Rune has tipped the scale between accuracy and computing power

Saturday 28 Jan 17
|

Error Mitigation in Computational Design of Sustainable Energy Materials

Main supervisor: Tejs Vegge

Co-supervisor: Heine A. Hansen

Rune Christensen has with his Ph.D. thesis "Error Mitigation in Computational Design of Sustainable Energy Materials" ensured that better battery types have come closer.

Materials research has traditionally consisted of a process of assumptions, experiments, and analysis of data, followed by new assumptions and new experiments until there was a scientific breakthrough. Now, using supercomputers it has become possible to gain new insights on the atomic level by independent computer calculations.

Much of what is currently possible was not possible at all just a decade ago. Not only are computers significantly better, but the analytical methods are now so good that you have much more data available than ever before", says 28-year-old Rune Christensen from DTU Energy, who on 9 January defended his Ph.D. thesis on ”Error Mitigation in Computational Design of Sustainable Energy Materials” at DTU Energy.

Rune Christensen and his colleagues do research on batteries, and his thesis rethinks and improves innovative methods developed very recently by Stanford University and DTU Physics. These methods are used to make a single calculation in 2000 different ways with different assumptions, in the same time it previously took to make the same calculation in only one way. This allows the uncertainties of calculations to be determined.

A single calculation done in 2000 different ways demands large amounts of computer power, but with the help of supercomputers this is no longer a problem for the researchers. They can now obtain hitherto undreamt of amounts of information on the atomic level. This means for example that 1000 potential combinations of materials can be cut down to the 5-6 combinations which in theory have the best properties. Then only these materials need to be synthesized and characterized.

This saves huge amounts of time and resources in materials research and shortens the experimental phase significantly. However, this is only true if one has taken into account potential sources of error, since many data have built-in systematic errors. For years, scientists just accepted the general uncertainty caused by systematic errors in the calculations, but Rune Christensen discovered a method to identify and eliminate some of these systematic errors and described how to avoid them in his thesis.

“It was not a conscious choice, but we kept encountering frustrating errors, and then I got curious and started to identify them. Because errors are acceptable, if you know where they come from so you can navigate around them. So my goal was to identify them, not eliminate them, as there are already methods that remove errors, but they require much more computing power", Rune says.

“We are now able to make calculations which take into account the specific errors, enabling us to get more information out of the same amount of data and computing power. We now use the method for research into materials and catalysis, but the method can for certain be used in other research areas using DFT.”

Density Functional Theory (DFT) is a quantum-mechanical method of calculation the energy and a number of other properties derived from the electron density of an atomic system. It is necessary to balance the desired accuracy in the calculations of the required computing power and the size of the atomic system. With Rune Christensen's method, the balance has shifted so DTU researchers can now make more accurate, faster or more calculations than previously.

"When the errors are identified and excluded, you can look at larger atomic constellations and thus get far better results and a better understanding of the reactions in the system without having to invest in more computers. Of course it is still a balance between how large systems, you want to look at, and how accurate you want the data, but I have shifted the axis and enables us to get more accurate data from the same type of calculations”, says Rune Christensen.

After his defense he is now a research assistant at DTU Energy while he awaits his diploma. And although he is not quite sure about the future, he is confident that he will end up working in a place where his research is very close to applications in the real world, just like he has witnessed the development of new battery types at DTU Energy.

Read more about battery research at DTU Energy

"I want to know that my research is used for something and not just end up as dry theory somewhere on a bookshelf", he says.

News and filters

Get updated on news that match your filter.