Main and sub distribution boards - Circuit Breakers

Andy beat me to it really - if you really must have reliable discrimination between breakers for all fault levels, and not just discriminate for modest overloads, you have to have programmable delays for all but the ones nearest the load.  

Fault levels above the threshold for the magnetic ‘instant trip’ part of the larger circuit breaker do not give any sort of reliable discrimination. 

Actually usually this is OK in practice as zero ohm faults are quite rare, and with care about what shares the sub-main, the worst of the side effects can be designed out. And at the back, for really critical things we can decide to fit “death or glory”  fuses that only operate when the end of the world is nigh, as it were, rather than magnetic breakers. Chosen well this also relaxes the blast containment aspects and may allow for lighter switchgear, as the equivalent I2t after the fuse can be a lot less than it is in front of it.

If you must have delay, then you need MCCBs and the curves in this paper may help clarify how things are done much  better than I can in words.

Using a delay into any fault level with no upper bound “instant trip” current is scary and something to think very hard about, as even a fraction of a second feeding kA into a dead short is a lot of joules per ohm or I2t, and a lot of potential damage to cables / arcflash etc. Imagine waiting a third of a second for things to strat to think about  cut-off, it could be like a bomb going  off. Sometimes the delays at the top of a cascade can be as much as a few seconds, but it is not nice if it is ever called upon to perform its duty.

I'm not really joking  -  at the indoor substation level you may find that one wall  is deliberately weak, rather like the ‘tear here’ lines on paper forms, to give a controlled failure mode if the worst happens, and not to damage more important structure  Saving the equipment is not a consideration then.

If you can, try to design the requirement out, or at least only guarantee to fully discriminate for faults up to a certain maximum current, then say above that all bets are off.

Mike.

Andy beat me to it really - if you really must have reliable discrimination between breakers for all fault levels, and not just discriminate for modest overloads, you have to have programmable delays for all but the ones nearest the load.  

Fault levels above the threshold for the magnetic ‘instant trip’ part of the larger circuit breaker do not give any sort of reliable discrimination. 

Actually usually this is OK in practice as zero ohm faults are quite rare, and with care about what shares the sub-main, the worst of the side effects can be designed out. And at the back, for really critical things we can decide to fit “death or glory”  fuses that only operate when the end of the world is nigh, as it were, rather than magnetic breakers. Chosen well this also relaxes the blast containment aspects and may allow for lighter switchgear, as the equivalent I2t after the fuse can be a lot less than it is in front of it.

If you must have delay, then you need MCCBs and the curves in this paper may help clarify how things are done much  better than I can in words.

Using a delay into any fault level with no upper bound “instant trip” current is scary and something to think very hard about, as even a fraction of a second feeding kA into a dead short is a lot of joules per ohm or I2t, and a lot of potential damage to cables / arcflash etc. Imagine waiting a third of a second for things to strat to think about  cut-off, it could be like a bomb going  off. Sometimes the delays at the top of a cascade can be as much as a few seconds, but it is not nice if it is ever called upon to perform its duty.

I'm not really joking  -  at the indoor substation level you may find that one wall  is deliberately weak, rather like the ‘tear here’ lines on paper forms, to give a controlled failure mode if the worst happens, and not to damage more important structure  Saving the equipment is not a consideration then.

If you can, try to design the requirement out, or at least only guarantee to fully discriminate for faults up to a certain maximum current, then say above that all bets are off.

Mike.