Information-based formulation for Bayesian updating of the Eurocode 2 creep model

The disparity between theoretical and experimental results reveals that the creep of concrete is often underestimated by most, if not all, codes of design; this is particularly true in the case of Eurocode 2. Thus it is necessary to calibrate the present code models. Bayesian-type inferences turn out to be an especially suitable tool for the work needed in revising and updating design codes. This is by virtue of their capability to incorporate additional information resulting from current practice and research so as to improve existing models. In this paper, corrective coefficients are proposed for the Eurocode model, allowing better estimation of the long-term creep of concrete. To achieve this aim, the authors rely on a large database of experimental results compiled by collecting data from several research institutions in Europe. Two descriptive statistical methods are applied in order to compare the experimental results from the above-mentioned database with results calculated using the Eurocode 2 model for the same input parameters. A Bayesian-type statistical inference is then performed to evaluate the corrective coefficient for different categories of concrete strengths using various prior distributions. The approach presented here has proven to be an effective and systematic framework for the consideration of all possible types of uncertainties in model calibration. The results obtained are very interesting for engineers involved in design and supervision of structures. The adoption of such a design approach would improve long-term serviceability of structures subjected to creep.