• Conventional photovoltaic solar panels are flat, and can be installed horizontally, angled to face the average height of the sun, or mounted on a tracking mechanism.

    Full Screen
  • MIT professor Jeffrey Grossman used a computerized system to let possible shapes for solar panels evolve over time, starting with simple basic shapes. This is an example of a shape that was found to be quite effective because it could catch the sun lower in the sky, and also some surfaces could reflect sunlight onto others.

    Image courtesy of Jeffrey Grossman, Bryan Myers and Marco Bernardi

    Full Screen
  • Shapes that were less efficient were culled from the mix, and the best shapes were combined to produce new hybrid forms.

    Image courtesy of Jeffrey Grossman, Bryan Myers and Marco Bernardi

    Full Screen
  • Some of the shapes produced by the computer program were quite complex.

    Image courtesy of Jeffrey Grossman, Bryan Myers and Marco Bernardi

    Full Screen
  • The program explored shapes that included multiple curves and angles.

    Image courtesy of Jeffrey Grossman, Bryan Myers and Marco Bernardi

    Full Screen
  • The shape on left was one of the most efficient ones generated by the program, but also one of the most complex. That complexity means it would be impractical to make, so the team produced a simplified version (right) that performed almost as well. Such forms could be designed to be shipped flat, then unfolded at their installation site to their full 3-D shape.

    Image courtesy of Jeffrey Grossman, Bryan Myers and Marco Bernardi

    Full Screen
  • Simulations show that the higher the 3-D panels extend up from the horizontal, the greater their power output, and that they produce a relatively stable output over the course of a day (blue lines) as compared to flat, horizontal panels (red).

    Image courtesy of Jeffrey Grossman, Bryan Myers and Marco Bernardi

    Full Screen
  • These images illustrate how the genetic algorithm used by the researchers -- based on the principles of evolution -- can start from complete randomness and lead to complex, highly efficient shapes.

    Image courtesy of Jeffrey Grossman, Bryan Myers and Marco Bernardi

    Full Screen

Slideshow: Solar power, shaped up

3-D shapes covered in solar cells could produce more power than flat panels, MIT researchers find.


Flat solar photovoltaic panels are becoming more widespread, but the power they produce varies over the course of the day as the sun’s position changes — unless the panels are mounted on tracking systems to keep them pointed sunward, which adds complexity and expense.

Jeffrey Grossman, the Carl Richard Soderberg Associate Professor of Power Engineering at MIT’s Department of Materials Science and Engineering (DMSE), was inspired by the way trees spread their leaves to capture sunlight and wondered how efficient a three-dimensional shape covered in solar cells could be, and what its optimal shape would look like. He worked with a second-year DMSE graduate student, Marco Bernardi, to create a computer program that mimics biological evolution, starting with basic shapes and letting them evolve, changing slightly each time and selecting those that perform best to start the next generation. He found that such systems could produce relatively constant power throughout the day without the need for tracking, and produce significantly more power overall for a given area — for example two and a half times as much as a flat array when the height equals the length and width. He is continuing to work on finding the best shapes and teaming up with Professors Vladimir Bulović and David Perreault (EECS) to build a prototype system. The team believes that solar panels based on this concept could be shipped flat and then unfolded at the site to their complex shapes.

These images show some of the varied shapes with improved efficiency that emerged from the evolving simulation.


Topics: Energy, Materials science

Comments

Why sell those novel solar cells in flat shape and then fold them in 3d shape? A factory can produce a bunch of it in small sizes. You say as small as an apple. And sell all of them with a tree like base. Customer can connect all solar cells to arms of this tree and it can hold all of them in suitable height to produce maximum power. This way we can build them just like in trees.
The idea to use a genetic algorithm for finding optimized strcutures is clear and has been done by others before. An interesting talk by Bill Gross on the topic can be found here: http://www.ted.com/talks/lang/eng/bill_gross_on_new_energy.html
I disagree that the work of Bill Gross has anything to do with this. Certainly many things can be designed using genetic algorithms, and energy generation is no exception. The idea from Prof. Grossman deals with *photovoltaic* technology, while the video from Bill Gross discusses a solar concentrator (i.e. pieces of metal, not semiconductors) arranged to reflect and focus light and activate an engine. Here the 3D structure is the actual solar cell, and can be made of any material used in photovoltaics. The structures are more compact and more similar to a black-body in this work, rather than a wide flower (as in Bill Gross' work) opened at the top and behaving like a focusing parabola. From a technical and technological point of view the two works are completely unrelated. Good point for the similarity on the use of genetic algorithms though, in my opinion it just means it is becoming more and more relevant in engineering design!
photosynthetic cells 'if nature can use other then flat surfaces, so can we' -ka
Back to the top