PSI - Issue 62

Simone Celati et al. / Procedia Structural Integrity 62 (2024) 361–368 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

367

7

4. Sensitivity analysis A sensitivity analysis is carried out by computing the Spearman correlation coefficient for each input variable and the LSF. Fig. 3 illustrates the sensitivities of the variables during the initial and final years under consideration.

Fig. 3. Spearman correlation coefficients of the 8 main variables in different years

The analysis demonstrates that model uncertainties are significantly more important compared to the other variables when the structure is well-preserved. In particular, the resistance model uncertainty has a significant sensitivity over the service life ranging from 0.92 in year 0 and decreasing to 0.52 in year 100. On the other hand, as expected, the sensitivity of the corrosion rate is influenced by the rate of corrosion when active ( , , ) and the time of wetness ( ) variables, increases as the degradation process advances. Both are very low at the beginning of the service life and become more significant in year 40 reaching a sensitivity of respectively 0.47 and 0.16 in year 100. The low sensitivity of the yielding stress ( ) of the steel can be attributed to its very low variability. 5. Conclusions This paper aims to provide a reliability assessment for the widespread post-tensioned girder structures, aiding their service life modelling and effective management. For this aim, a methodology is developed to assess the time dependent reliability of these girders exposed to tendon corrosion. For tendon corrosion, a novel approach is developed accounting for different corrosion initiation mechanisms and the conditions for corrosion development. The developed approaches are demonstrated by calculating the reliability index over time for a typical structural configuration. Furthermore, a sensitivity analysis is performed highlighting the most important model variables over time and for progressing degradation. The developed approach constitutes a starting point for reliability analysis over the service life of a post-tensioned structure accounting for the extreme load events and the degradation of the capacity due to corrosion of the cables modelled as a system of wires. The developed approach allows for considering the experience acquired from existing structures namely that the active corrosion may not coincide with a high chloride concentration at the reinforcement surface. The proposed degradation model, in which both chloride diffusion and carbonation are considered, facilitates to account of the electrochemical conditions required for active corrosion. The exemplary study demonstrates that relying only on the chloride diffusion process to forecast corrosion development can significantly underestimate the service life of post-tensioned structures. The sensitivity analysis highlights that, for structures with low degradation, model uncertainties in both the capacity and demand models significantly influence the assessment. Therefore, narrowing the variability of these uncertainties can significantly contribute to quantify the structural reliability more realistically. As the degradation progresses, the corrosion rate becomes increasingly important, and updating this parameter may influence the structural reliability.