SOLAR FREAKIN ROADWAYS! We've all seen it plugged. And we've all either cheered it on or smacked our head in the ridiculousness of the proposal. Some of us, myself included, doing both.
It's a utopian vision. There's vastly more road surface in the world receiving solar energy than would be needed to power all of human society. So we just replace our roads with solar freakin roadways, and the climate problem is solved without using any more land. No need to get every house retrofitted for a custom-designed solar panel array installed by hand with its own transformer and all sorts of overhead, you just pave out your solar roads en masse whenever a new road is needed or an old road needs to be repaved, and viola, climate crisis solved and human impact on the planet vastly minimized. And think of all the side benefits. Optionally batteries in the road storing excess between day and night! LEDs providing customizable traffic messages and illumination! Heaters melting inches of snow away in minutes! Piezoelectricity capturing even more power! The proposal is incredible.
The proposal is also, as it stands, ridiculous. The goal of this article is to remedy that.
How realistic is the current proposal?
Let's start with addressing some of the arguments that have been made against the concept that aren't actually valid.
1. "Glass is slippery" - which is why you don't use smooth window glass, you use "traction glass" / "anti-slip glass", which is textured and can be made as rough as gravel. The texturing has a small but not problematic impact on light transmission, and can in some cases help with self-cleaning. In very high traction anti-slip glass the surface texture often ruins optical clarity by distorting the image, but solar cells don't care about image distortion, just transmission.
2. "Glass breaks" - it is amazingly difficult to break a thick pane of resistant glass lying atop a hard surface. Glass has a very high compressive modulus. Its weakness is in flexure, which is why you have joints between panels and don't make the panels too large. Some large buildings even have glass support exterior loads, in much more difficult circumstances than a road.
3. "Broken glass will scatter the road with dangerous shards" - Glass is not a single product with a single set of properties. Laminated glass retains the shards in roughly their original positions after breakage (which is why it's used in, for example, car windshields).
4. "Glass scratches" - which is the purpose of anti-scratch coatings, to raise the hardness above that of quartz and thus prevent scratches from all common materials (sand, steel, glass, etc)
5. "Cars shade the roads" - Go to Google Maps and zoom in on random roads around the country, you'll see that the impact is quite minimal, to the point of near irrelevance on most roads.
6. "Some roads go through forests or other shady spots" - which is why you don't begin with those roads.
7. "It's more efficient per unit area to angle panels" - There's no shortage of "unit area" in roads, so all that matters is economics (covered below). It's also more efficient per unit area to put panels on heliostats than on fixed roof slopes, but you don't see that stopping people.
8. "It makes more sense to put panels beside the road" - not if your goal is minimizing human impact (aka, surface area). Additionally, one of the benefits of a solar road is that you get "two things" (a road and a solar farm) by building only one thing (a solar road). Building a second solar farm next to a road is an additional step which loses this benefit.
9. "You don't know how this will work out in the real world" - which is why one builds pilot projects, which is what people are seeking. That's the whole point of a pilot project.
And of course let's cover the unspoken reasons for solar roadways:
1. "Hardware costs are less than overhead." We've reached the point where installation and other overhead costs are now greater than panel costs. With current trends, we can expect that in the future, installation costs will be vastly greater than materials costs. Hence it's perfectly fine if by building a solar roadway your materials costs rise and your efficiency declines, so long as you reduce the more critical installation costs.
2. "Home roofs are really pretty bad if your goal is to reduce the installation costs." They're at altitude, on top of a not-nearly-as-solid-as-the-ground structure, require custom design for each home and a lot of other corporate overhead, very difficult to automate, require an transformer for each home, and so on down the line. By contrast, it's hard to envision any way to get lower installation costs than laying down bulk panels from an automated truck in a straight line. And while there is some overhead (permitting, right of way, etc), it's the same overhead required for any road, and per unit area, it's orders of magnitude less than for houses.
But there are some quite legitimate arguments against the "Solar Freakin Roadways!" proposal.
1. "Starting with high traffic roads in cold climates versus, say, sidewalks, driveways, access roads, etc in temperate climates, is not exactly starting with the low-hanging fruit" - No question there.
2. "The 'melting snow' concept just doesn't add up, it'd cost a dollar per square foot to melt a couple inches of snow at 100% efficiency, when it only costs five cents a square foot to plow it" - Absolutely correct.
