Rail Base Corrosion Detection and Prevention (2007)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

D-1 APPENDIX D “Estimation of the magnitude of the stresses formed due to corrosion cracks in railway rails” Final Report Submitted by Gabriel Plascencia, PhD & David Jaramillo, PhD To the Transportation Technology Center, Inc. CIITEC – IPN Cerrada Cecati s/n México, D.F. C.P. 02250 México e-mail: gplascencia@gmail.com

D-2 CONTENTS 1. Presentation D-3 2. Experimental D-4 3. Metallographic Evaluation D-8 4. Corrosion Tests D-11 5. Numerical Simulation D-19 6. Summary D-29 7. Conclusions D-29 8. References D-30 Acknowledgements D-30

D-3 Presentation The scope of this project is to evaluate alternative solutions for a constant material damage problem that underground transportation systems face across North America. Such problem is the deterioration of the rails due to a severe corrosion attack. In this report, we are presenting experimental results (metallographic characterization and corrosion tests) as well as some numerical calculations of the state of mechanical stresses that rails with and without corrosion damage present. The samples used to evaluate the metallurgical structure, and the corrosion resistance of the materials used in subway systems, were obtained from two different sources: 1) Mexico City’s subway system and 2) St. Louis, Missouri, Metro system. While the Mexican system provided two rail samples (one deformed due to its wear and a second one from a completely new rail), the second supplier only provided a sample severely corroded. The three samples were evaluated in the same manner both for their metallographic features as well as for their corrosion resistance. On the other hand, the numerical estimation of the mechanical stresses was conducted by means of the Finite Element Method analysis, using commercial software. It was found that in spite of the presence of non-metallic inclusions the materials showed structural features that allow them to perform as expected when a train passes on them. They also exhibited similar corrosion resistance regardless of the media in which they are tested. The numerical calculations showed that unless a sharp edge crack develops at the base of the rail, the material can withstand the loads applied to it without seriously compromising it performance under service.

D-4 Materials reception Optical inspection Sample cutting & machining Metallurgical Metallographic preparation Sample cutting & encapsulation for corrosion testing Surface cleaning for corrosion testing - Grinding with sand paper grades 80 to 1000 - Polishing with an alumina emulsion Corrosion tests - Tests in different media: KCl (1 & 0.1 M) Na2SO4 (1 & 0 1 M) Experimental As mentioned before, metallurgical examination and corrosion tests were conducted; parallel to these experiments, the numerical modelling of the rails was carried out. Figure D1 shows a schematic of how the different experimental activities related to this project were done. Figure D1. Experimental sequence in the evaluation of the rails samples. Two samples of rails were received from Mexico City’s Sistema de Transporte Colectivo (STC). One of the samples (SW) is from a worn rail which exhibits severe deformation on its head, whereas the second sample (SU) was taken from a new rail. In each case, the rails exhibited some corrosion products onto their surface; the corrosion products are due to the exposure to rain in the warehouse facilities of the STC. Figure D2 shows pictures of the rails in the as received condition. After initial inspection, the rust was removed from the surface of the rails by means of a mechanical brush.

D-5 Figure D2. Rails from STC in the as received condition. It was noticed after cleaning the rails, that both of them have some porosity, especially in their bottom end. The worn sample exhibited more porosity than the un-worn one. The initial inspection suggests that this porosity is due to the manufacturing of the rails (rolling operations); further examination of the rails will lead to a more conclusive evaluation of the source of such porosity. Figure D3 shows a picture of the bottom end of the SW sample.

