7604 wrote:

According to the theoretical model I need to obtain a Arc fault @ 50% of bolted fault which is 8kA not 4kA resulting from the AC model. Somehow, base on I^2t curve the clearing of both Bussman and SIBA above will clear the fault of 8kA almost in no time or more specific less than 2 ms. I wonder if my reading is wrong?

Nope. Semiconductor fuses are amazing. The only exasperating part is that they are so fast that usually multiple fuses trip simultaneously.

**Quote:**

Also, I wonder if DC Arcflash in this particular case can occur @ 115% of the let-through current which takes almost a second to clear?

This is where it gets a lot more tricky. Theoretically, anything is possible. However, one thing that is definitely true and known with DC systems is that we have to be concerned with trading off arc current and arc voltage, knowing that arc resistance is not a constant. It happens that maximum power transfer to the arc occurs at approximately 50% of the maximum available short circuit current...which explains what the 50% factor is all about, and that's where the arc will stay at.

There is a huge difference between the world of DC and AC, and you definitely can't just use AC models as a proxy for DC arcs. The big difference is that with an AC arc, it extinguishes at every zero crossing of current and then reignites as the voltage gets above around 150 volts or so. With DC arcs, the arc is very stable over a wide range of voltages and currents. The arc resistance quickly adjusts and stabilizes at the point of maximum arc power which happens to be at 50% of terminal voltage and thus balances 50% power transfer between the arc and the system impedance. With AC, what was found was that after a large number of tests (over 300), is that the incident energy and AC arcing curents tended to follow a long tail type distribution with a "double hump". Adding a second calculation at about 85% of bolted fault current and taking the worst case from both calculations resulted in a significantly better correlation between actual test results and incident energy. It is my understanding that the next edition of IEEE 1584 will go even further in achieving good correlation between test results and calculations.

If you get a copy of the actual papers that Annex D refers to, you will get a lot of this detail and more. DC arcs have been around for much longer than AC arcs. Some of the information I have stretches back to around the 1920's and 1930's when there was a lot of testing done to determine arcing parameters for neon lighting design. Its now very old data but still very relevant today. One of the most important equations you can get from that old data is that for DC systems, the minimum current required to initiate an arc at a given voltage is very well quantified. Below 28 V for instance, arcing is just plain impossible. It is possible but unlikely up to around 54 volts. Above around 70-80 volts though it's almost impossible not to form an arc even with extremely low currents.