3. "Capturing piezoelectricity means having your panels give to some extent, which means you're just stealing power from the inefficient, polluting cars driving on it, which effectively must continually drive upwards to counter the give." - By and large, correct.
4. "All of the electronics they plan to put in there would make the panels too expensive even with mass production, and would require panel replacement whenever they fail." - Probably true.
5. "Bright LEDs in particular are very expensive, and its far more efficient to just have light bounce off road paint than to capture it with PV cells then turn it back to light, thus losing over 90% of the energy." - Yep. Not to mention that you can still let whatever wavelengths through the paint that you don't need to reflect rather than absorbing them.
6. "They really hand-waved most of the math and economics, didn't they?" - Yes. Yes they did.
So - where does that leave us?
An alternative propsal: Realistic Solar Roadways
First off, we need to accept that we must tackle the low-hanging fruit first. There's no shortage of new road construction, you don't need to start with something designed for highways in Idaho. Your best examples would be, for example, sidewalks or access roads in warm, sunny climates that get little traffic. We will analyze below the complex case, but the first actual projects would be much simpler tasks.
Secondly, we need to toss piezoelectric generation. Our goal is to minimize fossil fuel consumption and pollution, not to increase it.
Third, we need to throw away the concept of melting inches of snow and ice. It's just doesn't add up, financially. Instead, I propose an alternative solution. Instead of having drainage channels built into the cells to in-cell drains as the "Solar Freakin Roadways!" team proposes, I propose to handle drainage on the surface (as with a regular road), and instead use airflow that can leak out from over the cells through small (0,7mm) holes at regular (22mm) intervals, creating a very weak "air hockey table" effect over the road. This would be an amount of pressure effectively irrelevant to tire traction (a tiny fraction of a psi, versus the dozens of PSI a car exerts), not even nearly enough to slide a 20 gram air hockey puck, but enough to keep 3 milligram snowflakes from settling. As anyone who's driven in snow before knows, snow doesn't settle when it has to overcome air blowing against it, it flows to the side - even driving at low speeds is enough to keep snow off your car. But when you stop and there's no longer a cross-current, it begins to accumulate. Panel heating would only be used sporadically to remove thin layers of ice should any form. The ability to pump air through the panels would also provide additional solar cell cooling, which raises efficiency. The small hole size (and potentially a hydrophobic interior coating) would prevent water and dirt ingress. More on the calculations behind this proposal in a moment.
Fourth, we need to greatly simplify the panels. Storing energy in little batteries in hard-to-access areas is just a bad idea, and should be tossed - leave it to grid operators to handle that in a much more efficient manner (pumped hydro, peaking plants, flow batteries, whatever). The per-panel control electronics need to go too. The panels need to be simple, passive structures to minimize costs. All more complex activities, such as fans, transformers, and so forth should be done in periodic roadside utility stations with convenient access for maintenance.
Fifth, we must consider real-world constraints. We need to throw away the hex panels and go with simple rectangles. Hex panels are cute, but in the real world we deal with straight lines instead of zigzags - everything from conduits to shipping containers. Our panels should be 50 centimers by 60 centimeters (roughly 1 1/2 by 2 feet), bundled in stacks of four on a standard 1016 x 1219 mm pallet, 18 panels high, to be shipped 20 pallets per standard iso container. Our panel laying truck should receive pallets as-is, forklifted straight into its bed, and feed directly from them. If our goal is to minimize costs, we can't be having tons of wasted space or involving lots of manual labor at any step.
Now, let's diagram what we're needing here. First, your basic panel:
The cells are wired in series which connects in parallel to a primary lateral conduit, which also allows for airflow between the panels. All current here is DC. For the sake of this design, we will assume that the operating panel voltage is 600V, achieved by running 1200 cells together per panel in series (for the purposes of illustration, we're showing them as boards pre-wired to 40V). 600V is much higher than is used in home rooftop installs, but typical for PV usage with centralized transformers (some installs are now upwards of 1kV). Between the panels, the lateral conduits are kept roughly pressuretight by rubber seals, and the electrical wiring in them connects by snap connectors.
Due to our high voltage, we can use very fine wire, even as low as 40 gauge (0.08mm) if so desired, between the cells without encountering significant loss; our only constraints are on how rugged we want the wires. For our calculations, let us assume 30 gauge uninsulated copper wire (0.255mm, 0.338 ohms/m) between the cells and 20 gauge insulated copper (0.812mm, 0.0331 ohms/m) in the lateral conduits. Waste heat is current squared times resistance. With perhaps 125 centimeters wiring between our cells and an assumed peak cell efficiency of 10% with 90%++ of the panel covered in cells (27 watts, aka 5.4 watts per bank), running at 600V (aka, 0.045A) our resistance works out to a heat loss of 0.00034W in the cell wiring, aka, a mere 0.006% of the generated power.
