1. A Level Set-Discrete Element Method in YADE for numerical, micro-scale, geomechanics with refined grain shapes
- Author
-
Jérôme Duriez, Cédric Galusinski, Risques, Ecosystèmes, Vulnérabilité, Environnement, Résilience (RECOVER), Institut national de recherche en sciences et technologies pour l'environnement et l'agriculture (IRSTEA)-Aix Marseille Université (AMU), Aix Marseille Université (AMU)-Institut National de Recherche pour l’Agriculture, l’Alimentation et l’Environnement (INRAE), Institut de Mathématiques de Toulon - EA 2134 (IMATH), Université de Toulon (UTLN), and French Sud region to575 the LS-ENROC project
- Subjects
Level set (data structures) ,Level Set-DEM (LS-DEM) ,Computer science ,0211 other engineering and technologies ,020207 software engineering ,Signed distance function ,Fast Marching Method (FMM) ,02 engineering and technology ,[SPI.MECA]Engineering Sciences [physics]/Mechanics [physics.med-ph] ,Discrete element method ,Computational science ,Set (abstract data type) ,Polyhedron ,Discrete Element Method (DEM) ,particle’s shape ,0202 electrical engineering, electronic engineering, information engineering ,Particle ,ZABR ,Computers in Earth Sciences ,Distance transform ,Fast marching method ,ComputingMilieux_MISCELLANEOUS ,021101 geological & geomatics engineering ,Information Systems - Abstract
International audience; A C++-Python package is proposed for 3D mechanical simulations of granular geomaterials, seen as a collection of particles being in contact interaction one with another while showing complex grain shapes. Following the socalled Level Set-Discrete Element Method (LS-DEM), the simulation workflow stems from a discrete field for the signed distance function to every particle, with its zero-level set corresponding to a particle’s surface. A Fast Marching Method is proposed to construct such a distance field for a wide class of surfaces. In connection with dedicated contact algorithms and Paraview visualization procedures, this shape description eventually extends the YADE platform for discrete simulations. Its versatility is illustrated on superquadric particles i.e. superellipsoids. On computational aspects, memoryrequirements possibly exceed one megabyte (MB) per particle when using a double numeric precision, and time costs, though also significant, appear to be lighter than the use of convex polyhedra and can be drastically reducedusing a simple, OpenMP, parallel execution.
- Published
- 2021