Shower circuit design.

For single-phase supplies up to 100 A, you have 6 kA OCPDs that have been tested for 16 kA conditional short circuit rating by the manufacturer.

Quite so. 

Although reading the fine print, this is not really  that the 6kA devices are tested bare on a supply of 16kA, if they were they'd be rated at  16kA and charged for accordingly,  but rather that they are OK downstream of a 100A fuse, on such a supply. This means that above 6kA, the DNO 100A fuse is doing some, or all, of the heavy lifting, and opening, first (after pre-arc) to make a rapidly growing series arc to act as a current limiter, and then soon after going completely  open circuit. The MCB is rated to survive that I squared t , and by the way it will probably trip as well…

(also 100 amp DNO fuses vary quite a bit- the 

I2t for a 100KR85 is 73500 but  the 

100LR85 is only 57300 to fully open - the latter, friskier, fuse is preferred by SSE and maybe other DNOs too. )

The I2t for the MCB is also  a variable thing, as illustrated in these plots from a Dorman Smith paper  The let through energies of  larger value MCB's  and that to blow the main fuse cross over in the high kA regions of interest. Not that we need to care too much, but a really low z fault on the shower or cooker cct may well take the DNO fuse with it from a cold start on a  PSSC >6kA  supply.?

Wylex MCB curves show similar trends but are not quite the same… (note they use the pre-arc not the clearance I2T for their BS88 dotted lines, so these are minima, not maxima.) Note also the theoretical constant opening time for 10msec (1 half cycle) and how they are all get going quite a bit faster than that, at least to start opening, if not to completely break the circuit. (who knows how many hundred amps dribble through to give a relatively long low energy tail while the arc is dying out on the quencher)

Now the MCB may be cheesecloth tested only in it's makers enclosure, but I think an MCB that blew itself  to bits would soon lose its reputation so it is a fair bet it would be fine in almost any similar enclosure, or maybe even none at all, at least up to the 6kA figure. 

Given this is a loop plus fault impedance of 40milliohms ( which is not much more than there and back on 1m of 1mmsq, or 2.5m of 2.5mmsq or 10m or 10mm2 etc..) the proportion of 100A single phase installations where PSSC at origin exceeds 6kA is small, and then the proportion of real faults that exceed 6kA will be smaller still. 

We could handle those separately, and say no mixing of MCBs on installations of PSSC > 6000A .

Mike.

For single-phase supplies up to 100 A, you have 6 kA OCPDs that have been tested for 16 kA conditional short circuit rating by the manufacturer.

Quite so. 

Although reading the fine print, this is not really  that the 6kA devices are tested bare on a supply of 16kA, if they were they'd be rated at  16kA and charged for accordingly,  but rather that they are OK downstream of a 100A fuse, on such a supply. This means that above 6kA, the DNO 100A fuse is doing some, or all, of the heavy lifting, and opening, first (after pre-arc) to make a rapidly growing series arc to act as a current limiter, and then soon after going completely  open circuit. The MCB is rated to survive that I squared t , and by the way it will probably trip as well…

(also 100 amp DNO fuses vary quite a bit- the 

I2t for a 100KR85 is 73500 but  the 

100LR85 is only 57300 to fully open - the latter, friskier, fuse is preferred by SSE and maybe other DNOs too. )

The I2t for the MCB is also  a variable thing, as illustrated in these plots from a Dorman Smith paper  The let through energies of  larger value MCB's  and that to blow the main fuse cross over in the high kA regions of interest. Not that we need to care too much, but a really low z fault on the shower or cooker cct may well take the DNO fuse with it from a cold start on a  PSSC >6kA  supply.?

Wylex MCB curves show similar trends but are not quite the same… (note they use the pre-arc not the clearance I2T for their BS88 dotted lines, so these are minima, not maxima.) Note also the theoretical constant opening time for 10msec (1 half cycle) and how they are all get going quite a bit faster than that, at least to start opening, if not to completely break the circuit. (who knows how many hundred amps dribble through to give a relatively long low energy tail while the arc is dying out on the quencher)

Now the MCB may be cheesecloth tested only in it's makers enclosure, but I think an MCB that blew itself  to bits would soon lose its reputation so it is a fair bet it would be fine in almost any similar enclosure, or maybe even none at all, at least up to the 6kA figure. 

Given this is a loop plus fault impedance of 40milliohms ( which is not much more than there and back on 1m of 1mmsq, or 2.5m of 2.5mmsq or 10m or 10mm2 etc..) the proportion of 100A single phase installations where PSSC at origin exceeds 6kA is small, and then the proportion of real faults that exceed 6kA will be smaller still. 

We could handle those separately, and say no mixing of MCBs on installations of PSSC > 6000A .

Mike.