First question, will equipment connected to the exported phase' be within reach of the locally connected equipment? (if not, there might be the possibility of creating two separate earthing systems in building B). OK, probably not, but worth a quick consider as designing out the problem can sometimes be simpler than solving it.

Use earth A up to the building but earth B inside the building.

Feels uncomfortable to me. What if the DNO temporarily disconnect building B's supply (complete with its earth) - the imported phase would still be live but without an earth at all.  Ditto for some maintenance within building B. Next best might be to completely TT building B - but that's probably a last resort.

Connect the two earth's together.

Probably talk to the DNO just in case, but in effect that's probably happening anyway via main bonding if the two building share extraneous-conductive-parts (e.g. metallic water or gas supplies). Bonding wise (and interconnecting protective conductor) would hsve to be sized for the worst case (e.g. the larger supply and for PME conditions).

    - Andy.

AJ is much more qualified to answer this - so can I possibly add I think the regulation that is most relevant is 542.2.3.3 (Earthing) and 544.1.1 (Bonding) 

I think that the CPC must accompany the live conductors.

You must not have two simultaneously accessible different earthing systems. If one is TN-C-S and the other is genuinely TN-S, then I would say that they are different. However, if they are connected to the same transformer, and one is TN-C-S, the apparently TN-S one is also effectively TN-C-S.

100 m apart is quite a long way. Are you certain that the two service cables are branches of the same street mains?

How far apart are the buildings? If the supplies are both perpendicular to the road, that suggests 100 m, in which case voltage drop and EFLI might be considerations.

I did consider running the earth separately but decided it is not acceptable as there are some closely located db's which would potentially be on different earth's.  

I have seen elsewhere the argument that if one building is struck by lightning and the earths are connected surge currents could then be transmitted to the second site.  One of the buildings does have a real possibility of being struck by lightning (it's a church) and it has a lightning conductor.  However, I also take the point about the possibility that the earths may already be de facto connected via common services and will investigate this.  I will also investigate the connections to the lightning conductor.

I think it's highly likely the supplies come from the same transformer but will check with the DNO. I agree with the points about voltage drop and EFLI.

Thanks

Can I ask why you are exporting power from A to B as it were, and how much load it will be ?  As others have noted, this is tricky, and you cannot really avoid linking the earths if the two sources in the same building are within simultaneous reach, so unless one load is double insulated and in a plastic box, or up and out of reach, you really will have to use a common earth, and then whatever they were originally they are both acting like PME, unless you deliberately  make them both TT - a solution not without its own issues.
A church on PME sounds a bit unusual, but maybe the done thing in a built up area - I'm used to village churches being TT or TN-S.

As it sounds like you calling the DNO area engineers office, or whatever that role is called these days, you need to ask them about the options - you don't want to link things in a way that the armour of your cable ends up carrying some of the neutral current for other buildings on the same transformer - which is the sort of thing that can happen when supplies are sharing a building and bonding.

The lightning thing is an additional complexity - there should be a lightning conductor with its own earthing arrangements, and those earths should link to the building earth near the supply origin, at one point only, again to avoid diverted current. Cases with in effect 2 supplies require specialist input, from someone who can look at the plans and go 'aah'

Mike.

However, if they are connected to the same transformer, and one is TN-C-S, the apparently TN-S one is also effectively TN-C-S.

In theory, you could have a genuine TN-S and TN-C-S from the same transformer. Simple enough if the transformer had multiple secondary windings, but even if it had a single secondary, as long as the TN-S earth was kept apart from the PEN conductor all the way to the star point/substation electrode, I think it would still qualify as TN-S. In practice these days though, the DNOs will almost certainly treat any shared main as PME and feel free to interconnect N and PE during any alterations and repairs, so while any older cut-outs may look like TN-S, the normal position is to "treat as PME".

You must not have two simultaneously accessible different earthing systems. If one is TN-C-S and the other is genuinely TN-S, then I would say that they are different.

And conversely, even if you have two obviously TN-C-S supplies, that only means you have two systems of a common type, not one common system. Even if they happen to be connected together in the street at the moment, there's no guaranteed it'll stay that way - it's far from unknown for the DNOs to chop a few customers off the end of one main and extend the cable to be supplied from the next one. If you want to make them common, you need to connect them together yourself.

   - Andy.