Ohm value of transformer

[Cookie]

For those of you who work with onsite MV-LV transformers, does anyone know the line to earth ohm value of a transformer? I'd like to include that in the loop impedance path. Similarly when stepping down LV to LV.

https://imgur.com/a/PIsy0nt

View: https://Upload the image directly to the thread.com/a/PIsy0nt

https://imgur.com/JwUM0fv

View: https://Upload the image directly to the thread.com/JwUM0fv

 

[pc1966]

bit-shift? over flow? instruction cycles?

Unless you write DSP software (or need to approximate things on a very small integer microprocessor) you don't need to worry about it!

How does BS7671 tackle the issue of say of including a transformer in loop impedance calculations?

In the vast majority of cases the "wiring regulations" only deal with two effects:

• Voltage drop (L-N)
• Fault current (L-N and L-E)

Usually you simply have a value for PSSC & PFC at the source on an installation and accept it (or measure it if already installed) and from those you can set limits for final circuits to meed VD limits and to clear faults (and along the way not to blow up breakers).

If you get a copy of the IET's "Electrical Installation Design Guide" (ISBN 978-1-78561-471-2 for the 4th edition sitting here on my desk just now) it has some worked examples in chapter 6.

Again as this is looking at LV supply to final installations (not the HV or DNO's side) they are only really looking at an earthed star-point transformer.

An as aside, the IET book shows the "typical" let-through energy I2t of a fuse and a breaker in Figure 4.1, but the breaker's cure is somewhat over-optimistic! Compare it with the real world MCB characteristics from, say the Hager commercial catalogue here:

Hager Brochures, Catalogues & Documentations | Hager UK

Discover all the product documentation you need in our brochures & catalogues! Download what you need in one click straight from Hager Official website!

hager.com

Page 96 for MCB let-through curves.
Page 127 for MCCB curves

11.8% is good enough for me.

Increase the final number by 11.8% for PSC, decrease the final number by 11.8% for max disconnection times.

Or does assuming 70*C-75*C conductor temp for R work out enough for loop impedance?

No! Just compute it the obvious way:
Z = sqrt(R^2 + X^2)

For all numeric range most folk work with that is perfectly good (again, not an issue for this forum, but you can have issues of the square terms overflowing/under-flowing so doing Pythagoras correctly is subtly more difficult).

 

[Cookie]

Unless you write DSP software (or need to approximate things on a very small integer microprocessor) you don't need to worry about it!


In the vast majority of cases the "wiring regulations" only deal with two effects:

• Voltage drop (L-N)
• Fault current (L-N and L-E)

Usually you simply have a value for PSSC & PFC at the source on an installation and accept it (or measure it if already installed) and from those you can set limits for final circuits to meed VD limits and to clear faults (and along the way not to blow up breakers).

If you get a copy of the IET's "Electrical Installation Design Guide" (ISBN 978-1-78561-471-2 for the 4th edition sitting here on my desk just now) it has some worked examples in chapter 6.

Again as this is looking at LV supply to final installations (not the HV or DNO's side) they are only really looking at an earthed star-point transformer.

An as aside, the IET book shows the "typical" let-through energy I2t of a fuse and a breaker in Figure 4.1, but the breaker's cure is somewhat over-optimistic! Compare it with the real world MCB characteristics from, say the Hager commercial catalogue here:

Hager Brochures, Catalogues & Documentations | Hager UK

Discover all the product documentation you need in our brochures & catalogues! Download what you need in one click straight from Hager Official website!

hager.com

Page 96 for MCB let-through curves.
Page 127 for MCCB curves


No! Just compute it the obvious way:
Z = sqrt(R^2 + X^2)

For all numeric range most folk work with that is perfectly good (again, not an issue for this forum, but you can have issues of the square terms overflowing/under-flowing so doing Pythagoras correctly is subtly more difficult).

Alright, but my understanding is that transformer reactance will change the numbers in reality, so Z = sqrt(R^2 + X^2) is not technically accurate for mains and sub board faults.

I will use your equation of 1/2 the smaller number for now.

But still struggling to find the X value of trafos- I guess this is my main concern now.
[automerge]1590141316[/automerge]

You didn't, you converted all the sections from R and X into individual Z values, and then added the vector/phasor values using their magnitudes only.

Pc is stating the same as me, add all the Resistances to each other, and all the Reactances to each other and THEN convert to the Z value as sqrt(R^2 + X^2).

Ah! Ok, that makes more sense now!

