Just to clarify - are we talking “big” UPSs here - not the little under-desk ones (typically <3kVA single phase) that will typically operate in separated mode during mains fail events?

  - Andy.

NE bonding is a vexing thing, and the clear cut answer for one situation is equally strongly the ‘never do’ for another.

That said, I'll put my head over the parapet, and wait to be shot at…

So in the UK the ECSQR require that N and E are not bonded within a consumer installation. 

Diverted currents are raised as the reason, but many other places with 230V and PME/ MEN like installations (Eire, Oz, NZ, SA for a start) have no such restriction, only a requirement to size conductors for the likely currents. They do not seem to be overly beset by extra deaths or fires because of this.

Of course a private substation, or a large UPS are not really consuming equipment, more supply equipment, and a bond is expected and indeed required for correct operation of the ADS though it may be some distance from the transformer, depending how things are wired. Indeed no NE bond at all is very bad on a substation type transformer as the LV side can float up to a dangerous fraction of the HV via the capacitance between adjacent windings, and must be solidly earthed.

It is common to have multiple NE bonds on one site if for example your installation is large enough to have two or more HV to LV transformers, but the N and E would normally be interlinked in such a cross-section of conductor that a diverted current equal to the full load would not be an overheating issue.

A UPS is more like having a genset on-site, that perhaps only supports a fraction of the main load, and there one may need to switch in an additional NE link at handover to UPS, especially if one of the conditions of the genset/ups operating is total loss of supply, including severing of the mains earthing. Personally I am not a fan of the idea of switched links, and would rather have  solid ones and suffer a line that cannot be RCD protected. (the RCD seeing the other NE bond as a fault.)

A really small UPS, say a plug in one supplying perhaps one desk, is not such an issue if it runs as an IT (fully floating no earth at all), indeed if you pull out the plug, this is unavoidable.

I think these 3 cases, solid bond, switched bond, no bond, cover all situations between them, for large medium and small set-ups. On the boundary, roll the dice.

Mike.

AJJewsbury: 
 

Just to clarify - are we talking “big” UPSs here - not the little under-desk ones (typically <3kVA single phase) that will typically operate in separated mode during mains fail events?

  - Andy.

Just because the little “under-desk” UPS can float doesn't mean it's a good idea though. The issues for using this type of arrangement for a computer include accidental or fortuitous earthing (e.g. via connected peripherals that aren't UPS backed), and static charging of wired LAN connections.

Just to clarify - are we talking “big” UPSs here - not the little under-desk ones (typically <3kVA single phase) that will typically operate in separated mode during mains fail events?

  - Andy.

For clarification, the main query is with regards to a large UPS (>160kVA) supporting critical loads. The option is there to split into smaller UPS's of course, but this leads to an even larger minefield for neutral-earthing.

Of course a private substation, or a large UPS are not really consuming equipment, more supply equipment, and a bond is expected and indeed required for correct operation of the ADS though it may be some distance from the transformer, depending how things are wired. Indeed no NE bond at all is very bad on a substation type transformer as the LV side can float up to a dangerous fraction of the HV via the capacitance between adjacent windings, and must be solidly earthed.

Apologies for not being clear here, the question is in regard to sites with private substations where the NE link is solidly bonded through the LV switchboard NE links. Interesting points on DNO derived networks though.

It is common to have multiple NE bonds on one site if for example your installation is large enough to have two or more HV to LV transformers, but the N and E would normally be interlinked in such a cross-section of conductor that a diverted current equal to the full load would not be an overheating issue.

How is the ‘interlinking’ section achieved? Would this not contravene the BS7671 requirement of only one NE link being in place at any time?

Carpster_: 
 

It is common to have multiple NE bonds on one site if for example your installation is large enough to have two or more HV to LV transformers, but the N and E would normally be interlinked in such a cross-section of conductor that a diverted current equal to the full load would not be an overheating issue.

How is the ‘interlinking’ section achieved? Would this not contravene the BS7671 requirement of only one NE link being in place at any time?

Yes, it does. Regulations 444.4.6, 444.4.7 and 463.1.4 cover this specifically. See also Regulation 551.6.2.

The only other option is to arrange the installation upstream of the main swtichboard as full TN-C-S, but of course this option is not open to consumers as defined in ESQCR - see Regulation 8(4) of the ESQCR.

I could also play devil's advocate, by taking the stance that if you have multiple N-E connections, it's TN-C-S by any other name because the N and E are not separate between the sources of supply by virtue of the Neutral being "multiple-earthed". The reason for the quotes here is that I'm not saying it's “PME” - PME is strictly a TN-C-S system used for distribution to the public, although PME conditions may also apply to public supply PNB systems, these being the object of Regulation 8(3)(b) of ESQCR.

How is the ‘interlinking’ section achieved? Would this not contravene the BS7671 requirement of only one NE link being in place at any time?

Perhaps. 

However,  it is what happens in practice when transformers are arranged so that the HV sides are in parallel, and LV sides may end up in parallel or supporting each other's loads if for some reason one TX is to be taken out of service, or an external generator fed in. 

It is vastly preferable to have two bonds some of the time than the alternative of a system where the wrong sequence of operation may result in a transformer  without any NE bond at all and HV still applied. After a fashion it behaves like  a TNC, but it should be arranged so that the bits that might need to are bosky enough to take the full neutral current in either neutral or CPC, and as they are, or should be, before any circuit breakers for final circuits, it is not really part of the current consuming installation, more of a private distribution network.   This is not prohibited by BS7671. (The ECSQR however, maybe - but it happens, as does the practice of having an NE bond at the first breaker or fuse panel, rather than at the transformer - what do you call the cable from the TX to that panel - neutral, earth or PEN, and what does the transformer earth to ?)

Some thought is needed with the location of earth fault relays and so on to ensure they are only on the load side of all NE bond (s), and that ADS covers all cases. (get that wrong and the fault relay won't stay on…. )

Mike

Ask Ed.

Neutral handling and grounding of transformerless UPSs | Ask Ed & Eaton Insights - Bing video

Z.

‘Ed’ is right in the US - where the NE bond is on the consumer side of metering, and will not be lost if there is an external loss of supply for example due to a JCB hitting a cable.

In the UK it is not so clear - if the NE bond is off-site then loss of supply may sometimes mean loss of NE reference as well.

Mike.

mapj1: 
 

‘Ed’ is right in the US - where the NE bond is on the consumer side of metering, and will not be lost if there is an external loss of supply for example due to a JCB hitting a cable.

In the UK it is not so clear - if the NE bond is off-site then loss of supply may sometimes mean loss of NE reference as well.

Mike.

The O.P. mentions N-E bonding at the main intake switch panel and sub-boards as well, within the installation.

"So the way to achieve both requirements  would be to have a N-E link within the main LV switchboard and another within the essential board, and complete the changeover using a 4 pole device to ensure the neutral is switched.

So if Fig. 44.9 shows the correct arrangement, and Fig. 44.10 shows the wrong arrangement, can't large N-E cables be used, and a swift changeover switching arrangement thus preventing the problems mentioned in Notes 1 and 2 of 444.4.6?

Section 444 seems really only concerned with electromagnetic disturbances, and the possible effects on sensitive electronic stuff.

Z.