It is currently Sun Sep 19, 2021 10:02 pm



Post new topic Reply to topic
Author Message
ekstra   ara
 Post subject: A philosophical discussion of how to conduct arc flash.
PostPosted: Thu Aug 05, 2021 11:35 am 
Arc Level

Joined: Thu Jan 10, 2008 8:49 pm
Posts: 510
Location: New England
I have some basic issues with the way arc flash studies are conducted, especially when using software.

First, there seems to be a goal to arrive at an IE number out to two decimal places. In reality, these numbers were arrived by empirical means and do not follow any law of physics like F=mA. So they can't be precisely calculated. They are made to fit a formula based on testing and under those lab conditions. They are a best guess, and that best guess may be significantly different under a specific user condition. So is there a safety factor we need to apply?

Second, the formula to calculate IE incorporates arcing current linearly, but clearing time exponentially. So, in a way, having high fault currents are beneficial because they cause the breaker to respond instantaneously without delay and clear the fault quickly. This seems counter intuitive to having low fault currents. But this is the reason we look at arcing current and 80% of arcing current. But is the reduced 80% actually low enough to cover all possibilities.

Third, do breakers really trip in 2 cycles when the instantaneous range is reached? Isn't that an optimistic statement that all breakers when triggered in their instantaneous range will clear in 2 cycles. Yet, doesn't NFPA reference in their general task guideline fault current and 2 cycle clearing times? Aren't they here making a statement that the majority of industrial applications will clear in 2 cycles? Surely, they are not basing their task list on the few exceptions. 2 cycles is a standard we have all been taught, myself included. But if I am making a safety decision to protect a human, should I lean to the side of caution and approximate that 12 year old breaker to clear in 3 cycles, or 4, or even 5 cycles. That 'time' value which is squared is now changing the IE value a lot in the equation. But aren't the software calcs based on the manufacturers curves?

Fourth, what about the utility. How often has the utility called you to inform you of a change to their grid, and that the fault values you were given previously, have now changed. IE has been calculated out to two decimals based on the previous values, so are those 200 labels you have in the field now all wrong? Oh yes, every five years we have to re-evaluate the values to assure they are still correct. A lot can change in five years and this again is a safety practice to protect human life. Maybe we should have used a safety factor in our original calculations to compensate.

Fifth, arcing current is based on three phase fault current. How often is a worker going to create a three phase fault in comparison to a single phase fault to ground? Is he really going to accidentally have his screwdriver across all three phase conductors, or is he more likely to slip and have one phase bridged to ground? If its the later, how accurate are the IE values. Single phase to ground current is going to be significantly less than three phase bolted fault current. Did the lower arcing current value now shift the trip time into the short time portion of the breaker curve? Is that time squared value in the formula raising the value of IE faster than the lower linear value of arcing current is reducing it? We don't know, because we didn't calculate it.

So what this rant is trying to question, is if we are applying the calculations correctly with the goal of protecting the human, vs arriving at an IE value to two decimals. Is it wrong, if the IE value on the label is higher than the conventional calculation would predict? My response is NO. The goal is to protect the human, not to precisely calculated an IE value which in reality can't be precisely calculated.

So, with all this said, I want to question if we should do this work of IE calculations better, and propose a possible method.
I want to put the IE value on the label that constitutes the worse case scenario. But in order to do that, I have to explore multiple conditions. Is the higher fault current going to raise the IE, or, is the lower fault current going to raise the IE because the clearing time has been increased. Is the accidental arc contact going to occur when production is in full swing and all motors are on, or, is it during a shutdown and the majority of production equipment is turned off. So do I calculate with motor contribution or no motor contribution.

I haven't completely thought this out, and would welcome other opinions, but here is a possible course or action:
Run the numbers with the utility values at 125% of what the utility tells you they are.
Run the numbers with the utility values at 75% of what the utility tells you they are.
Run the numbers with full motor contribution
Run the numbers with no motor contribution.
Run the numbers in the above scenarios, but with single phase to ground fault current.

Then, whichever IE value is highest, put that value on the label.

