Analysis of the wear characteristics of multi-directional die forging and forming dies for a railway wagon bogie adapter

doi:10.21203/rs.3.rs-1596106/v1

Download PDF

Research Article

Analysis of the wear characteristics of multi-directional die forging and forming dies for a railway wagon bogie adapter

https://doi.org/10.21203/rs.3.rs-1596106/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

This paper evaluates the influence of the combined die motion scheme and forging process parameters on the evolution of the wear pattern of forging dies for the railway wagon bogie adapter based on an improved Archard model The simulation results show that the distribution of die wear parts during the hot forging process of the railway wagon bogie adapter is uneven but symmetrical, and the maximum wear parts are mainly distributed in the ring belt part of the top die and the shoulder part of the bottom die, and the maximum wear amount of the top die is about 1.4 times that of the bottom die. In addition, the influence of the initial forging temperature (T_w) and the preheating temperature (T_d) on the wear behaviour of the die during the forging and forming of the adapter was investigated, and a combination of high T_w and low T_d was found to be more beneficial in reducing the wear of the die. In order to obtain the optimal process parameters, a response surface model of die wear was established and combined with a Genetic Algorithm-Artificial Neural Network (GA-BP) to optimize the forging process parameters and proposed an optimal strategy to reduce die wear.

A railway wagon bogie adapter

Multi-directional die forging

Die wear

Die combination movement

GA-BP neural network

The forging of a railway wagon bogie adapter is a typical three-dimensional thermodynamic coupling problem, with typical characteristics of large deformation, high forming speed and high deformation temperature. ^[1–3] The quality of the adapter forgings in the die forging process is influenced by a variety of factors, while the quality of the die directly affects the quality, cost, productivity and market competitiveness of the forgings. In the actual forging process, forging dies usually work under extremely high dynamic or static loads repeatedly, their working stresses are extremely high and the working conditions are extremely harsh. The preheating ^[4], cooling ^[5], lubrication ^[6] and coating influence ^[7] of the die during work have a direct impact on the use of the die, so it is important to improve it as much as possible in order to prolong the service life of the die. The large deformation and high deformation temperatures result in increased wear of the forging dies, which in turn leads to a reduction in the forming accuracy of the forgings. ^[8–10] Previous research on die wear has mainly been based on the finite element method, mostly using the modified Archard model to predict die wear behaviour. ^[11–13]Rajiev, R^[14] carried out an experimental study on the preheating temperature of the die to predict the total wear depth of different parts of the hot forging cavity and eventually obtained the maximum significance of the wear depth. Paiva, Jose Mario ^[15] investigated the wear effects of different PVD coatings and combined the experimental results with numerical simulations to confirm that the main distribution of wear is in the radius region of the die cavity.ZHI ^[16] He studied the effect of temperature conditions on material deformation and die wear by employing a single control variable and concluded that the effect of process conditions on material deformation was relatively simple and the effect on die wear was more complex.ZHANG^[17] simulated the extrusion process of aluminium tubes and explored the law of extrusion speed, friction conditions and the effect of temperature on die wear, and concluded that the extrusion pressure increases significantly with the increase of plunger speed and friction coefficient and the decrease of billet and die temperature, and the pre-test results are in full agreement.QIAO^[18] researched the wear characteristics of automotive cover stamping dies and established a new method for predicting die wear considering the variation of contact pressure, wear coefficient and material hardness along the thickness direction of the treated layer. The prediction results of the new method and the traditional Archard method were compared with the actual wear situation, and the results showed that the die life prediction method adopted in this study is closer to the actual situation and significantly more accurate than the traditional Archard method. XU et al. ^[19] studied the wear mechanism of the arbor during wheel forging by SEM and gave numerical simulations and BP neural networks to optimise the design of the geometry of the arbor, verifying the accuracy of the wear calculations, which were in good agreement with the calculated results. AISI H13 die steel wear prediction based on response surface methodology and artificial neural networks by Jagota, Vishal et al. ^[20]

In the current research, the evolution of die wear patterns in forging processes has focused on the study of process parameters, including speed, temperature and friction conditions, and has concentrated on single-process, single-direction forging dies. ^[21–23] However, most products cannot be formed by a simple one-way forging process, especially for large and complex forgings. The adapter has a complex structure and many features, and is a typical large deformation, difficult to forge, multi-process forming forgings, the current research on the forging process of such parts die wear work is very little.

In this paper, a three-dimensional thermodynamic coupling model for hot forging of the adapter is developed to predict the wear of the forging die, linking the initial forging temperature of the workpiece, the preheating temperature of the die and the wear evolution of the forging die. On the basis of the simulation results, the influence of temperature on the evolution of the wear of the forging die is revealed. The way in which the combined motion of multi-directional die forging dies has a significant influence on the die wear pattern of forgings and is also highlighted for evaluation in this study. In view of the influence of the interaction of process parameters on the wear of forging dies for adapter, a response surface model of die wear was established and combined with a GA-BP model for optimisation, and an optimal solution for reducing die wear was proposed. This research can provide theoretical guidance for improving the forging accuracy of adapter parts and extending the life of the die.

2.1 Archard wear theory model

The model of die wear currently used by most scholars is the modified Archard model, as shown in Eq. (1). Compared to the traditional Archard model^[24], the modified model takes into account the effect of temperature and the die wear factor (K) and die hardness (H) are defined as temperature-dependent functions ^{[25, 26]}, with the specific relationships shown in Equations (2) and (3):

$$W(T)=K(T)\frac{{LP}}{{H(T)}}$$

$$K(T)=[29.29 \times lnT - 168.73] \times {10^{ - 6}}$$

$$H(T)=9216.4 \times {T^{ - 0.505}}$$

Where: is the wear depth; is the wear factor; is the hardness of the die; is the tangential relative slip length between the workpiece and the die; is the normal pressure between the workpiece and the die.

