Fast particles, both electrons and ions, play an important role for the success of the next generation of large tokamak devices, such as ITER, that will prove the feasibility of magnetically confined thermonuclear fusion as an energy source. Ions accelerated by external heating or born in fusion reactions can reach energies in the MeV range. Their primary role is to sustain the plasma temperature and the fusion reaction rate, thus lots of efforts have been put into the development of efficient heating schemes and in the improvement of their confinement. On the other hand, during fast terminations of plasma pulses on tokamaks, electrons can accelerate to relativistic velocities, entering the runaway regime. Runaway electrons have enough energy to seriously damage the plasma facing components of large tokamaks, thus mitigation techniques are under study in view of ITER operations. This thesis focuses on the implementation of deconvolution techniques for the reconstruction of the fast particles distributions from their emission in the MeV energy range. The problem was approached from two different perspectives: the unfolding of the runaway electrons velocity-space distribution from spectroscopic measurements of their bremsstrahlung emission, and the tomographic reconstruction of the density distribution of both fast ions and runaway electrons from the integrated measurement of their emission performed with multiple lines of sight. These algorithms were implemented in an open source Python library. Four deconvolution algorithms were implemented for the unfolding of runaway electrons energy distribution: singular value decomposition, maximum likelihood - expectation maximization, Tikhonov regularization and Poissonian regularization. The transfer matrix necessary for this inversion was calculated using the GENESIS code for estimating the probability of bremsstrahlung emission and the MCNP code for computing the detector response function. The detector response function was calculated for all the hard X-rays diagnostics systems installed at the Joint European Torus and ASDEX Upgrade tokamaks. The performance of the four methods wes then compared over both synthetic and experimental spectra, the latter being measured at ASDEX Upgrade. Maximum likelihood - expectation maximization was found to be the most accurate in the reconstruction of both the runaway electrons energy distribution and their average and maximum energies. The robustness of the four methods against experimental limitations, such as low-energy cut and low statistics, was also investigated. In the path towards the generalization of these unfolding algorithms to the reconstruction of the runaway electrons 2D velocity-space distribution, the transfer matrices in energy and pitch were calculated for all the hard X-ray diagnostics installed at JET. The weight-function formalism was adopted, which allows studying the sensitivity of the detectors to different energy and pitch regions. The matrices showed a sensitivity peak in the pitch axis which is determined by the angle between the line of sight and the magnetic field. Finally, the gamma camera upgrade installed at the Joint European Torus, with its 10 by 9 lines of sight that observe a poloidal section of the tokamak from two perpendicular projections, allows reconstructing the spatial distribution of fast particles. A tomographic algorithm that makes use of smoothing along the magnetic field lines was implemented. This tomography was first applied to recent three-ion radio frequency heating experiments in D-3He mixed plasmas, during which the gamma camera was able to detect the 16.4 MeV γ-rays from 3He(D,γ)5Li reactions. The spatial distribution of the α-particles born in 3He(D,p)4He reactions was reconstructed and the results were used to validate TRANSP simulations. The tomographic algorithm was also applied to the reconstruction of the runaway electrons spatial profiles during plasma disruptions.
Lo studio degli ioni ed elettroni veloci è fondamentale per il successo della prossima generazione di tokamak di grande dimensione, come ITER, che mirano a dimostrare la possibilità di produrre energia attraverso la fusione termonucleare. Siano essi accelerati dal riscaldamento ausiliario o nati da reazioni di fusione, gli ioni possono raggiungere energie nell’ordine dei MeV. Questa energia viene poi trasmessa al plasma attraverso collisioni, aumentandone l’energia media e il rateo delle reazioni di fusione. È quindi cruciale migliorare il confinamento degli ioni veloci e sviluppare schemi di riscaldamento efficienti. Gli elettroni, invece, possono raggiungere velocità relativistiche se accelerati da campi elettrici induttivi generati durante disruzioni di plasma. Questi elettroni runaway rappresentano una minaccia per l'integrità strutturale dei tokamak di grandi dimensioni e necessitano di essere mitigati. Questa tesi tratta di metodi di deconvoluzione applicati alla ricostruzione della distribuzione delle particelle veloci a partire dall’emissione di radiazione nel range di energia del MeV. La deconvoluzione è stata declinata in due modi: unfolding della distribuzione di velocità degli elettroni runaway a partire dalla loro radiazione di bremsstrahlung, e tomografia della distribuzione di densità di ioni veloci o elettroni runaway a partire dalla misura dei loro profili di emissione fatta con linee di vista multiple. Questi algoritmi sono stati implementati in una libreria Python open source. Quattro algoritmi per l’unfolding sono stati implementati: singular value decomposition, maximum likelihood - expectation maximization, regolarizzazione di Tikhonov and di Poisson. La matrice di probabilità necessaria per svolgere queste inversioni è stata calcolata usando il codice GENESIS per stimare la probabilità di emettere fotoni di bremsstrahlung, e il codice MCNP per calcolare la funzione di risposta del detector. Queste funzioni di risposta sono state calcolate per tutte le diagnostiche hard X-ray installate nei tokamak Joint European Torus ed ASDEX Upgrade. I quattro metodi sono stati comparati su spettri sintetici e sperimentali, questi ultimi misurati ad ASDEX Upgrade. Maximum likelihood - expectation maximization si è dimostrato il più accurato sia nella ricostruzione della distribuzione in energia degli elettroni runaway, sia nel calcolo della loro energia media e massima. È stata inoltre studiato l’effetto del taglio a basse energie applicato ai dati sperimentali e il numero minimo di conteggi nello spettro necessario per svolgere una ricostruzione. In vista di una futura ricostruzione in 2D della distribuzione delle velocità degli elettroni runaway, sono state calcolate le matrici di probabilità come funzioni dell’energia e del pitch degli elettroni per tutte le diagnostiche hard X-ray installate a JET. I risultati sono presentati col formalismo delle weight-function, che permette di studiare la sensitività del detector a elettroni con energia e pitch differenti. Le matrici riportate mostrano un picco di sensitività per particelle aventi pitch-angolo uguale all’angolo racchiuso tra linea di vista del detector e il campo magnetico.
