An interesting problem, we read "The load current in any part of the circuit should be unlikely to exceed for long periods the current-carrying capacity of the cable (Regulation
433.1.5 refers)." and it continues with suggestions, but the problem is with a ring final near to the origin a series of loads can cause a current of over 20 amp in one leg of the ring.
We when doing the C&G 2391 were told to look for a fig of 8 as this could allow one leg to become overloaded, before the RCBO came in I had considered using a double pole 16 amp MCB so if either leg exceeded 16 amp it would trip, however never done it or seen it done.
However this was one of the reasons why not to use the lollipop system, where a large cable brings the supply across the house, possibly a 6 mm² into a junction box, a cooker connection unit seem favourite, and then form a local ring from that point. By using heavier cable one can get over volt drop problems do to distance, and get a better loop impedance, and it has been used for kitchen ring finals, but then one needs to work out likely hood of overload near the split.
The common practice at one point was to use two 2.5 mm² cables to a grid switch, and then a series of radials to fixed items like fridge, freezer, and washing machine, main reason for two 2.5 mm² cables rather than one 6 mm² cable was could not get a pair of 6 mm² cables in the 20 amp switches. it was not because of overload of one leg, but it did however stop the overload of one leg as a result, often called the kitchen ring final.
My mothers house had a 10 mm² SWA cable around outside of house to feed a kitchen consumer unit, it was done because the house needed a re-wire, so could not fit RCD's in main consumer unit, but kitchen needed RCD protection, so 4 x RCBO's fitted in the kitchen, latter after a re-wire this remained, and fed with a non RCD protected MCB in main board, which the use of SWA allowed, had it been 10 mm² twin and earth cable, the RCD protection would be needed in the main CU.
What we are looking at is design, and to be frank we often don't design, we do it that way because we have always done it that way, we are told limit of 106 meters of 2.5 mm² in a ring, which is great, no more than one role in a ring, but once installed we can't measure the length using a tape measure, we need to calculate the length, by measuring the loop impedance or prospective short circuit current we can work out the length of a radial, but for a ring final we need to either measure the resistance or have a guess where the centre is.
We may have worked out in collage that the line - neutral loop needs to be around 0.94 Ω when the incomer is 0.35 Ω but how often do we really check? Yes we check below 1.38 Ω so a type B will trip with a short circuit on the magnetic part of the MCB, but how often do we really check the volt drop?
Is volt drop really that important? With fluorescent lamps with wire wound ballast, or fridges and freezers with single phase capacitor start motors maybe volt drop was important, I had one time where AC's were failing do to volt drop, but it was rather large, the Turkish electrician thought that two 110 volt to neutral supplies from a 3 phase supply = 220 volt, however is only equals 190 volt, it was not from a split phase, it was from a three phase supply.
However although most freezers today say do not use with an extension lead due to volt drop problems, they can stall on trying to start and in the end the overload fails due to repeated tripping, in the main they are designed to run on 220 volt, although we harmonised with Europe to 230 volt until we started using EV's and solar panels, we stayed at 240 and rest of Europe stayed at 220, so most equipment will work fine on a 220 volt supply. It can drop to 207 volt and still be within spec, and switch mode power supplies and pulse width modulated chips in LED lighting means volt drop is hardly a problem.
As to DIY how many can measure loop impedance to be able to work out volt drop? At one point I worried that if I extended a ring final, that some one in the future doing an EICR could work out it was too long, and I would be required to correct it. However once I had built the java script software to convert loop impedance readings into volt drop and ring final length, I realised the loop impedance readings are not accurate enough to show one is wrong. Remember two readings required supply loop impedance and centre of ring loop impedance, so 0.02 becomes 0.04 ohms +/- and also the incoming reading can alter due to loading, so before one could say an error had been made we would need it to be around 0.08 ohms out, so one could add around 14 meters of cable over the 106 meter permitted without anyone being able to prove you have made an error.
Clearly if well out we would be able to see it was, but would need to be over the reading required for short circuit protection (1.38 Ω) before one could really say there was an error, although we use with a ring final 20 amp at centre and 12 amp even spread so we work on 26 amp for Ib this is not in the rule book, so a designer could claim he used 13 amp at centre and 19 amp even spread giving a much lower Ib.
Oh what a tangled web we weave, this could go on for a record in thread length as it is clearly not well defined as to what is permitted. I will read with interest what others think. I have assumed
@Observer has done what ever years ago, so this is just academic.