(++ Ideally we'd like more than 90% cell coverage, but let's be pessimistic)
For the lateral conduits, if we assume that our road has four lanes paved plus enough shoulder space to count as a fifth, and an average lane width of 4m, then we have a road that's 20 meters wide made of 40 panels (1080W per row of panels), meaning an average of 20 meters run length for a full circuit. If we repeat our above calculations for the 600V in the lateral conduits at 20 gauge, we get 2.14W of waste heat, aka 0.018% loss. Hence even with these very thin wires, our total losses are less than two tenths of a percent at this stage.
All lateral channels connect both power and air to a primary conduit on the side of the road, which handles air flow, two-wire DC transmission, 3-phase insulated AC transmission, and includes a fiber-optic line (a small extra cost that can be rolled into the manufacture to provide net connectivity not just to your utility stations, but to the general public):
The DC lines only run to the next station in each case, which are spaced 250 meters apart; we assume 4/0 gauge insulated aluminum wire (11.7mm / 0.0000272 ohms/km) for this. With an average run length of 125 meters in each direction (250 meters total) at 600V and a peak power of 450kW per segment, we get 3.825kW waste heat, or 0.85% loss. So far our losses are under 1%.
Each of our utility stations then turns this maximum 450kW per segment into AC for distribution. The ideal voltage is difficult to decide, as it depends greatly on your use case (urban with immediate distribution or very long distance transmission out in the middle of nowhere). We'll assume a suburban-to-rural case with an ideal transmission voltage of 20kV and average transmission length of 15km on 20mm insulated aluminum wire. Beyond this length, power should be stepped up to higher voltages for long-distance transmission. Our sample 4-lane + shoulder road generates a peak of 1.8MW/km, so with an average AC transmission length of 15km we need to transmit 27MW, which at 20kV is 1350A, or 450A per phase. The AC resistance on our wire at 10km is 1.29 ohms, yielding a loss of 260kW or about 1%. This brings our net total to 2% loss. One can play with these numbers to customize them to a particular situation; doubling the average transmission distance doubles conductor diameter, while doubling the voltage halves it but doubles the required insulation thickness.
Lets use bulk industrial manufacturer prices from Alibaba.com. Our net total of wiring per kilometer of road is approximately 360 kilometers of uninsulated copper 30 gauge (about $5k), 70 kilometers of insulated (600V) copper 20 gauge (about $5k), 2 kilometers of insulated (600V) aluminum 4/0 gauge (about $5k), and 3 kilometers of insulated (20kV) aluminum 20mm wire per kilometer of road (about $15k), yielding a total of about $30k per kilometer. One could save money by going with aluminum wiring in the panels and conduits instead of copper.
Adding a fiberoptic bundle in the primary conduit will add perhaps $10k per kilometer. Fiberoptic lines are not particularly expensive, most of the cost of laying fiber is the act of laying it.
Next, our utility stations (whether above the surface or buried) perform a number of different functions, which I will briefly go into:
First off, we have our primary task: the 450kW, 600V DC to 20kV AC transformer. Such a transformer is a box roughly about 1 cubic meter; in bulk this should be about $6k per unit.
Next we have our fan unit, for cell heating/cooling, road dust removal, and snow accumulation prevention. Given a road + shoulder surface area of 250m * 20m = 5000 m² per box, and the desire to have air coming out of holes comprising 0.1% of the panel's area and moving at 5 meters per second, this yields a flow area over a panel of 0.005m² and a flow rate over the current segment of road of 25 m³/s. A bulk-purchased industrial blower of this sort is an approximately $2k box comprising about 2 cubic meters and requires about 25 kilowatts to run at full capacity (only about 5.6% of the peak output of the road, and only needed either during snowfall/blowing snow or when cell cooling calculations suggest it would be a net benefit). A 50kW blower-integrated electric (or otherwise) heating element sufficient to remove thin ice coatings on the road (not to melt inches of snow) would add perhaps $1k to the blower's cost. Heat within the system must be maintained at temperature sufficiently below that which would be hazardous for the system components, such as wire insulation and solar cells. Operating the heater, assuming 500 joules per gram of ice melted (versus the minimum 334 required by physics) would take about 700kWh per millimeter of ice per station (1 1/2 hours of peak road output per millimeter of ice, plus a bit extra to make up for blower consumption; about $300 per kilometer at residential power rates). The blower and heater could be omitted for the utility stations for roads in non-snowy climes, or even abandoned altogether in snowy climes in favor of plowing. If a blower is present, a cooling unit could be optionally included to increase cell efficiency if determined to be economically beneficial overall; however, just blowing outdoor air through the panels will usually provide some degree of cooling, especially on bright summer days when the panels heat up the most relative to their environment.
