Numerical Methods for Plasma Physics

My interest in scientific computing is in developing numerical approaches for physical problems, with much of my research devoted to understanding plasma behavior in fusion devices. At the Lawrence Livermore National Laboratory (LLNL) I worked on a multi-fluid model of counter-streaming plasmas, with the goal of improving upon the single-fluid approach and better simulate inertial confinement fusion plasmas. At the Princeton Plasma Physics Laboratory (PPPL) my work focused instead on particle-in-cell codes, in particular their scalabality and their application to modeling plasma propulsion systems.

Contents

Overview

Plasma physics is the science of charged particle behavior in the presence of electro-magnetic fields. It is a major field of study at many national labs in the United States as well as at international facilities such as ITER because understanding it is fundamental to most approaches to controlled nuclear fusion, which when done efficiently promises to be a virtually unlimited source of clean energy. Further applications of plasma physics include space propulsion and nanotechnology.

Plasma behavior is modeled by systems of partial differential equations, which in most non-trivial cases require numerical methods to solve. Thus much effort is devoted to mathematical modeling and developing scalable iterative solvers. As with other sciences, the numerics play a supporting role to experimental evidence; however, since experiments in areas such as nuclear fusion tend to be very expensive and can be damaging to the device, understanding when certain approximations can be made and predicting when instabilities may arise is of prime importance.

Multi-Fluid Modeling of Interpenetrating Hohlraum Plasmas

We develop a multi-fluid model for studying the interpenetration of counter-streaming hohlraum plasmas. This problem is crucial to understanding density build-ups and the scattering of laser light in the fusion device that can hamper experiments, but current single-fluid models are too simple while particle-in-cell simulations are costly. Our model is closer qualitatively to experimental results and can be extended to arbitrary ion species and various experimental conditions.

Poster

Joint with Dick Berger, Tom Chapman, and Jeff Hittinger.

Particle-in-Cell Simulations for Plasma Propulsion

We investigate the problem of plasma detachment in an expanding magnetic field using a massively-parallel particle-in-cell code. While individual charged particles travel parallel to magnetic field lines, high-energy dense plasmas are predicted to detach from them due to turbulence effects. Predicting when and how this occurs is important to understanding the thrust of some space-based propulsion systems, which eject charge particles from magnetic nozzles. Through three-dimensional simulations of the expansion region we show that this plasma detachment is likely to occur in experimental settings. 

Slides

Advised by Sam Cohen.