Can model updating tell the truth

This paper discusses to which extent updating methods may be able to correct finite element models in such a way that the test structure is better simulated. After having unified some of the most popular modal residues used as the basis for optimization algorithms, the relationship between modal residues and model correlation is investigated. This theoretical approach leads to an error estimator that may be implemented to provide an a priori upper bound of a model`s predictive quality relative to test data. These estimates however assume that a full measurement set is available. Finally, an application example is presented that illustrates the effectiveness of the estimator proposed when less measurement points than degrees of freedom are available.