Keith Cartwright - Academia.edu (original) (raw)

Papers by Keith Cartwright

Bulletin of the American Physical Society, Oct 28, 2014

Comparison of velocity distribution function errors introduced by particle reweighting schemes in... more Comparison of velocity distribution function errors introduced by particle reweighting schemes in PIC-DSMC simulations 1 CHRISTOPHER MOORE, JEREMY BOERNER, STAN MOORE, KEITH CARTWRIGHT, TIM-OTHY POINTON, Sandia Natl Labs -Many PIC simulations span many orders of magnitude in the plasma density and therefore a constant particle weight results in too few particles in regions (or time periods) of low density or too many particles when the density is high. The standard solution is to employ a reweighting scheme in which low-weight particles are merged in order to keep the number of particles per cell roughly constant while conserving mass and momentum. Unfortunately merger schemes distort a general velocity distribution function (VDF) of particles (one can conserve arbitrarily higher moments such as energy flux by merging N to M particles for N>M>1) and often merge routines act like artificial collisions that thermalize the distribution and lead to simulation error. We will compare the accuracy of the unique reweighting scheme used in our PIC-DSMC code and common reweighting schemes (e.g. redrawing from a constructed VDF or rouletting) through two benchmarks. The first compares the time varying VDF from various merge routines to an analytic solution for relaxation of a bimodal VDF to a Maxwellian through elastic collisions. The second benchmark compares error introduced in the VDF due to merging electrons during a breakdown simulation.

Bulletin of the American Physical Society, Oct 24, 2005

The Effect of Secondary Emission Cathode Parameters on (Near-) Brillouin Flow in Crossed-Field Di... more The Effect of Secondary Emission Cathode Parameters on (Near-) Brillouin Flow in Crossed-Field Diodes CHRISTOPHER FICHTL, TIMO-THY FLEMING, KEITH CARTWRIGHT, CHRISTOPHER LENYK, Air Force Research Lab, ICEPIC TEAM -The initial velocity that an electron has from the cathode can change the magnetic field needed to insulate a crossed-field diode 1 . For a secondary emitting cathode the distribution of emitted electron velocities depends on the velocity distribution of electrons returning to the cathode. We have studied the evolution of the Brillouin hub in a crossed-field diode in self-consistent 1d electromagnetic Particle-in-Cell (PIC) code with thermal emission and a secondary emission model by Vaughan 2 . The baseline simulations have thermal emission with 0.1 eV of temperature; this is compared to simulations that have both secondary emission and thermal emission. The fraction of reflected and back scattered primaries is varied to induce perturbations in the Brillouin hub. The temperature of the true secondaries is conveniently set to the thermal emission temperature of 0.1 eV. It is found that relatively few (∼10%) reflected and back scattered primaries allow the Brillouin hub to expand further across the diode as compared to the thermal emission cathode.

Bulletin of the American Physical Society, Oct 27, 2017

Experiments are being conducted using streaked, visible spectroscopy to obtain time histories of ... more Experiments are being conducted using streaked, visible spectroscopy to obtain time histories of plasma formation and propagation throughout the final power flow regions on Z, where currents and fields are at their highest. Plasmas draw current away from the load, causing the machine to be less efficient; depending on the specific load, losses of up to 20% can occur. This paper describes the first comprehensive attempt to characterize these plasmas on Z. We outline the experimental techniques used to make these measurements, provide results obtained to date, and draw comparisons with hybrid fluid-PIC simulations.

Bulletin of the American Physical Society, Nov 4, 2014

A kinetic electron-neutral collision model for particle-in-cell plasma simulation 1 TIMOTHY POINT... more A kinetic electron-neutral collision model for particle-in-cell plasma simulation 1 TIMOTHY POINTON, KEITH CARTWRIGHT, Sandia National Laboratories -Details of a kinetic electron-neutral collision model for particle-in-cell plasma simulation codes are presented. The model uses an efficient scheme to randomly select collision events -elastic, excitation and ionization -with the appropriate probability [H. Sugawara, et al., J. Comput. Phys. 223, 298 (2007).] Ionization events create electron-ion pairs, and the secondary electrons can themselves ionize the gas. To maintain a manageable particle count, a particle merger algorithm can be used to periodically replace all particles of a given species in a cell with a new, smaller set that conserves charge, momentum, and energy [D. R. Welch, et al., J. Comput. Phys. 227, 143 (2007).] Small-scale tests show that results with the merger are in good agreement with non-merged runs. Large simulations can only be done with the merger on, and typically show excellent merger efficiency (>90%).

arXiv (Cornell University), Sep 16, 2004

Interest in air breakdown phenomena has recently been re-kindled with the advent of advanced virt... more Interest in air breakdown phenomena has recently been re-kindled with the advent of advanced virtual prototyping of radio frequency (RF) sources for use in high power microwave (HPM) weapons technology. Air breakdown phenomena are of interest because the formation of a plasma layer at the aperture of an RF source decreases the transmitted power to the target, and in some cases can cause significant reflection of RF radiation. Understanding the mechanisms behind the formation of such plasma layers will aid in the development of maximally effective sources. This paper begins with some of the basic theory behind air breakdown, and describes two independent approaches to modeling the formation of plasmas, the dielectric fluid model and the Particle in Cell (PIC) approach. Finally we present the results of preliminary studies in numerical modeling and simulation of breakdown.

