Investigation of the Hot Stamping-in-Die Quenching Composite Forming Process of 5083 Aluminum Alloy Skin

Aluminum alloy has been used as the skin material for rail vehicles and automobiles to meet the requirements of environmental protection. The hot stamping-in-die quenching composite forming (HFQ) process is a promising technology to compensate for the poor formability of the aluminum alloy sheet at room temperature. In this paper, the high-temperature mechanical properties of 5083 aluminum alloy under various temperature (200 °C, 300 °C, 400 °C, 450 °C) and strain rate conditions (0.01 s−1, 0.10 s−1, 1.00 s−1) were investigated by uniaxial tensile tests. The finite element software of PAM-STAMP was employed to simulate the forming process of high-speed train skin. The effects of forming method and process parameters on the minimum thickness and springback of the skin were analyzed using the Response Surface Methodology (RSM). After parameter optimization, the forming experiment verified the simulation results and the test part met the quality requirements: the thickness above 3.84 mm and the springback within 1.1 mm. Mechanical properties of the sheet before and after HFQ were examined by uniaxial tensile tests at room temperature. It can be inferred from the comparison that the yield strength of the Al5083 sheet increases, but the elongation decreases from the HFQ process.


Introduction
The problems of environmental degradation and energy shortage are worsening with the continuous development of industry and transportation. The development and application of lightweight technology are indispensable in achieving the goals of energy saving and emissions reduction [1]. Light weight products are often achieved by optimizing the structure and using lightweight materials [2,3]. Compared with steel, aluminum alloys have the advantages of low density and high specific strength and are widely used in the field of lightweight materials [4][5][6]. In the aviation industry, aluminum alloys are mainly employed in components such as fuselage skins, wings, pressure chambers, and fairing [7][8][9]. Additionally, there is a trend for aluminum to be used as a body material [10,11]. The greenhouse gas emissions of passenger cars using aluminum alloys are lower than those of regular passenger cars from a life-cycle perspective [12]. Furthermore, aluminum alloy is a predominant material used in the body of high-speed trains [13][14][15]. Deformed aluminum alloys are divided into two types according to whether or not they can be strengthened by heat treatment. The 2xxx and 7xxx aluminum alloys are granted improved strength primarily by heat treatment operations to modulate the diffusely precipitated second phase [16,17]. The 5xxx aluminum alloy differs from them because the transition phase produced during aging is not coherent with the matrix, and the equilibrium phase is prone to distribution along grain boundaries [18]. This type of alloy is usually used only in the annealed or cold-hardened state.
Cold stamping represents the prevalent forming method for aluminum alloy sheets owing to its high production efficiency and low cost. The poor forming performance at room temperature is prone to defects, such as cracking and springback in the cold stamping process [19,20], which limits the use of aluminum alloys. Lin et al. [11] examined the formability of stamping an engine hood with an A6181-T4 aluminum alloy sheet and changed the stamping process parameters to eliminate the crack defect in the test part. Huang et al. [21] studied the effect of different yield functions on the springback prediction of automotive parts stamped from the 5754 aluminum alloy sheet.
The researchers have focused on the deformation behavior of aluminum alloys at elevated temperatures to overcome the drawback [22,23]. The superplastic forming (SPF) process is applied to form parts with high accuracy. However, low productivity and nonuniform thickness distribution are its major problems [24]. A new technology called the quick superplastic forming process combines the hot drawing pre-forming and superplastic forming processes. Some researchers have applied this manufacturing process to make the side wall outer panel of metro vehicles and bonnets [25,26]. The hot stamping-indie quenching composite forming process (HFQ) for aluminum alloys was proposed by Lin [27][28][29][30][31][32]. This process is an ideal method for solving the problems of aluminum alloys at room temperature and is more productive than SPF [33][34][35][36]. Fan et al. [37] investigated the HFQ process of 6A02 aluminum alloy sheets at different forming-die temperatures ranging from 50 • C to 350 • C and found the temperature should not be higher than 250 • C to obtain a sufficient strengthening effect. Harrison [16] used the method of hot stamping and then artificial aging to form the 7075 aluminum alloy B-pillar outer plate with little springback. Wang et al. [38] tested the ductility of AA2024 at different temperatures. It has very high formability at around 450 • C, as confirmed by the cup test.
In this paper, the HFQ process was used to form a large-size aluminum alloy highspeed train skin. The effect of the two processes (hot stamping and drawing) on the forming quality of the skin was compared by finite element methods. Response Surface Methodology (RSM) was adopted to explore the effect of parameters including the stamping speed, die clearance, and holding time on the minimum thickness and springback, and to optimize and obtain the appropriate process parameters. The forming experiment was performed using the simulation results to verify the accuracy of the FEA. Uniaxial tensile tests have verified the strengthening effect of this process on Al5083.