The resistance gives a voltage drop in phase with the current, but the reactance gives a voltage drop at 90degree to the current

Agree. And a DC off set during a fault if I'm on the right tack?

The maths for the most part is just simple stuff, for the lv side of transformers you would just use normal resistance and reactance addition that is basic ac theory you would have already done during basic training.

No basic training in this regard. As I said, electricians in the US do not calculate earth fault loop impedance as its not required or mentioned in the NEC. We do not carry a multi function tester and most of us have no clue what Ze, Zs or R1+R2 even means. A lot of us couldn't even tell you the per foot impedance of #12 (3.31mm2) or #10 (5.26mm2) even though we run yards of it every day.

We only size the wire based on code rules governing ampacity, terminal temps (60*C vs 75*C) and de-rating factors. We consider voltage drop after that fact, based only on our discretion. Voltage drop limits are not mandated by NFPA 70. We can legally have 20% voltage drop to any load if we deem it acceptable. Further there are no disconnection time requirements present in the NEC. If were to run 1000+ feet of 2.08mm2 wire to a swimming pool or light post we could (and can) legally do so...

In fact up until 10 years ago most US electricians mistakenly believed that a ground rod could trip a standard thermal magnetic breaker which resulted in countless tragic electrocutions and injury law suits- one example of a fence that kept becoming energized:

View: https://youtu.be/C5EiSEtxRKU?t=620


We now know better- ground rods don't clear faults- however that is only half the story. A very small minority is starting to realize that having an equipment grounding conductor isn't a guarantee. Things like railing, pools, fences, light poles, water slides, farm equipment, ect is remaining energized because the loop impedance of the circuit is to high to trip an ordinary breaker. If electricians knew about loop impedance and disconnection times these incidents would not be happening.

Lastly this was also what got AFCIs into the NEC, when it was theorized that disconnection times in excess of 6 half cycles were leading to fires. Older residential breakers had a magnetic pickup of at least 20x, some without it entirely.

Hence why I'm inquiring here with questions that are rather elementary and obvious to IEC folks. I mean no irritation hard headiness

You only need to use symmetrical components if you want to calculate the unbalanced fault levels across transformers including the high voltage network.

They would not normally be used. Typically you would just accept the fault current for an earth fault direct on a transformer is the same as the three phase fault current, however you specifically included a network with the HV included.

Good to know. I'm willing to assume infinite for the MV network.

However, I'd like to know if I need sequence components for a setup where 480 volts is going to 120/208Y, as is typical in buildings.

For a transformer, all the sequence components are fundamentally the same as the voltage impedance. The only difference is the zero phase sequence one, and the inclusion of this along with how it should be included in the calculation is dependent upon the winding group and connection.

This can either be worked out by understanding the connections or by reference to sequence diagrams such as these:View attachment 58306
View attachment 58307

In the US the calculation processes are described in the IEEE red book and the IEEE buff books (at least they are from my mid '80s versions)

I would assume they have typical values for the specific equipment you use there.

Thanks! But they are missing one of our most common pole top transformers

https://imgur.com/a/PAkh5XJ

View: https://Upload the image directly to the thread.com/a/PAkh5XJ


T-T, which gives 3 phase power via only two core and coils- lighter and less material over delta-wye from what I'm told.

I know, we like to be different!

 

[Julie.]

Agree. And a DC off set during a fault if I'm on the right tack?

Yes, it's a DC offset - starting high and decaying down to zero - like the discharge characteristic of a capacitor.

In the US you have a simplified method of calculating the first cycle and sustained fault currents called the direct method - I assume it still exists in ANSI/IEEE Std 242-xxxx - it was in 242-1986 when I was last forced to use it!

No basic training in this regard. As I said, electricians in the US do not calculate earth fault loop impedance as its not required or mentioned in the NEC.

In terms of basic maths I was referring to standard circuit theory - you would have done something along the lines of calculating the impedance of any circuit containing R, X(L), & X(c).

But still struggling to find the X value of trafos- I guess this is my main concern now.

…...

T-T, which gives 3 phase power via only two core and coils- lighter and less material over delta-wye from what I'm told.

I know, we like to be different!

From your photo the impedance is 1.9%, to be sure you would need to obtain the X/R from the manufacturer, in lieu of this I would just take the value from ANSI/IEEE Std 242 - which is around 1.8; given Z and the X/R ratio you can calculate both the X and R for the transformer.

