Refine
Document Type
- Doctoral Thesis (3)
Language
- English (3) (remove)
Has Fulltext
- yes (3)
Is part of the Bibliography
- no (3) (remove)
Keywords
- Magnetohydrodynamik (3) (remove)
Institute
Achieving commercial production of electricity by magnetic confinement fusion requires improvements in energy and particle confinement. In order to better understand and optimise confinement, numerical simulations of plasma phenomena are useful. One particularly challenging regime is that in which long wavelength MHD phenomena interact with kinetic phenomena. In such a regime, global electromagnetic gyrokinetic simulations are necessary. In this regime, computational requirements have been excessive for Eulerian methods, while Particle-in-Cell (PIC) methods have been particularly badly affected by the "cancellation problem", a numerical problem resulting from the structure of the electromagnetic gyrokinetic equations. A number of researchers have been working on mitigating this problem with some significant successes. Another alternative to mitigating the problem is to move to a hybrid system of fluid and gyrokinetic equations. At the expense of reducing the physical content of the numerical model, particularly electron kinetic physics, it is possible in this way to perform global electromagnetic PIC simulations retaining ion gyrokinetic effects but eliminating the cancellation problem. The focus of this work has been the implementation of two such hybrid models into the gyrokinetic code EUTERPE. The two models treat electrons and the entire bulk plasma respectively as a fluid. Both models are additionally capable of considering the self-consistent interaction of an energetic ion species, described gyrokinetically, with the perturbed fields. These two models have been successfully benchmarked in linear growth rate and frequency against other codes for a Toroidal Alfvén Eigenmode (TAE) case. The m=1 internal kink mode, which is particularly challenging in terms of the fully gyrokinetic cancellation problem, has also been successfully benchmarked using the hybrid models with the MHD eigenvalue code CKA. Non-linear simulations in this TAE case have been performed confirming the analytical prediction of a quadratic relationship between the linear growth rate of the TAE and the saturated amplitude of the TAE for a range of moderate values of the linear growth rate. At higher linear growth rate, a slower scaling of saturated amplitude with linear growth rate is observed. This analysis has been extended to include the non-linear wave-wave coupling between multiple TAE modes. It has been shown that wave-wave coupling results in a significant reduction in the saturated amplitude. It has been demonstrated that both plasma elongation and ion kinetic effects can exert a stabilising influence on the internal kink mode. A population of energetic particles can also exert a stabilising influence at low normalised pressure. At high normalised fast particle pressure the stabilised kink mode has been shown to give way to the m=1 EPM, which has been simulated both linearly and non-linearly (the "fishbone" mode). The first self-consistent simulations of global modes in the magnetic geometry of the optimised stellarator Wendelstein 7-X have been performed both linearly and non-linearly. Limitations have been encountered in performing simulations in 3D geometry. A hypothesis for the cause of these problems is outlined and ideas for mitigation are briefly described. In addition to the hybrid model simulations, some of the first utilisations of a new scheme for mitigating the cancellation problem in the fully gyrokinetic regime have been carried out in the framework of this thesis. This scheme, which was developed separately, is concisely described in this work. The new scheme has been benchmarked with existing gyrokinetic and hybrid results. The linear Wendelstein 7-X simulations and linear and single mode non-linear TAE simulations have been repeated with the new model. It is shown that bulk plasma kinetics can suppress the growth rate of global modes in Wendelstein 7-X. The results of fully gyrokinetic TAE simulations, the first to have been performed to our knowledge, are shown to be in close agreement with those results obtained using hybrid models. In the TAE case, the hybrid models are an order of magnitude less computationally demanding than the new gyrokinetic scheme, which is in turn at least an order of magnitude less computationally demanding than the previous gyrokinetic scheme.
