Structural model updating using dynamic data

Parameter estimation through model updating is a powerful technique that has gained widespread popularity for condition assessment of aging infrastructure. Among the available techniques, the sensitivity method is perhaps the most popular approach to perform model updating in industrial-scale systems. However, convergence issues are inherent to the method, due to ill-conditioning of the sensitivity matrix and measurement errors. To overcome the afore-mentioned concerns, regularization techniques were successfully applied in simulated and experimental scenarios. Model updating is performed using the University of Central Florida Benchmark Structure which is a laboratory specimen intended to capture the behavior of short to medium span bridges. The finite element model of the Benchmark is updated using (1) a modal parameters-based and (2) a frequency response function (FRF)-based error functions where both methods consider sparse data. In the simulated study, the aim is to evaluate the robustness of both error functions under random noise and ill-conditioning. Finally, the methods are successfully validated using experimentally obtained dynamic data. Based on the obtained results, it is concluded that the FRF-based method is more robust under simulated noise and can handle more unknown parameters in the optimization problem. However, advantages and challenges inherent to each method are presented only in light of those experienced by the authors.