The steel–concrete interface (SCI) is known to play a major role in corrosion of steel in concrete, but a fundamental understanding is still lacking. One reason is that concrete’s opacity complicates the study of internal processes. Here, we report on the application of bimodal X-ray and neutron microtomography as in-situ imaging techniques to elucidate the mechanism of steel corrosion in concrete. The study demonstrates that the segmentation of the specimen components of relevance—steel, cementitious matrix, aggregates, voids, corrosion products—obtained through bimodal X-ray and neutron imaging is more reliable than that based on the results of each of the two techniques separately. Further, we suggest the combination of tomographic in-situ imaging with ex-situ SEM analysis of targeted sections, selected based on the segmented tomograms. These in-situ and ex-situ characterization techniques were applied to study localized corrosion in a very early stage under laboratory chloride-exposure conditions, using reinforced concrete cores retrieved from a concrete bridge. Several interesting observations were made. First, the acquired images revealed the formation of several corrosion sites close to each other. Second, the morphology of the corrosion pits was relatively shallow. Finally, only about half of the total 31 corrosion initiation spots were in close proximity to interfacial macroscopic air voids, and > 90% of the more than 160 interfacial macroscopic air voids were free from corrosion. The findings have implications for the mechanistic understanding of corrosion of steel in concrete and suggest that multimodal in-situ imaging is a valuable technique for further related studies.

@en

The steel–concrete interface (SCI) is known to influence corrosion of steel in concrete. However, due to the numerous factors affecting the SCI—including steel properties, concrete properties, execution, and exposure conditions—it remains unclear which factors have the most dominant impact on the susceptibility of reinforced concrete to corrosion. In this literature review, prepared by members of RILEM technical committee 262-SCI, an attempt is made to elucidate the effect of numerous SCI characteristics on chloride-induced corrosion initiation of steel in concrete. We use a method to quantify and normalize the effect of individual SCI characteristics based on different literature results, which allows comparing them in a comprehensive context. It is found that the different SCI characteristics have received highly unbalanced research attention. Parameters such as w/b ratio and cement type have been studied most extensively. Interestingly, however, literature consistently indicates that those parameters have merely a moderate effect on the corrosion susceptibility of steel in concrete. Considerably more pronounced effects were identified for (1) steel properties, including metallurgy, presence of mill scale or rust layers, and surface roughness, and (2) the moisture state. Unfortunately, however, these aspects have received comparatively little research attention. Due to their apparently strong influence, future corrosion studies as well as developments towards predicting corrosion initiation in concrete would benefit from considering those aspects. Particularly the working mechanisms related to the moisture conditions in microscopic and macroscopic voids at the SCI is complex and presents major opportunities for further research in corrosion of steel in concrete.@en

Although the steel–concrete interface (SCI) is widely recognized to influence the durability of reinforced concrete, a systematic overview and detailed documentation of the various aspects of the SCI are lacking. In this paper, we compiled a comprehensive list of possible local characteristics at the SCI and reviewed available information regarding their properties as well as their occurrence in engineering structures and in the laboratory. Given the complexity of the SCI, we suggested a systematic approach to describe it in terms of local characteristics and their physical and chemical properties. It was found that the SCI exhibits significant spatial inhomogeneity along and around as well as perpendicular to the reinforcing steel. The SCI can differ strongly between different engineering structures and also between different members within a structure; particular differences are expected between structures built before and after the 1970/1980s. A single SCI representing all on-site conditions does not exist. Additionally, SCIs in common laboratory-made specimens exhibit significant differences compared to engineering structures. Thus, results from laboratory studies and from practical experience should be applied to engineering structures with caution. Finally, recommendations for further research are made.

@en

Where exposed to an aggressive environment, concrete structures will develop reinforcement corrosion at some point in their life. Chloride ions from de-icing salt or sea water induce pitting corrosion. Local cross section reduction may occur relatively quickly and expansion due to corrosion products precipitating may crack and spall concrete cover and reduce steel-concrete bond. With the aging of the existing stock, questions arise about the effects of corrosion on their structural behaviour, safety and reliability.TNOhas started a four-year programme to investigate this subject. One project concerns modelling of pitting corrosion from electrochemical principles, aimed at prediction of cross section loss with time and, in a next stage, corrosion product formation. Existing models for pitting corrosion are highly simplified and essentially two-dimensional, while pitting is clearly a three-dimensional phenomenon. The difference may have serious consequences for structural safety. New approaches are needed, including new conceptual models and addressing uncertainties on several levels. The paper presents some conceptual models for pitting corrosion and reports preliminary results of numerical (finite element) modelling of pitting corrosion, based on solving potential and current distributions and resulting in cross section loss as a function of time. Further work including experimental validation is foreseen.

@en