Particle-in-Cell Simulations of Electron Transport from Plasma Turbulence

Zhihong Lin, Gregory Rewoldt
University of California, Irvine

Development of a reliable predictive capability for plasma turbulence is crucial for achieving ignition conditions in the next step fusion experiments such as the International Thermonuclear Experimental Reactor (ITER). Massively parallel particle-in-cell (PIC) simulation has proven to be a powerful tool for studying turbulent transport in magnetized plasmas, which typically involves nonlinear kinetic effects, widely disparate spatial-temporal scales, and complex geometry. Excellent progress in understanding the nature of ion transport driven by electrostatic turbulence has been enabled by previous SciDAC-funded research together with coordinated efforts in simulation, theory, and experiment. Since a similar degree of understanding of electron transport remains an outstanding question, the priority in current SciDAC-funded fusion research activity has now shifted to this area with focus on driving mechanisms associated with smaller scale, electromagnetic turbulence. This is a much more formidable simulation task due to the large ion-to-electron mass ratio. Recently, important progress has been made on studying two forms of turbulence potentially responsible for the electron transport in fusion plasmas. First, full device simulations of small scale turbulence driven by the electron temperature gradient (ETG) mode has been enabled by efficient numerical algorithms and effective utilization of powerful massively parallel computers. These large scale simulations together with the associated development of analytic nonlinear theory have led to key physics insights on nonlocal spectral transfer and wave-particle interactions, which underlie the mechanisms of mode saturation and electron transport. The new global simulation results challenge the prevailing picture based on earlier simulations of ETG instabilities carried out for local geometry. Secondly, global simulations of electromagnetic turbulence have recently been enabled by the development of a finite element elliptic solver for complex toroidal geometry. In order to invert an extremely large sparse matrix, this solver utilizes the state-of-the-art parallel library PETSc which has been accelerated by an algebraic multigrid solver hypre as a pre-conditioner. This new capability also enables analysis of interactions between small-scale kinetic modes and large scale magnetohydrodynamic (MHD) modes, which are believed to be important in burning plasma experiments in ITER. Timely progress on these important investigations has been enabled in large measure by productive interdisciplinary collaborations with applied math and computational scientists within the framework of the fusion component of the SciDAC Program.