Solar ‘towers’ beat panels by up to 20x

Intensive research around the world has focused on improving the performance of solar photovoltaic cells and bringing down their cost. But very little attention has been paid to the best ways of arranging those cells, which are typically placed flat on a rooftop or other surface, or sometimes attached to motorized structures that keep the cells pointed toward the sun as it crosses the sky.

Now, a team of MIT researchers has come up with a very different approach: building cubes or towers that extend the solar cells upward in three-dimensional configurations. Amazingly, the results from the structures they’ve tested show power output ranging from double to more than 20 times that of fixed flat panels with the same base area.

The biggest boosts in power were seen in the situations where improvements are most needed: in locations far from the equator, in winter months and on cloudier days. The new findings, based on both computer modeling and outdoor testing of real modules, have been published in the journal Energy and Environmental Science.

“I think this concept could become an important part of the future of photovoltaics,” says the paper’s senior author, Jeffrey Grossman, the Carl Richard Soderberg Career Development Associate Professor of Power Engineering at MIT.

The MIT team initially used a computer algorithm to explore an enormous variety of possible configurations, and developed analytic software that can test any given configuration under a whole range of latitudes, seasons and weather. Then, to confirm their model’s predictions, they built and tested three different arrangements of solar cells on the roof of an MIT laboratory building for several weeks.

While the cost of a given amount of energy generated by such 3-D modules exceeds that of ordinary flat panels, the expense is partially balanced by a much higher energy output for a given footprint, as well as much more uniform power output over the course of a day, over the seasons of the year, and in the face of blockage from clouds or shadows. These improvements make power output more predictable and uniform, which could make integration with the power grid easier than with conventional systems, the authors say.

The basic physical reason for the improvement in power output — and for the more uniform output over time — is that the 3-D structures’ vertical surfaces can collect much more sunlight during mornings, evenings and winters, when the sun is closer to the horizon, says co-author Marco Bernardi, a graduate student in MIT’s Department of Materials Science and Engineering (DMSE).

The time is ripe for such an innovation, Grossman adds, because solar cells have become less expensive than accompanying support structures, wiring and installation. As the cost of the cells themselves continues to decline more quickly than these other costs, they say, the advantages of 3-D systems will grow accordingly.

“Even 10 years ago, this idea wouldn’t have been economically justified because the modules cost so much,” Grossman says. But now, he adds, “the cost for silicon cells is a fraction of the total cost, a trend that will continue downward in the near future.” Currently, up to 65 percent of the cost of photovoltaic (PV) energy is associated with installation, permission for use of land and other components besides the cells themselves.

Although computer modeling by Grossman and his colleagues showed that the biggest advantage would come from complex shapes — such as a cube where each face is dimpled inward — these would be difficult to manufacture, says co-author Nicola Ferralis, a research scientist in DMSE. The algorithms can also be used to optimize and simplify shapes with little loss of energy. It turns out the difference in power output between such optimized shapes and a simpler cube is only about 10 to 15 percent — a difference that is dwarfed by the greatly improved performance of 3-D shapes in general, he says. The team analyzed both simpler cubic and more complex accordion-like shapes in their rooftop experimental tests.

At first, the researchers were distressed when almost two weeks went by without a clear, sunny day for their tests. But then, looking at the data, they realized they had learned important lessons from the cloudy days, which showed a huge improvement in power output over conventional flat panels.

For an accordion-like tower — the tallest structure the team tested — the idea was to simulate a tower that “you could ship flat, and then could unfold at the site,” Grossman says. Such a tower could be installed in a parking lot to provide a charging station for electric vehicles, he says.

So far, the team has modeled individual 3-D modules. A next step is to study a collection of such towers, accounting for the shadows that one tower would cast on others at different times of day. In general, 3-D shapes could have a big advantage in any location where space is limited, such as flat-rooftop installations or in urban environments, they say. Such shapes could also be used in larger-scale applications, such as solar farms, once shading effects between towers are carefully minimized.

A few other efforts — including even a middle-school science-fair project last year — have attempted 3-D arrangements of solar cells. But, Grossman says, “our study is different in nature, since it is the first to approach the problem with a systematic and predictive analysis.”



  1. glad to see attention focused on redesign and increased efficiency. Now, if they could only improve wind turbines in the same way, the industry would have a lot more friends.

