Study on the wheel/rail adhesion restoration and damage evolution in the single application of alumina particles

Study on the wheel/rail adhesion restoration and damage evolution in the single application of alumina particles SHI, L. B.; LI, Q.; KVARDA, D.; GALAS, R.; OMASTA, M.; WANG, W. J.; GUO, J.; LIU, Q. Y. Wear 2019, vol. 426-427, Part B, April 2019, pp. 1807-1819 ISSN: 0043-1648 DOI: https://doi.org/10.1016/j.wear.2019.01.021 Accepted manuscript © 2019. This manuscript version is made available under the CC-BY-NC-ND 4.0 license (http://creativecommons.org/licenses/by-nc-nd/4.0/), doi: 10.1016/j.wear.2019.01.021 Final version available from https://www.sciencedirect.com/science/article/pii/S0043164819300328 dspace.vutbr.cz 1 Study on the wheel/rail adhesion restoration and damage evolution in the single application of alumina particles L.B. Shi1, Q. Li1, D. Kvarda2, R. Galas2, M. Omasta2, W.J. Wang1*, J. Guo1, Q.Y. Liu1 1.Tribology Research Institute, State Key Laboratory of Traction Power, Southwest Jiaotong University, Chengdu 610031, China 2. Faculty of Mechanical Engineering, Brno University of Technology, Czech Republic Abstract: The aim of this study is to explore the wheel/rail adhesion restoration performance of alumina particles and the comparison between the continuous and single application test strategies in twin disc simulation. The results indicate that the single application test strategy performs better on evaluating the adhesion restoring ability of solid particles. In the single application tests, the restored adhesion coefficient increases with the applied alumina quantity and final gets stable when the quantity exceeds 2 g (0.607 g/m). However, the increase of alumina quality could always extend the duration of restored adhesion. The alumina particles at the size in the range of 0.075 to 0.15 mm showed the best adhesion restoration and the longest duration was achieved at the size of 0.15 mm. In addition, it is observed that the serious damage is mainly caused in the particles crushing process, and will be gradually ground off in the following long running to have no threat on the serving life of the wheel and the rail. Keywords: Adhesion restoration; Alumina; Quantity; Particle size; Wear damage 1. Introduction Sanding is a common adhesion restoration strategy widely used in the railway to overcome the poor adhesion problems. Low adhesion coefficient generally takes place when water [1-4], oil [4, 5], leaf [4, 6-8], the third-body layer [9-11] or other contamination [12, 13] 2 presents in the wheel/rail interface preventing the contact between metal micro asperities in the wheel/rail interface. Sand particles create new contact asperities in the wheel/rail interface to provide the mechanical shearing force in the relative motion of the wheel and the rail surfaces [14, 15]. Besides sand, other solid particles with higher hardness than the wheel/rail steels can also help to improve the adhesion coefficient under low adhesion conditions [16, 17]. Among them, alumina particles have been proved to have better adhesion restoration performance than sand and have been used practically in Japan [18]. Although some parameters of sanding have been stated in some railway standards [19, 20], most of them are loose rules and some of them are not the same in different countries [21]. There have been many studies in the lab or field to evaluate the adhesion restoring performance of solid particles. The particle size and feeding rate were generally studied to estimate their effects on the adhesion restoring ability [4, 15, 17, 21-28], wear and damage deterioration [15, 17, 21, 22, 24-26] and contact conductance [29]. In the previous studies, except for Arias-Cuevas’s [22] field and Lewis’s [23] linear full-scale studies, most of them were conducted on some largeor small-scale twin disc rigs. They adopted the continuous or single application test strategy to simulate the practical sanding process. In the continuous application simulation tests, sand or other solid particles are applied into the rolling contact interface continuously at a small-scale sand flow rate (generally no more than 50 g/min) [15, 17, 18, 24, 25] taking no sand deposition rate into consideration or a larger one to meet a