arXiv subjects

R. M. Churchill

Publications and source records attributed to R. M. Churchill.

12 recordsLinked to original sources

arXiv2022

2022 Review of Data-Driven Plasma Science

Data science and technology offer transformative tools and methods to science. This review article highlights latest development and progress in the interdisciplinary field of data-driven plasma science (DDPS). A large amount of data and machine learning algorithms go hand in hand. Most plasma data, whether experimental, observational or computational, are generated or collected by machines today. It is now becoming impractical for humans to analyze all the data manually. Therefore, it is imperative to train machines to analyze and interpret (eventually) such data as intelligently as humans but far more efficiently in quantity. Despite the recent impressive progress in applications of data science to plasma science and technology, the emerging field of DDPS is still in its infancy. Fueled by some of the most challenging problems such as fusion energy, plasma processing of materials, and fundamental understanding of the universe through observable plasma phenomena, it is expected that DDPS continues to benefit significantly from the interdisciplinary marriage between plasma science and data science into the foreseeable future.

physics.plasm-ph↗

arXiv2022

Electromagnetic total-f algorithm for gyrokinetic particle-in-cell simulations of boundary plasma in XGC

The simplified delta-f mixed-variable/pull-back electromagnetic simulation algorithm implemented in XGC for core plasma simulations by M. Cole et al. [Phys. Plasmas 28, 034501 (2021)] has been generalized to a total-f electromagnetic algorithm that can include, for the first time, the boundary plasma in diverted magnetic geometry with neutral particle recycling, turbulence and neoclassical physics. The delta-f mixed-variable/pull-back electromagnetic is based on the pioneering work by Kleiber and Mischenko et al. [R. Kleiber et al., Phys. Plasmas 23, 032501 (2016); A. Mishchenko et al., Comput. Phys. Commun. 238, 194 (2019)]. An electromagnetic demonstration simulation is performed in a DIII-D-like, H-mode boundary plasma, including a corresponding comparative electrostatic simulation, which confirms that the electromagnetic simulation is necessary for a higher fidelity understanding of the electron particle and heat transport even at the low-beta pedestal foot in the vicinity of the magnetic separatrix.

physics.plasm-ph↗

arXiv2021

Machine learning accelerated particle-in-cell plasma simulations

Particle-In-Cell (PIC) methods are frequently used for kinetic, high-fidelity simulations of plasmas. Implicit formulations of PIC algorithms feature strong conservation properties, up to numerical round-off errors, and are not subject to time-step limitations which make them an attractive candidate to use in simulations fusion plasmas. Currently they remain prohibitively expensive for high-fidelity simulation of macroscopic plasmas. We investigate how amortized solvers can be incorporated with PIC methods for simulations of plasmas. Incorporated into the amortized solver, a neural network predicts a vector space that entails an approximate solution of the PIC system. The network uses only fluid moments and the electric field as input and its output is used to augment the vector space of an iterative linear solver. We find that this approach reduces the average number of required solver iterations by about 25% when simulating electron plasma oscillations. This novel approach may allow to accelerate implicit PIC simulations while retaining all conservation laws and may also be appropriate for multi-scale systems.

physics.plasm-ph↗

arXiv2021

Constructing a new predictive scaling formula for ITER's divertor heat-load width informed by a simulation-anchored machine learning

Understanding and predicting divertor heat-load width $λ_q$ is a critically important problem for an easier and more robust operation of ITER with high fusion gain. Previous predictive simulation data for $λ_q$ using the extreme-scale edge gyrokinetic code XGC1 in the electrostatic limit under attached divertor plasma conditions in three major US tokamaks [C.S. Chang et al., Nucl. Fusion 57, 116023 (2017)] reproduced the Eich and Goldston attached-divertor formula results [formula #14 in T. Eich et al., Nucl. Fusion 53, 093031 (2013); R.J. Goldston, Nucl. Fusion 52, 013009 (2012)], and furthermore predicted over six times wider $λ_q$ than the maximal Eich and Goldston formula predictions on a full-power (Q = 10) scenario ITER plasma. After adding data from further predictive simulations on a highest current JET and highest-current Alcator C-Mod, a machine learning program is used to identify a new scaling formula for $λ_q$ as a simple modification to the Eich formula #14, which reproduces the Eich scaling formula for the present tokamaks and which embraces the wide $λ_q^X{GC}$ for the full-current Q = 10 ITER plasma. The new formula is then successfully tested on three more ITER plasmas: two corresponding to long burning scenarios with Q = 5 and one at low plasma current to be explored in the initial phases of ITER operation. The new physics that gives rise to the wider $λq_^{XGC} is identified to be the weakly-collisional, trapped-electron-mode turbulence across the magnetic separatrix, which is known to be an efficient transporter of the electron heat and mass. Electromagnetic turbulence and high-collisionality effects on the new formula are the next study topics for XGC1.