Physics-regularized Machine Learning To Approximate 3D Ideal-MHD Equilibria At Wendelstein 7-X
(2024)
The magnetohydrodynamic (MHD) equilibrium model is one of the fundamental building blocks in the description of a magnetically confined plasma. The computational cost of constructing solutions to the 3D ideal-MHD equilibrium problem is one of the limiting factors in stellarator research and design; in particular, it limits the extent to which we can perform sample-intensive applications, applications which require many samples to be evaluated to yield meaningful results. Sample-intensive applications in stellarator research and design include, for example, equilibrium reconstruction, stellarator optimization, and flight simulators. In this thesis, we investigate how faithfully artificial neural networks (NNs) can quickly approximate ideal-MHD equilibria in stellarator geometries, starting with Wendelstein 7-X (W7-X), the world’s most advanced stellarator. In particular, we investigate (see section 1.7):
RQI: to what extent can NN models approximate the MHD equilibrium solution for different W7-X configurations and plasma profiles? What
is the speed-accuracy trade-off offered by NN models?
RQII: to what degree the NN model faithfully reproduces equilibrium quantities of interest (e. g., MHD stability)? To what extent can NN models meet the requirements of downstream applications (e. g., Bayesian
inference, stellarator optimization) in terms of equilibrium quantities
accuracy?
RQIII: whether we can exploit the implicit representation of a MHD equilibrium, i. e., the equilibrium solution should satisfy the ideal-MHD force
balance equation, to improve the NN approximation’s accuracy;
RQIV: the reconstruction of the full posterior istribution of plasma parameters and equilibrium quantities with self-consistent MHD equilibria; moreover, how does the adoption of MHD equilibria approximated by NN models affect the inferred plasma parameters?
A deep NN model is developed to learn the ideal-MHD solution operator in W7-X operational subspace, yielding 3D equilibria up to six orders of magnitude faster than currently available MHD equilibrium codes. Physics domain knowledge is embeded into the NN model: equilibrium solution symmetries are satisfied by construction, and the MHD force balance regularizes the NN model to satisfy the ideal-MHD equations. The model accurately predicts the equilibrium solution and it faithfully reproduces global equilibrium quantities and proxy functions used in stellarator optimization. Finally, the developed fast NN equilibrium model has been applied in downstream applications to obtain W7-X configurations with improved fast-particle confinement and to infer plasma parameters with self-consistent MHD equilibria at W7-X.
The collisionless tearing mode is investigated by means of the delta-f PIC code EUTERPE solving the gyrokinetic equation. In this thesis the first simulations of electromagnetic non-ideal MHD modes in a slab geometry with EUTERPE are presented. Linear simulations are carried out in the cases of vanishing and finite temperature gradients. Both cases are benchmarked using a shooting method showing that EUTERPE simulates the linearly unstable tearing mode to a very high accuracy. In the case of finite diamagnetic effects and values of the linear stability parameter Delta of order unity analytic predictions of the linear dispersion relation are compared with simulation results. The comparison validates the analytic results in this parameter range. Nonlinear single-mode simulations are performed in the small- to medium-Delta range measuring the dependency of the saturated island half width on the equilibrium current width. The results are compared with an analytic prediction obtained with a kinetic electromagnetic model. In this thesis the first simulation results in the regime of fast nonlinear reconnection~(medium- to high-Delta range) are presented using the standard gyrokinetic equation. In this regime a nonlinear critical threshold has been found dividing the saturated mode from the super-exponential phase for medium-Delta values. This critical threshold has been proven to occur in two slab equilibria frequently used for reconnection scenarios. Either changing the width of the equilibrium current or the wave number of the most unstable mode makes the threshold apparent. Extensive parameter studies including the variation of the domain extensions as well as the equilibrium current width are dedicated to a comprehensive overview of the critical threshold in a wide range of parameters. Additionally, a second critical threshold for high-Delta equilibria has been observed. A detailed comparison between a compressible gyrofluid code and EUTERPE is carried out. The two models are compared with each other in the linear regime by measuring growth rates over wave numbers of the most unstable mode for two setups of parameters. Analytical scaling predictions of the dispersion relation relevant to the low-Delta regime are discussed. Employing nonlinear simulations of both codes the saturated island half width and oscillation frequency of the magnetic islands are compared in the small-Delta range. Both models agree very well in the limit of marginal instability and differ slightly with decreasing wave vector. Recently, the full polarisation response in the quasi-neutrality equation was implemented in EUTERPE using the Padé approximation of the full gyrokinetic polarisation term. Linear simulation results including finite ratios of ion to electron temperature are benchmarked with the dispersion relation obtained from a hybrid model. Finite temperature effects influence the saturated island width slightly with increasing ion to electron temperature ratio which has been verified by both models.