RCD periodioc testing and reliability

[JohnD]

I remember seeing a link to a research article a couple of years back, showing that RCDs which are not regularly tested are much more likely to stick or seize, and fail to operate in an emergency.

Does anyone remember this and know where I can find the link?

TIA

 

[streetlighter]

Electrical Safety Council website ?

 

[JohnD]

I think it may have been a European (not UK) report.

 

[ricicle]

Try here

Click on the link in the article for the actual PDF report

 

[JohnW2]

Try here ... Click on the link in the article for the actual PDF report

I’m sure we have discussed this before, and the Italian figures (7.1% failure when not regularly tested, 2.8% if regularly tested) are certainly frequently quoted. A few quick observations about the ERA report:

1...Despite what the ESC say in their summary of the ERA report, ‘ingress of contaminants and moisture’ was only mentioned in relation to 3/10 failed RCDs examined by ERA (2 cases and 1 case respectively – see Table 2)

2...In contrast, ‘contact welding’ or ‘severe contact disruption’ was reported in 5 out of the 10 failed RCDs which were examined. That makes one wonder whether they are engineered adequately to be able to break circuits on-load without being damaged and hence whether it is safe to leave an RCD in service if it has operated ‘in anger’ in the presence of a high load.

3...The overall failure rate figures are appreciably affected by the fact that one Housing Authority (‘B’, which contributed 80/607 RCDs tested) had a very much higher failure rate than the others. If that HA is excluded, the overall failure rate falls from 3.8% to 2.1%.

4...Attempting to determine whether regular testing of RCDs affects reliability by asking retrospectively (at the time of testing) when the RCD was last user-tested is a very biased approach which may make the effect of regular testing appear considerably greater than it is. If most users who regularly test their RCDs have them replaced if they fail the test, this would considerably reduce the probability of a faulty RCD being found at any particular point in time, even if the user-testing had no effect on reliability. Indeed, if monthly user-testing were undertaken (per Italian figures) with replacement of units which failed the test, one would expect to find very few failed RCDs at all on 'random' testing at any point in time. What is needed is a prospective study – or, at least, a study which makes use of full retrospective records about the history of the RCDs – something which is acknowledged by the ERA.

If the ‘failure’ rate is roughly of the order of 2% - 4%, this clearly would represent an awful lot of faulty RCDs in service in the UK at any point in time, which is clearly a cause for concern. However, as I often say, one has to wonder what is the corresponding figures for OPDs (MCBs and RCBOs). OPDs probably operate in service far less commonly than RCDs, cannot be user-tested and, in most practical senses, cannot be tested at all – and one can but presume that ‘sticking due to lengthy periods of non-operation’ is a fairly likely failure mode. One therefore might speculate thatthe 'failure' rate of MCBs could easily be higher than for RCDs. Particularly in view of (2) above, one certainly has to wonder how safe it is to leave an MCB (which effectively cannot be tested) in service after it has operated ‘in anger’.

Kind Regards, John

 

[dingodeano]

Try here ... Click on the link in the article for the actual PDF report

I’m sure we have discussed this before, and the Italian figures (7.1% failure when not regularly tested, 2.8% if regularly tested) are certainly frequently quoted. A few quick observations about the ERA report:

1...Despite what the ESC say in their summary of the ERA report, ‘ingress of contaminants and moisture’ was only mentioned in relation to 3/10 failed RCDs examined by ERA (2 cases and 1 case respectively – see Table 2)

2...In contrast, ‘contact welding’ or ‘severe contact disruption’ was reported in 5 out of the 10 failed RCDs which were examined. That makes one wonder whether they are engineered adequately to be able to break circuits on-load without being damaged and hence whether it is safe to leave an RCD in service if it has operated ‘in anger’ in the presence of a high load.

3...The overall failure rate figures are appreciably affected by the fact that one Housing Authority (‘B’, which contributed 80/607 RCDs tested) had a very much higher failure rate than the others. If that HA is excluded, the overall failure rate falls from 3.8% to 2.1%.

4...Attempting to determine whether regular testing of RCDs affects reliability by asking retrospectively (at the time of testing) when the RCD was last user-tested is a very biased approach which may make the effect of regular testing appear considerably greater than it is. If most users who regularly test their RCDs have them replaced if they fail the test, this would considerably reduce the probability of a faulty RCD being found at any particular point in time, even if the user-testing had no effect on reliability. Indeed, if monthly user-testing were undertaken (per Italian figures) with replacement of units which failed the test, one would expect to find very few failed RCDs at all on 'random' testing at any point in time. What is needed is a prospective study – or, at least, a study which makes use of full retrospective records about the history of the RCDs – something which is acknowledged by the ERA.

If the ‘failure’ rate is roughly of the order of 2% - 4%, this clearly would represent an awful lot of faulty RCDs in service in the UK at any point in time, which is clearly a cause for concern. However, as I often say, one has to wonder what is the corresponding figures for OPDs (MCBs and RCBOs). OPDs probably operate in service far less commonly than RCDs, cannot be user-tested and, in most practical senses, cannot be tested at all – and one can but presume that ‘sticking due to lengthy periods of non-operation’ is a fairly likely failure mode. One therefore might speculate thatthe 'failure' rate of MCBs could easily be higher than for RCDs. Particularly in view of (2) above, one certainly has to wonder how safe it is to leave an MCB (which effectively cannot be tested) in service after it has operated ‘in anger’.

