Reconstructing the fast-ion velocity distribution in fusion plasmas from sparse datasets

Birgitte Madsen

Research output: Book/ReportPh.D. thesis

72 Downloads (Pure)


Fast ions play a crucial role in current experimental fusion reactors, and highly energetic fusion-born alpha particles are envisioned to be the main source of heat in future nuclear fusion power plants. However, in current-day reactors, energetic auxiliary heating-born deuterium ions are often found to be affected by plasma instabilities and resonantly interact with wave activity leading to anomalous fastion transport and loss. Through tomographic inversion of combined fast-ion measurements, the fast-ion velocity distribution can be reconstructed, allowing detailed studies on fast-ion transport. Often this is a sparse-dataset problem, and reconstructions of high-resolution images can typically only be achieved by including prior information. This work introduces novel and improved inversion strategies using Tikhonov regularization with the goal of enabling measurement-based reconstructions of fast-ion distributions from diagnostic setups where standard methods do not necessarily succeed. Reconstructions of fast-ion distributions are calculated using the proposed strategies for different scenarios and diagnostic capabilities: sawtooth activity in the MAST tokamak, Alfvén eigenmode activity in the DIII-D tokamak and various neutral beam heating schemes in the EAST tokamak. Additional preliminary tomographic results at the TCV tokamak are presented. The reconstructed distributions are subsequently compared to numerical models with the aim of improving the understanding and predictability of fast-ion transport in both current and future fusion reactors.
Original languageEnglish
PublisherDepartment of Physics, Technical University of Denmark
Number of pages147
Publication statusPublished - 2020


Dive into the research topics of 'Reconstructing the fast-ion velocity distribution in fusion plasmas from sparse datasets'. Together they form a unique fingerprint.

Cite this