How Long Did You Say That Bulb Would Last?

By Eric A. Taub
Left, two low-energy consumption bulbs and a traditional light bulb. (Credit:Vincent Kessler/Reuters)

How long is a piece of string? When talking about how long an LED lamp will last, that certainly seems to be the state of the question.

Manufacturers of LED lamps, which many regard as the next generation of lighting, destined to eventually replace today’s incandescent and compact fluorescent lighting sources, make wild claims as to product life.

Typical incandescent bulbs last 1,000 to 2,000 hours. But in speaking about LED replacements, lamp life is routinely quoted as 25,000 to 50,000 hours. Long lamp life, and the reduced power used to create the same amount of light, is what makes this technology so promising.

But what does a 25,000-hour life mean? As it turns out, no one is quite sure yet. The definitions surrounding LED lamps, a nascent technology, are still being made up as we go along.

One thing we do know: It means something different than when people think about the life of a regular light bulb.

When it’s said that a standard light bulb will last 1,000 hours, that is the mean time to failure: half the bulbs will fail by that point. And because lamp manufacturing has become so routine, most of the rest will fail within 100 hours or so of that point.

But LED lamps don’t “burn out.” Rather, like old generals, they just fade away.

When a manufacturer says that an LED lamp will last 25,000 or 50,000 hours, what the company actually means is that at that point, the light emanating from that product will be at 70 percent the level it was when new.

Why 70 percent? Turns out, it’s fairly arbitrary. Lighting industry engineers believe that at that point, most people can sense that the brightness isn’t what it was when the product was new. So they decided to make that the standard.

Of course, brightness is subject to the old frog in the boiling water syndrome. I’m sure that most people won’t even notice the lower level then, if they’ve lived with the same bulb for its entire life. (How many owners of rear projection DLP TVs only realize that a TV’s image has dimmed once they replace the bulb?)

If nothing else in the lamp fails, like its electronics, the product will continue to work until it becomes really dim. But some engineers are proposing a way to get around even that.

Their idea is that once the LEDs start to emit less light, increase the power to each one to increase its brightness. Unfortunately, that will also diminish the life of the lamp.

Good idea, or bad? “The utilities really don’t like this idea,” Fred Welsh, a Department of Energy consultant, told me on Thursday at a lighting conference sponsored by his federal agency.

Not only would contractors need to use thicker cables, but the utilities would need to create more power, partially negating the appeal of LED lighting in the first place.

But still, it’s in its early days, and no one yet knows how this will be settled, or how the consumer will be educated to think about “bulb life” in a different way than they have for the past 130 years. If consumers are going to switch to this new lighting technology, it’s an issue that needs to be settled.

And if it isn’t? “This is a potential black eye for the industry,” Mr. Welsh said.

Comments are no longer being accepted.

JD February 11, 2009 · 8:54 am

The same concern also applies to CFLs; they fade over time.

With current LEDs, there is also a significant chance the bulb will fail in other ways (electronics, cooling, etc) because of shoddy manufacturing, well before the LEDs themselves start to go.

DonO February 11, 2009 · 9:18 am

The individual LEDs in the City’s walk/don’t walk displays seem to die without fading. Perhaps this is a unique situation, but I thin k not. If it is, maybe LEDs used for illumination could have a suicide switch thtat would kill the bulb when the output falls below a certain level.

Larry February 11, 2009 · 9:21 am

Yes, of the seven CFLs I’ve installed, two have failed within three months. And I can’t get one to work outside in the winter.

NE February 11, 2009 · 9:35 am

From the title of the article I was hoping for some insight and specifics about CFLs. They last longer in a kitchen or outdoor application when they are on for a long period of time. I’ve had some die surprisingly fast in bathrooms when they are turned on for much shorter periods of time. The number of cycles seems to be an important factor in their life, as much or more than hours. This is sometimes mentioned in articles, but not often. Also, in settings like bathrooms, when the light is on for a shorter time, the warmup time of some CFLs is also sometimes annoying. In settings like kitchens and outdoors it doesn’t matter so much. Does anyone have a link to a website that reviews bulbs for quick warmup and their performance with a lot of on-off cycles? This would be valuable, especially if it included full-spectrum in addition to ordinary CFL bulbs. A CFL that is globe shaped and works well in bathrooms would be a valuable thing – my bathroom and all in my neighborhood have 2 bathrooms with 8 bulbs in a line above a long mirror, if regular bulbs are used this is 600 Watts – and a lot of waste heat to be counteracted during cooling season.

