Title :
Particle heating in magnetized plasmas with interpolation-induced divergence
Author :
Aldan, M.P. ; Verboncoeur, J.P.
Author_Institution :
UC Berkeley, Berkeley, CA, USA
Abstract :
Summary form only given: Numerical effects in particle-in-cell (PIC) simulations due to the interpolation of discrete magnetic fields are examined. Heating, cooling, and/or non-physical diffusion across field lines is a concern for magnetically confinement plasmas in applications such as magnetic confinement fusion, sputtering magnetrons, and many other devices. A classical axisymmetric magnetic-mirror configuration is used to investigate the effects of magnetic-field discretization. A single particle is tracked for roughly two bounce periods. Sinusoidal fields are used to ensure truncation error exists to all orders in the numerical methods used. The magnetic field is discretized on a square grid, and linearly interpolated within the grid. The standard Runge-Kutta fifth-order accurate (RK45) solver with a continuum magnetic field is used as a control method. Electric fields are neglected, reducing the leap-frog scheme to a rotation in the standard way. Solutions from discrete-field and continuous-field distributions are compared with particular interest in energy conservation and particle-trajectory variations. A coarse grid of roughly 20x20 is found sufficient to produce the same energy-conservation results as the continuous-field solution. Time steps are found to be a more restrictive parameter than grid size in producing accurate results, with leap-frog requiring roughly ten times more computations that RK45. Due to the second-order accuracy of the leap-frog method, overall numerical energy loss (cooling) can occur with leapAcircnot frog when compared to RK45. Solutions with leap-frog are found to consistently produce an overall numerical cooling in the cases tested. Numerical heating due only to discretization of the magnetic fields is found to be negligible; however, a marked discrepancy in particle trajectories is observed for coarse meshes. This discrepancy could have implications for cross-field diffusion and particle loss, and quantification of this effec- is currently being investigated.
Keywords :
diffusion; plasma confinement; plasma simulation; Runge-Kutta fifth-order accurate solver; classical axisymmetric magnetic-mirror configuration; continuous-field distributions; continuum magnetic field; cooling; cross-field diffusion; discrete magnetic fields; energy conservation; heating; interpolation-induced divergence; leap-frog scheme; magnetic-field discretization; magnetically confinement plasmas; magnetized plasmas; nonphysical diffusion across field lines; particle heating; particle loss; particle-in-cell simulations; particle-trajectory variations; sputtering magnetrons; truncation error; Cooling; Heating; Interpolation; Magnetic confinement; Magnetic devices; Magnetic fields; Plasma applications; Plasma confinement; Plasma devices; Plasma simulation;
Conference_Titel :
Plasma Science - Abstracts, 2009. ICOPS 2009. IEEE International Conference on
Conference_Location :
San Diego, CA
Print_ISBN :
978-1-4244-2617-1
DOI :
10.1109/PLASMA.2009.5227358