 |
Center for Nonlinear Studies |
Kinetic (non-equilibrium) effects are important in both low-temperature plasmas, like those in Hall thrusters, and high-temperature plasmas, like those in ICF implosions. Fluid models, while being computationally inexpensive, of- ten fail to accurately capture plasma instabilities and transport processes. In Hall thruster plasmas, kinetic instabilities are driven by the relative E × B drift of electrons with respect to the ions, resulting in anomalous electron transport across magnetic field lines. Understanding anomalous electron transport is still one of the unanswered questions within the low-temperature plasma community. In this talk, I will present linear kinetic theory that I derived to model these ki- netic instabilities, including the effects of 3D wave vectors, doubly-charged ions, and finite ion temperatures. A generic method of solving implicit dispersion re- lations is discussed and applied to the kinetic instabilities within Hall thrusters. I will also present results from a 3D PIC model that I developed to compare the behavior of the plasma waves in a cross-field system with the kinetic theory, demonstrating excellent agreement in the instability growth rates during the linear growth stage.
On the other hand, in the high-temperature plasmas within ICF implosion events, radiation-hydrodynamic codes struggle to capture kinetic effects, leading to the over-prediction of the fusion yield and fuel temperature during implosion events. Therefore, high-fidelity kinetic simulations of ICF events are needed. One such model is iFP, developed at T-5. Within iFP, the fusion reaction sink term is ignored. A self-consistent collisional source term is too computationally expensive to employ, as it requires the evaluation of a 5-dimensional integral in velocity space that does not parallelize well. I will present work in which a tensor-train decomposition of the reactivity integral is employed to accelerate the calculation of the collisional source term. A method was also created to interpolate tensor-train objects between different grids, which is necessary to handle the adaptive velocity grid of iFP. Fortran calculations of the reactivity integral in tensor-train format achieved a 23-times speed-up, while introducing a relative error less than 0.3% across all temperatures of interest.
Bio: Andrew Denig is a Ph.D. candidate at Stanford University in the Plasma Dynamics Modeling Lab, where he works on the development of kinetic theory and implementation of particle-in-cell models for various low-temperature plasma applications. He also worked with T-5 to develop a tensor-train method to accelerate high-dimensional reactivity integrals, with applications to self- consistent Vlasov-Fokker-Planck models. He received his MS degree in aero- nautics and astronautics from Stanford University and his BS degree in nuclear and radiological engineering from Georgia Tech.
Teams: Join the meeting now
Meeting ID: 256 247 418 232 0
Passcode: 7Ye9Qd2b
Host: Will Taitano (T-5)