Considering a label someting like this until the breakers are changed out:

DANGER
Overdutied circuit breaker.
Do not operate when energized.

Any thoughts on this problem?

They are not "properly installed", so they may pose a hazard even for people walking in front of them (if a fault happens at the same time).

First let me state: the breakers need to be replaced, per NEC110.9.

But, Arc Flash protection is about risk. The most likely fault on an operating system will be an arcing one, not a bolted fault.
Are the breakers overdutied based on the 100% arcing fault current?

Can your work practices simply limit the closing of the breaker, maybe something like 'On-Off is allowed, resetting is not permitted'

This question was discussed during an IEEE 1584 meeting some time ago. The concern was about liability. If you apply a normal arc flash label to equipment as part of an arc flash study yet acknowledge that the equipment is overdutied, the thought is it may open you up to a liability problem.

I would think any arc flash caculation would be considered invalid if the fault current exceeds the AIC rating of the breaker, so IMHO I would not put the arc flash label on that breaker.

Agreed but that creates a new problem that is still debated. With no label, does that mean the person performing the study "missed" that piece of equipment or does the end user know why there is not label.

alf's label might not be such a bad idea. However, I could see it stirring things up with the owner if this is provided by a consultant. For some reason owner's seem to dislike being told they have inadequate equipment that needs replaced.

If something is not adequately rated, review that again, possibly with outside perspective of another engineer (it is embarrassing, irresponsible and expensive to the customer, to indicate something is going to blow up, only to be pointed out by others that it is part of series rated system or some other plausible explanation). Then the shortcoming should be noted, to the customer, and documented.

Having said all that , I generally don't even check equipment ratings if my work scope does not include such.

An older and wiser engineer (former chair of 70E) told me years ago that the only place an overdutied warning label should go is on the Plant Manger's desk so that he would see if everyday reminding him of the risk to his employees that is in his facility. The ANSI Z535 Safety Sign Standard requires a message panel that indicates how to avoid the hazard and allow reaction time to do so. This indicates that any such label would need to be placed a great distance away on a barricade (rather than on the overdutied equipment which would be too close to allow reaction time with 700 mph shrapnel)--but just how far away is considered a "safe distance" for a barricade of overdutied equipment? Bottom line, at a minimum, barricade it until it's replaced and definitely do not put an arc-flash label on it making it appear the same as all the other equipment permitting someone to remove its covers and do work--the exact opposite of SAFETY!

Similar to what we do, [But our labels warn that there is a danger of being in the vicinity] but we apply similar labels at the location(s) protected by that breaker as well, up to two levels downstream.

If by "overdutied", you mean that the fault current exceeds the interrupting rating, then this really does not have anything to do with opening or closing the breaker manually under normal load conditions. The breaker may blow up while trying to interrupt a fault.

If this is a breaker that needs to be opened and drawn out to establish an electrically safe working condition, then what do you do?

OTOH, if both ends of a "series rated system" are not marked as required by code [110.22 (C)], then it isn't properly installed, and hence really isn't "series rated".
Admittedly those cases where it is a series rating between a separately located upstream device, and a panel board, are much more rare, than ones for a panelboard internal main and its feeders. But in none of those cases have I seen the proper FIELD marking as requried by the NEC, admittedly, I see far more of these as "submittals" than live equipment.

Yes it may be embarasing to have some one else point out that it is supposed to be part of a series rated combination, that is why you do your homework and see if it could be. You shouldn't be saying that it is "going to blow up" in any case, even the most horribly under rated system MIGHT never blow up. If it is part of an engineered system, then again it is required to be field marked [110.22(B)]

In my opinion, BOTH ends of the system [And the middle if it is a multipart system] are {by the code language} required to be marked , not just the "under rated" end.

The arc flash label on the overdutied breaker is based on the upstream protection. Unless the upstream device is also over-dutied, the arc flash label is straight forward. As Vincent B pointed out though, an interaction is not required to initiate an event. The over-dutied breaker should be roped off or otherwise isolated until replacement.

Open it and draw it out with proper PPE. If a fault occurs during this process it will be cleared by a non-over-dutied upstream device.

Remote racking

A CB that is overdutied is dangerous. Do not assume that because the theoretical arcing current is below the capability of the device the device is still OK for arc flash purposes. 1) you always want margin between a CB's capability and its operational requirements, 2) the arcinf fault hsa not read IEEE 1584 and hence may be of any magnitude smaller or larger than that predicted by the IEEE guide for a host of differnt reasons. 3) arcing faults often start with some sort of contact so the first current may be closer to bolted at least for a few milliseconds. 4) if the CB is overdutired chances are it is also old which also calls the capability to question... In short it is unsafe and circuits below it are unsafe it if fails to open as expected. Act accordingly.

PS: I have seen tests on actual equipment and arcing currents can be both larger and smaller than predicted by IEEE 1584. If the test protocol mimics the original IEEE 1584 protocol the arcing current is properly predicted, but if gaps or enclosures or starting medium are different the arc can be quite different.

Jim, I believe there are more concerns than just liability. Arcing currents can be quite different than those predicted by the current IEEE 1584 guide. Lower or higher. A CB that is underrated is probably also old. A fault may start with more conductor than the test wires and hence a first half cycle arc could be near bolted in magnitude.... In other words the CB is just plain dangerous. It should be treated like an in appropriate interruption device for arcing or bolted faults. I believe it is inaprpriate to attach an AF label to the CB or to downstream equipment till that device is replaced.

I believe that the arc flash label for the overdutied device and everything downstream of it should reflect incident energy assuming failure of the breaker to clear the fault.

If you are over-dutying the breaker, how can you be sure it will clear at all and not weld?

Realistically you would have to go to the next upstream device and hope the breaker doesn't just disassemble itself in a fault.

Larry, I believe that is what jghrist was saying since since the applied label should reflect only the characteristics of the upstream device(s) (assuming no gen or large motors downstream).

Marcelo and several others raise some good points. The arcing current calculation from IEEE 1584 is an estimate at best under specific lab conditions. A wrench has a "bit less impedance" than a #12 trigger wire and hence, initially could result in more current initially.

I believe alf is thinking correctly about the label. The problem I have had if something is not labeled is why? Was it because of a problem or because someone forgot.

I have seen people use the following method from time to time. It always gives me heartburn and in my opinion a pretty bad idea.

(bad) Example:
Breaker: 22,000 Amp interrupting rating
Bolted fault current: 30,000 Amps
Arcing Current (calculated): 16,000 Amps

Conclusion, during an arc flash the breaker will experience a current level below it's interrupting rating. This has a pretty large assumption that the arcing current will really be that low - which it may not be. Also, the breaker is still technically inadequate which is a violation of NEC 110.9