desired deposition rate (generally use the practical rate of 7.5 g/m) on the discs surface [4, 21, 26, 29]. In the single application tests, the dose of applied particles is also designed from the expected deposition rates [27, 28]. However, the continuous application strategy uses the 3 discs rolling distance per minute or second to design the required flow rate, while the single application just takes the disc circumference as reference. Considering the different motion and scale between the wheel/rail and simulation discs, the strategy adopted in the experiments to entrain the solid particles into the contact interface seems to be crucial for the measured results. This is why different conclusions might be drawn in different researches. However, there was not a systematic study to compare which type of experiments is better. Arias-Cuevas had ever adopted both of these two strategies in his twin disc sanding studies under leaf condition [21, 28]. In his continuous application study, the measured influence of particle size on the adhesion recovery was opposite to the field [22] and single application tests results [28], which seemed to be caused by the too rapid remove of leaf layer by the continuous entrainment of sand particles. To better clarify such an uncertain effect caused by the different test strategies, there should be some more comparative studies to take other simulated sanding parameters, surface conditions and damage into consideration. This work uses the alumina particles as adhesion enhancer in the adhesion restoration tests. At first, both continuous and single application tests will be conducted for comparison. Then, the effect of particle size and quantity of alumina particles on the adhesion restoration ability and duration will be explored. In the end, a new perspective on the surface damage evolution will be drawn using the single application test strategy. 2. Experimental details 2.1 Experimental apparatus 4 A JD-1 wheel/rail simulation facility was used to carry out the adhesion tests, shown in Fig.1. This facility has been frequently used in previous studies to reveal the wheel/rail adhesion characteristics under various conditions [3, 30]. There is a small disc with a diameter of 210 mm serving as the wheel roller and a large disc with a diameter of 1050 mm serving as the rail roller. The contact between the wheel and rail rollers is loaded hydraulically to achieve a preset normal load and the speed difference of the wheel/rail rollers is controlled to achieve the desired creep ratio. The adhesion coefficient is described as the ratio of the tangential friction force and normal force in the contact. (a) (b) Fig.1. Scheme of JD-1 wheel/rail simulation facility and rollers, (a) JD-1 wheel/rail simulation facility; (b) DC motor DC motor Wheel roller Rail roller Vertical loading cylinder 5 wheel/rail rollers. Tests for wear and damage evolution were conducted on an MJP rolling-sliding apparatus, which uses two 55 mm discs cut from the wheel tread and rail head serving as the wheel and rail rollers, respectively. Schematic structures of this machine and rollers are shown in Fig.3. The wheel/rail rollers are driven by two individual motors and the normal force of the rollers is loaded by a hydraulic system. The creep ratio could be realized by adjusting the rotational speed of the wheel roller to reach a precise difference. More details about this facility could be found in [31]. Servo motor Driving belt Bearing housing Shaft coupling Torque sensor Hydraulic cylinder and Load cell Servo motor Rail roller Wheel roller Bearing block Lever arm (a) (b) Fig.2. Scheme of MJP rolling-sliding wear and contact fatigue apparatus and rollers, (a) MJP rolling-sliding 6 wear and contact fatigue apparatus; (b) wheel/rail rollers. For the application of adhesion enhancers, a developed sanding system similar to the real sander box was used to blow the particles into the contact interface in the rolling direction, as shown in Fig.1b. By adjusting the compressed air and the valve of the nozzle, different flow rates could be achieved in the continuous application. For the single application of different quantity, a certain amount of alumina particles was put into the sanding hose in advance and then blown out of the hose during a fixed period of approximately 1 second. 