A New Particle Generation Method for Arbitrary 2D Geometries in SPH Modeling

Abstract

For smoothed particle hydrodynamics (SPH), homogeneous particle distribution is important to ensure the computational accuracy and stability, but it is hard to achieve this for complex geometries. In this paper, a new particle generation method is developed to generate particles for arbitrary 2D geometries. In the method, the geometry required for generating particles is orthogonally partitioned into a series of sub-domains. Among the resultant sub-domains, the most ones having standard area are directly converted into particles. The others are iteratively meshed into elements with nearly standard area and particles are placed according to these elements. The present method is implemented based on Abaqus. Examples of particle generation are given to compare various particle generation methods. It is found that the present method shows advantages over some existing methods in the approximation of geometric boundary as well as the regularity and homogeneity of particle distribution. Several physical problems are adopted to examine the influence of initial particle distribution on SPH solution. The calculated results show that particle distributions generated by the present method can lead to better accuracy and stability than those created by some existing methods.

Point Cloud Generation for Meshfree Methods: An Overview

“…Across meshfree literature, it is quite common to see meshfree simulations that begin with mesh generation on the computational domain, see for example [24,72,82,121,144,153,160,219]. Then the vertices of the mesh are used as the global point cloud on which simulations are performed.…”

Dealing with the Effect of Air in Fluid Structure Interaction by Coupled SPH-FEM Methods

“…Whereas SPH-FEM coupling is often seen in relevant scientific literature [37,38], different commercial software codes also implement this integrated method. For instance, ABAQUS uses SPH functions to overcome the mesh distortion problem [39], while a wider range of applications can be found as implemented in LS-DYNA [40].…”