Títol:
|
Domain decomposition and parallel direct solvers as an adaptive multiscale strategy for damage simulation in materials
|
Autor/a:
|
Everdij, Frank P. X.; Lloberas Valls, Oriol; Simone, Angelo; Rixen, Daniel J.; Sluys, Lambertus J.
|
Altres autors:
|
Universitat Politècnica de Catalunya. Departament d'Enginyeria Civil i Ambiental |
Abstract:
|
Understanding physical phenomena of heterogeneous materials is an ongoing active research field in the structural design of buildings and roads. The failure analysis of quasi-brittle materials such as concrete is a particular topic of interest in civil engineering. This process is characterized by the initial formation of cracks on a microscopic length scale which coagulate into macroscale cracks leading up to weakening and fracture. Because the fracturing process of these materials occurs on several different length scales, care must be taken to provide an accurate description which accounts for all the relevant mechanical processes while maintaining a reasonable computation cost. With
this reasoning in mind, we use a multiscale approach in our numerical simulation, switching between different meshes and material parameters depending on the local mechanical behaviour. In this contribution, we will present a non-linear finite element computation
involving a non-local damage model of a wedge-split test used for
evaluating fracture mechanics in concrete-like materials. We will apply the classical FETI framework (Farhat and Roux [1991]) to the non-linear gradient enhanced damage (GD) model (Peerlings et al. [1996]) using both iterative and direct solvers to the interface problem as well as using a direct solver for the entire set of equations of the fully dual assembled system. |
Abstract:
|
Peer Reviewed |
Matèries:
|
-Àrees temàtiques de la UPC::Enginyeria dels materials::Assaig de materials -Brittleness--Mathematical models -Fragilitat -- Models matemàtics |
Drets:
|
|
Tipus de document:
|
Article - Versió publicada Objecte de conferència |
Publicat per:
|
Springer
|
Compartir:
|
|