Welcome to CPF, ED! I punched-in here to post the same link!
So much for not being able to find my happy a** with both hands and a flashlight! (Do not look into Tank Searchlight with remaining eye!)
Even though I believe that LED lighting is the way of the future, I think this is a bad idea at this time for several reasons:
1) Low pressure sodium lighting sits at around 200 lumen/watt. Power LEDs just aren't there yet. CREE just last month announced 129 lumen/watt for cool-white and 99 lumen/watt for warm-white. So from a power point of view, LPS still wins by a large margin.
2) Light pollution! Low pressure sodium emits only two primary wavelengths of light so it's largely monochromatic. This makes it trivial for astronomers to filter out. Contrast this to a wide-bandwidth light such as a mercury vapor light or the fluorescent coating of a white LED which emits many, many wavelengths. You cannot effectively filter this type of light pollution because you'd have to block so many wavelengths that you wind up blocking lots of 'useful' information about whatever it is you're trying to image.
3) Longevity. A quote in the article mentions 2 year replacement schedule for existing lights or 10-year schedule for LED lights. I think that remains to be seen. 10 years works out to around 45000 hours which matches up pretty well with the 50k hour lifespan often found in PR brochures. The 2-year lifespan of the existing lights seems a little low, though. LPS bulbs are usually rated between 15k and 20k hours which works out to around 4 years at the same duty cycle.
So if Ann Arbor is seeing LPS lights fail at twice the expected rate, it's not unreasonable to wonder if LED lights might also have shorter lifespans. So, it might make financial sense to switch from LPS to LED if they really do have a 10 year lifespan (the 5x longer lifespan might offset the higher electricity costs). But is the same true if the LEDs turn out to only have a 5-6 year life span?
(edited: the editor ate all my formatting)
Last edited by jrminky; 10-22-2007 at 11:16 AM.
Now let's look at LEDs. LEDs are directional. 100% of the emitted light goes to the street. The ballast losses are lower for various reasons, perhaps around 10%. So if you start out with LEDs at 100 lm/W (currently state-of-the-art) you can have an overall system efficiency of 90 lm/W, or nearly twice that of the sodium vapor lamp. Also, the higher color temperature of LEDs has a higher apparent brightness. This means you can get by with fewer photopic lumens if you want to, or use the same number of lumens yet have the street look brighter. Finally, LEDs are still on a steep improvement curve. They will soon pass the raw efficiency of HPS, and within 5 years, LPS.
Again, most cities (except a few in Arizona IIRC) are currently using HPS, not LPS. This is fairly broad spectrum so it just as difficult to filter out as LED light. This problem of light pollution can be more easily solved by just designing fixtures which don't allow light trespass. This is actually easier with LEDs than with anything else since they inherently emit in a half sphere (or less).2) Light pollution! Low pressure sodium emits only two primary wavelengths of light so it's largely monochromatic. This makes it trivial for astronomers to filter out. Contrast this to a wide-bandwidth light such as a mercury vapor light or the fluorescent coating of a white LED which emits many, many wavelengths. You cannot effectively filter this type of light pollution because you'd have to block so many wavelengths that you wind up blocking lots of 'useful' information about whatever it is you're trying to image.
There are good reasons for NOT using LPS or HPS. Both cause a loss of peripheral vision, making driving less safe. LPS has zero color rendering, making ID of vehicles by color impossible (one reason police don't like them). Both are asthetically horrible compared to some sort of white light such as LED or metal halide. I remember in the early 1970s when a lot of cities went from mercury to sodium lights they were almost universally hated.
Your other points are valid. To me though I think LED streetlights should (and can) be made to last as long as the fixture (50 to 200 years) by underdriving them. There will also be significant amounts of efficiency gain which will more than offset the cost of the required additional emitters. Come to think of it, you'll save on the labor of replacing emitters plus the cost of the replacement emitters themselves.