How to determine Arc Flash level?

Tis complex. Very complex.

But there are very simple ways to put an upper limit on how bad it could get when it is all stacked against you..

First you can metricate calories to joules with a factor of 4.2 joules per calorie to get a rating for the overalls in real units, as opposed to Diet numbers.

Around 5 joules per cm2 is the onset of something more permanent  than a bad sun-tan for bare skin, though to set a safe limit that is quite a bit less is better and leaves you with more hair on your arms.  (some US literature quotes 1.2cal/cm2 which is about the same, as the onset of permanent scarring.)

edit to set an expectation, in the test facility where I work (not a mains arc test facility, but we can certainly make some noise..)  a figure of 1 joule is used as the ‘action limit’  i.e. below that no special precautions need be taken, above that we start to think properly about it.

You can estimate the energy of the arc (volts times pssc amps times disconnection time) to get total joules of energy that could ever be in the arc between the time of onset and ADS doing it's thing. Note that actually you can safely quarter that, as when drawing the PSSC, the voltage at the fault has fallen to zero, in reality the power for making an arc - the I*V product -is maximum near to half voltage and half PSSC flowing.

Then you need a geometric factor - in free space the energy of the arc will spread out in all directions, and you can imagine those joules smeared out over the inside area of a spherical shell (like flattening an onion skin and measuring the area) This gets you so many joules per square cm or square m  at any given range like hand length or arm length..

Arcs in confined spaces are worse, as the energy that was going towards  the back and sides of the box may be redirected forwards, and  while the energy total is the same, we have robbed Peter to pay Paul and it is now safe to get closer  behind the box (duh well yes) but more dangerous to a greater distance in front of it.

A factor of 2 or 3 in the worst case direction, relative to spherical,  is typical.

By you have worked out a few you will find that a typical lighting circuit (PSSC of a few hundred amps disconnection less than 100msec ) needs no skin protection at all, unless you are within about 6 inches, while a bare bus bar off a 100A fuse deserves arms length treatment, or the fancy overalls.

There are far more accurate methods that allow for not all the arc energy being turned into flash, and the extra voltage loss in establishing the arc, but they all reduce the exposure compared to the outline method above which gives a very safe first boundary estimate.

Mike.

PS its not my method, it is a chap called lee.

Late EDIT and here is a very spotty scan of Lee's ground breaking paper from the 1980s. note that by assuming Zs was mostly inductive rather than resistance, and that the arc is resistive in nature, his peak of I_fauIt and associated V_fault is more like 70% of PSSC. This is probably true for faults in or very near large transformers. However for faults at the end of any reasonable length of submain, the ‘let's just pretend its two resistors in series' assumption is more applicable.

A more up to date view with a less cautious set of assumptions from a 230V country (ZA) is here

Tis complex. Very complex.

But there are very simple ways to put an upper limit on how bad it could get when it is all stacked against you..

First you can metricate calories to joules with a factor of 4.2 joules per calorie to get a rating for the overalls in real units, as opposed to Diet numbers.

Around 5 joules per cm2 is the onset of something more permanent  than a bad sun-tan for bare skin, though to set a safe limit that is quite a bit less is better and leaves you with more hair on your arms.  (some US literature quotes 1.2cal/cm2 which is about the same, as the onset of permanent scarring.)

edit to set an expectation, in the test facility where I work (not a mains arc test facility, but we can certainly make some noise..)  a figure of 1 joule is used as the ‘action limit’  i.e. below that no special precautions need be taken, above that we start to think properly about it.

You can estimate the energy of the arc (volts times pssc amps times disconnection time) to get total joules of energy that could ever be in the arc between the time of onset and ADS doing it's thing. Note that actually you can safely quarter that, as when drawing the PSSC, the voltage at the fault has fallen to zero, in reality the power for making an arc - the I*V product -is maximum near to half voltage and half PSSC flowing.

Then you need a geometric factor - in free space the energy of the arc will spread out in all directions, and you can imagine those joules smeared out over the inside area of a spherical shell (like flattening an onion skin and measuring the area) This gets you so many joules per square cm or square m  at any given range like hand length or arm length..

Arcs in confined spaces are worse, as the energy that was going towards  the back and sides of the box may be redirected forwards, and  while the energy total is the same, we have robbed Peter to pay Paul and it is now safe to get closer  behind the box (duh well yes) but more dangerous to a greater distance in front of it.

A factor of 2 or 3 in the worst case direction, relative to spherical,  is typical.

By you have worked out a few you will find that a typical lighting circuit (PSSC of a few hundred amps disconnection less than 100msec ) needs no skin protection at all, unless you are within about 6 inches, while a bare bus bar off a 100A fuse deserves arms length treatment, or the fancy overalls.

There are far more accurate methods that allow for not all the arc energy being turned into flash, and the extra voltage loss in establishing the arc, but they all reduce the exposure compared to the outline method above which gives a very safe first boundary estimate.

Mike.

PS its not my method, it is a chap called lee.

Late EDIT and here is a very spotty scan of Lee's ground breaking paper from the 1980s. note that by assuming Zs was mostly inductive rather than resistance, and that the arc is resistive in nature, his peak of I_fauIt and associated V_fault is more like 70% of PSSC. This is probably true for faults in or very near large transformers. However for faults at the end of any reasonable length of submain, the ‘let's just pretend its two resistors in series' assumption is more applicable.

A more up to date view with a less cautious set of assumptions from a 230V country (ZA) is here