Is it common to find too-thin wire in an installation?

[Dave A]

As you probably realise, I'm playing the devil's advocate here.

Here's another one- if I ring-feed the plug points on each circuit, can I use 1.5 mm?

[AndyD]

Ah ha. Now you're talking my language with a ring main. This is a standard installation in the UK but the British Standard spec there is also 2.5 mm² cabling and the socket outlets are only rated at 13 Amps not 16 Amps as in SA. I think we can safely say the answer to your question is 'no'.....but nice try though

[murdock]

yeah right ring circuit...you would find plenty electrocuted electricians in sa...they cant even get a standard circuit right...now yu want confuse them...cascading systems are bad enough...

[berndj]

I find this thread fascinating.

Has anyone found scorched 2.5mm behind 20A protection (other than right next to a bad connection, of course)? The justification is it seems to do the job and it's the way it's always been done.

Dave, I've often wondered about standards vs practice, and the resultant need for sometimes very generous safety margins in standards. This might be a manifestation of this effect.

Another possibility is that, perhaps, the 2.5mm^2 wire is damaged, but in a way that you can't easily see it, and that damage is cumulative. If the user is very disciplined and never trips the circuit, there's no damage. But if you just plug in heaters and washing machines and dishwashers willy-nilly, your wire could be creeping through the PVC as it repeatedly goes past its vitrification point at 80 degrees C.

With some mad physics-fu (*) and a few assumptions, I calculated the temperature rise of a wire during a 0.1s 20x overcurrent condition. I think I saw some graphs somewhere that suggested this might be the worst-case behaviour of "curve 1" breakers.

2.5mm^2, 15A circuit: 7.5 degrees
2.5mm^2, 20A circuit: 13.4 degrees
2.5mm^2, 25A circuit: 20.9 degrees

4mm^2, 15A: 4.7 degrees
4mm^2, 20A: 8.4 degrees
4mm^2, 25A: 13.1 degrees

My guess is that the 10 degree headroom between the maximum operating temperature specified in the standards, and the vitrification point, is exactly the room you need to survive an overcurrent condition without damage. Note that each of the calcs above which result in less than a 10 degree rise, are allowed.

Interestingly, notice also how 4mm^2 in a 25A circuit is marginally "safer" in this perspective than 2.5mm^2 in a 20A circuit.

(*) For 0.1s, ignore heat loss through radiation, conduction, or convection. It all goes into the copper, and only the copper. Take resistance per unit length, to calculate I^2*R losses per unit length. Multiply by 0.1s to get thermal energy dumped into the copper. Divide by copper's heat capacity to get rise in T.

[AndyD]

Another issue is that during current flow through a copper conductor most of the current flows at the surface. During overload this is also the case and the surface of the conductor would be where most of the heat would be developed. With a single conductor such as a FT+e cable the conductor has a lower surface area ratio to mass than say house-wire or panel-wire which is a stranded conductor with a larger surface area hence the house-wire should generate less heat. No allowance is made for these different conductor constructions.

You mention conductor creepage and vitrification of the PVC insulation. From experience, the first sign that a wire in a cable has been overheated is that the PVC insulation is bonded to it by the heat generated. At the termination point the PVC insulation may have retreated and shrunk back but I would suspect a higher temperature at the connection or where a ferrule or lug is crimped on would be responsible for this. I wouldn't see conductor creepage occurring in a multistranded wire which would have a twisted construction.

yeah right ring circuit...you would find plenty electrocuted electricians in sa...they cant even get a standard circuit right...now yu want confuse them...cascading systems are bad enough...

Both sides of a ringmain are fed from the same MCB so there's only one point of isolation. This shouldn't increase the risk of shock when working.

[Dave A]

Ah ha. Now you're talking my language with a ring main. This is a standard installation in the UK but the British Standard spec there is also 2.5 mm² cabling and the socket outlets are only rated at 13 Amps not 16 Amps as in SA. I think we can safely say the answer to your question is 'no'.....but nice try though

That's belts and braces thinking for you - they're catering for breaking the ring, I expect. But humour me one more time with this one:

What if you did a normal branch reticulation doubling up on 1.5mm all the way through?

(I am actually heading somewhere with this).

[AndyD]

(I am actually heading somewhere with this).

Lol, and I thought you were still just poking the fire for the sake of it.

If you installed a standard skt cct in SA using 2 x 1.5mm² FT+E cables all the way through I think it would be okay as far as conductor sizing would be concerned. It's one I've never come across before but without diving into the regs it might however raise other issues such as the occupation percentage of the conduit for one. This is also indirectly a temperature issue.

[berndj]

If you installed a standard skt cct in SA using 2 x 1.5mm² FT+E cables all the way through I think it would be okay as far as conductor sizing would be concerned. It's one I've never come across before but without diving into the regs it might however raise other issues such as the occupation percentage of the conduit for one. This is also indirectly a temperature issue.

