PSI - Issue 54

Maricely De Abreu et al. / Procedia Structural Integrity 54 (2024) 143–148 De Abreu M. et al./ Structural Integrity Procedia 00 (2023) 000 – 000

144

2

those resulting from laboratory tests performed with a representative selection of prestressing wires in the damage condition.

Fig. 1. a) Strand-tendon without mortar injection with a large number of broken strands; b) in-service broken strands samples used in the failure analysis; c) cross-section of a failed broken strand far from failure. 2. Material and experimentation The strand pieces shown in Fig 1.b coming from broken strands similar to those provided in Fig 1.a were supplied by the supervising authority for road and transportation for failure analysis. Each strand piece consisted of one of the breaking ends of three seven-wire strands belonging to distinct post-tensioned tendons of the viaduct. The wires of the three pieces were strongly corroded, especially in the vicinity of the broken ends, with some of them with showing resistant cross-section losses. The initial impression is that the cross-section losses were caused by generalized corrosion which had continued after the wire broke, denoting that these remained exposed to an aggressive environment that might have damaged the steel regardless of the applied stresses. However, the synergic action of environment and stress state could not be excluded, in which case the first wires would have failed by stress corrosion cracking and the strand would have finally collapsed by overloading. In this last scenario, as explained by Nuernberger (2009) and observed by Iordachescu et al. (2020, 2021), corrosion would have contributed to failure by assisted cracking at the bottom of firstly initiated corrosion pits and would have continued to corrode the steel wires after collapse. This is confirmed by the geometrical condition of one of the strands shown in Fig. 1b, with the consequent cross-section far from the rupture zone that can be seen in Fig 1c. The resistant cross-section of the strand represents only 30% of the nominal section (15.2 mm diameter), with the rest being practically replaced by brittle and easily detachable corrosion products, consolidated into the helicoidal shapes that can be seen in dark color in Fig. 1c. A small sample of the supplied strand pieces was employed to determine the mechanical behavior of the damage free steel by means of tensile tests. To this end, the central straight wires of the three strand pieces were extracted far from the rupture zone and cylindrical tensile specimens with threaded heads were machined up to diameters of 2.50, 2.50 and 3.00 mm, so that any visual indication of surface damage became suppressed. The remaining dimensions (thread size and shank length greater than 25 mm) were adjusted to assure a valid tensile test with the gauge length of 12.5 mm provided by a resistive extensometer. Fig. 2a shows the obtained stress-strain curves and Table 1 the corresponding mechanical properties. The results of the tensile tests plotted in Fig. 2a indicate that the tensile behavior of the steel with which the strands were made does not differ from that of a current commercial prestressing steel. Neither the stress-strain curves nor the ruptures show any signs of embrittlement. In the three tests, the steel behavior changes from an elastic to an elastic-plastic regime with a smooth transition elbow and failure occurs by plastic instability and necking, which only causes the final fracture after a considerable reduction of the resistant area (the final elastic unloading observed in one of the tests is due to the necking occurrence outside the gauge length of the extensometer).