The message from snipped-for-privacy@care2.com contains these words:
And if we light 1500 rather than just 1 at once, we're going to see, on average 1500 failures per 1500 hours (though of course we would no doubt need to repeat the experiment many times over to reach such a figure empirically).
Now while that might appear to give an average failure of one per hour that has absolutely no relevance to the real world because it simply a piece of abstact arithmetic which ignores the other factors which come into play with regard to failures and it certainly has no menaingful relationship to lamp life which is the subject under discussion.
With respect, it's lamp life that we're interested in. If I install and turn on 1500 lamps with a rated life of 1500 hours, I am NOT going to be changing a bulb every hour.The vast bulk of them will blow between
1000 and 2000 hours, during which period I would expect to change more than one per hour. Your "average failure rate" simply doesn't belong to the real world and doesn't relate in any practical way to lamp life. Quod erat demonstrandum, as far as I'm concerned.
what other factors? Whatever they might be, each lamp still lasts
1500hrs, making them not very significant.
a direct mathematical one
not every hour, no, but on average 1 per hour.
I had a feeling this is where you were going. In real world use one does not install a set of new lamps and not replace them. Although initially an installation gets a new set, after a bit they are replaced at unsynchronised times, and the failure pattern settles toward the mean of, in this example one dead lamp per hour.
It describes precisely the _average_ time between relamps in the real world, once one has gotten past the first batch period, ie IRL.
from snipped-for-privacy@care2.com contains these words:
Each lamp does NOT last 1500 hours. 1500 hours is the average lamp lifespan.
And you have given no justification for saying that other factors which come into play with regard to failures are "not very significant." Cap-up or cap-down burning can make a considerable difference to lifespan of some types of lamp. Type of enclosure is also likely to make a significant difference with most types of lamp, Number of cold starts and the method of starting also makes a difference -- soft starts make a significant increase in lamp life, as does the ambient temperature from which the start is made.
A mathematical one which is not likely to be seen in a real-life situation. The parameters necessary to make it appear in real life are unlikely ever to be achieved.
The average family has 1. something children. 0.3 of a child (or whatever) is not a terribly meaningful entity to cope with -- no family actually has 0.3 of a child in a physical sense :-)
So you've got to introduce the concept of "a bit" to try to make your statistics relate to reality. "A bit" is hardly a very precise term, but it would be necessary, if your statistics were to relate to anything like reality, to define the "bit" more precisely. In point of fact, long before you got to the "bit" after which your statistics might bear some relationship to reality the batch of bulbs would be long gone, you would be long gone and your great-grandchildren would possibly be long gone as well!
You've also failed to take into account the rating specification, production engineering and quality control aspects of lamp manufacture. It's a bit like timing a watch -- all manufacturers -- and all watch repairers, untless requested otherwise -- set watchhes to run very slightly fast on the ground that nobody will tolerate a watch running slow, but most will tolerate a slight gain. With lamps, the specified figure for lifetime has got to be related to the bell-curve of failures. It really is unacceptable if more than 10% fail before the rated lifespan, though it doesn't matter if 90% of them last just a few hours more than the rated lifespan. Relamping can be a very expensive exercise in terms of access and disruption, so that relamping is very often carried out on a schedule related to rated life, whether or not the lamps have failed.
At least you're admitting that it couldn't apply to the first batch. But in point of fact it couldn't relate in the real world to the second or third batch either, for the reasons given above.
To me that confirms that it's a purely theoretical figure with no practical relationship to actual relamping and therefore at best irrelevant, but in practice extremely misleading and, without the qualifications which you have now introduced, so misleading as to be of the sort deemed useful by the average politican or banker :-)
I think we can take that as obvious. Perhaps I can rephrase it as an average of 1500hrs.
yes, but not to 1500hr lamps, ie filament lamps. Besides, we began by accepting the lamps lasted an average of 1500hrs...
We began by accepting the lamps lasted 1500hrs average. Either we're discussing 1500hr lamps or we're not. We know there may be small variations, but those aren't greatly significant when it comes to calculating mean relamping time.
You're saying 1500hr lamps are unlikely to last on average 1500hrs in real life use? I think the 1500hr figure is plenty close enough for our purpose here.
right - not that that changes anything. We both know what average means.
as I said after they've been replaced at unsynchronised times.
no :) 2000 hrs would be plenty
Not at all - 1500hr mean life means precisely that. There's no need to second guess the mfrs on that. Even if the figure were out by 10% for some reason, our end figure would still be correct to 10%, which is plenty good enough to make the point about increasing relamping frequency as the number of lamps goes up.
in a private house?
For commercial premises yes, but I think its already been made clear that's not the scenario being adressed here. We're talking about home use, where each lamp is replaced when it dies.
none of which show it couldnt apply
it _is_ a decription of practical relamping.
that just aint reality
Its a true statement of the mean relamping period for home use. If you dont think its of any use to know the mean relamping period for a person's lighting design that's your call. Its an issue thats asked here on ukdiy repeatedly, and understanding this basic aspect of lighting design helps people not to create problem installs, or to make an informed choice one way or t' other.
And if the electricity co record 253V as a maximum they are within the tolerance allowed. I have a substation in my garden and never seem to have to change lamps.
The message from snipped-for-privacy@care2.com contains these words:
1500 hours for the sake of argument. Specification, however, tends to be in terms of 1000 hours, 2000 hours etc.
But filament lamps are in fact designed for cap-up burning and burning cap down can affect life adversely.
With respect, these variations can be of considerable significance. Even with filament lamps. And in some cases, especially with filament lamps.
