The diversity of Diversity

Yes, trip curves for MCBs are available. For overloads as distinct from short circuits, the trip curve is SOMEWHAT similar to a fuse.

I would expect a 32 amp MCB to handle 35 amps for many hours, possibly forever. And 40 amps for about an hour.

MCBs are a low priced mass produced item, not a precision device. On a moderate overload the time to trip is very variable and depends on ambient temperature and on manufacturing tolerances.

MCBs have both a thermal element operating on a bimetallic strip (which trips faster on large overloads, slower on smaller ones - rather like a fuse) and a magnetic element which almost instantly triggers a trip once the current reaches a pre-determined high level (e.g. somewhere between 3x and 5x the MCB's rating for a B-type MCB).

   - Andy.

Thanks Andy. I found the curve below which suggests a current of 2x rated current is likely OK for 20 seconds or less. 1.5x for almost 2 minutes.

Not good to design around expected regular exceedance, but this demonstrates short 40A excursions on a 32A ring won't trip the MCB.

The next silly question is how much (if at all) of a no-no is it to have two rings off one MCB? I'd rather have separate rings for the two rooms as then the 26A rating of 2.5mm2 wire feels OK against a single point of failure.

broadgage: 
 

Yes, trip curves for MCBs are available. For overloads as distinct from short circuits, the trip curve is SOMEWHAT similar to a fuse.

I would expect a 32 amp MCB to handle 35 amps for many hours, possibly forever. And 40 amps for about an hour.

MCBs are a low priced mass produced item, not a precision device. On a moderate overload the time to trip is very variable and depends on ambient temperature and on manufacturing tolerances.

You are right, below is a clearer curve than from my previous post which crossed with your post:

Nothing prohibits connection of two rings to one MCB, it is however unusual.

Each circuit requires its own fuse or MCB, but if two circuits are connected to the same MCB, then by definition they become one circuit, of an unusual but permissible design.

The other caveat is whether the MCB terminal and the neutral busbar terminals can accept four conductors, each of 2.5mm.

Ah - the centre fed figure of eight - very common when there are no free spaces in the consumer unit - seen by many as a hopeless bodge, but quite safe if done well, even if deprecated ?.

To be serious, in the same way that radials are not all daisy chains like the text book, and may be centre fed, and may have more branches than a christmas tree,  the humble ring may become more complex . Realise that it does make it harder to test and debug if there is a fault.

Common ring variants are the lasso=lollipop,  the side fed figure of eight, and the normal ring with an oversized spur feeding several sockets (AKA the reverse lasso) ones with spurs that interlock and rejoin forming more of a railway siding … 

But so long as cables are not likely to be overloaded, and far end voltage drops are not excessive, then any wiring configuration is certainly permitted by BS7671. 

Permitted but non standard layouts are about as popular as nipple piercings in a nunnery in the sense of the opprobrium from the purists who like linear radials and rings without spurs, as these are in the onsite guide, and need no great design thought to justify.

This is with good reason, as fairly often a non standard layout is not in fact the result of some brilliant masterstroke  of design by a hitherto unrecognised genius,  but rather some hapless cock-up by an over-keen diy type, and in that case serves as an alarm bell that there may  be more horrors to be found.

Mike. 

WireWeHere: 
The next silly question is how much (if at all) of a no-no is it to have two rings off one MCB? I'd rather have separate rings for the two rooms as then the 26A rating of 2.5mm2 wire feels OK against a single point of failure.

I agree with the comments above, but why would you do it? You may as well either join the two parts as a ring and not bring the cables back to the CU; or make two rings. I assume that this is a new build so there will be no difficulty fitting a CU with umpteen ways.

Incidentally, I don't like the thought of points of failure, if the installation has been designed and erected properly, there should be none. (So called, “nuisance tripping” is usually due to bad design, bad installation, or both.)

I might add that the choice of circuits depends to some extent upon the layout of the house. If the CU is in the utility room and close to the kitchen, it really is no trouble to make two circuits; but if it is at the other end of the house, you may want to save the bother.

“This is with good reason, as fairly often a non standard layout is not in fact the result of some brilliant masterstroke  of design by a hitherto unrecognised genius,  but rather some hapless cock-up by an over-keen diy type, and in that case serves as an alarm bell that there may  be more horrors to be found.”

Wise words and put in a very amusing way.

Many times I see one ring final (or radial) feeding sockets and one or two lighting circuits on a cheap rewire and in practice they last for years with few problems. 

Personally I like to see two socket circuits and two lighting circuits as an absolute minimum in any installation. That is just to reduce the chances of losing all sockets or all lights at any one instance.

If you had two identical properties with identical use then the one with two of each circuit would not use any more than the one with one of each circuit. It is dependant upon likely use total. 

Therefore applying diversity calculations often do not match.

That is why the floorspace consideration (plus intended type of use) seems more appropriate.

I recently did a rewire for a friend of mine with 5 rings and 5 lighting circuits but two of each would have been OK . It just added a bit more resilience in case of a fault (or powering a circuit down for additions and alterations).

Yes it`s ok to have two rings (or radials) on one MCB it is still one circuit. Although it is sometimes an indication that someone has created a spare way to create a shower circuit or something and modified what was originally in place. It can be done, for example, to reduce ring end to end cable lengths of a massively drop fed circuit thereby having its merits if you do not have a spare way in the consumer unit and normally you`d be happy with one circuit feeding all of those sockets (or lights) . 

Whilst having loads of different circuits can indeed make it possible for the total load drawn, the really big question is, what is the likely max current draw for anything over X time (X might be 5 mins, 10 mins etc etc etc) . How do you calculate diversity? that is the question.

ebee: 
Personally I like to see two socket circuits and two lighting circuits as an absolute minimum in any installation. That is just to reduce the chances of losing all sockets or all lights at any one instance.

Agreed. Also with sockets and lights e.g. on each floor fed by a different RCD so either a room light or table lamp stays on in the event of a fault.

Agreed Chris, although I tend to favour the RCBO approach if possible, it can encourage larger “earth losses” in a system to go unoticed though, but not as much as the sole MCB/Fuse systems I`m used to because of my age

I am now moving away from Rings in favour of 20A and 32A radials, much easier to manage and more difficult to abuse!