My next step is to explore if there is a shortcut method, say just two scenarios, the highest fault current and the lowest.
So one run with 125% utility fault, plus full motor contribution, and three phase fault values.
The next run with 75% utility fault, no motor contribution, and single phase to ground fault values.

My apologies if you think I am way off base. I did not mean this to be a diatribe to the standard method. These are my personal opinions and I am truly seeking advice from others to validate or disprove these observations.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: A philosophical discussion of how to conduct arc flash.
PostPosted: Thu Aug 05, 2021 12:33 pm 
Arc Level

Joined: Tue Jan 13, 2009 5:00 pm
Posts: 566
Some very valid points.
Look for "A Graphical Approach to Incident Energy Analysis." The paper itself can be purchased from IEEE, but discussions here and elsewhere can get you the gist. With the processors we have now, why are we limited to one current at a time?


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: A philosophical discussion of how to conduct arc flash.
PostPosted: Fri Aug 06, 2021 11:37 am 
Arc Level

Joined: Mon Jan 18, 2010 11:35 am
Posts: 543
Location: Wisconsin
I agree 2 decimal places is silly. Is 7.99 really different than 8.01, yet the PPE requirements are significantly for most companies.

Yes modern designs for molded and insulated case breakers can clear within 2 cycles.

The engineering group I worked for often ran 4 different scenarios for each project. For industrial facilities a scenario with no motor loads is important to address those maintenance tasks performed during production outages.

Except for open air conditions NFPA70E assumes all single phase faults propogate into e phase faults almost instantly.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: A philosophical discussion of how to conduct arc flash.
PostPosted: Mon Aug 09, 2021 7:35 am 

Joined: Mon Nov 09, 2020 7:31 am
Posts: 1
All good considerations. I had another scenario that may question the 2 second rule. What if the electrician is in a bucket 30 feet in the air, trying to install a bus plug into a bus bar. They couldn't get away so quickly, and certainly not in 2 seconds. Just a thought experiment.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: A philosophical discussion of how to conduct arc flash.
PostPosted: Mon Aug 09, 2021 7:49 am 
Arc Level

Joined: Tue Jan 13, 2009 5:00 pm
Posts: 566
All good considerations. I had another scenario that may question the 2 second rule. What if the electrician is in a bucket 30 feet in the air, trying to install a bus plug into a bus bar. They couldn't get away so quickly, and certainly not in 2 seconds. Just a thought experiment.

If you carefully read the "2 second rule" in the IEEE standard, you'll see your thought experiment is not just a good idea, it is part of the rule. The standard requires us to use engineering judgement. Perhaps we should stop referring to the "2 second rule" and call it IEEE 1584 6.9.1.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: A philosophical discussion of how to conduct arc flash.
PostPosted: Mon Aug 09, 2021 8:36 am 
Arc Level

Joined: Thu Jan 10, 2008 8:49 pm
Posts: 510
Location: New England
Except for open air conditions NFPA70E assumes all single phase faults propogate into e phase faults almost instantly.

Is this guaranteed 100% of the time. Because I have seen accidents where a screwdriver bridges phase to ground without resulting in 3 phase fault. Plus, I am not arguing that the 3 phase fault doesn't result, I am advocating to check for the worse case in which it remains a single phase to ground arc with lower current and longer time to trip. You check both, and record the worse.

2 cycle clearing does occur and is being tested to that standard. But, its still a best case. Since time in the IE equation it squared, I'm just advocating some safety factor be applied. Isn't good engineering practice to incorporate a safety factor. Bridges have a safety factor of 2. Aircraft has a safety factor of 1.5. How comfortable would you be flying at 30,000 feet in an aircraft with a 1.0 safety factor? You'll be perfectly safe, as long as every assembly component is perfect.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: A philosophical discussion of how to conduct arc flash.
PostPosted: Wed Aug 11, 2021 8:09 am 

Joined: Fri Nov 06, 2015 1:05 am
Posts: 26
Location: Evansville IN
I am thinking there has been years of this discussion and the committee meets on a regular basis. The committee puts years of thought into this same topic. I believe many Ph.D. EE have input also. I believe you should join the committee or seek out minutes from the last five or six years of discussions. You defiantly have the knowledge to join this group. I always look forward to seeing these posts and responses. I know the group likes the input.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: A philosophical discussion of how to conduct arc flash.
PostPosted: Wed Aug 11, 2021 1:24 pm 
Plasma Level
User avatar

Joined: Mon Sep 17, 2007 5:00 pm
Posts: 1616
Location: Scottsdale, Arizona
Quite a bit here. You are simply stating what I'm sure a few others are thinking.

