Bulletin of the American Physical Society
75th Annual Gaseous Electronics Conference
Volume 67, Number 9
Monday–Friday, October 3–7, 2022;
Sendai International Center, Sendai, Japan
The session times in this program are intended for Japan Standard Time zone in Tokyo, Japan (GMT+9)
Session GR4: Modeling - New Algorithms and Machine Learning |
Hide Abstracts |
Chair: Satoshi Hamaguchi, Osaka University; Jan Trieschmann, Kiel University Room: Sendai International Center Shirakashi 2 |
Thursday, October 6, 2022 1:30PM - 1:45PM |
GR4.00001: High-order moment closure for partially-ionized plasmas Alejandro Alvarez Laguna, Kentaro Hara Fluid models for the charged species in gas discharges are often based on the drift-diffusion approximation that neglects the inertial terms and uses transport coefficients that are obtained by solving the spatially-homogeneous electron kinetic in the two-term approximation. We propose an alternative model that is based on velocity-moments of the kinetic equation. In addition to conservation of mass, momentum, and energy, the moment models consider evolution equations for higher-order moments such as the heat-flux, pressure tensor, etc., which allow for capturing stronger non-equilibrium conditions. In this work, we will determine the closure of the higher-order moments while considering the relevant collisions in low-temperature plasmas, e.g., between the electrons and the gas (both elastic and inelastic) and Coulomb collisions (collisions between charged particles). We will compare our closure to the solutions of a Monte Carlo solver and a two-term Boltzmann model. The convergence with number of models as well as the numerical difficulties of these models will be discussed. |
Thursday, October 6, 2022 1:45PM - 2:00PM |
GR4.00002: Development of a 10-Moment Multi-Fluid Model for Low-Temperature Magnetized Plasmas Derek Kuldinow, Kentaro Hara Fluid moment models are an attractive option to model devices because they can describe the collective behavior of an ensemble of particles without needing to track individual trajectories. In low-temperature plasmas, non-Maxwellian effects may arise due to the coupling between plasma-wall interactions, collisions, and instabilities. While kinetic models can describe distributions far from equilibrium, these models are often computationally more expensive than conventional fluid models. In this study, a 10-moment multi-fluid model is developed to capture non-equilibrium effects in low-temperature magnetized plasmas. The present model solves for the number density, three components of bulk velocity, and six components of a symmetric pressure tensor. A one-dimensional model is developed to simulate the dynamics of ions, electrons, and neutrals. The 10-moment model is applied to the discharge plasma in a Hall Effect Thruster channel and compared to the results obtained from a 5-moment study. The off-diagonal terms of the pressure tensor allow for a direct modeling of shear, which can give a better understanding of shear-induced transport and finite non-Maxwellian effects in low-temperature magnetized plasmas. |
Thursday, October 6, 2022 2:00PM - 2:15PM |
GR4.00003: Recent progress on asymptotic preserving finite-volume methods for fluid models in low-temperature partially-magnetized plasma applications involving instabilities. Louis Reboul, Alejandro Alvarez Laguna, Anne Bourdon, Marc Massot Multi-fluid plasma models are able to represent the scale disparity between the different species within plasmas while being theoretically less expensive than kinetic approaches. Nevertheless, stability constraints, which imply that the time step must be smaller than the inverse of the electron plasma frequency and that the mesh size should be below the Debye length, are extremely restrictive and prevent finite volume method from outperforming PIC methods in terms of computational cost. We have recently proposed a series of so-called asymptotic preserving (AP) scheme that remains stable even when these conditions are not met and yield very accurate results even in the low Mach regime for the electrons. This approach thus allows for a significant reduction of the simulation time. Following the same path, we focus in this contribution on the extension to second order of these schemes while maintaining the AP properties, as well as introduce a well-balanced version of the scheme specifically designed to correctly balance the source terms and the convective parts of the equation. Special care is devoted to the implementation of boundary conditions. The approach is assessed through a thorough comparison to reference PIC simulation obtained via the LPPic code. |
Thursday, October 6, 2022 2:15PM - 2:30PM |
GR4.00004: Plasma Chamber Design Method Combined with Plasma Deep Learning Model and Optimization Algorithm JungMin Ko, Jinkyu Bae, Byungjo Kim, Hyunjae Lee, Younghyun Jo, Sangki Nam As plasma process becomes more complex, plasma simulation is becoming important but has several disadvantages such as high computation time. In addition, it is difficult to find a solution that satisfies multiple purposes at the same time. In this study, a deep learning model was implemented based on HPEM (The Hybrid Plasma Equipment Model) plasma simulation to ensure convergence and shorten the calculation time. The neural network consists of two encoders and a decoder, and each encoder distills process recipe and geometry information into a dense vector through fully-connected layers and convolution layers. MOPSO (Multi-Objective Particle Swarm Optimization) algorithm is also applied so that an optimized solution can be derived automatically as the iterations are repeated. The final algorithm can automatically split the selected variables and analyze the results to find optimized conditions for multiple objectives. In addition, the optimizing process which takes dozens of days was reduced to tens of seconds due to the deep learning model, maintaining 95% consistency of HPEM data. |
