Skip to main navigation Skip to search Skip to main content

Level-set based pre-processing techniques for particle methods

  • Technical University of Munich
  • Xi’an Research Institute of High Technology

Research output: Contribution to journalArticlepeer-review

26 Scopus citations

Abstract

Obtaining high-quality particle distribution representing clean geometry in pre-processing is essential for accurate and stable simulation with particle methods. In this paper, several level-set-based techniques for cleaning up ‘dirty’ geometry automatically and generating isotropic particle distribution are presented. First, under a given resolution, an identification method for non-resolved structures based on a level-set field is employed to detect the tiny fragments which dirty the geometry. Second, a re-distance algorithm is proposed to remove these fragments and reconstruct clean, smooth geometries. Third, a ‘static confinement’ boundary condition is developed for particle relaxation. By complementing the kernel support for the near-surface particles, the boundary condition achieves improved body-fitted particle distribution near the highly-curved or narrow region. Several numerical examples are given to demonstrate the efficient cleanup capabilities of the present method as well as the improvement of body-fitted particle distribution for complex geometries. In addition, numerical simulations have been carried out for the fluid-structure interaction (FSI) of an elastic airfoil NACA6412 at various resolutions to show that, when the unresolved structures affect or even fail the simulation, the cleaned geometry and improved particle distribution help to stabilize and smooth the simulations.

Original languageEnglish
Article number108744
JournalComputer Physics Communications
Volume289
DOIs
StatePublished - Aug 2023

Keywords

  • Kernel support completing
  • Level-set
  • Particle methods
  • Static confinement
  • ‘Dirty’ geometry cleaning

Fingerprint

Dive into the research topics of 'Level-set based pre-processing techniques for particle methods'. Together they form a unique fingerprint.

Cite this