ICCM Conferences, The 12th International Conference on Computational Methods (ICCM2021)

Font Size: 
Performance evaluation of centroidal Voronoi tessellation for initial particle distribution of particle methods
Masao Ogino, Yutaka Ogisu

Last modified: 2021-06-10

Abstract


The particle methods such as the SPH method and the MPS method have been widely used for the analysis of free-surface flows. This study focuses on the initial distribution of particles for the particle method. The initial particle distribution based on the regular Cartesian grid is generally adopted, however, it is difficult to represent slopes and curved surfaces of the wall boundary and then has negative effects on the precision of the numerical calculation.

To solve this issue, this study proposed a new regular particle distribution determined by the centroidal Voronoi tessellation (CVT), which is a special type of Voronoi diagram. By the famous Gersho's conjecture, Voronoi cells of the CVT asymptotically agree to regular hexagon grid in the two dimensional case. Therefore, proposed method uses the site of the Voronoi region as the particle and then is expected to space out particles evenly.

In this study, computational performances of CVT are demonstrated in cases of parallel computation and large-scale computation.

Keywords


numerical methods; machine learning

An account with this site is required in order to view papers. Click here to create an account.