2.2 Experimental parameters and procedure In this study, the creep ratio in both adhesion and wear tests was set as 5%, at which the macro sliding or spinning of wheels would be determined in the field and sanding would be triggered automatically to prevent these problems. The creep ratio in the tests is defined as the following formula. (1) Where, ω rail and ω wheel are the rotational speed of rail and wheel rollers; r rail and r wheel are the radius of the wheel and rail rollers, respectively. To achieve the desired creep ratio, the rolling speed of the rail roller was set as 120 rpm and the wheel roller speed was controlled to 570 rpm on JD-1 wheel/rail simulation facility. While the speed of wheel and rail rollers on MJP rolling-sliding apparatus was set as 190 and 200 rpm. The normal load applied in adhesion and wear tests were 1760 and 410 N, respectively, simulating the maximum wheel/rail contact pressure of 1100 and 1500 MPa. In the tests, water kept dropping on the rail surface to simulate low adhesion condition and the alumina particles with Al2O3 content more than 96% were used as the adhesion 7 enhancers. The alumina particles had been well sieved into different sizes before the tests. The wheel and rail rollers on JD-1 wheel/rail simulation facility are made of the wheel and rail steel applied in the field. Their chemical compositions could be seen from [30]. The rollers used on MJP rolling-sliding wear and contact fatigue apparatus were directly cut from the tread of a CL60 wheel and the head of an U71Mn rail, which are similar with the material of JD-1 rollers. The chemical compositions and hardness of the wheel/rail rollers are given in Table 1. Table 1. Chemical compositions and hardness of wheel/rail rollers (wt %). Roller C Si Mn P S Hardness/HV0.2 Wheel 0.55~0.65 0.17~0.37 0.58~0.80 ≤0.04 ≤0.045 310 Rail 0.62~0.77 0.15~0.37 1.35~1.65 ≤0.04 ≤0.050 287 First, the alumina particles with different flow rates (ranging from 1.5 to 500 g/min) were applied continuously into the wheel/rail interface under wet condition, in order to study the effect of flow rate on the improved adhesion characteristics, as listed in Table 2. Then, tests were conducted by single application of alumina particles to explore the effect of particle size (ranging from 0.02 to 0.8 mm) and applied quantity (ranging from 0.03 to 25 g) on the improved adhesion coefficient and the duration of restored adhesion. More details about the tests have been listed in Table 3. These tests were repeated twice in the study to ensure the results are reliable and repeatable. The single application tests of 2 g alumina particles at the size of 0.3 mm were also conducted on the MJP to trace the evolution of wear damage in the process of before-when-after applying the alumina particles. Four pairs of wheel/rail rollers started from the running-in process with fresh surfaces but would be stopped at different stages, as shown in Fig.3. Furthermore, the morphologies of the worn surface and subsurface 8 of the rollers were observed using optical microscopy (OM) (OLYMPUS, Japan) and scanning electron microscope (SEM), (JSM-7001F, Japan). Table 2. Details for the adhesion tests with continuous application of sand/alumina particles. Condition Particles Size/mm Flow rate/g/min Duration Wet/5ml/min Alumina 0.3 1.5, 12, 30, 60, 125, 200,500 240 Cycles Sand 1.5 Table 3. Details for the adhesion tests with single application of alumina particles. Condition Particle size/mm Quantity/g Duration Wet/5ml/min 0.3 0.03, 0.2, 0.5, 1, 2, 7, 14, 25 Last until the improved adhesion coefficient drops to the low level again 0.02, 0.038, 0.075, 0.1, 0.15, 0.3, 0.4,0.8 2 Wet/1ml/min 0.3 2 Wet/5ml/min Oil/1ml/min 02000 4000 6000 8000 0.0 0.1 0.2 0.3 0.4 0.5 0.6 CD B Adhesion coefficient μ Cycles Apply 2g alumina with size of 0.3mm A Fig.3. Wear tests stopped at different stages (A, B, C and D) in the process of before-when-after applying the alumina particles. 