Ever wanted to see a nuclear reactor core in action? A new computeralgorithm developed by researchers at the U.S. Department of Energy's(DOE) Argonne National Laboratory allows scientists to view nuclearfission in much finer detail than ever before.
An elevation plot of the highest energy neutron flux distributionsfrom an axial slice of the reactor is shown superimposed over the sameslice of the underlying geometry. This figure shows the rapid spatialvariation in the high energy neutron distribution between within eachplate along with the more slowly varying, global distribution. Thefigure is significant since UNIC allows researchers to capture both ofthese effects simultaneously.
A team of nuclear engineers and computer scientists at ArgonneNational Laboratory are developing the neutron transport code UNIC,which enables researchers for the first time to obtain a highlydetailed description of a nuclear reactor core. The code could prove crucial in the development of nuclear reactorsthat are safe, affordable and environmentally friendly. To model thecomplex geometry of a reactor core requires billions of spatialelements, hundreds of angles and thousands of energy groups -- all ofwhich lead to problem sizes with quadrillions of possible solutions. Such calculations exhaust computer memory of the largest machines,and therefore reactor modeling codes typically rely on variousapproximations. But approximations limit the predictive capability ofcomputer simulations and leave considerable uncertainty in crucialreactor design and operational parameters. "The UNIC code is intended to reduce the uncertainties and biases inreactor design calculations by progressively replacing existingmultilevel averaging techniques with more direct solution methods basedon explicit reactor geometries," said Andrew Siegel, a computationalscientist at Argonne and leader of Argonne's reactor simulation group. UNIC has run successfully at DOE leadership computing facilities,home to some of the world's fastest supercomputers, including theenergy-efficient IBM Blue Gene/P at Argonne and the Cray XT5 at OakRidge National Laboratory. Although still under development, the codehas already produced new scientific results.
|