Bulletin of the American Physical Society, Oct 15, 2015

Numerical parameter constraints for accurate PIC-DSMC simulation of breakdown from arc initiation... more Numerical parameter constraints for accurate PIC-DSMC simulation of breakdown from arc initiation to stable arcs 1 CHRISTOPHER MOORE, MATTHEW HOPKINS, STAN MOORE, JEREMIAH BOERNER, KEITH CARTWRIGHT, Sandia Natl Labs -Simulation of breakdown is important for understanding and designing a variety of applications such as mitigating undesirable discharge events. Such simulations need to be accurate through early time arc initiation to late time stable arc behavior. Here we examine constraints on the timestep and mesh size required for arc simulations using the particle-in-cell (PIC) method with direct simulation Monte Carlo (DMSC) collisions. Accurate simulation of electron avalanche across a fixed voltage drop and constant neutral density (reduced field of 1000 Td) was found to require a timestep ∼ 1/100 of the mean time between collisions and a mesh size ∼ 1/25 the mean free path. These constraints are much smaller than the typical PIC-DSMC requirements for timestep and mesh size. Both constraints are related to the fact that charged particles are accelerated by the external field. Thus gradients in the electron energy distribution function can exist at scales smaller than the mean free path and these must be resolved by the mesh size for accurate collision rates. Additionally, the timestep must be small enough that the particle energy change due to the fields be small in order to capture gradients in the cross sections versus energy.

Bulletin of the American Physical Society, Oct 29, 2019

CESTA -As part of the validation effort for Sandia's new EM PIC-DSMC plasma code EMPIRE, we simul... more CESTA -As part of the validation effort for Sandia's new EM PIC-DSMC plasma code EMPIRE, we simulate high intensity electron beam propagation through low pressure gas cells. Specifically, we model the CESAR and RKA beam experiments [1]. We compare the accuracy and performance of a fully kinetic PIC-DSMC scheme [2] and a hybrid fluid-kinetic scheme for modeling the electron transport through the Ar background gas. The kinetic PIC-DSMC model represents charged and neutral species as computational particles allowing for self-consistent evolution of the neutrals as the beam interacts with the Ar gas. In the hybrid scheme PIC-MCC collisions generate mass, momentum, and energy source terms for the evolution of the neutral fluid. 1. Gardelle, J. et al., "Revisiting the propagation and focusing of a high intensity electron beam in a low-pressure gas cell", 44th

Bulletin of the American Physical Society, Nov 6, 2018

CESTA -As part of the validation effort for Sandia's new EM PIC-DSMC plasma code EMPIRE [1], we h... more CESTA -As part of the validation effort for Sandia's new EM PIC-DSMC plasma code EMPIRE [1], we have begun to simulate high intensity vacuum beam propagation. Specifically, we are modeling the CESAR and RKA beam experiments [2], starting with vacuum propagation and proceeding to beam propagation through a low-pressure Ar background gas. EMPIRE models both charged particles and neutrals as computational particles that can move and collide with one another allowing for self-consistent evolution of the neutral gas as the e-beam propagates and interacts with the background gas. In the current work we will show comparisons for the current and beam radius to the CESEAR beam experiments. In addition, we will investigate EMPIRE's performance/scaling on multiple architectures (CPU's, MIC's, and GPU's) for the simulations. 1. Markosyan, A. et al, "Method of manufactured solutions for verification of particle-in-cell simulations", 45 th ICOPS, June 24-28 2018. 2. Gardelle, J. et al., "Revisiting the propagation and focusing of a high intensity electron beam in a low-pressure gas cell", 44th ICOPS, May 21-25, 2017.

Bulletin of the American Physical Society, Nov 17, 2008

The American Physical Society Boundary Integral Treecode (BIT) as a sub-cell method in Particle-I... more The American Physical Society Boundary Integral Treecode (BIT) as a sub-cell method in Particle-In-Cell (PIC) 1 ANDREW CHRISTLIEB, Michigan State University, KEITH CARTWRIGHT, AFRL/RDHE -The Boundary Integral Treecode (BIT) is a method for computing long-range forces in O(N log N ) without making use of an underlying mesh. The method use the point cluster form of fast summation. BIT has been shown to exhibit less numerical heating with higher accuracy than PIC and has been recently proposed as a sub-cell method in PIC as a way of extending efficient legacy PIC codes to dense plasma problems, where numerical heating is a challenge. The idea is that sub-cell BIT can extend PIC by circumventing the need to follow the traditional rule of thumb of ∆x < λ D , which controls numerical heating in traditional explicit PIC codes. This has been demonstrated in 1D periodic test problems. To use high order explicit time stepping, BIT and BIT corrected PIC use a regularized force kernel. In the work on BIT corrected PIC, the regularization was found to have a negative impact near boundaries. To overcome this issue, a systematic approach to localization of the kernel, based on Taylor expansions, and rigorous error bounds for the error near a boundary were developed. This paper discusses the extension of BIT and BIT corrected PIC to non-periodic domains through the use of adaptive regularization to control the error near boundaries. The 1D virtual cathode problem is investigated.

