New approaches are developed that use measured data to adjust the analytica
l mass and stiffness matrices of a system so that the agreement between the
analytical modes of vibration and the modal survey is improved. By adding
known masses to the structure of interest, measuring the modes of vibration
of this mass-modified system, and finally using this set of new data in co
njunction with the initial modal survey, the analytical mass matrix of the
structure can be corrected, after which the analytical stiffness matrix can
be readily updated. By manipulating the correction matrices into vector fo
rms, the connectivity information can be enforced, thereby preserving the p
hysical configuration of the system and reducing the sizes of the least-squ
ares problems that need to be solved. Solution techniques for updating the
system matrices are introduced, and the numerical issues associated with so
lving overdetermined and underdetermined least squares problems are investi
gated. The effects of round-off errors are also studied, and heuristic crit
eria are given for determining the minimum number of modes that need to be
measured in order to ensure sufficiently accurate updated mass and stiffnes
s matrices. Numerical experiments are presented to validate the proposed mo
del-updating techniques, to illustrate the effects of the number of measure
d modes on the quality of the updated model, to show how the magnitudes and
locations of the added masses influence the updated matrices, and to highl
ight the numerical issues discussed in this paper. Copyright (C) 2001 John
Wiley & Sons, Ltd.