I will try to offer some additional insight

Legal Disclaimer: These are my personal opinions/thoughts not representing any particular committee - required legal disclaimer due to my various committee positions
.
First, hard to imagine but arc flash calculations are still in the early years (yes 20+ years) and what began as a technical paper based on tests in the late 1990s morphed into the 2002 edition of IEEE 1584. A lot was learned since then which led to the 2018 Edition. So there have really only been 3 iterations and each one is better than before.

An issue that I see is this is somewhat analogous to predicting lighting. If you know the charge of a cloud, the earth, the humidity and a host of variables at any specific moment, you could probably accurately predict lightning but, for any given event, that info is mostly unknown. Similar to an arc flash, what is the actual gap distance (it can vary in one piece of equipment) what is the electrode configuration (again, there can be a few types in one piece of equipment) where did the fault actually occur etc. etc.

(my opinion), the 2018 edition of IEEE 1584 addresses more variables than ever before but… for a given arc flash, what was the specific condition at the moment. I use the phrase that we are becoming very precise at calculating of a very random event.

Regarding your five points, I will try to address them. Read on….

1) I agree 2 decimal places may be a bit much but many do this. The 2018 model was created with open boxes for repeatability of the tests/results. The data tracks well with the equations but… actual equipment may behave differently than an empty box.

There was an IEEE paper years ago that illustrated actual equipment yields lower incident energy than a box/calculations. I can’t say if there are outliers, that is just one paper. I believe the thought was equipment in the enclosure may play a significant role. That’s about all that can be said – anything else is speculation or an assumption (which can cause legal trouble.) So open boxes were used for repeatability of the tests, data and equations.

2) Actually arcing current is quite nonlinear. The latest edition of IEEE 1584 uses large polynomials and logarithms. (sorry for the math flashback) However the duration is linear so as an example 2 cal/cm2 over 3 cycles would be 4 cal/cm2 over 6 cycles (twice the duration).

3) Yes, I believe software uses TCCs without any extra time margin. This was discussed years ago in the working group and the conclusion was IEEE 1584 provides the model and the input data is up to the person performing the study. Therefore, if someone wants to add a margin, they are free to do so although I’m not sure if people do.

4) Changes = Relabeling. This could happen and can be expensive and frustrating (as you point out) which is why I introduced and have been a strong advocate of an alternative labeling strategy (that still complies with NFPA 70E). It is easier (my opinion) to list the arc rating of the PPE on the labels and standardize the Arc Flash Boundary (AFB). That way as long as the calculated incident energy and arc flash boundary don’t exceed the arc rating and standardized boundary on the label, no need to re-label.

An example, the label lists 8 cal/cm2 as a minimum PPE rating and a standardized AFB of 10 feet is used, then if the calculations change from let’s say 4.3 cal/cm2 to 5.2 cal/cm2, the 8 cal/cm2 PPE on the label would still be considered sufficient based on the rating. Same if the AFB changes. When the label is reviewed and deemed still in compliance, than add a maintenance sticker much like relays stating it was reviewed and have it list the review date. Good to go for another 5 years or when a major change occurs. In a similar sort of way, this is what is done with a short circuit study – calculated the fault currents and make a comparison to device interrupting ratings (usually listed on the equipment) in the report. For an arc flash study, calculate the incident energy and arc flash boundary and compare to the arc rating and standardized AFB on the label. Make a table to show what is still OK and flag (and relabel) what is not. Re-labeling will be greatly minimized.

Here is a video illustrating this strategy: Video: Alternative Labeling Strategy

Also, from experience heading up the short circuit studies group of a large utility in a past life, the short circuit current can change over time. If the facility is fed from a transformer, the transformer impedance will serve to buffer the system somewhat but it can still vary.

