Rooftop solar is about as static as technology gets. You bolt the panels down, they point at the sun, and the most exciting thing that happens in a normal year is a hailstorm. They stay where you put them.<\/p>\n
There’s an array in the Swiss Alps that doesn’t. For roughly half of every year it isn’t floating at all. The water drains out from underneath it, 1,400 solar panels come down onto the bottom of the lake, and they’re meant to.<\/p>\n
The plant sits on Lac des Toules, a reservoir above Bourg-Saint-Pierre in the Valais, at 5,940 feet (1,810 meters). Romande Energie switched it on December 3, 2019, after ten months of construction and more than six years of studies. It’s 448 kilowatts, about what you’d put on a large warehouse roof.<\/p>\n
Because Lac des Toules isn’t really a lake. It’s storage for a hydropower operation, which means the operator draws it down and refills it on a yearly cycle, the way you’d cycle a battery. Where the array sits, the surface moves about 50 feet (15 meters) between full and empty.<\/p>\n
So the array isn’t fixed to anything solid. It’s a mat of aluminum and high-density polyethylene floats chained to weights on the bottom, and it rides the water up and down. Take enough water away and there’s nothing left to ride. Romande Energie’s report to the federal energy office puts the structure aground for close to six months a year. Compare that with a German floating array on a Bavarian lake<\/a>, where the water level barely moves and the panels stand upright all year.<\/p>\n What it lands on isn’t the natural lake bed either. Romande Energie built a pad out of fill dredged from the lake, close to three acres of it (about 12,000 square meters), standing nearly 23 feet (7 meters) above the bottom at its highest point. It was meant to give the floats somewhere flat to settle.<\/p>\n It hasn’t behaved. The fill has a wide spread of grain sizes, so it hasn’t compacted evenly, and parts of it have dropped by as much as 5 feet (1.5 meters). Structures built to rest on all four corners ended up hanging off two, which shoved their load onto the frames next to them. Andy Kaufmann, who runs floating solar development at Romande Energie, listed what that did in the Swiss electrotechnical journal Bulletin<\/a>: deformation, buckling failures, welds torn out.<\/p>\n Finding out that your foundation moves is a rough day at the office. It’s also the kind of thing you only learn by leaving a real array out there through four real winters, which is arguably the entire point of building a demonstrator.<\/p>\n The pitch for alpine solar is snow. Fresh snow throws a lot of light back upward, so a panel on a mountain gets worked from below as well as from above, and cold silicon converts more of what hits it than hot silicon does. Bifacial modules have cells on both faces, and they’re there to catch the bounce.<\/p>\n Romande Energie tested that pitch on shore first, with a small rig running from 2012. It measured almost 1,800 kilowatt-hours per installed kilowatt per year, about 50 percent better than the same hardware in the Swiss lowlands, and that’s the number that went into every article about the project. It’s the same argument behind the 4,872 panels bolted to the Muttsee dam<\/a> in eastern Switzerland.<\/p>\n So how much of that survived the move onto the water?<\/p>\n Less than you’d hope, and the engineers knew it before they built anything. Their own simulation for the floating array came out at 1,470.<\/p>\n Three things ate the gap, and none of them are exotic. The rows on the floating platform sit 6.4 feet apart instead of the 9.8 feet used on shore, to get more panels onto less structure. The array is about 1.2 miles further south, close enough to the mountains that they take roughly 40 minutes of sun off an average day. And the snow bounce mostly stops working while the thing floats, because what’s under it then is water.<\/p>\n That last one is my favorite, because the selling point cancels itself for part of the year. The mirror is there in February, when the array is sitting on a frozen gravel pad surrounded by snow. In July it’s a boat.<\/p>\n Run the measured yield against 448 kilowatts and you land at about 636 megawatt-hours a year, which is what Kaufmann published. The project’s public page<\/a> still advertises 800,000 kilowatt-hours. That was the pre-construction estimate and nobody has gone back to update it.<\/p>\n The modules sit at 37 degrees while the array floats, which leaves them at 32 degrees once it’s aground. They’re steep on purpose. The canton made winter output a permit condition, so at least 40 percent of the year’s electricity has to arrive in the winter half, and a steep panel catches the low sun and sheds snow on its own.<\/p>\n Shedding snow only works if the snow has somewhere to go. Wind-driven drifts pile up under the array faster than the clearance underneath can swallow them, and once that gap is full, whatever slides off the glass has nowhere to land. Modules stay buried for hours, sometimes days. Kaufmann puts it at 20 to 45 lost days per winter, mostly between mid-December and the end of February. The rig on shore loses 4 to 15.<\/p>\n Drifting has also snapped about ten modules outright, and a windbreak went in to cut it down.<\/p>\n Every three years the team runs flash tests on the modules, which is a controlled light pulse that tells you how much performance a panel has lost since it left the factory. We only have what Romande Energie has published, and so far the Toules modules are degrading more slowly than equivalent panels in the lowlands, with nothing showing up that looks specific to water or altitude. Heat is what kills solar panels, and there isn’t much of it at 5,940 feet. That tracks with a long-running Swiss study of panel aging by altitude<\/a>, where the high, cold sites held up best.<\/p>\n The lake seems fine with it too. Fraunhofer ISE and the University of Freiburg spent three years on three floating plants, Toules included, and found no clear effect on water quality<\/a>. The water under the panels ran cooler in summer and lost slightly less heat in winter.<\/p>\n All of the above was a demonstrator. The real plan is an extension covering about 49 acres (20 hectares), roughly a third of the lake: 13.6 megawatts and 21 gigawatt-hours a year, with 8.6 of those between October and March. The structure is designed for 200 feet of water level movement and winds near 112 mph, with a 60-year service life and a module replacement halfway through.<\/p>\n Bourg-Saint-Pierre voted for the extension unanimously in June 2023 and the cantonal offices were happy with it. Then it reached the Federal Office of Energy’s dam surveillance section, which couldn’t issue an opinion at all. A dam is a regulated safety structure, so anything sitting in the water behind it becomes the dam regulator’s problem, and Switzerland has no technical or legal basis for a floating structure inside a reservoir. The regulator had nothing to measure it against.<\/p>\nThat 50 percent never turned up<\/h2>\n
Snow gets in underneath<\/h2>\n
The panels are aging better than the ones in the lowlands<\/h2>\n
So where’s the big one?<\/h2>\n