Don Wiss February 11, 2009 · 9:37 am

I’ve bought LED night lights from Lowe’s. They claim 25 years. A couple died completely after a few years. They flickered first (probably due to cheap capacitors) and then went totally dead. The others after a few years were so dim to be unusable. The 25 years bit is total BS.

James February 11, 2009 · 9:37 am

I had an amber colored incandescent lightbulb that lasted for about 25 years. I forget the wattage, probably 100 or 75. I went back to visit this apartment in Manhattan in the mid-nineties and was astounded when my friend who I gave the apt. to in 1973 said she never had changed the bulb, which was used regularly as a bathroom light.

J February 11, 2009 · 9:39 am

Wait does this mean it is actually possible to engineer a dimable LED?

gattrell February 11, 2009 · 9:55 am

My 5 year curly bulb lasted about 5 months. Of course, I left it on constantly in my pantry because I hated going into a dark room, pulling a cord, and having to stand there waiting for the bulb to come on and warm up so I could see… I really want to see immediately when I pull the cord. If not, I have to take a flashlight to see what I am doing. I live in a really old antique house, and there is always the possibility of sharing space in the dark with a mouse.

I am really hoping they improve the quality of the curly bulbs, make them normal looking so my lamp shades fit on them, and improve the color so I can tell what color things are. I pulled something out of the washer one day that looked brown, and I thought I had somehow ruined the garment, but when it got to real light is was red, as always. I hate to think what my skin must look like in that light!

Watts my name? February 11, 2009 · 9:55 am

As has Larry, I have had three or four CFLs, (with a seven year lifespan marked on the package), fail in less than a year.
One died in a manner that tripped a circuit breaker, and it was the only thing on that circuit at the time.

CPH February 11, 2009 · 10:05 am

“When it’s said that a standard light bulb will last 1,000 hours, that is the mean time to failure: half the bulbs will fail by that point. ” If half the bulbs fail by a certain point in time, that’s the median time to failure, not the mean time to failure.

don February 11, 2009 · 10:05 am

My laptop failed two weeks after the three-year warranty expired. That’s the American way isn’t it? So I’m sure they can do that for LEDs.

rcw February 11, 2009 · 10:16 am

Someone should do a serious investigation of the claimed life of CFLs. In my informal study involving maybe 20 bulbs so far 2 were essentially DOA (lasted less than a month) 10 have burned out in about the same time frame as incandescents sharing a lighting circuit (8-14 months) and the rest are still going, including a couple that are 7-8 years old. How does the energy consumption change if the real world lifespan of these much more complex and toxic lamps is not much more than an incandescent? Why are the manufacturers and CFL proponents allowed to claim such inflated lifespans?

LarryS February 11, 2009 · 10:19 am

Energy savings traded for reliability and MERCURY

Every CFL produced has mercury in it.Mercury causes madness.

Google “mad hatter”

Incandescents Forever !

Jeff Prestemon February 11, 2009 · 10:24 am

Another issue not evaluated is that, in many cases, there is a non-zero probability at all times that a bulb will break accidentally (e.g., the lamp gets knocked over, the light gets bumped with a piece of lumber, whatever). Simple probability theory or statistical simulations will reveal that the observed ratio of the lifespans of CFL or LED to incandescent bulbs may be half or less than the advertised ratio of their lifespans.

I did a simple simulation in a spreadsheet using a macro. Assumptions were that (1) the incandescent’s advertised lifespan was exactly 1488 hours and the LED’s lifespan was 25 times longer (i.e., exactly 37200 hours); (2) the probability of bulb breakage was 0.001 (one out of a thousand) for each hour of its operation until its random breakage or its advertised lifespan, whichever came first; (3) the probability of a random breakage was the same for the LED and the incandescent; (4) random bulb breaks were random draws from a Normal probability distribution; and (5) each random draw was independent across bulb types and over time.

