| Popis: |
Due to spatial and temporal variations of environmental, loading conditions, material properties, and geometries of large-scale reinforced concrete (RC) structures, spatial reliability analysis of corroded RC structures is crucial to perform informed decision-making in maintenance, retrofitting and serviceability enhancement of existing infrastructure. In reality, the probability of collapse of the entire structure is much less than that of local failures that may require interruption or delay of service that may result in more substantial loses of opportunity costs. Therefore, precise maintenance and decision-making require estimations of local and spatial reliability in the level of components rather than in the level of structures. In this thesis, in order to overcome current challenges faced by the spatial reliability analysis of large-scale corroded reinforced concrete structures, first, analytical corrosion models and numerical simulations have been implemented to simulate initiation of corrosions, reductions of rebar diameters, corrosion-induced crack propagations, and eventually toward degradation of elastic moduli properties of RC structures. In the established analytical and numerical analysis, critical environmental and geometrical variables are modeled as spatially varying random fields to simulate uncertainty propagation and predict service life of structures according to the limit state functions defined based on reduction of rebar diameter, crack width and degradation of elastic modulus. As an example, this study uses the established analysis method to probabilistically predict the service life of a reinforced concrete slab subjected to corrosion-induced cracks. The results are analyzed and insights on probabilistic service life are provided. Also, the sensitivity of statistical parameters (mean, coefficient of variation etc.) for selected corrosion damage related spatially random variables on the service life prediction are examined. In addition, a new spatial reliability updating framework based on incomplete monitoring data is proposed in order to realize more accurate probabilistic service life prediction of structures during structure service period. In this proposed spatial reliability updating framework, a recently developed self-optimizing inverse parameter identification method (Stochastic SELF-OPTIM) is used to identify the material parameters of structure from the incomplete monitoring data. The pre-established analytical and numerical analysis model of corrosion-induced reinforced concrete degradation and the Bayesian updating method are also incorporated so that the spatial reliability of structure can be updated. |