Material
The material of the high-speed train skin is Al5083 in H111 condition with a thickness of 4 mm. The main chemical composition is shown in Table 1. Electron Backscattered Diffraction (EBSD) detection was performed on the RD-TD surface of the specimen. The initial microstructure of Al5083 is shown in Figure 1. Figure 1a was based on the inverse pole figure (IPF) of (001) and was color coded according to its grain orientation. It can be seen that Al5083 is composed of the equiaxed grains with an average grain size of 12.87 µm. Figure 1b shows the distribution of misorientation angle. The misorientation angle is defined as lower than 15 • as low-angle grain boundaries (LAGBs), and higher than 15 • as high-angle grain boundaries (HAGBs). It presented with 70% as the HAGBs fraction and 30% as the LAGBs fraction, which indicates that the material exhibits good toughness.

Material Model
To investigate the mechanical properties of Al5083 at high temperature, the hot uniaxial tensile tests were performed in a WDW-300 electronic universal testing machine. The size of the tensile specimen is shown in Figure 1. Forces and displacements were automatically collected during the tensioning process. The PID thermostat automatically controlled the temperature with an accuracy of ±2 °C in the heating process. At present, there are many studies on the hot forming process of aluminum alloys, and the deformation temperature ranges basically between 200 °C and 450 °C [39][40][41]. Moreover, the strain rate of the sheet is high during hot stamping and the deformation mechanism is a glide-controlled thermally activated dislocation mechanism, instead of the Grain Boundary Sliding (GBS) and the Solute Drag creep (SD) that prevail in SPF and Quick-Plastic Forming (QPF), respectively [42]. Therefore, the tensile tests were conducted at different temperatures of 200 °C, 300 °C, 400 °C, and 450 °C. The strain rates at each temperature were 0.01 s −1 , 0.1 s −1 , and 1 s −1 . To investigate the anisotropy of the material under these temperatures and strain rate conditions, the tensile specimens were cut in directions that are 0°, 45°, and 90° from the rolling direction (RD), as shown in Figure 2b. The true stress-true plastic strain curves for Al5083 at different temperatures are shown in Figure 3. It is clearly seen that increasing the strain rate or decreasing the deformation temperature results in a monotonic increase in the flow stress before reaching the tensile strength. At the same temperature, elongation rises as the strain rate decreases. Furthermore, the higher the temperature, the greater the elongation with the same strain rate. The elongation exceeds 230% when the specimen was subjected to tensile stress at 450 °C with a strain rate of 0.01 s −1 . Because the larger strain rate leads to a higher dislocation density, the dynamic recovery process cannot reduce the dislocation density effectively within the limited deformation time. Moreover, the deformation at low-temperatures leads to more stored energy, which increases the driving force for grain boundary sliding but also reduces the ability of recovery and grain boundary sliding. It can also be found that under the same deformation conditions, the higher the temperature, the smaller the strain rate and the greater the difference in mechanical properties caused by the tensile direction. These curves were input to PAM-STAMP as discrete data points. The

Material Model
To investigate the mechanical properties of Al5083 at high temperature, the hot uniaxial tensile tests were performed in a WDW-300 electronic universal testing machine. The size of the tensile specimen is shown in Figure 1. Forces and displacements were automatically collected during the tensioning process. The PID thermostat automatically controlled the temperature with an accuracy of ±2 • C in the heating process. At present, there are many studies on the hot forming process of aluminum alloys, and the deformation temperature ranges basically between 200 • C and 450 • C [39][40][41]. Moreover, the strain rate of the sheet is high during hot stamping and the deformation mechanism is a glide-controlled thermally activated dislocation mechanism, instead of the Grain Boundary Sliding (GBS) and the Solute Drag creep (SD) that prevail in SPF and Quick-Plastic Forming (QPF), respectively [42]. Therefore, the tensile tests were conducted at different temperatures of 200 • C, 300 • C, 400 • C, and 450 • C. The strain rates at each temperature were 0.01 s −1 , 0.1 s −1 , and 1 s −1 . To investigate the anisotropy of the material under these temperatures and strain rate conditions, the tensile specimens were cut in directions that are 0 • , 45 • , and 90 • from the rolling direction (RD), as shown in Figure 2b.