Result: The ratio to simulated observed lifespan for the LED compared to the incandescent was about 13. Compare this with the advertised ratio of about 25. Of course, what any particular person would experience would depend on the probability of breakage and how those probabilities differ between the LED (or the CFL) and the incandescent.

wg February 11, 2009 · 10:24 am

The issue for manufacturers is the diminishing revenue from a replacement product with a substantial operating life over existing products. When businesses determine a operating life span that can sustain revenues, then the product will be marketed to the masses with an expected use span. The masses will accept it as true and purchase based on this cycle

Blonde Guy February 11, 2009 · 10:24 am

The idea of increasing power as the LED ages is brilliant. The LED lamp is such a good idea. I’ll be replacing all my CFs with LEDs once the price comes down.

I’ll also replace the last of the incandescents; the ones I kept because they are in the cold, or where I need light to come on faster than a CF.

Electric cars are going to be using LED lamps to save power, too. Keeping a constant brightness in that application is legally required.

When the LED is old enough to need replacing, then engineers should make it flash a couple of times when it comes on, to signal that it’s time for replacement.

Matt February 11, 2009 · 10:28 am

“a standard light bulb will last 1,000 hours, that is the mean time to failure: half the bulbs will fail by that point” – if half the light bulbs will fail by that point, 1,000 hours is the median, not the mean. The mean is the average of the lifetimes of all the bulbs. No reason to dumb it down to the point of being wrong.

Jeff Prestemon February 11, 2009 · 10:29 am

Addendum to my last post: My random draws were from a rectangular, not a normal probability distribution. Sorry. But it does not change the story.

PaB February 11, 2009 · 10:43 am

LEDs seem to be more durable if I judge by the ones that I have already installed. I was concerned 2 years ago by CFLs high failure rate but after having noticed a 5% rate of failure within the first 3-4 months, failure rate dropped significantly to be almost unnoticeable. I didn’t see much improvement on my ConEd bills since any savings are absorbed by the steady increases of their rates.

Martin February 11, 2009 · 11:09 am

One answer: Have the LED (or CFL) announce when it has reached its “end of life” (70% intensity, or whatever). It could start flashing a little, beep like a smoke alarm, turn purple, etc. (These days, it could twitter you a “replace me” message.) Users could even set their own EOL threshold when they install the device.

There is a good reason to do this, because if your LEDs are dim, you will turn on more of them to get acceptable illumination — and you lose the efficiency benefits. Same with CFLs.

Dan Schnitzer February 11, 2009 · 11:12 am

@Larry –
CFLs are not recommended for lighting a space where lighting is intermittent, as this significantly reduces the lifetime of the bulb. I used to have one in my bathroom, but it burned out in about 6 months because it was switched on and off so frequently. They’re better suited to kitchen, living room or bedroom spaces where your lights are kept on for greater duration.

LEDs also hold great promise for the developing world. The lifetime cost of a solar-LED lamp is significantly less than a CFL or incandescent. My non-profit organization, InterIntel (www.interintel.org), works with communities in least developed countries to improve energy and environmental quality. One of our current projects is making these lamps an option to those whose only choice is to use kerosene or candles for lighting.

am February 11, 2009 · 11:29 am

I’ve also found that CFLs don’t last nearly the “10x” as the hype says they do. LEDs won’t either. To make them inexpensive enough to sell widely, companies will have to cheapen the components and manufacturing to the point where they don’t last any longer than a normal bulb. We should only expect the LEDs to use less energy, not be some light-emitting miracle.

BobT February 11, 2009 · 11:53 am

Honestly, this makes sense. I’ve got some CFLs that seemed fainter recently, but I figured it was just my imagination or that they seemed fainter when it was dark outside. Maybe they’re just fading…..

V February 11, 2009 · 12:13 pm

how does the light know that it is dimming out? The sensing circuitry would add to the manufacturing cost. LEDs use so little power that the heavier wire argument is absurd. Just use heavy enough wire to begin with…which would still be 4 guages smaller than for incandescents of the same light output.