Sensitivity Analysis of Corrosion Rate Prediction Models Utilized for Reinforced Concrete Affected by Chloride

Chloride-induced reinforcement corrosion is one of the major causes of premature deterioration in reinforced concrete (RC) structures. Given the high maintenance and replacement costs, accurate modeling of RC deterioration is indispensable for ensuring the optimal allocation of limited economic resources. Since corrosion rate is one of the major factors influencing the rate of deterioration, many predictive models exist. However, because the existing models use very different sets of input parameters, the choice of model for RC deterioration is made difficult. Although the factors affecting corrosion rate are frequently reported in the literature, there is no published quantitative study on the sensitivity of predicted corrosion rate to the various input parameters. This paper presents the results of the sensitivity analysis of the input parameters for nine selected corrosion rate prediction models. Three different methods of analysis are used to determine and compare the sensitivity of corrosion rate to various input parameters: (i) univariate regression analysis, (ii) multivariate regression analysis, and (iii) sensitivity index. The results from the analysis have quantitatively verified that the corrosion rate of steel reinforcement bars in RC structures is highly sensitive to corrosion duration time, concrete resistivity, and concrete chloride content. These important findings establish that future empirical models for predicting corrosion rate of RC should carefully consider and incorporate these input parameters.