Material Model
To investigate the mechanical properties of Al5083 at high temperature, the hot uniaxial tensile tests were performed in a WDW-300 electronic universal testing machine. The size of the tensile specimen is shown in Figure 1. Forces and displacements were automatically collected during the tensioning process. The PID thermostat automatically controlled the temperature with an accuracy of ±2 °C in the heating process. At present, there are many studies on the hot forming process of aluminum alloys, and the deformation temperature ranges basically between 200 °C and 450 °C [39][40][41]. Moreover, the strain rate of the sheet is high during hot stamping and the deformation mechanism is a glide-controlled thermally activated dislocation mechanism, instead of the Grain Boundary Sliding (GBS) and the Solute Drag creep (SD) that prevail in SPF and Quick-Plastic Forming (QPF), respectively [42]. Therefore, the tensile tests were conducted at different temperatures of 200 °C, 300 °C, 400 °C, and 450 °C. The strain rates at each temperature were 0.01 s −1 , 0.1 s −1 , and 1 s −1 . To investigate the anisotropy of the material under these temperatures and strain rate conditions, the tensile specimens were cut in directions that are 0°, 45°, and 90° from the rolling direction (RD), as shown in Figure 2b. The true stress-true plastic strain curves for Al5083 at different temperatures are shown in Figure 3. It is clearly seen that increasing the strain rate or decreasing the deformation temperature results in a monotonic increase in the flow stress before reaching the tensile strength. At the same temperature, elongation rises as the strain rate decreases. Furthermore, the higher the temperature, the greater the elongation with the same strain rate. The elongation exceeds 230% when the specimen was subjected to tensile stress at 450 °C with a strain rate of 0.01 s −1 . Because the larger strain rate leads to a higher dislocation density, the dynamic recovery process cannot reduce the dislocation density effectively within the limited deformation time. Moreover, the deformation at low-temperatures leads to more stored energy, which increases the driving force for grain boundary sliding but also reduces the ability of recovery and grain boundary sliding. It can also be found that under the same deformation conditions, the higher the temperature, the smaller the strain rate and the greater the difference in mechanical properties caused by the tensile direction. These curves were input to PAM-STAMP as discrete data points. The The true stress-true plastic strain curves for Al5083 at different temperatures are shown in Figure 3. It is clearly seen that increasing the strain rate or decreasing the deformation temperature results in a monotonic increase in the flow stress before reaching the tensile strength. At the same temperature, elongation rises as the strain rate decreases. Furthermore, the higher the temperature, the greater the elongation with the same strain rate. The elongation exceeds 230% when the specimen was subjected to tensile stress at 450 • C with a strain rate of 0.01 s −1 . Because the larger strain rate leads to a higher dislocation density, the dynamic recovery process cannot reduce the dislocation density effectively within the limited deformation time. Moreover, the deformation at low-temperatures leads to more stored energy, which increases the driving force for grain boundary sliding but also reduces the ability of recovery and grain boundary sliding. It can also be found that under the same deformation conditions, the higher the temperature, the smaller the strain rate and the greater the difference in mechanical properties caused by the tensile direction. These curves were input to PAM-STAMP as discrete data points. The mechanical properties of the material at other temperatures and strain rate conditions during forming would be calculated by the interpolation. The yield function of the Hill48 model considering only the normal anisotropy [43] is expressed as follows: where R is the normal anisotropy index, which is strongly influenced by temperature and strain rate. Therefore, for convenient calculation and safe engineering design, and the average normal anisotropy index R was set to 0.7 [44].
Materials 2023, 16, x FOR PEER REVIEW 4 of 14 mechanical properties of the material at other temperatures and strain rate conditions during forming would be calculated by the interpolation. The yield function of the Hill48 model considering only the normal anisotropy [43] is expressed as follows: where R is the normal anisotropy index, which is strongly influenced by temperature and strain rate. Therefore, for convenient calculation and safe engineering design, and the average normal anisotropy index R was set to 0.7 [44]. Micro-hardness tests were performed on specimens that had undergone tensile tests, and Figure 4 shows the Vickers hardness of the Al5083 at different temperatures and strain rates. The Vickers hardness initial sheet is 93 HV. The Vickers hardness basically decreases with increasing temperature and strain rate. Overall, the Vickers hardness shows little change compared to the as-received state. Micro-hardness tests were performed on specimens that had undergone tensile tests, and Figure 4 shows the Vickers hardness of the Al5083 at different temperatures and strain rates. The Vickers hardness initial sheet is 93 HV. The Vickers hardness basically decreases with increasing temperature and strain rate. Overall, the Vickers hardness shows little change compared to the as-received state.

