I was recently handed this article for comment

http://esps.ca/wp-content/uploads/2016/ ... 016-A5.pdf

A question on "normal operation." We've done energy calculations for most of our LV equipment, and have some half-arsed broad brush values based on ArcPro for MV equipment, from which I wrote a guidance document (what seems like yesterday but was actually an eternity ago in 2009) to tell people how to behave.

Back when I put the guidance document together and had lots of discussion (mostly I pestered Hugh Hoagland, but also a few others) and after much agonizing decided, based on the "interaction with" trigger, to tell people to suit up even for opening and closing almost all breakers and switches. I made one exception which was for moulded case breakers because nobody I talked to could think of one blowing up in their face (various other failures, yes, but not causing an arc flash), I considered that if I made people wear PPE for this out of convenience for producing the document, that compliance to be a problem (since everyone would recognize the probability to be low).

I have an issue now related to a poorly configured generator panel where regular machine testing requires an operator to stand in front of the very first 2000 A 600 V breaker cabinet (no upstream protection whatsoever) to control and monitor generator function. Right now they're not allowed to "interact with" the generator controller on the breaker while standing in front of 100+ cal/cm^2, so they turn it on with a delay allowing them to leave the room, then once running they come back in and take readings, but it's a stupid situation where they're in and out and don't really have an emergency stop etc.

We'll be adding some upstream fusing that will knock down the energy somewhat, but maintenance staff are also quite keen on duplicating controls out of the room so that they're not exposed to arc flash at all during this "normal operation." How are others handling this sort of activity? How do you practically interpret the risk analysis that equipment that (to paraphrase slightly) is properly installed and maintained and enclosed (though not a tested arc resistant enclosure) does not have a significant probability of causing an arc just regular open/close? Do you have any equipment that requires suiting up for normal operation?

what about:
- Medium voltage non-arc-resistant metalclad? (Medium voltage inherently being of higher risk of flashover - I don't have the 2015 CSA Z462, but in 2012 CB operation in table A was HRC 2)
- Old equipment that hasn't given indications it might fail, but is still kinda looks scary?
- Any styles of breaker or switch mech that might seem to present more of a hazard, but haven't really caused any problems?


Thanks,
Jody Levine
Hydro One
Toronto, Ontario

First off, the full phrase is interacting with the equipment in such a way that is can create an arc. Examples of ways that it would not be a problem for instance is operating the normal control functions, reading panels, paging through data on a display, and operating the selector switch and getting readings from amp gauges and volt gauges.

Moving up from there, disconnect switches are just about as safe as you can get. The likelihood of an arcing fault from a disconnect switch is around 10^-12 incident/year according to IEEE 493 data. Breakers are actually down at the bottom of the scale and borderline on what most people consider safe enough for industrial work, around 10^-5 to 10^-6 incidents per year depending on how it is constructed.



Theoretically the idea is that metal clad gear would be safer since theoretically the phases are isolated. However I've never seen metal clad gear that is isolated all the way up to each phase of the breakers and all the way into the back around all the bus bars...so metal clad is super expensive but seems to have no real additional value. Even if the arc does start within the isolated section, it will be magnetically propelled down to the ends, which are of course where the isolation doesn't exist. Under the 2015 edition they did away with all this silly stuff and now a circuit breaker is a circuit breaker. The only guy still hung up on the idea that medium voltage gear is somehow mysterious and less safe (not proven in theory or practice) is Roberts.



What do you mean by "looks scary"? The problem with CSA Z462 as well as 70E is that the definitions are CRAP. They don't really give you any guidelines as to what is acceptable and what isn't. Check out www.osha.gov under 1910.269, in the annexes. They give you a really good list in there that is much more concrete and gives more definition to what "impending failure" really should be all about.



First off, breakers can and do fail pretty often. I've had many of them flash, especially if it was questionable in the first place. You might not have any experience with this happening but it definitely happens. Molded case breakers are generally better than their draw out cousins because molded case breakers are intended to be in service for decades with almost no maintenance whatsoever whereas at least traditionally the open steel frame ANSI style breakers would get routine maintenance...so they have a lot more problems with especially grease gumming up so that the breaker operates much more slowly and arcs itself to death, at which point it can spread into an arcing fault and blow up. Second, the larger the breaker, the more likely it is to have problems with bearings. Letting a breaker sit for years on end allows the pressure of even gravity to push the lubricant out from between the bearing surfaces so you end up with metal-on-metal. Then it operates slowly or seizes up and...boom. Small breakers like you have in a lighting panel are unlikely to have this problem but especially above about 600 A frames IEEE 493 again has documented a 300% increase in failure rates for even molded case breakers. When I go in to test switchgear breaker (the steel frame types), typically we find about 10% failure rates. Not all of them are potential arcing faults but they definitely happen.

