Applying alternative identification methods in eccentric mass shaker experiments

Abstract

When eccentric mass shakers are used for forced vibration testing of actual structures, the Peak Picking Method is generally used for identifying modal parameters. Here we investigate the applicability of two time domain methods, namely the Eigensystem Realization Algorithm applied with Auto Regressive Exogeneous Models and the Covariance Driven Stochastic Subspace Method, and one frequency domain method, namely the Frequency Domain Decomposition Method, as alternatives to Peak Picking. To this end, a finite element model of a 3-storey building is prepared and forced vibration tests are simulated on this model based on the properties of the eccentric mass shaker present at the Boǧaziçi University Structures Laboratory. Biases in modal parameter estimates are analyzed via Monte Carlo simulations and it is observed that identified mode frequencies and damping ratios fall within ± 3% of the actual values, and that the identified mode vectors have MAC numbers higher than 0.95. © RILEM 2013.