Finite Element Modeling of the HFQ Process
The geometric model of the aluminum alloy skin is shown in Figure 5. The part is large and there are some complex space shapes with local sharp edges and corners. It is

Finite Element Modeling of the HFQ Process
The geometric model of the aluminum alloy skin is shown in Figure 5. The part is large and there are some complex space shapes with local sharp edges and corners. It is difficult to form a part of high quality using cold forming because the sharp corner seems to be at risk of cracking and the springback may lead to poor shape accuracy.

Finite Element Modeling of the HFQ Process
The geometric model of the aluminum alloy skin is shown in Figure 5. The part is large and there are some complex space shapes with local sharp edges and corners. It is difficult to form a part of high quality using cold forming because the sharp corner seems to be at risk of cracking and the springback may lead to poor shape accuracy. In the process of numerical simulation, a commercial finite element software PAM-STAMP was developed with the coupled temperature-displacement deformation mode. The tools and the blank were discretized using Belytschko-Tsay (BT) shell elements. The initial size of the mesh was 5 mm, and the maximum adaptive level was 4 to guarantee both calculation speed and accuracy. The Coulomb friction model was used in the hot stamping process with a friction coefficient of 0.13.
In order to obtain the die surface, the part model was supplemented with a process that guarantees the same curvature of the part. The part can be produced by either the hot stamping or the drawing process. The FE simulation procedures for the two processes are shown in Figure 6. The hot stamping process includes gravity loading, stamping, quenching, springback, and trimming. When the hot stamping process is adopted, the sheet would wrinkle or show springback deformation because it is difficult to control the sheetʹs flow. Drawing has an additional step of holding before stamping compared to the hot stamping process. The sheet would crack because of the low flow stress at high temperature if the drawing process is employed. Therefore, it is necessary to discuss the effect of the two forming processes on formability. In the process of numerical simulation, a commercial finite element software PAM-STAMP was developed with the coupled temperature-displacement deformation mode. The tools and the blank were discretized using Belytschko-Tsay (BT) shell elements. The initial size of the mesh was 5 mm, and the maximum adaptive level was 4 to guarantee both calculation speed and accuracy. The Coulomb friction model was used in the hot stamping process with a friction coefficient of 0.13.
In order to obtain the die surface, the part model was supplemented with a process that guarantees the same curvature of the part. The part can be produced by either the hot stamping or the drawing process. The FE simulation procedures for the two processes are shown in Figure 6. The hot stamping process includes gravity loading, stamping, quenching, springback, and trimming. When the hot stamping process is adopted, the sheet would wrinkle or show springback deformation because it is difficult to control the sheet's flow. Drawing has an additional step of holding before stamping compared to the hot stamping process. The sheet would crack because of the low flow stress at high temperature if the drawing process is employed. Therefore, it is necessary to discuss the effect of the two forming processes on formability.