Considering that the relative sliding speed, pressure and temperature of the contact surfaces of the forging and the die vary with the contact position and contact time, Eq. (1) can be deformed as follows:

$${W_i}={K_{ij}}(T)\frac{{{V_{ij}}{P_{ij}}}}{{{H_{ij}}(T)}}d{t_j}$$

Where: ${P_{ij}}$ is the normal pressure at the point of and in time of ;${K_{ij}}(T)$and ${H_{ij}}(T)$ are the variation functions of the die wear factor and the die hardness the point of and in time of , respectively.

At the end of the forging process, the wear of the die at a specific point can be obtained from Eq. (5):

$${W_i}=\sum {_{{j=1}}^{{j=n}}} {K_{ij}}(T)\frac{{{V_{ij}}{P_{ij}}}}{{{H_{ij}}(T)}}d{t_j}$$

2.2 Finite element model building

Figure 1 shows a schematic diagram of a adapter. It can be seen that the forming of the adapter mainly lies in three parts: adapting surface annulus, the pedestal and the locating guard. Among these three parts, the adapting surface annulus is a semi-circular area and is in direct contact with the axle, which has a small machining allowance and requires a high degree of forging accuracy. Therefore, the wear of the forging dies in this area plays a decisive role in the accuracy of the adapter forgings and the service life of the forging dies.

In order to effectively visualize the distribution of die wear evolution in the ring belt area of the saddle face, simplify the finite element model and improve the efficiency of the simulation, a 3D thermodynamic coupled finite element model of the final forging process of the hot forging of the adapter was established based on a modified Archard wear theory model using a 3D local geometry model, as shown in Fig. 2. The precision forging and forming equipment for the adapter is shown in Fig. 2(a). The die is simplified and the wear behaviour of the tool during forging is analyzed in Deform, taking into account the symmetry of the adapter part, the 1/4 model is used for calculation and the geometric model created after the simplification is shown in Fig. 2(b).

The simplified model was imported into the Deform software in STL file format and a finite element model of the precision forging of the adapter was established. Table 1 gives the main simulation parameters, which are in agreement with those used in the actual forging of the adapter. In order to make the established finite element model more accurate in predicting the wear of hot forging dies for adapters, the finite element model was developed with the following characteristics:

The number and fineness of the meshes affect the simulation results, so the meshes are refined for the areas where the workpiece and the die are in direct contact. The mesh type is tetrahedral, with a finer mesh for the parts of the die that are in direct contact with the workpiece, where both the top and bottom dies have 150,000 meshes and the workpiece has 200,000 meshes, as shown in Fig. 3(b) and Fig. 3(c).

Considering that the forging of a adapter is a multi-physical coupling process, related to the temperature, hardness, lubrication state, contact stress and strain rate of the contact surface ^[27–29], the friction type is chosen as shear friction, because there is lubrication in the forging process, and the friction coefficient is set to 0.3.

Importing an Arrhenius intrinsic model with strain compensation into the software to predict the stress flow behaviour of AISI 1035 steel at high temperatures ^[30].

In the forging process of the adapter, the temperature change of the forging die exists in two processes: heating and holding. In the initial stage of forging, the temperature of the die is low, heat absorption is greater than heat dissipation, and the temperature gradually increases; after the temperature rises, the die reaches thermal equilibrium with the surrounding environment, and then the temperature is maintained in a higher range, while the surface hardness decreases. This process is similar to the tempering process ^[31–33]. Therefore, the hardness of AISI H-13 steel at different temperatures (Fig. 4) was entered into Deform as a function to obtain more accurate wear results ^[34].

DEFORM-3D software has the capability to simulate the wear of a die during a forging process, but the model that comes with the software is an unmodified Archard wear model, which is not applicable to the hot forging process. Therefore it is necessary to modify the subroutine on the basis of the original model, to write the relevant parameters of the modified Archard model and to introduce and , as functions of T into the Equations.

Table 1

Main simulation parameters
Matetial	Workpiece	Forging Die
Matetial	AISI 1035	AISI H-13
Initial temperature(℃)	1000	200
Convection coefficient to environment (N/(s mm °C))	0.02	0.02
Friction factor	0.3
Forging speed(mm/s)	30
Enviromment temperature(℃)	25
Heat transfer coefficient between workpiece and die (N/(s mm °C))	11
Hardness(HRC)	------	55

3.1 Influence of the combination of die movements on the evolution of die wear

The multi-directional forging process of adapter is a complex, non-linear, time-varying rheological process. Combined with the characteristic parts of the adapter and the principle of the existing multidirectional die forging equipment, it is clear that the forging process can be divided into vertical and horizontal forging.In this paper, four schemes (as shown in Table 2) have been designed to study the evolution of the wear of the die carrying the saddle. The dies used in the four schemes are the same type of die, the difference lies in the sequence of movements of the top die and left and right punches and the combination of movements. The actual displacement of the top die is 54 mm and the actual displacement of the left and right punches is 27.17 mm, based on the dimensional requirements of the adapter and taking into account subsequent machining allowances.

Table 2

Different schemes for multidirectional die-forging of the adapter
Programme	Die movement sequence
1	The top die is moved until the end of the stroke and the left and right punches are moved until the forging is fully formed.
2	The left and right punches are moved until the end of the stroke, then the top die is moved until the forging is fully formed.
3	The left and right punches move to the designated position first, the top die starts to move until the end of the stroke, and the left and right punches move until the forging is fully formed.
4	The top die and the right and left punches move simultaneously according to the set stroke until the forging is fully formed.

