Uncertainty Quantification and Calibration of Dynamic Models of Existing Bridges Considering Difference of Modeling Strategy

This study presents one of strategies to deal with the uncertainties in the Model Verification and Validation (V&V) for the modeling of existing structures by applying the Bayesian uncertainty quantification (UQ) and the model calibration. Authors especially verified how the uncertain model parameters could be calibrated in different modeling strategies that must be determined by a person who were making the model based on the purpose of the numerical analysis. The target bridge was an existing seismic isolated bridge, and compared the performances of the UQ and the calibration between two finite element (FE) models; a detail and a frame models, using measured dynamic characteristics. It was then shown that the performance of UQ varied with the difference of the modeling because descriptions of model parameters and their prior uncertainties were different.

Furthermore, the results indicated some requirements in the nominal modeling of the structure for appropriate Model V&V.