In FE model updating, as in any identification procedure, we select so
me parameters in the model, and try to fine-tune them to minimize the
discrepancy between the model predictions and the measured data. The p
aper compares the performance of the generic element matrices, recentl
y introduced by the authors, with other selection strategies for finit
e element model updating. The updated models obtained from these metho
ds are compared with the measured data and rated according to their ab
ility to produce the measured data within and beyond the frequency ran
ge used in the updating, and more importantly, according to their abil
ity to predict the effect of design changes.