Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Application of Neural Ordinary Differential Equations for ITER Burning Plasma Dynamics

Zefang Liu, Weston M. Stacey

TL;DR

The paper addresses ITER burning plasma dynamics by introducing NeuralPlasmaODE, a multinodal Neural ODE framework that learns inter-region diffusivities to capture multi-timescale energy transfer between core and edge regions. It leverages transfer learning from DIII-D data to initialize diffusivity parameters and applies fine-tuning against ITER design-phase results across inductive, hybrid, and non-inductive scenarios. The model integrates particle and energy balance equations with external heating, fusion reactions, radiation, and transport, demonstrating that radiative and transport processes effectively prevent thermal runaway in all studied cases. This approach offers a data-driven, transferable, and efficient method to simulate burning plasmas, aiding ITER planning and control by providing mechanistic insights into heat removal and stability margins.

Abstract

The dynamics of burning plasmas in tokamaks are crucial for advancing controlled thermonuclear fusion. This study applies the NeuralPlasmaODE, a multi-region multi-timescale transport model, to simulate the complex energy transfer processes in ITER deuterium-tritium (D-T) plasmas. Our model captures the interactions between energetic alpha particles, electrons, and ions, which are vital for understanding phenomena such as thermal runaway instability. We employ neural ordinary differential equations (Neural ODEs) for the numerical derivation of diffusivity parameters, enabling precise modeling of energy interactions between different plasma regions. By leveraging transfer learning, we utilize model parameters derived from DIII-D experimental data, enhancing the efficiency and accuracy of our simulations without training from scratch. Applying this model to ITER's inductive and non-inductive operational scenarios, our results demonstrate that radiation and transport processes effectively remove excess heat from the core plasma, preventing thermal runaway instability. This study underscores the potential of machine learning in advancing our understanding and control of burning plasma dynamics in fusion reactors.

Application of Neural Ordinary Differential Equations for ITER Burning Plasma Dynamics

TL;DR

The paper addresses ITER burning plasma dynamics by introducing NeuralPlasmaODE, a multinodal Neural ODE framework that learns inter-region diffusivities to capture multi-timescale energy transfer between core and edge regions. It leverages transfer learning from DIII-D data to initialize diffusivity parameters and applies fine-tuning against ITER design-phase results across inductive, hybrid, and non-inductive scenarios. The model integrates particle and energy balance equations with external heating, fusion reactions, radiation, and transport, demonstrating that radiative and transport processes effectively prevent thermal runaway in all studied cases. This approach offers a data-driven, transferable, and efficient method to simulate burning plasmas, aiding ITER planning and control by providing mechanistic insights into heat removal and stability margins.

Abstract

The dynamics of burning plasmas in tokamaks are crucial for advancing controlled thermonuclear fusion. This study applies the NeuralPlasmaODE, a multi-region multi-timescale transport model, to simulate the complex energy transfer processes in ITER deuterium-tritium (D-T) plasmas. Our model captures the interactions between energetic alpha particles, electrons, and ions, which are vital for understanding phenomena such as thermal runaway instability. We employ neural ordinary differential equations (Neural ODEs) for the numerical derivation of diffusivity parameters, enabling precise modeling of energy interactions between different plasma regions. By leveraging transfer learning, we utilize model parameters derived from DIII-D experimental data, enhancing the efficiency and accuracy of our simulations without training from scratch. Applying this model to ITER's inductive and non-inductive operational scenarios, our results demonstrate that radiation and transport processes effectively remove excess heat from the core plasma, preventing thermal runaway instability. This study underscores the potential of machine learning in advancing our understanding and control of burning plasma dynamics in fusion reactors.
Paper Structure (18 sections, 18 equations, 9 figures, 1 table)

This paper contains 18 sections, 18 equations, 9 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Burning Plasma Dynamics Model
  4. ITER Plasma Geometry
  5. Assumptions for ITER Plasmas
  6. Particle and Energy Balance Equations
  7. Particle Balance Equations
  8. Energy Balance Equations
  9. Diffusivity Models
  10. Computational Framework
  11. Simulations for ITER Plasmas
  12. Simulation Scenarios
  13. Simulation Methodology
  14. Simulation Results
  15. Inductive Operation Scenario
  16. ...and 3 more sections

Figures (9)

  • Figure 1: Multinodal model geometry of tokamak plasmas, where the first figure shows the cross section of an ITER plasma, and the second figure is the simplified geometry in the multinodal model.
  • Figure 2: Computational framework of NeuralPlasmaODE, including cylinders as datasets, squares as modules, solid lines as forward flows, and dashed lines as back propagation processes.
  • Figure 3: Typical radial profiles of plasma temperatures and densities in the ITER inductive and non-inductive operation scenarios (reproduced with permission from iaea2002iter).
  • Figure 4: Densities and temperatures of ITER inductive operation scenario 2.
  • Figure 5: Powers of ITER inductive operation scenario 2.
  • ...and 4 more figures