Multi-Objective Design Optimization of Cusped Field Thruster via Surrogate-Assisted Evolutionary Algorithms

Research output: Contribution to journalArticlepeer-review


Cusped field thrusters are an efficient cost-effective electric propulsion technology that offer promise for small spacecraft platforms. They offer various advantages, including simple design and low wall erosion, and hence long-life operation owing to high plasma confinement using permanent magnets over other electrostatic alternatives. Accurate physical modeling and performance characterization are crucial to achieve successful downscaling of cusped field thrusters while maintaining high performance. Multiobjective design optimization is performed by incorporating magnetic simulation coupled with an improved power balance model into evolutionary algorithms assisted by surrogate modeling, aiming to simultaneously maximize thrust, total efficiency, and specific impulse. Physical insights have been gained into key design factors to maximize thruster performance by probing into the plasma behavior inside the channel and the plume region. It has been found that altering the magnetic field distribution using an additional magnet can effectively enhance thruster performance by suppressing the plume divergence loss. Global sensitivity analysis has identified the anode current as the most influential design parameter on the performance parameters due to the active role it plays in the ionization process. Uncertainty analysis has found the considerable influence of the variations of anode potential and mass flow rate on those of thrust and specific impulse.

Original languageEnglish
Pages (from-to)973-988
Number of pages16
JournalJournal of Propulsion and Power
Issue number6
Publication statusPublished - Nov 2022

All Science Journal Classification (ASJC) codes

  • Aerospace Engineering
  • Fuel Technology
  • Mechanical Engineering
  • Space and Planetary Science


Dive into the research topics of 'Multi-Objective Design Optimization of Cusped Field Thruster via Surrogate-Assisted Evolutionary Algorithms'. Together they form a unique fingerprint.

Cite this