|
||||||
Home | Mission | about SciDAC | Contact Us | ||||||
Alumni ProjectThe Plasma Microturbulence ProjectW.M. Nevins, B.I. Cohen, A.M. Dimits
– LLNL; J. Candy, R. Waltz – General Atomics SummaryPlasma microturbulence is a critical issue to the magnetic fusion
program because it controls the energy confinement and thereby determines
the performance of a burning plasma experiment. Our goal is to study
plasma microturbulence through direct numerical simulation, which is
a powerful tool to study plasma microturbulence. It allows far better
diagnostics than those available on experiments, while existing supercomputers
achieve excellent resolution of the microturbulent eddies. Our effort
includes: code development (to enhance the fidelity of our numerical
models, increase their efficiency, and couple them to a common system
for data analysis and visualization) code validation (against each other,
against experimental measurements, and against theory) and expanding
our user community through web-based applications and collaborations
with the MFE theory and experimental communities. The development of more powerful computers and of efficient simulation algorithms provides us with a new means of studying plasma microturbulence – direct numerical simulation. Direct numerical simulation complements analytic theory by extending its reach beyond simplified limits. Simulation complements experiment because non-perturbative diagnostics measuring quantities of immediate theoretical interest are easily implemented in simulations, while similar measurements in the laboratory are difficult or prohibitively expensive. The Plasma Microturbulence Project (PMP) is addressing this opportunity through a program of code development, code validation, and by expanding our user community. Code Development. We support a 2x2 matrix of plasma microturbulence codes which use either particle-in-cell (PIC) or continuum methods, and model the plasma in either flux-tube or global geometry. We have completed (or are in the process of) implementing multi-species kinetic plasma models, including both the turbulent electric and magnetic fields. We have done this in a realistic equilibrium geometry (see Fig. 1). Our global codes are able to simulate plasmas as large as those in present experiments, and the even larger plasmas foreseen in burning plasma experiments. Code efficiency is a key issue. The PMP is the largest user of computer
cycles within OFES, and we are working with the Performance Enhancement
Research Center to further improve the performance of our codes. Our
algorithms scale nearly linearly with the number of processors, and
our production codes have achieved overall efficiencies of 10-20%. Figure 1. The (turbulent) electrostatic potential from a GYRO simulation of plasma microturbulence in the DIII-D tokamak.
Code Validation is a critical step in our program to understand plasma micro-turbulence. The PMP code validation effort includes comparisons both between simulation codes (see Fig. 2) and between these codes and experiments. Developing a wider User Community. A large user community is necessary if the magnetic fusion community is to benefit from the PMP codes. This user community has been engaged through our efforts to validate the PMP codes against experiments, and has been trained in workshops held in Dec. 2002 (for the GYRO code) and January 2003 (for the GS2 code and the GKV post-processor). The GS2 code is currently available as a web-based application. Key Accomplishments and Plans. The principal accomplishments of the project so far are (1) the successful implementation of electromagnetic drift-wave turbulence algorithms in both continuum and PIC codes; (2) the use of these code to simulate experimentally relevant discharges, and to perform parameter studies; (3) successful validation between PMP codes; (4) the propagation of PMP codes (particularly GS2) to a wider community of fusion researchers for routine use in modeling experiments; and (5) detailed investigations of the physics of velocity-shear stabilization and its relation to the internal transport barriers seen in experiments. These accomplishments resulted from computing resources and research support by both SciDAC and OFES. In the next phase of the project, we plan to continue to expand our user community, and our database of simulation results, with particular attention to simulations of electron-scale drift-wave turbulence. Such studies justify a doubling of both computing and staff resources. For further information on this subject contact:
|
Home | ASCR | Contact Us | DOE disclaimer |
|