ABSTRACT
The paper presents an application of the novel methodology for the assessment of structures using a semi-probabilistic approach exploiting advanced probabilistic modeling and experimental results. The selected existing bridge is represented by a costly finite element model, which reflects the non-linearity of concrete and the construction process. Due to a significant computational burden of each simulation, it is not feasible to perform a Monte Carlo simulation and a semi-probabilistic approach was thus adopted. In this study, we investigate the possibility of a Gram-Charlier expansion described by the first four central moments efficiently obtained directly from Polynomial Chaos Expansion metamodel together with the uncertainty quantification of input random variables described by a joint probability distribution obtained from experimental data combined with prior assumptions from codes. Obtained results are compared to the standard approach assuming a Lognormal probability distribution of structural resistance.