Effect of the Forming Process on the Forming Result
The thickness and plastic strain distribution of the skin formed by the two processes are shown in Figure 7. As seen in Figure 7a, the thickness of the sheet at the sharp central corner is thinnest, where the equivalent strain is 0.055 when the stamping is completed (Figure 7b). However, the minimum thickness of the skin formed by drawing is 3.619 mm, located at the entrance fillet of the die as the effect of the blankholder (Figure 7c). From Figure 7d, it can be seen that the equivalent plastic strain at the corresponding location is 0.104. Moreover, the thickness distribution of the skin formed by hot stamping is more

Effect of the Forming Process on the Forming Result
The thickness and plastic strain distribution of the skin formed by the two processes are shown in Figure 7. As seen in Figure 7a, the thickness of the sheet at the sharp central corner is thinnest, where the equivalent strain is 0.055 when the stamping is completed  Figure 7b). However, the minimum thickness of the skin formed by drawing is 3.619 mm, located at the entrance fillet of the die as the effect of the blankholder (Figure 7c). From Figure 7d, it can be seen that the equivalent plastic strain at the corresponding location is 0.104. Moreover, the thickness distribution of the skin formed by hot stamping is more uniform than another.

Effect of the Forming Process on the Forming Result
The thickness and plastic strain distribution of the skin formed by the two processes are shown in Figure 7. As seen in Figure 7a, the thickness of the sheet at the sharp central corner is thinnest, where the equivalent strain is 0.055 when the stamping is completed (Figure 7b). However, the minimum thickness of the skin formed by drawing is 3.619 mm, located at the entrance fillet of the die as the effect of the blankholder (Figure 7c). From Figure 7d, it can be seen that the equivalent plastic strain at the corresponding location is 0.104. Moreover, the thickness distribution of the skin formed by hot stamping is more uniform than another. Springback is an important indicator to judge the quality of the part. The greater the springback, the worse the shape accuracy. The springback distribution of the skin formed by the two processes is shown in Figure 8a,c. The springback increases gradually from the middle to the two edges because it is hard for the center of the sheet to deform owing to the inhibition of the free end sheet. When using the drawing process, the maximum springback is 2.581 mm, as shown in Figure 8c. The reason can be analyzed by looking at the temperature distribution of the sheet before removing it from the die (Figure 8b,d). The greater the difference in temperature between the pressing and the deformed area, the larger the deformation by releasing thermal stress. Springback is an important indicator to judge the quality of the part. The greater the springback, the worse the shape accuracy. The springback distribution of the skin formed by the two processes is shown in Figure 8a,c. The springback increases gradually from the middle to the two edges because it is hard for the center of the sheet to deform owing to the inhibition of the free end sheet. When using the drawing process, the maximum springback is 2.581 mm, as shown in Figure 8c. The reason can be analyzed by looking at the temperature distribution of the sheet before removing it from the die (Figure 8b,d). The greater the difference in temperature between the pressing and the deformed area, the larger the deformation by releasing thermal stress. To further illustrate the reason for the springback distribution, the von Mises st distribution of the blank before and after springback were analyzed for the two proces To further illustrate the reason for the springback distribution, the von Mises stress distribution of the blank before and after springback were analyzed for the two processes, as shown in Figure 9. Springback is a stress-release process, thus the von Mises stress is much lower after springback. The maximum stress in the sheet formed by hot stamping before springback is 0.166 GPa (Figure 9c), which is 0.01 GPa greater than that of the sheet formed by drawing (Figure 9a). The hot stamping process causes a smaller springback. To further illustrate the reason for the springback distribution, the von Mise distribution of the blank before and after springback were analyzed for the two pr as shown in Figure 9. Springback is a stress-release process, thus the von Mises much lower after springback. The maximum stress in the sheet formed by hot st before springback is 0.166 GPa (Figure 9c), which is 0.01 GPa greater than that of th formed by drawing (Figure 9a). The hot stamping process causes a smaller spring Considering the thinning and springback, it is more suitable for the high-spe skin to form by hot stamping.

Effect of Forming Parameters on the Forming Result
In hot stamping, the stamping speed affects the sheet temperature during mation. Thus, it directly changes the deformation capacity of the material. The d ance is closely related to the shape accuracy of the part and the wear of the die. The back is strongly influenced by the holding time. The minimum thickness refle formity, and the springback represents the shape accuracy of the workpiece. The R Surface Methodology (RSM) could derive a continuous functional relationship b the responses and multiple factors [45]. Therefore, RSM was adopted to explore th of parameters including the stamping speed, die clearance, and holding time, Considering the thinning and springback, it is more suitable for the high-speed train skin to form by hot stamping.