Yeah, we don't use the Tee - Tee connection over here - as a connection it has some advantages, lower iron loss, and smaller construction, the disadvantages are internal due to the interleaving required on the windings, but for us the main issue is how it fits into the greater network - a single phase fault translates to an unbalanced fault across all three phases on the primary, we tend to use Delta - Star, in this case a single phase fault this translates to a circulating current in the primary winding and a balanced fault on the HV.

We tend to run more balanced HV networks - no distributed neutral, so the Tee-Tee just isn't popular
[automerge]1590148405[/automerge]
Oh, forgot to mention:

With a Tee-Tee type connection when you have a phase to earth fault, it isn't the whole of the winding that's in the circuit, so it has lower values of X & R so for example in your case 1.9% with an X/R of 1.8 would give X=1.66% and R=0.923% which you would use for three phase faults.

However for single phase faults then ANSI/IEEE Std 242 indicates typical values are 0.6x X and 0.75x R for a 75kVA transformer

So for a single phase earth fault - you ought to use
1.66% for X(Three phase)
0.923% for R(Three phase)
0.6 x 1.66% = 0.996% for X(single phase)
0.75 x 0.923% = 0.692% for R(single phase)

I would suggest investing in the latest buff book if you are thinking about doing these calculations!

 

[Cookie]

Can I ask about conductors in parallel? What is the proper way to calc impedance for them?

 

[pc1966]

Can I ask about conductors in parallel? What is the proper way to calc impedance for them?

For resistance it is simple - just the usual parallel connection as as you would make them identical runs as far as practical (to share current evenly) the R term is going to be R/2 or R/3 for 2 or 3 in parallel, etc.

However, the inductive term depends on the layout so it is not going to be so simple. For example if you have a bunch of conductors in parallel and significantly separated from another bunch then the inductance won't change by much (i.e. no parallel reduction in X).

But more generally I don't know of any simple formulae for those cases, but probably @Julie. will know from her past work on high power stuff.

 

[Cookie]

For resistance it is simple - just the usual parallel connection as as you would make them identical runs as far as practical (to share current evenly) the R term is going to be R/2 or R/3 for 2 or 3 in parallel, etc.

However, the inductive term depends on the layout so it is not going to be so simple. For example if you have a bunch of conductors in parallel and significantly separated from another bunch then the inductance won't change by much (i.e. no parallel reduction in X).

But more generally I don't know of any simple formulae for those cases, but probably @Julie. will know from her past work on high power stuff.

Reactance has me again. Some parallel runs will be in the same conduit.

 

[Julie.]

No there isn't an easy answer unfortunately if you have two trifoil cables in parallel then you would treat the whole lot as two parallel paths of Resistance-Reactance in parallel.

If you can calculate using complex numbers it would be the normal equation R+JX = 1/((/R1+jX1)+(1/R2+jX2)) - in the same way you would just use R=1/((1/R1)+(1/R2)) for two resistors in parallel - this is the same thing, only the maths required isn't as simple as it is for resistance alone.

You cannot work out the resistance and the reactance separately.

If you are running single conductors then you can't do this as they influence each other.

The IEEE buff book has a simplified "guesstimate" method using a Reactance factor depending on the size of conductor and Conduit which allows you firstly to calculate each cable's Reactance based on the non-conduit value so for a conductor 250MCM (or kcmil) this would change the overall reactance by a factor of 1.149 , the Reactance itself though is made up of two parts - the self reactance Xa - which is available in a set of tables - around 11.45mOhm per 100ft for 250MCM - and the mutual reactance Xb which depends on the distance between them - say -7.95 for 0.4" spacing so Xa + Xb would be ~ 3.5 x 1.149 .

We had to do this once, a rather large generating station in India had a long run between the generator and the GSU transformer - I think it was around 800 MVA at ~22kV, so we had to run multiple parallel single core cables - the problem is if you think about it it would go something like L1-L1-L1-L1-L2-L2-L2-L2-L3-L3-L3-L3 for the four cables per phase over three phases - the problem is that only the centre two cables per phase are the same (one same phase to the left and one to the right) all the others either have no cable to one side, or have a cable of a different phase - so they would have a different reactance to each other! and therefore would not share the current equally!

We had to calculate the reactance from first principles - it gave such an unbalance that we had to transpose the conductors per phase - four times such that each one cable of each phase was next to the alongside phase once, the free air once (or the other phase) and one of the centre two twice - this caused a balanced reactance between conductors and therefore balanced sharing of currents!

