Experimental Modal Analysis
Dynamics and Vibration Group, Department of Mechanical and Manufacturing Engineering, University of Melbourne
staff: Dr J.M.Krodkiewski, Dr T.J.Chalko

MODAL ANALYSIS CONCEPT. The purpose of a modal analysis algorithm is to identify the system parameters (eigenvalues and eigenvectors) from the experimentally measured data. Theoretical articles listed below describe advanced methods for modal analysis based on a measured subset of the Frequency Response Functions (FRF) of the real structure and their application to bridge identification.

First mode of vibration of a concrete bridge, identified from experimental data using DSMA

Third mode of vibration of a concrete bridge, identified from experimental data using DSMA

Fuges concrete bridge: Mode 1, 10.51 Hz damping ratio 3.07 %

Fuges concrete bridge: Mode 3, 13.522 Hz damping ratio 3.89 %

Articles:

Due to highly mathematical content of these articles, they are in the LaTeX (TeX) format. To enjoy reading them, you need to download the LaTeX plugin (Techexplorer from IBM) for your bowser first.

DIRECT SIMULTANEOUS MODAL APPROXIMATION METHOD (DSMA). ABSTRACT. A direct method for estimation of modal parameters from a set of experimentally measured Frequency Response Functions (FRF) is presented. All modal parameters (eigenvalues and eigenvectors) in a given frequency range are evaluated directly and simultaneously from all available FRF data. The method employs a gradient technique to minimise approximation errors for all measured FRFs simultaneously. Any fragment of the FRF matrix can be used, provided that it contains sufficient data about all eigenvectors sought. The paper demonstrates the superior accuracy of the method and its excellent performance for data containing up to 50% measurement noise. The formulation of a criterion enabling comparison of various modal parameter estimation methods is also given.

MODAL PARAMETER ESTIMATION WITH UNIFORM TOLERANCES. ABSTRACT. This paper deals with the evaluation of tolerances for modal parameters (eigenvalues) reconstructed from measured spectral data containing measurement noise. A modified approximation criterion is presented which provides control of the tolerances of reconstructed eigenvalues so that a similar accuracy can be achieved for all considered modes. Absolute as well as relative accuracy of estimated eigenvalues is discussed. A method of evaluating tolerances of reconstructed eigenvalues on the basis of approximation errors for measured data is given. Analysis is presented for the case of a simple spectrum (separated eigenvalues).

 Dynamics and vibration homepage | Department homepage