Stefek Zaba wrote:

[...] and then turn on the isolator. That gets you all elements
drawing their cold inrush currents at once - but they very soon get
warm, settle down to "merely" their rated draw, [...]

Nichrome has a very low temperature coefficient of resistivity - there
is no significant "cold inrush current" - unless you're talking about
tungsten-halogen heat sources...

[...] barely 1.6 times the 30A MCB's nominal current. Yer typical
Type B/type " domestickle MCB will take 15-30 minutes to trip on so
"small" an overload [...]

.... Which is not long enough to count as a long-term overload in the
context of regulation 433-01-01, especially considering that this
overload scenario would not be repeated on a regular basis.

[Fault current]
Kitchens are typically not a huge distance from the CU; with say 10m of
6mmsq and a couple of "wimpy" 2.5mmsq flex at the end, total loop
resistance down the 'worst case' path of L-to-E (so as to put the fault
current down the thinner protective conductor of the 6mmsq) will be in
the range of 0.1 ohms.

0.15 ohm is nearer the mark for those lengths - Table 9A in the OSG
gives a handy reference for 'R1 + R2' (phase + CPC) resistances.

Let's quadruple that to allow for increased
contact resistance in a couple of connections along the way:

Eeeek - if you had 0.3 ohm contact resistance in a couple of connections
you'd have over 300 W being dissipated in inappropriate places (at 32
A). The short-circuit then would be the one that occurred as a result
of the fire...

that's still a rather low 0.4ohms resisting the 240V, giving us a
rather-MCB-persuasive current of 600A.

You've forgotten the source impedance at the point of supply. The worst
case condition here is 0.8 ohm external earth fault loop impedance (Ze)
for TN-S earthing, so the total earth fault loop impedance (Zs) becomes
0.95 ohm, making the fault current only about 240 A.

That's 15 times the nominal rating of the 40A MCB, which takes you
into the fast solenoid-based 'gross overload' part of the MCB's
operating regime (rather than the slow thermal-based 'smaller
overload' part), so you'll get disconnection within 0.1s or so - not
nearly long enough for your 'wimpy' cable to heat up towards the 140
or so degrees which is the 'don't go there' temp for PVC. From which
semi-quantitative argument, we conclude that your sparky's uprating
of the MCB from 30 to 40A - while not best practice - doesn't cause a
keep-you-awake-at-night level of risk.

It doesn't take you into risk at all if you do the sums, provided that
the MCB is Type B. For Type B 'instant tripping' (i.e. 0.1 s) will
take place at = 5 * In, which is 200 A for the 40 A device. Since our
fault current is more than that, even with the worst case Ze value, we
know that the fault will clear within 0.1 s and thus that the
disconnection time requirement of BS 7671 is met.

It now just remains to check that the 1.5mm^2 CPC in the bit of 2.5 T&E
won't fry. The relevant limiting temperature for a PVC cable, by the
way, is 160 deg. 140 deg. applies for cables over 300 mm^2, which would
normally be thought of as beyond DIY territory :-). The simplest way to
tackle this is to use the adiabatic equation backwards to determine for
how long a 1.5 mm^2 conductor will stand 240 A. The equation is given
in BS 7671 as S = sqrt(I^2 * t)/k. Turning it round to find t gives t =
(k * S / I )^2. (k is the constant from Table 54C, S is the CSA of the
conductor and I is the fault current.) Hence t = (115 * 1.5 / 240)
which works out at a tad over half a second. With the B 40 A MCB we've
already determined that the fault will clear in 0.1 s, so all is well.

If you work through the same process using the same cable lengths, but
for different devices you'll find as follows:

- 30 A BS 1361 (cartridge) or BS 3036 (rewireable) fuse - OK
- 45 A fuse of either kind - not OK
- Type C MCB - not OK.

If the supply is PME we can take the max. Ze value to be 0.35 ohm rather
than 0.8. Zs now becomes 0.5 ohm and the fault current is 460 A. The
1.5 mm^2 CPC will stand this for 0.14 s. This doesn't change much
though. The 45 A fuses would still fail to protect the CPC, but a Type
C MCB would be OK (which is a bit academic because there's no reason
whatever to need to use a Type C device in this application).

There is now a clear answer to Tony's question (as modified):

- provided that the original cooker circuit is wired in 6 (or 10) mm^2
T&E, and

- the circuit length from the CU is not vastly in excess of 10 m, and

- the circuit is protected by a 30 A fuse or 32 A or 40 A Type B MCB,

- then the proposal to use short 4 mm^2 T&E cable 'tails' to the
appliances is perfectly OK. (I've said 4 mm^2 here because of the
concerns over the current rating of 2.5 when ambient temperature and
grouping factors are applied.)

QEF
--
Andy