AUTOMATIC ADAPTIVE MESH REFINEMENT WITH CONTROLLED ACCURACY IN A MULTISCALE CONTEXT FOR NONLINEAR SOLID MECHANICS

Daria Koliesnikova (1) , Isabelle Ramière (1) , Frédéric Lebon (2)

1 SESC - Service d'Etudes de Simulation du Comportement du combustibles
2 LMA - Laboratoire de Mécanique et d'Acoustique [Marseille]

During irradiation in reactors, a variety of phenomena of different natures are occurring at different scales. It is crucial to understand such phenomena and to be able to simulate their evolution in time with a controlled accuracy. In this context, we propose to develop an efficient numerical tool enabling us to perform accurate nonlinear quasi-static mechanical simulations with evolving local phenomena. This tool is based on adaptive mesh refinement (AMR) techniques. We adopt a fully automatic algorithm governed by user-prescribed tolerances on errors. We show that the multilevel Local Defect Correction (LDC) method is the most efficient AMR approach in terms of computation time to achieve desired precisions. Its key advantage lies in the separate resolution of problems of limited sizes on a hierarchy of meshes evolving in time

Domaines: Matériaux; Analyse numérique [math.NA]; Physique mathématique [math-ph]

Daria Koliesnikova, Isabelle Ramière, Frédéric Lebon. AUTOMATIC ADAPTIVE MESH REFINEMENT WITH CONTROLLED ACCURACY IN A MULTISCALE CONTEXT FOR NONLINEAR SOLID MECHANICS. International Conference Nuclear Energy for New Europe, Sep 2021, BLED, Slovenia. ⟨hal-03448112⟩