physics.plasm-ph↗

arXiv2020

Encoder-decoder neural network for solving the nonlinear Fokker-Planck-Landau collision operator in XGC

An encoder-decoder neural network has been used to examine the possibility for acceleration of a partial integro-differential equation, the Fokker-Planck-Landau collision operator. This is part of the governing equation in the massively parallel particle-in-cell code, XGC, which is used to study turbulence in fusion energy devices. The neural network emphasizes physics-inspired learning, where it is taught to respect physical conservation constraints of the collision operator by including them in the training loss, along with the L2 loss. In particular, network architectures used for the computer vision task of semantic segmentation have been used for training. A penalization method is used to enforce the "soft" constraints of the system and integrate error in the conservation properties into the loss function. During training, quantities representing the density, momentum, and energy for all species of the system is calculated at each configuration vertex, mirroring the procedure in XGC. This simple training has produced a median relative loss, across configuration space, on the order of 10E-04, which is low enough if the error is of random nature, but not if it is of drift nature in timesteps. The run time for the Picard iterative solver of the operator scales as order n squared, where n is the number of plasma species. As the XGC1 code begins to attack problems including a larger number of species, the collision operator will become expensive computationally, making the neural network solver even more important, since the training only scales as n. A wide enough range of collisionality is considered in the training data to ensure the full domain of collision physics is captured. An advanced technique to decrease the losses further will be discussed, which will be subject of a subsequent report. Eventual work will include expansion of the network to include multiple plasma species.

physics.plasm-ph↗

arXiv2020

Comparison of edge turbulence characteristics between DIII-D and C-Mod simulations with XGC1

The physical processes taking place at the edge region are crucial for the operation of tokamaks as they govern the interaction of hot plasma with the vessel walls. Numerical modeling of the edge with state-of-the-art codes attempts to elucidate interactions between neoclassical drifts, turbulence, poloidal and parallel flows that control the physical set-up of the SOL region. Here, we present post-processing analysis of simulations from the gyrokinetic code XGC1, comparing edge turbulence characteristics from a simulation of DIII-D against one of C-Mod. We find that the equilibrium $E \times B$ flux across the separatrix has a similar poloidal pattern in both discharges which can be explained by magnetic drifts and trapped ion excursions. However, collisionality is noted to play a major role in that it prevents local charge accumulations from having global effects in C-Mod. In both cases, turbulent electron heat flux is higher than the ion one. This seems to be a universal characteristic of the tokamak edge. We identify turbulent frequencies and growth rates of the dominant mode in both simulations. In C-Mod, these numbers point to the presence of a drift wave. In DIII-D, linear simulations with Gene reveal a trapped electron mode. Furthermore, we present the amplitude and size distributions of the blobs from both simulations. Amplitude distributions are in qualitative agreement with experimental observations while size distributions are consistent with the fact that most blobs are not connecting to the divertor plates and suggest that they are generated by the shearing of the turbulent modes.

physics.plasm-ph↗

arXiv2019

Deep convolutional neural networks for multi-scale time-series classification and application to disruption prediction in fusion devices

The multi-scale, mutli-physics nature of fusion plasmas makes predicting plasma events challenging. Recent advances in deep convolutional neural network architectures (CNN) utilizing dilated convolutions enable accurate predictions on sequences which have long-range, multi-scale characteristics, such as the time-series generated by diagnostic instruments observing fusion plasmas. Here we apply this neural network architecture to the popular problem of disruption prediction in fusion tokamaks, utilizing raw data from a single diagnostic, the Electron Cyclotron Emission imaging (ECEi) diagnostic from the DIII-D tokamak. ECEi measures a fundamental plasma quantity (electron temperature) with high temporal resolution over the entire plasma discharge, making it sensitive to a number of potential pre-disruptions markers with different temporal and spatial scales. Promising, initial disruption prediction results are obtained training a deep CNN with large receptive field (~30k), achieving an $F_1$-score of ~91% on individual time-slices using only the ECEi data.

physics.plasm-ph↗

arXiv2018

Analysis of equilibrium and turbulent fluxes across the separatrix in a gyrokinetic simulation