D-6 Figure D3. Bottom end of the SW sample, showing some porosity. After receiving STC’s rails, we obtained a sample from St. Louis, Missouri, Metro system through representatives (Dr. Francisco C. Robles Hernández) of TTCI, Inc. This sample (USA) exhibited a severe corrosion attack on its base and web. Figure D4 shows this rail as received. Figure D4. USA rail sample in the as received condition. It is noticeable the uniform corrosion attack on the web and base of the USA rail Uniform attack Samples were taken from these locations

D-7 From the first two rails, two slices with 1 cm of thickness were taken. From the first slice 5 samples were cut to evaluate them metallographically; whereas from the second slice a sample was taken for corrosion tests. In the case of the USA rail sample, due to its condition, 2 samples were taken from its base, both samples were prepared for metallographic evaluation. After completing its structural characterization, one of the samples of this rail was used for the corrosion evaluation of this sample. Figure D5. Actual samples from each rail to analyze. Rail slice (1 cm thick) Samples to analyze from each slice

D-8 Metallographic Evaluation As mentioned in the previous section, 5 samples from each the SW and SU rails were evaluated, while only two samples from the USA rail was analysed. Every sample was prepared for metallographic examination firstly by girding them with sand paper (grades 80 to 1000) and then by polishing them with alumina emulsions (10, 5, and 1 μm). Once polished, the samples were analysed under an optical microscope to verify the size and level of non metallic inclusions. Figure D6 shows the latter. Using the NMX-B-308-1987 [1] and ASTM-E-45-1985 [2] standards the microstructures were evaluated to determine their microcleanlines As noticed in Figure D6, the SW sample shows grouped globular oxides type D2, whereas the SU sample shows discontinuous alumina inclusions type B4. The USA sample presents globular oxides type D1. It is likely that the globular oxides reported for both the SW and the USA sample actually are mainly alumina, which indicates that these three rails are made from killed steel. Now a days all rails are manufactured under vacuum treated casting conditions. After this initial inspection, the samples were etched with Nital 4 solution to verify their micro-structure. The microstructure of these materials is shown in Figure D7. It is clear from Figure D7 that the three samples exhibit the same microstructural features. All the samples have a homogeneous pearlitic matrix. Therefore is expected that both, the steel used in the USA or that used in Mexico would behave in the same manner under similar working and environmental conditions.

D-9 Figure D6. Non-metallic inclusions in the different rail samples: (a) SW, (b) SU and (c) USA (A) (B) (C)

D-10 Figure D7. Metallographic examination of the different rails samples: (A) SW, (B) SU, (C) USA. (A) (B) (C)

D-11 The average grain sizes for these materials are within 5 & 6 according to ASTM-E-45- 1985 [3] standard. The micrographs shown above were taken randomly since for every sample it was found the same type of microstructure in each of the 5 pieces evaluated (2 pieces for the USA rail). It must be mentioned that although the micrographs shown in Figures D6 and D7 are not at 100 X, the determination of the microstructure and the level of non-metallic inclusions were conducted under 100 X as required by the ASTM standards [2,3]. The metallographic examination confirmed that the porosity in the bottom of the base of the Mexican samples corresponded to the rolling stages in manufacturing them. No cracks attributed to the pores were detected nor any corrosion product deposited into the pores. After the initial polishing of the rail samples, all the rust from weathering disappeared, leaving the surfaces free of any residue or imperfection. Corrosion Tests The corrosion resistance of the steel used in the rails used in St. Louis, Missouri, Metro system was measured and compared to those installed in Mexico City’s subway facilities. The corrosion resistances of such steels were measured by means of the linear polarization technique. Again the three rails (SW, SU and USA) were tested. The linear polarization technique was chosen to measure the rate of corrosion because it is easily conducted while it allows for continuous data collection under different conditions. Since it takes only a few minutes to carry out one of these tests, the potential of the corroding metal is sufficiently stable during the test to act as a reference [4]. In our particular case, we carried out our corrosion tests by applying ± 20 mV than the corrosion potential of our reference electrode. In these tests we used a silver-silver chloride electrode as the reference one. The corrosion potential of such electrode is + 799 mV respect to the normal Hydrogen electrode at 25 °C. The corrosion tests were conducted using different electrolytes (1M KCl, 0.1M KCl, 1M Na2SO4, 0.1M Na2SO4), such electrolytes were chosen after reviewing the analysis of the soils in which US subway systems are installed. It should be mentioned that the aim of the corrosion tests was to evaluate the effect of ions such as SO42- and Cl- on the corrosion resistance of the rails.

