Smoothed Particle Hydrodynamics vs Lattice Boltzmann for the solution of steady and unsteady fluid flows

Angelantonio Tafuni, Maria Grazia De Giorgi, Alessandro De Rosis

Research output: Contribution to journalArticlepeer-review

Abstract

Numerical simulations of steady and unsteady viscous flows are presented by adopting two different numerical methodologies: the Smoothed Particle Hydrodynamics formulation implemented in the open-source code DualSPHysics and an in-house lattice Boltzmann code based on a concise central-moments scheme. Both methods employ a weakly compressible assumption to simulate incompressible flow, which means the fluid is assumed barotropic and the density and pressure are related through an equation of state. The accuracy of the two approaches is evaluated against well-defined and consolidated benchmark tests. Advantages and disadvantages of the two methodologies are discussed and substantiated by quantitative comparisons that focus on accuracy and efficacy of the two methodologies against other well-established computational methods. Overall, both formulations proposed herein are able to capture the relevant flow physics with a good level of accuracy when compared to other more affirmed techniques. Remarkably, this is observed in spite of the proposed two methods lacking key strategies commonly used in grid-based methods, such as adaptive mesh refinement.

Original languageEnglish (US)
JournalComputational Particle Mechanics
DOIs
StateAccepted/In press - 2021

All Science Journal Classification (ASJC) codes

  • Computational Mechanics
  • Civil and Structural Engineering
  • Numerical Analysis
  • Modeling and Simulation
  • Fluid Flow and Transfer Processes
  • Computational Mathematics

Keywords

  • CFD simulation
  • DualSPHysics
  • Lattice Boltzmann
  • Smoothed Particle Hydrodynamics
  • Viscous flow

Fingerprint

Dive into the research topics of 'Smoothed Particle Hydrodynamics vs Lattice Boltzmann for the solution of steady and unsteady fluid flows'. Together they form a unique fingerprint.

Cite this