@pc1966 thanks for your post.
My understanding is that the OP has 120mm tails at the incomer, then SWA of various sizes from 25 to 50 feeding sub mains in other buildings.
Just noticed the OP's comments later on about other buildings. That gets trickier as not so easy to guarentee services are the same (i.e. one water pipe for all site, etc) and structures are obviously not the same!
I think we all agree the main bonding needs to be upped to 35 sq mm in the building the incomer is in.
Yes, agreed.
That is basically my question too, expressed another way. I have never been content that I had full understanding on this point.
If there is a situation where there are no parallel paths between the incomer building and the remote buildings via services (no linked extraneous conductive parts such as a common metal water pipe) does the csa of the bonding at the remote building need to meet the onerous bonding requirements of the origin?
One could argue that table 54.8's value can't vary within an installation as it references the supplies PEN conductor.
But equally if the only things being bonded are localised services, then having a bonding conductor bigger than the line conductor seems over-zealous, and this might be one for a departure and a risk assessment.
The thing about Table 54.8 is the assumption that under open-PEN faults the max bond currents depend on the supply cable size.
Now clearly there is some dependence but not as obvious as the usual fault-clearing adiabatic that is down to OCPD and clearance time. In the open-PEN case you could have 1A fuses in the DNO cut-out and it might make
little or no difference!
How big the earth fault current would depend on the point the PEN opened (specifically the unbalance between phase loads down-stream of the fault), the DNO cable impedance (so cable size is a factor here) and the impedance from extraneous parts back to the origin (transformer neutral).
I guess that 58.8's argument is along the lines of "bond big enough only to roast if the DNO cable is roasted" with maybe the guide that under most cases (effective site Ra of the order of a few ohms or higher) you won't see the max possible unbalance current as when the neutral goes floppy the load currents will adjust to reduce it (and light-loaded phase blow electronics...) and the Ra will dominate the answer.
I suppose you might argue that the site load itself is the closest an open-PEN fault could be, and then it is down to a similar bond size to whatever reduced-size neutral that would be acceptable for a supply cable. If you have a big load that could be the case, but you might have
more fault current that your own site if there is a lot of unbalance on the supply from multiple sites. Presumably the folks who drew up 54.8 have knowledge of the sorts of imbalances the DNOs might see along segments of their network to come to those sort of figures. Which is where the sub-main and remote building bonding issue get tricky...
I follow that logic and agree - I arrived at 95 sq mm too from a purely CPC perspective.
This is where I get a little lost.
Is it right to use k1/k2 with k1 as 46 and k2 as 143 from tables 43.1 and 54.4 respectively; this gives the answer as 92.9 sq mm.
The linked table shows 4 core 240 SWA having 289mm of steel (agreed) and 92.9 sq mm copper equivalent as it uses the same calculation.
Or are there better reasons to adopt the multiple by 8 method, something I've read many times on forums but never to my knowledge seen in the regs.
The thing with the adiabatic equivalence is it takes in to account both the higher resistivity of steel
AND the greater thermal capacity of the bigger conductor that it points to.
So the approx factor of 8 for conductance leads to similar R and so similar I^2R losses under current flow, i.e. same sort of heat per unit length. But under the short-duration fault case when you make the adiabatic assumption that heat is not going in/out of the system, what gives to the change in temperature is both the energy input (i.e. I2t*R) and the thermal capacity (which greater mass of material from lower conductance for a given R increases). Hence the factor of about 3 instead of 8 (so your resistance might still be almost 3 times that of the copper equivalent but you now have almost 3 times the steel material to soak up the heat pulse).
But the open-PEN case is not short-lived. You could be running the PEN current via the bond for potentially hours or even days before it is fixed, and supply OCPD or switching of the site isolator could make no difference (unless you are breaking regs generally-speaking by opening the CPC) depending on the fault location, etc. In that case the thermal capacity of the CPC/bond is irrelevant, it is now the dissipation (I^2R) and resulting ability of the cable to get rid of heat (i.e. "method") that sets the conductor temperature and resulting risk of insulation damage or even a fire.