This paper presents a time-dependent reliability analysis of prestressed concrete girders subjected to degradation caused by pitting corrosion. The procedure presented includes the effects of both spatial and temporal pitting corrosions on prestressing steel, as well as the degradation of the strengthening CFRP laminate used for the rehabilitation of the member. Results indicate that the correlation of corrosion in different segments of the prestressing tendons impacts on the computed safety index for the deteriorated structure. Ditlevsen bounds are proposed for a better approximation of the correlation between failure modes in the spatial discretisation. Results show that such approach produces adequate estimates of the reliability index over the full range of analysis in comparison with other tested models. It is also observed that the degradation of the CFRP laminates does not affect the reliability as significantly as corrosion, and that traffic loads, models uncertainties, corrosion error and corrosion rate are the most relevant variables in the analysis, followed by prestressing strength and concrete cover. The significance of the variables changes with time: the corrosion rate, corrosion model error and concrete cover increase in importance with the development of corrosion, whereas traffic loads become gradually less important.

Fibre reinforced polymer (FRP) composites have been increasingly used worldwide in the strengthening of civil engineering structures. As FRP becomes more common in structural strengthening, the development of probability-based limit state design codes will require accurate models for the prediction of the mechanical properties of the FRPs. Existing models, however, are based on small sample sizes and ignore the importance of the tail region for analyses and design.Addressing these limitations, this paper presents a probabilistic-based characterisation of the mechanical properties of carbon FRP (CFRP) laminates using a large batch of tension tests. The analysed specimens were pre-cured laminates of carbon fibres embedded in epoxy matrices, which is the most commonly used laminate for the strengthening concrete beams and slabs. Based on the existing data, probabilistic models and correlations were established for the Young's modulus, ultimate strain and tensile strength. Analyses demonstrate the suitability of the Weibull distribution for the estimation of CFRP properties. Results also show that the statistical characterisation of the mechanical properties should be performed with a focus on the tail region. The proposed distributions constitute a set of validated probabilistic models that can be used for performing reliability analyses of structures strengthened with CFRP laminates.

No abstract

No abstract

No abstract