Development of Nuclear Radiation Based Tomography Methods for Runaway Electrons and Fast Ions in Fusion Plasmas
PANONTIN, ENRICO
2022
Abstract
Fast particles, both electrons and ions, play an important role for the success of the next generation of large tokamak devices, such as ITER, that will prove the feasibility of magnetically confined thermonuclear fusion as an energy source. Ions accelerated by external heating or born in fusion reactions can reach energies in the MeV range. Their primary role is to sustain the plasma temperature and the fusion reaction rate, thus lots of efforts have been put into the development of efficient heating schemes and in the improvement of their confinement. On the other hand, during fast terminations of plasma pulses on tokamaks, electrons can accelerate to relativistic velocities, entering the runaway regime. Runaway electrons have enough energy to seriously damage the plasma facing components of large tokamaks, thus mitigation techniques are under study in view of ITER operations. This thesis focuses on the implementation of deconvolution techniques for the reconstruction of the fast particles distributions from their emission in the MeV energy range. The problem was approached from two different perspectives: the unfolding of the runaway electrons velocity-space distribution from spectroscopic measurements of their bremsstrahlung emission, and the tomographic reconstruction of the density distribution of both fast ions and runaway electrons from the integrated measurement of their emission performed with multiple lines of sight. These algorithms were implemented in an open source Python library. Four deconvolution algorithms were implemented for the unfolding of runaway electrons energy distribution: singular value decomposition, maximum likelihood - expectation maximization, Tikhonov regularization and Poissonian regularization. The transfer matrix necessary for this inversion was calculated using the GENESIS code for estimating the probability of bremsstrahlung emission and the MCNP code for computing the detector response function. The detector response function was calculated for all the hard X-rays diagnostics systems installed at the Joint European Torus and ASDEX Upgrade tokamaks. The performance of the four methods wes then compared over both synthetic and experimental spectra, the latter being measured at ASDEX Upgrade. Maximum likelihood - expectation maximization was found to be the most accurate in the reconstruction of both the runaway electrons energy distribution and their average and maximum energies. The robustness of the four methods against experimental limitations, such as low-energy cut and low statistics, was also investigated. In the path towards the generalization of these unfolding algorithms to the reconstruction of the runaway electrons 2D velocity-space distribution, the transfer matrices in energy and pitch were calculated for all the hard X-ray diagnostics installed at JET. The weight-function formalism was adopted, which allows studying the sensitivity of the detectors to different energy and pitch regions. The matrices showed a sensitivity peak in the pitch axis which is determined by the angle between the line of sight and the magnetic field. Finally, the gamma camera upgrade installed at the Joint European Torus, with its 10 by 9 lines of sight that observe a poloidal section of the tokamak from two perpendicular projections, allows reconstructing the spatial distribution of fast particles. A tomographic algorithm that makes use of smoothing along the magnetic field lines was implemented. This tomography was first applied to recent three-ion radio frequency heating experiments in D-3He mixed plasmas, during which the gamma camera was able to detect the 16.4 MeV γ-rays from 3He(D,γ)5Li reactions. The spatial distribution of the α-particles born in 3He(D,p)4He reactions was reconstructed and the results were used to validate TRANSP simulations. The tomographic algorithm was also applied to the reconstruction of the runaway electrons spatial profiles during plasma disruptions.File | Dimensione | Formato | |
---|---|---|---|
phd_unimib_761719.pdf
accesso aperto
Dimensione
19.03 MB
Formato
Adobe PDF
|
19.03 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/169777
URN:NBN:IT:UNIMIB-169777