Novel cross wedge rolling method for producing railcar axles

Railcar axles are large-size parts manufactured in large batches using forging methods. More efficient methods of manufacturing railcar axles are currently being sought. A promising line of research in this respect is connected with the use of cross wedge rolling (CWR). Nonetheless, there exist no rolling mills that would make it possible to manufacture such large-size parts as railcar axles. A solution to this problem was inspired by the symmetric shape of a railcar axle, which makes it possible to roll this part in two passes using the same tools. Taking advantage of the potential offered by numerical modelling, simulations were made of two rolling processes for producing a railcar axle type BA302, i.e. one process was conducted with the use of two rolls and the other with two flat tools moving in the opposite directions. Numerical results demonstrated that railcar axles produced by the proposed rolling process had neither external nor internal (cracks) defects. The effectiveness of the developed CWR method was ultimately confirmed by the high quality of railcar axles that were produced (in a scale of 1:5) under industrial conditions at the Lublin University of Technology.


Introduction
The development of railways in the nineteenth century led to a great demand for railcar axles. These parts were at first produced by open die forging conducted on steam hammers or hydraulic presses, but it was not long before they came to be manufactured by rolling. The first machine for rolling car axles was patented by Cooper in 1865 [1]. This machine had two rolls with eccentric working surfaces. In 1885, the American inventor and industrialist Simonds patented a rolling machine with two flat wedges moving in the opposite directions [2] and wedge tools for rolling car axles [3]. In light of today's knowledge, it can be claimed that the tool design he proposed was correct. Since in his Fitchburg-based plant Simonds had at his disposal only rolling machines that made it possible to produce parts with a maximum length of 12 inches, he therefore decided to roll smaller axles. The obtained axles were presented at the stated meeting of the Franklin Institute in 1888 where he presented his universal metal rolling machine [4]. In 1892, Slick patented a method for the manufacture of axles with the use of three rolls [5]. Nevertheless, the tool design did not make it possible to roll axles of the required shape. A more effective tool design for the manufacture of axles was proposed by Schneider [6], whose invention assumed that axles would be formed in two separate operations. The first one consisted of shaping the axle body, while the axle journals would be shaped in the other one. The above solutions were not, however, applied in practice due to overall dimensions of railcar axles (they were over 200 mm in diameter and over 2000 mm in length), which meant that their production required using machines of very large size.
The idea of producing railcar axles by cross wedge rolling (CWR) resurfaced again after over a century. Despite a rapid development of the CWR method in the twentieth century [7,8], there existed no rolling mills that would allow the production of railcar axles by standard CWR. For that reason, studies were undertaken on synchronous cross wedge rolling in which the workpiece was deformed by several pairs of wedge tools at the same time. This solution made it possible to reduce tool length (and thus their diameter) to the acceptable value. Shu et al. [9] conducted a FEM study on synchronous cross wedge rolling of solid railcar axles using rolls with a diameter of 1600 mm. They also conducted experiments on producing railcar axles in a scale of 1:5, demonstrating that the axles produced thereby had numerous internal and external defects. Sun et al. [10] conducted a FEM study on synchronous rolling of hollow railcar axles with the use of three pairs of wedges. A similar numerical study of the rolling process conducted with the use of five pairs of wedge tools was conducted by Hu et al. [10]. Obtained numerical results were satisfactory. Nevertheless, experiments conducted by Peng et al. [11] demonstrated that the above numerical solutions were ineffective (the problem of excessive cross-sectional ovality of axles was not solved). Similar results were obtained by Zheng et al. [12]. A study by Bulzak [13] demonstrated that the probability of internal crack formation was considerably higher in synchronous cross wedge rolling than in the standard CWR process.
Due to failure modes occurring in CWR conducted by the synchronous method, the standard CWR method became of interest to researchers again. Pater and Tomczak [14] proposed a solution in which axles would be rolled in two operations, with the axle body shaped in the first operation and axle journals in the other. This solution made it possible to decrease the diameter of the rolls to an acceptable value of 1200 mm, as well as to considerably reduce the power consumption of the rolling mill. A similar solution in which the rolling process was conducted with the use of three rolls was presented in [15]. A shortcoming of the above solutions is that their implementation requires the construction of new special machines. It is worth mentioning that the use of wedge tools with convex working surfaces can also bring positive results. According to Pater [16], this solution makes it possible to produce railcar axles via the standard rolling process conducted using rolls with a nominal diameter of 1800 mm. What is more, the use of such tools reduces the risk of internal crack formation in the axial region of the workpiece.
