Finite element structural updating using frequency response functions

In this paper, a sensitivity-based updating method is proposed to reduce the discrepancies between the finite element (F.E.) model representation of a structure and its experimental counterpart thus allowing engineers the use of such numerical representation for structural modification prediction, dynamic model synthesis, aeroelastic analysis, and system qualification. The proposed method iteratively minimizes a residual vector of correlation functions, defined on the Frequency Response Functions (FRF), in order to find the missmodeled regions of the F.E. model through the identification of the unknown vector of the updating parameters. The approach provides an enhanced formulation of the Bayesian-based least-square solution technique and a proper numerical relaxation method allowing thus the updated F.E. model to go beyond the dynamic equivalence with the experimental findings, that is the increase in the correlation between the corresponding modal parameters, by identifying updating parameters that well represent the physical properties of the structure under investigation. A formulation of the weighting matrices that takes into account the uncertainty of the experimental data and the use of the L-curve criterion to numerically relax the solution are presented in the paper. Results from both numerical analyses and experimental investigations are reported to validate the proposed approach and show its robustness.