Computational & Technology Resources
an online resource for computational,
engineering & technology publications
PROCEEDINGS OF THE NINTH INTERNATIONAL CONFERENCE ON COMPUTATIONAL STRUCTURES TECHNOLOGY
Edited by: B.H.V. Topping and M. Papadrakakis
A Robust Component Mode Synthesis Method for Stochastic Vibroacoustic Problem
Q.H. Tran, M. Ouisse and N. Bouhaddi
FEMTO-ST Institute, UMR 6174, Applied Mechanics Laboratory R. Chaléat, Besançon, France
Q.H. Tran, M. Ouisse, N. Bouhaddi, "A Robust Component Mode Synthesis Method for Stochastic Vibroacoustic Problem", in B.H.V. Topping, M. Papadrakakis, (Editors), "Proceedings of the Ninth International Conference on Computational Structures Technology", Civil-Comp Press, Stirlingshire, UK, Paper 86, 2008. doi:10.4203/ccp.88.86
Keywords: component mode synthesis, vibro-acoustics, uncertainties, viscoelastic damping, poro-elastic damping.
In the vibro-acoustic domain, industrial problems lead to large finite element models which is time-consuming in the updating or optimization processes. Model reduction is mainly based on the component mode synthesis (CMS) technique.
In this paper, first, an overview of the vibro-acoustic problem allows one to choose the formulation. Then, an alternative CMS method using (u,p) formulation is proposed and compared to the classical CMS method. The reduced bases ensure the accuracy of responses of complex vibro-acoustic problems including visco-elastic and poro-elastic materials and they can be easily extended to Craig-Bampton CMS in which the treatment is considered by zones. The simulation of an acoustic cavity coupled with a flexible plate shows the efficiency of the proposed CMS method.
Next, based on those efficient basis the construction of robust Ritz vectors is proposed, defining the so-called robust CMS method, to solve stochastic models. This robust reduced basis answer the question of prohibitive numerical cost in the optimization and the propagation of uncertainties. In practice, this method uses the parametric approach. The dispersion of random variables can be considered as a modification of the deterministic model and can be approximatively represented by a set of random response vectors. These random residual responses are used for enriching the deterministic bases to obtain the robust bases.
To validate the proposed method, an exhaust line is simulated. Uncertainties are introduced in the parameters of visco- and poro-elastic patches. The performance of the proposed robust bases is compared to the Monte Carlo simulation, illustrating the drastic reduction in calculation time, with a efficient prediction level.
purchase the full-text of this paper (price £20)