Commun. Comput. Phys.,
Plasma Edge Kinetic-MHD Modeling in Tokamaks Using Kepler Workflow for Code Coupling, Data Management and Visualization
J. Cummings 1*, A. Pankin 2, N. Podhorszki 3, G. Park 4, S. Ku 4, R. Barreto 5, S. Klasky 5, C. S. Chang 4, H. Strauss 4, L. Sugiyama 6, P. Snyder 7, D. Pearlstein 8, B. Ludascher 3, G. Bateman 2, A. Kritz 2, the CPES Team 91 California Institute of Technology, Pasadena, CA 91125.
2 Lehigh University, Bethlehem, PA 18015.
3 University of California at Davis, Davis, CA 95616.
4 Courant Institute of Mathematical Sciences, New York University, NY 10012.
5 Oak Ridge National Laboratory, Oak Ridge, TN 37830.
6 Massachusetts Institute of Technology, Cambridge, MA 02139.
7 General Atomics, San Diego, CA 92186.
8 Lawrence Livermore National Laboratory, Livermore, CA 94550.
9 SciDAC FSP Center for Plasma Edge Simulation.
Received 2 November 2007; Accepted (in revised version) 28 January 2008
Available online 21 April 2008
A new predictive computer simulation tool targeting the development of the H-mode pedestal at the plasma edge in tokamaks and the triggering and dynamics of edge localized modes (ELMs) is presented in this report. This tool brings together, in a coordinated and effective manner, several first-principles physics simulation codes, stability analysis packages, and data processing and visualization tools. A Kepler workflow is used in order to carry out an edge plasma simulation that loosely couples the kinetic code, XGC0, with an ideal MHD linear stability analysis code, ELITE, and an extended MHD initial value code such as M3D or NIMROD. XGC0 includes the neoclassical ion-electron-neutral dynamics needed to simulate pedestal growth near the separatrix. The Kepler workflow processes the XGC0 simulation results into simple images that can be selected and displayed via the Dashboard, a monitoring tool implemented in AJAX allowing the scientist to track computational resources, examine running and archived jobs, and view key physics data, all within a standard Web browser. The XGC0 simulation is monitored for the conditions needed to trigger an ELM crash by periodically assessing the edge plasma pressure and current density profiles using the ELITE code. If an ELM crash is triggered, the Kepler workflow launches the M3D code on a moderate-size Opteron cluster to simulate the nonlinear ELM crash and to compute the relaxation of plasma profiles after the crash. This process is monitored through periodic outputs of plasma fluid quantities that are automatically visualized with AVS/Express and may be displayed on the Dashboard. Finally, the Kepler workflow archives all data outputs and processed images using HPSS, as well as provenance information about the software and hardware used to create the simulation. The complete process of preparing, executing and monitoring a coupled-code simulation of the edge pressure pedestal buildup and the ELM cycle using the Kepler scientific workflow system is described in this paper.
Notice: Undefined variable: ams in /var/www/html/issue/abstract/readabs.php on line 163
PACS: 52.65.-y, 52.65.Kj, 52.65.Tt, 52.65.Ww
Key words: Plasma simulation, magnetohydrodynamic and fluid equation, gyrofluid and gyrokinetic simulations, hybrid methods.
Email: email@example.com (J. Cummings), firstname.lastname@example.org (A. Pankin), email@example.com (N. Podhorszki), firstname.lastname@example.org (G. Park), sku@cims. nyu.edu (S. Ku), email@example.com (R. Barreto), firstname.lastname@example.org (S. Klasky), email@example.com (C. S. Chang), firstname.lastname@example.org (H. Strauss), email@example.com (L. Sugiyama), snyder@ fusion.gat.com (P. Snyder), firstname.lastname@example.org (D. Pearlstein), email@example.com (B. Ludascher), firstname.lastname@example.org (G. Bateman), email@example.com (A. Kritz)