So candidate problem breakers would be:
1. Any that have recently tripped on a fault. I know a guy that has been burned severely from an arc flash from a molded case breaker where it was repeatedly closed onto a fault and nobody bothered to inspect the breaker for this until they tried to close it one last time. That is why NEMA AB-4 specifically states to inspect after every fault (a quick 10 second visual inspection) and why every manufacturer references NEMA AB-4.
2. Any breaker that has sat for more than a couple years without being exercised (opened and closed).
3. Any molded case breaker over 600 A that hasn't been recently tested.
4. Any open frame breaker that has not been recently PM'd within the typical PM timing (3-8 years depending on manufacturer information and what NFPA 70B recommends...older breakers and those lubricated with Mobil 98 red grease more likely than those lubricated with Dupont 3451).

That's just a rundown from what I typically see and I see a lot of failed breakers in a year.

First off I have to get the 2015 CSA Z462 - so hopefully this isn't all too out of line.



I have a nice picture of a cabinet where a 3-ph arc had been burning on the end of the bus.

We don't consider the thin insulation on bus and taped up cable terminations in MV metalclad to be AC insulation or in any way for personnel safety. I have never talked to swtichgear designers about it, but from my high-voltage test lab experience, I always thought it was there to achieve lightning impulse withstand (BIL).



Sorry that's a bit of a tongue in cheek term I use for gut-feel risk assessment. I'm often faced with people coming to me and saying that they think something is unsafe because of something's age or design issues or whatever, which suggest it's worse than other equipment, and discussion will usually lead to agreement on possible failure modes. But without adequate failure data, we don't know how likely the bad outcome is in absolute terms. We sometimes have no choice but to rely on people's comfort level as the means of assessing risk.

Where are your failure stats coming from?



don't hold back!



Wow, I had no idea that detail was there.

I'm really surprised by the face protection recommendations in Appendix E. Way less conservative than had been in NFPA 70E. Is that the usual industry practice now?



Ya sure we've had breaker failures. And racking accidents. In that scheme of things we considered LV moulded case breakers to be not scary enough to suit up.



Ah, of course, and I can imagine a situation on the distribution system where people might be troubleshooting a line while repeatedly reclosing, and we have some new gear with the reclosers at ground level where, if we allow "normal operation" with no additional barriers, someone might consider that activity "normal" and OK to stand in front of it and close into faults.

Great stuff thank you. So nice not to have to reinvent the wheel.

\Thanks!

Jody Levine

Noticed that now. IEEE 493.

NFPA 70E and CSA Z462 are nearly word-for-word identical except for two areas. The annexes (and the mix of annexes) is somewhat different...CSA Z462 has more. And the numbering under the CSA version follows the CSA numbering scheme while the NFPA document follows NFPA formatting rules. With that being said, you can get on the NFPA web site and read the document online for free. So it's not the same as having a free copy of either document in your hands but if all you want is access to the task tables, it's not a bad way to go.



This is where it gets interesting. First off solid insulation has a significantly higher dielectric property compared to just air. Rubber tape that we use for insulation on motor terminations for instance is about 10 times more insulative per mm compared to air...check the technical specifications and you'll see what I mean. Busbar insulation is usually FBE (fusion bonded epoxy) or some kind of wrap. Either way, a little goes a long way compared to air.

For this reason it went around the academic circles for quite a while that insulated bus bar would stop an arc flash dead in it's tracks, or at least prevent arc tracking. Turns out that this is not true, as experiments have shown.