  2. Nature is fundamentally built on topology yet these numbskulls in solar power never thought the arrangement of cells would be the key to harvesting photons???

  3. A friend is fond of recalling an article he saw (upwards of a decade ago) about a photovoltaic collector built in the shape of a sphere. Purportedly it was extraordinarily productive too.

  4. Excellent article. Like many inventions, this discovery must’ve been preceded by a facepalm when the researchers arrived at the conclusion. Though I don’t think I or anyone I know will appreciate a tower or cube sticking out of their roof, I do see a lot of potential in using the tower design in a farm setup or complex 3d shapes adorning a commercial building.

  5. The base area should include the shadow if we are to compare to a flat panel. Power output per available area will be comparable, unless the neighboring area is considered. If the neighbor doesnt care, then that last row .casting its shadow would give a small advantage to towers.

  6. No, they are not making the base solar cells more efficient per volume of silicon. Since so much of the silicon deployed is at suboptimal angles, they’re almost certainly getting lower efficiency per silicon area.They are generating more power per square footage of base area. This is significant “because solar cells have become less expensive than accompanying support structures, wiring and installation.” It’s about getting more power out of the area of rooftop (or wherever) you can devote to solar panels.

    I don’t think it’s that hard to follow. If you guys who claim to not understand the article are just trolling, hats off, I guess. I honestly can’t tell.

  7. A flat panel tracking the sun seems to be the ultimate in many peoples’ minds.

    I think of it as more of a 3D problem. Light comes from different directions (there are clouds), panels reflect some of the light. It seems a volume that accepts light from different directions, and doesn’t reflect any back out, would beat a flat panel no matter how well it tracks. That is why I wonder about copying the fractal design of green plants.

  8. Sorry, almost everybody posting above is wrong. Yes, they really are comparing footprint, so their tower has lots more cells than than a flat panel, albeit most facing away from the sun at any moment. Yes, it’s impressive, because (did you get this?) most of the cells are facing away from the sun at any moment. As actually was pointed out in the article, cells are getting cheaper. Not everybody has a huge, south-facing roof. If you can arrange lots of cells to get enough power anyway, you can use solar where you thought you couldn’t.

    It was far from obvious that a pile of cells facing every-which-way could be made to do much better than a flat panel under common conditions, just by clever arrangement. It was far from obvious what such an arrangement would be, although comparing the shapes of trees at different latitudes offers a hint.

  9. What a ridiculous and misleading article!!

    More solar panels = more output. Wow. Whowoodathunkit?

    In tomorrows news – “solar panels revealed to be more efficient facing the sun, rather than left facedown on the ground.”

    PLEASE tell me I am missing something……

    • Interpret the experiment as it was likely conceived, not as it is described in the article. See my response below.


  10. Uh guys, when the article states –

    “Amazingly, the results from the structures they’ve tested show power output ranging from double to more than 20 times that of fixed flat panels with the same BASE AREA.”

    You do realize the author means the same overall area of active solar cells was used for both configurations (flat and cube), not an entire cube of solar cells vs. a flat panel only as big as the “base” structure the cube was mounted on…..

    • Rob,
      I think the latter is correct. By base area they mean the foot print. The towers have more cells than the conventional flat plate. The tower cells are stacked at the unusual angles. Otherwise it would be very odd indeed.

      • Hello Jiro, thank you for your response! I both completely agree and disagree with you at the same time! :) I agree with you that they meant the “footprint” of the conventional flat solar panel, I just disagree with what how that footprint is defined. Haha

        The “footprint” by definition has to be the total square footage of the solar cells used by the flat panels as equal to the total square footage used by the the cube configuration; not just the cube’s structural mounting “footprint” vs. all of the panels they could stack on the cube.

        Measuring equal square footage vs. equal square footage of active solar panels between the two deployment strategies is the only way to control this experiment. Otherwise, the results would be entirely arbitrary and meaningless in percentages of increased efficiency. As the others below stated, there would be nothing remarkable at all about mounting 10x or more solar panels above what would just be one panel and getting increased power production.

