A posteriori constitutive relation error estimators for nonlinear finite element analysis and adaptive control - Archive ouverte HAL Accéder directement au contenu
Chapitre D'ouvrage Année : 2005

A posteriori constitutive relation error estimators for nonlinear finite element analysis and adaptive control

Nicolas Moës

Résumé

Today quite a lot of a posteriori error estimators are available to control finite element calculations. We describe here error estimators based on constitutive relation residuals which have been developed for 20 years, in particular at Cachan. This approach has a strong physical meaning and is quite general. Different errors on constitutive relation can be easily introduced to measure the quality of finite element computations of plastic and viscoplastic structures whose behavior is described by internal variables. These measures take into account, over the studied time interval, all the classical sources of error involved in the computation: the space discretization (the mesh), the time discretization and the iterative technique used to solve the nonlinear discrete problem. To quantify more specifically each source of error, we introduce quantities called indicators. Numerical experiments show that the indicators are linked to the error in a limit sense. The errors and the indicators may then used to simultaneously adapt the mesh, the time discretization and the stopping criteria to meet a prescribed accuracy. The adaptive control is illustrated on two plane stress problems using the Prandtl-Reuss plastic model and its viscoplastic version.
Fichier non déposé

Dates et versions

hal-01007464 , version 1 (16-06-2014)

Identifiants

  • HAL Id : hal-01007464 , version 1

Citer

Pierre Ladevèze, Nicolas Moës. A posteriori constitutive relation error estimators for nonlinear finite element analysis and adaptive control. Advances in Adaptive Computational Methods in Mechanics, pp.231-256, 2005, 9780080433271. ⟨hal-01007464⟩
135 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More