学术成果更新于 2026-09-16
In this thesis, the role of the “marker weight” (or “particle weight”) used in gyrokinetic particle–in–cell (PIC) simulations is explored. Following a review of the foundations and major developments of gyrokinetic theory, key concepts of the Monte Carlo methods which form the basis for PIC simulations are set forth. Consistent with these methods, a Klimontovich representation for the set of simulation markers is developed in the extended phase space $\{ \mathbf { R } , v _ { \parallel } , v _ { \perp } , W , P \}$ (with the additional coordinates representing weight fields); clear distinctions are consequently established between the marker distribution function and various physical distribution functions (arising from diverse moments of the marker distribution). Equations describing transport in the simulation are shown to be easily derivable using the formalism. The necessity of a two–weight model for nonequilibrium simulations is demonstrated, and a simple method for calculating the second (background–related) weight is presented. Procedures for arbitrary marker loading schemes in gyrokinetic PIC simulations are outlined; various initialization methods for simulations are compared. Possible efects of inadequate velocity–space resolution in gyrokinetic continuum simulations are explored. The “partial–f” simulation method is developed and its limitations indicated. A quasilinear treatment of electrostatic drift waves is shown to correctly predict nonlinear saturation amplitudes, and the relevance of the gyrokinetic fluctuation–dissipation theorem in assessing the efects of discrete–marker–induced statistical noise on the resulting marginally stable states is demonstrated.
学术成果更新于 2026-09-16
DIVERTING ENERGY FROM ENERGETIC α-particles to waves (α-channelling) would be extremely beneficial for a magnetically confined deuterium tritium fusion reactor. If these waves were to damp on fuel ions, a hot ion mode would result, doubling the fusion power of the reactor at the same confined pressure. Alternatively, if these waves damp preferentially on electrons traveling in one direction, current would be driven. In both cases, the pressure profile could be modified and ash could be removed to advantage.
These potentially significant benefits motivate a detailed study of the implementation of α-channelling. This thesis identifies and explores issues in realizing α-channelling, making the following advances:
1. Calculating, through 0-dimensional reactor simulations, the substantial benefit in α-channelling which motivates the subsequent work.
2. Framing the α-channelling problem as a diffusion problem, with absorbing and reflecting boundaries. To develop insight, α-channelling is considered in a 2-dimensional phase space associated with a simple slab geometry. To solve realistic problems we pose α-channelling in the 3-dimensional constants of the motion space associated with particle orbits in tokamaks.
3. Developing a rapid Monte Carlo simulation in constants of motion space to keep
track of wave-induced and collisional effects on the energetic particle distribution.
This approach is equivalent to the full energetic particle dynamics in the limit of
small changes during a single bounce time and diffusive wave-particle interactions.
4. Identifying the wave characteristics necessary to produce the channelling of
fect. which we discover are available in a combination of the mode converted ion-
Bernstein wave - MCIBW and the toroidal Alfron eigenmode - TAE:
5. Demonstrating how two waves can be combined in a reverse shear tokamak reactor
to absorb 2 of the energy from the $93^{\circ}$ of the a-particles ejected!…