3. Results 3.1 Adhesion coefficient in the continuous application of alumina particles 15 quantity of 2 g). It is worth noting that when the applied alumina particles are infinitesimal (0.03 g, Fig.6a) or micro (0.02 mm, Fig.8a), the peak adhesion coefficient μ1 is almost same with the following stable adhesion coefficient μ2. The decreasing process from μ1 to μ2 could not be distinguished, although the actual quantity and particle size of the particles at these two stages are certainly different. That indicates the adhesion restoring ability of the particles won’t be remarkable when either of the size or quantity is too small. 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 1.0 μ2 Adhesion coefficient μ Particle size/mm μ1 Fig.9. Effect of the particle size on μ1 and μ2. 3.3 The duration of adhesion restoring effect after single application The variation processes of adhesion coefficient from the immediate increase to re-dropping to low level in different single application tests have also been recorded in Fig.10. Fig.11 gives an example to introduce the division of the different stages in the restored adhesion duration and to illustrate the good repeatability of tests. Three stages are distinguished in the adhesion restoring process: the first short duration t1 with the immediate increase and rapid decrease of adhesion coefficient, the second stage t2 with weak adhesion restoring effect but long duration and the third stage t3 with the disappearance of the adhesion 16 restoring effect. As summarized in Fig.12, both the applied quantity and particle size have a large effect on the duration of different stages. The duration of the whole process t0 increases with the applied quantity, but the increase slows down after the quantity exceeding 2 g. This trend could also be observed in the variation of duration of the second stage (t2). However, the duration of the first (t1) and third stage (t3) seem to gets stable after the applied quantity exceeding 2 g. As shown in Fig.12b, the maximum duration of t, t1, t2 and t3 are all achieved at the size of 0.15 mm. The duration increases with the grain size from 0.02 to 0.15 mm and then descends with the continuously increasing particle size and finally tends to be stable after the size exceeding 0.3 mm. 0.35 0.35 0.35 0.35 0.35 0.35 0.35 0500 1000 1500 2000 2500 3000 0.35 0.03g 0.2g 0.5g 1g 2g 7g 14g Adhesion coefficient μ 25g Cycles 0.35 0.35 0.35 0.35 0.35 0.35 0500 1000 1500 2000 0.35 0.35 Adhesion coefficient μ 0.02mm 0.1mm 0.38mm 0.75mm 0.3mm 0.42mm 0.8mm Cycles 0.15mm (a) (b) Fig.10. The whole variation process of adhesion coefficient after the application of alumina particles to the disappearance of the adhesion restoring effect, (a) with different quantities in the size of 0.3 mm; (b) in different sizes with the quantity of 0.3 g. 17 0400 800 1200 1600 2000 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 t3 t0 t1t2 μ2 μ0 μ1 Adhesion coefficient μ Cycles μ0 Fig.11. An example of the variation process to present the division of different durations (applied quantity: 2 g, particle size:0.3 mm) 0 4 8 12 16 20 24 28 0 500 1000 1500 2000 2500 3000 3500 t3 t1 t2 Duration/cycles Quantity/g t0 0.0 0.2 0.4 0.6 0.8 0 500 1000 1500 2000 2500 3000 t3t1 t2 Duration/cycles Particle size/mm t0 (a) (b) Fig.12. Effect of the applied quantity and particle size on the duration, (a) applied quantity; (b) particle size. An interesting phenomenon is found in Fig.13a that the duration of restored adhesion could be shortened significantly under strong water flow condition. The restored adhesion coefficient μ1 is almost identical in different water flow cases, while the following adhesion coefficient μ2 is obviously larger in the smallest water flow case than that in other cases. In addition, when the alumina particles are applied under oil condition, as shown in Fig.13b, restored adhesion coefficient μ1 is below 0.4, far lower than that under water condition. Moreover, there is only the first stage t1 could be found after the application of alumina 18 particles, meaning the adhesion restoration ability could not be held under oil condition. This may be caused by the larger film thickness at the contact interface under oil conditions. 0400 800 1200 1600 2000 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 1672 1520 Adhesion coefficient μ Cycles Water flow 1ml/min Water flow 5ml/min Water flow 20ml/min 1185 0100 200 300 400 500 0.0 0.1 0.2 0.3 0.4 0.5 Adhesion coefficient μ Cycles t1 (a) (b) Fig.13. The variation process of adhesion coefficient after the application of alumina particles under different low adhesion conditions (applied quantity: 2 g, particle size: 0.3 mm), (a) wet conditions with different water flows; (b) oil condition. 