Figure 5 shows the distribution of die wear on the top die at the end of forging for the four scenarios. As can be seen from the figures below, the degree of wear on the top die varies in the four scenarios and the locations of the wear are not identical, but the greatest wear is found on the adapter face ring belt and is basically symmetrical in distribution. The top die of scenarios 2 and 4 have more wear in the adapter ring band area, followed by scenario 3 and scenario 1 has the least wear. The top and bottom closing of the dies in programme 1 is the first stage of movement. The metal flows mainly along the transverse direction under the extrusion of the top die. The left and right dies are not closed at this time, so there is a large sliding speed between the workpiece metal and the top die and the contact pressure is low, so the wear at this time is low at the saddle face ringband position. At the closing stage of the left and right dies (second stage of movement), the saddle face ringband section of the workpiece has been formed, at which point the metal flow capacity is reduced, the interface pressure increases and wear continues to increase. Programme 2 and 3 are both left and right dies first, during this process the metal flow of the workpiece is mainly in the vertical direction, when the top and bottom dies are closed, the flow of metal is restricted and the friction with the ring belt part of the top die increases, resulting in an increase in the interface pressure and a rise in the amount of wear. In programme 4, where the left and right pounches closing is carried out simultaneously with the top and bottom dies closing, the metal flow is similar to that of programme 2, where the metal flow is influenced by the cavities after the die closing and the increase in interfacial pressure leads to an increase in wear. From the perspective of maximum wear, it can be seen that Programme 2 has a maximum wear of 1.09E-05mm, which is the largest of the four programmes, and is approximately 1.43 times the maximum wear of Programme 1 (7.62E-06mm), Programme 3 (9.86E-06mm) has a maximum wear of approximately 1.29 times that of Programme 1, and Programme 4 (1.07E-05mm) has a maximum wear is approximately 1.4 times that of programme 1.

Figure 6 shows the maximum wear distribution of the left and right punches for the four programmes. It can be seen that the wear areas of programmes 1, 2 and 3 are very similar, and the wear areas of programme 4 are different from the first three programmes. From the location of the wear parts, the wear parts of the left and right punches also show a symmetrical distribution, with the larger wear parts being distributed on both sides of the die and decreasing towards the middle, which is related to the deformation of the metal. From the point of view of the amount of wear, the amount of die wear for the four programmes is increasing in order, with Programme 4 having the highest amount of wear, because the flow of metal in Programme 4 is restricted, with the top and bottom dies and the left and right dies extruding the workpiece at the same time, with the metal flowing vertically and horizontally in a staggered manner, so the value of die wear in Programme 4 is considerably higher than in all the other programmes, and the distribution of wear areas is more uneven.

In order to quantify more accurately the evolution of the die wear during hot forging of the adapter, 45 measurement points were taken along the adapter face annulus of the top die as shown in Fig. 7(b). Figure 7(c) shows the final results of the point trace of the selected measurement points for the four programmes at the end of the hot forging. It can be seen that the evolution of the wear volume is basically the same for all four programmes and is consistent with the distribution pattern shown in Figure (5): the wear areas are unevenly but symmetrically distributed. Overall, Programme 1 has the least amount of wear and tear and Programmes 2 and 4 have more. In all forging programmes, the maximum amount of wear occurs at point 14 and at the location of point 32. The reasons for this phenomenon have already been mentioned in the previous analysis and will not be repeated here. From the edge of the ring belt to the middle of the ring belt, the wear of the die tends to rise and then fall. An analysis of the forging process of the adapter shows that the flow of the metal of the workpiece is limited by the movement of the cavity and the convex die of the die. As a result, the relative sliding velocity between the workpiece metal and the cavity surface remains low in the hard-to-deform area, resulting in a smaller amount of cavity wear in this area. The middle part of the workpiece has less wear due to the restriction of the die cavity, where the metal flow capacity is also poor. As the largest part of the ring belt wear parts, combined with the forging process of the adapter can be seen, this part has been in contact with the workpiece, resulting in the workpiece has been in a large deformation state, the sliding friction brought about by the flow of the workpiece metal makes the interface pressure in a high state for a long time, the amount of die wear here has been increasing.

According to the previous analysis, the multi-directional die-forging solution of programme 1 has a lower amount of wear compared to the other programmes. Figure 8 gives the evolution of the wear of the die with the forging stroke and the distribution pattern on the forging process of the adapter in Programme 1. It can be seen that the wear of the top die expands gradually with the increase of the forging stroke. The main reason for this is that as the stroke increases, the contact area between the adapter part and the top die increases, the contact pressure between the workpiece and the die increases and the flow of metal in the adapter is impeded, leading to an increase in the wear of the top die. At the beginning of the forging stage, wear occurs first on the part of the top die that is in first contact with the workpiece, and increases with the amount of downward pressure, and the wear area begins to spread to the rest of the top die. After the forging process is completed, the top die wear areas are not evenly distributed, but show a roughly symmetrical distribution. The greatest wear is found on the left and right sides of the ring band of the top die (marked by the red line box), with minimal wear in the middle ring band.

The deformation level of 75% in the figure corresponds to the first stage of the adapter forging, i.e. the closing stage of the top and bottom dies, and it can be seen that the maximum wear on the top die is 7.52E-06 mm when the top and bottom dies are completed. The deformation degree of 100% corresponds to the entire forging process of the adapter has been completed. At this time, the maximum wear amount of the top die is 7.62E-06mm. Compared with the top and bottom dies closing stages, the wear amount of the top die in the second stage of forging increases by 1E-07mm. It can be seen that during the multi-directional forging process of the adapter, the wear of the die is mainly concentrated in the top and bottom closing stages, while the wear of the top die also increases in the left and right closing stages, the degree of die wear is much less than in the top and bottom closing stages.

