Circuit Wire For Long Run

Cookie · Apr 2, 2021

Ok, say you've got a lighting run with the last fixture being 1,280 meters away. 1,800 watts per phase 3 phase, 277/480Y volts though in this example we can use 230/400Y for the ease of discussion.

My question is- what size wire do I run for each phase and earth? What disconnection time do I use? Do I use a Type B, C or D breaker? What should my max R1+R2 come out to be?

cliffed · Apr 2, 2021

Street lighting would be an example in which the REC uses a concentric cable looping of the main ring supply cable.
Theres loads of software to calculate size required to local cutout in each fixture.

pc1966 · Apr 2, 2021

I suspect this is a specialised case as above.

But...if you simply followed the UK domestic requirements for 3% drop and assuming the lights are spaced sort-of uniformly, the feed cable comes out at 25mm, for 5% drop (reasonable for a case like this with LED or similar lights that are not that voltage-sensitive) it is 16mm. Risers to each lamp could be much smaller of course!

As a final circuit you would be looking at 0.4s disconnection so a B or C curve breaker. With a lot of simultaneously-switched lights (central control) then probably a C-curve for that and 10A C-curve MCB has max Zs of 1.75 ohms.

Assuming 4C 16mm with armour as CPC, then R1 = 1.15 mOhm/m, and R2 = 3.1 mOhm/m so your R1+R2 = 5.4 ohms which is too much, as would 5C cable as well (2.9 ohm). In this case I would probably put on a 100mA instant RCD to meet disconnection, as then still selective if the main DB has a 300mA delay incomer for fire reasons, etc.

But I don't design street lighting, so the above might be complete nonsense!

Cookie · Apr 2, 2021

pc1966 said:
I suspect this is a specialised case as above.

But...if you simply followed the UK domestic requirements for 3% drop and assuming the lights are spaced sort-of uniformly, the feed cable comes out at 25mm, for 5% drop (reasonable for a case like this with LED or similar lights that are not that voltage-sensitive) it is 16mm. Risers to each lamp could be much smaller of course!

As a final circuit you would be looking at 0.4s disconnection so a B or C curve breaker. With a lot of simultaneously-switched lights (central control) then probably a C-curve for that and 10A C-curve MCB has max Zs of 1.75 ohms.

Assuming 4C 16mm with armour as CPC, then R1 = 1.15 mOhm/m, and R2 = 3.1 mOhm/m so your R1+R2 = 5.4 ohms which is too much, as would 5C cable as well (2.9 ohm). In this case I would probably put on a 100mA instant RCD to meet disconnection, as then still selective if the main DB has a 300mA delay incomer for fire reasons, etc.

But I don't design street lighting, so the above might be complete nonsense!

Thnaks, but I don't want to use an RCD for disconnection. RCDs can fail and all disconnection in TN systems must be done by the breaker (MCB) itself.

pc1966 · Apr 2, 2021

Cookie said:
Thnaks, but I don't want to use an RCD for disconnection. RCDs can fail and all disconnection in TN systems must be done by the breaker (MCB) itself.

If the lights are separately switched (e.g. each has its own photocell) you could get away with a 10A B-curve MCB fine as no massive inrush surge. That would allow you to meet the disconnection with 5-core 16mm (£9.37/m from Superlec) or just 4-core 25mm cable (£9.24/m).

1280m * 4-core 16mm at £6.84/m = £8,755 + RCD
1280m * 4-core 25mm at £9.24/m = £11,827
1280m * 5-core 16mm at £9.37/m = £11,994 (easier to wrangle, more reliable earth)

Of course that difference probably is trivial compared to the project cost.

Cookie · Apr 2, 2021

pc1966 said:
If the lights are separately switched (e.g. each has its own photocell) you could get away with a 10A B-curve MCB fine as no massive inrush surge. That would allow you to meet the disconnection with 5-core 16mm (£9.37/m from Superlec) or just 4-core 25mm cable (£9.24/m).

1280m * 4-core 16mm at £6.84/m = £8,755 + RCD
1280m * 4-core 25mm at £9.24/m = £11,827
1280m * 5-core 16mm at £9.37/m = £11,994 (easier to wrangle, more reliable earth)

Of course that difference probably is trivial compared to the project cost.

