Diversity

[ban-all-sheds]

Following a bit of a discussion prompted by yours truly failing to do due diligence ....

Fair enough. Does that mean that you agree with my interpretation (which, as I said, seems to be the basis of most advice one sees being given about diversity)?

Yup.

I was assailed by doubt, and got more uncomfortable with the idea that you could put a 19.1kW appliance on a 4mm² cable. Plus "diversity" is (almost?) always covered in "Maximum Demand and Diversity".

But, of course, you can't put a 19.1kW appliance on a 32A circuit - the 10A+(fl-10)*.3 calculation does not change the characteristics of the MCB. 19.1kW may well result in a post-diversity current of 31.9A, but on switch-on a B32 isn't going to put up with 83A for very long.

But up with what will it put?

"1.45In within 1 hour" is satisfied by 1.45In within 10ms. Can you rely on it being able to pass 1.45In for any length of time at all? Is the only figure you can properly rely on the non-tripping current of 1.13In? Should you probably not put anything larger than an 8.316kW appliance on a B32?

We've all seen the time-current curves for BS EN 60898 devices, but are they typical? Are they the maximum times (i.e. worst performance) you can expect from them? Is there any official guarantee that any device won't work much faster than those curves?

If I built a very sophisticated device which would pass 1.13In indefinitely (as required) but tripped "instantly" at 1.14In (ticks the box for 1.45In within 1 hour) would it comply?

Interestingly, when I got out my copy of the IET's Electrical Installation Design Guide, I read that (with my emphasis) "The demand of a circuit or of an installation is the current taken by the circuit or installation over a period of time, say 30 minutes", which does put my original concern firmly to bed.

Without any tea.

Actually, what was the interesting bit, and this probably goes a long way to answering previous questions about leeway on cable ratings, tolerances, all the malarkey about how one must use actual voltage rather than nominal or ones house will explode etc, at Iz a PVC cable takes 1 hour to get to 70°C.

 

[JohnW2]

Following a bit of a discussion prompted by yours truly failing to do due diligence ....

Fair enough. Does that mean that you agree with my interpretation (which, as I said, seems to be the basis of most advice one sees being given about diversity)?

Yup. ... I was assailed by doubt, and got more uncomfortable with the idea that you could put a 19.1kW appliance on a 4mm² cable. Plus "diversity" is (almost?) always covered in "Maximum Demand and Diversity".

OK.

But, of course, you can't put a 19.1kW appliance on a 32A circuit - the 10A+(fl-10)*.3 calculation does not change the characteristics of the MCB. 19.1kW may well result in a post-diversity current of 31.9A, but on switch-on a B32 isn't going to put up with 83A for very long.

Well, for a start,if such a thing as a (domestic) 19.1 kW cooker existed, it presumably would have a large number of hobs/ovens/grills/whatever, and the probability of them all being switched on, from cold, simultaneously would be pretty remote - and, if we are to accept the diversity concept (and formula), even if all were on simultaneously, once they were all on, they would settle down (under thermostatic control) to an average load of around 31.9A.

But up with what will it put?"1.45In within 1 hour" is satisfied by 1.45In within 10ms. Can you rely on it being able to pass 1.45In for any length of time at all? Is the only figure you can properly rely on the non-tripping current of 1.13In? Should you probably not put anything larger than an 8.316kW appliance on a B32?

In some senses (the sense of safety), one could argue that it doesn't really matter. Taking into account the characteristics of MCBs and the various build-in 'safety margins', it has been deemed that a (clipped direct) 4mm² cable will be adequately protected by a B32 MCB. That presumably remains the case whether the current flowing, or trying to flow, is 37A, 50A or 100A.

We've all seen the time-current curves for BS EN 60898 devices, but are they typical? Are they the maximum times (i.e. worst performance) you can expect from them? Is there any official guarantee that any device won't work much faster than those curves?