Figure 9 shows the distribution of the wear pattern of the bottom die cavity during the forging of the adapter. Similar to the die wear pattern of the top die, the wear distribution of the bottom die is also basically symmetrical and unevenly distributed on one side. Analysis of the evolution law of the wear part of the bottom die shows that the wear part of the bottom die starts from the position of the bottom top block. With the increase of the pressing amount of the top die, the wear part gradually spreads to the position of the cavity of the bottom die, and the wear amount also increases. The maximum wear on the bottom die is 4.6E-06 mm when the top and bottom dies are closed and 5.32E-06 mm when the entire forging process of the adapter is completed. Compared to the top and bottom die closing stage, the wear on the bottom die during the left and right punches extrusion stage increases by 7.2E-07 mm, an increase of approximately 15.7% of the first stage. At the end of the forging, the greatest wear on the bottom die occurred in the shoulder area (marked by the red box), with a maximum wear value of 5.32 x 10 − 6 mm, while the maximum wear on the top die was 7.62E-06 mm, approximately 1.43 times the maximum wear on the cavities of the bottom die. Considering the structural schematic of the adapter shown in Fig. 1, the machining accuracy of the ring belt part of the adapter face is much greater than the machining accuracy of the rest of the parts, and the requirements for forging accuracy are very high. Therefore, considering all aspects, the amount of wear on the ring belt part of the top die is the key to the life of the forging die and the forming accuracy of the adapter forgings.

3.2 Influence of billet starting forging temperature on die wear distribution

In order to further investigate the effect of temperature on the amount of wear in the forging die of the adapter, the single variable method was used to set different gradients of the starting forging temperature T_w of the workpiece and the preheating temperature T_d of the die respectively, as shown in Table 2, while the settings of the remaining parameters were kept in line with Table 1.

Table 2

Finite element simulation temperature gradient settings
Number	T_w (℃)	T_d (℃)
1	900,950,1000, 1050,1100	250
2	1000	150,200,250, 300,350

Figure 10 shows a three-dimensional surface plot of the wear values at the measurement points as a function of the initial forging temperature of the part. As can be seen from the figure, the maximum wear of the top die is negatively related to the starting forging temperature of the workpiece, the higher the starting forging temperature of the workpiece, the lower the maximum wear of the top die. As can be seen from the figure below, when the starting forging temperature is 1100°C, the maximum wear of the die is 7.91E-06 mm. When the starting forging temperature of the billet is 900°C, the maximum wear depth of the die on the adapter is 1.59E-05 mm, which is approximately 200% of the wear at 1100°C. This phenomenon is mainly related to the high temperature stress flow behaviour of the workpiece material. ^[35] The increase in the initial forging temperature of the workpiece effectively reduces the flow stress of the material, resulting in a reduction in the contact pressure between the adapter forging and the cavity surface of the die and a consequent reduction in the amount of wear, which is consistent with the trend shown in the figure.

As can be seen from Fig. 11, when the initial forging temperature of the workpiece increases from 900°C to 950°C, the maximum wear of the top die decreases by 15.7% and the sliding speed of the metal increases by 28.5%. The increased flow and filling capacity of the metal reduces the difficulty of forming the workpiece and reduces the resistance to forming. As the amount of pressure under the die increases, the interface pressure plays a dominant role in the friction between the workpiece and the die. The increase in the preheating temperature of the workpiece makes it easier for the metal to flow, resulting in a reduction in the interface pressure between the workpiece and the die and a reduction in the amount of die wear. Therefore, the following conclusion can be drawn: in the process of hot forging and forming of adapters, increasing the initial forging temperature of the forging can effectively reduce the amount of wear of the forging dies.

3.3 Influence of die preheating temperature on die wear distribution

In order to ensure the toughness of the die and reduce the risk of early service cracking, the hot extrusion dies are often preheated in actual production. If the preheating temperature is too low, the die material will not be tough enough and the difference between the surface temperature and the internal temperature will cause thermal stress when the die is in service. If the preheating temperature is too high, it will cause the die material to soften in advance, reducing the wear resistance of the surface and shortening the service life of the die. Figure 12 shows a graph of the amount of die wear correlation for different preheating temperatures of the die, it can be seen that the maximum wear of the top die basically shows a positive correlation with the preheating temperature of the die. As the preheating temperature of the die increases, the maximum wear of the top die also increases, which is the opposite of the influence of the initial forging temperature of the workpiece on the amount of die wear. However, when the preheating temperature of the die is increased from 300°C to 350°C, the wear of the die decreases instead. The friction and mutual heat transfer during the forging process lead to an increase in the surface temperature of the die. When the preheating temperature of the die is 300°C, the maximum temperature of the die surface reaches about 465°C, and when the preheating temperature is 350°C, the maximum temperature of the die surface reaches about 500°C. Combining the relationship between the hardness of the die material and the temperature shown in Fig. 4, it can be seen that the hardness of the die increases when the temperature of the die is between 400 and 500°C and therefore the wear of the top die decreases, with the trend shown in the figure.

In addition, the three-dimensional response surface plot in Fig. 13 shows that the maximum wear depth of the top die is 1.05E-05 mm when the preheating temperature of the die is 150°C. As the preheating temperature of the top die increases to 300°C, the maximum wear depth of the top die reaches 1.13E-05 mm, an increase of nearly 7.6% compared to the wear at a preheating temperature of 150°C. The reason for this phenomenon is that the change in die preheating temperature is related to the wear factor K(T) and the hardness factor H(T) of the die. According to Eq. (2) and Eq. (3), K(T) increases with increasing temperature and H(T) decreases with decreasing temperature. It can therefore be concluded that within a certain range, the amount of die wear increases as the preheating temperature of the die increases, but when the preheating temperature of the die exceeds a certain value, the hardness of the die material drops sharply resulting in a sharp increase in die wear. Therefore, from the point of view of reducing the amount of wear, the lower the preheating temperature of the die, the better. Furthermore, the effect of the preheating temperature on the die is less than the effect of the initial forging temperature on the amount of die wear.

