Using Markov Chain Monte Carlo Simulation for Robust Reliability Assessment of Existing Structures Based on the Inspection Results

Determining the material properties and structural detailing in existing structures is subject to a signif-icant level of uncertainty. Therefore, instead of a unique structural model, a set of plausible structuralmodels can be identified. A robust assessment of structural reliability takes into account a whole set ofpossible structural models that are weighted by their corresponding plausibility. Moreover, performingin-situ tests and inspections can improve the state of knowledge about the structure. A bayesian updat-ing framework can be implemented in order to update both the structural modeling properties and thereliability based on test results (Beck and Katafigiotis, 1998).The present study is aiming at addressing the issue of modeling uncertainty in existing structures, inthe context of the European and Italian seismic guidelines (e.g., reference number 4). These guidelinessynthesize the effect of modeling uncertainties in the so-called confidence factors which are applied tothe mean material properties. Evaluation of these confidence factors is rather qualitative and dependson the acquired level of knowledge about the structure. These guidelines define three increasing levelsof knowledge, for each of which, they prescribe a certain set of verifying tests and inspections to beperformed. The objective of the present study is to take into account the structural modeling uncertain-ties in a Bayesian framework where the results of tests and inspections can be implemented in order toupdate the structural modeling parameters’ probability distribution and to calculate the robust reliability.