We'll call this a hypothetical question. Say you are running a short circuit and arc flash study on a 50 year old building as part of a multi-building study. You discover you have about a half dozen branch panelboards that are over-dutied in this building. These are all 225A rated, 208Y/120V. Some are flush in block walls in finished areas. Breakers in these panels are 50 year old, long discontinued 10KAIC rated. The feeders are all protected with FRN or LPN fuses with no published series ratings available and no higher rated replacement breakers located. Artificially adding feeder length and/or reducing feeder sizes and fuses to reduce available fault current is possible but just as costly as changing the panels out.

You learn this building is supposed to be torn down within the next 5 years.

Now then switch hats, you're the Owner and aware of the issue. Would you:

1. Do nothing other than proceed with arc flash labeling the building since the study is practically done. Afterall, this building is 50 years old and nothing has failed in all that time plus the feeders are protected with current limiting fuses so "'yer' prolly good. Yeehaw!"
2. Replace the panels with new with 22KAIC breakers even though some of them will be a bear and new guts and covers won't fit in the old tubs and you're tearing down the building in a few years.
3. Lock the panel covers and label them something like "Danger. Do not open door while energized."
4. Other - maybe there is another option or code exception somewhere that can help you resolve this.

Thanks.

I would think the owner's insurance company would want to have the old stuff changed out. Not sure of the size of the service or load but would it be more feasible to change out the transformer to a smaller size with a higher impedance?

Not feasible in this instance. Err, I mean hypothetically this building has a 1500kva 480v unit sub and a 750 kva 208v unit sub. 15kv S&C switches bussed to the transformer primaries secondaries have feeder bus down to the basement switchboards. Huge $$$ to re-do that in comparison to replacing some 200A panels.

Short circuit current rating and arc flash can be treated as individual concerns.

The AIC rating of a device is based on the Bolted Fault current flowing through it.
The Incident energy at a location is based on the amount of Arcing Fault current flowing through it. For <600V, the IEE1584 formulas always(?) result in a calculated AF that is significantly less than the bolted fault.

For a risk analysis viewpoint.
The chance of a bolted fault primarily exists only when the circuit wiring has been 'worked on', therefore the act of switching a properly operating molded case breaker is not likely to result in a full rated bolted fault.

How many feet of circuit conductor is needed to lower the fault current below the AIC rating? For example, a customer was willing to accept the risk associated with overdutied breakers once we determined that a minimum branch circuit length of 30' lower the 'bolted through fault current' to
<10kA.

Consider current limiting fuses. Check out the peak let thru charts from Cooper Bussman or Littelfuse or Mersen. This might be a reasonable solution. It's possible that using this kind fuse would reduce the maximum current enough to allow you use the 10 KAIC breakers. I attached a pdf from Cooper Bussmann.

I believe you can't use the current limiting effect of fuses to protect lower rated breakers. Circuit breakers have a dynamic impedance incompatible with this method.

Before throwing out the fuse idea, try calling Bussmann or another of the fuse companies. They'll likely ask you for particulars and tell you if current limiting fuses might be a reasonable solution.

1. Be careful what short circuit model you use. Some of them make a LOT of assumptions. Also as with arc flash be sure to include realistic conductor lengths and models. If the problem hasn't shown up in almost 50 years, chances are that there's simply an overestimation going on. I've run into this before...fixing the engineer's mistakes in the model "fixed" the problem.
2. Overdutied equipment means that busbars have the potential in a dead short to break free of the supports and fly apart. For breakers this means that the ensuing arc can destroy them or that they simply won't open...pushing the fault to the next upstream device. That's where the fuse advice comes into play. It may be beneficial to delete the device out of the model to check what would happen if the overdutied equipment fails.
3. Been there, done that with the insurance companies. They really hate it when you ask how much it will lower your premiums if you correct the problem (or conversely raise your premiums if you do nothing). With equipment this old what you will probably found out is that they've already factored that into their actuarial number so it makes no difference at all as far as insurance goes. So then it comes down to risk management in terms of injury to employees (if anything). As I've posted elsewhere, there seems to be a threshold right around 240-250 VAC where arcs just become self-sustaining. Below that point (208/120 was mentioned) it isn't as big of a problem. There has been a fatality at these levels of incident energy but the risk here is extremely low...hence the reason that currently IEEE 1584 gives a pass to 208/120 equipment.

Regarding the 125 KVA and below 240 volt "exception".
That exception was removed from in 2012 70E Standard. If the 2012 NFPA 70E Standard pointed that out, I missed it. I learned about the deletion because Mersen Fuse sent out a news letter detailing the changes in the 2012 Standard. They made a point about the exception being removed.

