Finite Element Model Calibration of a Full Scale Bridge Using Measured Frequency Response Functions

A frequency response function based parameter estimation method is used for finite element model calibration of a full-scale bridge utilizing measured dynamic test data. Dynamic tests are performed on the bridge to obtain measured frequency response functions. Data quality is ensured by comparing measured data with the finite element model and removing erroneous data. A coherence data selector is developed to remove the noise contaminated portions of measured frequency response functions. An unrefined finite element model is created using design information for the geometry and structural parameters. This model is improved to a refined model by using concrete cylinder property data, as-built drawing geometry, and the addition of components that participate in the dynamic response of the bridge. The model of the bridge is then successfully calibrated using measured frequency response functions. An increase in the negative bending region concrete deck rigidity is found during the calibration and verified by increased reinforcement in that area. Examining the second norm of the residual between the different models and the measured data resulted in an improvement from the unrefined to the calibrated finite element model. The proposed method proved to be robust in the presence of modeling and measurement errors and computationally efficient for use with full-scale structures.