Right, but a trip and reclose from then utility would engage all of them at once I think.

Do you have the ohms value for each conductor?

pc1966 · Apr 2, 2021

Last page of this has both conductors and armour resistances:

https://uk.prysmiangroup.com/sites/default/files/atoms/files/BS5467.pdf

Yes, if you had a supply interruption at night it might well power all lamps simultaneously. I know some UPS allow random-timed start-up to avoid that sort of issue, no idea if there is a similar idea for lamps to get round the inrush problem without having to push the MCB trip point up.

Which you can do, just it gets very difficult/expensive for the cable if you do it that way and still want to meet the required disconnection Zs at the end of a long cable.

If your lights were in a region so they could be ring-connected that would reduce the max Zs value (and drop as well, of course).

Or you live with the RCD to avoid the Zs limit of the MCB being an issue and put in two in series for redundancy, accepting they don't have selectivity so both need reset on fault.

But as mentioned at the start, there probably are folks/companies/tools that specialise in street light solution. I'm just guessing!

Cookie · Apr 2, 2021

Your guess is as good as mine

FWIW I'm getting a minimum size of 33.62mm2 copper Phase and Earth.

pc1966 · Apr 2, 2021

Also to add if the supply is not RCD protected (so no selectivity issues) you could even use a 300mA delay RCD, still meet the 0.4s disconnection time for shock, and be pretty robust against false trips.

So you could have a 300mA delay then a 100mA instant and retain selectivity unless the 100mA fails.

Cookie · Apr 2, 2021

Does BS7671 allow an RCD to meet disconnection times in TN? I just can't bring myself to do it.

Cookie · Apr 2, 2021

Reason I ask about disconnection times is that the ground around an energized lamp goes up in voltage (not my pic):

https://www.reddit.com/r/electricians/comments/mgze4b

Strima · Apr 2, 2021

Here's what your R1+R2 would be (Column D):

pc1966 · Apr 2, 2021

Cookie said:
Your guess is as good as mine

FWIW I'm getting a minimum size of 33.62mm2 copper Phase and Earth.

For what aspect, volt drop or meeting MCB disconnection R1+R2?

pc1966 · Apr 2, 2021

Cookie said:
Does BS7671 allow an RCD to meet disconnection times in TN? I just can't bring myself to do it.

It allows it but I would normally see it as poor practice unless there are no reasonable alternatives.

With over 1km of cable going to something as big as 35mm to get MCB disconnection and good inrush tolerance is pushing cost and assembly/repair effort up massively, so it moves towards "no reasonable alternative" territory.

It is possibly worth looking at other aspect normally taken for granted. For example the usual Zs limits allow a 0.8 factor for cable heating - that is not an issue for a heavy cable needed to meet VD when very long.

While usually seen as old fashioned, sometimes it is worth considering fuses (e.g. switched-fuse feed) as they often have a better combination of surge handling and Zs limit. Also one fuse blowing means you still have 2 out of 3 lights on!

pc1966 · Apr 2, 2021

Here are some examples for SWA data and for various OCPD assuming no significant cable heating and 1280m length:

OCPD	Zs (from OSG)	Zs No heating (Ohm)	R1+R2 (mOhm/m)
10A BS88-2 fuse	3.7	4.63	3.61
10A B-MCB	3.5	4.38	3.42
16A B-MCB	2.2	2.75	2.15
16A BS88-2 fuse	1.9	2.38	1.86
10A C-MCB	1.75	2.19	1.71
SWA choice	R1	R2	R1+R2 (mOhm/m)
16mm 4C	1.15	3.1	4.25
16mm 5C	1.15	1.15	2.30
25mm 4C	0.727	2.3	3.03
25mm 4C + 16mm	0.727	1.15	1.88
25mm 5C	0.727	0.727	1.45

pc1966 · Apr 2, 2021

Your 1.8kW/phase is about 8A, so 10A or more OCPD should be fine, subject to inrush limits.

If you control inrush then the 10A B-curve MCB is OK and it needs cable below 3.42 mOhm/m so all examples except 16mm 4C are fine.

