Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Evaluation of corrosion level of naturally corroded bars using different cleaning methods, computed tomography, and 3D optical scanning

Evaluation of corrosion level of naturally corroded bars using different cleaning methods,... Materials and Structures (2018) 51:78 https://doi.org/10.1617/s11527-018-1206-z ORIGINAL ARTICLE Evaluation of corrosion level of naturally corroded bars using different cleaning methods, computed tomography, and 3D optical scanning . . Ignasi Fernandez Karin Lundgren Kamyab Zandi Received: 8 August 2017 / Accepted: 6 June 2018 The Author(s) 2018 Abstract Reliable methods are necessary to assess that occurs in existing impaired reinforced concrete the corrosion level to establish links between struc- structures. Chloride ion (Cl ) and carbon dioxide tural performance and reinforcement corrosion in (CO ) penetration from the structure’s surrounding concrete structures. Hence, in this study, a set of environment leads to the destabilization of passivity naturally corroded bars were subjected to metallic conditions provided by the surrounding concrete to the brushing, acid immersion, and sandblasting for rust steel bar. This destroys the steel protective layer, removal. Additionally, 3D optical, CT scanning, and termed as the passive layer, and subsequently initiates weight loss measurements were used to evaluate the corrosion of the steel. Hence, volumetric expansion of levels of corrosion. The results indicate that sand- corrosion products and cross-section reduction of the blasting is an optimal cleaning method. Weight loss steel bar leads to damages in the structure. Rust measurements are sufficient when detailed informa- expansion inside concrete generates significant inter- tion about corrosion is not required, and 3D scanning nal pressure that induce splitting stresses in the is preferred if information on corrosion variation along concrete along the corroded reinforcement and harm the bar is needed. the surrounding concrete. Splitting stresses are not well tolerated by concrete and result in cracking and Keywords Corrosion level measurement  Cleaning eventually spalling of the concrete cover. The corro- methods  3D optical scanning  Computed sion rate may increase when the reinforcement tomography scanning  Sandblasting  Acid cleaning becomes more exposed, thus facilitating the deterio- ration processes. Corrosion of reinforcement in concrete is examined widely by several previous studies. For example, 1 Introduction several previous studies [1–4] addressed the structural effects of corrosion. Various studies investigated local Corrosion of reinforced steel bars continues to be one aspects that addressed the effects of corrosion on bond of the most frequent and significant type of damage behaviour [5–7], change in mechanical properties [8–11], and other corrosion related phenomena [12–15]. However, most existing studies were con- I. Fernandez (&)  K. Lundgren  K. Zandi ducted under accelerated corrosion conditions. There Division of Structural Engineering, Department of Civil is a paucity of research examining natural corrosion and Environmental Engineering, Chalmers University of circumstances due to various difficulties [16–19]. Technology, 412 96 Go¨teborg, Sweden e-mail: [email protected] 78 Page 2 of 13 Materials and Structures (2018) 51:78 Therefore, there is a growing demand to continue different techniques. The study involved examining experimenting with naturally corroded specimens to and comparing the following three different cleaning further validate, and even extend existing knowledge. methods to remove attached mortar and rust from the However, irrespective of the origin of corrosion, all aforementioned specimens: (1) mechanical wire the fore-mentioned studies are based on the measure- bristle brushing, (2) sandblasting, and (3) chemical ment of the actual corrosion level with respect to the cleaning. This was followed by applying and compar- affected steel reinforcement. Thus, it is extremely ing the following three measurement techniques to important to obtain consistent methods to assess the evaluate the corrosion level: (1) weight loss, (2) 3D corrosion of the steel bar and to examine its detailed scanning, and (3) three-dimensional micro-computed impact on the steel surface. The methods should allow tomography (CT scanning). The study presents the acquisition of precise information detailing pit char- scope, applicability, and accuracy of each cleaning acteristics and a faster and detailed measurement of method along with its combination with the different the level of corrosion for a specific bar length. measurement techniques. Furthermore, it details rec- Simultaneously, the availability of reliable and more ommendations to obtain reliable levels of corrosion detailed information about the corrosion distribution for the corroded steel bars. will allow the establishment of increasingly trustwor- thy links between these measurements with respect to 2 Experimental programme the structural response. Typically, existing studies have based the assessment of the corrosion level on the ASTM G1—Standard Practice for Preparing, Clean- The specimens presented in the study are part of a ing, and Evaluating Corrosion Test Specimens [20]. A larger experimental campaign conducted at Chalmers few studies attempted to incorporate new techniques University of Technology in which specimens from and technologies from other fields that theoretically edge beams of the Stallbacka Bridge in Sweden were provided a better description of the corrosion level used. The specimens were taken from the bridge when along a steel bar [9, 21–26]. Nevertheless, all the it was under repair following approximately 30 years aforementioned techniques are yet strongly dependent of exposure to different natural deterioration phenom- on the cleaning methods performed on the steel bars. A ena such as corrosion induced by chlorides from de- recent study by Tahershamsi et al. [27] pointed out icing salts. In an earlier study, beams were tested in significant discrepancies between obtained results four-point suspended bending tests to obtain anchor- using three-dimensional optical measurement (3D age failure [16, 17]. scanning) and weight loss measurement following Table 1 presents an overview of all specimens metallic brush cleaning of naturally corroded steel included in this study on which different techniques bars. There is a paucity of similar studies that relate were applied to measure the corrosion level. Each row and compare different cleaning methods and corrosion indicates a specific enforced cleaning method includ- level measurement techniques. An exception is the ing acid immersion, sandblasting, metallic brush, and study conducted by Tang et al. [24] that presented a no-cleaned before scanning and later cleaned using direct comparison between gravimetric and 3D scan- sandblasting. The various columns describe the num- ning to assess the level of corrosion; the results ber of specimens and the method used for the indicated a very reasonable agreement; artificially corrosion level evaluation for each group, specifically, corroded bars and sandblasting were used as cleaning weight loss, 3D scanning, and CT scanning. methods. Hence, the aim of the present study includes 2.1 Reinforcement steel bars evaluating the scope and applicability of common cleaning methods used in existing studies as well as Tensile reinforcement steel bars were carefully different measurement techniques to evaluate the extracted from a non-critical section of the beams. corrosion level of naturally corroded bars. Conse- The beams contained to two different 16-mm diameter quently, a set of corroded steel bars extracted from a bar types; both types of steel class Ks60 [28]. Each real bridge that is more than 30 years old was cleaned, type involved different rib patterns, namely skewed and their levels of corrosion were assessed utilising Materials and Structures (2018) 51:78 Page 3 of 13 78 Table 1 Overview of specimens, cleaning methods, and corrosion level evaluation methods Cleaning method Specimens (bars) Corrosion level method Weight loss 3D scanning CT scanning Acid 7 Yes Yes Yes Sandblasting 7 Yes Yes Yes No-cleaned 7 Yes No Yes Metallic brush 17 Yes Yes No Specimens were first scanned and later cleaned by using sandblasting Specimens presented in a previous study [27] and straight ribs. Only straight rib pattern bars were 2.1.2 Chemical composition considered in this study. The chemical composition of the extracted steel was investigated by using scanning electron microscopy 2.1.1 Geometrical description of the bars (SEM). The values presented in Table 3 correspond to the average values obtained for each component in The steel bars were cut in lengths of 300 mm. This length was selected as a compromise between differ- different scanning points throughout the bar cross- section. Iron makes up to 100% of the steel ent requirements with respect to performed tasks. This included the maximum length that is allowed for composition. subsequent CT scanning by using reasonable resolu- tions, the minimum length necessary for mechanical 2.2 Removal of corrosion products properties characterization by means of tensile tests, and assessment of the corrosion level and a reasonable length for representativeness. Some studies have shown that pits are distributed stochas- The extracted steel was cut into suitable parts with a length of 300 mm, and the bars were cleaned by using tically along the bar length in [29]. Hence the choice of the specimen length should not have a relevant impact the following three most common methods found in literature: metallic bristle brushing, acid immersion, in the presented results; moreover, considering that the bars where extracted from the same structure, from and sandblasting. Metallic bristle brushing is the most commonly used cleaning method [27, 30, 31]; this is zones that presented clear signs of deterioration, similar damage levels, and which had been exposed mainly because it entails a low number of require- to similar aggressive environment, they are considered ments for its use. Conversely, immersion in an acid to be comparable. solution is less common [7, 15, 32, 33]. As required by The following bar specifications were measured the specified standard [20], the weight loss was measured after several cleaning cycles. Additionally, from the uncorroded samples by using a Vernier Calliper with a resolution of 10 lm, as shown in Fig. 1 some of the specimens were subject to two different scanning techniques, namely 3D and CT scanning; this and Table 2. The values given correspond to the measurement averages. allowed a description of the outer surface of the Fig. 1 Geometry of ribbed reinforcement bars 78 Page 4 of 13 Materials and Structures (2018) 51:78 Table 2 Measured parameters and standard deviation to describe the geometry of the ribbed reinforcement bar (in mm) Steel class Nominal diameter D r a r h r h r l r l rh () 1 2 1 2 Ks60 Ø16 15.72 0.03 1.91 0.02 1.23 0.10 1.23 0.10 9.00 0.36 2.27 0.10 90 Table 3 Chemical composition of the steel bars C O Si Mn Cr Ni Cu Ks60 2.84 4.63 0.22 1.08 0.19 0.17 0.51 Each parameter is given in % corroded bar and thereby permitted an evaluation of the corrosion level. Specifically, 3D and CT scanning techniques that are used widely in fields, such as industrial engineering or medicine, are not commonly Number of Cleaning Cycles used in civil engineering. Both techniques are rela- tively new, especially with respect to their application Fig. 2 Standard recommendations for cleaning corroded steel to deteriorated structures and their different structural bars [20] elements. Thus, only a limited number of studies supply, and the engine was in charge of continu- explored these methods and their applications [9, 22, 24, 30, 34, 35]. Furthermore, a group of ously rotating the brush at the same speed. The recommendation [20] did not specify any cycle or specimens were scanned with the CT scanning tech- nique prior to cleaning to assess whether the CT exposition time, and thus each cycle was not systematically measured. Instead, each cycle was scanning technique was sufficiently effective to obtain a corrosion level without cleaning. This would involve distinguished when perceptible changes on the surface due to the rust removing were observed time efficiency as well as a method to avoid possible after the bar was swept from end to end. According induced inaccuracies of the cleaning methods. to this criterion, the necessary time for each cycle was approximately in the range of 10 min to 2.2.1 Cleaning methods and weight loss measurement 15 min based on the actual level of corrosion. • Sandblasting was performed in an individual The same procedure was followed for each of the cleaning methods used in this study. Reiterated cabinet designed for the purpose. The sand was blasted at 5–7 bars of pressure. The sand employed cleaning cycles were applied to each specimen until the mass loss was lower than 0.2% of the previous for the rust removal corresponded to silica sand. Similarly, the recommendation followed for measurement. Consequently, it was possible to clearly distinguish two different slopes as shown in Fig. 2. metallic brush [20] did not specify any cycle or exposition time. Hence, the same described crite- This necking point was not quantified in the ASTM rion was used, and the bar was sandblasted from recommendations [20] although this is used in the end to end until perceptible changes on the surface present study based on the harshness of the cleaning were observed. Accordingly, the necessary appli- methods as detailed in subsequent sections. cation time for each cycle was approximately less A short description of each cleaning method than 5 min. performed on the steel bars is as follows: • A wide range of chemical cleaning based methods • Mechanical brush was performed by utilising a are found in the ASTM recommendation [20]. The rotational metallic wire bristle brush. It was present study involved using the chemical cleaning attached to an engine plugged to an electricity method by repeated immersion of the bars in a Mass Loss Materials and Structures (2018) 51:78 Page 5 of 13 78 solution of hydrochloric acid and utropine (500 ml/l of solution hydrochloric acid, sp gr 1.19, 3.5 gr/l of solution hexamethytene tetramine and regent water) in cycles that approximately lasted for 10 min. The selection choice was motivated by both safety rules and practical reasons since the other methods used carcinogenic products, high temperature environment, or very long exposition times. The mass loss was measured after every cycle and the cycles were repeated until the aforementioned threshold was reached for each method. Fig. 3 3D surface generated from the 3D scanning Additionally, in order to obtain the scope of harshness of each cleaning method, uncorroded steel (X, Y, Z), was established and referenced to the end of bars were cleaned, and the loss of sound steel was the bar. The high resolution of the surface mesh assessed with respect to each cleaning method. Acid allowed for a sufficiently detailed description of the and sandblasting cleaning yielded similar levels of geometry of the bar to obtain information on features sound steel removal that corresponded to approxi- including pit depth and length, pit distribution, and mately 0.2% of the initial weight. In contrast, metallic loss of cross-sectional area along the bar length. bristle brush had no significant impact in the non- corroded steel removing. Subsequently, the value of 2.2.3 Micro-computed tomography technique 0.2% weight loss between cycles was used in the present study as the threshold beyond which the bar Sound steel was determined using three-dimensional was considered as fully cleaned. In addition to this micro-computed tomography (CT scanning), which is threshold, it was also visually confirmed that the bars used widely in image diagnostic medicine and is a looked clean. promising technology that is also applicable to other fields such as civil engineering. Previous studies 2.2.2 3D optical measurement technique [22, 37, 38] used this technology to assess the effect of corrosion in concrete and mortar phases although The 3D scanning of the corroded bars was performed very few studies focused on obtaining a comprehen- by means of optical measurement. An industrial stereo sive surface of corroded reinforcing bars [39]. device with two cameras of 5 Megapixels was used. Corroded steel bars were placed in a Metrotom The maximum accuracy provided by the cameras machine that projected X-Ray beams. As the steel bars corresponded to 2 lm, which allows the description of were constituted by two clearly different materials, i.e. imperfections over the steel bar surface due to sound steel and corrosion products, different specific corrosion. A correction of the measurement inaccura- amounts of the X-Ray beam were absorbed by each cies, such as polygon spikes removing and mesh holes one. It is possible to obtain a 2D image of the object by closing, and data treatment was performed using the collecting the remaining transmitted intensity through TM post processing software Geomagic Wrap 2014 the machine detector. The required intensity of the [36]. X-Ray beam was adjusted based on the material The outcome of the optical measurement corre- density, i.e., it must be sufficiently strong to penetrate sponded to a very fine mesh of triangular surface the full thicknesses of the different components. The polygons connected by nodes, see Fig. 3. The average process is repeated several times conveyed to the size of the element corresponded to 0.014 mm with a rotation of the object within specific angles, and side length of approximately 0.15 mm. The number of subsequently post-processing of the obtained images triangular elements in each scanning was between is performed to build a full 3D interpretation of the 2,000,000 elements and 3,000,000 elements depend- body. A larger difference between the densities of the ing on the corrosion level. A global coordinate system, materials that conform the object increases the degree 78 Page 6 of 13 Materials and Structures (2018) 51:78 of ease and clarity of the result. The actual resolution A complete description of the detailed steps from of the flatbed-detector corresponded to a frame of the initial 3D polygonal mesh to the graph that shows 1024 9 1024 pixels or a voxel of 174 lm for 3D-CT the cross-section variation is provided in a previous scanning. Accordingly, in order to increase the final study [27]. resolution of the final surface mesh of the steel, two scans together covering the total volume were per- formed which allowed the maximum accuracy by the 3 Results and discussion equipment used. Subsequently, the two steel outcome surfaces were digitally stitched to form a unique full 3.1 Cleaning methods volume that contained all the defining points. The technology possesses the potential to describe the Figure 4a shows the average mass loss of all bars outer surface of the corroded steel bar in detail without cleaned with each cleaning method. The same pro- cleaning the corrosion products in advance as they posed methodology [20] was followed for each present very dissimilar densities. cleaning method with reiterated cleaning cycles until The same type of surface mesh as that described for the mass loss was lower than 0.2% of the last measured the 3D scanning technique that was previously weight. However, large differences were observed in presented was obtained by means of the CT scanning. the final measurements between the three selected options as shown in Fig. 4a. The sandblasted speci- mens exhibited the largest measured corrosion level, 2.2.4 Evaluation of the level of corrosion using systematically both for all the specimens, and for the scanning techniques average level of corrosion in each group. This was followed by the acid cleaned specimens. Finally, the A method that was developed in a previous study [27] mechanically brushed specimens exhibited the small- was used to determine the corrosion level variation est measured corrosion level. The methods were along a bar based on the scanning measurements: expected to result in similar levels of the average • The resulting outer geometry based on the scan- mass loss, since the bars were randomly obtained from ning output is postprocessed, cleaned and repaired. the tested beams, from areas which had been exposed • The coordinates of the nodes are transformed into a to similar aggressive environment and subjected to similar damage (all the specimens presented clear polar coordinate system, and a contour plot of the corrosion penetration along the bar surface is signs of deterioration). Thus, the fore-mentioned created from the new points, which allows to discrepancies indicate that the proposed recommen- visually observe the corrosion pits along the bar dations did not provide information to a user with surface respect to the degree of cleanness of the bar at the end • The cross-sectional area at specific sections uni- of the process, as opposed to whether the performed formly separated is obtained by integration of the cleaning method reached its maximum cleaning coordinates. capacity for a set of specific conditions including steel • The cross-sectional area along the bar is calculated type, initial amount of mortar, and corrosion products and plotted. The bar corresponded to a ribbed bar, attached or cleaning agent (such as chemical, or and thus the cross-sectional area varies along the brush). This assertion is backed by Fig. 4b, in which the average number of cycles applied to each specimen bar. The effect of the ribs is eliminated by a smoothing fit that used cubic splines; this results in to reach the necking point together with its standard another curve. The uncorroded zone/s of the bar is/ deviation is shown. It was observed that the number of are identified, and the average cross-sectional area cycles remained constant regardless the final level of is used as a reference. corrosion. • The normalized cross-sectional reduction is plot- As shown in Fig. 4b, the number of cycles neces- ted by dividing the measured cross-sectional area sary to obtain a weight loss difference lower than 0.2% with respect to the uncorroded area. within cycles (which was defined by the authors as a • Finally, the corrosion level variation in percentage reference value based on the degree of harshness of the is plotted along the bar. cleaning methods) ranged between 4 cycles and 7 Materials and Structures (2018) 51:78 Page 7 of 13 78 (a) 16% Brush Sand. 14% Acid 12% 10% 8% 6% 4% 2% 4.14 3.29 6.43 0% Brush Sand. Acid Brush Sand. Acid Number of cyles (b) 0.0% 0.2% 0.4% 0.6% 0.8% 1.0% Brush 1.2% Sandblasting Acid 1.4% Fig. 4 a Average mass loss for the different cleaning methods. b Examples of cleaning cycles for each cleaning method cycles according to the method for representative solution. Thus, the individual effective time with specimens. Sandblasting reached its highest cleaning respect to the bar was considered lower and approx- capacity in fewest number of cycles whereas acid imately corresponded to 15–20 min. The results immersion needed the highest number of cycles to revealed that sandblasting exhibited a very high speed reach the defined threshold. The cleaning speed of the and only required less than 20 min of application per metallic brush method ranged in between that of single bar to reach the same cleaning capacity. Finally, sandblasting and acid immersion on an average, metallic brush required the longest cleaning time per although the scatter observed among the different bar and the average corresponded to approximately bars increased in contrast to those observed in the 40 min based on the actual corrosion level. other methods. This scatter could indicate higher However, direct conclusions to assess the actual dependence of the brush cleaning method on the actual accuracy of the method could not be extracted from corrosion level than the others. the results after cleaning since the results only With respect to the cleaning time necessary to reach expressed a relative comparison with respect to the the proposed necking point, the results indicated that initial weight and did not provide any indication as to the maximum time to reach the proposed necking whether or not the bar was completely cleaned. point corresponded to that of acid solution that Nevertheless, the maximum cleaning capacity was required more than 70 min of immersion in addition definitely reached for each method. Figure 5 shows a to the drying and weighting time of the bars within few bars after rust removal. cycles. However, acid cleaning allowed multiple bars Important differences were observed in the final to be simultaneously cleaned, i.e., several bars could shape of the surface after cleaning. The finishes for the be placed in the same recipient containing the acid different methods corresponded to shiny and smooth Weight loss differrence [%] Average level of corrosion Average nr. of cycles 78 Page 8 of 13 Materials and Structures (2018) 51:78 Fig. 5 Cleaned bars. From top to bottom, the figure shows metallic brush, sandblasting, and acid immersion Remaining corrosion products for metallic brush, matt and granulated for sandblast- bar were only unveiled after a significant period of ing, and dark brown for acid cleaning. All the methods time elapsed as shown in Fig. 5. This was potentially showed corrosion pits along the bar irrespective of the due to the drying out of the acid solution and the finish for each method. However, the depiction of pits occurrence of slight corrosion in the sound steel that was significantly more clear after sandblasting when changed the surface shade. compared to the other methods, as shown in Figs. 5 Conversely, sandblasted specimens always exhib- and 6. Subsequently, a more comprehensive impact of ited remaining rust clusters that were distributed along the corrosion on the bar surface was observed in which the bar during the cleaning cycles. Specifically, the larger and deeper pits were found in the sandblasted mentioned clusters were also used as additional visual bars as shown in Fig. 6. This was not obtained when criteria to define the exposition time in each cycle. the bars were extracted or when the bars were cleaned Nevertheless, there was a significantly lower presence with the two other methods. of rust clusters at the necking point in the case for It should be noted that neither metallic brush nor sandblasting when compared with acid cleaning as acid immersion specimens portrayed any remaining clearly shown in Fig. 5. The remaining corrosion corrosion products after the bars were cleaned. The products on the bar surface were a result of the shiny and smooth surfaces due to the metallic brush compromise between the corrosion products and covered the remaining corrosion products attached to sound steel removed in each cycle, which must always the bar. However, few pieces jumped off and the bars be lower than the method harshness. Corrosion exhibited the aforementioned products when the bars products, in addition to sound steel must be removed were tested under a tensile load [27]. With respect to in the cycle in order to maintain the accuracy of the acid immersion, the surface after cleaning exhibited a measured level of corrosion. homogeneous dark brown finish and did not allow the detection of the presence of remaining corrosion 3.2 Assessment of the corrosion level, products. Clusters of corrosion products indicating the and a comparison between weight loss, CT, presence of a significant amount of rust attached to the and 3D scanning Table 4 lists the levels of corrosion obtained with the different techniques that are applied on each bar. Each value represents the average corrosion level along the specimen length, which corresponds to 300 mm. Hence, it was expected that presented values evaluated with the different techniques would be in agreement. However, major differences were observed and are discussed in the following section. Figure 7 shows an overall description of the obtained results by means of depicting the average measurements. As shown in the figure, 3D scanning Fig. 6 Pits unveiled after sandblasting Materials and Structures (2018) 51:78 Page 9 of 13 78 Table 4 Corrosion level results Cleaning method Specimen Corrosion level (%) Weight loss Average 3D scanning Average CT scanning Average Acid CA-1 12.1 8.78 10.3 6.96 9.9 6.66 CA-2 13.3 10.8 10.7 CA-3 2.0 1.5 0.8 CA-4 4.5 2.8 2.6 CA-5 16.2 13.6 13.5 CA-6 4.4 3.2 2.8 CA-7 9.0 6.5 6.3 Sandblasting CA-8 17.1 11.05 16.5 10.64 15.0 9.29 CA-9 11.1 10.6 9.3 CA-10 1.7 1.8 0.6 CA-11 13.9 13.4 12.1 CA-12 14.2 13.9 12.7 CA-13 11.9 11.3 9.5 CA-14 7.4 7.0 5.8 Not-cleaned CA-15 12.2 8.7 – – 3.9 2.085 CA-16 8.4 – 0.8 CA-17 3.3 – 0.1 CA-18 7.4 – 2.1 CA-19 14.3 – 4.3 CA-20 9.5 – 2.2 CA-21 5.8 – 1.2 Metallic brush CA-22 7.0 4.32 4.6 2.74 – – CA-23 2.9 2.5 – CA-24 1.6 0.8 – CA-25 0.0 0.1 – CA-26 6.7 3.7 – CA-27 2.3 2.0 – CA-28 9.0 4.3 – CA-29 6.3 3.5 – CA-30 2.3 1.9 – CA-31 5.7 3.6 – CA-32 7.8 6.2 – CA-33 4.0 2.5 – CA-34 2.6 1.4 – CA-35 1.5 0.9 – CA-36 1.4 0.7 – CA-37 8.6 5.0 – CA-38 3.7 2.8 – Specimens that were scanned prior to cleaning and subsequently cleaned using sandblasting Specimens presented in a previous study [27] resulted in higher values in all cases when compared to measurement. Average 3D scanning and weight loss those obtained in CT scanning. However, the obtained values described very good agreement with respect to values were consistently below the weight loss sandblasting cleaning method, and the difference 78 Page 10 of 13 Materials and Structures (2018) 51:78 12% brush. This is potentially related to the efficiency of Weight loss the cleaning method. 3D scanning 10% Generally, sandblasting cleaning yielded the best CT scanning agreement irrespective of the actual corrosion level as 8% shown in Fig. 8b by the ratio of 3D scanning to weight 6% loss measurement. Acid cleaning exhibited a slightly better agreement when the corrosion level increased, 4% and this indicated that the accuracy of the cleaning method was less relevant for increases in the corrosion 2% level. Conversely, metallic brush cleaning displayed large scatter and a clear trend was not observed. 0% Brush Sand. Acid No-Clean The same comparison for the CT scanning tech- nique is presented in Fig. 9. The CT scanning Fig. 7 Average corrosion level for each presented method measurements resulted in smaller corrosion levels when compared with the weight loss measurements. between the average weight loss and that in CT The same behaviour was observed for all the cleaning scanning was higher for all the cleaning methods. methods as well as for the bars scanned prior to A direct comparison between single weight loss and cleaning. In a manner, similar to the 3D scanning 3D scanning corrosion levels for the three proposed results, the best agreement between CT scanning and cleaning methods is shown in Fig. 8a. As shown, weight measurements was exhibited by the sand- gravimetric weight loss typically resulted in higher blasted specimens while the bars scanned prior to corrosion levels when compared with those from 3D cleaning showed the maximum disagreement. scanning measurements irrespective of the utilised On average, corrosion levels measured for the bars cleaning method. However, the results indicated better that were cleaned using sandblasting as preferred agreement between the weight loss and those of 3D cleaning method were higher, followed by the bars scanning for the sandblasted bars wherein there was a cleaned by acid immersion and the uncleaned bars in minor difference between both methods. In contrast, terms of both weight loss and CT scanning measure- both acid and metallic bristle brush exhibited higher ments as shown in Fig. 8. This result along with the deviations when compared to sandblasting. Metallic fact that the different bars were considered arbitrarily brush cleaning exhibited the maximum difference, and within the tested beams indicates that the accuracy of the obtained values followed a trend in which a higher the rust removal was strongly dependent on the corrosion level led to a higher discrepancy between employed cleaning method. both measurements. Additionally, the highest corro- The accuracy of both cleaning methods (acid and sion level measured using both weight loss and 3D sandblasting) seemed to follow a trend that indicated scanning was observed for the sandblasted specimens, that the accuracy of the cleaning method became less and this was followed by the acid and the metallic Fig. 8 Comparison of 3D 1.1 18% scanning versus weight loss 1.0 measurements for different 15% 0.9 cleaning methods 0.8 12% 0.7 9% 0.6 0.5 6% 0.4 Brush Brush 3% Sandblasting Sandblasting 0.3 Acid Acid 0.2 0% 0% 3% 6% 9% 12% 15% 18% 0% 3% 6% 9% 12% 15% 18% Weight loss 3D scanning method Level of corrosion Weight loss method 3D scanning /Weight loss [-] Materials and Structures (2018) 51:78 Page 11 of 13 78 Fig. 9 Comparison of CT 18% 1.1 scanning versus weight loss 1.0 15% measurement for different 0.9 cleaning methods and prior 12% 0.8 to cleaning 0.7 9% 0.6 Before cleaning 6% 0.5 Sandblasting Acid 0.4 Before cleaning 3% Sandblasting 0.3 Acid 0% 0.2 0% 3% 6% 9% 12% 15% 18% 0% 3% 6% 9% 12% 15% 18% Weight loss CT scanning method consuming, and possessing a lower removal relevant when higher corrosion levels were observed as shown in Fig. 9b, i.e. the more corroded the bar, the capacity when compared to sandblasting. The surface finish also hid the remaining rust clusters, better the performance of the cleaning method. Nevertheless, the ratio of CT scanning to weight loss and this led to wrong conclusions with respect to the cleaning capacity since rust clusters were measurement did not correspond to values exceeding 0.9, which clearly showed a limitation of the accuracy unveiled as time elapsed. of CT scanning. Furthermore, the corrosion level 3. Sandblasting corresponded to the most efficient evaluation results of the uncleaned bars by using CT and reliable corrosion removal method. Larger scanning technique revealed that the selected accuracy and deeper pits were detected as in addition to a value was not sufficient to distinguish between the rust better definition of the corrosion impression on the and steel, and consequently, it is necessary to carefully bar surface. The obtained weight loss results agreed well with the different corrosion level clean the bars prior to scanning. attainment techniques. 4. The harshness (i.e. unintended removal of sound steel) observed in acid immersion and sandblast- 4 Conclusions ing was similar and not significant. Brush cleaning presented almost no harshness. The following conclusions were obtained from the study: • With respect to the different measurement • Major differences were observed in the results of methods: the different cleaning methods: 1. The 3D scanning displayed high reliability in the 1. The results indicated that metallic brush was not assessment of the corrosion level when the sufficiently strong to remove all the corrosion corroded bar was cleaned well although the 3D products for naturally corroded bars. The fact that scanning results were very sensitive to the clean- it was impossible to visually observe remaining ing method used. The main advantage of this corrosion products after cleaning, and the cleaned method is that it enables a detailed description of surface was shiny and smooth led to misleading the pit geometry and the corrosion level variation interpretations. A comparison of the results with relative to the studied length. those obtained for the other methods, in conjunc- 2. The results revealed that CT scanning was not tion with the observation that the pieces jumped sufficiently accurate when applied to these types off during tensile tests, strongly suggests that of large specimens. It is possible to use CT there were remaining corrosion products. scanning with a higher accuracy if smaller spec- 2. Acid cleaning described reasonably good results. imens are used. However, the specimens are then However, the method exhibited clear disadvan- considered as too small to be used for tensile tests, tages such as requiring more cycles, being time and a larger number of scans per bar are required. Weight loss method CT scanning /Weight loss [-] 78 Page 12 of 13 Materials and Structures (2018) 51:78 The CT scanning technique appeared to be less 2. Law D, Du Y, Cairns J (2008) Structural performance of corrosion-damaged concrete beams. Mag Concr Res sensitive to the cleaning method. Unfortunately, 60(5):359–370 the technology used in this work did not possess 3. Biondini F, Vergani M (2014) Deteriorating beam finite the capacity to obtain reliable corrosion levels element for nonlinear analysis of concrete structures under corrosion. Struct Infrastruct Eng 11(4):519–532 when the specimen was completely uncleaned. 4. Maalej M, Ahmed SFU, Paramasivam P (2003) Corrosion 3. The corrosion level attained by weight loss durability and structural response of functionally-graded measurement corresponded to the most efficient concrete beams. J Adv Concr Technol 1(3):307–316 and trustworthy measurement. The density of rust 5. Lundgren K, Gylltoft K (2000) A model for the bond between concrete and reinforcement. Mag Concr Res products is lower than that of the sound steel, and 1:53–63 this strongly reduces the associated error of the 6. Casanova A, Jason L, Davenne L (2012) Bond slip model cleaning methods. However, it was necessary to for the simulation of reinforced concrete structures. Eng employ high efficiency cleaning methods, such as Struct 39:66–78 7. Fernandez I, Etxeberria M, Marı´ AR (2016) Ultimate bond sandblasting and acid cleaning, to obtain reliable strength assessment of uncorroded and corroded reinforced corrosion levels. recycled aggregate concretes. Constr Build Mater 111:543–555 As a general conclusion, commonly used recommen- ´ ´ 8. Fernandez I, Bairan JM, Marı AR (2015) Corrosion effects dations in previous studies [20] only warrant that the on the mechanical properties of reinforcing steel bars. Fatigue and r–e behavior. Constr Build Mater 101:772–783 selected cleaning method has reached its maximum 9. Ou Y-C, Susanto YTT, Roh H (2016) Tensile behavior of cleaning capacity, which does not unequivocally naturally and artificially corroded steel bars. Constr Build correspond to the actual corrosion level. Hence, the Mater 103:93–104 recommendations of the present study indicate that 10. Lundgren K, Gylltoft K (2000) A model for the bond between concrete and reinforcement. Mag Concr Res sandblasting should be used as a cleaning method for 52(1):53–63 naturally corroded bars. Weight loss measurements 11. Apostolopoulos CA, Papadopoulos MP, Pantelakis SG are sufficient when detailed information on corrosion (2006) Tensile behavior of corroded reinforcing steel bars BSt 500s. Constr Build Mater 20(9):782–789 is not required while 3D scanning is recommended if 12. Michel A, Solgaard AOS, Pease BJ, Geiker MR, Stang H, information about the corrosion variation along the bar Olesen JF (2013) Experimental investigation of the relation is needed. between damage at the concrete-steel interface and initia- tion of reinforcement corrosion in plain and fibre reinforced Funding This work was undertaken at Chalmers University of concrete. Corros Sci 77:308–321 Technology, Division of Structural Engineering, Concrete 13. Michel A, Otieno M, Stang H, Geiker MR (2016) Propa- Structures. The authors would like to acknowledge the funding gation of steel corrosion in concrete: experimental and from the Swedish Transport Administration (Trafikverket). numerical investigations. Cem Concr Compos 70:171–182 14. Malumbela G, Alexander M, Moyo P (2010) Variation of Compliance with ethical standards steel loss and its effect on the ultimate flexural capacity of RC beams corroded and repaired under load. Constr Build Conflict of interest The authors declare that they have no Mater 24(6):1051–1059 conflict of interest. 15. Berrocal CG, Fernandez I, Lundgren K, Lo¨fgren I (2017) Corrosion-induced cracking and bond behaviour of cor- Open Access This article is distributed under the terms of the roded reinforcement bars in SFRC. Compos Part B Eng Creative Commons Attribution 4.0 International License (http:// 113:123–137 creativecommons.org/licenses/by/4.0/), which permits unrest- 16. Tahershamsi M, Zandi K, Lundgren K, Plos M (2014) ricted use, distribution, and reproduction in any medium, pro- Anchorage of naturally corroded bars in reinforced concrete vided you give appropriate credit to the original author(s) and structures. Mag Concr Res 66(14):729–744 the source, provide a link to the Creative Commons license, and 17. Lundgren K, Tahershamsi M, Zandi K, Plos M (2015) Tests indicate if changes were made. on anchorage of naturally corroded reinforcement in con- crete. Mater Struct 48(7):2009–2022 18. Zhu W, Franc¸ois R (2014) Experimental investigation of the relationships between residual cross-section shapes and the References ductility of corroded bars. Constr Build Mater 69:335–345 19. Dang VH, Franc¸ois R (2013) Influence of long-term cor- ´ ´ 1. Fernandez I, Herrador MF, Marı AR, Bairan JM (2016) rosion in chloride environment on mechanical behaviour of Structural effects of steel reinforcement corrosion on stati- RC beam. Eng Struct 48:558–568 cally indeterminate reinforced concrete members. Mater 20. ASTM G1 (2011) Standard practice for preparing, cleaning, Struct 49(12):4959–4973 and evaluating corrosion test Materials and Structures (2018) 51:78 Page 13 of 13 78 21. Apostolopoulos CA, Demis S, Papadakis VG (2013) Chlo- performance and pit depth analysis. Constr Build Mater ride-induced corrosion of steel reinforcement—mechanical 38:139–146 performance and pit depth analysis. Constr Build Mater 31. Du Y, Cullen M, Li C (2013) Structural performance of RC 38:139–146 beams under simultaneous loading and reinforcement cor- 22. Fernandez I, Baira´n JM, Marı´ AR (2016) 3D FEM model rosion. Constr Build Mater 38:472–481 development from 3D optical measurement technique 32. Zhang R, Castel A, Franc¸ois R (2009) The corrosion pattern applied to corroded steel bars. Constr Build Mater of reinforcement and its influence on serviceability of 124:519–532 reinforced concrete members in chloride environment. Cem 23. Mancini G, Tondolo F, Iuliano L, Minetola P (2014) Local Concr Res 39(11):1077–1086 reinforcing bar damage in r.c. members due to accelerated 33. Moreno Ferna´ndez E, Cobo Escamilla A, Ferna´ndez corrosion and loading. Constr Build Mater 69:116–123 Ca´novas M (2007) Ductility of reinforcing steel with dif- 24. Tang F, Lin Z, Chen G, Yi W (2014) Three-dimensional ferent degrees of corrosion and the ‘equivalent steel’ crite- corrosion pit measurement and statistical mechanical rion. Mater Constr 57:5–18 degradation analysis of deformed steel bars subjected to 34. Mancini G, Tondolo F, Iuliano L, Minetola P (2014) Local accelerated corrosion. Constr Build Mater 70:104–117 reinforcing bar damage in r.c. members due to accelerated 25. Wang X, Zhang W, Gu X, Dai H (2013) Determination of corrosion and loading. Constr Build Mater 69:116–123 residual cross-sectional areas of corroded bars in reinforced 35. Wang X, Zhang W, Gu X, Dai H (2013) Determination of concrete structures using easy-to-measure variables. Constr residual cross-sectional areas of corroded bars in reinforced Build Mater 38:846–853 concrete structures using easy-to-measure variables. Constr 26. Kashani MM, Crewe AJ, Alexander NA (2013) Use of a 3D Build Mater 38:846–853 optical measurement technique for stochastic corrosion 36. 3D Systems, ‘‘Geomagic Wrap 2014’’. http://www. pattern analysis of reinforcing bars subjected to accelerated geomagic.com/en/ corrosion. Corros Sci 73:208–221 37. Beck M, Goebbels J, Burkert A (2007) Application of X-ray 27. Tahershamsi Mohammad, Fernandez Ignasi, Lundgren tomography for the verification of corrosion processes in Karin, Zandi Kamyab (2016) Investigating correlations chloride contaminated mortar. Mater Corros 58(3):207–210 between crack width, corrosion level and anchorage 38. Savija B, Lukovic M, Hosseini SAS, Pacheco J, Schlangen capacity. Struct Infrastruct Eng 13(10):1294–1307. https:// E (2015) Corrosion induced cover cracking studied by doi.org/10.1080/15732479.2016.1263673 X-ray computed tomography, nanoindentation, and energy 28. Institudes SS (1971) Armeringssta˚ng. Kamsta˚ng Ks 60 och dispersive X-ray spectrometry (EDS). Mater Struct Ks 60S (STD-1250), pp 1–4 48(7):2043–2062 29. Zhu W, Franc¸ois R, Poon CS, Dai J-G (2017) Influences of 39. Lim S, Akiyama M, Frangopol DM, Jiang H (2017) corrosion degree and corrosion morphology on the ductility Experimental investigation of the spatial variability of the of steel reinforcement. Constr Build Mater 148:297–306 steel weight loss and corrosion cracking of reinforced con- 30. Apostolopoulos CA, Demis S, Papadakis VG (2013) Chlo- crete members: novel X-ray and digital image processing ride-induced corrosion of steel reinforcement—mechanical techniques. Struct Infrastruct Eng 13(1):118–134 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Materials and Structures Springer Journals

Evaluation of corrosion level of naturally corroded bars using different cleaning methods, computed tomography, and 3D optical scanning

Download PDF
13 pages

Loading next page...
 
/lp/springer-journals/evaluation-of-corrosion-level-of-naturally-corroded-bars-using-q1Vjj835r4

References (34)

Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Engineering; Structural Mechanics; Materials Science, general; Theoretical and Applied Mechanics; Operating Procedures, Materials Treatment; Civil Engineering; Building Materials
ISSN
1359-5997
eISSN
1871-6873
DOI
10.1617/s11527-018-1206-z
Publisher site
See Article on Publisher Site

Abstract

Materials and Structures (2018) 51:78 https://doi.org/10.1617/s11527-018-1206-z ORIGINAL ARTICLE Evaluation of corrosion level of naturally corroded bars using different cleaning methods, computed tomography, and 3D optical scanning . . Ignasi Fernandez Karin Lundgren Kamyab Zandi Received: 8 August 2017 / Accepted: 6 June 2018 The Author(s) 2018 Abstract Reliable methods are necessary to assess that occurs in existing impaired reinforced concrete the corrosion level to establish links between struc- structures. Chloride ion (Cl ) and carbon dioxide tural performance and reinforcement corrosion in (CO ) penetration from the structure’s surrounding concrete structures. Hence, in this study, a set of environment leads to the destabilization of passivity naturally corroded bars were subjected to metallic conditions provided by the surrounding concrete to the brushing, acid immersion, and sandblasting for rust steel bar. This destroys the steel protective layer, removal. Additionally, 3D optical, CT scanning, and termed as the passive layer, and subsequently initiates weight loss measurements were used to evaluate the corrosion of the steel. Hence, volumetric expansion of levels of corrosion. The results indicate that sand- corrosion products and cross-section reduction of the blasting is an optimal cleaning method. Weight loss steel bar leads to damages in the structure. Rust measurements are sufficient when detailed informa- expansion inside concrete generates significant inter- tion about corrosion is not required, and 3D scanning nal pressure that induce splitting stresses in the is preferred if information on corrosion variation along concrete along the corroded reinforcement and harm the bar is needed. the surrounding concrete. Splitting stresses are not well tolerated by concrete and result in cracking and Keywords Corrosion level measurement  Cleaning eventually spalling of the concrete cover. The corro- methods  3D optical scanning  Computed sion rate may increase when the reinforcement tomography scanning  Sandblasting  Acid cleaning becomes more exposed, thus facilitating the deterio- ration processes. Corrosion of reinforcement in concrete is examined widely by several previous studies. For example, 1 Introduction several previous studies [1–4] addressed the structural effects of corrosion. Various studies investigated local Corrosion of reinforced steel bars continues to be one aspects that addressed the effects of corrosion on bond of the most frequent and significant type of damage behaviour [5–7], change in mechanical properties [8–11], and other corrosion related phenomena [12–15]. However, most existing studies were con- I. Fernandez (&)  K. Lundgren  K. Zandi ducted under accelerated corrosion conditions. There Division of Structural Engineering, Department of Civil is a paucity of research examining natural corrosion and Environmental Engineering, Chalmers University of circumstances due to various difficulties [16–19]. Technology, 412 96 Go¨teborg, Sweden e-mail: [email protected] 78 Page 2 of 13 Materials and Structures (2018) 51:78 Therefore, there is a growing demand to continue different techniques. The study involved examining experimenting with naturally corroded specimens to and comparing the following three different cleaning further validate, and even extend existing knowledge. methods to remove attached mortar and rust from the However, irrespective of the origin of corrosion, all aforementioned specimens: (1) mechanical wire the fore-mentioned studies are based on the measure- bristle brushing, (2) sandblasting, and (3) chemical ment of the actual corrosion level with respect to the cleaning. This was followed by applying and compar- affected steel reinforcement. Thus, it is extremely ing the following three measurement techniques to important to obtain consistent methods to assess the evaluate the corrosion level: (1) weight loss, (2) 3D corrosion of the steel bar and to examine its detailed scanning, and (3) three-dimensional micro-computed impact on the steel surface. The methods should allow tomography (CT scanning). The study presents the acquisition of precise information detailing pit char- scope, applicability, and accuracy of each cleaning acteristics and a faster and detailed measurement of method along with its combination with the different the level of corrosion for a specific bar length. measurement techniques. Furthermore, it details rec- Simultaneously, the availability of reliable and more ommendations to obtain reliable levels of corrosion detailed information about the corrosion distribution for the corroded steel bars. will allow the establishment of increasingly trustwor- thy links between these measurements with respect to 2 Experimental programme the structural response. Typically, existing studies have based the assessment of the corrosion level on the ASTM G1—Standard Practice for Preparing, Clean- The specimens presented in the study are part of a ing, and Evaluating Corrosion Test Specimens [20]. A larger experimental campaign conducted at Chalmers few studies attempted to incorporate new techniques University of Technology in which specimens from and technologies from other fields that theoretically edge beams of the Stallbacka Bridge in Sweden were provided a better description of the corrosion level used. The specimens were taken from the bridge when along a steel bar [9, 21–26]. Nevertheless, all the it was under repair following approximately 30 years aforementioned techniques are yet strongly dependent of exposure to different natural deterioration phenom- on the cleaning methods performed on the steel bars. A ena such as corrosion induced by chlorides from de- recent study by Tahershamsi et al. [27] pointed out icing salts. In an earlier study, beams were tested in significant discrepancies between obtained results four-point suspended bending tests to obtain anchor- using three-dimensional optical measurement (3D age failure [16, 17]. scanning) and weight loss measurement following Table 1 presents an overview of all specimens metallic brush cleaning of naturally corroded steel included in this study on which different techniques bars. There is a paucity of similar studies that relate were applied to measure the corrosion level. Each row and compare different cleaning methods and corrosion indicates a specific enforced cleaning method includ- level measurement techniques. An exception is the ing acid immersion, sandblasting, metallic brush, and study conducted by Tang et al. [24] that presented a no-cleaned before scanning and later cleaned using direct comparison between gravimetric and 3D scan- sandblasting. The various columns describe the num- ning to assess the level of corrosion; the results ber of specimens and the method used for the indicated a very reasonable agreement; artificially corrosion level evaluation for each group, specifically, corroded bars and sandblasting were used as cleaning weight loss, 3D scanning, and CT scanning. methods. Hence, the aim of the present study includes 2.1 Reinforcement steel bars evaluating the scope and applicability of common cleaning methods used in existing studies as well as Tensile reinforcement steel bars were carefully different measurement techniques to evaluate the extracted from a non-critical section of the beams. corrosion level of naturally corroded bars. Conse- The beams contained to two different 16-mm diameter quently, a set of corroded steel bars extracted from a bar types; both types of steel class Ks60 [28]. Each real bridge that is more than 30 years old was cleaned, type involved different rib patterns, namely skewed and their levels of corrosion were assessed utilising Materials and Structures (2018) 51:78 Page 3 of 13 78 Table 1 Overview of specimens, cleaning methods, and corrosion level evaluation methods Cleaning method Specimens (bars) Corrosion level method Weight loss 3D scanning CT scanning Acid 7 Yes Yes Yes Sandblasting 7 Yes Yes Yes No-cleaned 7 Yes No Yes Metallic brush 17 Yes Yes No Specimens were first scanned and later cleaned by using sandblasting Specimens presented in a previous study [27] and straight ribs. Only straight rib pattern bars were 2.1.2 Chemical composition considered in this study. The chemical composition of the extracted steel was investigated by using scanning electron microscopy 2.1.1 Geometrical description of the bars (SEM). The values presented in Table 3 correspond to the average values obtained for each component in The steel bars were cut in lengths of 300 mm. This length was selected as a compromise between differ- different scanning points throughout the bar cross- section. Iron makes up to 100% of the steel ent requirements with respect to performed tasks. This included the maximum length that is allowed for composition. subsequent CT scanning by using reasonable resolu- tions, the minimum length necessary for mechanical 2.2 Removal of corrosion products properties characterization by means of tensile tests, and assessment of the corrosion level and a reasonable length for representativeness. Some studies have shown that pits are distributed stochas- The extracted steel was cut into suitable parts with a length of 300 mm, and the bars were cleaned by using tically along the bar length in [29]. Hence the choice of the specimen length should not have a relevant impact the following three most common methods found in literature: metallic bristle brushing, acid immersion, in the presented results; moreover, considering that the bars where extracted from the same structure, from and sandblasting. Metallic bristle brushing is the most commonly used cleaning method [27, 30, 31]; this is zones that presented clear signs of deterioration, similar damage levels, and which had been exposed mainly because it entails a low number of require- to similar aggressive environment, they are considered ments for its use. Conversely, immersion in an acid to be comparable. solution is less common [7, 15, 32, 33]. As required by The following bar specifications were measured the specified standard [20], the weight loss was measured after several cleaning cycles. Additionally, from the uncorroded samples by using a Vernier Calliper with a resolution of 10 lm, as shown in Fig. 1 some of the specimens were subject to two different scanning techniques, namely 3D and CT scanning; this and Table 2. The values given correspond to the measurement averages. allowed a description of the outer surface of the Fig. 1 Geometry of ribbed reinforcement bars 78 Page 4 of 13 Materials and Structures (2018) 51:78 Table 2 Measured parameters and standard deviation to describe the geometry of the ribbed reinforcement bar (in mm) Steel class Nominal diameter D r a r h r h r l r l rh () 1 2 1 2 Ks60 Ø16 15.72 0.03 1.91 0.02 1.23 0.10 1.23 0.10 9.00 0.36 2.27 0.10 90 Table 3 Chemical composition of the steel bars C O Si Mn Cr Ni Cu Ks60 2.84 4.63 0.22 1.08 0.19 0.17 0.51 Each parameter is given in % corroded bar and thereby permitted an evaluation of the corrosion level. Specifically, 3D and CT scanning techniques that are used widely in fields, such as industrial engineering or medicine, are not commonly Number of Cleaning Cycles used in civil engineering. Both techniques are rela- tively new, especially with respect to their application Fig. 2 Standard recommendations for cleaning corroded steel to deteriorated structures and their different structural bars [20] elements. Thus, only a limited number of studies supply, and the engine was in charge of continu- explored these methods and their applications [9, 22, 24, 30, 34, 35]. Furthermore, a group of ously rotating the brush at the same speed. The recommendation [20] did not specify any cycle or specimens were scanned with the CT scanning tech- nique prior to cleaning to assess whether the CT exposition time, and thus each cycle was not systematically measured. Instead, each cycle was scanning technique was sufficiently effective to obtain a corrosion level without cleaning. This would involve distinguished when perceptible changes on the surface due to the rust removing were observed time efficiency as well as a method to avoid possible after the bar was swept from end to end. According induced inaccuracies of the cleaning methods. to this criterion, the necessary time for each cycle was approximately in the range of 10 min to 2.2.1 Cleaning methods and weight loss measurement 15 min based on the actual level of corrosion. • Sandblasting was performed in an individual The same procedure was followed for each of the cleaning methods used in this study. Reiterated cabinet designed for the purpose. The sand was blasted at 5–7 bars of pressure. The sand employed cleaning cycles were applied to each specimen until the mass loss was lower than 0.2% of the previous for the rust removal corresponded to silica sand. Similarly, the recommendation followed for measurement. Consequently, it was possible to clearly distinguish two different slopes as shown in Fig. 2. metallic brush [20] did not specify any cycle or exposition time. Hence, the same described crite- This necking point was not quantified in the ASTM rion was used, and the bar was sandblasted from recommendations [20] although this is used in the end to end until perceptible changes on the surface present study based on the harshness of the cleaning were observed. Accordingly, the necessary appli- methods as detailed in subsequent sections. cation time for each cycle was approximately less A short description of each cleaning method than 5 min. performed on the steel bars is as follows: • A wide range of chemical cleaning based methods • Mechanical brush was performed by utilising a are found in the ASTM recommendation [20]. The rotational metallic wire bristle brush. It was present study involved using the chemical cleaning attached to an engine plugged to an electricity method by repeated immersion of the bars in a Mass Loss Materials and Structures (2018) 51:78 Page 5 of 13 78 solution of hydrochloric acid and utropine (500 ml/l of solution hydrochloric acid, sp gr 1.19, 3.5 gr/l of solution hexamethytene tetramine and regent water) in cycles that approximately lasted for 10 min. The selection choice was motivated by both safety rules and practical reasons since the other methods used carcinogenic products, high temperature environment, or very long exposition times. The mass loss was measured after every cycle and the cycles were repeated until the aforementioned threshold was reached for each method. Fig. 3 3D surface generated from the 3D scanning Additionally, in order to obtain the scope of harshness of each cleaning method, uncorroded steel (X, Y, Z), was established and referenced to the end of bars were cleaned, and the loss of sound steel was the bar. The high resolution of the surface mesh assessed with respect to each cleaning method. Acid allowed for a sufficiently detailed description of the and sandblasting cleaning yielded similar levels of geometry of the bar to obtain information on features sound steel removal that corresponded to approxi- including pit depth and length, pit distribution, and mately 0.2% of the initial weight. In contrast, metallic loss of cross-sectional area along the bar length. bristle brush had no significant impact in the non- corroded steel removing. Subsequently, the value of 2.2.3 Micro-computed tomography technique 0.2% weight loss between cycles was used in the present study as the threshold beyond which the bar Sound steel was determined using three-dimensional was considered as fully cleaned. In addition to this micro-computed tomography (CT scanning), which is threshold, it was also visually confirmed that the bars used widely in image diagnostic medicine and is a looked clean. promising technology that is also applicable to other fields such as civil engineering. Previous studies 2.2.2 3D optical measurement technique [22, 37, 38] used this technology to assess the effect of corrosion in concrete and mortar phases although The 3D scanning of the corroded bars was performed very few studies focused on obtaining a comprehen- by means of optical measurement. An industrial stereo sive surface of corroded reinforcing bars [39]. device with two cameras of 5 Megapixels was used. Corroded steel bars were placed in a Metrotom The maximum accuracy provided by the cameras machine that projected X-Ray beams. As the steel bars corresponded to 2 lm, which allows the description of were constituted by two clearly different materials, i.e. imperfections over the steel bar surface due to sound steel and corrosion products, different specific corrosion. A correction of the measurement inaccura- amounts of the X-Ray beam were absorbed by each cies, such as polygon spikes removing and mesh holes one. It is possible to obtain a 2D image of the object by closing, and data treatment was performed using the collecting the remaining transmitted intensity through TM post processing software Geomagic Wrap 2014 the machine detector. The required intensity of the [36]. X-Ray beam was adjusted based on the material The outcome of the optical measurement corre- density, i.e., it must be sufficiently strong to penetrate sponded to a very fine mesh of triangular surface the full thicknesses of the different components. The polygons connected by nodes, see Fig. 3. The average process is repeated several times conveyed to the size of the element corresponded to 0.014 mm with a rotation of the object within specific angles, and side length of approximately 0.15 mm. The number of subsequently post-processing of the obtained images triangular elements in each scanning was between is performed to build a full 3D interpretation of the 2,000,000 elements and 3,000,000 elements depend- body. A larger difference between the densities of the ing on the corrosion level. A global coordinate system, materials that conform the object increases the degree 78 Page 6 of 13 Materials and Structures (2018) 51:78 of ease and clarity of the result. The actual resolution A complete description of the detailed steps from of the flatbed-detector corresponded to a frame of the initial 3D polygonal mesh to the graph that shows 1024 9 1024 pixels or a voxel of 174 lm for 3D-CT the cross-section variation is provided in a previous scanning. Accordingly, in order to increase the final study [27]. resolution of the final surface mesh of the steel, two scans together covering the total volume were per- formed which allowed the maximum accuracy by the 3 Results and discussion equipment used. Subsequently, the two steel outcome surfaces were digitally stitched to form a unique full 3.1 Cleaning methods volume that contained all the defining points. The technology possesses the potential to describe the Figure 4a shows the average mass loss of all bars outer surface of the corroded steel bar in detail without cleaned with each cleaning method. The same pro- cleaning the corrosion products in advance as they posed methodology [20] was followed for each present very dissimilar densities. cleaning method with reiterated cleaning cycles until The same type of surface mesh as that described for the mass loss was lower than 0.2% of the last measured the 3D scanning technique that was previously weight. However, large differences were observed in presented was obtained by means of the CT scanning. the final measurements between the three selected options as shown in Fig. 4a. The sandblasted speci- mens exhibited the largest measured corrosion level, 2.2.4 Evaluation of the level of corrosion using systematically both for all the specimens, and for the scanning techniques average level of corrosion in each group. This was followed by the acid cleaned specimens. Finally, the A method that was developed in a previous study [27] mechanically brushed specimens exhibited the small- was used to determine the corrosion level variation est measured corrosion level. The methods were along a bar based on the scanning measurements: expected to result in similar levels of the average • The resulting outer geometry based on the scan- mass loss, since the bars were randomly obtained from ning output is postprocessed, cleaned and repaired. the tested beams, from areas which had been exposed • The coordinates of the nodes are transformed into a to similar aggressive environment and subjected to similar damage (all the specimens presented clear polar coordinate system, and a contour plot of the corrosion penetration along the bar surface is signs of deterioration). Thus, the fore-mentioned created from the new points, which allows to discrepancies indicate that the proposed recommen- visually observe the corrosion pits along the bar dations did not provide information to a user with surface respect to the degree of cleanness of the bar at the end • The cross-sectional area at specific sections uni- of the process, as opposed to whether the performed formly separated is obtained by integration of the cleaning method reached its maximum cleaning coordinates. capacity for a set of specific conditions including steel • The cross-sectional area along the bar is calculated type, initial amount of mortar, and corrosion products and plotted. The bar corresponded to a ribbed bar, attached or cleaning agent (such as chemical, or and thus the cross-sectional area varies along the brush). This assertion is backed by Fig. 4b, in which the average number of cycles applied to each specimen bar. The effect of the ribs is eliminated by a smoothing fit that used cubic splines; this results in to reach the necking point together with its standard another curve. The uncorroded zone/s of the bar is/ deviation is shown. It was observed that the number of are identified, and the average cross-sectional area cycles remained constant regardless the final level of is used as a reference. corrosion. • The normalized cross-sectional reduction is plot- As shown in Fig. 4b, the number of cycles neces- ted by dividing the measured cross-sectional area sary to obtain a weight loss difference lower than 0.2% with respect to the uncorroded area. within cycles (which was defined by the authors as a • Finally, the corrosion level variation in percentage reference value based on the degree of harshness of the is plotted along the bar. cleaning methods) ranged between 4 cycles and 7 Materials and Structures (2018) 51:78 Page 7 of 13 78 (a) 16% Brush Sand. 14% Acid 12% 10% 8% 6% 4% 2% 4.14 3.29 6.43 0% Brush Sand. Acid Brush Sand. Acid Number of cyles (b) 0.0% 0.2% 0.4% 0.6% 0.8% 1.0% Brush 1.2% Sandblasting Acid 1.4% Fig. 4 a Average mass loss for the different cleaning methods. b Examples of cleaning cycles for each cleaning method cycles according to the method for representative solution. Thus, the individual effective time with specimens. Sandblasting reached its highest cleaning respect to the bar was considered lower and approx- capacity in fewest number of cycles whereas acid imately corresponded to 15–20 min. The results immersion needed the highest number of cycles to revealed that sandblasting exhibited a very high speed reach the defined threshold. The cleaning speed of the and only required less than 20 min of application per metallic brush method ranged in between that of single bar to reach the same cleaning capacity. Finally, sandblasting and acid immersion on an average, metallic brush required the longest cleaning time per although the scatter observed among the different bar and the average corresponded to approximately bars increased in contrast to those observed in the 40 min based on the actual corrosion level. other methods. This scatter could indicate higher However, direct conclusions to assess the actual dependence of the brush cleaning method on the actual accuracy of the method could not be extracted from corrosion level than the others. the results after cleaning since the results only With respect to the cleaning time necessary to reach expressed a relative comparison with respect to the the proposed necking point, the results indicated that initial weight and did not provide any indication as to the maximum time to reach the proposed necking whether or not the bar was completely cleaned. point corresponded to that of acid solution that Nevertheless, the maximum cleaning capacity was required more than 70 min of immersion in addition definitely reached for each method. Figure 5 shows a to the drying and weighting time of the bars within few bars after rust removal. cycles. However, acid cleaning allowed multiple bars Important differences were observed in the final to be simultaneously cleaned, i.e., several bars could shape of the surface after cleaning. The finishes for the be placed in the same recipient containing the acid different methods corresponded to shiny and smooth Weight loss differrence [%] Average level of corrosion Average nr. of cycles 78 Page 8 of 13 Materials and Structures (2018) 51:78 Fig. 5 Cleaned bars. From top to bottom, the figure shows metallic brush, sandblasting, and acid immersion Remaining corrosion products for metallic brush, matt and granulated for sandblast- bar were only unveiled after a significant period of ing, and dark brown for acid cleaning. All the methods time elapsed as shown in Fig. 5. This was potentially showed corrosion pits along the bar irrespective of the due to the drying out of the acid solution and the finish for each method. However, the depiction of pits occurrence of slight corrosion in the sound steel that was significantly more clear after sandblasting when changed the surface shade. compared to the other methods, as shown in Figs. 5 Conversely, sandblasted specimens always exhib- and 6. Subsequently, a more comprehensive impact of ited remaining rust clusters that were distributed along the corrosion on the bar surface was observed in which the bar during the cleaning cycles. Specifically, the larger and deeper pits were found in the sandblasted mentioned clusters were also used as additional visual bars as shown in Fig. 6. This was not obtained when criteria to define the exposition time in each cycle. the bars were extracted or when the bars were cleaned Nevertheless, there was a significantly lower presence with the two other methods. of rust clusters at the necking point in the case for It should be noted that neither metallic brush nor sandblasting when compared with acid cleaning as acid immersion specimens portrayed any remaining clearly shown in Fig. 5. The remaining corrosion corrosion products after the bars were cleaned. The products on the bar surface were a result of the shiny and smooth surfaces due to the metallic brush compromise between the corrosion products and covered the remaining corrosion products attached to sound steel removed in each cycle, which must always the bar. However, few pieces jumped off and the bars be lower than the method harshness. Corrosion exhibited the aforementioned products when the bars products, in addition to sound steel must be removed were tested under a tensile load [27]. With respect to in the cycle in order to maintain the accuracy of the acid immersion, the surface after cleaning exhibited a measured level of corrosion. homogeneous dark brown finish and did not allow the detection of the presence of remaining corrosion 3.2 Assessment of the corrosion level, products. Clusters of corrosion products indicating the and a comparison between weight loss, CT, presence of a significant amount of rust attached to the and 3D scanning Table 4 lists the levels of corrosion obtained with the different techniques that are applied on each bar. Each value represents the average corrosion level along the specimen length, which corresponds to 300 mm. Hence, it was expected that presented values evaluated with the different techniques would be in agreement. However, major differences were observed and are discussed in the following section. Figure 7 shows an overall description of the obtained results by means of depicting the average measurements. As shown in the figure, 3D scanning Fig. 6 Pits unveiled after sandblasting Materials and Structures (2018) 51:78 Page 9 of 13 78 Table 4 Corrosion level results Cleaning method Specimen Corrosion level (%) Weight loss Average 3D scanning Average CT scanning Average Acid CA-1 12.1 8.78 10.3 6.96 9.9 6.66 CA-2 13.3 10.8 10.7 CA-3 2.0 1.5 0.8 CA-4 4.5 2.8 2.6 CA-5 16.2 13.6 13.5 CA-6 4.4 3.2 2.8 CA-7 9.0 6.5 6.3 Sandblasting CA-8 17.1 11.05 16.5 10.64 15.0 9.29 CA-9 11.1 10.6 9.3 CA-10 1.7 1.8 0.6 CA-11 13.9 13.4 12.1 CA-12 14.2 13.9 12.7 CA-13 11.9 11.3 9.5 CA-14 7.4 7.0 5.8 Not-cleaned CA-15 12.2 8.7 – – 3.9 2.085 CA-16 8.4 – 0.8 CA-17 3.3 – 0.1 CA-18 7.4 – 2.1 CA-19 14.3 – 4.3 CA-20 9.5 – 2.2 CA-21 5.8 – 1.2 Metallic brush CA-22 7.0 4.32 4.6 2.74 – – CA-23 2.9 2.5 – CA-24 1.6 0.8 – CA-25 0.0 0.1 – CA-26 6.7 3.7 – CA-27 2.3 2.0 – CA-28 9.0 4.3 – CA-29 6.3 3.5 – CA-30 2.3 1.9 – CA-31 5.7 3.6 – CA-32 7.8 6.2 – CA-33 4.0 2.5 – CA-34 2.6 1.4 – CA-35 1.5 0.9 – CA-36 1.4 0.7 – CA-37 8.6 5.0 – CA-38 3.7 2.8 – Specimens that were scanned prior to cleaning and subsequently cleaned using sandblasting Specimens presented in a previous study [27] resulted in higher values in all cases when compared to measurement. Average 3D scanning and weight loss those obtained in CT scanning. However, the obtained values described very good agreement with respect to values were consistently below the weight loss sandblasting cleaning method, and the difference 78 Page 10 of 13 Materials and Structures (2018) 51:78 12% brush. This is potentially related to the efficiency of Weight loss the cleaning method. 3D scanning 10% Generally, sandblasting cleaning yielded the best CT scanning agreement irrespective of the actual corrosion level as 8% shown in Fig. 8b by the ratio of 3D scanning to weight 6% loss measurement. Acid cleaning exhibited a slightly better agreement when the corrosion level increased, 4% and this indicated that the accuracy of the cleaning method was less relevant for increases in the corrosion 2% level. Conversely, metallic brush cleaning displayed large scatter and a clear trend was not observed. 0% Brush Sand. Acid No-Clean The same comparison for the CT scanning tech- nique is presented in Fig. 9. The CT scanning Fig. 7 Average corrosion level for each presented method measurements resulted in smaller corrosion levels when compared with the weight loss measurements. between the average weight loss and that in CT The same behaviour was observed for all the cleaning scanning was higher for all the cleaning methods. methods as well as for the bars scanned prior to A direct comparison between single weight loss and cleaning. In a manner, similar to the 3D scanning 3D scanning corrosion levels for the three proposed results, the best agreement between CT scanning and cleaning methods is shown in Fig. 8a. As shown, weight measurements was exhibited by the sand- gravimetric weight loss typically resulted in higher blasted specimens while the bars scanned prior to corrosion levels when compared with those from 3D cleaning showed the maximum disagreement. scanning measurements irrespective of the utilised On average, corrosion levels measured for the bars cleaning method. However, the results indicated better that were cleaned using sandblasting as preferred agreement between the weight loss and those of 3D cleaning method were higher, followed by the bars scanning for the sandblasted bars wherein there was a cleaned by acid immersion and the uncleaned bars in minor difference between both methods. In contrast, terms of both weight loss and CT scanning measure- both acid and metallic bristle brush exhibited higher ments as shown in Fig. 8. This result along with the deviations when compared to sandblasting. Metallic fact that the different bars were considered arbitrarily brush cleaning exhibited the maximum difference, and within the tested beams indicates that the accuracy of the obtained values followed a trend in which a higher the rust removal was strongly dependent on the corrosion level led to a higher discrepancy between employed cleaning method. both measurements. Additionally, the highest corro- The accuracy of both cleaning methods (acid and sion level measured using both weight loss and 3D sandblasting) seemed to follow a trend that indicated scanning was observed for the sandblasted specimens, that the accuracy of the cleaning method became less and this was followed by the acid and the metallic Fig. 8 Comparison of 3D 1.1 18% scanning versus weight loss 1.0 measurements for different 15% 0.9 cleaning methods 0.8 12% 0.7 9% 0.6 0.5 6% 0.4 Brush Brush 3% Sandblasting Sandblasting 0.3 Acid Acid 0.2 0% 0% 3% 6% 9% 12% 15% 18% 0% 3% 6% 9% 12% 15% 18% Weight loss 3D scanning method Level of corrosion Weight loss method 3D scanning /Weight loss [-] Materials and Structures (2018) 51:78 Page 11 of 13 78 Fig. 9 Comparison of CT 18% 1.1 scanning versus weight loss 1.0 15% measurement for different 0.9 cleaning methods and prior 12% 0.8 to cleaning 0.7 9% 0.6 Before cleaning 6% 0.5 Sandblasting Acid 0.4 Before cleaning 3% Sandblasting 0.3 Acid 0% 0.2 0% 3% 6% 9% 12% 15% 18% 0% 3% 6% 9% 12% 15% 18% Weight loss CT scanning method consuming, and possessing a lower removal relevant when higher corrosion levels were observed as shown in Fig. 9b, i.e. the more corroded the bar, the capacity when compared to sandblasting. The surface finish also hid the remaining rust clusters, better the performance of the cleaning method. Nevertheless, the ratio of CT scanning to weight loss and this led to wrong conclusions with respect to the cleaning capacity since rust clusters were measurement did not correspond to values exceeding 0.9, which clearly showed a limitation of the accuracy unveiled as time elapsed. of CT scanning. Furthermore, the corrosion level 3. Sandblasting corresponded to the most efficient evaluation results of the uncleaned bars by using CT and reliable corrosion removal method. Larger scanning technique revealed that the selected accuracy and deeper pits were detected as in addition to a value was not sufficient to distinguish between the rust better definition of the corrosion impression on the and steel, and consequently, it is necessary to carefully bar surface. The obtained weight loss results agreed well with the different corrosion level clean the bars prior to scanning. attainment techniques. 4. The harshness (i.e. unintended removal of sound steel) observed in acid immersion and sandblast- 4 Conclusions ing was similar and not significant. Brush cleaning presented almost no harshness. The following conclusions were obtained from the study: • With respect to the different measurement • Major differences were observed in the results of methods: the different cleaning methods: 1. The 3D scanning displayed high reliability in the 1. The results indicated that metallic brush was not assessment of the corrosion level when the sufficiently strong to remove all the corrosion corroded bar was cleaned well although the 3D products for naturally corroded bars. The fact that scanning results were very sensitive to the clean- it was impossible to visually observe remaining ing method used. The main advantage of this corrosion products after cleaning, and the cleaned method is that it enables a detailed description of surface was shiny and smooth led to misleading the pit geometry and the corrosion level variation interpretations. A comparison of the results with relative to the studied length. those obtained for the other methods, in conjunc- 2. The results revealed that CT scanning was not tion with the observation that the pieces jumped sufficiently accurate when applied to these types off during tensile tests, strongly suggests that of large specimens. It is possible to use CT there were remaining corrosion products. scanning with a higher accuracy if smaller spec- 2. Acid cleaning described reasonably good results. imens are used. However, the specimens are then However, the method exhibited clear disadvan- considered as too small to be used for tensile tests, tages such as requiring more cycles, being time and a larger number of scans per bar are required. Weight loss method CT scanning /Weight loss [-] 78 Page 12 of 13 Materials and Structures (2018) 51:78 The CT scanning technique appeared to be less 2. Law D, Du Y, Cairns J (2008) Structural performance of corrosion-damaged concrete beams. Mag Concr Res sensitive to the cleaning method. Unfortunately, 60(5):359–370 the technology used in this work did not possess 3. Biondini F, Vergani M (2014) Deteriorating beam finite the capacity to obtain reliable corrosion levels element for nonlinear analysis of concrete structures under corrosion. Struct Infrastruct Eng 11(4):519–532 when the specimen was completely uncleaned. 4. Maalej M, Ahmed SFU, Paramasivam P (2003) Corrosion 3. The corrosion level attained by weight loss durability and structural response of functionally-graded measurement corresponded to the most efficient concrete beams. J Adv Concr Technol 1(3):307–316 and trustworthy measurement. The density of rust 5. Lundgren K, Gylltoft K (2000) A model for the bond between concrete and reinforcement. Mag Concr Res products is lower than that of the sound steel, and 1:53–63 this strongly reduces the associated error of the 6. Casanova A, Jason L, Davenne L (2012) Bond slip model cleaning methods. However, it was necessary to for the simulation of reinforced concrete structures. Eng employ high efficiency cleaning methods, such as Struct 39:66–78 7. Fernandez I, Etxeberria M, Marı´ AR (2016) Ultimate bond sandblasting and acid cleaning, to obtain reliable strength assessment of uncorroded and corroded reinforced corrosion levels. recycled aggregate concretes. Constr Build Mater 111:543–555 As a general conclusion, commonly used recommen- ´ ´ 8. Fernandez I, Bairan JM, Marı AR (2015) Corrosion effects dations in previous studies [20] only warrant that the on the mechanical properties of reinforcing steel bars. Fatigue and r–e behavior. Constr Build Mater 101:772–783 selected cleaning method has reached its maximum 9. Ou Y-C, Susanto YTT, Roh H (2016) Tensile behavior of cleaning capacity, which does not unequivocally naturally and artificially corroded steel bars. Constr Build correspond to the actual corrosion level. Hence, the Mater 103:93–104 recommendations of the present study indicate that 10. Lundgren K, Gylltoft K (2000) A model for the bond between concrete and reinforcement. Mag Concr Res sandblasting should be used as a cleaning method for 52(1):53–63 naturally corroded bars. Weight loss measurements 11. Apostolopoulos CA, Papadopoulos MP, Pantelakis SG are sufficient when detailed information on corrosion (2006) Tensile behavior of corroded reinforcing steel bars BSt 500s. Constr Build Mater 20(9):782–789 is not required while 3D scanning is recommended if 12. Michel A, Solgaard AOS, Pease BJ, Geiker MR, Stang H, information about the corrosion variation along the bar Olesen JF (2013) Experimental investigation of the relation is needed. between damage at the concrete-steel interface and initia- tion of reinforcement corrosion in plain and fibre reinforced Funding This work was undertaken at Chalmers University of concrete. Corros Sci 77:308–321 Technology, Division of Structural Engineering, Concrete 13. Michel A, Otieno M, Stang H, Geiker MR (2016) Propa- Structures. The authors would like to acknowledge the funding gation of steel corrosion in concrete: experimental and from the Swedish Transport Administration (Trafikverket). numerical investigations. Cem Concr Compos 70:171–182 14. Malumbela G, Alexander M, Moyo P (2010) Variation of Compliance with ethical standards steel loss and its effect on the ultimate flexural capacity of RC beams corroded and repaired under load. Constr Build Conflict of interest The authors declare that they have no Mater 24(6):1051–1059 conflict of interest. 15. Berrocal CG, Fernandez I, Lundgren K, Lo¨fgren I (2017) Corrosion-induced cracking and bond behaviour of cor- Open Access This article is distributed under the terms of the roded reinforcement bars in SFRC. Compos Part B Eng Creative Commons Attribution 4.0 International License (http:// 113:123–137 creativecommons.org/licenses/by/4.0/), which permits unrest- 16. Tahershamsi M, Zandi K, Lundgren K, Plos M (2014) ricted use, distribution, and reproduction in any medium, pro- Anchorage of naturally corroded bars in reinforced concrete vided you give appropriate credit to the original author(s) and structures. Mag Concr Res 66(14):729–744 the source, provide a link to the Creative Commons license, and 17. Lundgren K, Tahershamsi M, Zandi K, Plos M (2015) Tests indicate if changes were made. on anchorage of naturally corroded reinforcement in con- crete. Mater Struct 48(7):2009–2022 18. Zhu W, Franc¸ois R (2014) Experimental investigation of the relationships between residual cross-section shapes and the References ductility of corroded bars. Constr Build Mater 69:335–345 19. Dang VH, Franc¸ois R (2013) Influence of long-term cor- ´ ´ 1. Fernandez I, Herrador MF, Marı AR, Bairan JM (2016) rosion in chloride environment on mechanical behaviour of Structural effects of steel reinforcement corrosion on stati- RC beam. Eng Struct 48:558–568 cally indeterminate reinforced concrete members. Mater 20. ASTM G1 (2011) Standard practice for preparing, cleaning, Struct 49(12):4959–4973 and evaluating corrosion test Materials and Structures (2018) 51:78 Page 13 of 13 78 21. Apostolopoulos CA, Demis S, Papadakis VG (2013) Chlo- performance and pit depth analysis. Constr Build Mater ride-induced corrosion of steel reinforcement—mechanical 38:139–146 performance and pit depth analysis. Constr Build Mater 31. Du Y, Cullen M, Li C (2013) Structural performance of RC 38:139–146 beams under simultaneous loading and reinforcement cor- 22. Fernandez I, Baira´n JM, Marı´ AR (2016) 3D FEM model rosion. Constr Build Mater 38:472–481 development from 3D optical measurement technique 32. Zhang R, Castel A, Franc¸ois R (2009) The corrosion pattern applied to corroded steel bars. Constr Build Mater of reinforcement and its influence on serviceability of 124:519–532 reinforced concrete members in chloride environment. Cem 23. Mancini G, Tondolo F, Iuliano L, Minetola P (2014) Local Concr Res 39(11):1077–1086 reinforcing bar damage in r.c. members due to accelerated 33. Moreno Ferna´ndez E, Cobo Escamilla A, Ferna´ndez corrosion and loading. Constr Build Mater 69:116–123 Ca´novas M (2007) Ductility of reinforcing steel with dif- 24. Tang F, Lin Z, Chen G, Yi W (2014) Three-dimensional ferent degrees of corrosion and the ‘equivalent steel’ crite- corrosion pit measurement and statistical mechanical rion. Mater Constr 57:5–18 degradation analysis of deformed steel bars subjected to 34. Mancini G, Tondolo F, Iuliano L, Minetola P (2014) Local accelerated corrosion. Constr Build Mater 70:104–117 reinforcing bar damage in r.c. members due to accelerated 25. Wang X, Zhang W, Gu X, Dai H (2013) Determination of corrosion and loading. Constr Build Mater 69:116–123 residual cross-sectional areas of corroded bars in reinforced 35. Wang X, Zhang W, Gu X, Dai H (2013) Determination of concrete structures using easy-to-measure variables. Constr residual cross-sectional areas of corroded bars in reinforced Build Mater 38:846–853 concrete structures using easy-to-measure variables. Constr 26. Kashani MM, Crewe AJ, Alexander NA (2013) Use of a 3D Build Mater 38:846–853 optical measurement technique for stochastic corrosion 36. 3D Systems, ‘‘Geomagic Wrap 2014’’. http://www. pattern analysis of reinforcing bars subjected to accelerated geomagic.com/en/ corrosion. Corros Sci 73:208–221 37. Beck M, Goebbels J, Burkert A (2007) Application of X-ray 27. Tahershamsi Mohammad, Fernandez Ignasi, Lundgren tomography for the verification of corrosion processes in Karin, Zandi Kamyab (2016) Investigating correlations chloride contaminated mortar. Mater Corros 58(3):207–210 between crack width, corrosion level and anchorage 38. Savija B, Lukovic M, Hosseini SAS, Pacheco J, Schlangen capacity. Struct Infrastruct Eng 13(10):1294–1307. https:// E (2015) Corrosion induced cover cracking studied by doi.org/10.1080/15732479.2016.1263673 X-ray computed tomography, nanoindentation, and energy 28. Institudes SS (1971) Armeringssta˚ng. Kamsta˚ng Ks 60 och dispersive X-ray spectrometry (EDS). Mater Struct Ks 60S (STD-1250), pp 1–4 48(7):2043–2062 29. Zhu W, Franc¸ois R, Poon CS, Dai J-G (2017) Influences of 39. Lim S, Akiyama M, Frangopol DM, Jiang H (2017) corrosion degree and corrosion morphology on the ductility Experimental investigation of the spatial variability of the of steel reinforcement. Constr Build Mater 148:297–306 steel weight loss and corrosion cracking of reinforced con- 30. Apostolopoulos CA, Demis S, Papadakis VG (2013) Chlo- crete members: novel X-ray and digital image processing ride-induced corrosion of steel reinforcement—mechanical techniques. Struct Infrastruct Eng 13(1):118–134

Journal

Materials and Structures – Springer Journals

Published: Jun 14, 2018

There are no references for this article.