ProMbrooke:

gkenyon:

That's interesting.


So, how should we approach TT systems?

Delayed RCD main in a plastic enclosure with 30ma sub-main RCD. However, clearing longer than 0.035 seconds of the main RCD may not be fast enough in areas of lowered skin conductivity.  

In all honesty I'd like developed countries to move away from TT supplies. TT, TN-C-S and PME are simply a choice between cons in a world where DNOs do not want to provide more conductors than it takes to get the lights on. TN-S provides all the advantages of superior fault clearing given by TN-C-S, but without parallel earth currents or the risk of exposed parts becoming live should the neutral be lost. 

In PME systems, the main issue we have is when the PEN breaks.

If we use a 4-core cable, with the armour or outer wrap as earth, we will have the same problem in TN-S systems. It will be a broken PE rather than PEN - although this time you won't know about it until you need it. At least with broken PEN, most of the time it's known about before it's a real issue.

There are things you can do to monitor PE - but there are also things you can do to monitor PEN (perhaps less reliable, though, if you have a single-phase installation in a three-phase network).


Because of this, the only advantage I can see that TN-S has over TN-C-S is that there are no diverted neutral currents in "normal operation". Diverted neutral currents can cause "tingles" in certain conditions, and of course not recommended in locations with explosive atmospheres. Where it may cause tingles, or livestock are involved, a buried earth grid can help control voltage effects of diverted neutral currents.


 

Stray voltage is the byproduct of TN-C-S.


Any idea why the UK started off with TN-S but moved away from it?

"Sub-main only requires 2 s disconnection time. Also, in TT systems, even in final circuits you're not limited to 30 mA RCD for some circuits, provided earth electrode resistance is sufficiently low to ensure 2x residual current rating. You can achieve 0.04 s disconnection with any non-delayed RCD if you can achieve 5x residual current rating."


 

Right, but you still have the issue of remote earth, whereby the user is subjected to full 230 volts for 2 seconds while outside the zone of bonding extraneous bonding. Where as with a TN supply, sub-mains typically result in some level of voltage sag across the terminals of the transformer.      




.

"If we use a 4-core cable, with the armour or outer wrap as earth, we will have the same problem in TN-S systems. It will be a broken PE rather than PEN - although this time you won't know about it until you need it. At least with broken PEN, most of the time it's known about before it's a real issue.

There are things you can do to monitor PE - but there are also things you can do to monitor PEN (perhaps less reliable, though, if you have a single-phase installation in a three-phase network)."

For a TN-S supply, in order for the earthing system to become live you need 3 failures: loss of the incoming PE,  failure of an RCD, and a fault in the system. Where as in TN-C-S, you only require one failure, the rupture or disconnection of the PEN.  

 

 

Because of this, the only advantage I can see that TN-S has over TN-C-S is that there are no diverted neutral currents in "normal operation". Diverted neutral currents can cause "tingles" in certain conditions, and of course not recommended in locations with explosive atmospheres. Where it may cause tingles, or livestock are involved, a buried earth grid can help control voltage effects of diverted neutral currents.

 

The required bonding and earthing to control ever increasing neutral currents ends up being more costly in the long run. You also have the issue of magnetic fields, where a gamble is being waged on the hopes of eventually discovering 50Hz fields have no effect on organisms. 





 

 

"cost"


I am a bit puzzled by this, in that I would think that material resources were more scarce post WWII than they are today.


Forgive the bold. I don't know how to work the quoting system on their forum. Still new to it. My apology ?  

mapj1:

TT can be done well, and is useful for situations where an offset voltage between CPC and terra-firma is not desirable - animals on earth floored barns are one example, and humans in hot tubs outdoors are another.

I agree on the reliability of RCDs being a concern, but it does not have to be perfect, just as good as, or better than, the alternative.

In a TNs system, the classic double independent fault to danger is the live wire to the case of the washing machine, and an undetected  break in CPC path back to the substation. Earthing rotting off is not common, but not unknown, especially on underground cables.

There are easy ways to reduce the impact of a defective RCD, though we'd need data I do not have to say at what risk level the failed CPC  becomes more likely that a dead RCD.


One way  to reduce the risk of un-detected RCD fault  is regular checking,  every 3 months, perhaps 6 months , or as is more likely in a domestic setting, never in a million years, unless there is a real fault. The regs do include this, but I do not think it actually works in many situations.


If an RCD has say a 5% chance of failure in the first 10 years of its life (and this is a pure PIDOOMA figure for the sake of explanation   define PIDOOMA - I do try and give my sources....- if we have better figures we should use them,) then we can reduce the chance of failure from 1 in 20 to 1 in 400 by having 2 independent RCDS in cascade.

Of course if the RCDS were the same age and make and in the same box this advantage is partly vitiated as they may suffer the same design weakness and the failures are no longer independent.

This is the approach taken on on a UK campsite, where the site hookup, and the caravan, are both supposed to contain a double pole RCD, and are almost always of different age and make.


For the same reason I am in favour of the 100mA   ~ 1/10 second time delay (S type) in an up-front plastic box, and then the 30mA instant types or RCBOs in the final circuits, though this is only really mandated in agricultural settings.

In a large TT system you may see things tiered, with perhaps 300mA 300mS  earth fault detection at the origin,  then  a few 100mA S types feeding  various sub-mains, then 30mA instant devices on final circuits.  the chance of a failure at all levels in the chain is very low, probably lower than that of an undetected loss of CPC.