As a result of the development in CWR machine design, it is nowadays possible to produce parts with their diameter as large as that of railcar axles. The China-manufactured largest rolling mills (type D46x1600) are used for producing axisymmetric parts with diameters of up to 180 mm and lengths of up to 1300 mm. The largest flat-wedge rolling mill (type WRL20035TS) is provided with two tools moving in the opposite directions, each tool with a length of 3.5 m. This machine can produce parts with diameters of up to 200 mm and lengths of up to 1200 mm. Although these diameters are similar to the diameter of a railcar axle (approx. 220 mm), the length of a standard axle is almost twice the length of parts that can be rolled on the above-mentioned rolling mills. To remedy this problem, a novel CWR method for producing railcar axles in two passes, using the same tools, was developed at the Lublin University of Technology. In the first pass, the tools shape the journals on one side of the axle and half of the axle body. After that, the workpiece is rotated by 180°, and the remaining portion of the body is shaped in the second pass of the tools. This paper reports the results of a study undertaken to assess the effectiveness of this novel CWR method for producing railcar axles. The study involved performing numerical simulations of the CWR process conducted with the use of both two rolls and two flat wedges, as well as carrying out experiments in which railcar axles were produced in a scale of 1:5, which was due to the size of the rolling mill available in the laboratory.

Numerical analysis
The object of this study was a railcar axle of type BA302. A rolled axle with machining allowance set to 5 mm for diameter and 10 mm for end face is shown in Fig. 1.
A wedge roll used for producing the above-mentioned railcar axle is shown in Fig. 2. The tool has a roll face with a diameter of 1180 mm and a length of 1450 mm. The spacing between the axes of two mating rolls is 1400 mm. There are two wedges on the roll face. The first one is described by a spreading angle β = 6.8° and a forming angle α = 30°; this wedge is used for shaping the axle journals. The other one is described by the angles β The above values for the angles α and β were chosen arbitrarily so that the following relationship is fulfilled: 0.04 ≤ tgαtgβ ≤ 0.08 used in the design of CWR tools. In the early stage of the rolling process, the wider wedge has an undercut at an angle of β = 1.6° in order to achieve a higher-quality end face between the axle journals. The roll is also provided with a side cutter for cutting off discard material and two guide paths for maintaining the workpiece in position in the early stage of the rolling process.
A flat wedge tool with the overall dimensions of 1450 × 3500 mm is shown in Fig. 3. This tool also has two wedges described by the angles β = 7° and α = 30° (for shaping the axle journals) as well as β = 15° and α = 15° (for shaping the axle body). On its end, the narrower wedge has a 600 mm long side cutter for cutting off defective material on one end of the workpiece.
The above dimensions of the wedge tools are either smaller or equal to the dimensions of the previously mentioned rolling mills D46x1600 and WRL20035TS, which means that the proposed rolling process can be performed under industrial conditions.
The designed tools were used to develop two geometrical models of CWR which are shown in Fig. 4. Each model consisted of a billet described by a diameter of 216 mm and a length of 1650 mm, two tools (rolls or flat wedges), and guides for maintaining the workpiece in stable position during the rolling process. The billet was made of 42CrMo4 steel, the material model of which was taken from the material database library of ForgeNxT v.1.1, and it was preheated to a temperature of 1240 °C. The temperature of the wedge tools was 250 °C, the guides had a temperature of 350 °C, and the heat transfer coefficient was set equal to 10,000 W/m 2 K. The rotational speed of the rolls was set to 3 rev/min, whereas the flat wedges moved in the opposite directions with a speed of 175 mm/s each. Friction on the contact surface was described by the Tresca model, with the friction factor value set to 0.9 for the wedge and 0.4 for the guide. Changes in the shape of the workpiece in the CWR process conducted with the use of two rolls are illustrated in Fig. 5. In the first operation (which corresponds to the first revolution of the rolls), the tools shape journals on one end of the axle and slightly more than a half of the axle body. After that, the workpiece is rotated and put between the rolls again. The second operation (which corresponds to the second revolution of the rolls) consists of shaping journals on the other end of the axle and the remaining portion of the axle body. Excess material (defective ends) is cut off by the side cutters in the final stage of each operation. No significant differences in the forming process can be observed after tool change, from rolls to wedges (Fig. 6). In both cases, the produced railcar axle has the required shape, and the rolling process is not disturbed by the occurrence of uncontrolled slipping. Figure 7 shows one of the railcar axles obtained from the numerical simulation. This figure also shows the cross-sectional shapes of different zones on the axle. Measurements made with the Forge NxT software demonstrate that the axle dimensions have an accuracy of ±1 mm, which is a very good result.
