I'm glad you brought up the energy required to heat the panels. I'll copy a calculation I made the other day about this.
It takes 330 Joules of energy to melt 1 g of ice (if it starts out at 0C, more if the ice is colder). Lets say in a snow storm (with the large snowflakes) the average mass of a snowflake is .02g. That means the energy required to melt a single snowflake is 6.6 joules. To generate 6.6 joules of energy, you would need 6.6 watts of thermal energy for 1 second. Solar Panels produce about 130 watts per meter square under peak conditions.
Lets assume that in a 1 square meter area (9 square feet, or about the size of a small coat closet) 10 snow flakes land every second in a snow storm. That means that if 100% of the electricity consumed is perfectly converted to heat (which it wouldn't be) and 100% of that heat was transferred to the snow flakes (it would be hilariously less than 100% in reality, you still have to warm up all that air), then you would need 66 watts of electricity per square meter continuously drawn to keep the road just barely thawed. And that's assuming its just below freezing outside. That is HALF the electricity that those solar panels would produce at high noon in the middle of the summer without a cloud in the sky. It would take nearly a megawatt of electricity with 100% perfect conversion per mile. If you had 50% conversion, which is still insanely high, you would just barely break even if it snowed at high noon in the middle of august and had perfectly clean panels.
If you get a heavy snow storm (around a foot), you would have to melt over 30,000 grams of ice per square meter, which would take over 10 MJ of energy, or 2.75 KW-hrs per square meter of road. If that snow falls over a period of 10 hrs (just 1 inch per hour, a blizzard can produce snow at 3 times this rate), your solar panels need to contribute 275 watts of electricity for the entire time with 100% efficiency. That is more than what they can deliver under optimal conditions. If solar panels were that efficient we would be able to power the average house in America using just 20 or 30 square meters worth of solar panels on the roof. The maximum snowfall rate that these panels could handle on their own (assuming all of the energy generated gets transferred strictly to the snowflakes) is about a half inch per hour. When you factor in how much of the thermal energy would escape into the atmosphere rather than the snow it becomes clear just how inefficient the process is.
In the northeast, the roads would HAVE to be plowed. Every time those roads get plowed the glass gets scratched up and less light is able to find its way to the solar cells. Even if the glass holds up in a way that it is still safe to drive on, how much power will these super expensive roads produce after a few winters? Part of the claim of this project is that they will eventually pay for themselves, but they can't do that if they aren't producing electricity. They can't have put ANY thought into this claim, which makes you wonder how much thought they put into the rest of this project.
Just look how opaque the glass the glass is from where they textured it to give it suitable traction properties.
Nobody can seriously think that this sheet isn't blocking a ton of sunlight from reaching the solar cells. And this is from a nearly straight on point of view! What happens when it is scratched up from wear and tear in the mornings and evenings when incident angle of the light from the sun is shallow and even more light gets reflected? This is just one of those projects that make sense when you don't think about.
EDIT: I really shouldn't beat a dead horse, but I have another thought. As you can see from the image below I'd say at least 1/3rd of each tile is exposed PCB.
That is a ton of wasted space and would reduce the power produced per square meter by 1/3rd. Makes you wonder why they chose hexagons doesn't it? Maybe it serves some structural purpose, but I don't see anything explaining why he changed shape. His phase one prototype seems like a more ideal shape to me.
