Optimisation de structures par couplage métamodèles multi-fidélité et modèles réduits

Abstract : Computation time is the main limitation to use optimisation toolbox on mechanical models, especially in the nonlinear case. The use of multi-fidelity metamodels fed by numerical simulations of variable precision makes it possible to reduce time computation in areas which are far from the optimum. Hence to use as the best these metamodels, it is necessary to have an efficient solver, allowing to generate simulations with a given level of precision. The idea developed here is to use PGD models reduction technique. They product naturally solutions whose precision levels varies according to the number of modes used. In the nonlinear case, the PGD used here is a "space-time" decomposition coupled with the LATIN method for the treatment of material nonlinearities.
Liste complète des métadonnées

Cited literature [15 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01672813
Contributor : Stéphane Nachar <>
Submitted on : Wednesday, December 27, 2017 - 12:28:40 PM
Last modification on : Saturday, March 23, 2019 - 1:29:25 AM
Document(s) archivé(s) le : Wednesday, March 28, 2018 - 12:16:40 PM

Files

PapierCFM.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01672813, version 1

Citation

Stéphane Nachar, Pierre-Alain Boucard, David Néron. Optimisation de structures par couplage métamodèles multi-fidélité et modèles réduits. 23ème Congrès Français de Mécanique [CFM2017], Aug 2017, Lille, France. ⟨hal-01672813⟩

Share

Metrics

Record views

185

Files downloads

164