The only problem that needs to be solved in practice concerns accurate positioning of the workpiece between the wedges, particularly prior to the second operation. This position is affected by i.a. billet diameter deviations that may cause differences in workpiece elongation. This means that the position of the workpiece will have to be set individually for every case of the rolling process. Figure 8 shows the distribution of effective strains in a rolled railcar axle. The strain distribution is typical of cross rolling processes, i.e. it has a ring-shaped pattern [8]. One can observe that higher strains are located in the surface layers that are affected by the friction forces which make the workpiece rotate. A higher reduction in the cross   The forming stage in which the workpiece enters into contact with the much colder tools takes 20 s in each operation of the CWR process. The inter-operational interval when the workpiece is rotated takes additional dozens of seconds. In spite of such a long forming time, there are no excessive drops in the temperature of the workpiece. This is proved by the temperature distribution shown in Fig. 9. It can be observed that the temperature of the workpiece is maintained within the hot forming temperature range. The highest temperature value which is similar to that of the billet can be observed in the axial zone of the workpiece. In contrast, the lowest temperature is located on the external surface of the axle, which is where the workpiece had contact with the tools. This temperature distribution results from both a high heat capacity of the workpiece and great quantities of heat generated during the forming stage when the deformation work is exchanged into the friction work.
Material fracture in the axial zone of the workpiece is a frequent failure mode in CWR processes. The probability of material fracture is predicted by means of so-called damage functions, the value of which should be lower than the critical value [17][18][19]. In this simulation, the damage function was calculated via the Cockcroft-Latham criterion, which had been employed in previous studies on CWR [20][21][22].
Based on a study conducted by Pater et al. [23], it was assumed that the critical value of the damage function causing the fracture of 42CrMo4 steel at 1200 °C was equal to 3.55. An analysis of the damage function distribution in the produced railcar axle shown in Fig. 10 demonstrates that the damage function in the axial zone of the workpiece is lower than 1. The highest damage function values can be observed on the end faces of the axle, i.e. where excess material was cut off by the side cutters. It can therefore be claimed that railcar axles produced by the proposed CWR method should be devoid of internal cracks.
Appropriate estimation of force parameters is important for the design and optimisation of the production process [24,25]. An analysis of force parameters obtained in the analysed CWR processes provides important data too. Figures 11 and 12 show the radial load in CWR conducted with the use of two rolls and two flat wedges, respectively. This load causes mill stretch, which affects manufacturing accuracy [26]. The numerical results demonstrate that the maximum obtained loads are similar in both analysed CWR processes, their value being 2.50 MN in CWR conducted with two rolls and 2.52 MN in CWR conducted with flat wedges. These load values are considerably lower than the value predicted for a single-operation CWR process, i.e. 7 MN [27]. It must also be mentioned that slightly higher radial loads occur in the second operation of CWR, which should be explained by a decrease in the temperature of the workpiece. Figures 13 and 14 show the torque on one of the rolls and the forming load (the force which makes the wedge move) in CWR conducted with the use of flat tools. Based on the data given in the figures, it is possible to determine energy that is required for forming a railcar axle. In CWR Fig. 10 Distribution of the damage function calculated in accordance with the normalised Cockcroft-Latham criterion in a railcar axle at the final step produced by two-operation CWR Fig. 11 Radial load on a roll obtained from numerical simulations of a two-operation CWR process for a railcar axle   conducted with the use of two rolls, the work is 1758 kJ in the first operation and 1926 kJ in the second operation. In CWR conducted with flat wedges, the work is 1534 kJ and 1554 kJ, respectively. The higher work value in CWR conducted with the rolls results from both the resistance generated due to friction of the tools against the guides and the increased diameter of the roll, a factor which must take into consideration an additional space for billet loading and finished product unloading. The increased diameter of the roll leads to a higher torque value, which affects the energy consumption of the rolling process. Based on the obtained maximum torques and forming loads, it is possible to make a rough estimate of the power required for the rolling mill, which amounts to 2 × 175 kW.

Experiments
Experiments on producing railcar axles by the proposed CWR method were conducted in a scale of 1:5 on a flat wedge rolling mill available at the Lublin University of Technology. Given the fact that this machine has only one moving tool (top tool), the rolling process was slightly more difficult to perform because it was impossible to use a tubular guide for maintaining the workpiece in stable position during forming.
Tool segments (wedges) were constructed in compliance with the design shown in Fig. 3, and their linear dimensions were reduced by five times. This means that the tools had the overall dimensions of 700 × 290 mm. Also, the tools used in the experiments were provided with guide paths for maintaining the workpiece in stable position in the early stage of the process, i.e. when the wedge tools begin to cut into the workpiece.