The exception was removed from 70E. It was not removed from IEEE 1584. But that's just one of the 8 arc flash models referenced in Annex D and the only one where the exception exists. The other models do not have an exception so it would be inappropriate to have it in 70E except with respect to the table-based method that 70E specifically supports. Thus it was entirely appropriate to remove the exception but it is not appropriate to conclude that this means that a significant risk exists for low voltage arc flash incidents.

I've summarized the exception as well as a whole lot more data here:
https://drive.google.com/open?id=0B6mGR ... Dl5VzlDSWM

The big thing to realize is that even though it is possible to die from an arc flash injury below 250 VAC, the likelihood of such a thing is extremely rare when you consider that OSHA has only documented one fatality in 20 years..that amounts to millions of worker-years and vastly below the generally accepted actionable threshold of "1 in a million" used by government agencies.

As a similar example, consider being struck by lightning. It should be pretty obvious that being struck by lightning has a pretty significant chance of being fatal, if not severely injured. The likelihood is about 1 in a million per year. This is close to the point where OSHA would require us to address it except with some simple common sense recommendations such as not playing golf in a lightning storm, or similar activities.

In the case of an arc flash injury, the odds of a fatality (or severe injury) from a system voltage that is under 250 VAC is much less than being struck by lightning (probably closer to being struck by lightning twice) so there's no compelling requirement to address it so long as workers are not taking undue and unnecessary risks such as attempting to disassemble a temporary construction panel while wearing flip flops and shorts instead of waiting for the lineman to arrive to open the cutouts (the incident that actually happened).

The rest of the review that I did attempts to quantify the hazard. The issue with using the IEEE 1584 empirical equation or really any existing calculation method for that matter is simply that all of them start with data from stable arcing faults. The model "works" because the incident energy is proportional to time once we can take into account the arcing voltage and arcing currents. Below 250 VAC and especially at 215 VAC and below, this assumption is violated. Arcs tend to spontaneously self-extinguish. None of the models have an upper bound on arcing time other than a rough rule ot thumb of 2 seconds. So they will grossly overestimate the incident energy for real world situations. The best way to data it is to upper bound it based on experimental data. Thus IEEE 1584 for instance upper bounds everything 208 VAC or below fed by a single 125 kVA transformer at 1.2 cal/cm2. IEEE C2 upper bounds everything below 250 VAC at 4 cal/cm2 or less.

But again...this is going down the road of ignoring the likelihood and focusing purely on the hazard side of things. If we look at it from a pure risk point of view, no significant action is required. This is also why for instance even though the IARC has concluded that red meat and processed meats "probably" cause cancer (along with some occupations like airline pilots and painters), we choose not to address it outside of some basic recommendations because the risk simply isn't there (injury is highly unlikely).

It seems like the thread has sort of diverted from the original question, so I'll try to bring it back.

I would hypothetically go with something like option 3. Though if I were to find something overdutied, I would label it "Do not operate while energized". The intention being that you go to the upstream device that is overdutied, de-energize the downstream equipment, then do your switching at the panelboard. There's a low probability of an event by just opening the door, so that's why we'd be more specific with "operate."

This may be a bit overboard for a 208/120V panel. In our case, if we were to find hypothetical overdutied equipment it would probably be old 480V MCCs or bus duct, so we have to go upstream to a 480V breaker.

The problem with being overdutied is that there is a chance that it won't survive during a fault. Force on a bus bar for instance is proportional to the square of the current so for instance fault currents that are 50% higher than the rating results in 225% of rated force on the bus bars. When it comes to breakers either the breaker will be incapable of interrupting the current at all or it may open but be destroyed in the process.

So unless there is a very high likelihood of creating a fault during energization, "do not operate" really makes little sense. As the panels were described, the AIC was 10 kA while the normal current is probably well under 1 kA. Such currents are also unlikely to occur during an arcing fault (arc flash) so the real risk is during a dead short condition. I haven't found any original sources but there are plenty of statistics suggesting that these types of faults account for less than 10% of the faults that occur, and it would not occur during normal operation in the first place except perhaps with utility reclosers and high current test labs where equipment closes into a fault by design.

I have a secondary question:
Now that you know this and you have informed your client (assuming you are working out side your company) what obligation do you have if they decide to do nothing?

I would document it in writing to them and make sure that you have proof that it was received. What they do with it, is up to them but you have done your obligation.

I agree with this.