Title :
MHD Turbulence Studies Using Lattice Boltzmann Algorithms-Physical Simulations Using 9,000 Cores on the Air Force Research Laboratory HAWK Supercomputer
Author :
Vahala, G. ; Yepez, Jeffrey ; Soe, Min ; Vahala, Linda ; Carter, Jonathan ; Ziegler, Sibylle
Author_Institution :
Dept. of Phys., Coll. of William & Mary, Williamsburg, VA
Abstract :
Most non-spectral computational fluid dynamics (CFD) algorithms do not scale to many thousands of cores. Here, we examine a particular mesoscopic representation of free decaying magnetohydrodynamics (MHD) turbulence that is amenable to massive parallelization to the full 9,000 cores available on Air Force Research Laboratory (AFRL) SGI Altix 4700 (Hawk). Moreover, our mesoscopic lattice Boltzmann (LB) algorithm will automatically enforce the important nabla* B = 0 constraint without any need for magnetic field divergence cleaning. Isosurfaces of vorticity and current for LB simulations on a spatial grid of 18003 show the long time existence of large scale magnetic and velocity structures. This is in stark contrast to the long time behavior of the velocity isosurfaces which all disintegrate into small scale structures.
Keywords :
computational fluid dynamics; lattice Boltzmann methods; magnetohydrodynamics; parallel machines; turbulence; Air Force Research Laboratory HAWK supercomputer; MHD turbulence; computational fluid dynamics algorithms; free decaying magnetohydrodynamics; lattice Boltzmann algorithms; massive parallelization; vorticity; Computational fluid dynamics; Computational modeling; Heuristic algorithms; Isosurfaces; Laboratories; Lattice Boltzmann methods; Magnetic cores; Magnetic fields; Magnetohydrodynamics; Supercomputers;
Conference_Titel :
DoD HPCMP Users Group Conference, 2008. DOD HPCMP UGC
Conference_Location :
Seattle, WA
Print_ISBN :
978-1-4244-3323-0
DOI :
10.1109/DoD.HPCMP.UGC.2008.32