I was much younger I am not sure I could do it now!

 

[Vortigern]

Wow something just went over my head! Reading (most) of the above reminds of theory we did but of course have never had to use much in the normal work environment hence such understanding of the kind of thing you are talking about tends to atrophy as the years go by. Interesting though about the answer to the power station in India, transposing the phase cores to achieve phase balance. Who thought of that Julie? Ingenious.

 

[Julie.]

Wow something just went over my head! Reading (most) of the above reminds of theory we did but of course have never had to use much in the normal work environment hence such understanding of the kind of thing you are talking about tends to atrophy as the years go by. Interesting though about the answer to the power station in India, transposing the phase cores to achieve phase balance. Who thought of that Julie? Ingenious.

It's actually a fairly standard technique, if you look along very long transmission lines, you will find odd towers where the conductors are transposed so the phases end up of similar impedance.

Long before I was a puppy!

Transposition tower:

 

[pc1966]

It's actually a fairly standard technique, if you look along very long transmission lines, you will find odd towers where the conductors are transposed so the phases end up of similar impedance.

A broadly similar idea to Litz wire.

I have not see it since radios built pre-60s but a quick search shows there are still folk making it!

 

[Cookie]

Transposition towers are semi common in the US, so I am familiar with that concept.

But going back- 6-500 kcmils in a 4 inch steal conduit. Or 32 600kcmil sets in 8 separate PVC conduits.

If anyone is curious, here are US cable sizes in mm2:

https://imgur.com/a/lZvLk2D

View: https://Upload the image directly to the thread.com/a/lZvLk2D


500kcmil = 253mm2

600kcmil= 304mm2

 

[pc1966]

Transposition towers are semi common in the US, so I am familiar with that concept.

But going back- 6-500 kcmils in a 4 inch steal conduit. Or 32 600kcmil sets in 8 separate PVC conduits.

The sort of places I have seen conductors that large tend to have them on open trays, probably for ease of installation as much as for air cooling. But equally it is not really my area.

 

[Cookie]

The sort of places I have seen conductors that large tend to have them on open trays, probably for ease of installation as much as for air cooling. But equally it is not really my area.

In the EU I've heard that being the norm, but in North America conduit- (either steal or PVC)- rains supreme.

Here is an example of a chiller breaker feeding two parallel runs in separate conduit:

https://imgur.com/JPeRNMq

View: https://Upload the image directly to the thread.com/JPeRNMq


US power companies deliver 120/208Y to large buildings as much as 277/480Y, so high current circuits (and their up-size for voltage drop) is very common. That and NEC article 220 load calcs tend to make feeders and services near double their actual peak current draw.

 

[pc1966]

It looks like they keep the 3 phases in each conduit, presumably to reduce magnetic interaction a bit.

However the earth wires look astonishingly thin in comparison so I presume they are using a RCD for earth fault protection!?

 

[Cookie]

It looks like they keep the 3 phases in each conduit, presumably to reduce magnetic interaction a bit.

Good eyes- correct. IF the conduit is steal it must be done that way.

The only typical exception is PVC conduit when installed between the padmount transformers and the service disconnecting means.

However the earth wires look astonishingly thin in comparison so I presume they are using a RCD for earth fault protection!?

No RCD. We struggle earth fault loop impedance lol.

Earth wires are sized based on NEC Table 250.122:

https://imgur.com/YQzWJKS

View: https://Upload the image directly to the thread.com/YQzWJKS


These values are basically calculated off the adiabatic method... the bare minimum size that won't cause the insulation to melt off during a ground fault.

So translating:

15 amp fuse or breaker requires 2.08mm2 earth wire

20 amp fuse or breaker requires 3.31mm2 earth wire

Here is where it gets rather small:

60 amp = 5.26mm2

100 amp = 8.36mm2

200 amp = 13.30mm2

300 amp = 21.15mm2

400 amp = 26.67mm2

500 amp = 33.62mm2

600 amp = 42.41mm2

800 amp = 53.49mm2

1000 amp = 67.43mm2

 

[pc1966]

These values are basically calculated off the adiabatic method... the bare minimum size that won't cause the insulation to melt off during a ground fault.

But if (as you say) they don't really check Zs, how do they know the let-through I2t for the adiabatic calculation?

 

[Cookie]

But if (as you say) they don't really check Zs, how do they know the let-through I2t for the adiabatic calculation?

Can you educate me on this concept? I have no idea how let-through in relations to Zs works in regards to EGCs.
[automerge]1590333865[/automerge]
If curious in regards to me starting this thread I ask based on 250.4 (A) (5) and 250.4 (B) (4) as indicated in Table 250.122:

https://imgur.com/t5RbR6p

View: https://Upload the image directly to the thread.com/t5RbR6p


Code mandates that impedance be low enough to trip the breaker- but does not specify the maximum time or method to determine that hence why I am turning to IEC 60364-4-41 Table 41.1 and BS7671 as a guide.

I know asking about IEC standards while simultaneously working with NFPA-70 is creating some confusion so I hope I can clear that up. Plus its fun learning new stuff

 

[pc1966]

Basically if you have a given fault, and lets assume it is a dead short at the end of your circuit, then if you know Zs you can find the lowest PFC from the minimum supply voltage as:
PFC = Umin / Zs
From this level of fault current you then look up the current-time curve of the breaker or fuse and that gives you the time to disconnect at that level. Once you know the fault current and corresponding time you have the let-through I2t "energy" and that allows you to compute what it will do to the cable.

The usual assumption is the adiabatic case - then the input heat energy is over a short period of time so the out going heat energy can be neglected, which is reasonable for a short circuit type of fault - and so if you assume a closed thermal system, then a given temperature rise depends on the material constant and input energy.

This is normally simplified even more by assuming only a couple standard conductors (Cu, Al, and Fe) and a few common thermal limits (e.g. for thermoplastic and for thermosetting insulation) as a table of a few fudge-factors to use.

It might be easier to give an example. Say you have 63A fuse (common for domestic in UK) and you find your Ze value was 0.35 ohms (typical upper limit for TN-C-S) and you wanted to size a copper earth wire. Lowest fault current would be 95% of U / Ze

PFC = 0.95 * 230 / 0.35 = 624A

Here is a typical fuse curve:

If you look up 624A prospective current for the 63A curve (as far as practical) you see the corresponding time is around 0.4 seconds, so we can compute our let-through energy as:
I2t = 624^2 * 0.4 = 156k (A2s units)

Looking at the values for copper and say 30C initial and thermoplastic (70C cable) we have k = 143 so we can size our earth using:
S >= sqrt(I2t) / k = 395 / 143 = 2.76 mm^2 CSA minimum

If we had Ze = 0.7 ohms (double the fault impedance) then our current is half at 312A but then our disconnect time is 7s hence I2t = 312^2 * 7 = 681k (A2s units) which is about 4.4 times larger and now our minimum earth conductor is given by:
S >= sqrt(681E3) / 143 = 5.77 mm^2 CSA

For a fuse (at least our BS88 ones) they limit the I2t let-through and so worst-case is at lowest fault currents when disconnection takes a long time. Of course once you get in to the ten seconds or more the adiabatic assumption no longer holds so eventually you end up with a steady-state current carrying requirement. But for a fuse the least fault energy is at max PFC.

For a breaker it is more complicated, and you have a massive difference between faults that hit the "instantaneous" trip and those that don't. Also MCB and MCCB are not fault-limiting to the same degree as fuses, so as PFC increases from that trip point you see a moderate increase in let-through energy.
[automerge]1590337182[/automerge]
Just to add - that is why you do the lowest PFC (minimum supply voltage) for the adiabatic check as it gives the longest disconnect time and hence usually the biggest let-through energy.

 

[Cookie]

I wish I could give this reply 100 likes. Love it!

I think I might use this to compare our tables, especially with motor OCPD sizing rules.

 

[Cookie]

But how does this work out with sputtering or arcing faults?

 

[pc1966]

But how does this work out with sputtering or arcing faults?

Poorly I suspect.

It is always a bit of a gamble under-sizing the CPC on the assumption that any fault is a clean hard one that trips the OCPD as quickly as you assumed. For high current circuits an arc is likely, but that should mean you deduct tens or even a hundred volts from the supply to correct for the imperfect fault.

In the UK we don't do so much on arc-fault calculations, which is a big thing in the USA. Not sure if it is down to our systems generally having fast-disconnect or your systems having much higher PFC creating the greater risks, or just we don't see enough thermal injuries compared to shock to push it up the priority list.

But returning to CPC calculations, usually I would be conservative and go for the higher size (towards the phase/2 "rule-of-thumb") to help cope with such events. Also we do see RCD being used for fire control, but not as common generally (more of a thing for agricultural buildings, etc).

