If the PCE has an internal impedance of 5 Ω to 6 Ω, then ignoring whether this is Ze or not, how can any Zs be satisfactory?

You just have to accept a high Zs and design accordingly for shock protection. If using ADS then pick an appropriate device (most likely an RCD - just like for TT), or use double/reinforced insulation (likely upstream of the 1st RCD), or even rely on the characteristics of the inverter to collapse the voltage to acceptable levels in the case of a L-PE fault of negligible impedance (section 419 style).

   - Andy.

How does the G99 relay work when the connection is made at the DB ? 

I know that’s not a requirement in your system, but a typical on-grid system.

AJJewsbury said:

You just have to accept a high Zs and design accordingly for shock protection. If using ADS then pick an appropriate device (most likely an RCD - just like for TT)

Thank you, Andy. Yes, I can see the need to protect in a similar way to TT, so is Zs relevant?

AJJewsbury said:

Ze is defined as the loop impedance external to your installation - if your system is entirely self-contained (off grid) then there is no external part - so the test and concept simply don't apply. I'd record it as "N/A".

Yes, I think that sums up the situation.

Hugh Frater said:

but this will just give the impedance of the output stage of the inverter, currently around 5-6ohms. This value being a fail because the supply from the PCE to the distribution board is T-N-S and this requires a lower value.

Which is why I mentioned Zs. Is the installation really TN-S?

Hugh Frater said:

Earth rod is connected to the MET, N-E link is made in the PCE (although code of practice states it should be in the distribution board). All circuits are on 30mA rcbo’s.

Still doesn’t help with the original issue/question though.

I think that Andy has answered it. In Section I of the EICR, put "N/A" for Ze, but you must also give details of the earthing in Section J including the resistance of the rod.

With a public TT supply, you can estimate the resistance by undertaking a Ze test, which ignores the resistance of the transformer's earth. You cannot do that in your case so the rod's resistance needs to be measured properly.

Chris Pearson said:

Thank you, Andy. Yes, I can see the need to protect in a similar way to TT, so is Zs relevant?

It is relevant in TT systems also ... even if RCDs are used ... see 411.5.3 (i).

Regardless of "very high" Zs, it's still a requirement to know that the RCD will operate in the stated disconnection time. To achieve 0.1 s disconnection times, we are talking in excess of 1 k ohm for 100 and 300 mA RCDs ... but can be as low as 250 ohms Zs for 500 mA RCDs.

I don't believe, though, that for island mode there is a common standard for protections built into the inverter, so I'm not sure that you can always guarantee the RCD will operate before the inverter current limits ?

Chris Pearson said:

I think that Andy has answered it. In Section I of the EICR, put "N/A" for Ze, but you must also give details of the earthing in Section J including the resistance of the rod.

This is not part of the earth fault path unlike a TT system, so it's not appropriate if you form a TN-S system in island mode (which is the preferred option and regardless of the technicalities will operate protective devices more quickly).

So, whilst it's correct to record the earth electrode resistance (it must be measured before energization in initial verification in any case), I don't think Ze is "N/A" and you can't rely upon 411.5.3 (ii) ... the value of RA (which is the sum of R2 from exposed-conductive-parts to MET, resistance of protective conductor between MET and  electrode, and the earth electrode resistance itself).

So similar to a typical earthed generator configuration 

AMK said:

So similar to a typical earthed generator configuration 

Well ... yes, but I can measure a loop impedance with a rotary generator (although noted the results can be variable with the wrong test instrument). With an inverter, the loop impedance/prospective fault current measurement is meaningless because the rms voltage is kept pretty constant during the test by the inverter.

It might mean, for an inverter, that Ze ≈ 0 for currents below the current limit, and Ze →∞ [or at least a very large number] for currents above the current limit, after a particular time (which may vary from manufacturer to manufacturer).

Thank you, Graham. Could you please confirm if it is appropriate to treat a PV system in the same manner as a standard generator with respect to earthing arrangements and related considerations? The value of the earth electrode should be low enough to allow sufficient fault current to flow in the event of a fault to earth; such as May arise from insulation damage on a cable.

AMK said:

Thank you, Graham. Could you please confirm if it is appropriate to treat a PV system in the same manner as a standard generator with respect to earthing arrangements and related considerations?

Yes, in so far as there is currently no distinction in BS 7671, AND Regulations 551.1.1 and 551.1.2 pull in "static converters" ...

AMK said:

The value of the earth electrode should be low enough to allow sufficient fault current to flow in the event of a fault to earth

Well, which 'earth' are you talking about/

'Earth' (capital E) is the 'general mass of the earth':

BUT, BS 7671 considers 'single fault conditions' which would be L to PE ... but not L to 'Earth' unless we are talking about additional protection? However, ADS is not 'additional protection' in general ...

AMK said:

The value of the earth electrode should be low enough to allow sufficient fault current to flow in the event of a fault to earth; such as May arise from insulation damage on a cable.

According to?

You see ... as far as BS 7671 is concerned, an insulated and sheathed cable can't have a single fault to Earth.

But, an armoured cable, or a cable with a cpc, can have a single fault to PE ... which in a TN-S or TN-C-S system, does not include Earth in the fault path ?