Effect of corroded reinforcement in RC structures: from cracking to bond performance

Reinforced concrete (rc) is the most used building material. The natural ageing of the rc structures moves the attention to the durability aspect. The latter depends on the environmental conditions, which are getting always more severe, but also it depends on the poor detailing and material characteristics during casting. Hence, the deterioration of rc structures becomes a matter of primary importance. The most common cause of deterioration of rc structures is the reinforcement corrosion: it affects the steel reinforcement reducing the sectional area and modifying the mechanical properties. Moreover, corrosion provokes the concrete section loss because of the cracking induced by the expansive phenomenon; hence the resisting section is reduced and the deteriorated reinforcement is further exposed to the aggressive environment. Finally, corrosion deeply changes the characteristics of the interface layer between steel and concrete because of the oxides formation; therefore, bond between the materiala changes. In the present dissertation, steel reinforcement corrosion and its mechanical implications were studied; in particular, the investigation was developed both experimentally and numerically. As far as the former is concerned, pull-out test of corroded RILEM type specimens was performed. The experimental variables were three: the corrosion level; the main reinforcement dimension and the presence of the transversal reinforcement. The maximum corrosion level reached was 20% in terms of mass loss, this values is far greater than those available in the literature which rarely overtake the 10% of mass loss; further corrosion nominal levels were 2, 5 and 10%. The main reinforcement, whose diameter was of 12 and 16 mm, were artificially corroded by means of the application of a low density current in order to reach the nominal corrosion level with a reliable accelerated mechanism. Confined specimens were provided with closed stirrups which were corroded up to the level of the main reinforcement. During the corrosion phase, the specimens were daily soaked in a tank containing water and then stored in a humidity controlled room, those operations allowed to accomplish the wet and dry condition. The cracks were measured every 48 hours and the collected data were merged to other outcomes obtained from the literature. An interpolation curve was proposed: it is able to forecast the corrosion level starting from the crack width measure and taking into account geometrical and mechanical characteristics of the rc elements. The core of the experimental program was the pull-out test stage which allowed to obtain the bondslip curves of the specimens firstly corroded and then tested. At the end of the pull-out test, the reinforcing bars and the stirrups were extracted and the actual corrosion rates were determined. Results of the test integrate the present literature and highlighted the role of the inspected parameters. Furthermore, the outcomes were used for calibrating the numerical model. The numerical study lead to the definition of a simple and modular analytical model which allows to take into account the presence of the confinement and the corrosion products. The modelling was realized in the longitudinal and in the transversal direction: hence the model can be considered three dimensional with axial symmetry assumption. In the longitudinal modelling, the boundary value problem is solved using a local bond-stress similar to that one proposed by the MC2010. The transversal modelling was based on the thick-walled cylinder theory which allowed to determine the pressure acting at the steel to concrete interface. In this context, a softening behaviour of the cracked concrete was assumed. Transversal and longitudinal modelling were linked by means of a friction criteria whose parameter depend on the corrosion level and on the concrete tensile strength. Finally, the model was used to simulate some experimental results in the literature. Taking into account the great scatter of the outcomes from the bond test and also due to corrosion phenomenon, a good agreement between the experimental outcomes and the numerical results is observed.