Here is my view on a possible work around without utility data: Utility Short Circuit Current/Work Around

5) Agreed a single-phase fault accounts for an overwhelming majority of faults. I equate this to a flat tire. It is more likely for one tire to fail than all 4 simultaneously – basic probability. Similarly, it is more likely that as you state, that one phase shorts out and not all three.

However, the arc flash is a conducting plasma so it is possible (not guaranteed) that the plasma from a single-phase arc flash escalates into a phase to phase event and then escalates to a three phase arc flash in a few cycles. It is impossible (unlikely?) to be able to determine what will escalate and what will not so we use the worst case assuming everything escalates to three-phase.

I've seen pretty weird things happen in the lab. One was an MCC where the arc flash shot to the bottom away from the source on top, then re-initiated at the top again - we assume from the plasma rising back to the top.

With all that said, the 2018 edition is better than anything we’ve had but there is always room for more work and improvements. We will continue to learn more every year. The committee is made up of all volunteers and most also have day jobs so the research and effort move quite slow. In addition, the consensus standard process is also slow moving.

The good news is that according to Bureau of Labor Statistics shown in Annex K of NFPA 70E, the data indicates there is downward trend in electrical incidents and fatalities so electrical safety, arc flash awareness and most important – reduce/eliminate energized work has a positive benefit.

May not be the answer you were looking for but I hope it provides additional insight on where we are and no doubt, where we need to go in the future.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: A philosophical discussion of how to conduct arc flash.
PostPosted: Fri Aug 13, 2021 11:26 am 
Arc Level

Joined: Thu Jan 10, 2008 8:49 pm
Posts: 510
Location: New England
Jim,
Your comments always provide a great deal of insight into this topic as most all here will acknowledge you as the de facto expert.
But if I can clarify some of my points.

2) Actually arcing current is quite nonlinear. The latest edition of IEEE 1584 uses large polynomials and logarithms. (sorry for the math flashback) However the duration is linear so as an example 2 cal/cm2 over 3 cycles would be 4 cal/cm2 over 6 cycles (twice the duration).

My point here is that the single phase fault will significantly reduce arcing current, and, just possibly, the trip point moves from the instantaneous region of the curve (say 2 cycles) to the short time portion of the curve (say 6 to 8) cycles. So IE can increase linearly, but, it could be 2 to 4 times that which was calculated 3ph arcing current. Hence, when you mention 'the worse case'
we don't know what the worse case is by just following the suggested format of 3 phase fault. We have to run multiple iterations to determine the worse case.


4) Changes = Relabeling. This could happen and can be expensive and frustrating (as you point out) which is why I introduced and have been a strong advocate of an alternative labeling strategy (that still complies with NFPA 70E). It is easier (my opinion) to list the arc rating of the PPE on the labels and standardize the Arc Flash Boundary (AFB). That way as long as the calculated incident energy and arc flash boundary don’t exceed the arc rating and standardized boundary on the label, no need to re-label.

Whole heartily agree with this. When I was doing arc flash, I would try to convince management to agree to categories of IE levels for the labelling, IE's of 4,8, 25, 40 that happens to correspond to the old PPE levels. So if you were doing any live work, the minimum PPE was 4.0 cals. Which today would still be fine, as we no longer have the level 0 and -1 PPE levels. But when NFPA went with the push to label IE values, the market place seemed to change to this precision method of calculating IE to two decimal places. In the companies I have worked, the company supplied the PPE, so we just matched the labels to the PPE level of garments supplied. General workers had 4.0 cal coveralls, while electricians had 7.8 cal coveralls. Electricians also got 40 cal suits. So our labels were either 4.0, 7.8, or 40. Of course, they always overstated the actual.