For completeness here are some example let-through curves for Hager B-curve MCB. As a point of reference the high fault limit for a 63A BS88 fuse is around 18k A2s (so at 10kA fault less than a 6A MCB) but from above you see the fault energy in the region from "instantaneous" trip to a few kA is much better in the MCB case:

 

[Cookie]

FWIW, UL found that if a circuit's minimum bolted fault current was 125% greater than the breaker's magnetic pickup threshold the circuit was protected from parallel arc faults.

The UK's wiring regs already provide parallel arc fault protection in that values are adjusted by 80%:

View: https://youtu.be/5rHXBD9UYg4?t=365

Remember that wiring under full load wire doesn't actually reach 70*C.

On the other hand we don't assure disconnection times by any measure- so it was much easier for manufacturers to sell AFCIs to the US. It was EU breakers that actually got the ball rolling:

https://paceforensic.com/pdfs/Circuit_Breakers_The_Myth_of_Safety.pdf

15 and 20 amp breakers in general got a magnetic pickup of around 150 amps:

http://paceforensic.com/pdfs/newsletter/KeepingPace-15.pdf

 

[pc1966]

FWIW, UL found that if a circuit's minimum bolted fault current was 125% greater than the breaker's magnetic pickup threshold the circuit was protected from parallel arc faults.

The UK's wiring regs already provide parallel arc fault protection in that values are adjusted by 80%

Yes, that would be reasonable for a clean fault.

Remember that wiring under full load wire doesn't actually reach 70*C.

On the other hand we don't assure disconnection times by any measure- so it was much easier for manufacturers to sell AFCIs to the US. It was EU breakers that actually got the ball rolling:

15 and 20 amp breakers in general got a magnetic pickup of around 150 amps:

Some systems run closer to max temperature than others, but unlikely in most cases for sure. I remember seeing that paper before, really quite a surprise. Do USA breakers now have something closer to the EU style of 3-5 In for the magnetic trip?

A lot of the Zs requirements are being eroded by widespread RCD use, but I am always a bit paranoid about electronics failing so prefer to know it would clear OK on the MCB trip alone!

 

[Cookie]

Yes, that would be reasonable for a clean fault.

UL says it will take care of a parallel arc fault. Our AFCIs stop looking for parellel arcs below 75 amps as this was the lowest short circuit value determined (theorized really) to be found in a US home's in-wall wiring- after research was conducted on dwelling properties by UL.

Some systems run closer to max temperature than others, but unlikely in most cases for sure. I remember seeing that paper before, really quite a surprise.

A lot of people are surprised or straight up don't believe me when I compare the NEC with VDE, BS7671, ect.

The thing is the NFPA is purely a reactionary, litigation responsive document. Change typically does not come about unless something can be proven well beyond a doubt as presenting a hazard- usually after one or more documented tragedies.

For example, the NEC specifically added verbiage forbidding the earth (TT) as being used as an effective ground fault current path and re-word the definitions of grounding, bonding and "effective ground fault current path" after numerous incidents where ground rods alone failed to clear a fault.

Still then manufacturer reps sitting on code making panels have the final word, often voting on specific products instead of rules rooted in electrical theory.

Do USA breakers now have something closer to the EU style of 3-5 In for the magnetic trip?

Its around 6-10x for today's for single pole 15 and 20 amp breakers. Originally the idea was to have all North American 15 and 20 amp single pole breakers trip at 75 amps, but discovered such a threshold would result in nuisance tripping from appliance inrush. This resulted in electronic AFCIs being developed, tested and then mandated by NFPA 70.

The thing is NFPA-70 could have just mandated loop impedance calcs which would have done the job just as well with today's 6-10x breakers which aren't causing any problems, even with highly inductive loads like refrigerators, washing machines, and microwave ovens (which typically lack inrush suppression in North America).

A lot of the Zs requirements are being eroded by widespread RCD use, but I am always a bit paranoid about electronics failing so prefer to know it would clear OK on the MCB trip alone!

Never trust electronics. Your thinking is correct. An RCD is nothing more than a backup CPC- and a rather wonky one.

The 2020 NEC just mandated surge protection devices on all dwelling unit services and has been doing so for several code cycles in regards to emergency circuits. Under 2020 practically every single dwelling unit circuit requires either an AFCI, GFCI, or dual function GFCI/AFCI breaker.

GFCIs were mandated in the US due to the large number of two prong metal framed tools which were becoming live while in use. GFCI latter expanded to other scenarios where EGCs were being compromised like cord and plug connected pool pump motors.