I'm not aware that any Standard imposes a minimum operating time. However, some manufactures do publish curves which show both 'minimum' and 'maximum' times - the latter of which corresponds to 1.45n. However, as above, although it might be an inconvenience if an MCB operated much quicker than the maximum in the face of a diversity-dependent situation, it would still not represent any safety issue.

Actually, what was the interesting bit, and this probably goes a long way to answering previous questions about leeway on cable ratings, tolerances, all the malarkey about how one must use actual voltage rather than nominal or ones house will explode etc, at Iz a PVC cable takes 1 hour to get to 70°C.

Indeed. I think that, over the years, we have decided/concluded, and largely agreed, that there clearly is a lot of 'safety leeway' built into most of the calculations we undertake.

Kind Regards, John

 

[ericmark]

I think I have missed something here? I have looked at my cooker.
Volts 220 - 240.
Watts 10466 -12455.
Recommended supply size 32A.
So 47.5 - 52 amp on a 32A MCB following manufactures instructions.

 

[JohnW2]

Interestingly, when I got out my copy of the IET's Electrical Installation Design Guide, I read that (with my emphasis) "The demand of a circuit or of an installation is the current taken by the circuit or installation over a period of time, say 30 minutes", which does put my original concern firmly to bed. ... Without any tea.

I missed that bit. I'm not sure that it concerns me too much. For a start the words are "...say 30 minutes". More to the point, those who produced the diversity formula we are using presumably believed that, at least after one has got beyond any initial 'switch on everything from cold' stage, the current will average out as a maximum of 10A+(fl-10)*.3 over 'any reasonable period of time' (and I would have thought that 30 mins or so was more than adequate) - and, furthermore, that the average current over a reasonable period of time is probably roughly what matters in terms of thermal operation of an MCB.

Kind Regards, John

 

[JohnW2]

I think I have missed something here? I have looked at my cooker. Volts 220 - 240. Watts 10466 -12455. Recommended supply size 32A. So 47.5 - 52 amp on a 32A MCB following manufactures instructions.

I imagine the most likely explanation is that they have invoked diversity when arriving at their 32A MCB - using the normal formula, the after-diversity current with 12,455W would be around 22.6A, so a 20A MCB would not be adequate - hence they specify 32A (since 25A ones are not available for many CUs).

Kind Regards, John

 

[ban-all-sheds]

Well, for a start,if such a thing as a (domestic) 19.1 kW cooker existed, it presumably would have a large number of hobs/ovens/grills/whatever, and the probability of them all being switched on, from cold, simultaneously would be pretty remote

Yes, but the probabilities of all sorts of things are pretty remote but that doesn't stop there from being regulations aimed at preventing or mitigating them.

and, if we are to accept the diversity concept (and formula), even if all were on simultaneously, once they were all on, they would settle down (under thermostatic control) to an average load of around 31.9A.

Indeed they would, but until they reached operating temperature the appliance would be drawing 83A, 2.6x the rating of the OPD, and possibly 2.6x the CCC of the cable.

In some senses (the sense of safety), one could argue that it doesn't really matter. Taking into account the characteristics of MCBs and the various build-in 'safety margins', it has been deemed that a (clipped direct) 4mm² cable will be adequately protected by a B32 MCB. That presumably remains the case whether the current flowing, or trying to flow, is 37A, 50A or 100A.

My point there was not about safety, it was what formal assurance does a designer have, in the form of ratified national or international standards, that a protective device will not operate at > 1.13In so quickly that his circuit will not work, and therefore is a faulty design?

I'm not aware that any Standard imposes a minimum operating time. However, some manufactures do publish curves which show both 'minimum' and 'maximum' times - the latter of which corresponds to 1.45n. However, as above, although it might be an inconvenience if an MCB operated much quicker than the maximum in the face of a diversity-dependent situation, it would still not represent any safety issue.

Well there might be safety issue - it depends on what the circuit supplies. But as per above, safety wasn't what I was thinking of - it was whether the circuit would actually work. If in reality a lot of MCBs did work such that a B32 operated near instantaneously at 36.5A and therefore there was a good chance that you could not turn all of an 8.4kW cooker on at once, would you really be happy with having a circuit like that because your inability to turn your cooker on was not a safety issue?

