I think we are pretty much in agreement John, RCD protection is usually supplementary in TN systems therefore the disconnection time is relying on the fuse/MCB primarily and that`s the time and Zs we work to . As a back up we add an RCD but do not place such reliance upon it as it is merely supplementary.
I still think that you may be thinking that I said something I didn't say. I'm obviously not suggesting that one should do away with the OPD and rely only on an RCD for 'L-E' fault protection - quite apart from anything else, one obvioulsy needs the OPD for overload and L-N fault protection.
Having said that, although there is always a lot of debate about this, I think you would be hard-pressed to find anything in the UK regs which explicitly says that one cannot 'rely on' an RCD to provide adequate disconnection times (even with a TN system) - although, as above, one would still need an OPD for other reasons. Indeed, Table 41.1 of BS 7671 (defining required disconnection times) actually has a footnote which starts "When compliance with this regulation is provided by an RCD ....", without saying that this only applies with TT systems.
You go on to raise the important point about the reliability of protective devices. As you say, we have some handle on the in-service 'failure rate' of RCDs, but haven't a clue about the corresponding in-service failure rate of MCBs. If one really had to choose between the two (which we clearly don't), I must say that I might be more inclined to 'rely' on an RCD. It can be tested, both by user and electricians (and replaced if found to be faulty) and the act of testing it regularly 'exercises' it and therefore reduces the chances of 'stiction'. I personally suspect that, if they could be routinely tested, the reliability of MCBs may well be found to be at least as poor as that of RCDs.
Furthermore, in response to a L-E fault of negligible impedance (which is what BS 7671 considers), the current (or current imbalance) resulting in the device operating, with a TN system' will be dramatically higher in relation to the device's 'trip threshold' for an RCD than for an MCB - which one might intuitively feel makes it more likely that the RCD would operated 'when required' in response to an L-E fault - with a PFC of, say, 640A, that would be less than 20 times the 'trip threshold' of a 32A MCB, but over 21,000 times the 'trip threshold' of a 30mA RCD.
It has also been stated that a (correctly functioning ) RCD might be expected to save 95% of the population - the frightening thing is that this implies it would not save 5%.
Very true, but you're going off at a bit of a tangent here. The only RCD which, when correctly operating, would get close to eliminating 100% of fatalities due to 'L-E' currents through human beings would have such a low I
Δn that everyone would be driven mad by 'nuisance trips' (some of which might themselves result in dangers to life and limb) - and, of course, no circuit protective device under the sun can (or ever will be able to) protect people who come in contact with both L and N. As you say, RCDs are certainly no panacea, but are nevertheless worth having.
Kind Regards, John.