2026-09-19T16:54:47Z https://www.ideals.illinois.edu/oai-pmh

oai:www.ideals.illinois.edu:2142/120312 2023-09-05 col_2142_5131 col_2142_14837 com_2142_5130 com_2142_14836 com_2142_234

Curreli, Davide Kozlowski, Tomasz 2023-05 application/pdf en text Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2023-09-01 without embargo terms The student, Mohammad Mustafa, accepted the attached license on 2023-04-21 at 14:21. The student, Mohammad Mustafa, submitted this Thesis for approval on 2023-04-21 at 16:32. This Thesis was approved for publication on 2023-04-24 at 14:21. DSpace SAF Submission Ingestion Package generated from Vireo submission #19098 on 2023-09-01 at 17:09:20 Development of reduced-order models of the ion impact distribution function in magnetized plasma sheaths Mustafa, Mohammad Abdul-Jalil Mohammad 2023-04-24 Plasma Physics Plasma Sheath Plasma-surface Interactions Reduced-order Model Sensitivity Analysis Machine Learning Uncertainty Quantification Surrogate Modeling Ion Energy-angle Distributions In magnetic-confinement fusion devices, high-fidelity models of the energy-angle distribution of the ions impacting on material walls are crucial for characterizing ion-surface interactions and impurity release. Typically, the Ion Energy-Angle Distributions (IEADs) are simulated using plasma kinetic models (e.g. Particle-In-Cell PIC codes, such as hPIC), which are usually computationally intensive. In this work, we constructed an effective surrogate model for the IEADs by means of a data-driven strategy in high-dimensional parameter space. The surrogate model considers up to four input parameters of relevance to the problem: (1) electron-to-ion temperature ratio, (2) magnetic field inclination, (3) magnetic field strength, and (4) plasma density. The hPIC2 code was utilized to generate necessary training and testing data sets. A sparse grid was employed to reduce the cost of the surrogate model construction without compromising accuracy. Least square approximation using data from a denser sparse grid was utilized to mitigate the effect of particle noise. Fitting of IEAD was performed in a transformed coordinate system of the distribution. Additionally, an artificial neural network strategy based on zero-inflated models was employed to construct a surrogate model of plasma sheath potentials. The surrogate models constructed as part of this work provide computationally efficient tools to emulate the output of hPIC with limited errors. A variance-based sensitivity analysis using samples drawn from the surrogate models was performed. The sensitivity analysis of plasma potentials showed that the floating wall potential is solely affected by the electron-to-ion temperature ratio, whereas the potential drop across the magnetic presheath and Debye sheath were significantly affected by the magnetic field inclination angle. The sensitivity analysis of the IEAD moments also showed a strong dependency of ions' impact energy on the electron-to-ion temperature ratio with insignificant dependence on the other physical parameters. On the other hand, the analysis revealed significant dependencies of ions' angle of impact on electron-to-ion temperature ratio and magnetic field inclination angle, with a much lower effect of the remaining parameters. Thesis eng https://hdl.handle.net/2142/120312 Copyright 2023 Mohammad Mustafa M.S. Thesis Nuclear, Plasma, Radiolgc Engr University of Illinois at Urbana-Champaign Nuclear, Plasma, & Rad Engr