If have touched on this subject before but I want to talk about it again, the dynamic characteristic of an arc flash. I have made 2 calculations on the same system to use as an example.

For the first calculation I used the IEEE1584 formulas and got an arc flash with an energy of 195 cal/cm2. When I simulate the system in SKM I get the same answer.

If both are the same then it must be correct right? Well no I do not think so. This situation has 6 generators it has an initial bolted fault current of around 71kA. However just before tripping (0,632 seconds after the start off the arc flash) the current is ‘only’ 21kA. So how can an answer based solely on the initial bolted fault value be true if it is in fact so dynamic.

I have used Excel to calculate the short circuit current with intervals of 0,001 seconds for all generators. I added these together and then used the formula from the IEEE1584 to calculate the arc flash energy with intervals of 0,001 seconds. After adding all those I get an arc flash with about 67 cal/cm2. The question is how correct is this value, this 67 cal/cm2. I did use the correct formula’s, the only thing I changed was instead of using the initial bolted fault current and the total duration of the arc, I split it in a lot off small sections. The result is a drop of 65% in energy released from the arc. Well not actually the arc but the calculated energy.

I do this because I do not believe in the static form of calculating an arc flash in such a dynamic system.

I would like to see the opinion of you guys and maybe get a discussion going.

The dynamic simulation that you modeled is a more correct approach. Taking a fixed short circuit current and treating it as it lasts a long time is not realistic when there are motors, generators or other sources of short circuit current that decay over time. In addition, there is the situation of devices tripping sources off line at different times so that is another situation where a more dynamic simulation is necessary.

SKM does model some dynamics but the situation you described pushes the limits of its capabilities.

The next edition of 1584 is supposed to mention the piecewise approach.

Thanks for the replies Jim and Paul!
Paul I have a question for you: you say that the piecewise approach is supposed to get a mention in the next edition, but where do you read things like this?
Because if I tell people about this at my company I do want to have something to fall back on so to say.

I mean that I can say ‘this and that will probably come in the next edition because it is on the IEEE1584 website’. Not to be insulting but I can’t say ‘this and that will probably come in the next edition because someone on the internet told me’. I know you guys are not random people but the person I tell this to will think that either way.

Hope I did not insult you guys, because that is not my intension.

Jim and you are correct. Assuming the full short circuit current for the full duration of the event is not realistic. As Jim states some level of decremtation of fault current over time would be more accurate. However, I am not aware current models do this properly at all. As far as I know short circuit calculation models assume the fault is bolted and the motors or generators are providing power to a bolted fault. An arcing fault is a high impedance fault, so the generator contribution will be different. I would think it is less to start out with, and decrements slower. I have not seen a model of what the expected behavior really is, however it would need to take into account arcing voltage and the associated decrement curve. Seems complex, and to some degree speculative.

I am not aware that 1584 will directly address this in any meaningful way.

Nothing published. With both IEEE and NFPA the only way to get "official" information is when they publish draft editions of an upcoming standard for public comment. If you communicate with the various committee/subcommittee members though they will frequently reveal privately some direction as to where the conversation is going.

IEEE 1584 currently does not really address the input (current) at all. What we have is that several tests were performed and incident energy was measured. The data was then curve fitted to a formula, resulting in an empirical equation. The tests were performed in the lab with a controlled (or at least measured) level of fault current. It has nothing at all to do with the power system analysis side of things. It is simply based on empirically derived equations that match roughly a 300 test database.

When applying this to an actual power system, you need to determine an available fault current. In the lab, the current source is a constant and there is only one source. In an actual power system it is not constant and there are multiple power sources. Thus one should do piece-wise fitting of the current to the arc flash calculation to arrive at an accurate estimate. As I said...IEEE 1584 does not mention this but it is implicit in the analysis.

Both SKM and ETAP (the two I'm familiar with) do SOME modelling of motors and generators. There are specific settings for this in both systems. However it is not all that sophisticated in implementation.

You may consider doing a dynamic arc flash calculation using EMTP simulation.

Regards