Corrosion of steel in cracked concrete

Abstract

Reinforcement corrosion is frequently considered as the predominant degradation mechanism affecting reinforced concrete structures. Reinforced concrete structures are commonly subject to harsh environmental and loading conditions in which aggressive species can penetrate. Chlorides, present in seawater or de-icing salts, are the main responsible for the initiation of reinforcement corrosion in reinforced concrete structures. Chlorides penetrate through the concrete cover by different transport mechanisms, i.e. capillary and/or diffusive processes. During time, the chloride concentration at the surface of steel reinforcement reaches a level known as the critical chloride content Ccr i t . The time necessary for chlorides to reach Ccr i t is known as the initiation period. Service life models that consider reinforcement corrosion as the governing degradation mechanism are composed of two periods: initiation and propagation. Numerous investigations have been dedicated to the study of chloride ingress and corrosion initiation and propagation. Most of them, however, have been conducted in crack-free concrete. In practice, different types of cracks can be present in concrete structures originated from construction (i.e. drying shrinkage cracks) to service (e.g. mechanical load) actions. The presence of cracks in concrete structures is commonly considered as irrelevant for their mechanical capacity. However, the influence of cracks upon the structures durability remains uncertain. Commonly, surface crack widths are limited and controlled by design codes based on exposure conditions. Often, these guidelines provide recommendations of ’acceptable’ surface crack width values which. Recent findings have shown that the presence of secondary cracks at the concrete-steel interface could be more detrimental to the durability of reinforced concrete structures. This thesis is devoted to the study of the influence of cracks on the initiation of reinforcement corrosion including aspects on the determination of chloride concentration in cement paste and concrete and the incorporation of cracks on service life predictions. One of the most important, and difficult to determine, parameters related to the chloride induced reinforcement corrosion is the corrosion inducing chloride concentration Ccr i t . In this thesis, the experimental determination of Ccr i t was carried out following a proposed test method by the RILEM Technical Committee 235-CTC "corrosion initiating chloride threshold concentrations" in which a test method for determining Ccr i t . Portland (PC - CEM I) and blast furnace slag (BFS - CEM III/B) cements were employed in the fabrication of reinforced concrete specimens with two different cover depths of 15 mm and 10 mm, respectively. After fabrication, the specimens were exposed to a chloride rich solution (3.3% NaCl) while monitoring of the electrochemical potential vs a reference electrode (SCE). Corrosion initiation was determined by detecting a shift of the electrochemical potential towards less noble values for a period of at least 7 days. Subsequently, chloride profiles were obtained by acid digestion followed by wet chemical analysis. Also, a Round Robin test (RRT) among the participating laboratories in the RILEM Committee was studied. In PC specimens, an average Ccr i t of 0.56% by weight of cement was found whilst in BFS specimens chloride contents were below 0.1% by mass of binder after 6 months of exposure. In the case of the RRT, concentrations of up to 1.3% by weight of cement were measured without signs of corrosion (potential shift) after 1 year of exposure. Results of electrochemical measurements and chloride concentrations for each series are discussed. Based on the obtained results, the test method is critically analysed and potential improvements to it are presented. Compared to bulk techniques (viz. acid digestion and titration), X-ray microanalysis can provide quantitative determination of element concentrations at high spatial resolutions. In order to perform fully quantitative microanalysis, X-ray spectra from rock forming minerals must be used as reference. Six minerals were studied as reference minerals for quantification of chlorine in cement paste: chlorapatite, halite, marialite, scapolite, tugtupite and zunyite. Cement paste samples with added chloride concentrations of 0.5, 1.0, 2.0, 3.5 and 5.0% by mass of cement were fabricated. Subsequently, X-ray spectra collected from C-S-H rich phases was analysed with each mineral. Results show that scapolite provided to be the most accurate concentrations of chlorine amongst the employed minerals. Deviations from acid digestion and wet chemical analysis are attributed to matrix effects (¯Z AF) between rock-forming minerals and cement paste such as the presence of chemically bound water in C-S-H, porosity and poorly crystalline structure. Due to the lack of appropriate minerals with similarities to C-S-H, the synthesis of a cement-like mineral is presented. For this, high pressure and high temperature (HPHT) experiments were performed at VU Amsterdam department of Planetary Evolution. Cement paste with a chloride concentration of 5% by weight of cement was subject to HPHT conditions of 1 GPa and temperature range between 1000 and 1600 ±C for 30 minutes. After quenching, the resulting material was analysed with a Wavelength Dispersive X-ray Microprobe and EDS. Results showed that an homogeneously glass porous matrix could be obtained. Subsequently, it was found that the synthetic glass provided better quantitative results compared to the previously tested minerals due to reduced matrix effects. Cracking of concrete was assessed by electrical resistance and image analysis in reinforced concrete specimens subject to a Modified Wedge Splitting Test. Concrete specimens with Portland cement (PC) and blast furnace slag (BFS) were fabricated with dimensions of 150 x 150 x150mm3. Cracking of concretewas performed under controlled displacement conditions while the electrical resistance across the crack was monitored. Results showed that the electrical resistance across the crack increased with increasing COD (crack opening displacement) at the concrete surface. Afterwards, cracks were impregnated with a fluorescent epoxy resin. Concrete slices of 20mmthickwere sawn from the specimens and the geometry of the cracks was assessed. Secondary cracks (roughly parallel to the rebars)were observed with COD values of 200 micrometer or higher. Image analysis of binary images obtained from the impregnated slices showed that the crack volume increased with increasing COD. The corrosion behaviour of cracked concrete specimens was monitored in reinforced cracked concrete specimens. The specimens were exposed to cyclic wet-dry cycles with a chlorine rich solution. Monitoring of the electrochemical potential, corrosion rate and electrical concrete resistivity were carried out. After 36 weeks, results showed that the corrosion rate of all PC specimens was higher than 1 ¹m/year whilst in BFS they were below this value. At the end of the cyclic exposure, reinforcement bars were mechanically extracted for inspections of the pit locations and dimensions. Stereomicrographs showed the presence of secondary cracks in specimens with COD values higher than 200 micrometer. Samples of the concrete-steel interface in the vicinity of corrosion pits were analysed with EDS. The concentration of chlorine was found to be higher in specimens with higher corrosion rate values. Estimations of accumulated pit volume were higher with increasing COD in all cases. Finally, the incorporation of cracks on chloride transport and service life predictions of concrete elements is presented. The model incorporates a correction to the diffusion coefficient of uncracked concrete determined by performance tests (natural diffusion or RCM). Prediction of the service life of concrete elements under marine exposure was performed with the DuraCrete model, including the correction to the resistance to chloride ingress. Estimations of ti as a function of crack width are studied. Results of the predicted ti from the model are then compared to observations of the ti of specimens studied in Chapter 7. For PC specimens, difference between estimations of ti were up to 4 times higher than the observed corrosion behaviour. In the case of BFS, the difference is larger. It seems that predictions of ti in BFS concrete are more sensitive to the increased resistance to chloride ingress as shown in the value of D0 which is one order of magnitude lower than in PC concrete.