NERSC Initiative for Scientific Exploration (NISE) 2011 Awards

Breaking Magnetic Field Lines with Turbulence

James Drake, University of Maryland

Associated NERSC Project: Turbulence, Transport and Magnetic Reconnection in High Temperature Plasma (mp217)
Principal Investigator: William Dorland

Magnetic reconnection negatively impacts energy confinement in laboratory fusion experiments and explosively releases energy in space and astrophysical plasmas. Dissipation is required to break field lines and release magnetic energy. It has long been hypothesized that the intense current layers produced during reconnection could drive turbulence. We have identified and will explore a new class of turbulence that dominates the dissipation during reconnection and will produce a new understanding of how magnetic energy is released.

Magnetic reconnection is the dominant process by which magnetic energy in a plasma system is dissipated. In laboratory fusion experiments reconnection the onset of reconnection negatively impacts confinement and in extreme cases can lead to the termination of the plasma discharge. Solar flares on the sun and substorms in the magnetosphere lead to explosive releases of magnetic energy that can negatively impact power grids, satellites and communication. A key scientific issue of over 50 years has been to understand the mechanism that provides the dissipation that is necessary to break magnetic field lines and enable reconnection to proceed. The dissipation produced by classical collisions is too weak to explain the laboratory and space observations. A widely discussed mechanism for producing this dissipation is through the development of turbulence in the intense current layers that form during reconnection. It had been hypothesized but never demonstrated that this turbulence could scatter electrons and therefore produce an "anomalous resistivity" that facilitates reconnection. We have demonstrated in a series of simulations and in a paper that is being reviewed by Nature that "anomalous resistivity" is not the dominant mechanism. Instead the narrow current layers that are produced during reconnection breakup into a filamentary web as a result of an electron shear-flow instability. This behavior can be represented by a turbulence-induced viscosity rather than a resistivity. Thus, "anomalous viscosity" is the dominant mechanism facilitating reconnection in the high temperature regime of greatest interest in the laboratory and space.

Over the past year we have carried out a series of simulations of reconnection that first uncovered the filamentation process and then tested the range of parameters for which it was important. A surprise was that a system with a modestly strong guide field, which is the most relevant casefor fusion and space and astrophysics, exhibits a greater tendency for filamentation than reconnection with no guide field. Thus, with simulations and modeling to date we have established the robustness of the instability but because the development of turbulence requires the exploration of a 3-D system (the turbulence develops along the symmetry direction of a traditional 2-D reconnection model), the simulations carried out thus far are only marginally separating the characteristic scale lengths of the turbulence from that of the macroscale reconnecting system. Further the small size of the total system limits the time over which reconnection takes place before depleting the available free energy so that there is insufficient time to be sure that the turbulent-induced breakup of the current layer has come into balance with the tendency for reconnection to strengthen and narrow the current layer.

Thus, we can not be certain about the impact of turbulence on the rate of reconnection until the turbulence has come into balance with the reconnection drive. The tendency for turbulence to overshoot is a common characteristic of the onset of turbulence in plasma systems.

Thus, the important next step in our exploration of this exciting new area of research is to carry out simulations with a greater separation between the turbulence and system scales. Such simulations will enable us to pin down how turbulence influences the rate of reconnection by following the development of turbulence and the rate of reconnection until both have come to a balance in the full 3-D system and by comparing the results to those obtained in the computationally simpler 2-D system.

The specific simulations that will be carried out will have a computational domain of 16 x 8 x 8 in units of ion inertial lengths, and an ion-to-electron mass ratio of 100. The box would be 2048 x 1024 x1024$ grid points and contain roughly around 10^{11} particles. Our code has shown essentially perfect scaling up to 8192 processors. Extrapolating its performance to 32,768 processors we expect such runs to take around 20 wall-clock hours, for a total of 650,000 processor-hours. Runs with at two different values of the ambient guide field (with a strength that is equal two and twice the reconnecting magnetic) are presently planned.