More about cooling: solar cells are sensitive to temperature, losing about 1% efficiency for every degree celcius they rise. If by blowing a small amount of outdoor or chilled air through the cells one could lower the operating cell temperature by 10 degrees (C) or so, they could gain a (roughly) 10% efficiency boost, well more than the energy needed to run the blower. Whether such flow rates could achieve such cooling is a complex problem to model and will not be attempted in this article; I wish only to point out that the possibility exists.
Continuing onward: The utility box includes a weather/monitoring station to assess current conditions - snowfall, rainfall, temperature, wind, icing, vehicle counting, whatever is determined to be useful for either the road itself, traffic planners, meteorologists, drivers, etc; in bulk, as part of a utility station, this hardware would probably run about $500. The station also has a small central control computer controller, probably nothing more than a $100 unit (its job isn't very complex).
Each utility station has a fiberoptic access point to provide itself net access and to provide net access to local consumers. In addition, the station provides a connection point for local consumers to get electricity (aka, to connect a line to local step-down transformers). All of this, plus the housing shed, etc probably adds another $10k or so to the total cost. If one wishes to bury the substations (so that only a cooling air intake and possibly feeds to local power distribution are visible and noise would be minimized), excavation costs must be added. This, however, should not be a large percentage of the total cost, as one can hire a backhoe and operator for under $100 an hour and it shouldn't take more than couple ours of digging (plus, of course, overhead). It would likely however increase the cost of the station's housing as well. Having them underground would probably increase the costs by about $5k per station. Let's just say that our total net cost is probably in the ballpark of $22.5k per station, or $90k per kilometer of solar road.
This is for a 4-lane road with utility stations spaced every 250 meters. We could reduce the size of these utility stations and their hardware by spacing them more frequently, which would also reduce the diameter of our DC conductors. A 2-lane or smaller road would have smaller station hardware requirements, while a larger road would have larger requirements.
Now, let's take a closer look at our panels, in comparison to our sample road (4 lanes, shoulder, surface area of 20k m² per kilometer).
The base layer on which everything is built is simple 3mm aluminum sheeting, which is about $10 per square meter, which works out to $200k per kilometer. This has the side effect of reflecting any non-absorbed light back for a second pass at the cells, which should provide a small but relevant boost to their efficiency. It will also help draw away heat from the cells and lose it to the ground, which further increases their efficiency.
Next comes the most important and expensive layer - the solar cells. These are wafers, not panels, so they're cheaper than panel prices - about $0.25 per nominal watt in bulk. At 15% nominal efficiency this works out to around $40 per square meter. Assuming 90% coverage the total cost here is $720k per kilometer.
Glass is the second most expensive element. We'll assume 1.5cm thick float glass, laminated on one side, anti-slip on the other, with an anti-scratch coating on the anti-slip side (note that the drainage channels, air holes, interlocks, etc will need to be cut before the anti-scratch coating is applied, but we'll cover that in manufacturing later). The base anti-slip laminated glass cost here is approximately $8 per square meter, hardly more than the price of regular laminated glass. So scratch-resistance seems to be the price-dominating factor. High-end anti-scratch glass, like the gorilla glass that's used in cell phones, costs about $60 per square meter, but we don't need anything that extreme; low end anti-scratch glass of this thickness is around $16 per square meter and it should suffice, so let's just say $18 per square meter all together after accounting for anti-slip surfacing. Our total here is $360k per kilometer.
Our ends will have to be sealed with a synthetic rubber to keep the panels watertight and to provide a pressuretight seal where the air channels meet. Let's say $30k per kilometer.
Our cells will fit together with rubber-coated aluminum pins that slots into the glass. Let's say $30k per kilometer.
Our electrical snap connectors for the power cables will probably add another $30k per kilometer.
