A hierarchical Bayesian framework for model updating regarding structural systems

Resumo

The inherent variability of parameters that define engineering systems result in uncertainties for the
predicted response. This scenario is even more noticeable for civil engineering structures, where environment
effects, acting along the whole lifespan of the system, can accelerate the structure deterioration process. In this
context, the structure can reach a safety level lower than the acceptable minimum and, if no corrective actions be
taken by responsible authorities, catastrophic scenarios could occur. Thus, for such responsible authorities, it is of
paramount interest that a reliable bound of the current safety level be available. In this context, the present work
proposes the application of a hierarchical Bayesian framework to perform model updating regarding structural
engineering. By this approach, it is enabled the estimate of the inherent variability of the own parameter to be
updated, avoiding the underestimation of its total uncertainty. In order to assess the suitability of the proposed
approach, it is compared with the classical Bayesian approach in a set of numerical simulations regarding the
response of a cantilever beam due to a point load. Results showed that the hierarchical Bayesian approach is able
to provide more reliable estimates for the total uncertainty regarding the parameter of interest.