Sensitivity Analysis of a Fully Kinetic Code for Physical Parameters

Abstract

A sensitivity analysis of a fully kinetic particle code was conducted to investigate the importance of uncertainties associated to physical parameters. A 500 W-class laboratory model magnetic-layer Hall thruster was used as the testbed. The sensitivities of the physical parameters, including thermal accommodation coefficient, anode/wall temperature, Bohm diffusion coefficient, electron injection current, cathode coupling voltage, and background pressure, were quantified one-byone on a conservative possible range. The results suggest the wall erosion prediction is more sensitive to the physical parameters than the thrust or the discharge current. Among the physical parameters, sensitivity to the Bohm diffusion coefficient and parameters related to the neutral flow (i.e., thermal accommodation coefficient and anode/wall temperatures) were dominant. It was hence found that uncertainties in the physical parameters related to the neutral flow had comparable influence on the Bohm diffusion coefficient despite the low attention they attracted.

Effect of auxiliary gas injection on the operation of a Hall current plasma accelerator

“…The two-dimensional (axial and radial) axisymmetric fully kinetic particle-in-cell (full-PIC) code used in this article is the same as the one described in previous publications [17,18]. The full-PIC code relies on an artificial mass ratio model to speed up computations and a semi-implicit method to derive electric field from Poisson's equation permitting stable simulations for larger time steps than electron plasma frequency (ω p ∆t ≫ 1) and larger grid spacing than Debye length (∆x/λ D ≫ 1).…”

“…The computational time-step (∆t = 5 × 10 −10 s) was chosen such that the plasma oscillation and electron cyclotron motion were captured. The grid spacing and macroparticle size were decided according to the parametric study conducted previously for this specific accelerator [18]. A grid spacing of 0.2 mm was sufficient to accurately capture the sheath potential drop, even though it is larger than the Debye length (λ D = ∼0.05 mm), and a macroparticle size of 1 × 10 7 was a good trade-off between the numerical error from so-called discrete particle noise and the computing cost.…”

“…The full-PIC code's prediction of the accelerator performance (16.0 mN, 1.1 A, %29 and 1201 s) matches very well with the measured values with a maximum discrepancy less 3% for the thrust and 10% for the discharge current. Note that the numerical discretization errors as well as the uncertainties in model physical parameters such as TAC and wall temperature may affect prediction accuracy of the fully kinetic codes [18]. Effects of vacuum chamber background pressure [23] and cathode coupling voltage are not included in the simulation, which may account for the small discrepancy between measurements and predictions.…”

Analysis of influencing parameters in ion thruster plume simulation

“…10 Karadag et al used a one-factor-at-a-time sensitivity analysis method to study uncertainty quantification of fully kinetic Hall thruster modeling. 11 However, in previous studies, little work has systematically analyzed the influencing parameters of the ion thruster plume simulation. Based on the work of the predecessors, this paper analyzes parameters that affect the plume simulation of ion thrusters.…”

“…used a one-factor-at-a-time sensitivity analysis method to study uncertainty quantification of fully kinetic Hall thruster modeling. 11…”