I've been doing a "data mining" exercise. I will post the results. The goal is to dig through OSHA accident investigation summary data to look for typical and/or likely tasks and activities that induce an arc flash. One potential issue with this data is that it is not comprehensive...it cannot be used for statistics such as average number of arc flash related fatalities per year. However one of the statistics really stands out so far: so far I have gone back to 2008 and have 68 serious injuries (requiring hospitalization), 14 minor injuries, and 4 fatalities.

This tells me two things:
1. ESFI reports about 1 arc flash fatality per year per 100,000 workers based on screening BLS data. Clearly not every case is investigated, or at least not reported in some way because I'm only finding about 1 fatality per year.
2. I can easily understand why the minor injuries are under-reported. Most of the time I'm sure that there are arc flash incidents that don't rise to the level of a "recordable" and thus go unreported. Further I'm sure that OSHA does not have the staff or time to investigate every case and thus picks the more serious injuries. Still, what is very surprising is the lack of fatalities relative to serious injuries. If the database is in any way representative on a relative basis it seems to suggest that there are about 17 serious arc flash injuries for every fatality. Does this resonate with those who watch this data closer than I do?

Second interesting take on the data is that so far out of 66 reported incidents I've only found 3 that involve switching activities. One involved a lighting panel that was repaired improperly and fell apart when an unqualified person went to switch the breakers (for lighting). Two involved medium voltage switchgear operated by utilities. It seems so far that the odds of failure under normal operating conditions are very low. Slightly more cases involve faulty equipment, and those additional cases are ones in which maintenance/repairs/modiifcations were being done to the equipment. But the vast majority, over and over again, are definitely caused by human actions.

Third take is that several cases involve using meters that are definitely NOT rated for the service. Examples of stupid things that have happened:
1. Multiple cases of using a low voltage (<1000 V) meter on a high/medium voltage system (>1000 V) with the predictable result that frequently the meter initiates an arc flash. No discussion about "Category" ratings so I can't tell even if the meters were capable of surviving this.
2. One case had a utility crew working on a 12 kV system that used a proximity tester followed by a low voltage (<1000 V) meter. After finding 85 volts to ground on each phase, they proceeded to fool around trying to eliminate the induced voltage until something happened and the meter created an arc flash. Never mind the fact that the correct approach is to use the proximity tester, then attach temporary grounding, and that the meter should not have been used at all. This is a rookie mistake and one that I've run into multiple times even on lower voltages (not understanding differences in ground potentials and induced or capacitively coupled voltages).
3. At least one case of using a proximity voltage tester on what sounds like either low voltage or shielded cabling and then cutting into the cable with predictable results.

It would seem that 70E is pretty darned vague on specifics on meters, and maybe contributes to a lack of understanding of just how to actually test for absence of voltage, and what "absent" means (0 volts? <50 volts? <250 volts?)

There have been several similar analyses conducted that I'm aware of - I think one was presented at the Electrical Safety Workshop a few years ago, and one was presented at an IEEE 1584 working group meeting several years ago. Have you seen any of these? If not I can dig them up and send them to you, perhaps they have the data you are collecting and could simply be updated.



NFPA 70E-2012 section 110.4 has an informational note about meter ratings. Sections 110.4 and 130.2(A) seem to imply that we need to be more excited about voltages >50 volts than those voltages lower than 50 volts. Granted this is vague, but it has convinced my company to disallow the common low voltage capacitive voltage detectors because most of the ones rated to detect up to 600V are NOT rated to detect below 90V.....

From personal experience - No

I encountered 6600 VAC where I was assured by the manufacturer there would be only 600 VAC.

I was using a Fluke 77 with leads rated at 1000 V. The tip of the test probe was burned off, the busbar had a 1/4" divot in it, and the meter was totally fried.

PaulEngr, your data seems to be in line with a cursory investigation I did a couple of years back. Not dismissing the devastating effects of arc flash and the seriousness of the injury, I do not see arc flash incidents justifying every bit of effort being put into it. Especially based on the current data. I agree with an awareness, training, and basic protective equipment but let those doing the work decide.

Arc flash standards, equipment, and discussion will never go away as there is way too much money involved. I just wish we could be more practical based on the research. Either that or add side air bags and computers to monitor your every move!

Not sure I can go there with you.

The money involved by the way is insurance money. I hate to say it but those major injuries that involve hospitalization in burn units for extended periods of time, plus all the skin grafts and reconstructive surgeries that follow, are vastly more expensive than fatalities. I hate to say it but if it is a pure financial argument, the fact that arc flash is usually survivable is even worse.

Generally the way that risk analysis methods treat it, if the likelihood of a major injury is about 10 times more likely than a single fatality, then they are treated with the same level of concern.

Unfortunately though arc flash injuries themselves are pretty rare. Arc flash fatality rates are half of electrocution fatality rates. After just going through the accident investigations for the years 2007-2011, I was specifically looking for cases where even if NFPA 70E had been followed, the injury still occurred. I found a few where the verbiage did not lead to any particular conclusion that proper safety practices were not followed, but these were in a very small minority.