Experimental and numerical investigation of chloride ingress in cracked concrete

Abstract

Chloride induced corrosion of reinforcing steel is recognized as the most common deterioration mechanism affecting reinforced concrete structures. As such, it has been in focus of research for more than thirty years. Numerous studies of chloride ingress, corrosion initiation, and corrosion propagation have been conducted. Most studies of chloride ingress focused on sound (uncracked) concrete. In reality, however, concrete is almost never crack free. Cracks form either in the construction phase (early age cracks, for example shrinkage cracks), or during the use of a structure (e.g. cracks caused by mechanical loads). While these cracks are usually not detrimental to the load bearing capacity of a structure, they are potentially a threat to its durability. Cracks occurring in the concrete cover diminish its protective capabilities, and present fast routes for ingress of deleterious species (e.g. chloride ions). While national and international design codes provide guidelines and limits for maximum crack widths in aggressive environmental conditions, these are often empirical and based on rules of thumb. As a result, only the surface crack width is considered. However, recent findings seem to indicate that an even more important factor could be the zone of debonding which occurs at the steel/concrete interface due to cracking of the cover. In this thesis, an attempt is made to increase the body of knowledge related to chloride ingress in cracked concrete. Laboratory experiments and numerical simulations were used during the study. Experimental data available in the literature is, at the moment, inconclusive. For years researchers have been trying to find a so-called ''threshold'' crack width for chloride (or water) transport, below which concrete can be treated as sound (uncracked). In this quest, mostly plain concrete specimens were used. While this approach resulted in increased understanding of chloride transport in cracks, it failed to address an important mechanism which affects only reinforced concrete - debonding occurring at the steel/concrete interface. Only recently have researchers focused their attention on the effect of damage at the steel/concrete interface on transport behavior and corrosion of reinforcement in concrete. In this thesis, compact reinforced specimen geometry is adopted, which mimics (with respect to crack geometry) the behavior of reinforced concrete beams. Cracked specimens were subjected to weekly cycles of salt water wetting and drying for a prolonged period of time. After the exposure, two-dimensional chloride maps were obtained by means of LIBS (Laser Induced Breakdown Spectroscopy), in collaboration with BAM Federal Institute for Materials Research and Testing in Berlin, Germany. The results showed that, once damage occurs at the steel/concrete interface, chloride ions penetrate parallel to the reinforcement, which could possibly be very harmful with respect to reinforcement corrosion. It has been frequently reported in the literature that autogeneous healing of cracks can reduce chloride ingress in concrete. To examine this, experiments were performed to investigate the influence of the curing ("healing") regimen on chloride penetration depth in cracked specimens. Two regimens (submerged in water and in 95% relative humidity) both enabled the tested specimens to reduce chloride ingress in cracks, compared to the control series. It was found that, under favorable conditions, autogeneous crack healing does have a positive effect on chloride penetration resistance. Corrosion induced cover cracking has been extensively studied in recent years. When reinforcing steel corrodes, it causes expansive pressure on the surrounding concrete. As a consequence, concrete cover cracks. In the thesis, cracking induced by accelerated corrosion was studied using X-ray computed tomography. Mechanical properties of the rust layer were determined using the nanoindentation technique. Valuable insights were obtained, especially for fine tuning of numerical models. Numerical models can be of use in understanding complex problems. In this thesis, a model for chloride ingress in cracked concrete, based on the lattice modeling approach, was developed. Concrete is discretized as a set of one dimensional "pipe" elements through which the transport takes place. The model is coupled to the lattice fracture model, which enabled simulating chloride penetration around cracks in concrete. The model was validated using experimental results from the literature. Numerous experiments make use of external electrical field to accelerate chloride ingress. Therefore, the developed model was extended to enable modeling chloride ingress in accelerated experiments. Based on the lattice modeling approach and utilizing the Characteristic Galerkin scheme, the approach enables using smaller elements compared to the conventional Galerkin approach. Apart from its computational efficiency, it showed that the, under accelerated conditions, chloride front ahead of a crack may be much more sharp compared to the natural (i.e. diffusion) conditions. Therefore, it is questionable if findings from accelerated experiments on cracked specimens can be directly applied to "real" exposure conditions. Cracking of the concrete cover caused by reinforcement corrosion is also frequently modeled. In this thesis, the two-dimensional Delft lattice model was used to model it. First, the developed model was validated using a set of well-documented experiments from the literature. It was found that penetration of corrosion products into pores and open cracks needs to be considered in order to obtain good match with the experiments. Furthermore, it was observed that the internal pressure which causes cracking is not a deterministic value, but that it depends on local mechanical properties inside the concrete. Also, two hypothetical pitting scenarios were tested and compared to then uniform corrosion case. It appears (according to the 2D model) that pitting is more detrimental for the concrete cover than uniform corrosion. This means that cover cracking will occur at lower internal pressure compared to the uniform corrosion case. Techniques employed in this thesis (such as LIBS, X-ray computed tomography, and nanoindentation) can be used in the same or similar way for studying other deterioration mechanisms. Also, the chloride transport model can be easily modified to model other transport processes, such as moisture transport, sulfate transport, or carbon dioxide ingress. Employing a staggered scheme used in this thesis, these transport processes can be coupled with the mechanical analysis to model the influence of combined actions on reinforced concrete structures.