I missed that bit. I'm not sure that it concerns me too much. For a start the words are "...say 30 minutes". More to the point, those who produced the diversity formula we are using presumably believed that, at least after one has got beyond any initial 'switch on everything from cold' stage, the current will average out as a maximum of 10A+(fl-10)*.3 over 'any reasonable period of time' (and I would have thought that 30 mins or so was more than adequate) - and, furthermore, that the average current over a reasonable period of time is probably roughly what matters in terms of thermal operation of an MCB.

Yes but, yes but, yes but.

I do not believe that it is acceptable to base design on "in practice you'll get away with it".

I do not believe that anything less than formal proof of the correctness of a design will do.

So if there is nothing formal to say that an MCB will allow 1.13In < Ib < 1.45In for any length of time then IMO a designer must not assume that 1.13In < Ib < 1.45In will be OK for any length of time.

Yes, we all know that MCBs don't in practice work like that, and therefore we all know that in practice 1.13In < Ib < 1.45In will work for long enough.

But that is not an acceptable, scientific basis for design - you have to have formal guarantees that it will work. Without that it's not engineering, it's not a design methodology, it's art, it's styling, it's superstition, folk lore, old wives tales.

 

[ban-all-sheds]

I imagine the most likely explanation is that they have invoked diversity when arriving at their 32A MCB - using the normal formula, the after-diversity current with 12,455W would be around 22.6A, so a 20A MCB would not be adequate - hence they specify 32A (since 25A ones are not available for many CUs).

There is absolutely no reason why they could not specify "at least 25A".

 

[JohnW2]

Yes, but the probabilities of all sorts of things are pretty remote but that doesn't stop there from being regulations aimed at preventing or mitigating them.

The common thread through nearly all you have written here is that you do not seem to accept that very many (probably most) things in this world are necessarily to some extent, and in some senses, probabilistic, rather than absolute or 'guaranteed' - and that any regulations relating to such things often have to work within that framework.

If a designer chooses to invoke the concept of diversity, (s)he is knowingly adding a layer of statistical uncertainty - hence has to accept (and, if competent, presumably regards it as reasonable to accept) that they are introducing a further, finite, probability that things will not work out in accordance with the 'expected' (i.e. 'average') way predicted by statistical/probability theory. Provided that does not impact appreciably on safety, the designer may regard that as an acceptable risk - but if they want to avoid that additional risk, they should not rely on diversity. Given enough information and data, the magnitude of the risks can be at least approximately quantified to aid those decisions. In practice, it's usually much less formal than that - in the context of the way this discussion started, practising electricians know, from their experience of callouts (or the lack of them), that the probability of a cooker circuit 'not working' because of the application of diversity is extremely small (and, in many of their minds, 'acceptably small').

My point there was not about safety, it was what formal assurance does a designer have, in the form of ratified national or international standards, that a protective device will not operate at > 1.13In so quickly that his circuit will not work, and therefore is a faulty design?

Again, there could never be any absolute "formal assurance" (at best only a probabilistic one) - and the designer who does not feel that the increased risk (of the circuit 'not working') is acceptable, should not rely on diversity.

If in reality a lot of MCBs did work such that a B32 operated near instantaneously at 36.5A and therefore there was a good chance that you could not turn all of an 8.4kW cooker on at once, would you really be happy with having a circuit like that because your inability to turn your cooker on was not a safety issue?

No - and, yet again, if it had emerged (through formal estimation of probability or by the experience of electricians) that the probability of this situation arising was appreciable ('unacceptably high'), then designers would not invoke diversity.

Yes but, yes but, yes but. ... I do not believe that it is acceptable to base design on "in practice you'll get away with it". ... I do not believe that anything less than formal proof of the correctness of a design will do. ... But that is not an acceptable, scientific basis for design - you have to have formal guarantees that it will work. Without that it's not engineering, it's not a design methodology, it's art, it's styling, it's superstition, folk lore, old wives tales.

