Protracted Arcing In Substation 2

[Mbrooke]

Video of what looks like arcing at a Con Edison substation, possibly East 13th st:

https://www.reddit.com/r/SubstationTechnician/comments/h9d07n/no_no_no/

From my vantage point it looks like a series arc on a 69 or 138kv disconnect switch. My question is- why didn't relaying catch it? I'm assuming differential protection is present.

 

[Mbrooke]

Yup. Disclaimer though: series arcing is the end stage of joule heating, not the beginning, assuming a fire has not already started.

 

[prc]

Thank you itsmoked for the links. I liked the operating principle. An electronic algorithm that detects arc by sensing the change in current.Why not such a device for protecting large transformers? It should sens low impedance inside faults quickly and trip the breaker so that tank rupture and fire can be prevented..

 

[Mbrooke]

But remember detecting arcs isn't as easy as it looks considering 1) load can look like arcing 2)breakers arc momentarily when opening 3) its difficult to detect in zone vs out zone faults though in theory the presence of CTs at each point could help.

Because of this AFCIs are nothing but trouble for electricians. Lots of nuisance tripping even on UL listed equipment, trips for utility faults, and lots of cases where connections where seen sparking or burned up without tripping said AFCI.

The computing power required is obscene, and programming it to discriminate between waveforms is another level of mind numbing obscenity.

If you ask me they aren't worth the trouble, when for example, IR readings detect hot spots before they reach an arcing stage.

 

[crshears]

I thought I had finally gone mad lol.

Hadn't that happened quite some time ago?

We now have to use arc-detection breakers in our houses. Why isn't there an arc-detect relay that 'listens' for the unique electrical noise generated by them?

Good question...

Note on residential AFCBs: these are non-directional...a couple of weeks ago a distribution insulator downstream of our residence let go, leading to an interruption of a few hours duration until a line crew got there to replace it. What I found noteworthy was that when power was restored, part of the house; viz., the bedrooms, remained dark; found all the AFCBs tripped...reset them all no problem, but still...

The only thing that makes sense is that these AFCBs "heard" the electrical noise generated by the fault out on the line.

CR

"As iron sharpens iron, so one person sharpens another." [Proverbs 27:17, NIV]

 

[Mbrooke]

AFCIs can not distinguish arcing between line or load, and typically can't discriminate between normal arcs, normal current ripples, and dangerous conditions.


AFCIs are nothing more than pure marketing hubris.

 

[itsmoked]

I'm not convinced Mbrooke..

Decades ago I was working on power line communications. The signals had a hard time reaching very far down power lines even with clever modulation schemes. The high frequencies in arcs radiate thru the air well but not down wires very far being attenuated rapidly with distance starting with the higher frequency components.

A $5 signal processor could easily analyze an arc signal for amplitude, duration, AND distance from the processor's sensing location. So, while some loads could look like arcs they would be at excessive distances, more than a hundred feet. Local switching arcing is easily ignored simply by duration. Arcs shorter than (for example) 5 seconds could be ignored.

A power spectrum taken and applied to a characterized power line could tell you within tens of feet how far away the arc is. I suspect there would be reflections of the arc signal as the signal confronts everything from line bends, breaker housings, insulators, and transmission towers. Sifted these would show large variations in power- spectrums providing much finer distance discrimination (feet).

Basic trip settings on a unit would be:
1) Arc distance from the detector
2) Arc duration

It would log all arcs recording their finger prints.
I'd keep all events in a database to refine tripping details.

The fact that crshears bedroom AFCBs tripped on a utility arc is caused by the total focus on how little, and how cheap, they can make consumer junk.



Keith Cress
kcress -

http://www.flaminsystems.com

 

[Mbrooke]

Not only Crshears, but others as well. Its routine with weekly threads on electrician based forums:

https://forums.mikeholt.com/threads/afci-faults-caused-by-rural-power-outages.2554509/#post-2590436

The truth is you can only get so much out of a $35 AFCI. Any legit signature monitoring would require the computing power of at least an Iphone. UL also requires tripping within 8 half cycle.

I like the emoji btw Didn't know we had that.

 

[bacon4life]

A specific marketing lie is at 1:28 when the Eaton video shows the AFCB detecting a loose wire at terminal. A wire coming loose from a screw is an example of a series fault. Is a series fault, the current will between zero and the normal load current. The test requirements for AFCIs do not cover this common arcing situation, and UL listed AFCIs generally fail to detect this sort of arc.

