I have a fundamental issue with how Arc Flash calculations are done. After having done quite a few studies and putting alot of detail into my work, I realize that most of the effort I have spent is wasted because I do not have faith in the results or premise of the calculations. There are quite a few reasons but I want to highlight the main 3 issues and then summarize my thoughts on how I think they impact the overall results:

1. Current Limiting fuses- IEEE1584 provides calculations for modeling current limiting fuses as a load. These are pretty simple calculations and while some may find fault with the calculations, I believe they may be accurate for what they are intending to show. What I do have a problem with is that the impact of current limitation is not reflected on the remainder of the downstream system. During a fault the impedance of the current limiting fuse rises, limiting the fault current to downstream devices. In an arc flash model this does not exist.

2. Utility Short Circuit current-
Often utilities will not provide a precise short circuit current and often do not provide any impedance value. Generally, the fault current is given higher than expected so that contractors can properly rate their equipment duty in a conventional short circuit study. This short circuit current is minimally useful for these arc flash studies, since the higher clearing times often

3. Using momentary fault current for a high clearing time fault-
Using a momentary fault that includes high asymmetrical fault contribution from reactive sources and assuming that that fault current lasts the duration of the fault is a bad assumption. In addition to the asymmetrical decay, the symmetrical decay of longer faults approaching the 2-3 second max threshold for low voltage clearing times is also not taken into account.

A combination of an overestimate of the short circuit current and a reduction in downstream fault current for services protected by a current limiting fuse might cause your model's fault current to be up to 50% less in the real world than what is modeled. When the fault current falls under the short time pickup, BAM you have one huge incident energy because momentary fault current is assumed to last for a full 2 seconds (assuming a 208v arc even lasts that long).

I believe in order for realistic results to exist, that the model should generate a 2-second waveform that includes any current limiting effects of a CL fuse for a range of fault current values based on a user defined standard deviation. This waveform should then be run through a TCC function that determines the clearing time and then can determine the amount of energy dissipated in the arc Empiricially. Generally speaking, the results of performing these calculations are way low with moderate fault current, higher than reality with high and low fault current.

Either that or you hand a contractor some simple 70E tables and I'm sure he might get closer than the engineer once in awhile and save the country alot of time, money, and wasted effort because arc flash calculations are not saving lives alone- PPE availability, NFPA70 changes like arc flash reduction mode, labeling 1200A and higher services with fault current, arc flash rated switchgear, lawyers pushing this stuff, and having management pay attention to personal safety and not requesting live work is what is making people safer.

Those are for ONE vendor. Granted it's a major one (Mersen). And it doesn't address downstream effects as stated. EPRI also has some similar simplified models for handling for instance relatively long arcs. No software implementation I'm aware of uses these equations but then again they are so simple, you probably don't need software.



It is getting better since as of 2015 utilities can't ignore arc flash anymore either. That being said, they have a point...they have systems that are not fixed in any way and configuration can change quite flexibly at any time. So the values you are given are pretty much an educated guess. I'm not sure why this disappoints you. Imagine if you are effectively a private utility complete with a large cogen, transmission lines, miles of distribution lines...like a large mine. And customers that CONSTANTLY move their equipment around. It all becomes an educated guess.

[/quote]3. Using momentary fault current for a high clearing time fault-
Using a momentary fault that includes high asymmetrical fault contribution from reactive sources and assuming that that fault current lasts the duration of the fault is a bad assumption. In addition to the asymmetrical decay, the symmetrical decay of longer faults approaching the 2-3 second max threshold for low voltage clearing times is also not taken into account.[/quote]

Better research your software closer. SKM does it the way you expect. It's piecewise linear, not quite continuous but close enough.



This is like adding up all your motors into one fault contribution load. When you study this stuff by hand and learn what all the assumptions and estimation tricks are, you get more comfortable with it and all these assumptions and shortcuts don't seem so bad when you see the actual effects. If you want, you can use EMTP to do waveform level estimating and see for yourself. You can get a free license to do some modelling.

[/quote]

This is called a boundary method. Several have been proposed. There is even two standard ones (one in NESC, one in 70E) as well as a commercial one (Arc Flash Tables). It's not as unrealistic as it seems either.. Imagine a service engineer like me that goes to lots of plants and works on power equipment all the time where often the plants can't even afford an electrician, never mind an arc flash study. I have a choice between the tables and just winging it.

Hey, thanks for the response. I think you're right- I need to understand what is going on under the hood a little bit better (easypower) and get a look at their 'integrated' method which I don't use but it's basically what you describe, although the results seem inconsistient within the program and they give the user very limited information to look at. Maybe what i'm also sour about is that easypower's scenario manager is pretty poor and what I want to do is to model a range of utility short circuit values to determine the worst case- rather than just a max and a min.

I completely understand why it is so difficult for utilities to have a calculated fault current at every bus throughout their system. I think there's a better way such as plugging into their database to get a real time value if you are a big client or being given a range of values..

I wasn't aware that Mersen was the only CL fuse manufacturer- or is there something proprietary that they do that i'm missing?

The formulas in IEEE 1584 for fuses are based on incident energy tests that were performed using fuses from Mersen. There are plenty of other manufacturers. The fuses are "standard" but when it comes to fuses like an RK5 fuse package the standard is very flexible so there is no way to say that a Mersen RK5 works identical to another manufacturer's RK5. So this is the reason that those formulas aren't used.

Furthermore there is nothing in IEEE 1584 to itself to suggest that a piecewise approach to calculating incident energy is even valid. There is really nothing within the standard suggesting that this is even possible. The calculation method as given takes a single current which is determined through some other method and calculates an incident energy. Piecewise approaches that are done by the software vendors are technically outside of IEEE 1584. The next edition will certainly bless this approach but right now technically it's not a valid result. When you realize that we are calculating "arc power" and then integrating it over time (multiplying by the number of seconds) then obviously a piecewise approach is perfectly acceptable but right now IEEE 1584-2002 does not explicitly state this.

I am glad I joined this forum, thanks for the explanations. Any idea when the next edition will be released?

Good question. The first round of balloting was just completed a few weeks ago. So far, so good. Now lots of comments need addressed over the next several months, then a new draft, then a new round of balloting. There will probably be a few rounds of balloting so I'm guessing maybe 1 to 2 years barring anything unexpected - which has happened on a few occasions.

And the usual disclaimer... this is a personal comment from me and is not anything representing IEEE 1584 as Vice Chair. <sounds like a broken record but I'm required to make the statement.