FREQUENCY–DOMAIN CRITERIA FOR CORRELATING AND UPDATING DYNAMIC FINITE ELEMENT MODELS

This paper discusses the use of frequency–domain criteria for finite element (FE) model correlation and updating. The criteria, which can be local or global, are either shape or amplitude based. The former are sensitive to mode shape differences but not to relative scales, and the latter depend on the actual response amplitudes. The updating objective function is formulated in such a way that two global correlation criteria are satisfied simultaneously. A sensitivity-based updating problem is expressed as a set of linear equations where the coefficient matrix contains the partial derivatives of the correlation functions with respect to the chosen design variables. The formulation has the advantage of being able to deal with incomplete measurements and there is no critical selection of updating frequency points. The correlation criteria were used successfully to quantify the initial closeness between 4 increasingly detailed FE models of an automotive oil pan. The updating of the second most detailed model, using simulated response functions from the most detailed model, exhibited quick convergence properties and yielded a very good match between the target and updated frequency response functions. However, the initial FE model was corrected in a curve-fitting/minimisation sense since actual modelling deficiencies were compensated by adjusting a small number of pre-selected design parameters.