If you used a 16A B-MCB for more inrush tolerance then it is 2.15 mOhm/m, so either 25mm 4-core and a supplementary 16mm earth, or 25mm 5-core is needed, etc.

pc1966 · Apr 2, 2021

I'm sure there is proper software to do this, just never got round to finding/learning a package!

Cookie · Apr 2, 2021

pc1966 said:
For what aspect, volt drop or meeting MCB disconnection R1+R2?

Disconnection times.

pc1966 · Apr 2, 2021

Should also add that I have not included and Ze or Zs at DB in the above table, assuming it is quite small for this discussion.

Equally you might just be lucky and find the lamp-posts are having 10-20 ohms when in the ground resulting in a small reduction in the R2 value.

I also did not do the parallel CPC & armour calculation as usually the regs assume the worst-case of the better conductor being used. Unless explicitly paralleled and then it is often assumed they are all similar conductors.

So there are possible factors making it better or worse!

Cookie · Apr 3, 2021

pc1966 said:
I'm sure there is proper software to do this, just never got round to finding/learning a package!

You kidding? I love doing this all by hand. Software doesn't teach you theory.

Cookie · Apr 3, 2021

pc1966 said:
OCPD
Zs (from OSG)
Zs No heating (Ohm)
R1+R2 (mOhm/m)
10A BS88-2 fuse
3.7
4.63
3.61
10A B-MCB
3.5
4.38
3.42
16A B-MCB
2.2
2.75
2.15
16A BS88-2 fuse
1.9
2.38
1.86
10A C-MCB
1.75
2.19
1.71
SWA choice
R1
R2
R1+R2 (mOhm/m)
16mm 4C
1.15
3.1
4.25
16mm 5C
1.15
1.15
2.30
25mm 4C
0.727
2.3
3.03
25mm 4C + 16mm
0.727
1.15
1.88
25mm 5C
0.727
0.727
1.45

By any chance, would you know the max Zs for this 15 amp MCB?

https://library.industrialsolutions.abb.com/publibrary/checkout/DES-079?TNR=Time%20Current%20Curves%7CDES-079%7CPDF&filename=DES079.pdf

pc1966 · Apr 3, 2021

Cookie said:
By any chance, would you know the max Zs for this 15 amp MCB?

https://library.industrialsolutions.abb.com/publibrary/checkout/DES-079?TNR=Time%20Current%20Curves%7CDES-079%7CPDF&filename=DES079.pdf

Looking at the data sheet, and applying the UK regulation's approach but with USA value, it would be:

Max current for 0.4s trip time = 30 * In (slightly less, but that is marked on horizontal scale) so that = 30 * 15A = 450A
Assuming you are looking at the USA supply of 277V nominal line-neutral and that is +/-10% then the lowest supply is 277 * 0.9 = 249V
So max Zs = 249V / 450A = 0.55 ohms
As for the table I did earlier, I am not applying the typical 0.8 factor for hot cable resistance compared to cold design/test values.

That is about 4 times lower than the European style of B-curve MCB would require, and that puts you cable size up massively! Assuming copper and matching CPC, that is no more than 0.27 ohms per conductor, and over 1280m that is 0.215 mOhm/m so you are looking at 3/0 AWG!

Realistically if you are faced with that sort of a breaker curve I would be looking at putting in a high quality RCD of 300mA or even up to 5A trip point as it would deliver far faster disconnection in the event of an earth fault, not having inrush trip risk, and allowing a sane cable size based on 5% VD of something like #5 AWG copper.

pc1966 · Apr 3, 2021

As mentioned before, two RCD in series could be used for "no single point of failure" and even at the $1k MCCB style pricing it is going to work out cheaper than that extra copper (and the time & effort to install and terminate it).

Cookie · Apr 3, 2021

pc1966 said:
Looking at the data sheet, and applying the UK regulation's approach but with USA value, it would be:

Max current for 0.4s trip time = 30 * In (slightly less, but that is marked on horizontal scale) so that = 30 * 15A = 450A

Assuming you are looking at the USA supply of 277V nominal line-phase and that is +/-10% then the lowest supply is 277 * 0.9 = 249V

So max Zs = 249V / 450A = 0.55 ohms

As for the table I did earlier, I am not applying the typical 0.8 factor for hot cable resistance compared to cold design/test values.