D-12 The effect of such ions on the corrosion resistance of the rails are of importance due to the fact that soils across North America contain considerable amounts of sulfates, and in the other hand, the effect of the Cl- ion becomes also important for Subway systems located nearby the ocean. Figure D8, shows a picture of corrosion products in New York subway system. This system needless to say is installed in the shore of the Atlantic Ocean and the soil found there present considerable amounts of sulfate salts. Figure D8. Corrosion products found in New York’s subway system. As evident from Figure D8, the environment in which the rails are installed plays a very important role in terms of the corrosion resistance of the rails. In view of this, we conducted several tests in 4 different electrolytes. The experimental set up for the corrosion tests consisted of an electrolytic cell attached to a potentiostat – galvanostat apparatus, which in turn was connected to a CPU through a data acquisition system. The potentiostat used in our tests was an EG & G Princeton Applied Research apparatus model 273. This equipment has a built in corrosion software M352 which enables to automatically run the corrosion experiments while it collects data and sends it to a CPU. Figure D9, shows the experimental assembly. Rust from rails in New York’s subway system Corrosion

D-13 Figure D9. Experimental set up for corrosion tests. Reference Electrode (Ag/AgCl) Counter electrode (Graphite) Working electrode (sample) Electrolyte Potentiostat/Galvanostat Experimental Electrochemical cell

D-14 Figure D10. Probes used for corrosion testing. The picture shows the probes after being corroded in 0.1 M KCl electrolyte. As seen in Figure D9, every corrosion test consisted in immersing the sample to be evaluated into the different electrolytes. Once immersed, the voltage and current were applied during 20 minutes. Recording of the voltage drop and current density were taken at a scan rate of 20 readings/minute. At the end, for every single test, a total of 400 data set was obtained. At the same time every probe was tested 5 times in each electrolyte, so in total 60 experiments were conducted. The corrosion rate values reported for each sample in every electrolyte, corresponds to the average value of the different experiments. Figures D11 and D12 present such average values for tests in both KCl and Na2SO4. SU USASW

D-15 Figure D11. Corrosion resistance of the different rails in KCl. (A) 1M KCl electrolyte, (B) 0.1 M KCl electrolyte. 1E-08 1E-07 1E-06 1E-05 1E-04 -640 -620 -600 -580 -560 -540 -520 Potential (mV) C ur re nt D en si ty ( μ A /c m 2 ) Worn Unworn USA Linear Polarization test in 1M KCl @ 25 °C 1E-08 1E-07 1E-06 1E-05 1E-04 -650 -625 -600 -575 -550 -525 Potential (mV) C ur re nt D en si ty ( μ A /c m 2 ) Worn Unworn USA Linear Polarizat ion test in 0.1M KCl @ 25 °C (A) (B)

D-16 Figure D12. Corrosion resistance of the different rails in Na2SO4. (A) 1M Na2SO4 electrolyte, (B) 0.1 M Na2SO4 electrolyte. 1E-08 1E-07 1E-06 1E-05 1E-04 1E-03 -675 -650 -625 -600 -575 -550 -525 -500 Potential (mV) C ur re nt D en si ty ( μ A /c m 2 ) Worn Unworn USA Linear Polarization test in 1M Na 2 SO 4 @ 25 1E-06 1E-05 1E-04 -625 -600 -575 -550 -525 -500 Potential (mV) C ur re nt D en si ty ( μ A /c m 2 ) Worn Unworn USA Linear Polarizat ion test in 0.1M Na 2 SO 4 @ 25 °C (A) (B)

