Diesel Generator Fault Levels

If you are looking at setting the Vigis to 0.4 second delay on over current,  you have to ask why - there is no need for fast disconnections  on  L-N and LL faults, you may as well let the lower tier of breakers pick up if it can. this can be set to up to 5 seconds if you fancy, though 1 is probably enough, but set the In multiplier low, less than 2000A

LE faults will be picked up by the I delta N function, which can be set to quite a high value - a few amps, to allow any lower tier faults to be picked up by local RCDs where fitted, but obviously a shorter delay, perhaps 250-300msec.

I have no idea what number Amtech is expecting, though I fear the authors were not really aware of the full subtlety of genset supplies.


By all means use the software, but be aware that you may need to do the sums twice, once with a high value for all the fast breakers, and again with a much lower value for the slower ones, and being a bit grey of hair, I'd recommend to do some sanity check estimates the old way to make sure it is not miles out for some reason.



In general an LLL fault is about twice the LN fault current, and the effect of an asymmetric transient  (phase varying when in the cycle the fault comes on )is also about 2:1 at the start, but settles after the first couple of cycles.

perspicacious:

If I've read the graph correctly, the set will provide just 2000 A at 0.1 sec. Will the preceding higher current be enough to operate the primary protective device and/or how far downstream? 


Dare I mention RCDs?!


Regards


Bad BOD and Bed BOD till tomorrow

Which would tend to agree with a rule of thumb that suggests (for a turbo diesel) that you'll get at best about 2.5 x FLC at 100milliseconds - which won't open most protective devices at Tier 1 or Tier 2 unless they are equipped with earth fault shunt trips (or are time stopped 100ms units with much lower pick ups (say 5 or 10 A)


I suspect the problem here is Amtech can't model the fault current curves anything like as well as the manufacturers data  - it probably needs ERACS or similar to get some certainty


Regards


OMS