Corrosion of steel in concrete in concrete due to the ingress of chloride is a major factor in the premature failure of steel reinforced concretes. Thus changes in the concrete mixture to reduce permeability to chloride are highly desirable. Lowering the water-to-cementitious ratio (w/cm) is a very effective means of reducing permeability. Pozzolans such as fly ash and silica fume are very effective in reducing the apparent diffusion coefficient for chloride at a given w/cm. Typically, most of the reduction in diffusion with silica fume occurs within the first 28 days, whereas fly ash containing concretes see a more gradual reduction over time. Several life-cycle predictions models use a time dependent value of the apparent diffusion coefficient. In this paper, data are presented that can be used to estimate how the apparent diffusion coefficient and corrosion behavior changes in time, thus providing useful information for modeling. The laboratory results in experiments that have been in progress for over 10 years are in good agreement with results from larger specimens exposed to simulated tidal conditions using ocean water in an outdoor test facility in Daytona Beach Florida. The data indicate that the reductions in diffusion coefficients for fly ash concretes primarily occur in the first 2 to 3 years, and that the overall slope on a log- log scale for diffusion versus time is less than ?0.35 and mostly under ?0.2.
A major aspect of the modeling of the corrosion of steel in concrete exposed to chlorides involves predicting the ingress of chloride as a function of concrete mixture proportions and the environment. Many early models relied upon one-dimensional diffusion with a constant apparent diffusion coefficient and surface concentration of chloride [1-4]. These models were quite good in indicating the benefits of reducing the chloride diffusion coefficients through reduced water-tocementitious ratio (w/cm).
Realizing that the chloride concentration often increased in time, and that geometries were not always one-dimensional, led to further refinements in the modeling of chloride ingress [5-7]. These models have been extensively used in recent years to help in the design of new structures [7-9]. Several of these models incorporate chloride binding with the cement and differentiate between bound and free chloride [10-11]. Though more exact than total chloride only models, the final predictions are quite similar [3-6,12-16].
In addition to changing chloride concentrations and 2-dimensional modeling, several researchers have noted that the diffusion coefficient can decrease in time due to continued hydration of the concrete [13-19]. In many cases the degree of reduction in diffusion is based upon long-term data on concretes made with cements significantly different than those used today. Specifically, cements produced before the 1960?s were coarser and had a lower ratio of tricalcium silicate to dicalcium silicate, and thus had less early hydration, but more hydration over time. Thus, some of the data used to predict the decrease in the diffusion coefficients could be misleading by overestimating the time for chlorides to reach the steel reinforcement.