Uncertainty quantification in mid-frequency range simulations using the statistical modal energy distribution analysis - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2017

Uncertainty quantification in mid-frequency range simulations using the statistical modal energy distribution analysis

Résumé

The Statistical Modal Energy Analysis (SmEdA) is a variant of the Statistical Energy Analysis (SEA) developed to predict the high frequency behaviour of structures by dividing them into subsystems without requiring a modal energy equipartition. The method is based on the modal bases of uncoupled subsystems, and coupling loss factors are derived from Finite Element Analysis. Uncertainty Quantification can thus be applied in such a configuration at either the subsystem level, with respect to the physical input parameters (eg material properties and dimensions), or at the coupled model level with respect to the coupling factors or the modal data used to compute them. For UQ to be physically meaningful, it is necessary that uncertainty modeling at the coupled model level be representative of uncertainty at the subsystem level. A strategy based on sampling at the coupled model level using a covariance matrix computed at the subsystem level is proposed here. The methodology is formulated and applied to a four-subsystem structure. The UQ performed at the two levels is shown to be coherent but with reduced computational costs at the coupled model level allowing a higher number of UQ simulations.
Fichier non déposé

Dates et versions

hal-02131350 , version 1 (16-05-2019)

Identifiants

  • HAL Id : hal-02131350 , version 1

Citer

Emeline Sadoulet, Kendra Buren, Morvan Ouisse, Scott Cogan, Laurent Maxit. Uncertainty quantification in mid-frequency range simulations using the statistical modal energy distribution analysis. Euro-Mediterranean Conference on Structural Dynamics and Vibroacoustics, Apr 2017, Sevilla, Spain. ⟨hal-02131350⟩
24 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More