Here's an interesting phenomenon that results in huge IEs. See the TCC on the attached PDF.

The red curve is for a SEL 351 on the 115kV side of transformer T46. The predicted arcing fault current on T46's 7.2kV secondary is 21.7kA. Note that the 21.7kA line is to the left of the end of the SEL curve data and to the right of the longtime pickup line (short red line at top near R-122 P).

According SKM, the arcing fault 'threaded the needle' of available data for the relay. The program can't find the intersection of the AF amps and the relay TOC curve. The IE result is limited by the maximum arcing time.

We actually had a 'squirrel incident' at that location, which resulted in an arcing current of 39kA and did eventually trip the relay in about 3 seconds.

Is the secondary voltage 7.2 kV Ø-Ø or 7.2 Ø-grd? I assume 7.2 Ø-Ø from the current ratio, but that's an unusual voltage. The transformer damage curve would indicate a base rating of 80 MVA. A fault current of 21.71 kA would indicate a transformer impedance of nearly 30%. Even considering some 115 kV system impedance and arcing current being lower than bolted current, it seems that the transformer impedance is at least 25%. Something is wrong - that's much too high.

If the fault current is correct, you need some secondary protection to prevent transformer damage during a through fault.

It is an unusual situation. The relay and CB are located 1600 feet (750KCM, Pb Sheathed, 3 -1/C + G) from the transformer (85,750kVA, Z=15%, three winding, 115-7.2/7.2). There is differential protection.

Can you set the pickup of the SEL-351 lower? It is too slow to adequately protect the transformer for the fault shown. As you noted, it also results in high arc hazard IE.

If for some reason you can't set the pickup lower, then you need to have an overcurrent relay on the low side to provide faster protection for low side faults outside the differential zone.

There is no needle eye, since the VI curve does not simply end as shown. If you use the equation rather than the curve, you can plot time values much closer to 1 X PU. 1 X PU itself is a bit difficult, since t goes to infinity.

stevenal,
What you say is true. However, I posted this in [b]Software for Arc Flash Studies Where we can share experiences about various Arc Flash Programs[/b] because SKM's data ends as shown on the TCC. I thought others should be aware of this unusual effect.

I'm curious. Do EasyPower, ETAP and other AF software have similar limitations?