3-D simulation of ion thruster plumes using octree adaptive mesh refinement

Burak Korkut, Zheng Li, Deborah A. Levin

Research output: Contribution to journalArticlepeer-review


The physics of neutral and ion species in plume expansions to vacuum is challenging due to the wide variation in length and time scales. To address the multiscale nature of these flows, a grid approach known as adaptive mesh refinement (AMR) combined with the octree method has been used with direct simulation Monte Carlo (DSMC) and particle-in-cell solvers to model the plumes of small ion thruster expansions to vacuum. Single- and triple-thruster geometries have been simulated with the inclusion of the well-known charge exchange (CEX) reactions that transfer energy from energetic xenon ions to neutrals. A multispecies timestep, species weighting factors, and local timestep adaptation are employed to simulate the interactions between major and trace, ion species. Three grids for modeling particle collisions, interactions with the electric field, and visualization have been developed in a parallel code that has good scalability. Comparison studies are presented between a two-level Cartesian-based DSMC solver and our new code in terms of accuracy of results and simulation cost. CEX collisions are shown to produce back flow in both single- and triple-thruster configurations and increase the ion number density up to 10% near the thruster exit. For the 3-D, triple-thruster case, the combined plume interaction region was automatically captured by the AMR-octree mesh.

Original languageEnglish (US)
Pages (from-to)1706-1721
Number of pages16
JournalIEEE Transactions on Plasma Science
Issue number5
StatePublished - May 1 2015


  • Adaptive mesh refinement (AMR)
  • direct simulation Monte Carlo (DSMC)
  • electric propulsion
  • octree
  • parallel computing
  • plasma.

ASJC Scopus subject areas

  • Nuclear and High Energy Physics
  • Condensed Matter Physics


Dive into the research topics of '3-D simulation of ion thruster plumes using octree adaptive mesh refinement'. Together they form a unique fingerprint.

Cite this