Relay-Fuse coordination CTI/Grading Time

[Modula2]

IEEE 242-2001, p605, suggest a coordination interval of 0.12 between fuse and upstream static relay. Others, suggest similar values such as 0.10, for example,

http://siemens.siprotec.de/download...ation_of_Overcurrent_Relais_with_Fuses_en.pdf

I assume these are referring to the total clearing curve. Although some manufacturers, for example Siba and others, mostly present the one average curve.

I believed the fuse total clearing curve was already the maximum including all tolerances (at the stated temperature). So, I am wondering what these time intervals are accounting for, and if it is actually needed at all. Could it simply be from a previous era when the starting point was only average curves. My consideration is indoor, so cold weather is not a factor.

Does someone have some insights on this item? Thanks

 

[davidbeach]

I'd certainly only use 0.1s from a total clearing curve, average could be much worse. Instantaneous elements (or time overcurrent for that matter) in numeric relays don't drop out instantaneously. 0.1 s is probably longer than absolutely necessary, but unless you have the ability to test under actual fault conditions, or you are willing to allow the occasional over trip, I'd be rather leery about tightening it up too much.

 

[dpc]

Yes, it is based on total clearing time.

Some allowance is required for:

CT Error
Relay error

 

[Modula2]

Yes, dpc, it must be that. I see others suggest fuse-fuse as "daylight", and the same for LVCB-LVCB. That seems to settle the fuse total curve as including all tolernaces. So, it must be about CT and relay error. I also now see that some experienced people have a lower interval for relay Inst, and definite time regions. If my app is for certain in the Inst/DT region, with no chance of moving into the time overcurrent region, then that eliminates the CT error. Remaining is the relay timing error, which is really difficult to imagine as 100-120 ms (6 - 7.2 cycles on 60 Hz). I see one prominent relay listed as +/-0.25 cycle timer accuracy for 50 elements (inst/DT).

If the above makes sense, I would conclude that Ieee 242-2001 is conservative, and is a broad brush. Further, we can have tighter intervals in the Inst/DT region. I'll probably stay with 100 ms for now (fear is my valuable assistant!). However, perhaps 50 ms is not so unreasonable, also.

 

[dpc]

I normally try to maintain 0.15 sec, but I'm OK with 0.10 sec. I would not want to go less than that. I don't trust the fuse curves that much.