Sensitivity analysis and parameter identification to railway bridge structural model updating

Resumo

One of the major challenges in the management of infrastructure systems is to ensure their safety and
structural integrity throughout its useful life. This is due to the fact that the material and structural properties loss
and the eventual failure of these structures has catastrophic consequences. In that context, this project aims to
contribute to the application of methods that allow the safety and structural integrity assessment of railway bridges
based on experimental modal data. The objective is to apply the sensitivity analysis of a real railway bridge
structure using finite element model update to its unknown structural parameters. For this purpose, the Sobol‘
indices are studied to describe how the variability of the model response is affected by the variability of each input
parameter or combination thereof. Usually, these indices are computed by Monte Carlo simulation. However, they
are practically not applicable to CPU-intensive models such as finite element models. An alternative approach to
overcome this scenario is the use of surrogate models to speed-up the calculations, such as a polynomial chaos
expansion, a robust framework to compute Sobol’ indices. Thus, for the sensitivity analysis of this paper, it brings
a connection between these approaches. Finally, the role of the parameter identification is discussed, based on the
results of the sensitivity analysis.