jbrameld said:

Does anyone support the theory that use of a separate CPC is no longer allowed, and

That wouldn’t be a bad thing if it were true!

I would suggest that the separate cpc with swa has been part of the furniture for many years. Done automatically by both consultants and contractors alike. 
My question would be why? I don’t believe there are too many circumstances where simple selection from table 54.7 would be breached and if it was, reference to the adiabatic would be the get-out. 
Agree that there may be times when a separate conductor might be unavoidable but for the vast majority of cases, it can be done without.

jbrameld said:

Does anyone support the theory that use of a separate CPC is no longer allowed, and

That wouldn’t be a bad thing if it were true!

I would suggest that the separate cpc with swa has been part of the furniture for many years. Done automatically by both consultants and contractors alike. 
My question would be why? I don’t believe there are too many circumstances where simple selection from table 54.7 would be breached and if it was, reference to the adiabatic would be the get-out. 
Agree that there may be times when a separate conductor might be unavoidable but for the vast majority of cases, it can be done without.

Coming back to the question at the top.  It all rather depends what you want to happen during a fault, and the touch voltage you are prepared to be exposed to at the far end while the fault clears.

In terms of operation of prompt ADS, there are indeed very few cases when the armour alone is not adequate. (ignoring some folks inability to connect to it in a solid and reliable manner)

But if fault currents are high, or there is a desire for a lower than normal touch voltage then firstly the armour in parallel with an internal CPC is more elegant, but cables with enough cores are not always readily available, in all sizes, so  the external  CPC version is then the next neatest solution.  The magnetics are not in your favour then, but the steel armour wires of SWA  do not form an unbroken ring of steel around the cores, but rather  a core with lots of air gaps, so it is not as bad as an unbalanced steel conduit - the problem the excess inductance is not accurately calculable in advance, nor is it constant between apparantly quite similar cable lay-ups. But again, at high enough currents the problem dissappears - a quick consideration of the currrnt needed to fully saturate the solid steel band case shows this.

(numbers for the rusty of memory - take the amps the middle - perhaps 100, and divide by the circumference around the SWA - say pi times diameter of the core bundle, so 6cm for a 2cm dia cable,  to get the H and convert to amps/metre
(in this case 100/ 0.06 ~  2000 amp -metres )    B = u0ur H. well u0 is 1.2 E-6 and uR for steels is in the range 500- 10000. But with air gaps of 10% , probably less than 50. and saturation occurs at say 1Tesla. )

M.