RCD failure rate

[pc1966]

I was watching one of the SparkNinja videos recently on (I think) earthing and bonding and in the related chat they mentioned doing an inspection on a caravan park and almost half of the 100 or so RCDs had failed. Two years later when they went back for another inspection (and the owner had been doing regular testing with the built-in test button) it was around 10% that had failed.

I was quite surprised by this and wondered if they were a duff batch, duff brand, or maybe that is really the real-world expectation

Related to this was my curiosity in the Best Practice Guide #4 for coding faults giving a C2 for an RCD self-test failure but only for one in a TT system:

• The main RCD or voltage-operated earth leakage circuit-breaker on a TT system fails to operate when tested with an instrument or integral test button

I am puzzled by why they specify it for TT and not in all cases, given RCD protection is there for a safety reason and it being omitted from certain places a system (bathroom, outside socket, etc) is also cause for a C2 code.

Q1) When you do inspections and test what sort of RCD failure rate do you see?

Q2) How do you code an RCD trip test failure?

 

[radiohead]

I rarely see RCD's fail the correct test procedure. It's not uncommon for test buttons to fail to operate though, particularly with 4293 units which have not been operated for centuries . Personally on a failure rate of almost 50% of 100 RCD's I would question the test procedure or meter accuracy, simply because I have never experienced anything like that failure rate. I think even the 10% apparently found on the next visit is unrealistic.
Regarding coding on a TN system (obviously RCD failure on a TT is a potentially very serious defect and at least a code 2)
Socket outlets not expected to supply equipment outdoors......C3
Socket outlets which may supply equipment outdoors....C2
Bathroom circuits where supplementary bonding is in place.....C3
Bathroom circuits where supplementary bonding is not in place.....C2
Buried cables without earthed metallic sheath/conduit protection...C3
Domestic lighting circuits.....C3
This would apply to any aspect of RCD failure, not just the test button.

 

[Lucien Nunes]

Agree with all the above. I think I could count all the genuine failures I've seen on one hand, although bad contact at the test button is more common. I had to replace dozens at a marina where the RCDs themselves worked fine but the test buttons were intermittent. Failure rate both of the RCD itself and the test button might be higher when exposed to adverse climates though, such as in a hot damp outdoor cabinet at a caravan park or marina.

I think RCDs are wrongly accused all the time because:
a) The old VOELCBs got bad press and people expect the same of RCDs
b) Many N-E faults go undiagnosed and the RCD is changed and not pardoned when the N-E fault is finally located.
c) Incorrect test procedures including testing at the screwheads where the balance transformer surrounds the cable entry tunnel - that one caught me out a few times.
d) They are a good scapegoat for any other kind of fault that someone can't be bothered or doesn't know how to locate.

A similar phenomenon occurred with radio and TV valves, which people came to believe were the main cause of problems with their sets. Some valve types did eventually wear out and there were occasional random failures, but they were often changed needlessly because they were a quick buck. Customers expected to be told a valve was faulty and it wasn't in the interests of the dealers to dispel that myth. Sometimes the less honest service people and shops would charge for a new valve but polish up the existing one, and not mention the tuppeny capacitor that was the actual fault.

 

[pc1966]

Thanks folks, your experience is more in line with what I would expect (a couple of percent at most). My initial reaction to the comment was to wonder if there had been something seriously wrong with all the RCD (by batch, but brand, or a peculiar aspect of installation).

@radiohead Yes, to me the loss of the main (or possibly only) RCD on a TT system is almost in the C1 category as you more or less lose all ability to disconnect on a fault to earth.

@Lucien Nunes Your point about RCD faults being assigned to the RCD itself and not initially found as E-N or similar short is also a good one!

 

[Wilko]

Hi - fwiw, I’ve only had two fails that I can recall. Both failures did not seem adequately IP protected so I’m wondering if the unusually high failure rate was due to Caravan Park weather ingress?
[automerge]1588091904[/automerge]

Hi - fwiw, I’ve only had two fails that I can recall. Both failures did not seem adequately IP protected so I’m wondering if the unusually high failure rate was due to Caravan Park weather ingress?

Oops, 4 now recalled - 2 possibly IP, 1 Installer Error (oops) and one that might be manufacturer defect.

 

[123]

I recall doing EICR's on 3 food factory sites in West London about 11 years ago...a lot of MK RCD's failed, somewhere in the region of 50-100. These generally fed commando sockets so pretty easy to disconnect load and rule out N-E faults and the like.

 

[littlespark]

When I first started working on a caravan park, as manager, not electrician, I went round and tried every rcd. There was 300 to go through and they had never been tested by the staff.
All in all, there was about 20% had some form of fault. Test button didn’t work, water ingress...wrong manufacturer. Just as many MCBs.
Found one mcb had been changed by the caravan owner to a 32, and he had sawn off the switch and superglued a different colour one off the original 16A. ! He got a nasty letter for that.
Another had been changed to a 20A by the old electrician that used to come in. His reasoning was that it was a new static, with a washing machine and tumble drier... needed more power.
Had only used a 1.00mm Artic flex though.