D-17 On the other hand, it can be seen in Figure D10 that the different samples are uniformly corroded by the different media tested. This also indicates that the steel used in the different rails behave quite similarly in every case. After reviewing the plots shown in Figures D11 and D12, it becomes more evident that the different specimens corroded at a very similar rate, there is only a slight deviation in terms of the voltage drop for every sample, however, the current density required for the corrosion of each specimen is in the same order of magnitude. The latter indicates that the materials tested present a similar corrosion resistance regardless of the medium in which they are evaluated. With the information from the plots in Figures D11 and D12, we were able to determine the current density for corrosion. Such current density is found by intercepting the slopes of the anodic and cathodic portions of each plot. Since the current density is directly proportional to the rate of corrosion, we were able to estimate such rate by means of the following equation [4]: nF jrcorr 0−= (1) Where jo is the current density for corrosion, n is the number of electrons transferred during the oxidation of the metal and F is Faraday’s constant (96500 C/mole). It must be noticed that equation 1 expresses the corrosion current as current density. Since there is no net reaction, since the rate of oxidation and reduction within the electrochemical cell are equal. Therefore the exchange reaction is equivalent to either the rate of corrosion or the rate of reduction, thus the corrosion rate can be conveniently expressed in terms of the current density [4]. Since we experimentally know the current density for corrosion, we can estimate the rate of corrosion in terms of mm/year, which is a more used expression to estimate the material losses due to corrosion. Table 1 shows such values. According to Table 1, it is clear that since the corrosion rate of the different samples is between 0.002 and 0.07 mm/year, thus it can be said that these materials exhibit excellent corrosion resistance except for the USA sample tested under 1M Na2SO4 solution, whose corrosion resistance value suggests that in sulfate media this sample only posses fair corrosion resistance [4]. The data obtained in Table 1, clearly shows that the steel used for this application is able to withstand the attack of different chemicals under normal conditions. If an over voltage is applied to the corroding system, then the rate of corrosion of the steel will increase quite significantly.

D-18 Table D1. Rate of corrosion of the different samples in every tested electrolyte. Electrolyte Sample i corr (A/cm2) Corrosion rate (mol/cm2/s) Corrosion rate (g/cm2/hr) Corrosion rate (mm/year) Worn 2.5x10-6 - 8.6 x10-12 - 1.7 x10-7 - 1.8 x10-3 Unworn 9.5x10-6 - 3.3 x10-11 - 6.6 x10-6 - 0.07 KCl 1M USA 2.5x10-6 - 8.6 x10-12 - 1.7 x10-7 - 1.8 x10-3 Worn 4x10-6 - 1.4 x10-11 - 2.8 x10-6 - 0.03 Unworn 6x10-6 - 2.1 x10-11 - 4.2 x10-6 - 0.04 KCl 0.1M USA 3x10-6 - 1.0 x10-11 - 2.1 x10-6 - 0.02 Worn 3.5x10-6 - 1.2 x10-11 - 2.4 x10-6 - 0.03 Unworn 5.5x10-6 - 1.9 x10-11 - 3.8 x10-6 - 0.04 Na2SO4 1M USA 9x10-5 - 3.1 x10-10 - 6.2 x10-5 - 0.64 Worn 2x10-6 - 7.0 x10-12 - 1.4 x10-6 - 0.01 Unworn 2.5x10-6 - 8.6 x10-12 - 1.7 x10-7 - 1.8 x10-3 Na2SO4 0.1M USA 3x10-6 - 1.0 x10-11 - 2.1 x10-6 -0.02

D-19 Numerical Simulation The estimation of the stresses generated in the rails with and without severe corrosion damage was done. Such estimations were conducted by means of the finite element method, using COMSOL Multiphysics™ software. The magnitude of the stresses estimated reveals that it is likely that the rails with severe corrosion attack will drastically fail under current subway traffic conditions if they develop very sharp edge cracks. This means that localized corrosion attack must be avoided. The calculations show that even in the event of uniform corrosion damage, the rails are still able to withstand the stresses developed under normal traffic conditions; therefore, replacing the damaged rails becomes a critical issue in terms of subway safety and maintenance programs. Results from these calculations are shown in the following figures. To solve the stresses equations, the mesh generated contained at least 15000 elements. To prove the accuracy of the method, mesh tests were conducted by doubling the number of elements. In that case, the results from the solver were identical to those with the initial mesh (it should be noticed that computer time ~ nodal points2). So in order to save computational time, we ran the simulations with a minimum of 15000 elements. More details on the simulations are shown in following.

