Optimization of buildable stellarator coils to confine electron-positron plasmas and advance fusion reactor design

Gil, Pedro (2026): Optimization of buildable stellarator coils to confine electron-positron plasmas and advance fusion reactor design. Dissertation, LMU München: Faculty of Physics

Abstract

The stellarator is considered a front-runner for a future fusion power plant. Unlike axisymmetric magnetic confinement fusion devices like tokamaks, stellarators ely on complex, numerically optimized three-dimensional magnetic fields to confine the plasma, enabling stable, steady-state operation. The stellarator concept, therefore, does not require pulsed induced currents in the plasma for confinement; it instead delegates the magnetic field generation to external magnetic coils. This should help to avoid disruption events and large magnetohydrodynamic instabilities, which can present significant obstacles to achieving commercial nuclear fusion. In addition to being useful for a fusion reactor, a stellarator can also serve to confine electron-positron plasmas. However, the shapes of stellarator coils can become extraordi-narily complex and still require very tight manufacturing and assembly tolerances. Finding cost-effective, buildable, and accurate coil sets for a given stellarator is cur-rently a bottleneck to the viability of stellarator designs. This results in a scarcity of stellarator experiments worldwide. This thesis focuses on advancing stellarator optimization, the first fundamental step in shaping magnetic coils and plasma. First, we propose a new method for designing stellarators with increased toler-ances. Deviations originating in the manufacturing and assembly of the magnetic coils infamously led to the cancellation of the National Compact Stellarator eX-periment project and imposed strict 2mm tolerances on the construction of the 3m-diameter magnets for the Wendelstein 7-X stellarator. The requirement to accurately build magnetic coils to exactly reproduce finely tailored magnetic fields renders the construction task particularly demanding for current production technologies. Therefore, a novel method combining recently established techniques uch as stochastic coil optimization and single-stage optimization was developed as part of this dissertation and presented here. This results in a numerical framework that simultaneously optimizes the coils and plasma shape, balancing engineering and physics constraints and relaxing the requirements on coil tolerances. This is done by taking a well-defined cloud of perturbed coils, computing their field accuracies, and optimizing them together with the plasma equilibrium. Next, this method was integrated in an optimization scheme to design the to-be-built tabletop-sized stellarator experiment: Electrons and Positrons in an Optimized Stellarator (EPOS). This experiment aims to confine pair plasmas in steady-state. This sets strict requirements regarding size, tolerances, and magnetic field strength. Unlike its fusion energy siblings, EPOS is required to have a small volume to reach high enough densities and high magnetic field strength (∼2T) to cool down the particles via cyclotron cooling. This should result in the observation of collective effects. It will be a first-of-a-kind device, as pair plasmas have not et been confined. Balancing all the physics and engineering requirements, therefore, necessitates a dedicated optimization framework. The latter is presented in this work, together with an analysis of possible candidates that were produced with this method. A final configuration of EPOS is proposed which meets all the design requirements. Finally, a new approach to optimizing stellarator coils is presented in this work. Traditional optimization methods struggle to balance multiple physics and engineering constraints, requiring significant computing resources and manual input tuning, yet still yielding suboptimal or infeasible configurations. We present a new optimization method based on an augmented Lagrangian scheme and show that it efficiently generates stellarator coils that exhibit enhanced physics performance and meet essential design requirements. This has a high potential to accelerate progress towards commercially viable stellarator fusion reactors.

• Item Type: Theses (Dissertation, LMU Munich)
• Keywords: fusion, positron, electron, plasma, stellarator, magnetic, coil
• Subjects: 500 Natural sciences and mathematics; 500 Natural sciences and mathematics > 530 Physics
• Faculties: Faculty of Physics
• Language: English
• Date of oral examination: 19. May 2026
• 1. Referee: Zohm, Hartmut
• Signature of the printed copy: 0001/UMC 32064