Dynamic Particle Weight Remapping in Hybrid PIC Hall-effect Thruster Simulation
Abstract
Control of macroparticle growth and distribution and statistical noise are key challenges for particle kinetic models such as particle-in-cell (PIC). For hybrid fluid-PIC models such as those commonly used in Hall-effect thruster (HET) simulation, the statistical noise adds an additional challenge due to the stiffness imposed by the use of gradients of the mobility tensor in the solution of Ohm's law. Past results indicated the direct hybrid PIC is approach is essentially intractable in more than 1D without advanced altering for noise reduction. Though increasing the ratio of computationally represented macro particles to real physical particles is one means of variance reduction used to improve the fidelity of particle-based simulations, the high dynamic range necessary for problems involving chain-branching reactions such as ionization tend to require balancing regions of high statistical noise with regions of extreme computational cost. In this work, we use a recently developed method of particle weight remapping that strictly conserves mass, momentum, and energy while simultaneously unlike most prior standard particle merging techniques to adapt macroparticle weights within a simple pseudo-spectral HET simulation. The technique also avoids thermalization by remaining faithful to the original velocity distribution function through the use of octree binning in velocity space. Though demonstrating effective control over macroparticle density in an axial-azimuthal hybrid PIC HET simulation with little impact on solution quality, the method was found to be unnecessary because ions exit the small computational domain too rapidly to impose severe computational challenges in the baseline implementation. However, the method remains a viable candidate to enhance solution tractability as problem complexity increases.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 2015
- Accession Number
- ADA627145
Entities
People
- David Bilyeu
- Justin Koo
- Robert R. S. Martin
Organizations
- Air Force Research Laboratory