Kind Regards, John

And despite all your quoted f acts and figures I've not heard of a single fatality caused where a correctly installed mcb failed to operate because it was faulty

 

[JohnW2]

And despite all your quoted f acts and figures I've not heard of a single fatality caused where an installed mcb correctly failed to operate because it was faulty

I'm not quite sure what you mean by "correctly failed to operate because it was faulty" - do I take it that we should ignore the word 'correctly'?

I haven't heard of such a case, either - but nor have I heard of a single proven case of a fatality caused by failure of an RCD to operate when it should, despite the fact that all the figures suggest that there are certainly tens of thousands, quite possible hundreds of thousands of 'faulty' RCDs in service in the UK at any point in time. One of the problems of anything to do with electrical fatality statistics is that there are (I would say surprisingly) so few such fatalities that it's very difficult to notice the effects of anything on the fatality figures.

Common sense certainly suggests that the combination of an OPD and (separate) RCD gives some useful redundancy to guard againstthe possibility of failure of of one of them (as regards low impedance L-E faults) - in that respect, an RCBO is a less attractive option, since (I think) it puts all one's eggs in the basket of the same set of contacts.

Kind Regards, John

 

[Iggifer]

And despite all your quoted f acts and figures I've not heard of a single fatality caused where an installed mcb correctly failed to operate because it was faulty

I'm not quite sure what you mean by "correctly failed to operate because it was faulty" - do I take it that we should ignore the word 'correctly'?
Kind Regards, John

I think it should read:

I've not heard of a single fatality caused where an installed mcb correctly, failed to operate because it was faulty

 

[stillp]

I'm a little disappointed by the fact that ERA don't seem to have reported on the make of the faulty RCDs found. Of course they would have had to describe them as "manufacturer A2, "manufacturer B" and so on, but hat might have helped explain the differences between HAs, in that different HAs might specify different makes.
A colleague in Italy told me that many of the failed RCDs in their oft-quoted report were of the same low-cost make.

I'm not sure that regular use of the test button is a good idea, as unless the householder switches off all loads before using the test button they will be accelerating the contact wear, which could lead to contact welding.

 

[dingodeano]

Sorry typo.

I have come across many an rcd new and old which does not trip in recommended times.
It's a testament to BS7671. As long as an installation is designed constructed and tested as per regs it's as safe as its going to be.

 

[dingodeano]

Sorry typo.

I have come across many an rcd new and old which does not trip in recommended times.
It's a testament to BS7671. As long as an installation is designed constructed and tested as per regs it's as safe as its going to be.

 

[JohnW2]

I think it should read:
I've not heard of a single fatality caused where an installed mcb correctly, failed to operate because it was faulty

I don't think I'd understand that version, either - do you mean "...a correctly installed MCB...."?

As I said before, I assume it refers to an MCB failing to operate when it should, because it is faulty.

Kind Regards, John

 

[JohnW2]

I'm a little disappointed by the fact that ERA don't seem to have reported on the make of the faulty RCDs found. Of course they would have had to describe them as "manufacturer A2, "manufacturer B" and so on, but hat might have helped explain the differences between HAs, in that different HAs might specify different makes. A colleague in Italy told me that many of the failed RCDs in their oft-quoted report were of the same low-cost make.

Yes, I agree that would be useful information.

I'm not sure that regular use of the test button is a good idea, as unless the householder switches off all loads before using the test button they will be accelerating the contact wear, which could lead to contact welding.

Indeed - I've often thought that - and, as I just wrote, these findings we're discussing make one wonder whether it's safe to leave an RCD in service after it has disconnected a high load. Regular operation of any mechanism (be it a water stop cock, a car starter solenoid, a hard drive, tape drive or whatever) in order to avoid 'sticking' makes sense, but I've always thought that householders should be advised to switch off as many loads as possible before testing an RCD.

Kind Regards, John

 

[JohnW2]

I have come across many an rcd new and old which does not trip in recommended times.

... as have most of us (certainly in-service ones, I'm not so sure about 'new') - that's really what we're talking about.

It's a testament to BS7671. As long as an installation is designed constructed and tested as per regs it's as safe as its going to be.

I don't really understand that statement. No matter how well designed, constructed and tested an installation may be, it's not going to remain 'as safe as it's going to be' for very long if there is a significant probability that protective device(s) may fail fairly soon after the electrician walks out of the door - even if the BS7671-basedpaper work is pristine and happy!

Kind Regards, John
Edit: red bit added

 

[Iggifer]

I think it should read:
I've not heard of a single fatality caused where an installed mcb correctly, failed to operate because it was faulty

I don't think I'd understand that version, either - do you mean "...a correctly installed MCB...."?

As I said before, I assume it refers to an MCB failing to operate when it should, because it is faulty.

Kind Regards, John

Yea technically 'a correctly installed mcb' I just added the comma without rewording