D-20 (A) (B) (C) Figure D13. (A) Geometry and grid used in the calculation of mechanical stresses in a normal rail. (B) Tensile stresses developed in the rail when the load is applied. (C) Compression stresses in the rail when the load is applied. Along with the stresses, the figure shows flow lines along with the direction of the possible displacement. 5 cm Applied load

D-21 (A) (B) (C) Figure D14. (A) Geometry and grid used in the calculation of mechanical stresses in a corroded rail. (B) Tensile stresses developed in the rail when the load is applied. (C) Compression stresses in the rail when the load is applied. Along with the stresses, the figure shows flow lines along with the direction of the possible displacement. 5 cm Applied load Corrosion damage

D-22 Applied As seen from Figures D13 and D14, the effect of corrosion on the mechanical behaviour of the rails may have severe consequences. The stress components (tension and compression) in the stresses generated in the normal rail when the subway passes, show that they are under equilibrium and no mechanical failure is expected. On the other hand, the stresses developed in the corroded rail seem to create zones of compression and tension nearby the corroded area, such zones of tension or compression tend to accelerate or induce a catastrophic failure along the railway. The presence of cracks, especially those with sharp edge, result in a non uniform distribution of stresses in the vicinity of the crack. If high tensile stresses are developed in these cracks (stress concentrators), then it is likely that the cracks will propagate a faster rate. On the other hand if shear stresses develop alongside the cracks, then slip will occur [5, 6]. In either case, any stress system in which large tensile stress components combines with small shear stress components develops, will favour cleavage. Such stress state consideration is important when considering any possible fracture. On a different set of calculations, with a lateral view of the geometry, the statements above are more evident. Even with a uniform attack the material is able to withstand the loads applied, whereas the formation of sharp cracks encourages the failure of the rail when a load is applied. Figure D15. Actual situation to model. As seen in Figure D15, the new set of calculations was conducted taking in consideration the physical situation shown in the picture above. In this case 3 points of applied load will be considered. From the data of Mexico City’s survey, the load used in our calculations was in the order of 45000 N.

D-23 Figure D16. Numerical results for a load applied to an un-attacked rail, the loads are applied on top of every beam. (A) Shear stresses developed along with the deformation of the rail, (B) Tensile stresses developed along with the tensile strain. (C) Compression stresses developed along with the path of strain. (A) (B) (C)

D-24 Figure D17. Numerical results for a load applied to an un-attacked rail, the loads are applied on top of every beam except from that in the middle. (A) Shear stresses developed along with the deformation of the rail, (B) Tensile stresses developed along with the tensile strain. (C) Compression stresses developed along with the path of strain. (A) (B) (C)

D-25 Figure D18. Numerical results for a load applied to a rail with uniform attack, the loads are applied on top of every beam. (A) Shear stresses developed along with the deformation of the rail, (B) Tensile stresses developed along with the tensile strain. (C) Compression stresses developed along with the path of strain. As seen in this figure high compressive stresses develops along the crack formed, however these stresses do not affect drastically the mechanical behavior of the rails. (A) (B) (C)

D-26 Figure D19. Numerical results for a load applied to a rail with uniform attack, the loads are applied on top of every beam except from that in the middle. (A) Shear stresses developed along with the deformation of the rail, (B) Tensile stresses developed along with the tensile strain. (C) Compression stresses developed along with the path of strain. (A) (B) (C)

