LED street lighting: Not as good as you think.


This post on the Greater City: Providence blog is excited about LED street lighting. It links to this post on Red Green and Blue, about an LED-street-lighting pilot program in New York, which mentions that they're apparently replacing high-pressure sodium lamps with LEDs.

That doesn't seem like a very good idea to me.

LED street lamps could work very well. But the numbers don't look good yet.

I can believe the part where the Greater City blog quotes ScienceDaily as saying "If all of the world's light bulbs were replaced with LEDs for a period of 10 years...", vast amounts of power could be saved.

But that's talking about replacing incandescent-filament light bulbs, whose luminous efficacy - amount of light produced per watt of power you put into them - is miserable, down around 17 lumens per watt.

Almost no street lights use incandescent bulbs, for exactly this reason. Instead, street lights use fluorescent tubes and gas-discharge lamps of one kind or another - often low-pressure and high-pressure sodium vapour lamps. The NYC pilot program is replacing high-pressure sodium lamps with LEDs.

Low-pressure sodium lamps are highly recognisable, because they output monochromatic orange light. Single-colour light like that only lets you see the world in shades of orange (in other words, its colour rendering index approaches zero), but you get a whole lot of light per watt - up to 200 lumens per watt.

High-pressure sodium lamps give white light with reasonable colour rendering (though their spectrum is still a long way from being smooth). They can have luminous efficacy as good as 150 lumens per watt.

And then there are fluorescents. Fluoro streetlights generally use the highest-efficiency fluorescent tubes in existence, which are the "triphosphor" tubes whose output has a distinctive greenish-white look. (This is why anywhere lit by cheap triphosphor fluoros, like warehouses and public toilets, will make people look zombie-ish.) Triphosphor is close enough to white for government work, though. Triphosphor fluoros manage about 100 lumens per watt.

So existing, common, street-light technologies have luminous efficacy ranging from 100 to about 200 lumens per watt.

Thus far, white LEDs have managed about 100 lumens per watt.

Only a few years ago, the best white LEDs were only achieving about 25 lumens per watt, the same as halogen incandescent lamps. There's been a lot of market pressure to create better white LEDs, and the technology is leaping ahead.

But this doesn't change the fact that if you switch all of your fluorescent street lights to LED now, you'll save no power at all. If you switch discharge-lamp street lights to LED, you'll use more power to get the same illumination.

The one fact about LEDs that everybody latches onto, which leads to things like that Greater City post, is that many LEDs need very little power to operate. A normal 5mm white LED will work very nicely from a twentieth of a watt.

But a 5mm white LED also outputs very little light, by street-light standards. And LEDs are not magic hyper-efficient light sources; they waste energy as heat just like every other kind of lamp. It's just that it's hard to notice that wastage, when the total lamp power is only a twentieth of a watt. So people often seem to think that LEDs waste no power, and must thus be the best light source in the world.

To be fair, LEDs do have one unique advantage over all conventional lamps: They're inherently directional. The light comes from a little metal pit inside the LED, and it comes out of only the top of the pit.

This means that it's quite easy to make an LED lamp that throws light in only the direction you want it to, with no efficiency-sucking reflectors or wasted light shooting up into the night sky to pollute it. So the effective luminous efficacy of an LED lamp, for street-lighting purposes, may be higher than its raw efficacy number might suggest.

I presume it's this fact that makes the NYC pilot program worthwhile. The Red Green and Blue piece mentions that "the light footprints can be tailored for parks, street corners or mid-block", which implies that they're replacing sodium-vapour lamps with an unnecessarily wide throw with LEDs that light up only what needs to be lit. If this is the case, then even replacing 150-lumen-per-watt sodium lamps with 100-lumen-per-watt LEDs could yield a net improvement. Even if you just want the usual round-pool-of-light, a well-designed LED luminaire could work just as well, if not better, than a technically-brighter vapour lamp.

But LEDs are not, yet, the slam-dunk winners that so many people seem to think they are.

Here's another problem: White LEDs wear out.

Nobody's yet made a "native" white LED. All white LEDs so far are actually blue LEDs, with a phosphor layer over the blue die that eats some of the blue and emits the other colours needed to create light that looks white. And the phosphor slowly burns out and becomes opaque, which reduces the LED's brightness.

There's seldom a clear point where a white LED "dies", but you shouldn't expect street-light white-LED lamps to last more than a few years. Fluorescent tubes will probably need replacing more often - and they really do die, not just get dimmer and dimmer - but fluoro tubes are very cheap. I suspect the value-for-money difference between LED and fluorescent in this case would hinge on how much it costs to send people up ladders to change the lamps.

One solution to the white-LED-lifespan problem is to not use white LEDs, but a combination of red, green and blue coloured LEDs. They should last far longer...

Mixed coloured LED light

...and can decently approximate white light.

They have higher luminous efficacy, as well. Coloured-LED luminous efficacy hasn't been improving nearly as rapidly as white-LED efficacy has, but an array of red, green and blue LEDs should still be highly competitive, in lumens-per-watt, with other street-light lamp types.

(This is also why LED traffic signals work so well. LEDs can natively emit red, amber or green light, and you want a traffic signal to be directional, too. LED traffic lights are just hilariously better, in every important respect, than the old type, which uses low-efficacy incandescent bulbs with coloured filters in front of them that eat most of their output.)

The Greater City: Providence piece dreams of street lights that use so little power that a solar panel on top of each light can charge it up with all the power it'll need to work all night.

That, I'm afraid, is going to remain a dream for some time yet.

