Parametric Updating of Finite Element Models by Minimizing Response Residuals at Resonances

Abstract

In this paper a two step Computational Model Updating method is investigated. In the first step, physical mass and stiffness parameters are updated. In the second step, the modal damping parameters as well as physical damping parameters (allowing to update global non-proportional damping) are open for updating. The updating techniques are based on the inverse sensitivity approach and use frequency response data only at the resonance peaks. The optimization process is solved by minimizing an objective function containing two residuals. The first represents the differences between the analytical and the experimental resonance peaks utilizing the fact, that the analytical and the experimental FRF peaks occur at different resonace frequencies. The decision on what test/analysis FRF peak differences to include in the objective function is based on the Frequency Domain Assurance Criterion (FDAC). The resonances must not necessarily be identical with the eigenfrequencies since they are determined by more or less rough estimates of the eigenfrequencies using a mode indicator function.