Modal identification of bridges via cost-effective approaches and advanced decomposition techniques

The present PhD thesis investigates the use of cost-effective strategies in combination with advanced decomposition techniques for the modal characterization of existing bridges. A usual strategy for the health condition assessment of existing structures relies on vibration-based approaches; specifically, the most sought-after ones exploit long-term continuous monitoring records under ambient vibrations, processed via operational modal analysis techniques. However, the costs associated with the network of sensors, its protection and maintenance as well as the time and effort required to carry out the tests and to process large amounts of data make the application of this strategy profitable only in the case of strategic structures. This PhD thesis presents the results of several research studies with the main goal of proposing alternative approaches for the modal identification of existing bridges that mitigate the time and cost-related limitations of current ambient vibration monitoring, making it affordable also to monitor the health conditions of ordinary structures. In this regard, two possible solutions are studied. The first one is based on free vibration tests performed with a limited number of sensors directly mounted on the bridge; the second one explores the possibility of providing an approximate estimation of modal parameters via an indirect approach based on the dynamic response of moving vehicles. Aiming at processing the recorded signals, identification strategies based on two advanced signal decomposition techniques, namely the Variational Mode Decomposition (VMD) and the Empirical Fourier Decomposition (EFD), are presented. The proposed approaches are validated with numerical benchmark applications. Further, their performance is also assessed and compared with other traditional approaches in real case-studies concerning roadway and railway bridges.