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



Different equipment/environment is the argument given. But the real reason for this is that wearing a balaclava and/or face shield was a VERY heated and contentious issue and OSHA eventually went with a less conservative approach compared to NFPA 70E. One of the strongest arguments was that the face shield sticks out and would get in the way when attempting to climb a pole. The arguments came most strongly from linemen. These are the same guys that are already being subjected to wearing gloves AND SLEEVES (even when sleeves serve no safety purpose), already stuck often wearing various hi-visibility vests/jackets/pants, already being stuck wearing FR clothing head-to-toe, already required to wear logger style boots with right-angle heels for climbing ladders, and it gets a lot worse still if you are pole climbing which is still the practice where bucket trucks can't go like in swamps. So understandably the same group pushed back HARD on the idea of having to reduce their vision and being wrapped in a sweaty head covering all day long unless it was absolutely necessary.

There are a number of other assumptions (15" vs. 18", 1.2 vs 2.0 cal/cm2) that are different as well. But before you get to the point of saying that 1910.269 is somehow more or less valid than 70E, consider this. When you really drill down into the "model" under 70E, it is really just a set of assumptions and guesses that has withstood the test of time. For instance the 18" working distance is assumed to be the distance from the point of fault out to the head/chest area (a point roughly between the chin and chest). If you actually run the anatomical calculations, it doesn't hold up. OSHA used 15" because if you are holding onto an energized component with rubber gloves, it is usually ergonomically about 15" away from the chest. That being said, it's somewhat ridiculous because the only spot on the body that applies is the chest area. The arms and hands are considerably closer to a potential arcing fault, while the rest of the body is subjected to considerably less incident energy either by virtue of being further away or by virtue of the fact that the cosine of the angle plays a role in the amount of thermal radiation absorbed, never mind the fact that the PPE is "thicker" if we look at it at any angle other than perpendicular. Never mind the fact that the "threshold" of a 2nd degree burn has less than a 25% chance of fatality even over a large portion of the body, or arguing about "energy absorbed" vs. thermal flux (radiation...watts/meter2, not joules/meter2). The reality is that the model defined by 70E is just that...an arbitrarily contrived model for estimating required protection from thermal energy emitted by an electrical circuit. OSHA made different assumptions and came to different conclusions both with respect to the input parameters for IEEE 1584 for instance and as to the required PPE.

When none of this stuff is based on "hard science", we open ourselves up to a negotiated safety standard and that's what OSHA 1910.269 is. NFPA 70E is no different in this way. Someone proposed a "model". The details were argued over and eventually we arrived at a set of assumptions that we have worked with ever since. But if you dare pull the curtain back, you quickly find out that the Wizard of arc flash is an illusion of belief...a belief that works for what it is but it's still just a set up assumptions at the end of the day.

While I got involved in arc flash in my utility because of my work with grounding and it doesn't take a lot of my time, I spent a large chunk of my career playing with insulation in a high-voltage lab so I would delight in discussing the details of why dielectric strength is only one factor in the success of an insulation system. But we should probably move it to the "off topic" forum. Or by email or phone if you're interested.


Just skimmed it -- looks really interesting, thank you!


So people are generally sticking to requiring some face protection above 1.2 cal/cm^2 in industrial/stations environments?


I said that solid insulation was off topic but I'll suggest that standards philosophy is not.

This is our dirty little secret as engineers, eh? Just how much stuff we've pulled out of a hat (or worse) and shoved into a standard, and then the technology changes and nobody can even remember how it got there decades later it and it generates a lot of heat flux when somebody suggests re-examining a test or a model or even a number?

When maintenance staff have asked me about the arc flash numbers, I tell them it's based in an incomplete skin model and idealized fabric test and a somewhat wrong arc model and uniform radiated heat with no directionality or superhot gas or metal droplets, assumed arc electrode gap distance that's effectively wrong, a working distance that doesn't seem to make any sense. But if you put all that together, the experience is that if people suit up based on these practices and an arc flash accident happens, they tend to walk away, and if they don't quite get it right, then bad things can happen. (This opinion largely based on the Dupont/EPRI study - if there's newer better accident survey data now I'd love to know about it).

Some of you may know Francois Martzloff (I know him from LV surge arresters on IEEE SPD) and I remember one occasion over a glass of wine or beer or three talking about what makes for a good standard - it's how well does it predict performance. A test or a model doesn't have to be physically realistic, as long as the results predict performance in the field. If you can say that a thing that passes the test works well and if it doesn't pass that it fails, then the standard is good. That's what we're all aiming for.