What happens if exactly one of the conductors fails open? Perhaps you got a bad batch of PVC glue, the conduit segments moved apart when the alarm technician installed his cobweb, and a rat fancied a nibble at the wire insulation. What has to happen in order for you to know that it has failed?

[Dave A]

If you installed a standard skt cct in SA using 2 x 1.5mm² FT+E cables all the way through I think it would be okay as far as conductor sizing would be concerned. It's one I've never come across before

At least not in plug circuits - realistically because of cost efficiency. But in situations with big current, having multiple cables in parallel is not uncommon at all.

This now begs the question - doubling up in parallel is probably OK (from a regs point of view at least), so why isn't doing much the same thing but in a ring feed configuration a problem?

What happens if exactly one of the conductors fails open?

Precisely. In such a situation a ring feed reduces the risk of such a failure causing catastrophic results.

[Sparks]

My plug circuits are minimum 2.5 but depending on load and length I do go thicker. As for ring circuits, not allowed in domestic installations, exactly as has been stated too many idiots in the field let alone the DIY crowd. The only time I will accept a 1.5 cable with a socket outlet on it is in a mixed circuit, obviously then it will not have a 20A CB.

[AndyD]

I have seen ring circuits before in SA but very few and a long time ago. I don't see them as illegal according to the regs and I don't see them as a hazard as Murdock suggested. They just have redundancy built-in and are obviously more capable of handling fault currents etc.

Not being specified as standard would make them cost prohibitive to install as they require one extra cable and conduit back to the DB but in the context of this problem of a high temperature roof space it would be a reasonable alternative solution in my book. The cost difference between a 2.5mm ring circuit with 20mm conduit and a standard circuit on 4mm cable and 25mm conduit would probably be negligible.

[Sparks]

Spec yes, but in this country definately not "safe" therefor not meeting criteria for my COC.

[AndyD]

Spec yes, but in this country definately not "safe" therefor not meeting criteria for my COC.

Please explain why it's 'definitely not safe'. Which part of the SA regs would make it unacceptable to install a ring main? Why would it be unsafe here in SA yet safe elsewhere?

[berndj]

This now begs the question - doubling up in parallel is probably OK (from a regs point of view at least), so why isn't doing much the same thing but in a ring feed configuration a problem?

Precisely. In such a situation a ring feed reduces the risk of such a failure causing catastrophic results.

My worry would be that if exactly one conductor failed, your circuit would still appear to be "working". But that single conductor is now grossly undersized for the current that it might be carrying, but you don't find that out until either you do a (routine) check (but realistically, how often does that happen?) or your building burns down - because your circuit breaker doesn't have a problem with the current.

With single conductors, if one fails, the whole circuit fails - it makes the failure very obvious.

As for why parallel is fine but ring isn't, I'd say it's because you're far more likely to get an equal distribution of current through the parallel conductors than in the case of a ring. Think of a ring that's 50m long in total, with an outlet at 10m from the CB: those 10m will carry nearly all of the current, so to have a safe installation, you need the whole ring to be of a thickness capable of running the max current on its own.

[Dave A]

So let's sum up. I think we can acknowledge that:

• Bernd has a point.
• There is science behind the numbers.
• You can't rely on the current rating per the packaging.
• The potential scale of the issue could be huge.

Why it hasn't caused widespread problems so far is probably a combination of eating into the safety margin and the fact that it has to be really rare that a plug circuit runs continuously close to maximum protection rating, due in most part to the start-up current draw profile of nearly all plug-in appliances.

Fortunately the numbers are not much on the wrong side of where they should be, but they are on the wrong side of where they should be under certain circumstances which runs contrary to regs and COC standards - and therefor the issue shouldn't just be ignored (particularly bearing in mind the changed potential for claims with the introduction of the Consumer Protection Act).

Some exposure on the issue would deal with new installations. It's the existing installations that are troublesome - probably on a number of fronts.

In terms of raw flack, there's probably enough fudge in the history of the widespread adoption of surfix to keep fingers pointing round in circles for decades.

(And here's where I was heading in the 1.5 mm ring-feed questions)

At a technical level, it's a case of either downrating the protection (much more nuisance tripping complaints), or upping the wire where it counts. Now given that the problem is really confined to roof spaces, I'd suggest a simple ring feed solution (could even be confined just to the roof loom) - adding a 2.5mm feed from the db to the furthest point on the circuit.

I only raised ring circuits at the 1.5mm level to demonstrate how powerful this solution would be.

Anyway - rough sketch of my thinking - I'm sure it needs ripping to shreds, refining and squaring away of the details/fine print etc.

Two more things perhaps worth mentioning:

1. If line resistance is primarily a function of surface area, round conductors are the most inefficient shape. You could up the rating of surfix simply by using flattened copper conductors.

2. I asked my IE what the maximum circuit breaker rating can be on plug circuits with 2.5mm wire - he said 25A. Only when I said Aha! did he say "Are we talking GP wire?"