We all know that when it comes to household electrical lighting, LED lights consume the least amount of energy for a given amount of light, followed by fluorescent tubes, halogens, and Edison-age tungsten-filament light bulbs.

We also know that roughly the same ordering applies when looking at endurance; LED's will last until the end of the universe, flurescent tubes will last thousands of hours, etc.

We've also all seen the graphs that show how much money you'll save for all of these types of electrical lighting. Looking at initial investment (purchase) and energy consumption over a year, the break-even point when compared to tungsten-filament bulbs lies for LED lights somewhere in the order of ~10-15 months, fluorescent tubes ~6 months and halogens ~4 months.

This is all very nice, but this all has a nasty sales-pitch/green-washed marketing aftertaste in all of this; way too much focus is on cost for the consumer, and very little is being said on actual overall energy expenditure.

I want to prove to myself and everyone else that LED's are really the way to go.

What I mean is the following. I would like to create a graph, plotting total energy consumption versus time (so money is not considered), with four lines in them (one for each of the aforementioned types), that takes into account:

  • The energy it takes to create a single bulb. See this related question. For example, to create a tungsten-filament light bulb, you need tungsten, glass, some metals, some plastics, a mixture of specific gases, cardboard and plastics for the packaging, ... Extraction/production/transportation of these materials takes an amount of energy per kg. Shaping it into a filament/glass housing/metal fitting/etc. takes an amount of energy, etc. In other words, from the "stuff that's in the ground" to "finished product in the store".

  • Replacement time (this is why LED's will probably have a smooth line, and tungsten-filament bulbs-line will probably show frequent upward jumps due to replacement). The nominal amount of burn hours stated on the packages (mean time before failure, MTBF) is not really sufficient here; these are averages somehow, and frequently come from calculations or laboratory setups rather than actual real-life usage (how else can they claim that LED's can be lit for 10 years continuously?). Also, there is a finite chance that the average user will accidentally drop the bulb, crash into it with a ladder, etc, which is far worse for tungsten-filament bulbs and luminescent tubes than for LED's in protective casings. So, a far more realistic statistical model should be assumed here, which I can't find any reliable, authorative sources for...

  • Energy expenditure due to disposal and/or recycling

  • Transportation. This is a broad topic and strongly depends on geographical location, but let's assume some worst-case scenario (where tungsten-filament bulbs have the advantage)

  • anything else I've forgotten :)

Of course, such detailed calculations can only be done when making some assumptions. Let's assume:

  • the most energy-efficient, state-of-the-art models for each kind
  • specific models with an equal or at least comparable amount of lumens produced (say, equivalent to a 40W tungsten bulb)
  • Daily average use of 5 hours in Winter, 2 hours in summer, sinusoidal in between (or some better, comparable model)

I find it very hard to find sufficient details on this subject to be able to do this comparison. I find it hard to believe that I'm the only one thinking this, but I'm just not finding them.

Does anyone know some sites or papers I should look into?

  • Why do you want real-world longevity tests (or at least statistical studies to back up longevity claims),for each light source but you (seemingly) pulled 5 hours/winter & 2 hours/summer usage out of thin air? Wouldn't it be better to use an average annual figure ? Most LED lightbulb replacements seem to have a published lifetime of around 25,000 - 50,000 hours, have you seen some that promise 2 million hour lifetime? – Johnny May 6 '13 at 23:36
  • @Johnny: Obviously, exact usage depends on latitude; these numbers are indeed just a random use case for my particular latitude. However, this is still how I want to set up my study, because the timings for the replacements will be different, it's easier to bring noise and outliers into the equation (we all forget to turn off the light once in a while), it's easier to model differences in wear due to the different load cycles, etc. But I guess it's mostly because I come from the simulation world, where theory and lab results never equal results found in practical applications :) – Rody Oldenhuis May 7 '13 at 5:33
  • @Johnny: As far as the lifetimes; googling around a bit will show that most modern LED bulbs have lifetimes well in excess of 100.000 hours (in other words, 11.4 years of continuous operation). If you look up the numbers for individual LEDs, the numbers easily exceed 1 million hours. Granted, this is for low-power, unpackaged LEDs without any AC conversion circuitry involved, but I do expect the developments in the lighting industry to catch up with these numbers in the next 5 years. – Rody Oldenhuis May 7 '13 at 5:41
  • 1
    This would be better broken into quite a number of separate questions that are reasonably answerable, such as "What is the energy it takes to create a single bulb". You'll probably get more upvotes doing it that way too! ;-) – Highly Irregular May 7 '13 at 8:18
  • 1
    @HighlyIrregular: indeed; I would like to see some form of validation of such extrapolation methods. I realize it's virtually impossible given the extremely long lifetimes of LEDs to do the actual experiment, but I've never seen really good experiments showing that the extrapolations actually have some merit... – Rody Oldenhuis May 7 '13 at 8:28