Combining the simulation results with the quantitative analysis of the process parameters, the process parameters were optimised with the objective of reducing wear as follows: increasing the billet temperature is conducive to reducing the hardness of the billet, enhancing the fluidity of the billet metal and improving the quality of the filling, although it will cause the surface temperature of the die to rise but overall it will reduce the wear on the surface of the die. From the simulation results, high temperature workpiece should be selected as far as possible, provided that the heating temperature range of the workpiece is met. The simulation results show that the lower the preheating temperature the better it is for reducing wear, and as the preheating temperature rises the effect of increasing wear becomes more and more obvious.

The combined effect of the initial temperature of the workpiece (T_w) and the preheating temperature of the die (T_d) on the wear of the die is given in Fig. 14. The overlap of the closed area formed by the parameters in the figure is the area of the optimal solution for the thermal parameters based on the minimum wear of the die. As can be seen from the figure, the optimal thermal parameter solution area is made up of six points, including three points at the start of the forging process and three points at the preheating temperature of the die. When the optimum range is chosen: T_w =1000 ~ 1100℃,T_d =150 ~ 250℃ in the thermal parameters for hot forging and forming of forgings, the amount of wear in the die cavity can be effectively reduced. In addition, the figure shows that the amount of die wear during the forging of the adapter is more sensitive to changes in the starting forging temperature of the workpiece compared to the preheating temperature of the die.

4.1 Response surface model analysis

The above has examined the effect of individual factors on the amount of wear in forging dies for adapter, but the effect of the interaction between these factors is also not negligible. Therefore, the effect of forging speed on die wear is taken into account, and the response surface model for die wear is established using the workpiece initial temperature T_w, the die preheating temperature T_d and the forging speed V as factors affecting the amount of die wear. Each factor has three levels, as shown in Table 3.

Table 3

Table of levels and factors
Factors	Levels
Factors	-1	0	1
T_w (℃)	900	1000	1100
T_d (℃)	150	250	350
V (mm/s)	25	27.5	30

Based on the parameter design in Table 3, the response surface data obtained for the die wear depth is summarised in conjunction with the established finite element model, and Table 4 shows the values of the die wear depth for different deformation parameters. Figure 15 shows the response of the depth of wear of the die as a function of the initial forging temperature of the part, the preheating temperature of the die and the forging speed. In Fig. 15(a), it can be seen that the depth of wear of the die decreases significantly with the increase of the initial forging temperature of the workpiece. The preheating temperature of the die also has an effect on the amount of wear, but this effect is not as strong as the effect of the initial forging temperature of the workpiece. The maximum amount of die wear occurs when the forging speed is 27.5 mm/s, when the workpiece initial forging temperature is 900°C and the die preheating temperature is 350°C. Figure 15(b) shows the effect of the interaction of the workpiece initial forging temperature and forging speed on the amount of die wear when the die is preheated to 250°C. It can be seen that the effect of the forging speed on the amount of wear is not as significant as the effect of the initial forging temperature of the workpiece, but it generally shows that under certain conditions, an increase in the forging speed leads to an increase in the amount of wear. In contrast, the interaction of die preheating temperature and forging speed in Fig. 15(c) does not have as significant an effect on the amount of die wear as demonstrated in Figs. 15(a) and 15(b). As can be seen from Fig. 15, the change in die wear has a more pronounced effect on the change in the initial forging temperature of the workpiece, while the effects of die preheating temperature and forging speed are not as significant as those of the initial forging temperature of the workpiece, but there is a clear interaction between the three. Therefore, in order to obtain the optimum wear depth, the interaction between the process parameters should be reasonably considered and the degree of influence of the parameters should be clarified so that the utilisation rate of the die can be increased and its wear depth reduced in a reasonable and effective manner.

In summary, considering the combined effect of the three process parameters on the amount of die wear, the following process combinations should be chosen: a higher workpiece initial forging temperature, a lower die preheating temperature and a lower die forging speed.

Table 4

Response surface model design
Number	T_w (℃)	T_d (℃)	V (mm/s)	Wear depth (mm)
1	1000	250	27.5	1.241E-05
2	1000	250	27.5	1.131E-05
3	1000	350	25	9.262E-06
4	1000	250	27.5	1.177E-05
5	1000	250	27.5	1.085E-05
6	900	350	27.5	1.64E-05
7	900	150	27.5	1.683E-05
8	1000	250	27.5	1.154E-05
9	1000	150	30	1.221E-05
10	1100	150	27.5	7.447E-06
11	900	250	30	1.661E-05
12	1100	350	27.5	7.94E-06
13	1100	250	25	8.811E-06
14	1000	350	30	1.188E-05
15	1000	150	25	1.0428E-05
16	900	250	25	1.65E-05
17	1100	250	30	8.448E-06

4.2 GA-BP neural network based process optimization

A BP neural network is an artificial neural network that uses error back propagation and contains an input layer, an output layer and an implicit layer, which can be used to solve classification and prediction problems. ^[36–38] In this paper, a BP neural network is used to predict the optimal process parameters for adapter forging. The input layer input is the factors affecting the die wear, the output layer output in the training process is the value of the die wear in the sample data set, and the output layer output in the prediction process is the predicted value of the die wear.

Genetic Algorithm (GA) is a method for finding optimal solutions by simulating the natural evolutionary process. The basic principle is to simulate the evolutionary phenomenon of superiority and inferiority in nature by mapping the search space into a genetic space, encoding the possible solutions in the search space into a vector, corresponding to biological chromosomes, with each element of the vector corresponding to biological genes; by continuously calculating the fitness value of each chromosome, the best chromosome is selected, thus obtaining the optimal solution. The main features of genetic algorithms are direct operations on structural objects, unconstrained by derivatives and function continuity; inherent implicit parallelism and better global search capability; a probabilistic search method that automatically acquires and guides the optimisation search space; and adaptive adjustment of the search direction without the need for defined rules. Genetic algorithms can well overcome the local optimum defects of BP artificial neural networks, optimize the initial weights and thresholds of BP artificial neural networks and improve their stability. ^[39–41]

The algorithm flow of the GA-BP neural network is shown in Fig. 17. First, the BP neural network is initialised and the coding length of the GA chromosome is determined according to the network structure; then, the BP neural network calculation error is used as the basis for constructing the fitness evaluation function, and the initial weights and thresholds of the optimised BP neural network are obtained through genetic operations; finally, the BP neural operation is executed again to obtain the calculation results.

The GA-BP model was built in MATLAB software. The BP neural network activation function was chosen as a single-level sigmoid function, as shown in Eq. (6). The fitness function of the genetic algorithm was constructed as the reciprocal of the cumulative error of the BP neural network, as shown in Eq. (7).

$$f(x)=\frac{1}{{1+{e^{ - x}}}}$$

Where, ${f_i}$ is the fitness value of the chromosome individual; ${d_k}$ and ${o_k}$represents the predicted and actual values of the die wear depth neural network generated from this individual; and $\theta$ is an offset with a small value greater than 0 so that the denominator is not zero. A larger ${f_i}$ indicates a smaller error and a higher fitness of the chromosome.

In this study, the inputs to the model are the initial forging temperature of the workpiece, the preheating temperature of the die and the forging speed, and the output of the model is the wear of the die, so the input layer in the established model is 3 and the output layer is 1. The hidden layer is usually determined by Eq. (8).。

$${h_1}=\sqrt {m+n} +a$$

Where, is the number of nodes in the input layer, is the number of nodes in the output layer, and is the regulation constant, which takes values in the range [1,10]. The number of neurons in the hidden layer can be calculated to be in the range [3,12]. The fit of the hidden layer neurons at different values was analysed by trial and error, and the network was more stable and better fitted when the number of hidden layer neurons was 8. Therefore, the BP artificial neural network structure in this paper is designed as 3-8-1.

Table 5 shows the relevant parameters of the genetic algorithm. Thirteen sets of the data were selected as training data and the rest as test data. In order to improve the speed,convenience of the network calculations and the accuracy of the predictions, the data were first normalised by using Eq. (9). Figure 18 shows the iterative process of adaptation in the genetic algorithm. It can be seen that the lowest value of adaptation is reached when there are about 30 iterations.

Table 5

Parameters of the genetic algorithm
Population size	Maximum iteration	Crossover probability	Variation probability
10	30	0.3	0.1

$$y=\frac{{\left( {{y_{\hbox{max} }} - {y_{\hbox{min} }}} \right) * \left( {x - {x_{\hbox{min} }}} \right)}}{{\left( {{x_{\hbox{max} }} - {x_{\hbox{min} }}} \right)}}+{y_{\hbox{min} }}$$

Where, ${y_{\hbox{max} }}$ and ${y_{\hbox{min} }}$ are the normalised higher and lower bounds; ${x_{\hbox{max} }}$ and ${x_{\hbox{min} }}$ are the maximum and minimum values in the set; and is the original data.

Figure 19 shows a comparison and error plot of the predicted and expected values of the GA-BP neural network. As can be seen from the figure, the prediction results of the GA-BP neural network are accurate. The results of the multiple regression analysis of the predicted results are shown in Fig. 20. The correlation coefficient for the training data was 0.998, indicating a good fit. Also, the R-value for the regression analysis used for both the validation data and the test data was 1, indicating that the model is quite reliable. The optimum process parameters predicted by the GA-BP neural network for the forging of the adapter were: the initial forging temperature of the workpiece was 1081.71°C, the preheating temperature of the die was 199.48°C and the forging speed was 25.33 mm/s. The predicted maximum wear of the die is 6.8723E-06mm. Combined with the response surface model (Fig. 16) and the comprehensive evaluation graph of thermal parameters on die wear (Fig. 14), it can be seen that the optimal parameters selected by the GA-BP model marshal are consistent with the most available process parameters derived from the response surface model and the comprehensive evaluation graph, which verifies the accuracy of the established neural network model from the side.

4.3 Experimental verification

A three-dimensional finite element model was developed to predict the die wear during hot forging of adapters by using T_w =1081.71 ℃,T_d=199.48℃,V = 25.33mm/s as the optimum parameters. Figure 20(a) and Fig. 20(b) show the wear distribution clouds of the top and bottom dies after the forging of the adapter is completed. It can be seen that the wear of the forging dies is reduced significantly after optimising the thermal parameters. According to the results of the finite element simulation, the maximum wear value of the top die is 7.38E-06mm and the result predicted by the GA-BP neural network is 6.8723E-06mm, the error between the two is 7.4%, indicating that the established GA-BP neural network model has a high accuracy in predicting the wear of the die.

Figure 22 shows the tool used for forging and forming the adapter. The process conditions in this condition are: T_w =1000 ℃,T_d =270℃,V = 25mm/s. It is worth noting that the process conditions are not optimised at this point.In particular, Fig. 22(a) and Fig. 22(b) show the top and bottom dies respectively, Fig. 22(c) shows a partial enlargement of the top die and Fig. 22(d) shows a test part of the adapter. As can be seen from Fig. 22(d), the various parts of the adapter forgings are fully formed and free from defects such as pits, especially the ring belt part of the saddle face is accurately formed to meet the subsequent machining requirements. However, as can be seen in Fig. 22(a), the surface of the forging die shows significant wear and even severe macroscopic cracking in localised areas of the die. As can be seen from the enlarged diagram of the top die, the wear areas of the top die are highly consistent with the FEM simulation results, with the greatest wear occurring on the left and right sides of the top die saddle ring band, with cracking occurring in the area marked by the red oval in the diagram. The simulation results show that the amount of wear in this area is always at a high level. The unreasonable process conditions and the accumulation of wear over a long period of time have led to an increase in wear in this area and ultimately to early cracking of the die.

Figure 23 shows a comparison of the dies under equivalent service conditions with using the optimised process parameters. Figure 23(a) and Fig. 23(c) show the top die before optimisation and its partial enlargement respectively, while Fig. 23(b) and Fig. 23(d) show the optimised schematic. It can be clearly seen that the optimised die does not crack at the maximum wear position under the same service conditions, indicating that the use of the optimised process parameters to produce the adapter can effectively reduce the amount of die wear and avoid the phenomenon of die cracking.

A three-dimensional thermodynamic coupled finite element model for predicting the wear behaviour of dies during precision hot forging of adapters was developed based on a modified Archard wear theory model. The analysis of the distribution and evolution of the die combination movement on the forging die wear during the precision hot forging of the adapters, and the influence of the forging process parameters on the amount of die wear, were investigated and the conclusions obtained are as follows:

In the case of the combined motion of multi-directional die forging with adapters, the solution proposed in this paper, in which the top die is moved to the specified stroke first and the left and right punches are moved to full forming, results in less wear on the die.

During the forging of the adapters, the wear areas of the dies are unevenly distributed but roughly symmetrical, the maximum wear of the top die is about 1.4 times that of the bottom die, and the serious wear areas of the top die are located in the ring belt on both sides, and the serious wear areas of the bottom die are located in the shoulder area.

The amount of wear in the cavity of the adapter forging die is negatively related to T_w and positively related to T_d. For T_w, a higher Tw reduces the flow stress of the forging material, resulting in a lower contact pressure between the workpiece and the die surface, effectively reducing the amount of wear in the die cavity. For T_d, a lower Td reduces the wear factor (K) and increases the hardness (H) of the die material, thus reducing the amount of cavity wear.

During the hot forging process of the adapter, the wear of the die is more sensitive to the change of T_w, and the use of high T_w and low T_d can significantly reduce the wear of the die. Based on the results of the single control variable it is clear that the optimum parameters for extending the service life of forging dies for adapters range from T_w=1000 ~ 1100℃ and T_d=150 ~ 250℃. Taking the influence of forging speed into account, the GA-BP model was used to further optimise this parameter range, and the optimal parameters for reducing the wear of forging dies for adapters were obtained as follows: T_w = 1081.71°C, T_d = 199.48°C and V = 25.33 mm/s.

Contribution

Hongchao Ji: Writing - original draft, Project administration, Conceptualization, Review & editing, Investigation,

Gang Song i: Review & editing, Investigation, Experimental,

Weichi Pei: Review & editing. Investigation,

Changzhe Song: Experimental, conceptualization,

Jingsheng Li: Experimental, Review & editing.

Weichi Pei: Investigation, Review & editing.

Baoyu Wang: Project administration.

Funding

This work is supported by the National Natural Science Foundation of China (51905501), this work is also supported by the Tangshan talent foundation innovation team (20130204D, 18130216A) and funded by S&P Program of Hebei (19012204Z), Key research and development plan of Tangshan science and technology (19140203F) and Science and Technology Project of Hebei Education Department (BJ2019010).

