The wires you have shaded in yellow, only need to be continuously rated for the (16A) load connected via W3. However, in a normal extensive  installation they must not be so thin that if there was a short just prior to the 16a breaker, that the wiring may fail before the upstream devices (125A) would operate.

However, there is a further clause that if the distance is short (usually 3m or less) and the mechanical arrangement means a fault anywhere along that thin link was not very likely, and if it did happen, then not an matter of life and  death, only more of an expensive nuisance, then even that condition need not always be met.

We are allowed to accept a very small risk that if the short piece of thin cable suffers a fault in the unprotected length, then it will be damaged and have to be replaced, subject to those other considerations.

Against a 125A breaker, assuming it is not an MCCB with a long delay, 2.5mm or 4mm will  meet 16 A load need and the tripping for shorts anyway,  however 1.5mm might be seen as pushing it !! .

If that is how you read it, you are correct ..

Mike.

This is a case of the difference between overload and fault protection. It is as Mike says above, but it is easy to provide fault protection as well, because you simply apply the usual adiabatic equation and a maximum conductor temperature of less than the melting point. Damage may occur but the 125A breaker will trip. You may like to check this, but as Mike says, failure is not a significant worry.

I suspect you might struggle to give 2.5mm² proper fault protection from a 125A OPD. The withstand for 2.5mm² is typically around S²k² - 2.5² x 115² = a little over 80,000A²s whereas a 125A BS 88 fuse has an energy let-through of somewhere between 100,000 and 200,000 A²s according to some manufacturers.  Other manufacturers may differ of course. I don't have any figures for MCCBs to hand, but it's not unusual for a circuit breaker to have a greater energy let-through than fuses when fault currents are high.

So 434.2.1 may well be your way forward.

On the other hand even a modest increase in c.s.a - say from 2.5mm² to 4mm² or 6mm² might well allow you to provide fault protection directly - all depending on the actual numbers of course.

   - Andy.

Thank you Mike,

All answers useful and aligning with my expectations.

I can use 434.2.1 as the way to "justify" not to have to increase the CSA, but ideally I could use a cross-section that can withstand a bit more S²k² in line with the external protection. But that can be a bigger task if I have an even bigger rating breaker upstream. Where do you draw a line?

All clear in my view.

Adding just one last point, which I consider relevant.

Being something wired inside a distribution board, will this still be in the scope of BS 7671?

Or if we are referring to a machine, should I be looking for BS 7671 or something else e.g. EN 60204?

Where does one overlaps the other?

David Stone 

AJJewsbury 

Thank you also for your answers.

Much appreciated.

The 80 k j/ohms for 2.5mm is not the copper melting, it is the plastic being warmed so that when cooled it is then embrittled, and might crack if flexed. That sort of cable damage is not ideal of course, but not perhaps the sheets of flame and ejection of molten metal that some might be expecting. For that you need the Onderdonk formulae or something more accurate.
Mike