        It’s easy to see really, and this experiment represents one of those times where we all say, “why didn’t I think of that?”. Solar panel efficiency is highly directional, meaning for every degree of deflection between direct sunlight and a fixed flat solar panel you get a measurable decrease in light absorption and thus, power production. The cube/accordion structure does a remarkable job of mitigating that flat/fixed panel constraint, more so than the engineers imagined. Lovely!


          • > But you now have to put these things 2 to 20 times further apart, so
            > that they’re not casting shadows on each other. :

            Or, let them cast shadows. The whole idea is based on the price per area of solar cells falling. Soak up more total sun light, even if watts per silicon is lower.

    • No – it is not clear. If EXACTLY the same surface area of PV panels is inferred – it is not made clear. In fact, “BASE AREA” infers the 2D footprint of the 3D model only.

      If it said “The team discovered that making a 3d “accordian” shape – out of exactly the same surface area of PC panels – increased efficiency by up to 20X.”

      Not difficult really…..and I STILL don’t know what this article means..!

      • Stunned – I agree, it is not made clear via the article at all, because the author employs the term “base” which is nebulously descriptive at best and totally incorrect at worst. Thus, we must use our brains a little to conceive how the experiment was configured and what it sought to prove….vs. taking the article quoted word for word.

        I believe the author used the words “base area” when he really meant “surface area”.

        • Cheers for the response. However I tend to disagree – since the claim of 2-20X efficiency seems off the scale. It just doesnt make sense – concertina the panel and you get THAT much of a jump in efficiency? Maybe if the entire surface was covered in thousands of tiny pyramid/prisms….

      • They say the cost is higher, so I assume they mean that it’s not more efficiency per semiconductor area. But seeing this article, I now think all or many flat panels need to be mounted in an elevated state with 2 axes of rotation so they can track the sun better.

  11. If the structure rotates, it can efficiently track the sun in the horizontal axis. It the tower accordians vertically, the angle of the solar panels changes, which can be used to track the sun in the vertical axis. The main benefit would be a reduced structural cost compared to other types of panel mounting and tracking.

  12. Geez…why is this “groundbreaking”….you mean someone was actually surprised stacking 20 solar panels on a tower makes 20X the energy as one panel laying on the ground? What next…you tell me I can store more files in a filing cabinet than putting one layer of files on the floor in that same space?!?

    The higher efficiency of the cubes aren’t terribly surprising either, considering even solar cells have quite a bit of reflectance from the surface. Obviously if you can arrange the cells to catch some of that reflected light, all the better.

    You might even beat a goofy cube geometry by using flat solar panels, but reflecting the light to a solar-thermal tower…sort of like the “Solar One” and “Solar Two” thermal projects, but instead of mirrors, use solar cells then reflect and concentrate the unwanted energy to the tower.

  13. I don’t really see this as that big of an advancement since they’re comparing their 3D structure to just a single panel with the same footprint. No doubt they are harvesting more energy, but their total cell square footage is much higher, and I would bet that their yield per sq inch of PV is lower than a motorized tracking panel.

    So this really only matters if the PV cells themselves become so cheap that we can pack a lot of relatively inefficient ones into a relatively small volume.

    I personally want to see further advancements into semi-transparent photovoltaics for use as films over windows.

  14. Sounds promising!
    Now if they could only improve on the industrial size wind turbine technology to make it 1) more efficeint 2) less obtrusive 3) more safe and friendly to avian and other wildlife species.
    Wish they’d stop giving these companies incentives and put the money into developing better systems and designs.

    • “Boy genius” has been debunked:

      [UPDATE Aug 22 2011. All may not be as it seems. According to Gadget Lab reader and grown-up Patrick Theiner, Dwyer made several schoolboy errors when making his experiments. An article debunking the experiment and results appears on the UVdiv blog. Apparently Dwyer was measuring the open voltage on the circuit, which “is practically independent of power output,” and stays all but constant regardless of light falling on the cells.

      This post also says that the theory is flawed, and that pointing the panels in different directions, most of which aren’t at the optimal angle to the incoming light, will yield less power than a flat panel. You can read the full math here. (Oddly, the post has itself disappeared, but you can read Google’s cache.]

  15. I wonder if these are dynamically rotating cubes, which rotates in a way such that one flat side of the cube always faces the sun to maximize the conversion rate.


Please enter your comment!
Please enter your name here