Conflict of interest: The authors declare no competing interests

LUO S, ZHU D (2017) Numerical analysis of die wear characteristics in hot forging of titanium alloy turbine blade [J]. Int J Mech Sci 123:260–270
BEHRENS B-A (2008) Finite element analysis of die wear in hot forging processes [J]. CIRP Ann 57(1):305–308
WILKUS M, POLAK S, GRONOSTAJSKI Z et al (2015) Modelling of the die wear in the hot forging process using the Archard model [J]. Comput Methods Mater Sci 15(2):311–321
SHARMA R K JAGOTAV (2021) SEHGAL R. Impact of austenitizing temperature on the wear behaviour of AISI H13 steel [J]. Proceedings of the Institution of Mechanical Engineers, Part J: Journal of Engineering Tribology, 235(3): 564 – 74
HILD R, BERGS T, MATTFELD P et al (2019) Analysis of wear phenomena during forward extrusion under dry friction conditions [J]. Wear 426:1362–1370
MORóN R, HERNáNDEZ-ONOFRE I, CONTLA-PACHECO A et al (2020) Friction and reciprocating wear behavior of borided AISI H13 steel under dry and lubricated conditions [J]. J Mater Eng Perform 29(7):4529–4540
BEHRENS B-A, BRUNOTTE K, WESTER H et al (2021) Multi-layer wear and tool life calculation for forging applications considering dynamical hardness modeling and nitrided layer degradation [J]. Materials 14(1):104
DWORZAK Ł, HAWRYLUK M (2021) The impact of the lubricant dose on the reduction of wear dies used in the forging process of the valve forging [J]. Materials 14(1):212
OGBEZODE J, AJIDE O (2021) Tool wear analysis of C-shaped equal channel reciprocating extrusion process of AISI-H13 steel die using finite element method [J]. J Brazilian Soc Mech Sci Eng 43(12):1–11
ZHANG C, LI P, WEI S et al (2019) Effect of tempering temperature on impact wear behavior of 30Cr3Mo2WNi hot-working die steel [J]. Front Mater 6:149
PENG Z, JI H, HUANG X et al (2021) Numerical Analysis and Parameter Optimization of Wear Characteristics of Titanium Alloy Cross Wedge Rolling Die [J]. Metals, 11(12): 1998
HAWRYLUK M, KASZUBA M, GRONOSTAJSKI Z et al (2020) Identification of the relations between the process conditions and the forging tool wear by combined experimental and numerical investigations [J]. CIRP J Manufact Sci Technol 30:87–93
DAVOUDI M, NEJAD A F, KOLOOR S S, R et al (2021) Investigation of effective geometrical parameters on wear of hot forging die [J]. J Mater Res Technol 15:5221–5231
RAJIEV R, SADAGOPAN P, PRAKASH RS (2020) Study on investigation of hot forging die wear analysis–An industrial case study [J]. Materials Today: Proceedings, 27: 2752-7
PAIVA JM, FOX-RABINOVICH G, LOCKS JUNIOR E et al (2018) Tribological and wear performance of nanocomposite PVD hard coatings deposited on aluminum die casting tool [J]. Materials 11(3):358
ZHI J, JIE Z, JIN-JIN J et al (2013) The effect of temperature condition on material deformation and die wear [J]. J Mater Eng Perform 22(7):2019–2028
ZHANG C, ZHAO G, LI T et al (2013) An investigation of die wear behavior during aluminum alloy 7075 tube extrusion [J]. J Tribol 135(1):011602
QIAO X, CHENG A, NIE X et al (2018) A study on die wear prediction for automobile panels stamping based on dynamic model [J]. Int J Adv Manuf Technol 97(5):1823–1833
XU W, LI W, WANG Y (2014) Experimental and theoretical analysis of wear mechanism in hot-forging die and optimal design of die geometry [J]. Wear 318(1–2):78–88
JAGOTA V, SHARMA R K (2020) Wear volume prediction of AISI H13 die steel using response surface methodology and artificial neural network [J]. J Mech Eng Sci 14(2):6789–6800
YU Y, ALBA D R, SCHAEFFER L et al (2019) Experimental evaluation of wear protection ability of sheet metal die covers in closed-die forging [J]. Prod Eng Res Devel 13(5):627–634
JIN Q, WANG W, JIANG R et al (2022) Wear and lifetime prediction of the roller cavities for the net-shape rolling blade [J].The International Journal of Advanced Manufacturing Technology, :1–11
HAWRYLUK M, ZIEMBA J, DWORZAK Ł et al (2018) Wear analysis of forging tools used in the hot forging processes using 3D reverse scanning techniques and cooling-lubricating system [J]. Int J Adv Manuf Technol 97(5):2009–2018
GROCHE P, HEß B (2014) Friction control for accurate cold forged parts [J]. CIRP Ann 63(1):285–288
KANG J, PARK I, JAE J et al (1999) A study on a die wear model considering thermal softening:(I) Construction of the wear model [J]. Journal of Materials Processing Technology, 96(1–3): 53 – 8
LEE R-S, JOU J-L (2003) Application of numerical simulation for wear analysis of warm forging die [J]. Journal of Materials Processing Technology, 140(1–3): 43 – 8
HU C, DING T, OU H et al (2019) Effect of tooling surface on friction conditions in cold forging of an aluminum alloy [J]. Tribol Int 131:353–362
DIXIT U S, RAJ A, PETROV P A et al (2021) Numerical simulations for studying the influence of friction in forging [J].Advances in Materials and Processing Technologies, :1–23
HAN X (2012) Friction behaviors in cold rotary forging of 20CrMnTi alloy [J]. Tribol Int 55:29–39
JIN J, SONG G, JI H et al (2021) Constitutive model of AISI 1035 at high temperature [J]. Metalurgija, 60(3–4): 257 – 60
WEI M, WANG S, WANG L et al (2011) Effect of tempering conditions on wear resistance in various wear mechanisms of H13 steel [J]. Tribol Int 44(7–8):898–905
ZHU J, ZHANG Z (2019) Improving strength and ductility of H13 die steel by pre-tempering treatment and its mechanism [J]. Mater Sci Engineering: A 752:101–114
NING A, MAO W, CHEN X et al (2017) Precipitation behavior of carbides in H13 hot work die steel and its strengthening during tempering [J]. Metals 7(3):70
GUANGHUA Y, XINMIN H, YANQING W et al (2010) Effects of heat treatment on mechanical properties of H13 steel [J]. Met Sci Heat Treat 52(7):393–395
JI H, SONG G, HUANG X et al (2022) Precision hot forging forming experiment and numerical simulation of a railway wagon bogie adapter [J].The International Journal of Advanced Manufacturing Technology, :1–19
LIU N, YANG H, LI H et al (2015) BP artificial neural network modeling for accurate radius prediction and application in incremental in-plane bending [J]. Int J Adv Manuf Technol 80(5):971–984
GAO L, LI F, DA HUO P et al (2021) Accurate prediction of the extrusion forming bonding reliability for heterogeneous welded sheets based on GA-BP neural network [J]. Int J Adv Manuf Technol 117(3):765–774
WAN P, ZOU H, WANG K et al (2020) Research on hot deformation behavior of Zr-4 alloy based on PSO-BP artificial neural network [J]. J Alloys Compd 826:154047
WANG S, ZHANG N, WU L et al (2016) Wind speed forecasting based on the hybrid ensemble empirical mode decomposition and GA-BP neural network method [J]. Renewable Energy 94:629–636
XU X, LIU F (2021) Optimization of online education and teaching evaluation system based on GA-BP neural network [J]. Computational Intelligence and Neuroscience, 2021
WANG Z, YE P, QIU F et al (2020) Crack characterization in ferromagnetic steels by pulsed eddy current technique based on GA-BP neural network model [J]. J Magn Magn Mater 500:166412

Download PDF

Editorial decision: Major Revisions Needed
14 Jul, 2022
Reviews received at journal
29 Apr, 2022
Reviewers invited by journal
29 Apr, 2022
Editor assigned by journal
28 Apr, 2022
First submitted to journal
26 Apr, 2022

You are reading this latest preprint version

Analysis of the wear characteristics of multi-directional die forging and forming dies for a railway wagon bogie adapter

Status:

Version 1

Abstract

Figures

1 Introduction

2 Numerical Model Building And Finite Element Simulation Analysis

2.1 Archard wear theory model

2.2 Finite element model building

3 Finite Element Simulation Results

3.1 Influence of the combination of die movements on the evolution of die wear

3.2 Influence of billet starting forging temperature on die wear distribution

3.3 Influence of die preheating temperature on die wear distribution

4 Process Optimization And Experimental Verification

4.1 Response surface model analysis

4.2 GA-BP neural network based process optimization

4.3 Experimental verification

5 Conclusion

Declarations

References

Status:

Version 1