EU has a variety schuko plugs which can mate with each other but not always with an earth connection. Thus RCDs were a good idea in the EU.

The UK was able to hold off into the 90s due to the fact all none double insulted tools and appliances were equipped with an EGC since WWII, and a missing earth pin would not open the socket shutters and not likely be broken in the first place due the robust design of UK plugs.

 

[pc1966]

The thing is the NFPA is purely a reactionary, litigation responsive document. Change typically does not come about unless something can be proven well beyond a doubt as presenting a hazard- usually after one or more documented tragedies.

Sadly that is not a surprise to folk on this side of the pond.

For example, the NEC specifically added verbiage forbidding the earth (TT) as being used as an effective ground fault current path and re-word the definitions of grounding, bonding and "effective ground fault current path" after numerous incidents where ground rods alone failed to clear a fault.

It has long been the case here that a TT setup would (to all practical purposes) need an RCD-style of device as the rod impedance is very unlikely to be low enough for operation of the OCPD.

Yes, up until probably the 70s they were the VOELCB style that were not particularly reliable in operation (not the units themselves, but the means of detection).

Its around 6-10x for today's for single pole 15 and 20 amp breakers. Originally the idea was to have all North American 15 and 20 amp single pole breakers trip at 75 amps, but discovered such a threshold would result in nuisance tripping from appliance inrush. This resulted in electronic AFCIs being developed, tested and then mandated by NFPA 70.

The thing is NFPA-70 could have just mandated loop impedance calcs which would have done the job just as well with today's 6-10x breakers which aren't causing any problems, even with highly inductive loads like refrigerators, washing machines, and microwave ovens (which typically lack inrush suppression in North America).

So they are similar to our C-curve MCBs (5-10 * In) then.

I guess in the UK we have the advantage of the ring & fused plug system, so our 32A B-breaker has a similar instantaneous trip point of 100-150A, but individual end appliances are often on a smaller fuse such as 13A for washing machine, 5A for may electronics devices like TVs, etc, and hence a lot less fault energy if they short out.

Never trust electronics. Your thinking is correct. An RCD is nothing more than a backup CPC- and a rather wonky one.

Here it is often referred to as 'additional protection' which is a good way of thinking about it - it is there for shock protection more than fault clearing.

But the regulations and a few design tools are using the RCD trip Zs as the design test, and not the OCPD requirement. Now in something like the TT incomer RCD case you have to accept that, but if it were down to me I would adjust the wording of the regs a bit to make it clear you should always design for the OCPD's Zs and only in justifiable cases where it is not feasible to meet that to then rely on RCD action in its place.

 

[Cookie]

Sadly that is not a surprise to folk on this side of the pond.

I'm just glad there is awareness of it.

It has long been the case here that a TT setup would (to all practical purposes) need an RCD-style of device as the rod impedance is very unlikely to be low enough for operation of the OCPD.

Yes, up until probably the 70s they were the VOELCB style that were not particularly reliable in operation (not the units themselves, but the means of detection).

Yup- though the high Z of earth rods was not readily known in the 70s.

So they are similar to our C-curve MCBs (5-10 * In) then.

Yes, similar. Two and 3 pole breakers are closer to a D curve in order to allow inrush on large multi-motor equipment.

If that is not enough we can also use Table 430.52 and increase the size of the breaker up to 250%, 175% for time delay fuses and 300% for none time delay fuses.

https://imgur.com/9Gh5YsY

View: https://Upload the image directly to the thread.com/9Gh5YsY


As an example 2.08mm2 copper line and earth can be protected with a 40 amp breaker.

I guess in the UK we have the advantage of the ring & fused plug system, so our 32A B-breaker has a similar instantaneous trip point of 100-150A, but individual end appliances are often on a smaller fuse such as 13A for washing machine, 5A for may electronics devices like TVs, etc, and hence a lot less fault energy if they short out.

Trust me, UL had fun studying your system when developing AFCIs. Damaged cords rather quickly blow plug fuses.

Here it is often referred to as 'additional protection' which is a good way of thinking about it - it is there for shock protection more than fault clearing.

But the regulations and a few design tools are using the RCD trip Zs as the design test, and not the OCPD requirement. Now in something like the TT incomer RCD case you have to accept that, but if it were down to me I would adjust the wording of the regs a bit to make it clear you should always design for the OCPD's Zs and only in justifiable cases where it is not feasible to meet that to then rely on RCD action in its place.

