I did a Google search asking "what percentage of arc flash accidents are human caused?� I got tons of hits, but the websites I visited stated that "most arc flash accidents are human caused". I was looking for the support data for their assertion. But no one had anything to back up their claim. For example, one website talked about a “typical arc flash injury�. They talked about an electrician who was injured because he “failed� to use an insulated screw driver when he was maintaining a 480 volt panel board. Okay, he was injured, but was what he did be classified as an accident?
I agree, it was human caused but he seemed to ignore the requirements found in the NFPA 70E. Either he chose to ignore the Standard or did not know about the Standard. If he had, he’d not tried to perform maintenance on this energized panel.
So my question is, should this type of incident really be classified as a human caused accident? Was this really an accident or someone ignoring safe practices? You could argue either way, but it seems the assertion this qualified as an accident muddies the waters as to what real human caused arc flash accidents are.
To me, there are several important pieces in how we define the incidents described. If a person knowingly ignores industry best practices, is it fair to call the results an accident? Or should it be called something else?
Thanks,

Hello Wil

I am not sure as to whether "Human Caused" may be the correct terminology? Perhaps it should be "Human Intreraction" is the correct terminology. Which opens up to a long list of items from turning on or off a disconnect switch to troubleshooting... Lets start with that. Also could you be a little more descriptive with the sentence "maintaining a 480V panelboard. What actually was the working performing?

You won't find anything. ESFI has published reports on the average incident rates for arc flash and shock and they are about equal to the overall injury rates for all sources of injuries in the work place. Electrical injury rates in general are less than 1% of all work place injuries. However when we look only at fatalities, electrical injuries are the 7th most common cause. ESFI's statistics also show that arc flash fatalities are between about 1 in 10 and 1 in 20 arc flash incidents, so clearly the 7th leading cause statistic is primarily driven by shock injuries. I would also argue that I disagree with the data. My own analysis of the data shows that arc flash incidents are actually pretty low, maybe about 10-20 incidents per year. The problem is that the ESFI data basically includes every incident of a burn that is caused by electricity whether shock or arc flash as an arc flash injury. This artificially inflates the number since it only takes a couple amps to burn flesh. But regardless of the validity of the data, it shows that the incident rates are really low. Specifically injury rates due to arc flash are given as about 0.1 incidents per 100,000 workers per year, or right at one in a million. The CPSC (Center for Process Safety) LOPA Standard gives a table of various industry and government standards for acceptable injury rates and gives a number o about 1 in a million per worker per year for fatalities. Thus the average arc flash injury rate is better than the rates given as acceptable standards.

Second source is that IEEE 493 which is now codified somewhere in the 3000 series of IEEE standards shows that failure rates due to arcing faults varies from about 1 in 100,000 pieces of equipment per year for large (800 A+) circuit breakers down to around 10^-12 for disconnects.

Taking these togehter it definitely proves that under normal operating conditions electrical equipment failure rates are equal to or better than comparable injuries from all sources. So it should be very clear that the average or "typical" case which we can summarize as normal operation of electrical equipment is safer than most industrial equipment. In terms of nromal operation I would go much further than most in suggesting that this should include anything where the failure is primarily dependent on the device itself. For example, a general LOTO (1910.147) should be considered normal operation when a breaker or a disconnect is opened or closed because the spring loaded mechanism engages or disengages it wihtout being dependent on the operator. This is very different for instance from a fused cutout or inserting a bucket into an energized MCC where operation is partly if not entirely dependent on the operator.

This means that we really only have to consider two more cases. The first is where the equipment is clearly not operating as it should and this should be a quick visual inspection to determine this. The inspection should be:
1. Has a fuse/circuit breaker tripped? If so these are indications of en electrical fault (something is not right, not normal operation).
2. Is there any evidence of overheating (bubbled up paint, burn/scorch marks, blown off doors, swelled up panels)?
3. Is there any evidence of contamination such as water or other fluids dripping out of a panel?
4. Is there any other evidence of impending failure such as physical damage to the electrical equipment, someone saw or thinks they saw a bright flash, severely rused out equipment, equipment racked, bent, or doors open?
5. Are there inspection stickers? Are they out of date, or has any inspection been done?

If any of these are true, then treat as no longer normal operation until thoroughly inspected and returned to normal service conditions.

Once we get past these, the rest of the cases are due to human performance. For lack of a better word, look out. OSHA Annex E describes the tasks where this should be a concern but to cast some like on it, the above mentioned LOPA standard as well as various human performance standards have all shown that under the absolute best case scenario human error rates might get as low as 1% and under severe stress conditions are as high as 40%. Similarly in quality control Deming suggests a general number of 10%. The same 10% number is often used as a simplified value in many risk assessment procedures. Regardless these are nowhere near approaching the "1 in a million" statistics demanded by the risk assessment standards. Since that is the case any time human performance directly determines the likeiihood of an injury, a hazard exists.

So in some ways these latter two cases are addressing "outliers" or situations where the likelihood of an injury is much higher than the expected average arcing failure rate for electrical equipment. If we can control those two issues then what's left is the basic inherent reliability of electrical equipment which as previously discussed is actually very good and not normally a cause for concern at all.

Hope that answers your question because there is no direct answer to it but in a roundabout kind of way it is pretty easy to get to the answer you are looking for.

Leonard,
I tend to agree with your comments. Human caused or interaction both can be seen as different. I was hoping to open up a discussion about how we classify accidents.

I mentioned that a website had said that most arc flash accidents are human caused. They went on to talk about one case. The particular case was human caused, but the mechanic should not have been doing any work on energized equipment. So does that accident really fall into human caused? Had the person followed NFPA 70E, he would not have done this work energized.
Seems like we could have better ways to classify individual events.
We had one person injured in an accident several years ago. He was installing a fuse holder back into the energized fuse base when it shorted out and when phase to phase. He was not wearing PPE even though he had it and knew he should have worn it. He was burned on his right arm up to his elbow. One could argue that this was both human caused and also a mechanical failure. The fuse base was poorly designed and it failed. I'd say this accident was mechanical and not human caused. But like I said, you could argue either.

Years ago, I attended a AVO training class on arc flash. The instructor sited a statistic stating that more than 70 percent of the arc flash accidents were caused by mechanical failures. I tend to agree in that most of the arc flash accidents I've heard about or seen were mechanical in nature. My point is this, it appears people may be wrongly relying on misleading statistics. May be we need a better way to classify arc flash accidents?