Diagnosis of concrete structures: the influence of sampling parameters on the accuracy of chloride profiles

For determining the chloride threshold level of steel in concrete, the withdrawal of concrete samples is not a trivial task. The elected sampling method, the associated step size of depth intervals and the depth position of the steel with regard to the interval limits can exert a significant influence on the accuracy of the results. Also local variations of chloride ingress may lead to significant deviations from the effective chloride concentration in the depth of interest. In this paper, the chloride profiles from grinding powder samples with small interval sizes of 2 mm are compared to drilling powder samples with interval sizes of 10, 15 and 20 mm. Regression curves according to Fick's second law were generated and the corresponding chloride diffusion coefficients were determined. The deviations of the curves and the resulting diffusion coefficients were illustrated. The assumption of considering the determined chloride concentration as constant and valid for the entire interval depth was found to lead to huge deviations from the effective profiles which are not acceptable. The implementation of regression approaches from drilling powder samples lead to an extreme reduction of deviations, thus resulting profiles may be suitable for certain inspection tasks, even though most profiles examined do not exhibit a sufficient number of data points from depth ranges completely controlled by diffusion. Additionally, a theoretical modeling of chloride penetration curves was performed. A quantitative approach for the election of depth intervals in practice was developed based on the input parameters diffusion coefficient and exposure time.