The lack of rigor makes for a lot of frustration, I agree. But frankly I think you all have done a pretty good job on this one making a silk purse out of a sow's ear. A lot of people have got to go home at the end of the day because of it.

No, it's very relevant. The basics is that the interface is what matters and as you said dielectric strength is just one factor. The other one determines how much voltage is seen by one insulation layer vs. the other in a multi-layer system, and then there's creep along the surface...all contribute but my major point is that there is a claim floating around the academic world, INCLUDING Part 1 of the article that I linked earlier, which claims that insulated bus is pretty much the solution to arc flash. Obviously Part 2 almost entirely refutes this based on experimental evidence. In industrial equipment from where I stand right now if you wanted to "fix" arc flash, I'd say that there are three keys to this:
1. Design equipment right up front to deal with arc flash by design. We can definitely get down to around 8 cal/cm2 in most cases without anything exotic.
2. Right now other than refinement of the basic IEEE 1584 model on a theoretical front, I think we can pretty much "nail" the amount of heat released. The trouble is how much is actually seen by a victim is an entirely different matter. Enclosure design is a big deal. Right now all the testing is pretty much based on an open box with nothing inside it except a couple bus bars. That is NOT reality. Every test in real equipment produces drastically lower results. The energy is either blocked or absorbed by equipment inside the enclosure. NOT figuring this in leads to grossly misleading numbers. And that's in a "typical" industrial scenario (e.g. MCC's and panelboards). It's another matter entirely in a way for switchgear and other larger equipment...approaching utilities. Right now all the testing is done under 5 feet away. The few "long range" tests that have been done show that incident energy quickly moves towards a spherical answer (exponent x=2) within about 5-6 feet, which makes most of the large enclosure testing in realistic real-world cases invalid because the current model has a very low exponent for this type of equipment...only valid for close-in working distances and relatively compact equipment such as 2-high switchgear and Class E1/2 MCC's.
3. We're not even close to handling plasma. There is a lot of preliminary data on it but so far nothing capturing the effects of plasma or gases or whatever you call jets of extremely hot material which tend to drive the incident energy well above IEEE 1584. On the other hand it has been shown that the PPE also has a big impact here...rain suits are about twice as effective.


Just skimmed it -- looks really interesting, thank you!



In Z462/70E in Annex H it seems to mandate this. In the older editions and the table-based approach it seems to only mandate this starting at 4 cal/cm2. And as you've just read the OSHA/utility version is also quite a bit higher. In fact some versions state the balaclava is only required at 8-9 cal/cm2 and the face shield starts at 4 cal/cm2. I've done some semi-theoretical models...basically look at the incident energy at the back of the head vs. the front of the head and what I found is that the drop in incident energy is very little. Couple this with the fact that the real reason for the balaclava is because the hot gases tend to come around the edges of the face shield and such, and it sure seems like if you need a face shield you probably need a balaclava.



It's all human nature stuff. Quite often when it comes to developing policy and procedure when we are faced with incomplete information, we do our best to come up with something that is at least plausible. One of the major difficulties in engineering circles is balancing theoretical nuances with practical reality. Right now there are a couple really good theoretically based arc flash models that require computer modelling to actually implement and there are still some finer points of the model that have not yet been completely solved so for now there is sort of an arbitrary equation there. There probably isn't enough money or time to fully research the issue at this point, and with all that being said, these models only improve accuracy over the existing IEEE 1584 empirical model by about 10%. So it's not "state of the art" but it's close enough for many cases.



It's not quite that bad but the idea is there. Put another way there is so much conservativism built into the model that it probably never gets "tested" in the real world. Values such as the above article with actual testing on MCC show that incident energy calculations are often off by a factor of 2 or more. Then there's the fact hat the arcing current is a "long tail" distribution so that if you just run the calculation once through with 100% arcing current, the incident energy is right about 75% of the time. The 85% factor gets it to somewhere around 95% of the time due to that "long tail" distribution issue, but also means that even if field conditions matched lab conditions, 75% of the time we're overstating incident energy. Then there's the fact that when you look at the testing data that actually calculates an ATPV value if you go about 0.5 to 1.0 cal/cm2 below the rated ATPV you get 100% protection, and that's with an arc that is perpendicular to the fabric (worst case). And since if you look at your actual data you will find that there are usually only a couple spots where you are within 0.5-1.0 points of the ATPV rating, then 90%+ of the time, you are grossly overprotected in reality even in a worst-case scenario, IF the actual incident energies were what is estimated. If we start multiplying all these various probabilities together to determine how often workers are actually running anywhere close to being exposed to the actual incident energy using PPE that is exposed close to it's ATPV, the odds are heavily weighted against this. So it's not that I think that the Dupont/EPRI study is invalid, and it is certainly one of those contrived ways of making the point to non-engineers, but that I think it is highly unlikely that a marginal scenario will occur very often, and that's with an event that is in itself pretty rare.