No, it's statistics and probability. As I've said, the underlying basis of your position is that 'uncertainty' is not acceptable - but some degree of uncertainty is often inevitable, and taking a course which involves accepting a (usually very small) degree of 'risk' due to a consequential increase in that uncertainty is often sensible or essential ...

... if we were not prepared to accept a finite probability that things would go wrong (i.e. deviate from the statistically 'expected') sensible domestic installations based on 60A/80A supplies would very often be impossible, 63/80/100A-rated main switches in many CUs would be unacceptable and the electricity generators would be in an impossible position, probably having to have a generating capacity 10 or more times greater than the usual maximum demand. Fortunately, it's almost amazing how well real-world things do work out very close to the 'statistically expected' - with the probability/risk of things deviating seriously from that, albeit inevitably finite, being very small.

Kind Regards, John

 

[JohnW2]

I imagine the most likely explanation is that they have invoked diversity when arriving at their 32A MCB - using the normal formula, the after-diversity current with 12,455W would be around 22.6A, so a 20A MCB would not be adequate - hence they specify 32A (since 25A ones are not available for many CUs).

There is absolutely no reason why they could not specify "at least 25A".

That's obviously true - but given that 25A ones are pretty rare, yet 32A ones ubiquitous, I'm not that surprised that (if my guess about their thinking is correct) they have specified 32A. Of course, somewhat in support of your suggestion, if a 25A were available for the CU in question, that might allow 2.5mm² cable to be used!

Kind Regards, John

 

[EFLImpudence]

Does this thread with the link supplied offer any favourable conclusion?

http://www.theiet.org/forums/forum/messageview.cfm?catid=205&threadid=59632&enterthread=y

 

[JohnW2]

Does this thread with the link supplied offer any favourable conclusion?http://www.theiet.org/forums/forum/messageview.cfm?catid=205&threadid=59632&enterthread=y

Unfortunately, I don't really think so - that link is talking about the calculation of 'thermally equivalent loads' (essentially 'after diversity' loads) when the is regular cyclical variation of load/current and both the duty ratio and currents associated with that cycle are known - which will virtually never be the case in domestic installations.

A lot of this is little more than applied common sense. Although BS7671 itself says virtually nothing about diversity, it is clearly permissible in appropriate situations and, as you know, there is a reasonable amount of detailed guidance in the OSG. However, although permissible, it obviously is not mandatory - so, as I said, it's for the designer to decide, on a case-by-case basis to what extent they regard it as 'reasonable' to introduce the inevitable (even if very small) uncertainties ('risk', although generally not of safety) of any probabilistic situation, and also how much 'diversity allowance' is reasonable in a given situation. Just because the OSG provides a 'guideline diversity formula' does not necessarily mean that a designer will feel, in a particular situation, that it is appropriate to invoke 'as much diversity' as that guideline. In some extreme circumstances, the designer might not feel that any allowance for diversity was appropriate. In other situations, a designer might even decide that invoking a larger 'diversity allowance' than is suggested by the OSG would be acceptable.

As I wrote last night, although it may sometimes be possible to actually undertake some approximate calculations in relation to diversity (as in the link you have just provided), in practice there is virtually never going to be enough information or data available to facilitate this (at least in relation to domestic installation) - the calculations described in that link rely on knowledge of the duty cycle, periodicity etc. and is therefore pretty 'solid' ('deterministic' if you want a technical term). In a domestic environment, all one usually has are guesses about variable things, hence the situation is inevitably probabilistic. Designers will therefore usually have to rely on guidance (like the OSG), common sense and, in particular, the experience of themselves and others. As I said, I strongly suspect that electricians have learned that, for example, application of the 'OSG diversity formula' for cooking appliances very rarely results in callouts (or even 'reports') relating to OPDs operating as a result of diversity having been utilised in the design - they therefore have strong empirical evidence that this particular diversity guideline is 'very reasonable' in practice.

Kind Regards, John