Knowledge-Based Performance Assessment of Existing RC Buildings

One of the most challenging aspects of the seismic assessment of existing buildings is the characterization of structural modeling uncertainties. Recent codes, such as Eurocode 8, seem to synthesize the effect of structural modeling uncertainties in the so-called confidence factors that are applied to mean material property estimates. The confidence factors are classified and tabulated as a function of discrete knowledge levels acquired based on the results of specific in-situ tests and inspections. In this approach, the effect of the application of the confidence factors on structural assessment is not explicitly stated. This work presents probabilistic performance-based proposals for seismic assessments of RC buildings based on the knowledge levels. These proposals take advantage of the Bayesian framework for updating the probability distributions for structural modeling parameters based on the results of tests and inspections. As structural modeling parameters, both the mechanical material properties and also the structural detailing parameters are considered. These proposals can be categorized based both on the amount of structural analysis effort required and on the type of structural analysis performed. An efficient Bayesian method is presented which relies on simplified assumptions and employs a small sample of structural model realizations and ground motion records in order to provide an estimate of structural reliability. As an alternative proposal suitable for code implementation, the simplified approach implemented in the SAC-FEMA guidelines is adapted to existing structures by employing the efficient Bayesian method. This method takes into account the effect of both ground motion uncertainty and the structural modeling uncertainties on the global performance of the structure, in a closed-form analytical safety-checking format. These alternative proposals are demonstrated for the case study structure which is an existing RC frame. In particular, it is shown how the parameters for the safety-checking format can be estimated and tabulated as a function of knowledge level, outcome of tests, and the type of structural analysis adopted.