That is about 4 times lower than the European style of B-curve MCB would require, and that puts you cable size up massively! Assuming copper and matching CPC, that is no more than 0.27 ohms per conductor, and over 1280m that is 0.215 mOhm/m so you are looking at 3/0 AWG!

Realistically if you are faced with that sort of a breaker curve I would be looking at putting in a high quality RCD of 300mA or even up to 5A trip point as it would deliver far faster disconnection in the event of an earth fault, not having inrush trip risk, and allowing a sane cable size based on 5% VD of something like #5 AWG copper.

Pure awesome! Though I would use -5% instead of -10% but it works either way.

Yup, I noticed that two. US breakers keep giving me a much lower Zs which kept throwing me off. Still confuses me to be honest. Basically most US breakers are a type E of F, if such a category existed after type D lol.

3/0 is a massive amount of copper. I'm basically calculating voltage rings around the post in hopes of being able to get away with a longer disconnecting time based on lower hand to foot voltage than 138 volts:

For your calcs, are you using 30*C or 75*C for the AWG wire system?

pc1966 · Apr 3, 2021

Cookie said:
Pure awesome! Though I would use -5% instead of -10% but it works either way.

The UK voltage tolerance is +10% / -6% but the calculations are actually on 0.95 so -5% of nominal. However Aus/NZ seems to use the nominal cases so is more "optimistic".

What is actually applied would depend on local code/regulations but I have no idea of the details of USA policy.

Cookie said:
Yup, I noticed that two. US breakers keep giving me a much lower Zs which kept throwing me off. Still confuses me to be honest. Basically most US breakers are a type E of F, if such a category existed after type D lol.

That seems to be the case, though to be fair the EU range of MCCB also come with that style (basically fixed magnetic trip point for the family, but different thermal curves) as well as ones more like MCB with the "instant" magnetic point being at a fixed ratio to 'In'.

As well as the fancy but expensive ones with adjustable electronic trip up to the very high energy-limiting magnetic "last resort" trip.

Cookie said:
3/0 is a massive amount of copper. I'm basically calculating voltage rings around the post in hopes of being able to get away with a longer disconnecting time based on lower hand to foot voltage than 138 volts:

With local rods you have no real hope of safe disconnection or protection, unless very deep and wet. Having a CPC is far safer but adds cost, though if you can use SWA armour and still disconnect it is not really making a difference to project cable cost (assuming armoured is used).

But from above it looks like an RCD is needed in any case to deal with that sort of length, even though it is not an RCD for direct-touch shock itself.

Cookie said:
For your calcs, are you using 30*C or 75*C for the AWG wire system?

I assume 30C (i.e. cold, as measured), just got the AWG size & resistance from Wikipedia!

Cookie · Apr 3, 2021

Agreed.

Can you post the link to the 30*C AC resistance? The NEC is only giving me 75*C.

pc1966 · Apr 3, 2021

Cookie said:
Agreed.

Can you post the link to the 30*C AC resistance? The NEC is only giving me 75*C.

I used this:

American wire gauge - Wikipedia

en.wikipedia.org

Looking at details it says 68F = 20C which is room temperature around my part of the world.

Also probably they are DC, starts to become a source of error for large conductors as skin effect kicks in.

telectrix · Apr 3, 2021

personally i'd be quoting at least less than half the price. how?...... fit gas lamps. no volt drop or disconnection issues then.

pc1966 · Apr 3, 2021

Or solar powered ones

Soi disant · Apr 3, 2021

pc1966 said:
Also probably they are DC, starts to become a source of error for large conductors as skin effect kicks in.

At 50/60Hz??

telectrix · Apr 3, 2021

if digging in for 1 lamp post, why not fix 2 and stickk a propeller on the other one. wind power.

pc1966 · Apr 3, 2021

Soi disant said:
At 50/60Hz??

