PLAINSBORO, NJ -- Scientists at the U.S. Department of Energy’s(DOE) Princeton Plasma Physics Laboratory (PPPL) report a majoradvance in the computer modeling of fusion plasmas in theSeptember 18 edition of Science magazine. The new results wereobtained utilizing the Massively Parallel Processing (MPP)capabilities of the DOE’s National Energy Research ScientificComputing Center (NERSC) at the Lawrence Berkeley NationalLaboratory (LBNL) in Berkeley, California.
In general, the formation of stars results from the dynamicbehavior of hot ionized gases known as plasmas. Fusionresearchers use magnetic fields to confine such hot gaseous fuelat the temperature, density, and duration required for thecontrolled production of significant amounts of fusion energy.However, turbulence can spoil the efficiency of this approach bycausing accelerated loss of particles and energy from the plasma.As pointed out in a Perspectives article in the same issue ofScience, good news from toroidal confinement experimentsindicates that the suppression of turbulence can enabletransitions to newly discovered enhanced confinement regimes.
In the work reported in Science this week, PPPL scientists haveused the full power of the SGI/Cray T3E supercomputer at NERSC tocreate three-dimensional nonlinear particle simulations ofmicroturbulence in the plasma. The use of NERSC’s massivelyparallel processor (MPP) capabilities enabled scientists toperform calculations involving 400 million plasma particles(i.e., 100 million guiding centers) in 5,000 time-steps -- anachievement impossible without the use of powerful MPP computers.
“The information obtained from these advanced computersimulations is providing valuable new physics insights andcorrelates well with trends observed in experiments. This workbuilds on the excellent knowledge base developed internationallyand complements related research efforts at other nationallaboratories such as Lawrence Livermore National Laboratory andGeneral Atomics (both in California), as well as at universitiessuch as UCLA and the University of Colorado,” noted PPPL’s ChiefScientist, William M. Tang.
“The results in our Science article help support the increasinglyaccepted position that high performance scientific computing hasmatured to a level where it can be considered a new tool fordiscovery complementing traditional theory and experiment,” Tangsaid. “It is evident that plasma science, along with many otherfields, will benefit greatly from the development of the advancedcomputational capabilities envisaged and strongly encouraged atthe DOE/National Science Foundation Workshop on AdvancedScientific Computing held this July at the National Academy ofSciences.”
NERSC was originally established in 1974 to support fusion energyresearch programs sponsored by the DOE. The center, one of thenation’s most powerful unclassified supercomputing centers,provides both computational resources and technical support toresearchers at PPPL, as well as a wide range of research effortsin various scientific disciplines at other DOE sites.
“These results are exciting in that they again demonstrate thevalue of computational science as a complement to experimentalscience. It’s particularly exciting that this significant fusionresult was achieved using a highly parallel computing system,”said William Kramer, Deputy Director of NERSC. “We’re alsopleased to see that NERSC continues to play a critical role inhelping plasma physics scientists make new advances in the field.Collaborations such as this really are the future of large-scalescientific research.”
NERSC (http://www.nersc.gov) provides high performance computingservices to DOE’s Energy Research programs at nationallaboratories, universities, and industry. Berkeley Lab(http://www.lbl.gov) conducts unclassified research and ismanaged by the University of California.
PPPL (http://www.pppl.gov), which is funded by DOE and managed byPrinceton University, is a collaborative national center forscience and innovation leading to an attractive fusion energysource.
Editor’s Note: The citation for the PPPL Science article is:“Turbulent Transport Reduction by Zonal Flows: Massively ParallelSimulations,” Z. Lin, T.S. Hahm, W.W. Lee, W.M. Tang, and R.B.White, Science (281), 1835 (1998). There is also a Perspectivearticle by Keith Burrell of General Atomics Corporation and ahighlight of the article in the same issue of Science.
The above post is reprinted from materials provided by Lawrence Berkeley National Laboratory. Note: Content may be edited for style and length.
Cite This Page: