Identification of the failure mode of corroding steel rebars in a viaduct in service through hardness measurements

Results in Engineering 13 (2022) 100331 Available online 7 January 2022 2590-1230/© 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Identification of the failure mode of corroding steel rebars in a viaduct in service through hardness measurements G. Ruiz-Men´ endez a , b , C. Andrade b , G. Carro-Sevillano a , C. Pe˜ na a , P. Adeva a , J. Medina a , R. Fern´ andez a , * a Department of Physical Metallurgy, Centro Nacional de Investigaciones Metalúrgicas (CENIM) C.S.I.C., Av. de Gregorio del Amo No. 8, E-28040, Madrid, Spain b International Centre for Numerical Methods in Engineering (CIMNE), Edifici C1 Campus Nord UPC C/ Gran Capit` a, S/N 08034, Barcelona, Spain ARTICLE INFO Keywords: Tilted fracture Rebar Integrity Ductility Corrosion ABSTRACT In the present work, rebars extracted from a reinforced concrete viaduct that had to be demolished has been studied because they broke in service with an unusual tilted pattern for the rebars without reduction in diameter. The structure was a bridge suffering alkali-silica reaction and reinforcement corrosion. The determination of rebar’s in service failure mode is very complex due, among other causes, to their high degree of degradation by corrosion. In this work it has been shown that by performing hardness measurements along a diameter in the cross section of the bars it is possible to determine their failure mode, tensile or fatigue. The results show a similar probability of in service fatigue or tensile plastic collapse failure. In both cases, a brittle fracture has occurred. The corrugations of the rebars generate a stress concentration that explains the observed tilted appearance of the fracture surface. The influence of different corrugation patterns, as for example B 500SD, in service fracture appearance should be studied to extend the conclusions reached in the present work. (172). 1. Introduction The construction of reinforced concrete bridges started in the 1850s and 1860s [1]. However, in spite of the knowledge accumulated from then serious structural failures or degradation levels of these bridges are still occurring today [2]. Recent reports describe that about a quarter of the reinforced concrete bridges in Canada and the USA are in poor condition and structurally deficient or obsolete [3]. Reinforced concrete bridges often deteriorate due to a combination of several factors as: ageing, materials and construction defects, exposure to aggressive environments, lack of ductility, and excessive loads [4] although two are the main causes of corrosion initiation. The penetration of chlorides through the pore network in marine areas or when chloride salts are used as deicer to avoid the ice in the roads. The other cause of corrosion is the penetration of carbon dioxide gas in the air through the not fully saturated pores which reacts with the alkaline substances leading into the formation of calcium carbonate that lower the concrete pH. Chloride and carbonation arrival to the steel surface makes the passive film to disappear and reinforcement corrosion starts. Moreover, the resulted rust expansion cracks the concrete cover and affects the residual load-bearing capacity of the concrete section [5]. In particular, the degradation of rebars associated with corrosion phenomena can compromise the structure to the point of deciding to demolish it for safety reasons [2]. The problem is usually very complex because bridges and viaducts are reinforced with multiple rebars and the correlation of corrosion among the rebars greatly influence their structural behavior [6]. However, the combined effect of non-uniform corrosion of multiple reinforcing bars has been barely considered in the literature [7]. The mechanical behavior of rebars used in bridges, mainly σ y and UTS, has been extensively studied under different stress states, tensile and fatigue, and different conditions associated with different corrosion levels [8–12]. Previous results show a sharp reduction in the UTS and ductility when the corrosion was in the range 8–25% [13]. Brittle failure of corroded reinforcement after laboratory fatigue tests have previously reported [14]. However, brittle failures of rebars have been identified only once as such in Spain [15] in a carbonated concrete due to cracking by thermal actions cycles in a wall without joints and in Japan [16] in bridges suffering also alkali-silica reaction. The case of steel failure considered here is also produced in a concrete affected by alkali-aggregate reaction and it is very important the identification of the mechanism leading to the rupture in service, as other bridges were made with the same aggregates in the same times. * Corresponding author. E-mail address: [email protected] (R. Fern´ andez). Contents lists available at ScienceDirect Results in Engineering journal homepage: www.sciencedirect.com/journal/results-in-engineering https://doi.org/10.1016/j.rineng.2022.100331 Received 4 October 2021; Received in revised form 22 December 2021; Accepted 2 January 2022 Results in Engineering 13 (2022) 100331 2 In previous works, the effect of corrosion degree on the stress-strain curve of rebars extracted from the demolition of reinforced concrete viaducts has been studied [17]. From all these studies, a number of conclusions have been drawn about the effect of corrosion on the mechanical properties of rebars. Firstly, when the corrosion of the rebar is uniform, it is found that the mechanical strength of the reinforcement decreases proportionally to the section loss [18] though, it is usually referred to the initial section and not to the actual one. If the corrosion degree is high and pitting occurs, they act as stress concentrators and produce a triaxial stress state that is assumed to embrittle the material [18]. The yield strength may show a slightly higher decrease than that of the UTS due to corrosion in rebars [9,10]. In fact, one of the possible effects of corrosion on the mechanical properties of rebars is the disappearance of the apparent yield point in the tensile curve. In addition, it is observed in general that corrosion of rebars drastically decreases the deformation under maximum load [9,10]. Ductility is usually defined by the ratio of maximum stress to yield strength (UTS/ σ y ) or by the strain under maximum load, ε max . The ductility of reinforced concrete structures is highly dependent on the ductility of the reinforcement. In particular, a ductile structure allows the bearing capacity of the elements to be redistributed. For this reason, the effect of corrosion on the variation of ductility in rebars is of great importance [18,19]. The loss in ductility due to corrosion can be attributed to the presence of pits which reduce the remaining section and increase the stress in the pit tip, making a premature local plastic deformation. Moreover, this loss of ductility might be as well due to the embrittlement of the steel due to hydrogen uptake. It is known that corrosion acidifies the corroding interface which favors the reduction of protons as cathodic reaction releasing hydrogen [20]. Sometimes, the reduction in ductility is such that the 5% strain usually considered as the value of high ductility in rebars is not reached [18]. In Ref. [9] it is concluded that, although the reduction in mechanical properties is not proportional to the degree of corrosion, the decrease in ductility for high degrees of corrosion is proportional to the rebar diameter. The presence of pitting at the rebar surface also has a great influence on the degradation of its properties, e.g. their fatigue life [9,21]. However, the fatigue influence on the rebar-concrete interface is difficult to determine. The bond stress-slip curves in fatigue testing is similar to those of specimens subjected to monotonic loading [22]. The decrease in properties is more pronounced in the case of inhomogeneous rebars such as those manufactured with the TEMPCORE process. Among these rebars, the B 500 S ones have a high ductility and are one of the most commonly used in many construction works in Spain. The understanding of the influence of corrosion of their mechanical behavior and ductility under an applied stress needs to combine both analytical [8–12] and finite element models, FEM, [10,23,24]. The FEM modelling is fundamental to explain the influence of bar geometry, including ribs and corrugations distribution along the bar on their mechanical behavior. However, during in-service conditions of a viaduct, rebars suffer a complex stress state that highly difficult, when combined to environmental degradation, to understand source of their final breakage tensile or fatigue. Moreover, the assessment of the service life of reinforced concrete structures is difficult because many parameters have been also changed along with the corrosion of steel reinforcements [25]. In the present paper are reported experimental and numerical results on reinforcements corroded and broken in a bridge near Madrid which had to be demolished due to its risky situation when the degradation was detected with the occasion of a maintenance inspection. It is out of the scope of the work a detailed description of the whole set of degradations mechanisms that were observed in the bridge. The bridge was demolished due to the number of rebar failures found and the expansion process suffering. The bridge deck was composed of spans that some were reinforced and other post-tensioned. Both reinforcements were corroding, but this work is restricted to the study of the failure mode of the rebars for reinforcing. This is because the unusual mode of failure of these bars, many of them at 45◦and brittle in appearance. The main objective of this work has been to determine the origin of rebars in service failure. The mechanical properties in tension and fatigue have been compared to establish the causes and nature of the mechanical stresses that could have caused the failure of the bars in service. Furthermore, the use of FEM has allowed to establish a protocol to characterize the mode of fracture of rebars under in-service conditions. 2. Materials and methods The samples used in this work are rebars were collected from the reinforced concrete viaduct. Significant damage to the viaduct was found during a maintenance inspection that put the safety of the structure at risk. The bridge was built in the late 1990s, so these bars have been in service for about 23 years. Fig. 1 shows a view during an inspection of the viaduct showing the broken rebars due to the combined action of the corrosion process and the stress loads during its service life. This corrosion was observed in rebars located below the concrete cracks (although not only and not in all cracks) that run parallel to the edge of the deck, and that were attributed to the expansion induced by alkalisilica reaction. The chemical composition of the rebars was carried out using optical spark emission spectroscopy. The results shown in Table I meet with the chemical composition requirements for a corrugated steel type B 500S according to the UNE-EN 10080 Standard. Rebars of 12, 16 and 20 mm diameter and different preservation states has been used for the microstructural and mechanical characterization. Some of the rebars were unaffected by corrosion, others corroded and some fractured. The bars have been tested as they were collected from the viaduct; this state is designated as-received condition (as-R). These bars collected from the viaduct can be divided in two groups, unaffected, U as-R, and corrosion affected, C as-R. Some of the U as-R bars collected from the viaduct were kept in fresh water for 7, 30 and 111 days respectively (designated by 7 FW, 30 FW and 111 FW) or 7 days in fresh water+3.5% NaCl (designated by 7 NaCl). Some new rebars were used for comparison purposes. The microstructure of the rebars as well as the fracture surfaces of the mechanically tested samples were studied by optical and scanning electron microscopy (SEM) and EDX microanalysis. Longitudinal and transversal sections of the bars were characterized both in the as-R state and after mechanical testing, for which a conventional metallographic preparation consisting on mechanical polishing and 2% Nital etching was used. The characterisation of the species generated on the bars surface due to the exposition to air and fresh water was obtained by X-ray diffraction analysis. These measurements were carried out with a Bruker AXS D8 diffractometer equipped with a Co X-ray tube, Goebel mirror optics and a LynxEye Linear Position Sensitive Detector for ultra-fast XRD measurements. This type of radiation is specially suited for iron-rich samples to avoid the strong fluorescence arising from copper radiation, and to produce high resolution data. A current of 30 mA and a voltage of 40 kV were employed as tube settings. Operational conditions were selected to obtain XRD profiles of sufficient quality namely, optimal counting statistics, narrow peaks and detection of the small diffraction peaks of minor phases. Three samples were selected, one in the U as-R condition with a good appearance (uncorroded), the second one in the C as-R condition, and the third one a 111 FW sample. The mechanical characterization at room temperature was carried out by means of hardness tests, tensile tests at constant strain rate up to breakage and fatigue tests. The hardness tests were carried out with a load of 5 N. Hardness sweeps along the transverse section of the bars were made from the outside to the core of the rebars, making indentations every 0.4 mm, to determine the hardness gradient. In the tensile tests, unmachined bar G. Ruiz-Men´ endez et al. Results in Engineering 13 (2022) 100331 3 specimens with lengths between 300 and 700 mm between grips were used in such a way that strain rates of 5.0⋅10 −4 s −1 and 2.4⋅10 −6 s −1 were achieved to study the strain rate effect on the mechanical properties of the bars. The elongation of the bars was calculated from the variation in the position of the machine crosshead. In the fatigue tests, unmachined bar specimens of 330 mm length were subjected to tensile conditions with a sinusoidal variation of the tensile load up to fracture and R =0.17 and 0.39 applying a displacement of 0.3 mm amplitude between 5 and 30 and 13/30 KN respectively and a frequency of 30Hz. The stress distribution in the bars during a tensile test has been simulated by FEM using COMSOL Multiphysics 5.6 software. A bar length four times the diameter (height: 40 mm, diameter 10 mm) has been considered to eliminate edge effects in the stress/strain predictions. Lateral square section ribs have been included in the geometry along the length of the bar, and the corrugations have been represented by ellipsoids placed in the positions corresponding to the B500S designation, respecting the different orientation of the corrugations on both faces of the bar, Fig. 2a). The stress state was generated by displacing one of the transverse faces of the bar at a 10 −4 s −1 up to 5% of strain while the other remained fixed. Fig. 1. In the left an aspect of the broken reinforcements in coincidence with a crack, likely produced by the expansion by silica-reaction. In the right a detail of one of the broken rebars (top) and details of highly corroded areas showing pits and holes (bottom). Table 1 Chemical composition, weight %, of one rebar extracted from the viaduct. Element, mass % Rebar B 500 S C 0,109 ±0006 Si 0,17 Mn 0,49 ±0,01 P 0,024 ±0001 S 0,054 ±0002 Cr 0,086 ±0002 Ni 0,18 Mo 0,043 ±0002 Cu 0,470 ±0002 V <0,05 N 0,007 Fig. 2. Rebar geometry a) and elastoplastic model describing the mechanical behavior of the rebar b) used in the FEM simulation. G. Ruiz-Men´ endez et al. Results in Engineering 13 (2022) 100331 4 The mechanical behavior of the material in tension has been described by means of an elastoplastic model where the hardening has been described by means of a polynomial fitting, Fig. 2b), from σ y and up to the UTS reached at 10% strain. 3. Results 3.1. Microstructure The cross section of a U as-R bar shows three areas very well differentiated, the external region corresponds to martensite, the middle region is bainite and the inner one is ferrite +perlite, Fig. 3. These different microstructures appear as a consequence of the thermal treatment produced during the manufacturing process of the rebars. Furthermore, there is an important volume fraction of MnS inclusions in the bars, mainly located at grain boundaries, as shown in Fig. 4. These inclusions are elongated and they are broken by the rolling process producing a mean inclusion size of 2 μ m. Another important microstructural detail of the as-R rebars is the presence of an iron oxide or hydroxide layer in the external region of the bars, Fig. 5. The thickness of this layer is different in each bar extracted from the viaduct. As can be observed in the backscattered electron image of Fig. 5a), the rebar of 20 mm in diameter in the U as-R condition has a layer of 20 μ m approximately. The cracks probably come from the TEMPCORE manufacturing process of the rebars. For comparative purposes the metallographic cross section of a new rebar (not mechanically tested or embedded in concrete previously) is shown in Fig. 5b) where an iron oxide/hydroxide layer of approximately 5 μ m is also appreciated. In the 30 FW bar (also collected from the bridge +30 days in fresh water) the external surface is fully covered by an iron oxide/hydroxide layer of 3 μ m in thickness approximately, similar to that shown in Fig. 5a). In addition, a significant number of cracks can be identified in the outer surface of the rebar. These cracks are covered by oxide, and initially propagate along the grain boundary, some of them more than 30 μ m in length as shown in Fig. 5c). The compared X-ray diffraction spectra, Fig. 6, indicate that the 3 bars have the same corrosion products such as Fe 3 O 4 and FeO(OH) although the U as-R rebar exhibits the highest content of Fe 3 O 4 and FeO (OH). Furthermore, peaks of Fe (base material of rebars) and CaCO 3 as concrete residue are also present. The fact that high intensity peaks of Fe 3 O 4 and FeO(OH) are observed in the U as-R rebar may be related to the higher adhesion of these products to this bar while in the case C as-R and the 111 FW the products detach from the surface which explains also the observation of Fe peaks of higher intensity than in the U as-R bar. The detachment of the oxides and hydroxides justifies the progress of the degradation process in C as-R and the 111 FW bars. 3.2. Mechanical properties 3.2.1. Tensile behavior The main mechanical characteristics of the rebars studied are summarized on Table 2 φ is the rebar diameter, L 0 is the original length, σ y the yield strength, UTS the ultimate tensile strength, ε max the maximum strain, and ˙ ε the strain rate. The tensile tests showed a clear influence of corrosion on the mechanical properties, Fig. 7. Healthy bars with no apparent corrosion, U as-R, reached an apparent σ y of around 550 MPa and at least 10% of ε max . Some of the C as-R and FW rebars present a σ y higher than 550 MPa, Fig. 7, due to the section reduction suffered by corrosion that overestimates the stress when the original section is used for the calculations. FW rebars, presented reduced ductility in tensile tests in comparison to healthy rebars extracted from the viaduct. All the FW rebars exhibited an ε max below 9%. The rebar diameter has almost no influence on the stress-strain curve, Fig. 7. FW and NaCl treatments induce homogeneous corrosion which also reduce homogeneously the rebar section. This process can also produce small pits that reduce the sample ductility but the mechanical resistance is maintained, Fig. 7. However, the C as-R Fig. 3. a) Optical Image and secondary electron images of transversal section of U as-R rebar showing the different microstructures through the section b) outer area, c) inter-medium area and d) Inner area. G. Ruiz-Men´ endez et al. Results in Engineering 13 (2022) 100331 5 Fig. 4. Metallographical section of the rebar in the U as-R condition showing MnS inclusions. The chemical composition of particle 1 is summarized in the Table. Fig. 5. Electron backscattering images of cross section of rebars: a) Sample in a U as-R condition b) new rebar c) 30 FW rebar. G. Ruiz-Men´ endez et al. Results in Engineering 13 (2022) 100331 6 rebars showing a high corrosion degree, deep pits, present a low σ y , 7 FW in Fig. 7a) and C as-R in Fig. 7c), and very reduced ductility, reaching in some cases values below 2% of ε max . The fracture surface of the bars tested in tensile tests, some of them close to 45◦following the corrugations, exhibit a ductile appearance showing the expected dimples structure, Fig. 8a). The ductility loss due to water immersion is clearly observed in the fracture surfaces, Fig. 8b) and c). The fracture mode changes from completely ductile in the U as-R rebar towards cleavage fracture for FW samples. This trend increases as the amount of time spent in the water increases. Thus, the fracture surface of 7 FW bars shows areas of ductile fracture coexisting with cleavage areas, Fig. 8b). However, for 111 FW bars there are no ductile areas in the fracture surface Fig. 8c). These results are independent of rebar diameter. 3.2.2. Fatigue behavior The fatigue behavior of rebars also depends to a great extent on their corrosion degree, Table 3. As it is well known, under fatigue conditions, a crack is generated which, if the appropriate conditions are met, will grow until the part breaks under a given number of cycles. The surface state greatly influences the resistance to fatigue. Stress concentrators such as corrugations, ribs or even pre-existing cracks, as those shown on Fig. 5, will shorten the fatigue life of the rebar as they are defects prone to nucleate cracks. The Table 3 shows the results of the fatigue tests performed on the bars of 12 mm in diameter. The fatigue resistance of the rebars studied is well below one million of cycles. This relatively low number of fatigue cycles is attributes in the as-R bars because they have surface defects (cracks). The as-R bars exhibit the highest fatigue resistance but the number of cycles decreases as rebar conditions worsen from lightly corroded to 7FW bars, so in this last condition the fatigue resistance decrease up to 90%. It is important to note that all samples tested in fatigue break at approximately 90◦respect to the sample longitudinal axe. Moreover, the fracture surfaces exhibit the typical beach marks and striations indicating the propagation direction of the cracks by fatigue as Fig. 6. X-ray diffraction patterns compared. □ Fe(bcc), Δ Fe 3 O 4 , ◊ CaCO 3 , ○ FeO(OH). Table 2 Mechanical characteristics of the studied bars obtained by tensile tests. φ (mm) L 0 (mm) σ y (MPa) UTS (MPa) ε max (%) ˙ ε (s −1 ) Delivery condition 12 360 559.3 631.9 10.0 5⋅10 −4 U as-R 16 380 522.9 562.7 0.0075 5⋅10 −4 C as-R 16 700 512.1 641.8 7.3 2.4⋅10 −6 U as-R 20 380 505.3 516.9 0.0032 2.4⋅10 −6 C as-R 20 700 505.1 654.5 7.8 2.4⋅10 −6 U as-R 20 300 537.7 690.2 12.0 5⋅10 −4 U as-R 20 700 569.3 654.1 8.0 2.4⋅10 −6 7 FW 16 690 562.0 690.6 5.3 2.4⋅10 −6 7 FW 12 740 592.4 613.3 0.0024 2.4⋅10 −6 7 FW 12 800 531.0 667.6 6.0 2.4⋅10 −6 111 FW 20 690 523.3 660.1 8.8 2.4⋅10 −6 111 FW 16 700 541.5 684.1 6.4 2.4⋅10 −6 7 NaCl 20 700 534.9 690.1 8.6 2.4⋅10 −6 7 NaCl Fig. 7. True stress vs. true strain curves for a) 12 mm, b) 16 mm, c) 20 mm rebars. G. Ruiz-Men´ endez et al. Results in Engineering 13 (2022) 100331 7 shows Fig. 9. However, Fig. 10, shows the fracture surface of a sample lightly corroded and tested that is very different to the previous case. The fibrous morphology corresponds to the fast and unstable propagation of the crack, which is compatible with a low resistance to fatigue due to a premature failure. 3.2.3. Hardness gradient The study of the mechanical properties of the rebars was complemented with microhardness testing along the transverse section diameter of the bars Fig. 11. The hardness gradient accounts for the microstructural variation, Fig. 3, of the U as-R rebars associated at micro level with the manufacturing process or due to the tensile testing, if so. The hardness measurement along a U as-R rebar diameter is affected by the hardening that the sample undergone close to the fracture after a tensile test at a given strain rate, Fig. 12a). In this case, there is an almost constant hardness increment close to the fracture, 7 mm, around 2530HV, respect to a section far away, »100 mm, from the fracture. However, in the fatigue test of a rebar, the hardness profile is indistinguishable close, 7 mm, and far, »100 mm, from the fracture for half the section of the rebar (right part in Fig. 12b), which correspond to the fatigue crack progression. The final section of the rebar (left part in Fig. 12b) breaks by plastic collapse. The dashed lines indicate the limits of the hardened zone due to plastic deformation close to the fracture surface of the rebar. The hardness patterns found for samples tested in tensile or fatigue mode, Fig. 12, were also found in rebars collected from the viaduct, Fig. 13. The hardness difference close and far from the fracture surface allows classifying the fracture mode of the viaduct members analyzed. The dashed lines in Fig. 13 indicate the limit of the hardened zone. Of the five samples removed from the viaduct, two of them, Fig. 13a) and b) show a hardness difference close and far from the fracture surface in their core region. This hardness difference is due to the plastic deformation associated to tensile in-service conditions close to the fracture surface. One sample Fig. 13 c) shows only hardness difference in Fig. 8. Secondary electron images showing the fracture surface after tensile tested rebars: a) U as-R, b) 7 FW c) 111 FW. Table 3 Fatigue test conditions of 12 mm diameter rebars indicating the number of cycles at fracture, N cycles. Stress ratio (R) N Cycles Condition State 0.39 130.000 U as-R 0.39 600.000 U as-R 0.39 100.000 C as-R 0.39 58.000 7 FW 0.17 160.000 C as-R 0.17 150.000 7 FW G. Ruiz-Men´ endez et al. Results in Engineering 13 (2022) 100331 8 half of the section, located in the outer region of the rebar what is interpreted as the existence of crack propagation by fatigue during inservice conditions of this rebar. Moreover, in two other samples, there are parts of the hardness profile that are very similar along sample diameter, indicating that these bars have undergone also a fatigue process during in-service conditions, Fig. 13d) and e). The mode of fracture of the viaduct rebars should be also reflected in the degree of fracture surface tilting with respect to the longitudinal axes of the sample. However, their high degree of deterioration makes this characterization sometimes very difficult. Nevertheless, it can be said that the fracture surface tilt ranges from 30◦to 45◦with respect to the longitudinal axes in all samples analyzed Fig. 13. An inhomogeneous reduction of the rebar cross-section is also observed in all the samples except in Fig. 13b) and e). 3.3. FEM Fig. 15 shows the stress distribution in a longitudinal section where a high stress concentration in the contact zone between the bar and the corrugations. This stress concentration introduces a stress gradient of up to an order of magnitude in about 500 μ m around the contact zone of the corrugation with the bar. This stress concentration is not observed at the rib-surface intersection, Fig. 15. 4. Discussion In the present study, rebars have been collected from a viaduct that has undergone a degradation process that has compromised its integrity until failure. The behaviour of a reinforced concrete structure and the degradation of its properties depend on several factors that have been extensively studied. Among them, the aggregate-alkali reaction in concrete [26], the corrosion of rebars [8,9] and frost attack are the most common ones. In a real structure, the combination of processes and stresses in a group of rebars make difficult to establish the influence of each of these mechanisms in the degradation process of the structure. What is widely agreed is that the degradation process of a reinforced concrete structure such as the bridge under study begins with a process of cracking of the concrete. Cracking can be due to several causes [26]. Fig. 9. Fracture surface of U as-R bar broken after 600.000 cycles. Fig. 10. Fracture surface showing the zone of crack propagation for a C as-R bar broken after 160.000 cycles. Fig. 11. Hardness profiles along a U as-R rebar diameter for 12, 16 and 20 mm diameter rebars. G. Ruiz-Men´ endez et al. Results in Engineering 13 (2022) 100331 9 Among the most common ones are shrinkage of the structure, thermal gradients, alkali-silica reactions, rebars oxide expansion, and mechanical loads. Cracks provide preferential pathways for external agents such as water or chlorides in the form of salt to penetrate the structure [3,4]. The most relevant consequence of this cracking is that the rebars exposed to the environment suffer typical corrosion by pitting progresses steadily by anodic dissolution in the bottom of the pit until the whole diameter is lost by the electrochemical dissolution. The type of failure found in rebars (Fig. 1) indicates a stress assisted fracture whose origin was uncertain. A similar unexpected failure type was found in bridges suffering alkali-silica reaction; AAR, in Japan by Dr. Miyagawa [16] who shared his experience with one of the authors of this paper. In the Japanese bridges failures, the fracture is tilted 90◦with respect to the rebar axis and was found in the bended zones of the stirrups. In the present case, the fracture surface is tilted between 30◦ and 45◦with respect to the rebar axis, Fig. 14. This pattern found in the five rebars broken in service analyzed from the viaduct can be explained by the stress gradient produced in the contact between the corrugation and the cylindrical part of the rebar. Neither the real geometry of the rebars nor the hardening during plastic deformation is usually considered in FEM models [23] except in recent works [24]. Those made of TEMPCORE carbon steel show a relatively high value of the hardening exponent, n, which decreases as the strain increases. As it is seen in Fig. 15, stress values that of σ y are found at the corrugation-cylinder interface while the corrugation center remains free of stress. This stress gradient is located in around 0.5 mm area close to the corrugation-cylinder intersection and act as a notch that facilitates crack progression in the range 30–45◦from the rebar axis. Basic research has been undertaken by the authors to try to understand deeper the effects of corrosion into the steel reinforcement behavior since the long-term impact of the cracks in the concrete on the corrosion propagation phase has not been clarified and is still under debate [10]. Current codes establish a maximum allowable crack width on the concrete between 0.2 and 0.4 mm [27,28] depending on the aggressiveness of the environment, in an attempt to minimise the penetration of environmental elements that degrade the rebar structure. The microstructural characterization of the in service fractured rebars shows, as described in the results section, the typical microstructure of corrugated steel type B 500S. Moreover, there is an oxide/hydroxide layer of Fe (Fig. 6) that covers their entire external surface that must be formed during production (mill scale) or the storage. In this layer, which is of different thickness depending on the history of the rebar, a significant number of cracks have been identified that initially propagate along the grain boundary, as shown in Fig. 5. The origin of these cracks must be associated with the rebar manufacturing process. Likewise, and as has been verified in the laboratory, after prolonged exposure of the rebars in water, the layer detaches, which implies a loss of section of the material but at the same time the action of the water causes both the layer and cracks to continue to grow inward. This phenomenon of degradation suffered by the rebars maintained in fresh water (free of chlorides) in the laboratory is similar to that observed in the real case of rebars in service, as has been shown in the micrographs of Fig. 1, when the concrete cracked. The reduction of mechanical resistance and ductility of a rebar under tensile conditions strongly depends on its corrosion degree [29]. For low and medium degrees of corrosion, up to 40%, ductility is reduced proportionally to the corrosion degree, but mechanical resistance is maintained. From 50% of corrosion degree, there is a progressive reduction in mechanical resistance and ductility [29,30]. In particular circular-cross section pits with large section reduction (~50%) greatly reduce the mechanical strength of rebars [29]. These results have been also found in the present work. The residual ductility is complicated to predict because it depends on the inhomogeneous distribution of residual sections associated with pitting [29]. For this prediction, a consideration of the weakest section provides a better approximation of residual ductility than the mass loss. The residual section is of interest for electrochemical dissolution as it gives the amount of residual metal, but for mechanisms in which the mechanical load is contributing, what is important is the maximum pit depth and the stress concentration depending on the geometry of such pit. In present case, the wide cracks generated by the AAR expansion have allowed the localization of attack in some narrow regions of the bars leading to some of them to a brittle failure. In the area of the rebars just below the crack in the concrete, a very aggressive corrosion attack can occur in the form of deep pitting and a heavy section loss. By the time the crack occurs and starts to propagate through the rebar, the localized section under the crack in the concrete is so degraded by corrosion in the pit residual section that the fracture is brittle, elongation <2% with high hardening during plastic deformation [29,30]. In other words, in either case, tensile or fatigue, the failure of the viaduct rebars under investigation, exposed to the aggressive environment, has occurred under brittle condition. In other structures where rebar corrosion and mechanical overloads have been combined, brittle failure of carbon steel rebars associated with the SCC mode has also been found despite this phenomenon is not very common [31]. Fatigue can be considered to start at pitting [32]. Once a pitting has been generated, a transition to crack generation occurs through a combination of mechanical and electrochemical processes, ending with crack propagation through the rebar section [32]. Corrosive media accelerate crack propagation by fatigue [32]. Regarding the microhardness measurements carried out on bars Fig. 12. Hardness gradient along a rebar diameter of a sample being tested in a) tensile mode and b) fatigue mode. Red color indicates sample position far, »100 mm, from fracture surface and black color indicates sample position close, 7 mm, from fracture. G. Ruiz-Men´ endez et al.