OK there are two shock scenarios we wish to avoid, 

1) a fault in our kit that raises the local metalwork to a nasty voltage relative to something bringing in a low impedance true terra-firma earth from outside.

Drain pipes,  building steel and fences and things.

2) a fault on the neighbours kit bringing a nasty voltage inside via some shared service (water main or similar) and being dangerous relatve to our own metalwork at near true earth voltage.

Both can be mitigated by bonding the CPCs of our local kit to the bits of metal that leave the zone.

But the peak current that may flow in fault is quite different.

I'd ague that if the metal conduit already connects to the metal stairs, then there is not a lot of point in an additional bond.  So there is a danger band of reistances - less than 22k - a dangerous current could flow, and more than perhaps an ohm, where the connection is not good enough to keep things below  50V for a large but credible earth current. The CPC of the water heater type situation, probably makes it fairly well bonded already and therefor unlikely to be at a dangerous voltage, unless there is so much fault current that the CPC is at risk of failure.

There is a very important infuence of the external impedance to ground, that we do not generally know. I have been told of cases where earth bonding has caught fire due to diverted neutral current, and much to the surprise of the victims pulling the company fuses did not stop it. At some very high level of current there is a danger  point when the big no-no of a fuse in the CPC suddenly looks attractive again.

The regs do not really distinguish between a shared service pipe that may have a sub-ohm connection to the substation star point, thanks to all the neighbours bonding, and really could take a large fraction of the street neutral current, and the random pipe in the ground that is more like an adventitious earth electrode, where the current is inherently limited by the resistance of the mud that surrounds it. They are however really quite distinct cases with different risks.

Mike

mapj1 said:

have been told of cases where earth bonding has caught fire due to diverted neutral

It is something we should be mindful of but try as I might, I cannot find any official documentary evidence of such occurrence. The LV network in ROI is almost entirely TNCS but none of my contacts in various authorities can point to a single incident of either fire or shock caused by loss of PEN. 
NI has also an extensive use of PME but similarly, HSE cannot point to an incident that caused injury. I am, however, aware of several loss of PEN incidents, one of which caused extensive damage to electronic kit in a large pharmacy.

Of course, the lack of incidents may be largely due to network integrity and effective bonding practice.

mapj1 said:

have been told of cases where earth bonding has caught fire due to diverted neutral

It is something we should be mindful of but try as I might, I cannot find any official documentary evidence of such occurrence. The LV network in ROI is almost entirely TNCS but none of my contacts in various authorities can point to a single incident of either fire or shock caused by loss of PEN. 
NI has also an extensive use of PME but similarly, HSE cannot point to an incident that caused injury. I am, however, aware of several loss of PEN incidents, one of which caused extensive damage to electronic kit in a large pharmacy.

Of course, the lack of incidents may be largely due to network integrity and effective bonding practice.

Actually some of the clearest pics that I have seen in the wild of a diverted current problem came from John Peckham 

It happens, but I think here at least it is often only partly recognized for what it is, and then silently fixed very quickly.

Mike