 

[Mike Johnson]

All in all, there was about 20% had some form of fault. Test button didn’t work, water ingress...wrong manufacturer. Just as many MCBs.

Interested in your comment re "Wrong Manufacturer" above can you elucidate?

 

[littlespark]

Interested in your comment re "Wrong Manufacturer" above can you elucidate?

Thinking back now, it was an MCB wrong manufacturer... Whatever it was, the old electrician that bodged everything had used whatever he could get to fit... so this particular MCB was mounted upside down, and not really sitting on the DIN rail..
Had been liker that for some number of years.

 

[David Prosser]

I often carry out RCD testing on remote sites that very rarely get visited. I find they can fail to trip (probably 30%) on the test button If you press that first up. However if you test them through the range using the MFT they will then trip on pushing the test button I've never come across one that has not work on the test button if you go in with the MFT first (and it passed the tested).

 

[Mike Johnson]

I have found it you push the button four or five times after a long time of not being operated they seem to re-set and work on the button and MFT? seems they just need a clean internally.

 

[Flanders]

I do come across some rcd failures but as david says if you press the test button it tends to wake up the rcd and will then trip when tested a second time . Although I believe the correct order of checking an RCD is functioning correctly is you should be testing with the tester first and then use the test button. So yes I would say that RCDs do fail more often if they are not manually operated ever 6 months. Edit just seen Mikes post and agree

 

[pc1966]

This issue of testing "waking up" the RCD is worrying, but not so unexpected. If it is only a case of tarnished test button contacts then no big deal, but if it is taking a couple of trip attempts before the RCD contacts move and it isolates the load that is not good!

• The early bird gets the worm.
• The second mouse gets the cheese!

 

[Mike Johnson]

But then you need to know which circumstance has occurred, I suppose the only way to tell is how hard the switch is to re-set, this could be deceiving on a dual pole RCD. where only one side is not activated, again the only way is probably by comparison with a known good RCD of the same make.

 

[pc1966]

I guess it comes down to how these are commonly tested. The usual assumption is they are either working or not, so it would not matter if you push the test button first (even once or twice to see if it is bad button action) or try an RCD tester.

But if there is a tenancy to stick until triggered more than once it is worrying as you are mouse #1 !

Maybe I'm just a bit paranoid, but with more reliance on RCD action for protection (i.e. allowing higher Zs, etc) then it would be interesting to see if any are "reluctant to trip" or simply the test button had tarnished contacts when they don't fire 1st time.

 

[radiohead]

I think regarding the original point that nearly half of 100 units had failed the wallet of the client needs a bit of sympathy when deciding how to proceed. Clearly having to replace that many RCD's would be a very serious cost burden, if they were genuinely duff then there would be no option. But if a few operations rendered them fit and feisty then that may be sufficient, with an instruction to carry out the 6 monthly test of each unit verbally and in writing on the cert.

 

[Mike Johnson]

It would be interesting to know the kind of failures experienced by the manufacturers, dirt ingress, burnt contacts, overloading, frequent operation, internal component failures or external influence's? Who's going to be first to take one to bits and have a look? unless you have already?

 

[pc1966]

I think regarding the original point that nearly half of 100 units had failed the wallet of the client needs a bit of sympathy when deciding how to proceed. Clearly having to replace that many RCD's would be a very serious cost burden, if they were genuinely duff then there would be no option. But if a few operations rendered them fit and feisty then that may be sufficient, with an instruction to carry out the 6 monthly test of each unit verbally and in writing on the cert.

That is going to be a stabbing pain in the wallet for sure! Though I guess the ten-grand-ish cost is probably not such a big deal compared to the other costs in running a 100 place park.

I was more shocked by that sort of failure rate!

They (SparkyNinja team for father & son) did not say any more about the details of it though (at least not in the video I saw, maybe they have another on RCDs and stuff that might have it as a tale to tell). As already mentioned if it is failure on self-test button but still tripping on fault it is not so bad as folk are still protected, but obviously they still need replacing so they can be demonstrated to work each time a client connects.

That was the other bit of missing information: are these static caravans and folk just don't ever look as they treat it as a fixed system, or berths for mobile ones? I would like to think the former as otherwise all of the folk plugging must not have ever tested (or did and ignored the faults).

 

[MJPD29]

I'd say I've had 50 odd square d rcbos out of about 500 fail this year in a chain of shops I've been testing in. The are about 10 years old and highly doubt they're been tested since they were installed. I expected better of square d however.

 

[Mike Johnson]

If you still have any, how about taking one to bits to try and establish what failed?