Thursday, October 6, 2022 2:30PM - 2:45PM |
GR4.00005: Exploring Physics Informed Neural Networks for Solving an Anisotropic Diffusion Equation Arising in Plasma Kinetics Vladimir I Kolobov, Lucius Schoenbaum The recent success of deep neural networks in artificial intelligence encourages their applications to solve high-dimensional Partial Differential Equations (PDEs) describing the particle kinetics in low-temperature plasma (LTP). Physics-informed neural networks (PINNs) have already been implemented in the NVIDIA’s SimNet and Modulus software and demonstrated for different physical systems. Although traditional numerical methods are still generally favored for forward problems, PINNs have been applied to inverse problems that cannot be solved with traditional techniques. In this work, we report on investigations to apply statistical factor analysis techniques to tune hyperparameters of PINNs for solving PDFs arising in plasma physics. The case is the anisotropic diffusion equation, which describes, for example, a surface diffusion over a sphere. This equation appears in modeling electron kinetics in phase space in collisional LTP. We test the effects of hyperparameters and platforms on the performance and accuracy of PINN solvers on modern GPU/CUDA platforms. Our experiments indicate an overall 8x speedup effect from the mean due to GPU and CUDA libraries versus CPU-based processing and indicate a correlation of the number of hidden layers with the number of epochs of training, showing that with an increase in hidden layers, the number of epochs of training can be lowered without impacting the accuracy of results. |
Thursday, October 6, 2022 2:45PM - 3:00PM |
GR4.00006: An Open Source, Three-Dimensional, Kinetic Code for Modelling Low-Temperature Plasmas on Modern Supercomputing Architectures Andrew T Powis, Johan A Carlsson, Stephane A Ethier, Alexander Khaneles, Grant Johnson, Maxwell Rosen, Igor D Kaganovich Over the past four years the Princeton Plasma Physics Laboratory has been developing a new kinetic particle-in-cell code designed for use by the low-temperature plasma community. The code, LTP-PIC models complex geometry on a uniform Cartesian mesh in two and three dimensions, incorporating a geometric multigrid linear algebra solver for the Poisson equation. LTP-PIC can handle an arbitrary number of charged species, which can interact with a uniform neutral background via elastic and inelastic collisional processes, including ionization and charge exchange. Surface interactions including secondary electron emission and charge accumulation on dielectric surfaces can also be modelled. |
Thursday, October 6, 2022 3:00PM - 3:15PM |
GR4.00007: N-body charged particle simulation in two- and three-dimensional systems Yasutaro Nishimura A computational tool is developed to investigate charged particle transport in the presence of magnetic fields [D. Bohm, The characteristics of electrical discharges in magnetic fields, New York: McGraw-Hill (1949)] based on the first principle based N-body particle simulation. Instead of employing conventional softening parameters at the binary collisions, we have employed the analytical Kepler solutions when the kinetic energy and the Coulomb potential's energy became comparable, which are successfully connected to the numerical solutions of the N-body simulation. We assume that the interactions from the rest of the N-2 particles are small compared to the binary interactions (the effect of the central forces from the N-2 particles can be incorporated perturbatively). The work is extended to a three-dimensional system. At the binary collisions, reminding Kepler's first law, we find out local planes where the two orbits reside. In a boundary-less system, particles' diffusion coefficients are estimated by varying the strength of the background magnetic field. |
Thursday, October 6, 2022 3:15PM - 3:30PM |
GR4.00008: The LisbOn KInetics Monte Carlo solver Tiago C C Dias, Antonio Tejero-del-Caz, Luís L Alves, Carlos D Pintassilgo, Vasco Guerra The LisbOn KInetics Monte Carlo (LoKI-MC) is an open-source code for the Monte Carlo simulation of electron transport in an arbitrarily complex gas mixture [1]. The simulation tool can address electron-neutral collisions with any target state (electronic, vibrational and rotational), characterized by any user-prescribed population, allowing to include the effects of superelastic collisions in a general manner. The influence of the thermal motion of the background molecules is considered, enabling the description of electron swarm properties at low reduced electric fields E/N. On output, the code provides: the electron energy and velocity distribution functions; flux and bulk swarm parameters; collision rates; power balance; and the spatiotemporal evolution of the electron swarm. The code will be released as part of the simulation tools provided by the N-PRiME group at Lisbon. We will also discuss the features of future versions: inclusion of time-dependent electric fields, magnetic fields (crossed at any arbitrary angles) and anisotropic scattering for any type of collision. |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700