Comparison between the modal idientification method and the POD-Galerkin method for model reduction in nonlienar diffusive systems

Olivier Balima (1) , Yann Favennec (1) , Manuel Girault (1) , Daniel Petit (1)

1 LET - Laboratoire d'études thermiques

This paper presents a comparison between the modal identification method (MIM) and the proper orthogonal decomposition-Galerkin (POD-G) method for model reduction. An example of application on a nonlinear diffusive system is used to illustrate the study. The study shows that in both methods, the state formulation of the nonlinear diffusive equation may be similar. However, the ideas behind both methods are completely different. The considered example shows that, for both methods, reducing the order up to 99,5 % gives enough accuracy to simulate the dynamic of the original system. It is also seen in this example that the reduced model given through the MIM are slightly faster and more accurate than the ones given through the POD-G method.

Mots clés: nonlinear model reduction; proper orthogonal decomposition; Galerkin formulation; modal identification method; optimization; adjoint methods

Domaines: Thermique [physics.class-ph]

Olivier Balima, Yann Favennec, Manuel Girault, Daniel Petit. Comparison between the modal idientification method and the POD-Galerkin method for model reduction in nonlienar diffusive systems. International Journal for Numerical Methods in Engineering, 2006, 67, pp.895-915. ⟨10.1002/nme.1620⟩. ⟨hal-00095910⟩