Prediction of Chloride Ingress into Concrete - Verifications from the Field Exposure

Paper i proceeding, 2006

This paper presents the verification results from four prediction models for chloride ingress in concrete: 1) the simplest error function solution to Fick’s second law (Model 1); 2) the DuraCrete model (Model 2); 3) a mathematically improved DuraCrete model (Model 2b) and 4) the mechanism-based model ClinConc (Model 3). Chloride profiles taken from 10 types of concrete from the field exposures are used to verify these models. The concrete compositions include Portland cement and blended with pozzolanic additives (silica fume, fly ash and their combination), with water-binder ratios from 0.25 to 0.40. The results show that the mechanism-based model ClinConc reveals good prediction for all the cases. The model using simplest error function to Fick’s second law significantly overestimates chloride ingress. The DuraCrete model in most cases underestimates chloride ingress, especially for the concrete with pozzolanic additives, due to its oversimplified mathematic expression in dealing with the age factor for chloride diffusion coefficient. The results suggest that care must be taken when using empiric models for service life prediction, because the actual chloride ingress in concrete with silica fume and fly ash after 10 years’ field exposure has in some cases already exceeded the 100 years’ prediction made by the DuraCrete model.