3.4 Damage evolution in the adhesion restoring process This part presents the surface damage of discs after a single application of alumina particles in the adhesion restoring process. The surface damage of wheel rollers at different stages is presented in Fig.14. Different from the slight damage produced after 5 000 cycles running under wet condition at stage A, the surface gets very rough and is covered with numerous embedded alumina particles at stage B, at which the maximum adhesion coefficient has just been achieved after the application of alumina particles. Furthermore, it could be observed most of the alumina particles have been removed away from the rolling track from stage B to stage C (less than 400 cycles), although the surface is still visibly rough. In the end of process (stage D), when the adhesion coefficient has dropped to the wet level, the surface 19 on the track recovers to be smooth again, while the rough surface with embedded alumina particles could still be found on the sides of the track. Fig.14. The surface topographies of wheel rollers at different stages. As shown in Fig.15, the surface wear damage is obviously aggravated after the application of alumina particles from stage A to stage B, due to the crushing of alumina particles at the contact interface. During the crushing process, initial particles with larger size will be crushed into lots of micro fragments when the normal load it withstands exceeds the particle strength. In addition, particles with sharp corners would be compressed into the wheel/rail materials for much higher hardness of alumina, leaving indentions and embedded fragments on the surface, as shown in Fig.15b. After a period of the mechanical remove of the crushed alumina particles in the rolling, only some micro particles, that have been embedded deeper, could be kept on the surface at stage C (Fig.15c). Thus, there are less alumina particles remaining on the track. In addition, it could also be found in the left figure of Fig.15c that there is a singular surface with darker color, which is peeling away with numerous micro cracks on it. When at stage D (Fig.15d), there are almost no embedded alumina particles and old surface found on the rolling track except in the border of the track, at which the 02000 4000 6000 8000 0.0 0.1 0.2 0.3 0.4 0.5 0.6 CD B Adhesion coefficient μ Cycles Apply 2g alumina with size of 0.3mm A 1mm A B D C 1mm 1mm 1mm 20 mechanical shearing is weakest owing to the pressure distribution. (a) (b) (c) Embedded micro fragments Embedded alumina particles Small micro pits Crushed micro particles Removing material Indention 50μm Numerous micro cracks Peeling Indentions and embedded alumina particles Rolling track Track side Track side Rolling track Track side Track side Remaining embedded alumina particles Rolling track Track side 500 μm 50 μm 50 μm 500 μm 20 μm 500 μm 21 (d) Fig.15. SEM micrographs of worn surfaces of wheel rollers at different stage (left: overall perspective; right: details), (a) stage A; (b) stage B; (c) stage C; (d) stage D. OM micrographs of longitudinal sections of wheel rollers at different stages are illustrated in Fig.16. There are no obvious cracks and plastic deformation on the subsurface at stage A. However, just a few cycles later after the application of alumina particles, obvious plastic deformation and embedded alumina particles could be observed at stage B. Except leaving indentation pits on the surface, another important adverse effect caused by the embedded particles is the severe plastic deformation of the material surrounding them, as shown in Fig.16b marked with red dot lines. The material over embedded particles will be highly strained in the rolling contact and get easy to be cut off, as shown in the picture attached in the lower left corner of Fig.16b. Therefore, lots of defects could be observed on the subsurface at stage C after most of the embedded alumina particles have been removed. Some micro embedded fragments with sizes less than 20 μm could still be found on the