Particle methods in Computational Fluid Dynamics

Published in Encyclopedia of Computational Mechanics. Second Edition. Edited by E. Stein, R. de Borst and T. J. R. Hughes. Chichester, UK: John Wiley & Sons, Ltd, pp. 1-41, 2018
DOI: 10.1002/9781119176817.ecm2113

Abstract

Particle methods in Computational Fluid Dynamics (CFD) are numerical tools for the solution of the equations of fluid dynamics obtained by replacing the fluid continuum with a finite set of particles. For mathematicians, particles are just points from which properties of the uid can be interpolated. For physicists the particles are material points, which can be treated like any other particle system. Either way, particle methods have a number of attractive features. One of the key attributes is that pure advection is treated exactly. For example, if the particles are given a determined colour and the velocity is specified, the transport of colours by the particle system is exact. The convection of properties also eases the solution of multi material problems, simplifying the detection of interfaces. The use of particles also allows to bridge the gap between the continuum and fragmentation in a natural way, for example in fracture or droplets problems. Since the computation domain, the particles, matches exactly the material domain of interest, the computational resources are optimized with the corresponding reduction in storage and calculation time compared to other methods. Finally, because of the close similarity between particle methods and the physics of the problems to be solved, it is often possible to account for complex physics more easily than with other methods.

Full document