D-27 Figure D20. Numerical results for a load applied to a rail with a sharp crack, the loads are applied on top of every beam. (A) Shear stresses developed along with the deformation of the rail, (B) Tensile stresses developed along with the tensile strain. (C) Compression stresses developed along with the path of strain. (A) (B) (C)

D-28 As seen in the previous figures, it is evident that the nature of the cracks developed on the rail change the state of stresses generated. Although the tensile and compressive stresses developed in the rail are below the yield point of the material, the shear components of the stresses are in the vicinity to cause a catastrophic failure in the rails. As the crack become sharper, the magnitude of such shear stresses increase quite drastically along side the crack, thus it is expected a major failure in such type of cracks. Regarding to the uniform attack, it is clear that some shear stresses develops, however their magnitude is less than that observed in the sharp cracks, so this type of attack may not induce a severe failure as the sharp crack may, however, due to the material mass lost, it is necessary to remove the attacked rail and install a new one. As expected, the un-attacked rail is able to comply with the mechanical demands imposed by the loads under normal operating conditions.

D-29 Summary This investigation examined the variables that may affect the service life of rails in transit systems. Since the actual problem that these transportation systems are facing is related to severe corrosion damage, an analysis of the microstructural features of rails installed in different transit systems was conducted. In addition were conducted the measurement of the corrosion rates in different media. Parallel to these activities, the numerical estimation of the stresses generated into the rails with and without corrosion damage was conducted. Results reveal that besides being used in different transit systems and also apart from having different geometry, the rails analyzed in this work show similar microstructural features, which lead to the conclusion that they must perform similarly under the investigated conditions. In terms of the mechanical properties it is expected that both rails will behave similarly. Regarding to their corrosion resistance, the tests conducted in similar media shows that the materials tested exhibited an excellent corrosion resistance, except for one sample; it was also found that the corrosion resistance slightly decreased in the electrolytes with chlorine ions as expected, however, such ions did not corroded significantly the different samples. Numerical results show that unless corrosion cracks with sharp edge form, the rails would be able to withstand the mechanical stresses associated with the rail traffic. Even in the case that uniform corrosion takes place, the rails are able to perform well under normal load conditions. Conclusions The present investigation lead to conclude that under normal conditions, the steel used in the fabrication of rails would withstand the effect of the environmental conditions. However, the use of direct currents significantly affects the corrosion rate, which makes the rail less corrosion resistant. It means that when the rail is subjected solely to environmental corrosion in the absence of direct currents it can sustain the effect of the environment. In the presence of localized corrosion attacks combined with cyclic stresses creates proper conditions to induce fatigue-corrosion failures. Furthermore, the return current from the train to the rail and the improper insulation of the rails to the ground may form galvanic cells. This affects considerably the service characteristics of the rails as well as its integrity, increasing the susceptibility of the rail to fatigue corrosion.

D-30 Acknowledgements The authors wish to acknowledge the cooperation of Mexico City’s subway system, especially to Mr. Erasmo Cuatecontzi (B. Eng.) and Mr. Marco Mercado (B. Eng.). We also would like to thank TTCI for their support, especially to Dr. Francisco Robles (Gracias Paco). References [1] Norma Oficial Mexicana NMX-B-308-1987. Steel making industry-Methods for determining the inclusion content of steel. SECOFI, México, D.F., 1987. [2] ASTM-E-45-1985. Standard practice for determining the inclusion content of steel. ASTM, USA, 1985. [3] ASTM-E-112-2004. Standard test methods for determining average grain size. ASTM, USA, 2004. [4] Fontna M.G., “Corrosion Engineering 3rd Edition”, McGraw-Hill, New York, USA, 1987. [5] Dieter G.E., “Mechanical Metallurgy 2nd Edition”, McGraw-Hill, New York, USA, 1976. [6] Reed-Hill R.E., “Physical Metallurgy Principles 2nd Edition”, Brookes/Cole, New York, USA, 1973.