The billet was a cylindrical bar with a diameter of 43 mm and a length of 330 mm, made of 42CrMo4 steel. The billet was preheated to a temperature of 1180 °C in an electric chamber furnace. After that, it was placed in the bottom tool's guide paths and one of its end faces was pressed to the end stop, the position of which determined the length of the axle body. Next, the top tool was set in motion. The tool moved in a linear manner and rolled the workpiece over the stationary bottom tool. During this motion, the wedges reduced the diameter of the workpiece. Excess material was cut off by the side cutters in the final stage of the rolling process. Following a complete work cycle, the top tool was retracted to the start point; the workpiece was rotated by 180° and put again on the guide paths of the bottom tool. After that, the top tool was set in motion again, and a finished railcar axle was produced. Successive stages of the proposed CWR process for producing railcar axles are shown in photographs in Fig. 15. Figure 16 shows the railcar axle produced by the proposed CWR method. The part is characterised by good manufacturing accuracy. The diameters of deformed regions are within a dimensional tolerance of d +1.0 0.0 . It is worth drawing attention to the fact that the inside diameter of the undeformed region increased. This is an effect of material upsetting on the wedge flank caused by the unbalance of loads acting in the workpiece axis direction. This defect is  16 Railcar axle (in a scale of 1:5) produced by a two-operation CWR process conducted under laboratory conditions at the Lublin University of Technology however negligible, as it can easily be removed by mechanical machining. Importantly, the axle produced by the proposed CWR process is free from internal defects (cracks) that occur in the standard rolling process [13]. Figure 17 shows the experimental forming load (the force which makes the wedge move) in the CWR process. It can be observed that in both passes of the wedge tool, the load rapidly increases in the early stage of the rolling process when the guide paths cut into the workpiece. Once the position of the workpiece becomes steady, the load decreases and its subsequent behaviour pattern agrees (in qualitative terms) with that obtained in the numerical simulation. The load value in the second pass of the wedge tools is approx. 50% higher than that in the first pass. This primarily results from a drop in the temperature of the workpiece that occurs during the inter-operational interval which is several seconds long. Furthermore, in the second pass of the tools, the workpiece is not as accurately guided over the bottom wedge as in the first pass, which results in higher ovality of the cross section. Nevertheless, the obtained experimental forming loads are low compared to the maximum achievable load of the laboratory rolling mill (280 kN).
Microstructure studies were carried out on metallographic specimens taken at cross sections of successive steps of the axle forging. The specimens were cut using an abrasive cutting machine with water cooling. The specimens were subjected to grinding on abrasive papers with grits of 80, 220, 320, 600, 1200, respectively, then the specimens were polished using diamond suspension with a grit of 3 μm and colloidal silica suspension with a grit of 0.05 μm. Nital 4% was used to etch the microstructure. The area located in the middle of the radius of the cylindrical specimen was analysed. The resulting microstructure images are shown in Fig. 18.
After rolling and cooling in air, 42CrMo4 steel has a ferritic-pearlitic structure and consists of regular grains, in the interior of which finely lamellar pearlite and ferritic areas are present. Analysing the microstructure on each step of the forging individually, it can be seen that it is homogeneous and does not show banding. Comparing the microstructure on each stage of the forging, it can be seen that the grains with the largest size are found in step B and, on the other hand, the grains with the smallest size are found in step A.
Step B of the forging was not deformed plastically during rolling and, as a result, heating the material to 1180 °C resulted in grain growth. The grains with the smallest size are found in stage A, where the highest value of plastic strain was recorded according to the data shown in Fig. 8. With an increase in the value of plastic strain, significant degradation of the pearlitic plate structure and mixing of the fine particles with the ferritic matrix was observed. As a Fig. 17 Experimental forming load in a CWR process for producing a railcar axle in a scale of 1:5 Fig. 18 Microstructure of a railcar axle forging for individual steps in accordance with Fig. 1: a step A ∅139 mm, b step B ∅216 mm, c step C ∅182 mm result, an increase in plastic strain resulted in a decrease in the number of ferrite grains.

Conclusions
The numerical and experimental results of this study lead to the following conclusions: • High-quality railcar axles can be produced by the proposed cross wedge rolling process conducted in two operations using the same tool sets. • To ensure manufacturing accuracy, it is recommended that the CWR process be conducted with the use of two rolls or two flat wedges moving in the opposite directions. • In the second pass of the tools, the forces (torques) must be higher than those applied in the first pass. • Given its high heat accumulation capacity, the workpiece does not have to be reheated between individual passes of the wedge tools. • The proposed CWR process for producing railcar axles can be performed using the largest rolling mills that are available on the market.