Effect of temperature field on impurity extrusion

The temperature field is the primary factor affecting the upsetting process and an appropriate temperature field distribution is the basis of the upsetting process. To study the influence of different temperature fields on the extrusion behavior of impurities, five different temperature gradients were obtained by changing the heating and holding time during heat transfer, as shown in Fig. 9. Under the same upsetting length (10 mm) and upsetting speed (10 mm/s), the displacements of the highest point of impurity in different temperature fields are listed in Table I. It can be observed from the data that the displacement of the impurities is larger as the high-temperature region of the rail is widened, and they are also extruded out from the rail more easily. This is because the high-temperature plastic zone is above 1000 °C for U75V steel [Reference Ren, Ji, Zhou, Yuan, Han and Li21]. The wider the high-temperature region in the temperature field, the more easily the metal is deformed on the upsetting surface and its vicinity, and the likelier it is that the plastic metal extrusion deformation of the rail end face brings out impurities under the action of the upsetting force. Simultaneously, the temperature field before upsetting should match the upsetting process. If the high-temperature plastic zone is too wide, it is easy to cause poor joint formation with a large upsetting force and upsetting length [Reference Ichiyama, Ichikawa and Saito5], which cannot guarantee the joint performance. Therefore, the temperature field distribution before the upsetting process should be considered comprehensively with the upsetting parameters, especially the upsetting length. Furthermore, the high-temperature plastic zone should be widened as much as possible on the basis of satisfying the good shaping and toughness requirements of the rail welded joint.

Figure 9:

Temperature distribution of different temperature fields.

TABLE I:

The displacements of the highest point of impurity in different temperature fields.

Effect of upsetting speed on impurity extrusion

The magnitude of the upsetting speed affects the joint deformation and the extrusion of oxide impurities. Therefore, the simulation of the extrusion of impurities at different upsetting speeds was performed under the same temperature field and the upsetting length of 10 mm. Considering the temperature field 3 as the initial temperature field, the displacements of impurities were calculated when the upsetting time is 0.6 s, 0.8 s, 1 s, and 1.2 s, i.e., the upsetting speed is 16.667 mm/s, 12.5 mm/s, 10 mm/s, and 8.333 mm/s, respectively.

The simulation results are shown in Fig. 10. The higher the upsetting speed, the greater is the displacement of impurities, i.e., the easier is the extrusion of the impurities. Furthermore, an increase in the upsetting speed is beneficial for the rapid healing of the gap between the rails, which avoids the continuous oxidation of the end faces. A significant reduction of the end face temperature is also prevented at a higher upsetting speed, which has an advantageous effect on the recrystallization process of the weld and the strength of the joint. It can also be observed from the curve that the slope becomes smaller when the upsetting speed is greater than 12.5 mm/s, which indicates that the extrusion effect on the impurities becomes no longer evident after the upsetting speed reaches a certain level. Moreover, the requirements for the upsetting hydraulic system are improved when the upsetting speed is too high, which requires a motor with a higher power to drive the oil intake.

Figure 10:

The displacements of the highest point of impurity of different upsetting speeds.

In addition, the influence of constant-speed and variable-speed upsetting on the extrusion of impurities is investigated in this study. As shown in Fig. 11, the upsetting length in both cases is 10 mm. The constant speed is set to 10 mm/s, and variable-speed upsetting adopts the method of slow upsetting after quick upsetting. The calculation results are shown in Table II. It is shown that the displacement of impurities during variable-speed upsetting is larger. As the decline of the rail temperature is fast, the temperature of the rail end face is higher in the initial stage of upsetting, and the rapid upsetting is beneficial for the plastic deformation of the metal. Therefore, in the actual welding process, variable-speed upsetting should be adopted instead of constant-speed upsetting.

Figure 11:

Upsetting of constant speed and variable speed.

TABLE II:

The displacement of the highest point of impurity of constant speed and variable speed.

The displacement curve obtained using the UN-150ZB1 flash welder during the welding process is shown in Fig. 12. It can be observed that the actual rail FBW process uses the variable-speed upsetting method. Rapid upsetting is used to close the end face and reduce oxidation in the initial stage of upsetting. The joint forms alternating crystallization by slowing the upsetting speed in the later stage. The simulation results correspond to the actual production, which indirectly proves the reliability of the model. Therefore, using variable-speed upsetting or increasing the speed of the early upsetting can reduce the probability of impurities in the welded joint of the rail under the premise of ensuring the toughness of the joint, thereby improving the quality of the welded joint.

Figure 12:

The displacement curve of rail FBW experiment.

