Roger Mills wrote:

Yes, the circuit is going at the extreme right - which currently has a blue
base for a 16A MCB.

I may well still have a 30A fuse - not sure about the BS1361 bit though - is
that something special? [The CU started off with fuses, and was converted to
MCBs!]

The BS 1361 fuse is a cartridge fuse, like this
http://www.tlc-direct.co.uk/Products/WYC30.html
as opposed to the rewireable one that you probably still have (which is
to BS 3036).

What is the rationale for using a fuse rather than an MCB? Is it that it
takes longer to blow, making it more likely that the one at the garage end
will trip instead?

In effect, yes. The problem with cascaded MCBs is that if the fault
current exceeds a certain figure both MCBs are on the 'instant' magnetic
tripping part of their characteristics, and they'll both trip together -
i.e. there's no discrimination.

If so, it makes sense, but it flies in the face of the advice given
in the earlier thread which said that I needed a Type B (and *not*
Type C) MCB on account of only having a 2.5mm^2 earth wire for most
of the length.

OK - I've probably confused you there. We're looking at different
aspects of the design in isolation - discrimination only in this
sub-thread - and have lost sight of the whole picture. Let's look at
some other aspects. I've found the June thread and assume that 10 m of
T&E plus 5 m of SWA still stands. I'll also assume that the 30 A BS
1361 cartridge fuse is to be used.

- Voltage drop: 15 m (total) of 6 mm^2 will have a voltage drop of under
3.3 V at 30 A; that's 1.4% of Uo, so OK there.

- Disconnection time: 10 m of 6/2.5 T&E will contribute about 0.13 ohm
to the earth fault loop impedance (Zs) and 5 m of 6 mm^2 2-core SWA will
contribute a further 0.06 ohm; taking the cautious assumption that the
supply is in fact TN-S we add 0.8 ohm for the external (supplier's side)
part of Zs, so Zs (total) is 0.8 + 0.13 + 0.06, which is 1 ohm, as near
as damn. The earth fault current it thus 230 A which will take out the
30 A fuse in somewhere between 0.2 and 0.3 seconds - so the 5 s
disconnection time requirement is met.

- CPC sizing: assuming 0.3 s disconnection then for the T&E section
S(min) = sqrt(230^2 * 0.3) / k, with k = 115 (from Table 54C). This
gives a min CPC size of just over 1 mm^2, so the available 2.5 mm^2 CPC
is OK. It's hardly necessary to do this calculation for the SWA
section, but for completeness the relevant k value is 51 (Table 54D,
assuming 70 deg.) so S(min) is just under 2.5 mm^2 (steel). The armour
CSA in 2-core 6 mm^2 cable to BS 5467 is 22 mm^2, so definitely no
problems there.

Unless I've missed anything, we can conclude that the submain design
complies with BS 7671.

--
Andy