surface or at the bottom of the indentations. Most importantly, the material at the top of the crack-like cavies has been highly strained to be cut off and this could be correlated to the peeling of the surface with numerous micro cracks presented in Fig.15c. At the end of the Rolling track Side of the track Peeling of old surface Rolling track Track side Track side Embedded alumina particles 500 μm 100 μm 22 process (stage D), almost all the embedded alumina fragments and the highly strained layer have been removed and there are only some shallow pits and micro cracks could be found on the subsurface, as shown in Fig.16d. Therefore, it could be inferred that serious damage caused by the application of alumina particles is mainly produced in the crushing process and the damage will get alleviated in the following long-time running. (a) (b) (c) (d) Fig.16. OM micrographs of longitudinal section of wheel rollers at different stage, (a) stage A; (b) stage B; (c) stage C; (d) stage D. 4. Discussion 50 μm 50 μm 50 μm 50 μm 25μm 25 μm 25μm 25 μm 70 μm Highly strained material Embedded particles Embedded particles Shallow pits Shallow pit Micro crack Broken surface material Highly strained material Left cavities Left cavities 23 4.1 Comparison of continuous and single application of particles in the laboratory This study has firstly explored the adhesion restoration of alumina particles using both continuous and single application testing strategies on the JD-1 wheel/rail test rig. Table 4 lists the corresponded sand deposition rates in these two strategies, which are calculated based on the different customs used in previous continuous [21, 26] and single application studies [27, 28]. It is obvious the results coming from comparing these two different test strategies are distinct. In the continuous application, the restored adhesion coefficient increases with the deposition rate until reaching the threshold of 0.076 g/m (30 g/min), above which the adhesion restoring ability of solid particles is weakened significantly. In the single application, there is also a threshold of the deposition rate. However, this threshold is much larger (0.67 g/m) and the adhesion restoring ability gets stable when the deposition rate further increases from this threshold to the limited 7.5 g/m. This trend seems to be more approximate to the practice and Lewis’s linear full-scale test results [23]. In the continuous application, there is always a part of the applied material remaining on the discs surface and coming into the contact repeatedly. So, the actual quantity of the active particles at the interface increase gradually and the true deposition rate becomes much larger than expected. At low flow rates, the elevating particles density helps to increase the adhesion coefficient and will lead its adhesion restoration overestimated. While, at high flow rates, this elevation will easily make the contact overdosed. Then a compacted crushed particles coating will be formed by the excessive solid particles in the contact, diminishing the adhesion restoring ability [27, 28]. In this case, the achieved effect could not be correlated to the actual tested deposition rate, because the wheels just pass through the laid particles once in the field. 24 In the single application tests, as the application of alumina particles takes a short time (less than 2 cycles), the followed improvement of adhesion coefficient could be directly attributed to the applied quantity. And considering the remove of particles in this transient process will not loss too much, it could be regarded such the applied particles are uniformly deposited on the disc circumference. Therefore, the improved adhesion coefficient could be correlated to a relatively precise deposition rate used in the field and the appropriate deposition rate achieved by this test strategy could be more convincing to guide the field application. Table 4. Comparison of the adhesion restoration using continuous and single application of alumina particles (particle size: 0.3 mm). Continuous application Single application Flow rate/g/min Expected sand deposition rate/g/m Average adhesion coefficient Applied quantity/g Expected sand deposition rate/g/m Adhesion coefficient 1.5 0.004 0.41 0.03 0.009 0.31 