero crossing. During the "bump" on the line, it measures whether or not a fault still exists before issuing a normal reclose operation. Thus the "bump" does indeed have significantly less fault energy involved because the line voltage during the bump is significantly reduced. Then again, the price on this breaker is double the competition so unless interrupting capacity is an issue, it seems awfully expensive for what you get.

 

[timm333]

There is another thing called blocking-mechanism whereby the 87 relay blocks the operation of the upstream breaker for some time. What would be the overall sequence of operation: will it be GF (or L-L) tripping the upstream breaker, then 87 tripping the upstream breaker, and then finally the upstream breaker trips automatically once the time delay caused by the blocking mechanism is elapsed?

 

[steve66]

I have a question about NEC 240.87(B). The upstream breaker would clear the fault anyway; so how do ‘zone selective interlocking’ and ‘bus differential protection’ reduce the fault clearing time?

Here is my very simple understanding of the two:

If meters or circuit breakers measure the current entering a switchboard, and the current leaving a switchboard, we can compare the two. If there is some large imbalance, it is probably caused by a fault at the switchboard. So it would be safe to reduce or eliminate any intentional delay, and trip the breakers much faster than normal.

This only protects a certain area (in this case the switchboard - which is where an arc flash incident may be most likely to happen). It doesn't respond any faster to a downstream fault at a subpanel that doesn't have interlocking or differential protection.

 

[timm333]

Thanks, I think have better understanding now. So coordination is desirable only in the case of overcurrent/overload protection, and the coordination is not desirable in the case of differential protection, is it correct?

 

[GoldDigger]

Thanks, I think have better understanding now. So coordination is desirable only in the case of overcurrent/overload protection, and the coordination is not desirable in the case of differential protection, is it correct?

No. It is just harder to arrange in the case of differential protection IMHO. Coordination is desirable any time you do not want to trip off the whole site because of a branch or feeder level fault, regardless of what type of branch fault it is.

 

[big john]

One big nit pick here....

So simply put: You're saying that with rare exception the fault current will always reach peak value before the operation of any protection? So the advantage to quick operation is entirely due to reducing incident energy and has no bearing on the amount of current flow?

 

[paulengr]

So simply put: You're saying that with rare exception the fault current will always reach peak value before the operation of any protection? So the advantage to quick operation is entirely due to reducing incident energy and has no bearing on the amount of current flow?

Google assymetrical fault current. During a fault, a purely resistive load has a symmetrical fault current. A reactive load produces a transient current that is superimposed on the symmetrical one. This usually doesn't last for more than 3-5 cycles though, but that's the time required to open with large breakers doing an instantaneous trip. Fault current doesn't grow over time unless the system is ungrounded and an arcing fault causes additional faults as it drives system voltage to shred insulation typically at 6x to 8x normal line voltage when motor insulation gives out.

is not possible to detect a fault in under a quarter cycle with current measurement alone. Only light sensors can detect an arc flash sooner. So the various arc termination devices on the market achieve <1 cycle arc quenching by creating a dead short on the line with 1 cycle tripping.