As much as I want LED, solar powered street lights, I keep thinking about a problem many southern dwelling designers never considered.
A few years ago -- far before LA's switch -- Wisconsin tried to switch to LED bulbs for power savings in their existing light fixtures. It was working great until a front of wet, sticky snow came through. Snow stuck to the fixtures, eventually covering the light and making it useless.
Why wasn't this a problem before? After all the light fixtures were the same and worked for years without this problem. The reality is that the inefficient bulbs in the fixtures gave of a lot of waste heat. Heat, that melted the snow and allowed the street light to illuminate the surrounding area uninterrupted.
I've wondered for years if there was a solution to this problem that we haven't considered. The simplest is that the shape of the fixture is the problem. Taking advantage of different materials (perhaps even hydrophobic) might reduce the problem. Changing the heat sink design on the high-powered LEDs might help redirect the waste heat more effectively.
It's still surprising to me that up to 40% of a city's municipal expense comes from streetlights. I'd love if we could power all of those off of solar panels installed on the light itself. The lights would power themselves, continue in to operate in the event of a power outage, and shut off early in the morning once power is exhausted. Of course, the problem with all of this is something designers don't really like.
Surface area. Solar panels thrive on surface area and need a considerable amount in order to support a high or low-but-long-term current demand. Surface area also plays havoc with snow, as anyone that's had to rake a roof in winter knows. It becomes a collection point for snow, covering the panel and halting power collection.
I could argue that future technology advances such as flexible or spray on panels will minimize such considers. As a tinkery-type person, I don't want to rely on magical R&D fairy dust of the next tech advance to come. It might not. It will often undeliver. We're far past the point in time where we can make huge tech leaps with each iteration. We've already done the easy stuff.