Do Distribution circuits require RCD protection?

[WilliamAtkin]

I'm doing a college design project and need to choose a suitable range of circuit protective devices for the whole contact. I've chosen and already done all the protective devices for all my final circuits but was curious if the distribution circuits need RCD protection aswell.
The project doesn't give to much info on the distribution circuits just that I had to label them on the diagram to show how I'm supplying all the db boards in the new hotel extension. The SPEC only gives info with specifics for one distribution circuit which is a 32A distribution circuit to supply a four way SP+N distribution board with 3 core SWA and for that specific distribution circuit I dont think it needs to be rcd protected as Its being supplied via SWA cable, but for the 8 distribution circuits in the new hotel extension there is no information. I asked my boss and he said he wouldn't as something to do with discrimination but he's not too good with the regs and says he aint really looked at them since he finnish college so he isn't the most reliable person for advice with college theory work . I'm pretty sure that these distribution circuits don't need rcd protection and that I can move on to the next question in the project but I just wanted to run it by here and double check before I moved on incase I was wrong

 

[Lucien Nunes]

The aim is normally to design and install distribution circuits so that they don't need additional protection by RCD at the main board, thereby avoiding selectivity a.k.a. discrimination problems. So you would avoid e.g. twin and earth buried less than 50mm in the wall, as this would require RCD protection of the same type as on the final circuits (non-delayed 30mA). In that case, a final circuit fault is as likely to trip the distribution circuit RCD as its own, taking the whole sub board down. A distribution circuit wired in SWA or other suitable method doesn't require additional protection, so the problem is avoided.

One other situation might be that the Zs is too high on the distribution circuit to rely on ADS for fault protection, in which case you might need to provide an upstream RCD, but in this case selectivity can be achieved because you're not limited to 30mA non-delayed.

 

[WilliamAtkin]

Ok thank you and noted , I will look in to the discrimnation part in the regs

 

[pc1966]

The aim of selectivity is making sure that only the closest protection trips in the event of a fault, reducing the impact of the fault on other healthy circuits.

It is not always easy/practical to achieve (e.g. MCB feeding MCB is troublesome), but good design tries to get an acceptable arrangement considering the use-case and risks arising from the supply failing.

 

[WilliamAtkin]

The aim of selectivity is making sure that only the closest protection trips in the event of a fault, reducing the impact of the fault on other healthy circuits.

It is not always easy/practical to achieve (e.g. MCB feeding MCB is troublesome), but good design tries to get an acceptable arrangement considering the use-case and risks arising from the supply failing.

what current rating mcb would be used normally to supply a 230v db board ?

 

[pc1966]

what current rating mcb would be used normally to supply a 230v db board ?

It depends on what the sub-board needs.

But the underlying issue with MCB feeding MCB is how they work. For "overload" the thermal curve is fine at a ratio of 1.6:1 or so, for example, a 10A MCB fed from a 16A MCB would reliably disconnect first if you had a small excess in the 20-30A region. But once you get in to the "instant" magnetic trip region for a hard fault there is very poor selectivity. Once you hit the trip point for the up-stream MCB then, while the down-stream MCB will be clearing the fault, it is almost certainly not fast enough to stop it going.

For example, if you look for the Hager commercial catalogue then in the later section it has tables of selectivity for various combinations of supply fuse/MCCB/MCB and downstream MCCB/MCB/RCBO/AFDD. They give you an idea of what will reliably work based on the final circuit's PFC/PSCC values and your choice of device.

Fuses are well-behaved and for BS88 series you typically get total selectivity for a 1.6:1 ratio (i.e. for all PFC up to the breaking limit for the fuse, and that is usually something bonkerly big like 80kA or more), but for most other devices you need to consult such tables.

For RCD it is simpler in that the design rule is usually a 3:1 ratio in trip current and a 200ms additional delay per level. So final might be 30mA "instant" RCD/RCBO, that is fed from a 100mA RCD that has a 200ms or more delay ("selective" type) and if it has a RCD further up it would be 300mA and 400ms+ delay setting, then 1A and 600ms, etc.

 

[WilliamAtkin]

It depends on what the sub-board needs.

But the underlying issue with MCB feeding MCB is how they work. For "overload" the thermal curve is fine at a ratio of 1.6:1 or so, for example, a 10A MCB fed from a 16A MCB would reliably disconnect first if you had a small excess in the 20-30A region. But once you get in to the "instant" magnetic trip region for a hard fault there is very poor selectivity. Once you hit the trip point for the up-stream MCB then, while the down-stream MCB will be clearing the fault, it is almost certainly not fast enough to stop it going.

For example, if you look for the Hager commercial catalogue then in the later section it has tables of selectivity for various combinations of supply fuse/MCCB/MCB and downstream MCCB/MCB/RCBO/AFDD. They give you an idea of what will reliably work based on the final circuit's PFC/PSCC values and your choice of device.

Fuses are well-behaved and for BS88 series you typically get total selectivity for a 1.6:1 ratio (i.e. for all PFC up to the breaking limit for the fuse, and that is usually something bonkerly big like 80kA or more), but for most other devices you need to consult such tables.

For RCD it is simpler in that the design rule is usually a 3:1 ratio in trip current and a 200ms additional delay per level. So final might be 30mA "instant" RCD/RCBO, that is fed from a 100mA RCD that has a 200ms or more delay ("selective" type) and if it has a RCD further up it would be 300mA and 400ms+ delay setting, then 1A and 600ms, etc.

Ok so if this db board that I'm feeding had 7 RCBOS in it: two 20amp RCBOs , four 16amp RCBOs and one 6amp RCBO what current rating should the distribution circuit thats feeding the db board be ?

 

[timhoward]

Ok so if this db board that I'm feeding had 7 RCBOS in it: two 20amp RCBOs , four 16amp RCBOs and one 6amp RCBO what current rating should the distribution circuit thats feeding the db board be ?

You need to think about the expected loads and consider diversity. i.e. what is actually likely to be drawn, as opposed to what would it take to trip every single breaker.
The 6 amp RCBO could for instance be a lighting circuit where with every LED light turned on less than 1 amp is drawn.
The Onsite Guide has some guidance on diversity. I'd encourage you to have a look at that and have a go at arriving at a number. By all means come back here and say "how does this sound" (with more details about what the circuits are for)