On the shroud over each light, a number of LED/photodiode pairs should be placed. The LED can be blinked every few seconds. The photodiode will detect significanly more light during the blink if there is snow packed into the shroud. If the controller detects packed snow it turns on a heater to melt it. Similar LED/photodiode pairs are used to detect paper in copiers and other similar situations.
Even though this may sound complex, the parts required are cheap, especially compared to the cost of the traffic light as a whole.
Of the ~40000 traffic fatalities in the US each year perhaps 0.01% are due to this problem, but the fix is easy.
FTA> Heating up the LEDs would help defeat the purpose of saving energy.
This will only be true if it snows every day! Or are they really proposing switching bulbs twice a year?
(There are plenty of people in MA who keep their incandescents year round 'for the free heating' - ignoring the fact that they have their air conditioning on for 6 months a year. People can be perplexing at times.)
And ignoring the fact that the heat wouldn't really be "free". It would be generated by excess energy consumption.
Electric heat (in any form or format in homeowner applications) is generally the most expensive way to turn a given quantity of dollars into a given amount of heat. The only MORE expensive option is burning the dollar bills themselves to produce heat.
No, use a thermostat. It is much simpler, less prone to errors, you only need one per light housing, and the energy cost required to keep the shrouds above 34 degrees is quite insignificant.
You could probably get away with having one per block; aren't most traffic lights all connected? Or have I just been watching too many movies where hackers take over streetlights?
Thanks, you beat me to it. The sensationalism in this article is incredible.
This is a simple engineering problem that does not justify an article titled "The Unintended Negative Consequences of LEDs". Put in a coil connected to a thermometer to melt snow if necessary. Problem solved, and energy is still conserved.
This article got written because economists have a limited vocabulary for talking about design.
If you build a machine to solve a problem, and the machine doesn't act the way you want it to, that's a bug. Fixing bugs is easier than starting over would be. Bugs that make you wonder are called design flaws.
You don't get unintended consequences until your system works, and things change in your environment (or maybe with your users) because of the way it works. Unintended consequences are hard to fix because they only show up in systems that are already doing what you want.
There is a good point in the comments that isn't made in the article. Unlike a traffic light outage due to power failure, the ice buildup may only affect one of the signals, so the conventional rule to treat a dark traffic signal as a four way stop fails. One person may see a green light, while another driver on a perpendicular course stops at the affected light, but then enters the intersection because he expects the other driver to stop.
You should never enter an intersection "expecting" another driver to stop under those kind of circumstances. After watching the driving abilities of the average American, I would actually wait at the intersection until no cars are in sight to continue on.
I had exactly this scenario happen years ago. I was driving to a four-way stoplight and the light was snowed over (only in my direction, as it turns out). The cars on the side road were stopped. I slowed to a near-stop and was trying to figure out what to do - some people coming from the opposite direction were going. Sure enough, the side traffic started entering the intersection and I tried to come to a complete stop and (in slow-motion) slid on ice into a car at about 5 MPH, who did not even bother to look to the sides, presumably because his light was perfectly clear.
Maybe they should sell led traffic bulbs with resistive wires to generate heat, like the rear-windshield defroster. They could be switched on during snow. Still saves a ton of money/energy.
It's a well known and solved problem, most areas are not willing to pay extra to have it installed when it's only a problem for a few days a year or even decade.
This seems silly. "Heating up the LEDs would help defeat the purpose of saving energy". Not if you only heat them when they need heated, which I imagine is trivial to detect. Even then you've got a bulb that runs at the same energy usage during the winter as the rest, and then at a lower rate for 10 months per year.
Not even "during the winter". You only need to run the heater during active snow or ice precipitation. That's what, a few dozen hours a year in most areas?
It seems strange that all the "solutions"---heating coils, diodes, whatever---seem to presume that the basic form factor of the light system would remain unchanged. But wouldn't it be easier to solve if you just rethought the shape of things? Instead of a three-light assembly with hoods over each light (where snow can build up on top of the hood for the next lower light), just reshape the traffic signal as a box with the lights on the back, the four sides opaque (to block glare and peripheral view), and the front side transparent plastic of some variety? There are surely engineering issues there too (like preventing glare off the clear plastic), but it seems like a more stable system, no worries about snow accumulation, and it's probably cheaper to manufacture too.
well worth reading for it's analysis of all the problems with horses that the CAR solved (and, as always, it created new ones) -- like piles of manure 40-60 feet high in cities! Now that problem's limited to D.C.!!
This is really funny maybe U.S. infrastructure decision makers should buy european technology, because a simple combination of sensors(temperature and or light)/heating coils would solve all of these problems pretty easily.
> None of the proposed solutions are foolproof
is just simply wrong. Traffic lights will be upgraded with little heating systems (coil/wires) and everything will be fine.