12 0.030 0.54 0.2 0.061 0.40 30 0.076 0.56 0.5 0.152 0.42 60 0.152 0.43 1 0.303 0.45 125 0.316 0.44 2 0.607 0.58 200 0.506 0.4 7 2.123 0.65 500 1.264 0.3 14 4.246 0.65 - - 25 7.582 0.64 Furthermore, sanding is carried out sporadically in the field and the wheels would pass through the deposited sand just for a short time in its serving life. Previous investigations on the wheel/rail wear and RCF damage during application of sand or alumina particles were mainly conducted in continuous application tests [15, 17, 21, 24-26]. The results in Section 31 Transactions of the ASME, 108 (1986) 141-147. [16]R. Galas, D. Kvarda, M. Omasta, I. Krupka, M. Hartl. The role of constituents contained in water–based friction modifiers for top–of–rail application, Tribol. Int. 117 (2018) 87-97. [17]W.J. Wang, T.F. Liu, H.Y. Wang, Q.Y. Liu, M.H. Zhu, J.X Jin, Influence of friction modifiers on improving adhesion and surface damage of wheel/rail under low adhesion conditions, Tribol. Int. 75 (2014) 16-23. [18]K. Ohno, T. Ban, T. Obara, K. Kawaguchi, Development a new method for adhesion improvement replacing traditional sanding, Proceedings of the international heavy haul association, 1994, pp. 57. [19]RSSB: GM/RT2461, Sanding Equipment Fitted to Multiple Units and On-Track Machines, Issue One: August 2001, Rail Safety and Standards Board, Block2, Angel Square, 1 Torrens Street, London EC1V 1NY. [20]TB/T 3254-2011, Sanding device of locomotive and motor car, Standards of the Railway Industry of the People's Republic of China. [21]O. Arias-Cuevas, Z. Li, R. Lewis, A laboratory investigation on the influence of the particle size and slip during sanding on the adhesion and wear in the wheel-rail contact, Wear 271 (2001) 14-24. [22] O. Arias-Cuevas, Z.L. Li, Field investigations into the adhesion recovery in leaf-contaminated wheel–rail contacts with locomotive sanders. Proc. ImechE, Part F: J. Rail Rapid Transit, 225 (2011) 443-456. [23]S. Lewis, S. Riley, D.I. Fletcher, R. Lewis, Optimisation of a railway sanding system, Part2: Adhesion Tests. in: Proceedings of the 10th International Conference on Contact 32 Mechanics and Wear of Rail/Wheel System (CM2015), Colorado, USA, 2015. [24] W.L. Huang, X. Cao, Z.F. Wen, W.J. Wang, Q.Y. Liu, M.H. Zhu, X.S. Jin. A subscale experimental investigation on the influence of sanding on adhesion and rolling contact fatigue of wheel/rail under water condition. Journal of Tribology, Transactions of the ASME, 2017, 139(1):11401-11408. [25] X. Cao, W.L. Huang, C.G. He, J.F. Peng, J. Guo, W.J. Wang, Q.Y. Liu, M.H. Zhu, The effect of alumina particle on improving adhesion and wear damage of wheel/rail under wet conditions, Wear 348-349 (2016) 98-115. [26] R. Lewis, R. S. Dwyer-Joyce, Wear at the wheel/rail interface when sanding is used to increase adhesion. Proc. ImechE, Part F: J. Rail Rapid Transit, 220 (2006) 29-41. [27] M. Omasta, M. Machatka, D. Smejkal, M. Hartl, I. Křupka, Influence of sanding parameters on adhesion recovery in contaminated wheel–rail contact, Wear 322-323 (2015) 218-225. [28] O. Arias-Cuevas, Z.L. Li, R. Lewis, E.A. Gallardo-Hernández, Laboratory investigation of some sanding parameters to improve the adhesion in leaf-contaminated wheel–rail contacts. Proc. ImechE, Part F: J. Rail Rapid Transit, 308 (2010) 139-157. [29] O. Arias-Cuevas, Z. Li, R. Lewis. Investigating the lubricity and electrical insulation caused by sanding in dry wheel-rail contacts. Tribol. Lett. 37 (2010) 623-635. [30] W.J. Wang, H.Y. Wang, J. Guo, Q.Y. Liu, M.H. Zhu, X.S. Jin. Experimental investigation of adhesion coefficient of wheel/rail under the track ramp conditions. Proc. ImechE, Part J: J. Engineering Tribology, 228 (2014) 808-815. [31] W.T. Zhu, L.C. Guo, L.B. Shi, Z.B. Cai, Q.L. Li, Q.Y. Liu, W.J. Wang, Wear and damage 33 transitions of two kinds of wheel materials in the rolling-sliding contact, Wear 298-299 (2018) 79-89. [32] M. Faccoli, C. Petrogalli, M. Lancini, A. Ghidini, A. Mazzu, Effect of desert sand on wear and rolling contact fatigue behaviour of various railway wheel steels, Wear 396-397 (2017) 146-161. [33] Y. Nakata, M. Hyodo, A.F.L. Hyde, Y. Kato, H. Murata, Microscopic particle crushing of sand subjected to high pressure one-dimensional compression, Soils Found, 41 (2001) 69-82.