For large conductor yes, but its only about 20%-ish for 1000mm^2 so usually can be ignored. But also the inductive term starts to become more significant relative to the DC R1+R2, but even that is only about 10%-ish for 150mm cables.

Cookie · Apr 3, 2021

pc1966 said:
For large conductor yes, but its only about 20%-ish for 1000mm^2 so usually can be ignored. But also the inductive term starts to become more significant relative to the DC R1+R2, but even that is only about 10%-ish for 150mm cables.

Oh man, you guys have some serious brains!

Cookie · Apr 3, 2021

pc1966 said:
I used this:

American wire gauge - Wikipedia

en.wikipedia.org

Looking at details it says 68F = 20C which is room temperature around my part of the world.

Also probably they are DC, starts to become a source of error for large conductors as skin effect kicks in.

I should know about this link! ????

Anyway, do you know of a table taking skin effect into account? For a while I've desperatly been searching for an AC resistance at 30*C.

Cookie · Apr 3, 2021

I take this to mean an RCD for disconnection is a no-no?

pc1966 · Apr 4, 2021

Cookie said:
I should know about this link! ????

Anyway, do you know of a table taking skin effect into account? For a while I've desperatly been searching for an AC resistance at 30*C.

At AC power frequencies the skin depth in copper is about 9mm, so any conductor that is less than around 20mm diameter it has quite a small effect. Somewhere there will be equations to allow its computation for different conductors but most folks in engineering take the easy route of getting a feel for when you can ignore something, and ignoring it.

For example it is unlikely you will know any AC system parameter to better than 5% uncertainty, let alone 1%, so once effects get down to the 5%-ish region they can be often ignored for most purposes.

The temperature coefficient for copper is around 0.00393 so each 10C increase in temperature has a 3.93% increase in resistance. Which is about 4%. Which is why the UK regs have a 0.8 factor for going from measured at 20C to Zs value working at 70C.

Yes, our regs do have detailed formulae for non-standard cases such as measurement very cold/hot and/or operation at unusually high temperatures, but that vast majority of industrial cases are just like commercial and domestic in the 20C test & 70C assumed max operation is perfectly applicable.

In terms of generic AC values we have tables in the regulations with the values to assume. For example this if for single cables in various configurations giving both the DC and AC "resistance" values (more correctly the voltage drop due to cable impedance):

Soi disant · Apr 4, 2021

pc1966 said:
For large conductor yes, but its only about 20%-ish for 1000mm^2

Just a tad bigger than I'm used to thinking about.... ?

pc1966 · Apr 4, 2021

Cookie said:
I take this to mean an RCD for disconnection is a no-no?

No, the UK regulations generally do not specify how you achieve disconnection, only the times, etc, needed for safety. But they do list approved device standards for doing so (e.g. BS / EN ones on fuses, MCB, RCD, etc).

That section you quoted (411.3.2.5) has been updated but really it was a clause to deal with cases where the usual fault disconnection (OCPD or RCD) are not feasible so other means of protecting against shock need to be considered (supplementary bonding, etc).

The note about "disconnection may be required for reasons other than protection against electric shock" is (I think) to point out you might need it for overload/fire protection anyway, even if that fails to meet the maximum disconnection times discussed in that section about shock protection.

But beyond that, there are other rules permitting no OCPD or special cases when disconnection might be more dangerous, e.g. support electromagnet in factory, etc.

pc1966 · Apr 4, 2021

pc1966 said:
In terms of generic AC values we have tables in the regulations with the values to assume. For example this if for single cables in various configurations giving both the DC and AC "resistance" values (more correctly the voltage drop due to cable impedance):

If you look closely at that table you can see where the skin effect comes in by comparing the DC volt drop with the resistive term 'r' in column 5, but more generally it is the total impedance 'z' that would be used for volt drop if dominated by the cable.

Though close to a transformer, etc, where reactance is also a big factor you would be separately adding the r & x terms and finally computing z = sqrt(r^2 + x^2) to get your impedance.

Soi disant · Apr 4, 2021

pc1966 said:
At AC power frequencies the skin depth in copper is about 9mm, so any conductor that is less than around 20mm diameter it has quite a small effect.