Add things like fire alarms,  flood alarms, and in an industrial setting all manner of pressure, temperature, pH, gas leak, liquid spill and general malfunction detectors.  There are a great many systems that may or may not fall under the same catch-all - obviously rather depending on how serious the consequences if the alarm condition  is missed.

To make the point, that sometimes really odd things can sometimes need to be treated as if they were  "mission critical"  - consider the supply to cooling fans in chicken sheds - without which in some designs, the poor creatures die after a few hours due to their own body heat. not safety of human life perhaps, but certainly livelihood, and worth having an alarm for.

It is not reasonable for the reg-writers to foresee all this, so the wording has to be a bit fuzzy so it can be interpreted for the local case.

You raise the spectre of a common problem with robustness of IP based services, as while local networked kit may be on UPS or the site on backup genset, quite often the link out is a single point of failure, largely out of user control, - and simple secondary links may not work as expected - e.g. the mobile phone network may not stay up if there is a power cut at substation or HV level,  yet at that very moment it may be good to raise an alarm if the local supply is lost to pumps or fans or similar.
A internet service provider guarantee to pay compensation within 30 days should the network go down may not be as useful as it sounds ;-)
Mike.

mapj1 said:

It is not reasonable for the reg-writers to foresee all this, so the wording has to be a bit fuzzy so it can be interpreted for the local case.

100%. Where the failure of a system resulted in injury, the enforcing authority’s expectation would be that recognised industry guidance was followed as a minimum. Where that guidance is “fuzzy” then suitably robust and sufficient risk assessment should support the design proposal. 
Because 7671 isn’t explicit on the matter is the very reason that the risk of dropping SPD protection should be carefully considered. Given the ALARP principle that underpins just about every thing we do, I cannot see how one could make a reasonable case for omitting SPDs in any of the systems you list.

Maybe it is time for a bit of a wider rethink about chapter 56 - as many of the systems mentioned are likely to be found in domestic installations and some of the provisions of that chapter seem to not sit at all well in that context - e.g. for switchgear and control gear, and safety sources only to be accessible to skilled or instructed persons. Should we really be locking out CUs and smoke alarm backup batteries so they're not accessible to householders?

   - Andy.