Sensitivity of chloride ingress modelling in concrete to input parameter variability

The Rosenblueth point estimator method is used to propagate uncertainty on stochastic parameters in a multi-ionic reactive transport model for concrete degradation. The degradation mechanism that was analysed is the ingress of chlorides into concrete. This point estimator method is used because it requires a minimum number of simulations to compute means, standard deviations and skewness for the solution of the transport model. Two exposure cases are considered in this article: the first is a saturated case, and the second is an unsaturated case with wetting (4 days) and drying (3 days at 50 % relative humidity) cycles. The sodium chloride exposure during the wetting cycle of both cases was 500 mmol/L. In both cases, predictions from the model were compared to experimental results. The Rosenblueth approach yielded results comparable to Monte Carlo simulations for both saturated and unsaturated cases. The relative sensitivity of model parameters on prediction results are investigated through a sensitivity analysis. The sensitivity results show the high importance of ionic diffusion parameters and of exposure conditions, while the calculated initial mineral phases in the hydrated paste have lower importance.