I'm with you on this. Plus you still have L-N faults.

 

[pc1966]

Yup- though the high Z of earth rods was not readily known in the 70s.

Oh I think the impedance of earth rods was known about in the 19th century!

Actually applying that to regulations and safe design, well...

As an example 2.08mm2 copper line and earth can be protected with a 40 amp breaker.

That is one of the marginal cases here, it would be acceptable only if the fault-clearing time is low enough to limit the energy. Which is back to Zs once more.

Generally we have two cases of related protection:

• Short circuit protection
• Overload protection

Short circuit protection is always required, with the exception of situations where the source is fundamentally current limited (e.g. some bell transformers, etc) or it is a short run and well protected (e.g. bus bar tap-off cables where a fault in that area is seen as unlikely)

Overload protection is not always required at the source end of a cable, as the load might provide that (if a fixed demand appliance, or cable terminates in a switch/fuse or a DB with a limited total current).

I'm with you on this. Plus you still have L-N faults.

Generally we treat that as above, and we look to Zs disconnection as worst-case.

That is not always true, as you might find parallel earth paths making PFC higher than PSSC, but if you have designed you L+E wires to meet disconnection times then L-N faults should be a slam-dunk (to borrow one of your phrases).

But going down the RCD path makes that less certain, though we do still have the volt-drop limit acting as another sanity check to cause fast disconnection under short conditions (e.g. If you are designing a 20A circuit with 5% max VD then your PSSC should be at least 20A / 0.05 = 400A at nominal voltage which is enough for even a D-curve breaker's instantaneous trip). But it is not as clearly defined as for the shock-protection case of L-E disconnection times, and is worded more in terms of conductor protection (and even for the L-E times, there is a note that faster than 0.4s/5s may be required for thermal reasons).

So in a sense we have all of the right bits in place, but personally I would like to see the OCPD requirement on max Zs pushed more in the RCD cases.

 

[Cookie]

Oh I think the impedance of earth rods was known about in the 19th century!

Actually applying that to regulations and safe design, well...

I'm not sure why it took 60 years for them to realize it. Its possible the constant use of the word "ground" "ground wire" "ground fault" the code gave rise to that myth... but I have no idea to be honest.

That is one of the marginal cases here, it would be acceptable only if the fault-clearing time is low enough to limit the energy. Which is back to Zs once more.

Generally we have two cases of related protection:

• Short circuit protection
• Overload protection

Short circuit protection is always required, with the exception of situations where the source is fundamentally current limited (e.g. some bell transformers, etc) or it is a short run and well protected (e.g. bus bar tap-off cables where a fault in that area is seen as unlikely)

Overload protection is not always required at the source end of a cable, as the load might provide that (if a fixed demand appliance, or cable terminates in a switch/fuse or a DB with a limited total current).

NEC is much the same way. It distinguishes between overload and short circuit protection. Motors and AC equipment are considered as having overload protection at the load end. Welders are allowed to be on smaller conductors relative to the OCPD rating (duty cycle exception). Circuits with fixed equipment can go the "next standard size up" in regards to OCPD selection. Same for calculated loads like sub-panels. 55 amp wire can go on a 60 amp breaker, ect.

In Canada (which uses the CEC, a close relative of NFPA-70) fixed electric heat circuits can take an OCPD 125% of the final circuit conductor rating.

The difference being no Z requirements.





Generally we treat that as above, and we look to Zs disconnection as worst-case.

That is not always true, as you might find parallel earth paths making PFC higher than PSSC, but if you have designed you L+E wires to meet disconnection times then L-N faults should be a slam-dunk (to borrow one of your phrases).

Borrow what you need Its the smallest favor I can re-turn.

But going down the RCD path makes that less certain, though we do still have the volt-drop limit acting as another sanity check to cause fast disconnection under short conditions (e.g. If you are designing a 20A circuit with 5% max VD then your PSSC should be at least 20A / 0.05 = 400A at nominal voltage which is enough for even a D-curve breaker's instantaneous trip). But it is not as clearly defined as for the shock-protection case of L-E disconnection times, and is worded more in terms of conductor protection (and even for the L-E times, there is a note that faster than 0.4s/5s may be required for thermal reasons).

So in a sense we have all of the right bits in place, but personally I would like to see the OCPD requirement on max Zs pushed more in the RCD cases.

You're way ahead of use. The thing is the NFPA knows that, gradually mandating GFCIs on all circuits... something we do not all agree with.