Yes, cheap LED garden lights work that way. But if you scale them up and put them on top of a pole, you'll either need an outrageously large solar panel, or have to settle for a very dim street light.

LEDs are not a miracle product for street lighting.

13 Responses to “LED street lighting: Not as good as you think.”

  1. Ziggyinc Says:

    Are we also going to see huge heat sinks attached to our lamp posts? I know that my little flashlight with a single hi-intensity led in it gets noticeably warmer with constant use.

  2. TwoHedWlf Says:

    I don't think, in this case, noticably warmer is going to be a problem. Though they'll probably have to design it for greater cooling than the sodium lights will have, they don't much care if they're really toasty hot. They'll have this great big lump of metal lamp post to dump heat into though.

  3. FuzzyPlushroom Says:

    I think you're right, Dan - why don't they just worry about traffic lights for now? We've still got the ridiculous incandescents in my small town, for example (New Hampshire, US).

  4. Changes Says:

    I hate incandescents with a passion and I'm a huge fan of LEDs, CCFLs and really any other illumination method, but there is one good thing about having hideously inefficient incandescent bulbs in traffic lights: if snow blocks the window, the heat from the filament will eventually melt it off.

  5. phrantic Says:

    And why are they still round? Art Lebedev has proposed square traffic lights that seem to make more sense. More surface area for the same amount of space.

  6. TwoHedWlf Says:

    I'd say they're still round because all others are still round. I don't think it would be a very good idea to just slowly change them over, you'd want at least consistency of shape within a city. People might not immediately notice square lights when they're used to round for example and end up running through them. So you'd have to replace an entire city worth of lights simultaneously...Imagine the chaos that would cause.

  7. Changes Says:

    Never let it be said that I trust humanity's intelligence, but I think even the dumbest people of our species would behave the same whether the lamp was square or round...

  8. TwoHedWlf Says:

    Sure, eventually. But the way your brain works while driving is that you don't look closely at much at all. You drive along glancing towards important areas and it's a matter of pattern recognition to alert you to anything important while ignoring anything that isn't. In this case you're likely to get a lot of people glancing towards the intersection and their brain instead of going "Round things! Traffic lights! Stop!" it'll go "Square things! Irrelevant."

    Now if it would mean a couple minor crashes and handful of panic stops at intersections over the first couple months or dozens fatal crashes, I don't know.

  9. Changes Says:

    Your argument isn't without sense, but I've seen some square traffic lights already (can't really remember where, but maybe it was in Zagreb, Croatia), and there weren't pieces of glass and metal everywhere, so I can only assume the end result is pretty much che same.

  10. awollangk Says:

    Two things--

    Thing the first:

    The main flaw in TwoHedWlf's arguments are that the shape of the lights are not the only clues that indicate traffic lights to drivers. If they went to a single light that just changed color from green to yellow to red (There are individual LEDs more than capable of doing exactly that) his argument would have more merit that it might not be interpreted as a stop light. Any arrangement of three lights with a lit red light on top, a lit yellow light in the middle or a lit green light on the bottom will most likely be interpreted as a stoplight. Also moderating this is the fact that people tend to look for traffic indicators when approaching an intersection. The only times I don't actively look for indicators is when I've been through the intersection enough times that I know exactly where to look. In both of these situations if the lights are square instead of round (or even if there is just one light that changes color) I am unlikely to be thrown. I'll notice the difference, sure, but it won't create an unsafe situation.

    Thing the second:

    Luxim has come up with a light that has achieved 140 Lumens per Watt (http://news.zdnet.com/2422-13748_22-192842.html?tag=nl.e539) and they're promoting it as if it's a pretty big deal. It's not an LED, but the hoopla is pretty convincing and the bulb is REALLY small. About 16mm long by 8mm round. They were running the one they had at about 250 Watts which suggests it was putting out about 35,000 Lumens. They had it side by side with a 400 Watt street lamp and it was significantly brighter.

  11. jeroen8 Says:

    The lumen/watt values mentioned here are more the theoretical values which have been measured in optimum conditions without taking in account the loses reflection.

    On the renewable energy OliNo.org website they have measured the real lumen output of both a Low-pressure sodium lamp and a new state-of-the-art LED street lamp. This LED street lamp is using the newest Cree 100 lumen/watt LEDs.

    Look here for the results:

    The Low-pressure sodium Indal Industria Aurora Streetlamp

    Measured value: 67 lumen/watt

    The LED Streetlight Lioris Aduro 52

    Measured value: 64 lumen/watt

    As you can the see the real measured values are a lot lower than the theoretical values. The LED street lamp has almost the same efficiency as the low-pressure sodium lamp.

    The advantages of the LED lamp compared to the low-pressure sodium lamp:
    - The light of the LED lamp more evenly spread on the road when multiple lamps are combined together with a fixed distance between them (no hotspots).
    - The light of the LED has a wider colour spectrum, so objects are better visible.

  12. Chris L Says:

    Do you know anything about "induction" street lighting? My town is in the middle of replacing the nasty orange sodium lamps with them, and they do look nice and white, and are supposed to pay for themselves. They are supplied by this company: http://www.luxlite.com/

  13. Popup Says:

    Talking about R-G-B LED configurations... Have you seen the Philips 'living colors' lights? (There are plenty of youtube videos around) A set of powerful R-G-B LED's in a cool package. (Comes in transparent or glossy black or white orbs. I think I've even seen it with an ipod dock.) You can control intensity and color with the wireless remote.
    The only drawback is the price. (It's around USD100-USD300 depending on size.) I can't wait for a cheap knock-off at dealextreme. (And if possible, with a USB interface to control the LEDs from an computer!)

Leave a Reply