Chloride induced reinforcement corrosion is widely accepted to be the most frequent mechanism causing premature degradation of reinforced concrete structures. Condition assessment and service life prediction is based on comparing the chloride content in the concrete at the steel depth – either measured in the field or computed by means of theoretical modelling – with the chloride content that is believed to be tolerable before corrosion starts. The latter is commonly referred to as critical chloride content or chloride threshold value. Owing to the considerable statistical variation of the parameters involved in service life considerations, probabilistic approaches are preferentially used since these aim at taking into account the uncertainties inherent to all parameters – at least on a theoretical basis.
The present thesis approached the issue of chloride induced reinforcement corrosion from various angles. First, a non-destructive chloride measurement technique was studied. Second, the critical chloride content was reviewed with particular focus on how to determine this value experimentally and on common practice of its application. In a third part, the mechanism of chloride induced corrosion was experimentally studied.
Regarding the measurement of chlorides, the application of ion selective electrodes (ISEs) as non-destructive chloride sensors in concrete was investigated. It was found that silver / silver chloride electrodes respond to the chloride ion activity in the pore solution as expected from theory and are functional also in highly alkaline environments. However, correct measurement of the sensor potential is the critical step and in this regard, the presence of diffusion potentials was identified as serious error source. These disturbing potentials arise from concentration gradients along the measurement path between reference electrode and ISE, particularly owing to pH gradients and chloride profiles. The error can be minimised by optimal placing of the reference electrode with respect to the ISE. Generally, in uncarbonated, alkaline concrete, the accuracy of this non-destructive chloride measurement method was found to be comparable to the accuracy of common procedures to determine the acid-soluble chloride content in concrete powder. On the other hand, when the pH of the concrete is on a lower level such as owing to the presence of pozzolanas, the adverse effect of diffusion potentials arising from chloride profiles increases and negatively affects the measurement accuracy.
A review on the critical chloride content has shown that this parameter scatters significantly in the literature and that the published data does not offer a basis to improve service life predictions. The reported values are not consistent, particularly regarding non-traditional binder types. This was, at least partly, explained by the wide variety of experimental methods and the pronounced effect of certain experimental parameters. It was concluded that there is a strong need for a generally accepted, practice-related test setup for the critical chloride content. Without reliable input data, the common practice of probabilistic service life modelling is highly questionable. Both based on experimental results as well as the literature review, recommendations were made for a realistic test setup; these include the use of ribbed steel in as-received condition, chloride exposure by cyclic wetting and drying as well as leaving the rebar at its free corrosion potential rather than subjecting it to potentiostatic control.
While it was from experimental work concluded that even in rather small laboratory specimens, the cathode is sufficiently large to provide realistic conditions for (early) pitting corrosion, probabilistic considerations have illustrated that the specimen size is likely to significantly influence the measured critical chloride content. More specifically, the smaller the specimens, the higher the expected mean critical chloride content and the larger the scatter of measured values. It was further discussed how the size effect influences the concept of critical chloride content and service life modelling in general. It was suggested that the size of specimens on which the critical chloride content is measured has to be taken into account when transferring the values to structures of real-life dimensions in probabilistic service life calculations. A procedure of how this can be done by considering structural behaviour was sketched (characteristic length).
Regarding corrosion performance, the steel/concrete interface was found to be the most important influencing factor. Investigations by means of scanning electron microscopy revealed microstructural differences of top and lower sides of rebars that were horizontally orientated during casting, in particular the presence of a bleed-water zone below the reinforcement. It was striking that chloride induced corrosion initiated preferentially on the rebar side with the bleed-water zone regardless of the direction of chloride ingress. Also entrapped air voids were frequently observed at the steel/concrete interface; however, these coincided never with the location of corrosion onset. It was suggested that the internal moisture state is decisive in determining which interfacial defects present a risk of corrosion initiation.
Last but not least, it was experimentally observed that steel embedded in concrete might depassivate/repassivate several times until stable pitting corrosion is achieved – at least under unpolarised conditions. After the first signs of corrosion onset, a marked increase in chloride content was often required to prevent repassivation and to enable stable pit growth. The time at which the chloride content is measured and taken as critical chloride content is thus decisive for the outcome of a laboratory test method. It was suggested that in order to obtain practicerelated chloride threshold values, this should be done as soon as stable pit growth is achieved (rather than at the first depassivation event).
Finally, measurements after depassivation provided insight into the mechanism of early pitting corrosion and lead to the conclusion that the corrosion kinetics are at this stage dominated by anodic diffusion control.