No, I'm not saying that 1500 hour lamps are unlikely to last on average
1500 hours in real life use. My quarrel is with your bringing this suprious failure every 150 hour figure into the discussion.
The question is whether it's meaningful in practical terms.
After 2000 hours it's most improbable that you would be replacing them on a frequency that related to anything like 150 hour interevals. There would be a considerable cluster effect for many generations of lamp changes. The tightness of the cluster, if they were all burning in identical conditons would bear a strong relationship to QC on the production line.
I'm not second-guessing the manufacturers -- any responsible manufacturer has production line tolerances and a reasonable degree of QC and will adjust spec until he gets the bell curve for lamp life just right. And make no mistake about it, it's a bell curve it is.
Lamps are not simply produced for private houses. They're designed for use in many locations.
As above -- lamps are produced for use in all sorts of locations.
Lamp failure rates following a bell curve does in fact indicate that it would take many generations of replacement and I thought you were keen a moment ago that we were to address a domestic situation?
Afraid it is reality I'm dealing with.
No-one wants a problem install. We agree that in general terms lamps with a 1500 hour rating will, on average, last less time than lamps with a 5000 hour rating, if each be used within the specified design parameters. However what you've produced has no relationship to the calls which the occupier of domestic premises can expect to have to replace light bulbs.
GLS lamp life must be measured cap-up (according to the BS). That doesn't mean they are designed specifically for cap-up operation, and I cannot think of any reason you'd expect shorter life cap-down. They are designed to operate in any orientation.
Candle and golfball lamps OTOH are specifically designed _not_ to be operated cap up, but cap horizontal through to cap down (and their life must be measured cap-down). If operated cap-up, the lamp base can exceed max permitted temperature. (This would be a problem for GLS lamps of 150W and up too, so they include an internal thermal barrier to prevent the lamp base overheating when operating cap-up, but there's no space to do this in golfball and candle lamps, and being much smaller, it's a problem for them at much lower power levels.)
I'd be interested to hear what mechanism would do that. Cap down would merely decrease cap temp, not a recipe for adverse effects on life afaik.
Recently we've been discussing a room that uses 10 lamps lit at once, and I've been saying that would give a mean relamping interval of 150 hrs to explain to people how number of lamps used at once affects relamping interval. It would take a very big deviation from the mean
1500hr figure to make a difference that is genuinely significant to the point being made here. And such large deviations are not normally seen in domestic use.
To be clear I'm saying with 10 lamps operated at once the mean relamping interval will quickly become 150 hrs - not that one fails every 150 hrs :)
I dont see how it can fail to be. Many people enquire here about this issue, it is relevant to them, and a mean relamping interval is meaningful. The fact that Jo Bloggs will inevitably relamp at intervals that vary each side of this figure is not a problem, it is a mean figure.
Real world filament lamp life is not so tighly clustered around
1500hrs that after generations you're still seeing masses of failures at close to 1500hr periods. Real world lamp life depends on random & semi-random processes, and varies a fair bit from one bulb to another
- and yes, still with a mean of 1500hrs.
Filament lamp mfr is an old art, you'd be real hard pressed to find batches that arent within spec.
precisely, and a wide one.
Quite so, but not that relevant - this is uk.d-i-y, not uk.traction.heavymother.16hz.
but this is uk.diy, people want to know what happens in their own homes, and sometimes other people's.
really it doesnt. Individual relamping intervals will vary, but the mean is still 150 hrs for a 10 lamp setup where all 10 are lit at the same time. We also have the fact that IRL its fairly normal for new installs to have some lamps replaced before failure with a different wattage, thus scattering things further.
yup
Even on that I dont think thats correct. Many people here do choose installations with eg lots of halogen lamps, which cause various issues. The aim is for them to understand what those issues are, so they can make an informed choice. Many do choose issueful installs.
good
calls?
And you still say that 10x 1500 hr lamps operated together having a mean figure of 10 failures per 1500hrs has nothing to do with reality? If so its time to move on.
The reply I wrote seems not to have gone through. I'll summarise...
The spec is 1500 hrs mean life.
not with filament lamps. CFLs yes. The only difference a cap down use makes is lower cap temp.
Filament lamp mfr is an old art, and mfrs have no difficulty meeting their specs, as is a legal requirement.
Environmental conditions may cause small changes in lamp life, but these are of trivial magniftude compared to the core point being made.
I said mean relamping interval is 150hrs, not one every 150 hrs.
It tells us how often joe bloggs can expect to relamp. Many ukdiyers ask about this, so yes its meaningful. Its also a fairly accurate figure.
Filament lamp failure is caused by random and semirandom processes, and individual lamp life varies considerably. IE the spread is wide. So the effect of all lamps being installed at one point in time quickly disappears.
Plus its common to change the wattage of some bulbs during the first
1500hrs in a new lighting install, which further breaks the initial failure pattern up.
None of these has any effect on the fact that mean relamping interval is still 150hrs though.
Filament lamp mfr is an old art... they're quite capable of meeting the specs.
true, but not relevant here. This is uk.d-i-y, and 99% of us are interested in domestic use.
Makes no difference to the fact that in a domestic setting....
If we were discussing this in uk.traction.heavymother.16hz we might want to consider other issues such as buffer impacts, but I dont think those apply to very many domestic installs.
it doesnt.
yes
In fact many ukdiyers choose them. Many choose them knowing the issues.
indeed
calls?
If you still dont believe that 1500hr filament lamps last a mean of close to 1500hrs, or that this doesnt translate to a mean relamping interval of 150 hrs when 10 are in use all lit at the same time, I'll agree to disagree and move on.
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.