The AFCIs instead protect against shunt faults, where the fault current is larger than the load value. Note that in all the other examples in the video, cutting through two different sets of insulation was required to produce the shunt fault.

 

[Mbrooke]

@Bacon: Modern combination AFCIs (different from dual function AFCIs) are required to detect series arcing above 5 amps. The issue is that arcing can take place below 5 amps (think lamp cord, 60 watt bulb), and even then its hit or miss above 5 amps.

Regarding shunt faults a magnetic coil set at 80% of the short circuit current will catch them. That was supposed to be the original AFCI (trying to mimic EU circuit breakers) but when it was discovered some high inrush loads might cause nuisance tripping they went to electronics.

 

[sushilksk]

How distance of arc from sensing location is measured, can you give some details?

 

[davidbeach]

The problem at the transmission level is that there might be 10 locations within say 2.5 ohms of the arc that see the arc signature. You want exactly one of those and one 4 ohms away to trip to clear the event. Trip any of other nine and you needlessly drop load. Fail to trip the one at 4 ohms and you may or may not extinguish the arc. I just don’t think there’s any practical means of finding and clearing these types of events at the transmission level without risk way over tripping.

But if we implement something that, at the risk of over tripping for series arcs, will clear all series arcs it will also trip for an awful lot of other things for which there is no desired tripping. For as rare as they are it may well be that dealing with the rare damage is better than over tripping.

It’s entirely possible that the damage to be repaired following a series arc event that never went to ground or phase to phase is essentially the same regardless of the duration. Probably a switch. Barely damaged to horribly damaged still require the same effort to repair. If we can put up with it without dropping customers why should we preemptively drop customers? Enjoy the light show when you see it.

 

[itsmoked]

Hi sushilksk; There are several ways. I was back-of-the-napkin describing one that's based on loss of the high frequencies with distance. The arc will have a typical spectral distribution of noise. The high frequencies are rapidly lost as power distribution is in-no-way able to keep them intact. That's why we use twisted pairs and coax cable to try to keep the high frequencies "in the wire". The more the high frequencies are missing the farther away the arc is.

Google images "electrical arc discrimination". You will see there's a bunch of studies and research in that area.


davidbeach; Such a stick-in-the-mud... I'm starting to think you'd be okay with simple video cameras looking at everything of interest keyed to "If there is an ARC in this camera view that lasts longer than x seconds trip this breaker". That would actually be pretty definitive and false trip resistant. You could use a lot of inexpensive weatherproof cameras and put a UV only filter in front of them. What d'ya say?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[davidbeach]

Anything could be done, it’s just money. On utility systems that money comes from the rate payers. Our recent event was quite notable simply because it’s the only substation example I’m aware of in over 13 years.

In radial systems the necessary discrimination is easy, in networked systems it would be much more difficult, meaning more devices. Could run a couple thousand devices on a system our size. If we’d had 2000 devices for the whole 13 years it would have taken over a quarter billion device hours of waiting and watching to respond to one event. A good use of rate payer money?

 

[Mbrooke]

I would also argue that infared scanning and preventative maintenance will greatly reduce the probability of a series event in a substation or even out on a line.

More complexity isn't better either- more that can go wrong biasing the system toward misoperation. Simplicity is key, where simply solutions often provide the same (if not greater) benefits compared to complex solutions.

https://www.youtube.com/watch?v=_2HyTRxzwXs

 

[crshears]

I like the emoji btw Didn't know we had that.

Not sure "we" do; I looked for that emoji in the ET list of them and didn't see it...other sources, perhaps?

CR

"As iron sharpens iron, so one person sharpens another." [Proverbs 27:17, NIV]

 

[Mbrooke]

Well, that would explain why I couldn't find it. But where Itsmoked got it from, I like it!

 

[itsmoked]

Welp, you guys make good sense in your arguments. Thanks for sharing your knowledge.

................^
............... |
...............me

Ah yeah, Mbrooke, crshears, I have a private library of emoji that I've collected over the years.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[crshears]

In the emoji presently in view, something is agley...the chain would be running clockwise, spraying fluids all over the operator instead of away from the saw, no?

CR

"As iron sharpens iron, so one person sharpens another." [Proverbs 27:17, NIV]

 

[davidbeach]

True. Hadn’t noticed that, but the under side of the blade does come back toward the operator.

 

[itsmoked]

Yes indeed. Nothing much gets past a bunch of engineers.

Keith Cress
kcress -

http://www.flaminsystems.com