Perhaps this article is what you are looking for? It's a study by the US Department of Energy on

the life-cycle environmental and resource costs in the manufacturing, transport, use, and disposal of light-emitting diode (LED) lighting products in relation to comparable traditional lighting technologies.

There is a similar study done by manufacturer OSRAM

The conclusion of both studies basically is that currently the life-cycle energy of LED is more or less the same as that of CFL but expectations are that it will become better and that the LED's life-cycle energy will decrease by approximately one half.

  • 3
    I'd like to add a comment related to answer about CFL and LED having similar life-cycle energy. This may be true, but given the fact that when we talk about sustainability we should think about pollution too, In this case LED bulbs would be preferable as their disposal do not contribute as much with harmful chemicals as the CFL bulbs do. – Bogdan Mar 20 '14 at 16:12
  • It also depends on region. The DOE did not factor in the A/C cost to remove the additional heat generated by CFL infrared and convection losses for hot climates. The loss has double impact there. For colder climates, the loss of the bulb heats the house and therefore reduces the heating costs so a filament bulb as close to the humans as possible may make the most sense for sustainability. – Douglas Daseeco Feb 21 '17 at 10:38

i signed up just so i could make this comment. THelper♦ provided a good link to a DOE report on this topic, which i was researching because my roommate is skeptical about the advantages of efficient bulbs.

the report THelper♦ cited has the following bar graph that pretty much answers the question:

enter image description here

  • Thanks, this is the clearest figure on the topic I've seen so far! – Rody Oldenhuis Mar 4 '16 at 18:51

There is a lot of efficiency variation that depends on the quality of the LEDs. Cheap LEDs are no better than CFLs.

High quality LEDs from brands like Cree or Philips can be over double the efficiency of CFLs.

Also under-powering LEDs will make them more efficient and last longer. They can last indefinitely when underpowered by a certain level.

  • Do you have any references about underpowering LEDs? – Earthliŋ Apr 17 '16 at 9:19
  • 5
    @Earthliŋ cree.com/~/media/Files/Cree/LED-Components-and-Modules/XLamp/… On page 10 you can see that the the power output graph flattens out with increasing input current. Basically it is diminishing returns at higher power. Also lower temperatures will make LEDs more efficient. This is true of all LEDs I think. LEDs are improving all the time, there is a lot of rivalry with the big brands to make the most efficient ones. The Cree CXB3590 and XP-L are the most efficient white LEDs available at the moment. – Nick Collier Apr 17 '16 at 10:15
  • But IME white LEDs also fade over time, so you're limited by the life of the phosphor at some point. I'd love to see someone work out the optimum power, but I suspect it would vary significantly between LEDs (even expressed as a % of rated power). – Móż May 4 '16 at 4:32

I believe you can also find more detailed info through phys.org (the link throws you in the website with the "incandescent light bulb" tag on).

Regarding "state of the art", I would refer you to NanoLeaf (formerly NanoLight). They just completed a Kickstarter campaign with a lot more money than they asked for, and have reviews in several renowned magazines and websites.

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.