Effect of upsetting length on impurity extrusion

To study the influence of upsetting length on the extrusion of impurities, the temperature field 3 is set as the initial temperature field and the upsetting speed is set as 10 mm/s. The upsetting length is set as 8, 10, 12, and 15 mm. Figure 13 shows the displacement of the highest point of impurity. It can be observed that the displacement of impurity increases linearly with the upsetting length. It shows that increasing the upsetting length is effective for reducing the number of defects such as gray spots, which also confirms the results obtained by Ziemian et al. [Reference Lu, Xu, Jiang, Ma and Shu14, Reference Ziemian, Sharma and Whaley16].

Figure 13:

Displacement of impurity in different upsetting length.

However, an excessive upsetting length will cause large lateral deformation and change the metal texture at the joint as the rail is formed by rolling, which will reduce the joint toughness [Reference Ichiyama, Ichikawa and Saito5, Reference Dai, Lv, Tan and Yan22]. The upsetting length is closely related to the temperature field. To achieve a large upsetting length, there must be a wider high-temperature plastic zone. Therefore, the width of the high-temperature plastic zone should be controlled to avoid large lateral deformation caused by the excessive upsetting length and to reduce the joint performance.

Extrusion behavior of impurities at different locations

In the rail FBW, oxide impurities such as gray spots mainly appear in the area below the rail web, especially at the base angle and the trigonum. To study the movement of impurities at different positions during the upsetting process, the temperature field 3 is set as the initial temperature field, the upsetting speed is set as 10 mm/s, and the upsetting length is set as 10 mm. The position of the impurities in the model is set as shown in Fig. 14, in which the center of C3 is 3 mm from the rail bottom, and the distance between impurities in the Y direction is 3 mm. The center of all impurities in the Z direction is 3 mm from the rail bottom, and the distance between impurities is 15 mm.

Figure 14:

The position of the impurities.

The simulation results are shown in Fig. 15, where the further the distance from the lower surface of the rail in the Y direction [Fig. 15(a)], the smaller is the displacement of the impurity. The displacement of the highest point of C3 in the Y direction is −8.79426 mm, whereas that of C3-Y15 is only −3.06682 mm. This is because the plastic metal at the trigonum of the rail bottom will flow simultaneously toward three surfaces of the rail during upsetting, and the force in the three directions will offset a part or even reach equilibrium, which is not conducive to metal flow and extrusion of impurities. At the center of the trigonum, the protective atmosphere formed by the metal vapor during the flashing process is effective, and the air is difficult to intrude, and hence, the probability of generation of oxide impurities such as gray spots is very low. Therefore, the protection of the flashing process should be strengthened to avoid the formation of oxides in the deep areas. While the areas close to the surface are prone to oxide inclusions, it is also easier to extrude them out.

Figure 15:

Displacement of the highest point of impurities at different locations: (a) impurities in the Y direction and (b) impurities in the Z direction.

In the Z direction, the displacement of the highest point of impurities is shown in Fig. 15(b). The further the distance from C3, the smaller is the displacement of the impurity. This is mainly due to the difference of temperature fields at the rail bottom. The relative contact areas between the different positions of the rail and air are different because of the particularity of the rail geometry, and the heat dissipation is significantly different. The relative contact areas of air with both sides of the rail bottom are larger, accordingly, the heat dissipation is faster. Therefore, the temperature field formed at the rail bottom is shown in Fig. 16. In the Z direction, the high-temperature range from the center to the sides of the rail bottom is gradually narrowed, and the corresponding high-temperature plastic zone is also narrowed, which is not conducive to the extrusion of impurities. In the experiment, the weld beading of the joint was immediately pushed after welding, and the macroscopic weld profile of the rail bottom was obtained as shown in Fig. 17. It is evident that the weld of the rail bottom is narrowed from the middle to both sides, which indicates that the width of the high-temperature plastic zone is different. The macroscopic weld profile corresponds to the temperature field of the rail bottom of Fig. 16, which also ensures the rationality and reliability of the temperature field.

Figure 16:

The temperature field of rail bottom.

Figure 17:

The macroscopic weld profile of the rail bottom.

Establishment of finite element model

After the flashing process, a very thin liquid layer is formed on the end surface. The oxide inclusions in the liquid metal are easily extruded out during the upsetting process as the fluidity of liquid is excellent. Simultaneously, craters are formed by the liquid-bridge blasting during the flashing process. The craters are solid phase, whose fluidity is not as good as that of the liquid phase. Therefore, it is difficult to squeeze out the oxide inclusions remaining in these craters [Reference Zheng, Tian and Wu10]. Thus, a three-dimensional finite element model of the solid-phase upsetting process is developed, whereas the liquid phase portion is neglected. The model is shown in Fig. 18.

Figure 18:

Three-dimensional model of rail upsetting process: (a) three-dimensional overall model, (b) built-in layer, and (c) loading method.

