But the steel wires of an SWA do not really form a continuous cylinder enclosing the wires  - really there are air gaps, slots  and bunching and the magnetic effect is more like split tube with a distributed 10% gap, rather than welded conduit.

Thinking out loud is it were - most SWA armours strands seems to be 'spiralled' around the bedding - so if there's a continuous gap between any two strands, what we'd then have, in effect, might look like a current transformer, running the length of the cable - the line conductor forming the single turn primary and the armour forming a multi-turn secondary. What impact might that have on an earth fault current ? (either opposing or encouraging it depending on which way the strands are spiralled).

not that significant, at least at 50Hz, 

Again, just pondering. If a fault occurred at a peak in the a.c. cycle, the rate of change of current over time would be very great at that moment - far faster than would be expected with a simple 50Hz wave starting at a zero crossing point. Would that be similar to components of a much higher frequency - if only for a short instant? Probably not an issue if the protective device is a fuse that'll take quite a few cycles to blow anyway, but if it's a fancy circuit breaker that's trying to make the decision to trip in the first few milliseconds in order to discriminate with upstream devices, might things not quite add up as we'd hope?

   - Andy.

It does not need to spiral - the magnetic field needs to enclose the conductor, but the conductors need to be parallel (as they are in a conventional transformer.) It is this one turn transformer effect that causes the current dependent voltage drop between the ends of the AWA singles that leads us to criss-cross the armours at link boxes  1/3 and 2/3 of the way along in really high power kit.
In terms of the really fast component to the fault current (*), the load or fault impedance is  not the immediate determine factor if the fault comes on at time 't1' the source cannot 'know' to put more electrons in, until 't2' where the length of cable and the speed of light sets t1-t2.

Initially the current to voltage ratio is limited by the L to C ratio of the cable, its 'characteristic impedance' or 'surge impedance'

After that the current rises in a stepsise fashion, with a step every triple transit, reaching a final state where the current at both ends is set by the load. This is usually assumed to be after a few dozen triple transit times.

(some nice pictures here )

Mike

* here I mean short compared to the wavelength of the highest frequency h component of the waveform - approximately the radian period of  the fastest rise time ( so a  pulse with a 1usec rise time, has a highest frequency component of approx 2.pi MHz. A cable of 100m or so will exhibit triple transit times of order microsecond. A 'short' cable at 1MHz is therefore a few metres to tens of metres.
All cables, unless a DNO, are short compared to 50Hz....

It does not need to spiral - the magnetic field needs to enclose the conductor, but the conductors need to be parallel (as they are in a conventional transformer.) It is this one turn transformer effect that causes the current dependent voltage drop between the ends of the AWA singles that leads us to criss-cross the armours at link boxes  1/3 and 2/3 of the way along in really high power kit.
In terms of the really fast component to the fault current (*), the load or fault impedance is  not the immediate determine factor if the fault comes on at time 't1' the source cannot 'know' to put more electrons in, until 't2' where the length of cable and the speed of light sets t1-t2.

Initially the current to voltage ratio is limited by the L to C ratio of the cable, its 'characteristic impedance' or 'surge impedance'

After that the current rises in a stepsise fashion, with a step every triple transit, reaching a final state where the current at both ends is set by the load. This is usually assumed to be after a few dozen triple transit times.

(some nice pictures here )

Mike

* here I mean short compared to the wavelength of the highest frequency h component of the waveform - approximately the radian period of  the fastest rise time ( so a  pulse with a 1usec rise time, has a highest frequency component of approx 2.pi MHz. A cable of 100m or so will exhibit triple transit times of order microsecond. A 'short' cable at 1MHz is therefore a few metres to tens of metres.
All cables, unless a DNO, are short compared to 50Hz....