The SOL width is a parameter of paramount importance in modern tokamaks as it controls the power density deposited at the divertor plates, critical for plasma-facing material survivability. An understanding of the parameters controlling it has consequently long been sought (Connor et al. 1999 NF 39 2). Prior to Chang et al.(2017 NF 57 11), studies of the tokamak edge have been mostly confined to reduced fluid models and simplified geometries, leaving out important pieces of physics. Here, we analyze the results of a DIII-D simulation performed with the full-f gyrokinetic code XGC1 which includes both turbulence and neoclassical effects in realistic divertor geometry. More specifically, we calculate the particle and heat ExB fluxes along the separatrix, discriminating between equilibrium and turbulent contributions. We find that the density SOL width is impacted almost exclusively by the turbulent electron flux. In this simulation, the level of edge turbulence is regulated by a mechanism we are only beginning to understand: $\nabla B$-drifts and ion X-point losses at the top and bottom of the machine, along with ion banana orbits at the low field side (LFS), result in a complex poloidal potential structure at the separatrix which is the cause of the ExB drift pattern that we observe. Turbulence is being suppressed by the shear flows that this potential generates. At the same time, turbulence, along with increased edge collisionality and electron inertia, can influence the shape of the potential structure by making the electrons non-adiabatic. Moreover, being the only means through which the electrons can lose confinement, it needs to be in a balance with the original direct ion orbit losses to maintain charge neutrality.

physics.plasm-ph↗

arXiv2017

A fast low-to-high mode bifurcation dynamics in a tokamak edge plasma gyrokinetic simulation

Transport barrier formation and its relation to sheared flows in fluids and plasmas are of fundamental interest in various natural and laboratory observations and of critical importance in achieving an economical energy production in a magnetic fusion device. Here we report the first observation of an edge transport barrier formation event in a gyrokinetic simulation carried out in a realistic tokamak edge geometry. The results show that turbulent Reynolds stress driven sheared ExB flows act in concert with neoclassical orbit loss to quench turbulent transport and form a transport barrier just inside the last closed magnetic flux surface.

physics.plasm-ph↗

arXiv2017

Pedestal and edge turbulence characteristics from an XGC1 gyrokinetic simulation

Understanding the multi-scale neoclassical and turbulence physics in the edge region (pedestal + scrape-off layer) is required in order to reliably predict performance in future fusion devices. We explore turbulent characteristics in the edge region from a multiscale neoclassical and turbulent XGC1 gyrokinetic simulation in a DIII-D like tokamak geometry, here excluding neutrals and collisions. For an H-mode type plasma with steep pedestal, it is found that the electron density fluctuations increase towards the separatrix, and stay high well into the SOL, reaching a maximum value of $δn_e / \bar{n}_e \sim 0.18$. Blobs are observed, born around the magnetic separatrix surface and propagate radially outward with velocities generally less than 1 km/s. Strong poloidal motion of the blobs is also present, near 20 km/s, consistent with $E \times B$ rotation. The electron density fluctuations show a negative skewness in the closed field line pedestal regions, consistent with the presence of "holes", followed by a transition to strong positive skewness across the separatrix and into the SOL. These simulations indicate that not only neoclassical phenomena, but also turbulence, including the blob-generation mechanism, can remain important in the steep H-mode pedestal and SOL. Qualitative comparisons will be made to experimental observations.

physics.plasm-ph↗

arXiv2017

Electrostatic gyrokinetic simulation of global tokamak boundary plasma and the generation of nonlinear intermittent turbulence

Boundary plasma physics plays an important role in tokamak confinement, but is difficult to simulate in a gyrokinetic code due to the scale-inseparable nonlocal multi-physics in magnetic separatrix and open magnetic field geometry. Neutral particles are also an important part of the boundary plasma physics. In the present paper, noble electrostatic gyrokinetic techniques to simulate the flux-driven, low-beta electrostatic boundary plasma is reported. Gyrokinetic ions and drift-kinetic electrons are utilized without scale-separation between the neoclassical and turbulence dynamics. It is found that the nonlinear intermittent turbulence is a natural gyrokinetic phenomenon in the boundary plasma in the vicinity of the magnetic separatrix surface and in the scrape-off layer.

physics.plasm-ph↗

arXiv2017

Total Fluid Pressure Imbalance in the Scrape-Off Layer of Tokamak Plasmas

Simulations using the fully kinetic neoclassical code XGCa were undertaken to explore the impact of kinetic effects on scrape-off layer (SOL) physics in DIII-D H-mode plasmas. XGCa is a total-f, gyrokinetic code which self-consistently calculates the axisymmetric electrostatic potential and plasma dynamics, and includes modules for Monte Carlo neutral transport. Previously presented XGCa results showed several noteworthy features, including large variations of ion density and pressure along field lines in the SOL, experimentally relevant levels of SOL parallel ion flow (Mach number~0.5), skewed ion distributions near the sheath entrance leading to subsonic flow there, and elevated sheath potentials [R.M. Churchill, Nucl. Mater. & Energy, submitted]. In this paper, we explore in detail the question of pressure balance in the SOL, as it was observed in the simulation that there was a large deviation from a simple total pressure balance (the sum of ion and electron static pressure plus ion inertia). It will be shown that both the contributions from the ion viscosity (driven by ion temperature anisotropy) and neutral source terms can be substantial, and should be retained in the parallel momentum equation in the SOL, but still falls short of accounting for the observed fluid pressure imbalance in the XGCa simulation results.

physics.plasm-ph↗