On top of that another interesting angle is that this is all about survivability in reality. The contrived model aims to make it a "walk away" scenario only 90-95% of the time because IEEE 1584 itself states that it is only accurate 90-95% of the time. Having first degree burns over most of your body (in the contrived case) or burned hands/arms (in reality) is hardly a "no injury" result. 70E and CSA Z462 are aiming for survival, not "no injury". What's interesting is that due to these various factors it is been remarked that even grossly under-rated PPE is unlikely to allow more than burn exceeding 25% of the body as long as it's FR (doesn't propagate a flame), which means that the chance of survival might drop from say 85% down to say 75% or something like that...still pretty good. Given that analysis of OSHA injury data shows that the fatality rate for arc flash injuries is between 10:1 and 20:1 (total reported injuries to fatalities) already, it seems like even if workers just wore FR PPE regardless of whether they were wearing the proper rating, their chance of survival goes up dramatically.

Personally I currently work as a field service contractor. I'm in and out of many plants every week. Most of my customers are plants that can't afford full time electricians never mind engineers. A lot of them have done anything about arc flash at all. Most of the time I have no idea what the incident energy of the equipment I'm working on is as a consequence. I'm almost always wearing FR pants and most of the time either an FR shirt or FR jacket or both at a bare minimum. I try to adhere to the table method in 70E because that's all that I have to work with. This is the best I can do given that I can't charge a customer $20,000 to do an arc flash study just so that I can come to their plant and test a motor.

If you are in the utility business, then you are facing the exact same issues. We don't honestly have a way of determining incident energy above about 15 kV as far as we know simply because we have NO test data at those voltages, only conjecture. On pole lines, arc tracking (the tendency for the arc to be magnetically propelled along the power line) has a major influence on incident energy. The vast majority of testing is on three phase and not single phase arcs, but single phase arcs are the vast majority of utility scenarios. Due to the fact that utilities tend to be network distribution and not nice, clean radial or even loop distribution, and the network configuration changes regularly, it's also hard to estimate incident energy at any given time. And it's hard to even gauge incident energy even if we could overcome these things because we don't have a limited number of specific locations to analyze...we can work on a power line anywhere along it's length. So although we can use some data from EPRI and some estimating approaches to at least estimate bolted fault current and thus incident energy to some degree for some substation cases, it's hard to have any kind of hard and fast rules. So the reality is that for most utilities they are no better than us in the contract maintenance business...most of the time you have no idea what the incident energy might actually be and all those standards are not very practical for the most part.



That's a major portion of my thesis defense when I was looking at efficiency (energy per ton) of full scale grinding mills which were grinding ore from 6" rocks down to 100 microns. The theoretical models with 4 or 5 adjustable parameters gave a much better curve fit but the confidence intervals were so wide that the results were meaningless compared to a very simple model (energy per ton=constant). I saw it first hand in the defense...the professors (academics) in the room were completely hung up on the idea that a theoretical model is far superior to my work and had a hard time coming to grips with the idea that the trivial model was actually superior in every way, even when the math was staring them in the face and making the point for me.

As I understand it from one of the IEEE 1584 committee members, that's also what's going on within that standards committee...you have the academics that want to study the issue for another few decades and who quibble with each other over details that don't really matter. Then you have the "do something" crowd that don't take into consideration the consequences of what they put in the standard, and the pragmatists that truly want to do something about it but facing either "do nothing" or "do something stupid", drag their feet and try to prevent any forward motion at all, or push for a theoretically unsound but practically workable solution. At least within NFPA, it's sort of an up/down voting process. The problem with that approach is that you can get "stuck" on some things that end up in the standard and are very hard to displace because the process doesn't lend itself well to large changes, only minor incremental ones.