Abstract Details
| status: | file name: | submitted: | by: |
|---|---|---|---|
| approved | abstract.pdf | 2025-02-19 12:38:34 | Arnaud Jonathan |
Abstracts
Author: Arnaud S Jonathan
Requested Type: Consider for Invited
Submitted: 2025-02-19 14:11:54
Co-authors: X.-Z. Tang, C. J. McDevitt
Contact Info:
University of Florida
1698 Gale Lemerand Dr
Gainesville, Florida 32611
USA
Abstract Text:
Research on mitigating damage from tokamak disruptions is limited by the inability of experiments to access relevant plasma conditions expected in future devices. Additionally, the multi-physics nature of disruptions makes first-principles modeling computationally prohibitive for many-query analyses, which requires a self-consistent treatment of plasma power balance, magnetohydrodynamic (MHD) activity, and runaway electron (RE) formation. To this end, we present a novel path towards an efficient and high-fidelity integrated model of a tokamak disruption. This approach leverages an adjoint treatment of the relativistic Fokker-Planck equation[1] together with recent innovations in physics-constrained deep learning. While incorporating a fully kinetic description of RE formation and evolution, the adjoint formulation of the relativistic Fokker-Planck equation employed is tailored to only predict quantities of interest needed to close the coupled MHD-RE system such as RE density or current. It is shown that such an adjoint problem can be solved across a broad range of plasma conditions using a Physics-Informed Neural Network (PINN)[2]. The resulting surrogate allows for near instantaneous online predictions of RE density, while incorporating a fully kinetic description of RE physics including essential physical processes such as partial screening and radiation. As an initial application, the RE surrogate is coupled with a reduced yet fully self-consistent model of a tokamak disruption. This efficient integrated model is used to explore the high-dimensional space of potential disruption mitigation strategies, thus motivating a path towards accelerating disruption research.
[1] C. F. Karney and N. J. Fisch, The Physics of Fluids 29, 180 (1986).
[2] J. S. Arnaud, T. B. Mark, and C. J. McDevitt, Journal of Plasma Physics 90, 905900409 (2024).
Characterization: 6.0
Comments:
This presentation is also relevant for category 5: Integrated
modeling of fusion plasmas.