My objective here is to get all of us to recognize that our ultimate goal is to protect the worker, and not become the ultimate engineer trying to design (or label) a system to an exact calculated IE. There is nothing wrong with performing multiple iterations and using the worse case, even if its a 'mix and match' taking the worse case of whichever iteration produces it.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: A philosophical discussion of how to conduct arc flash.
PostPosted: Mon Aug 16, 2021 7:00 am 
Sparks Level

Joined: Tue Jan 19, 2010 2:35 pm
Posts: 152
I agree with most of what Jim said in his reply. Your points all have some merit, however, it's important to consider this. There are way to many variables that are beyond our control and therefore cannot be accurately quantified. For example, the utility loading. Suppose you have a utility that has lots of capacity during part of the day and mot much reserve during another part of the same day. That will affect the fault current at the transformer and mean lower fault current and probably longer tripping times. Then consider circuit breakers. I agree that relying on breakers tripping in 2 cycles can be problematic. As we all know, unlike fuses breakers are mechanical devices and are subject to wear. So, how often do breakers get exercised and how many breakers are maintained to the manufacturers specifications? My point is, if you're looking for hard and fast rules to govern arc flash, you'll probably not going to get them. Yes, conducting an arc flash survey and study are inexact practices, but it's the best we have for now. I support your collection efforts. Unfortunately, you're probably in a small group who feel the same.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: A philosophical discussion of how to conduct arc flash.
PostPosted: Mon Aug 16, 2021 7:11 am 
Sparks Level

Joined: Thu Oct 18, 2007 7:10 pm
Posts: 263
Location: NW USA
Breaker operating time would seem to be 6 cycles and it is expected the software includes that clearing time when breakers are modeled based on an accurate library (medium voltage, where relays are modelled separately). In the software utilized (ETAP) it is possible to model a sequence of operations that verifies the 6 cycle delay. This was a big topic 14 years ago, and it was hypothesized that slower than spec breakers would compromise the all important interrupting rating. Those were the thoughts, it would be interesting to hear from a manufacturer on same.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: A philosophical discussion of how to conduct arc flash.
PostPosted: Mon Aug 16, 2021 7:19 am 
Arc Level

Joined: Tue Jan 13, 2009 5:00 pm
Posts: 566
Are the software folks listening? Human iteration on computers seems rather non-sensical. We should be able to enter a range, step size, three phase, single phase or both; and let the software graph it out showing where the highest IE lies and its magnitude.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: A philosophical discussion of how to conduct arc flash.
PostPosted: Mon Aug 23, 2021 8:43 am 
Arc Level

Joined: Thu Jan 10, 2008 8:49 pm
Posts: 510
Location: New England
I was helping a friend do an AF study on SKM, his company's software. After calling SKM we found out the following.
SKM can run up to 20 scenarios at a time, and you can check a box to have it take the worse case of each scenario.
So we built the following:
Base Case with utility actual values
-20% utility values
+20% utility values
+20% utility values with motor contribution (25% of supply KVA)

There is an input for maximum arc time, we used 2 seconds, but there is no ability to enter a minimum arc time - unfortunately.
But the tech said that when we run the report in table format, it would show the OCPD clearing time. If you right click on the time, you can unlink it from the automatic library and enter your own time. We had one molded case breaker, 100A, that showed a clear time of 0.012 secs, that node had an IE of 0.8. We though that was very optimistic so we changed it to 0.066, and IE went up to 3.6. We elected that 0.066 would be our minimum for all breakers (but for fuses we would keep the library).

The software automatically runs the 80% TO 85% OF Arcing Current.

One of the tables in the report provides footnotes, and also tells which scenario was used.
That column showed that a variety of the scenarios was used, including one that had the 85% of Arc Current used.

I want to go back and find out if there is a way to include single phase faults. Trying to do it manually is way time consuming as it changes at every local transformer and would have to be over-written manually. We did experiment with that, but didn't find it changed the values much. Maybe someone else can figure out a method to do this.

Overall, I feel more confident that the above method provides a more conservative evaluation.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: A philosophical discussion of how to conduct arc flash.
PostPosted: Mon Sep 13, 2021 8:02 am 
Arc Level

Joined: Thu Jan 10, 2008 8:49 pm
Posts: 510
Location: New England
We called SKM. They said that running the calcs for single phase fault is not possible.


Top
 Profile Send private message  
Reply with quote  
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 14 posts ] 

All times are UTC - 7 hours


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Jump to:  
© 2019 Arcflash Forum / Brainfiller, Inc. | P.O. Box 12024 | Scottsdale, AZ 85267 USA | 800-874-8883