Uncertainty quantification for viscoelastic composite structures in computational linear structural dynamics
Résumé
This paper deals with the analysis of the propagation of uncertainties in computational linear dynamics for strongly dissipative 3D linear viscoelastic composite structures in the presence of parameter uncertainties as well as model uncertainties. The approach used for modeling uncertainties is the probabilistic nonparametric approach, which consists in replacing the matrices of the nominal reduced-order model with random matrices following a probability distribution obtained using information theory. Special care is taken regarding the stochastic modeling of the random reduced stiffness and damping matrices. These two matrices are statistically dependent through a set of compatibility equations, implied by the causality of the system. This set of equations involves Hilbert transforms of the frequency-dependent part of the two matrices and are used in order to generate statistically independent realizations of each random matrix which satisfy the causality principle for the Monte Carlo stochastic solver.
Fichier principal
conference-2016-USD-Leuven-sept-19-21-capillon-descelier-soize-preprint.pdf (333.86 Ko)
Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...