Bayesian updating of the reliability of existing RC structures based on the inspection results

The seismic assessment of existing buildings is subject to uncertainties. These uncertainties can sometimes be comparable to that of the ground motion representation. Seismic codes such as the European code and the Italian Code seem to consider the overall effect of these uncertainties through confidence factors applied to nominal material strength. This work aims to classify and characterize these uncertainties and build a prior distribution of structural lateral load resistance capacity as a function of these uncertainties. The prior probability distribution model for the uncertain parameters is constructed based on the state of knowledge about the building before in situ inspections and tests are conducted. The uncertainties in the parameters are propagated to the structural resistance using an advanced simulation method known as subset simulation. Subset simulation allows for suitable grouping of parameters into groups in order to build a simplified model of correlation across different structural parameters. In the next step, the results of tests are used to update the probability distributions for parameters and to propagate the effects on to the structural resistance. The results of this study are aimed to make recommendations as to how confidence factors can be estimated taking into account all sources of uncertainty and be applied to the structural resistance. Finally, the effect of performing tests in increasing the level of knowledge about the structure are quantified in terms of a decrease in the confidence factor. As a case study, the two-dimensional finite element model of the central frame of an existing school built in the 1960’s in Avellino Italy has been used.