As shown in Fig. 18(a), the total length of each rail is 400 mm in this model. Two pairs of copper electrodes are clamped on both sides of the rail web and 110 mm away from the rail end faces. The large deformation zone is focused on the joint and its vicinity, and hence, finer meshes were designed in these areas, whereas relatively coarser grids were adopted in the region away from the joint. Furthermore, as shown in Fig. 18(b), a solid layer is used to embed gray spots for the upsetting simulation. In this model, three-dimensional eight-node elements are used for analysis in ABAQUS. The continuum model has 198,969 nodes and 166,044 hexahedral solid elements.

The material properties that must be determined in this model are thermal conductivity [mW/(mm °C)], specific heat capacity [mJ/(T °C)], density (T/m³), Poisson’s ratio, and elastic modulus (MPa). As the upsetting process is a deformation process at high temperature, the material properties use temperature-dependent data. Furthermore, the deformation of the rail is closely related to the temperature and the upsetting speed. The Johnson–Cook model is a classic empirical constitutive model proposed by Johnson and Cook in 1983 [Reference McClain, Batzer and Maldonado23]. The model shown as Eq. (1) describes the effects of material temperature and strain rate on the yield stress and strain of the material, and hence, it is suitable for investigating the upsetting process of rail FBW.

(1)$$\sigma = \left\bigg[ {A + B{{\left( {{\varepsilon ^{\rm{p}}}} \right)}^n}} \right]\left[ {1 + C\;\ln \left( {{{{\varepsilon ^{\rm{p}}}} \over {{{\dot{\varepsilon }}_{\rm{o}}}}}} \right)} \right]\left( {1 - {{\left( {{{T - {T_{\rm{r}}}} \over {{T_{\rm{m}}} - {T_{\rm{r}}}}}} \right)}^m}} \right)\quad ,$$

where σ is the stress, ε^p is the effective plastic strain, ${\dot{\varepsilon }_0}$ is the reference strain rate, T is the workpiece temperature, T _r is the room temperature, T _m is the melting temperature of the material, and A, B, C, n, and m are material-dependent constants.

According to the characteristics of FBW, the temperature field during upsetting is formed during the flashing stage. There is no suitable heat source model to characterize the process as the liquid-bridge blasting of joint during the flashing process is random. Ma and his team observed that the heating speed of the rail during flashing process is very fast, and the temperature is evenly transmitted from the flashing surface to the inside of the rail [Reference Ma, Cai, Huang, Deng, Murakawa and Pan18]. Hence, the initial temperature field is applied to the rail by means of superficial temperature load heat transfer from the upsetting surface to the rail inside in this model, and then the temperature field data are transmitted to the dynamics analysis (Dynamic, Temp-disp, Explicit) to simulate the thermomechanical coupling of the upsetting process.

The equation for transient heat transfer analysis during welding is given as follows [Reference Ma, Cai, Huang, Deng, Murakawa and Pan18]:

(2)$$\rho c{{\partial T} \over {\partial t}}\left( {x,y,z,t} \right) = - \nabla \cdot q\left( {x,y,z,t} \right) + Q\left( {x,y,z,t} \right)\quad ,$$

where ρ is the density of the materials, c is the specific heat capacity, T is the current temperature, q is the heat flux vector, Q is the internal heat generation rate, x, y, and z are the coordinates in the reference system, t is the time, and ∇ is the spatial gradient operator. The nonlinear isotropic Fourier heat flux constitutive equation is employed:

(3)$$q = - k\nabla T\quad ,$$

where k is the temperature-dependent thermal conductivity.

The following assumptions are made based on the above description:

1. (i) The room temperature is 20 °C;

2. (ii) The solid-phase temperature of the rail upsetting surface is approximately 1300 °C after the flashing stage as the liquid phase is not considered in this model [Reference Dai, Lv, Tan and Yan22];

3. (iii) As shown in Fig. 18(c), a fixed constraint is applied to a pair of copper electrodes, so that they cannot be displaced and rotated. The displacement load in the U1 direction (x-axis) is applied in the form of amplitude on the other pair of copper electrodes, and U2 = U3 = UR1 = UR2 = UR3 = 0 is set to limit the displacement and rotation in these five degrees of freedom.

FBW experiment

To ensure the reasonableness and reliability of the model, this study conducted a verification experiment. The experiment used the UN-150ZB1 flash welder, which is commonly used in Chinese seamless tracks to weld the U75V rail. The UN-150ZB1 flash welder is a mobile flash butt welder with a pulsating flash process. The chemical composition of the U75V rail is listed in Table III. The rail length is 700 mm and the upsetting parameters are consistent with the settings during the simulation. The welding parameters, such as the current, voltage, and displacement, are obtained using the data acquisition system of the welder in the welding process.

TABLE III:

The chemical composition of U75V.