SSRN Electronic Journal

In this work, we present a discontinuous-Galerkin method for evolving relativistic hydrodynamics.... more In this work, we present a discontinuous-Galerkin method for evolving relativistic hydrodynamics. We include an exploration of analytical and iterative methods to recover the primitive variables from the conserved variables for the ideal equation of state and the Taub-Matthews approximation to the Synge equation of state. We also present a new operator for enforcing a physically permissible conserved state at all basis points within an element while preserving the volume average of the conserved state. We implement this method using the Kokkos performance-portability library to enable running at performance on both CPUs and GPUs. We use this method to explore the relativistic Kelvin-Helmholtz instability compared to a finite volume method. Last, we explore the performance of our implementation on CPUs and GPUs.

Physics of Plasmas

An approach to numerically modeling relativistic magnetrons, in which the electrons are represent... more An approach to numerically modeling relativistic magnetrons, in which the electrons are represented with a relativistic fluid, is described. A principal effect in the operation of a magnetron is space-charge-limited (SCL) emission of electrons from the cathode. We have developed an approximate SCL emission boundary condition for the fluid electron model. This boundary condition prescribes the flux of electrons as a function of the normal component of the electric field on the boundary. We show the results of a benchmarking activity that applies the fluid SCL boundary condition to the one-dimensional Child–Langmuir diode problem and a canonical two-dimensional diode problem. Simulation results for a two-dimensional A6 magnetron are then presented. Computed bunching of the electron cloud occurs and coincides with significant microwave power generation. Numerical convergence of the solution is considered. Sharp gradients in the solution quantities at the diocotron resonance, spanning a...

Bulletin of the American Physical Society, 2014

Proposed for presentation at the 73rd Annual Gaseous Electronics Virtual Conference held October 5-9, 2020., 2020

The Unstructured Time-Domain ElectroMagnetics (UTDEM) portion of the EMPHASIS suite solves Maxwel... more The Unstructured Time-Domain ElectroMagnetics (UTDEM) portion of the EMPHASIS suite solves Maxwell's equations using finite-element techniques on unstructured meshes. This document provides user-specific information to facilitate the use of the code for applications of interest.

This report documents the outcome from the ASC ATDM Level 2 Milestone 6358: Assess Status of Next... more This report documents the outcome from the ASC ATDM Level 2 Milestone 6358: Assess Status of Next Generation Components and Physics Models in EMPIRE. This Milestone is an assessment of the EMPIRE (ElectroMagnetic Plasma In Realistic Environments) application and three software components. The assessment focuses on the electromagnetic and electrostatic particle-in-cell solutions for EMPIRE and its associated solver, time integration, and checkpoint-restart components. This information provides a clear understanding of the current status of the EMPIRE application and will help to guide future work in FY19 in order to ready the application for the ASC ATDM L 1 Milestone in FY20. It is clear from this assessment that performance of the linear solver will have to be a focus in FY19.

Journal of Computational Physics, 2000

Existing and new particle loading and injection algorithms for particle simulations are analyzed ... more Existing and new particle loading and injection algorithms for particle simulations are analyzed to determine numerical accuracy and computational efficiency. Emphasis has been placed on loading and emission of Maxwellian, drifting Maxwellian, and cutoff Maxwellian velocity distributions. Once a velocity distribution has been inverted for loading or injection, time-centering of the position and velocity is necessary in order to maintain second-order accuracy. Here, the accuracy of these methods is determined and compared to three analytic test cases with spatially varying, timedependent, and time-independent electric fields in a homogeneous magnetic field and a self-consistent crossed-field diode. The initial push is shown to be important in calculating the correct electric field at the boundary where particles are injected, in relaxing constraints on the time step, and in providing reliable field fluctuations due to particle statistics.

IEEE Transactions on Microwave Theory and Techniques, 2009

See http://barkley. ME. Berkeley. EDU/~ …, 1996

This report discusses an implementation of an Adaptive Mesh Refinement (AMR) Poisson solver which... more This report discusses an implementation of an Adaptive Mesh Refinement (AMR) Poisson solver which solves Poisson's equation using multigrid relaxation. Local refinement introduces several added issues. Special care has been taken to match the solution across coarse/fine interfaces so that the solution maintains global second order accuracy. The nested mesh hierarchy can be defined by the user, Richardson extrapolation, or a user supplied criterion. Extensive use of Boxlib [2] has reduced the bookkeeping needed by the authors.