Although precluded by UK regs at the moment, the use of sockets with a built in RCD feature are also good as another layer to the safety onion, and more likely to be tested, as the end user is less worried about just tripping off one appliance.

Done well such an approach can be incredibly reliable.

Mike.

PS and some developed world places idea of how to handle earthed neutrals in a non TT way is not ideal.

This would not raise eyebrows in NZ/Oz. I presume you can see the problem.



 

All those delays raise the duration of voltage to remote earth. 


In a TN-S supply, there is no NEV, which negates the complexity of it all. RCDs are simply a backup, and, prevent neutral to earth faults from going undetected.


There is no longer any reason what developed countries use TN-C-S. Like TN-C, it is one of once limited material and knowledge to be gained.

gkenyon:

When PME was introduced, we were still in the "post-war" period.


Now we are in a position where, in existing urban areas, PME is here to stay.


I understand some DNOs are offering TN-S for new-build - of course, there is a slight cost increase for the extra core.


With TN-S, you only need one failure after broken PE, on circuits where RCDs are not used. Where RCDs are used, I thought you said you weren't happy with reliance on the RCD?


Your point regarding earthing is very valid ... for all system types. The impact on the effectiveness of main bonding due to the change to plastic service pipes can't be underestimated. I have measured the effective combined earth electrode resistance of my water supply pipe, which is still lead from the street, and it's well under 4 ohms.


How should we deal with that? Germany insist on foundation earth electrodes, which achieve a similar result ... but that idea keeps being shot down here in the UK. I'd like to see some alternative approaches considered. Don't forget, the loss of plastic service piping is not a fault of the DNO, it's a result of health & safety (replacement of corroding metal gas mains for plastic) and public health (replacement of iron and lead water pipes for plastic) and far outside the DNO's control.

Two failures for TN-S without RCD, Three failures with RCD. RCDs aren't 100% reliable, hence why I advocate a foundation of low loop impedance and rapid disconnection. As has been said in this thread by others, if you have say 0.5% probability of failure, 1.5% and 5% chance summed together the odds of all 3 failing at the same time are much more slim vs relying on only one.   


Personally, I think there needs to be less focus on earthing and bonding, and more on loop impedance. I've run the numbers and a lot of time having an earth grid below one's feet bounded to the MET makes little difference when in circuits 32 amps and below 99% of the VD during a fault is on the circuit itself and not the supply between the transformer and consumer unit.

mapj1:
Any idea why the UK started off with TN-S but moved away from it?


I have in moments of irritation / dark humour suggested that it is because it makes it easier for the DNOs as their workers no longer need to be able to count any higher than four .. though there are some very capable folk in DNO land, so that is probably unfair to judge all by the actions of a few.


In reality the CPC came almost for free in the era of underground lead sheathed cables and sweated  (wiped solder) joints, and once the hessian serving had rotted off or been eaten by mice, the lead sheath also served as as a kilometer length distributed electrode.  Overhead 2 wire service was always TT from day one, but the lead water pipe serving as the electrode was miles long was quite capable of blowing the low current service fuses of the day even though they were the old  hot wire type. 

The lead sheath was never seen as a credible current carrying conductor, so was never used as  the neutral, and the TNS earth Zs is allowed to be slightly higher because of that.

OF course under tension or in heaving ground, it is possible the lead cracks all round and the cores stay connected, and so an 'off earth' fault is possible, and not always spotted straight away.


The modern cable dispensed with the lead sheath some time ago, and those cables are now failing and being replaced, in many cases by aluminium clad cable where the outer is as good a conductor as the cores, and so is used as the neutral. The problem is that in turn that aluminium jacket rots off quite fast once water gets in, hence the open PEN problems.


It would be perfectly possible to make a modern cable with 4 segmented cores inside a metal jacket instead of 3, but there is some economy in not having 2 conductors at (well, nearly) the same voltage. It is essentially about cost, and the fact that the open PEN fault is rare enough, like car crashes, to be acceptable.

Mike.

I remember on another forum someone posting a TN-S overhead earthing supply from the 40s or somewhere about there. I feel like back then people understood electrical theory to a great depth.


IMO, I would dispence of metallic sheaths and simply have 5 cores in an outer jacket. NYY cable, but for underground DNO use. 

ProMbrooke:

Personally, I think there needs to be less focus on earthing and bonding, and more on loop impedance.

That's interesting. There will be less control of touch voltage. Loop impedances in TN-S systems are typically higher than TN-C-S.


With microgeneration and other forms of network stability control that are necessary for embedded generation in the move to DSO, effective loop impedance can change. In addition, taking an accurate earth fault loop impedance reading will become a near impossibility ... and perhaps meaningless.


Guidance already advises a check of Ze only to confirm an external earth connection for supplies to  - otherwise, assume Ipf = 16 kA / Ze = 0.35 Ohm (TN-C-S) or 0.8 Ohm (TN-S).



What precisely are you advocating?

ProMbrooke:in circuits 32 amps and below 99% of the VD during a fault is on the circuit itself and not the supply between the transformer and consumer unit.

Huh? Let's take some typical order-of-magnitude values for a suburban house. Ze might be around 0.2 Ohm, while R1+R2 for a radial socket might be about 0.4 Ohm. A short at/near the socket will result in about 2/3 of the VD in the circuit itself.  For it to be 99%, R1+R2 would have to be around 20 Ohm.