The primary conduit on the side of the road needs to be rather large and sturdy, probably about $10 per meter. So $10k per kilometer
Our total thusfar is approximately $1,5m per kilometer. Next we must add manufacturing costs and any shipping/taxes required during manufacture. While it's tempting to dismiss them as insignificant, they're not, and there's reason in particular here to suspect that they'll be a significant portion of total costs - for example, the need to waterjet cut holes, channels to the holes on the underside of the glass, drainage channels on the exterior, interconnects between panels, etc. In mass production it'd be custom hardware designed to put the same pattern into each piece at the same time, not an expensive one-off cutting process, but it will still not be "cheap". The utility stations also need some assembly, although it'll be a lower portion of their total costs. I'm tempted to say that overall these manufacturing and other associated costs could more than double the overall cost per kilometer, despite the fact that we're using some moderately pricey components (solar cells, scratch-resistant glass, etc). So let's say our total suddenly jumps up to $3.2m per kilometer, with the largest component being the glass cutting, and the second largest being the solar cell wiring.
The tacky asphault surface on which all of this is laid will be accounted for in the roadbuilding process.
Now it's time to look at how we lay the road:
Just like regular roads are laid down by machines custom-designed for it in order to minimize costs, so should our solar road. Pallets straight out of the shipping container are delivered to the site, their bundling plastic is removed, and then they're loaded up in rows onto the paver truck. On the back of the paver is a ramp, which is maintained at the same height as the panels it's feeding from (this means that its angle declines with time, but is always steep enough for the smooth aluminum underside of the panels to slide down it). The paver does a final smoothing pass on the pre-laid road undersurface while tiny wheels at the back of the ramp keep it aligned with any adjacent panels (if any exist) that it might be connecting to. The end of the ramp is staggered so that the panels slide into place one after the next, allowing their pins to slot into place, and a wheel presses them together. Panels are pushed onto the ramp by adjustable high traction rollers, driven by the driveshaft to ensure a proper feed rate. One worker drives the paver, while one walks beside it at all times to make sure that everything is slotting into place correctly and take care of any problems that may arise. On one side of the road, the primary conduit is laid down at the same time, also pinned into place as its unreeled; a small blade digs a channel for it to slot into in the road undersurface, and buries it under at a reasonble depth after it latches into place. On the other side of the road, a seal is applied to the air channels to seal them off. On both sides it may be desirable to create additional structures - curbs, drainage gravel, etc, depending on the situation, as per with a conventional road. While the paver is laying down the road, a loader drives back and forth to fetch additional pallets of panels to load into the paver. An additional team follows the truck afterwards, installing the utility stations.
It's hard to say what the cost of a roadlaying process that does not yet exist will be; however, the overall process and hardware complexity looks similar to that of building traditional roads; it is, in short, a regular road with an alternative, more expensive top layer. Hence we should use traditional road costs per kilometer in our calculations (perhaps $3m per kilometer for such a road). Note that these traditional roadlaying costs also include surfacing materials that we're not using, as we're using the panels instead. We'll just eat those costs, saying that they're equivalent to whatever extra overhead exists for the utility stations.
The panels, as mentioned, come from the manufacturer on standard pallets. If we assume our panels stack 3 centimeters high each, then we can stack panels 18 high (half a meter) in four columns per pallet. Assuming an average panel density of 1500 kilograms per cubic meter then our pallets weigh 972 kilograms, under the 1000 kilogram max for standard pallets. At 20 pallets per container, our total load is 20 tonnes, below the 22 tonne standard maximum. The 1440 panels per shipping container pave an area of 432 square meters, meaning that we require 46 containers per kilometer of road. At a cost of around $10k per container (international shipping plus taxes), this works out to a panel shipping cost of about $460k per kilometer of road. To this we should add perhaps a third of that for domestic shipping, bringing us up to around $650k per kilometer of road. Let's add in another $150k for international and domestic shipping of the utility stations.
Our net figures, thus, are around $7m per kilometer of solar road, versus $3m per kilometer of conventional road, for a net surcharge of $4m (again, remember, the operating assumption is that it's built where a road was needed anyway, that first $3m would have had to be spent anyway).
Now, what about the generation figures?
