Hi All,

I had a fellow electrician ask me something up here North of the border. If electrical equipment maintenance has been performed, and it has been discovered through infrared scans perhaps that there is a hot spot, loose connection, etc. somewhere in the equipment or switchgear and a report is provided to the end user (client). If the necessary repairs for whatever reason cannot be completed rate away or immediately, by which means can the electrical worker be notified or advised that this particular piece of equipment could be problematic for arc flash hazard? In other words if there happens to be a time laps between the maintenance and necessary repairs?

Thanks

Best regards
Len

NFPA 70B proposes a sticker based approach to doing calibrations/PM's and this is generally the way it is done industry wide. So you have 3 stickers: pass, pass with problems, and fail. Makes it very easy to spot when something hasn't been calibrated or is out of calibration or there is some kind of problem. An example of pass with problems for instance would be if it was something cosmetic like flags not working correctly on a trip unit or it is out of calibration in that it trips too quickly. It does not impair the safety function, but it's still broken.

This is the way that we'd mark say a failed trip unit that has to be put back.

However I'd caution you against going to far with this. Yes there's a chance that the bus bars could cut loose and fail in an arcing fault. However are there personnel constantly in the area that are in danger? This is close to the old if a tree falls in the woods and nobody is around, does it still make a sound? There is an increased danger to be sure but the likelihood is still very low that it will cut loose when someone is around. Further, it does not impact the usual cause of arc flash...interacting with the equipment in a way that can cause an arcing fault. So no reason to treat the result differently.

Hi Paul

Thank you very much for sending this along. I will check 70B as you have suggested. I should have known that there would be something in 70B

Thanks again Paul

All the Best

Len

So the sound is the safest bet that something's wrong, Paul?

No. There is a philosophical existential argument about whether or not a phenomena occurs if it is not observed. And since it sounds like you are trying to poke the bear...

Equipment faults are of two kinds: hidden failures and nonhidden failures. Hidden failures are the kind where they cannot readily be observed without performing some kind of test. An example of a hidden failure is when the trip unit in a circuit breaker is out of calibration or the breaker has been neglected and it is seized up. Even when the breaker is energized or de-energized there is no external sign that the breaker has failed. However if it subsequently spuriously trips or worse never trips when it should and we can readily detect that the breaker has failed, the failure is no longer hidden. Breakers are a great example of hidden failures because even if it does trip (or doesn't), how would we know that it is not working properly? Again...most of the time, we can't tell without doing some kind of test on it.

This example points out the problem with almost all protective devices. Failures in protective devices are almost all hidden failures because for almost the entire operating life of almost all protective devices, they sit idle and do not operate (trip). We can immediately tell when production/operating devices have failed because we can readily observe them in operation. I can tell when a light bulb is burned out because it is no longer lit for instance or with fluorescent tubes they flicker. Any operator can easily issue work requests to fix these items. But with protective devices in almost all cases the only way to determine if it is working properly is by performing some kind of test to verify the condition. The worst case is fuses...we pretty much have to take it on blind faith that a fuse is working properly because aside from checking the fuse/fuse holder connection and very basic visual observations there is almost nothing we can do to test them. Fuses are type tested...representative samples are tested at the factory. For any given fuse a current test would render it useless and require a replacement.

Typically when we go in to do a lot of protective device testing this is part of the discussion and the testing protocol. We ask the customer whether or not they have spare breakers on hand. We ask the customer what tests they want performed. We ask the customer what the procedure is if we find a failure. For instance do we replace defective breakers? Do we suspend testing once we exhaust the spares inventory? What if there is no spare? Do we continue testing to at least determine the scope of the problem? Do we reinstall a failed breaker or leave it out of service? And what if it fails to where it will not close back in and can't be put back in service? The goal of these questions is to establish well ahead of testing what the protocol will be when failures are detected. That way there are no surprises on anyone's part and we don't spend hours waiting on a customer to agonize over a decision.

And now for the bear part...

So prior to testing what condition is a protective device in when hidden failures are possible? This is where Schrodinger's paradox comes in. The answer is that the protective device is simultaneously in good and bad condition at the same time. It is only after testing that we can determine the condition.

Next, testing is harmless, right? No. Testing interacts with the breaker and can affect it's condition. Not only can accidents happen during handling but the testing itself puts some small stress on the breaker and can damage it. Heisenberg's Uncertainty principle states that we cannot determine the condition without also changing the condition.

Finally, after testing we should expect no failures? Well, not so fast. First off testing is not 100% successful. Deming's quality principles states that 90% success rate at detecting failures is about as good as it gets. But worse still, Nowlan and Heap and others doing research that is the basis behind RCM (Reliability Centered Maintenance) have shown that electrical equipment failures are essentially purely random. Just because we do testing does not affect the failure rates at all, only reveals failures that have already occurred.

So does this mean we shouldn't test? On the contrary...we can be assured that the probability of a failed protective device operating when it is needed is reduced the more often we perform maintenance testing. And these arguments are purely statistical and do not address preventative maintenance such as clearing and greasing.