RCDs in series wont reliably discriminate unless the upstream RCD is of a time delayed type. Try to avoid more than two RCDs in series and make certain that the upstream device is a time delayed type.

MCBs in series wont reliably discriminate on short circuits, only on overloads. Avoid MCBs in series and use fuses for protecting upstream sub mains with MCBs reserved for final subcircuits..

BS88 fuses in series will usually discriminate if of different ratings, and will almost always discriminate if TWO standard sizes apart.

32 amp and 40 amp BS88 fuses in series will probably discriminate, 32 amp and 50 amp will almost certainly discriminate.

A very clear response thank you!

broadgage said:

MCBs in series wont reliably discriminate on short circuits, only on overloads.

Yes, if it's the mythical zero resistance fault. However, if it's at the business end of a hand tool, you might stand a chance.

This is unlikely to be a satisfactory design. The top drawing is far from clear, and the circuit descriptions below do not fully match, and cascaded submains may well give you loop impedance problems as well as lack of discrimination. Cascaded RCDs are not a good idea as you have them, where does the Customer look for a trip? Your design may be easy to install, perhaps be a bit cheaper, but is far from good. I suggest a rethink, and as Broadgage says BS88s until you get to a final circuit breaker will give less problems for the customer, and discriminate better. The changing from type Bs to Cs is curious, why? A plan of the property would also be useful, and if the different parts have different occupiers, then it is even worse.

 Thanks for your response David, I should have stated from the start that this is an existing installation on a large commercial unit which has been split down into various workspaces (offices and art workshops for example). It is evident from the installation that submains have been added over time. I created the illustration above to try and give a picture of the site.

The customer has requested a full inspection/ verification of the installation as its way out of date and he would like to start renting out the spaces and hopefully split them down further to individual metered units. The units currently don't have any tenants so there's barely any load on these boards, but as soon as space heaters are introduced tripping will occur for sure.

I am proposing a new three phase supply the length of the property then branching from that. As the property is just shy of 100m over two floors they might not go for this proposal so the BS88 fuses maybe the alternative solution.

In regards the RCDs I tested the furthest one and all upstream RCDs trip so clear they aren't effective.

What further information would be required for a clear picture.

Philip Conlan said:

The customer has requested a full inspection/ verification of the installation as its way out of date

Hm, what is "way out of date"? 40 years? Is the condition bad enough to get the customer to agree to a complete re-wire?

It depends on which board we look at, the only previous inspection evident is from march 2016 on DB6/7 and should have been tested again 2021.

The rest of the installation is well 10 years of age with no previous testing evidence/certification. 

For what the customer wants a rewire is the only way forward in my eyes, but I sense that isn't an option due to budget.

I would not expect that an installation only about ten years old needs re wiring, unless it was either badly defective originally, or if needs have increased SUBSTANTIALY.

For reasons already given, MCBs in series are poor practice, but if already existing and not in practice giving problems, are arguably acceptable for continued use as distinct from being installed like that on a new job.

Series RCDs likewise.

RCDs in series wont reliably discriminate unless the upstream RCD is of a time delayed type.

Agreed - but I would add that the upstream RCD must also have a much higher residual current rating (e.g. min 3x the downstream rating) and be of a compatible type (e.g. a A type upstream of a AC type can't be expected to discriminate half-wave (rectified) residual currents).

   - Andy.