As per above, we calculated a maximum generation per kilometer of around 1.8 MW. There's a couple percent losses in the road wiring, plus the transformer losses, plus the occasional blower or heater losses, plus whatever might be on the road that hasn't washed off / been blown off, plus glass transmission losses, plus shading, plus angle losses, plus clouds, plus night, and on and on. All of these things together make up the capacity factor. A typical capacity factor for rooftop installation is about 15%. While we do gain on some fronts, such as using higher voltages and more efficient, centralized transformers, we'll still end up with a significantly lower capacity factor, probably in the ballpark of around 9%. This means that we yield an average output of 162 kW, or an annual net of 1.4 GWh. At a net market rate of 10 cents per kilowatt hour (I think one could debate rates for this ranging from 5 to 30 cents, let's go with 10), that's an electricity sale value of 1.4 million per year, yielding a simple payback period of only 2.9 years.
What are our failure modes?
We should analyze what significance various system hardware failures would have, as one can always be assured of failures somewhere.
A short in a panel can eliminate the generation of power from that panel and to some degree other panels connected with it, turning it to heat. However, the low gauge of wire limits the potential loss from this mechanism; too much current flow means that the shorting circuit will melt through. Fuses could be added to handle this more cleanly.
There is no realistic fire risk from the road or lateral conduit, as there is little flammable material. The thick laminated glass and thick lateral conduit walls make the risk of electric shock low in the case of breakage, and the aluminum panel underside grounds the road. The primary hazard would be if the primary conduit was dug up and damaged, as there is a medium voltage (20kV) high capacity AC line in there. However, such buried lines exist all over the place and are not a hazard unique to solar roads.
A breakage of the primary conduit will take out all generation from a 250m segment due to the loss of its DC line. As for the AC, unlike when a regular transmission line is broken, it does not automatically imply that customers go without power, as generation is conducted on both sides of the breakage. Whether any outages occur depend on the local consumption versus local generation.
Any hardware failures in a utility box will be automatically reported, using their fiber connection. The most serious failure would be the failure of the station transformer, which would cut off all generation for that 250m segment of road. Fire is possible, although not particularly likely, and should be handled in the same way as any other fire in medium-voltage electrical equipment.
Now, what's the significance of all of this?
One can, and should, dispute the various numbers above. They're very rough, and there's no way to know for sure how everything would play out in the real world without a more detailed investigation and then a pilot project. But the fact that with what I felt were at least fair assumptions, we get a very reasonable price and payback period for a solar road.
Why do the numbers perform so well against a rooftop install? Two main reasons:
* The overhead is so much smaller than a rooftop install; everything here is done in bulk, from purchases to installation. And as time goes on, overhead will increasingly be the dominant factor in rooftop installs. That's not to say that roads don't have their own overhead - they do, and it's significant. But nothing compared to the sort of per-house, custom, hand-installation overhead you get with rooftop solar.
* The solar cell wafers are only a fraction of the cost of a solar panel; the rest is itself overhead - associated hardware, assembly, shipping, distribution, taxes, marketing, profit, etc. The fact that the solar road has a less efficient use of wafer and other materials per watt, and higher shipping costs due to the increased mass, indeed increases part of the costs per watt. However, it's not all of the costs per watt.
Why do the numbers perform well even versus a PV solar farm (something that has to use up land that otherwise would have been left wild)?
* A large part of the cost of any project (solar farms included) is overhead - permitting, right of way, environmental studies, contract negotiations, preparing the land, drainage, and on and on. It's a large part of the chunk of the cost of building any large project. But the key here is that a road would have had to have been built anyway where we're building the solar road - hence, these overhead costs are already paid for us. The only real difference is what sort of top layer we choose for our road, one that makes power or one that doesn't.
* The solar farm has some advantages over us, in that they can use cheaper panels and mount them at a more optimal angle or on heliostats (although this comes at the cost of mounting hardware, significant costs in the case of heliostats, and leaves them more vulnerable to weather, which increases maintenance). They're also less prone (but not immune) to dust, debris, and shading. However they're also usually built out in the middle of nowhere to get cheap land and avoid bothering people, meaning you have to build a high power transmission line to move all of the generated power, and, yes, you need to build an access road. It's ironic, but in some cases with small/remote solar farms, the surface area of the access road can be greater than that of the panels, and by making the access road a solar road and piggybacking onto its construction costs, you could actually get more power from the road than from the farm itself! A 30km, 4m-wide solar access road would have a nameplate yield of 120MW and a 9% capacity factor, equivalent to a 60MW PV farm with an 18% capacity factor.
So in summary?
Solar Freakin Roadways!? No.
Reasonable solar roadways? Yes.