Effect of Forming Parameters on the Forming Result
In hot stamping, the stamping speed affects the sheet temperature during deformation. Thus, it directly changes the deformation capacity of the material. The die clearance is closely related to the shape accuracy of the part and the wear of the die. The springback is strongly influenced by the holding time. The minimum thickness reflects uniformity, and the springback represents the shape accuracy of the workpiece. The Response Surface Methodology (RSM) could derive a continuous functional relationship between the responses and multiple factors [45]. Therefore, RSM was adopted to explore the effect of parameters including the stamping speed, die clearance, and holding time, on the minimum thickness and springback, and to optimize and obtain the appropriate process parameters. The experimental design was performed using the Box-Behnken Design (BBD) in order to avoid too many experiments reducing the computational efficiency. The experimental design and results are presented in Table 2.
A linear model was established between the minimum thickness and the process parameters after analysis, as shown in Equation (2). The model F-value of 7.4 implies the model is significant. The functional relationship diagram is shown in Figure 10. In this experimental scheme, the minimum thickness varies in a small range, from 3.717 mm to 3.865 mm, which meets thinning rate requirements of less than 10%. The minimum thickness is inversely correlated with the die clearance and directly correlated with the stamping speed. When the stamping speed becomes faster, the more serious the local plastic deformation and the more obvious the thinning. The die clearance is related to the deformation of the sheet, with the increase in the die clearance, the smaller the deformation. However, the effect of holding time on the minimum thickness is negligible. A linear model was established between the minimum thickness and the process parameters after analysis, as shown in Equation (2). The model F-value of 7.4 implies the model is significant. The functional relationship diagram is shown in Figure 10. In this experimental scheme, the minimum thickness varies in a small range, from 3.717 mm to 3.865 mm, which meets thinning rate requirements of less than 10%. The minimum thickness is inversely correlated with the die clearance and directly correlated with the stamping speed. When the stamping speed becomes faster, the more serious the local plastic deformation and the more obvious the thinning. The die clearance is related to the deformation of the sheet, with the increase in the die clearance, the smaller the deformation. However, the effect of holding time on the minimum thickness is negligible.
(2) Figure 10. The effect of the single factor on the thickness. A quadratic model with an F-value of 14.08 was developed between the springback and the process parameters, as shown in Equation (3). The influence of parameters on springback is complex and the interaction between the factors needs to be considered. The effect of the two-factor interaction on the springback is shown in Figure 11. The springback is related to the temperature distribution of the sheet at the end of the hot stamping. If the temperature of the sheet is low and evenly distributed before opening the mold, then the springback will be smaller. R2 = 41.345 − 0.033A − 17.051B − 0.379C + 0.004AB − 0.00003AC + 0.0828BC + 0.00007A 2 + 1.872B 2 + 0.0008C 2 (3) effect of the two-factor interaction on the springback is shown in Figure 11. The springback is related to the temperature distribution of the sheet at the end of the hot stamping If the temperature of the sheet is low and evenly distributed before opening the mold then the springback will be smaller. R2 = 41.345 − 0.033A − 17.051B − 0.379C + 0.004AB − 0.00003AC + 0.0828BC + 0.00007A 2 + 1.872B 2 + 0.0008C 2 (3)

Optimization of the Process Parameters
In order to obtain the part that meets the thickness and shape requirements, the process parameters were optimized using the hill climbing technique based on the established regression model. To maximize the minimum thickness and minimize the springback, the optimized parameter combination was stamping speed of 131 mm/s, die clearance of 4.16 mm, and holding time of 15 s.

Experimental Verification
The forming experiment was conducted by a box-type resistance furnace with a big chamber size of 2500 × 2000 × 1000 mm and a maximum heating temperature of 960 °C The sheet was formed at 450 °C since the plasticity was greatest at this temperature [46] The tools were made of H13 steel with high hardenability and resistance to thermal cracking (Figure 12a). When machining tools, a clearance of 4.16 mm was kept between the punch and the die. The friction between the sheet and the tool has a significant impact on the service life of the tool and the forming quality of the part [47,48]. Boron nitride is a common lubricant employed in aluminum alloy hot stamping, and its Coulomb friction coefficient under this condition is 1-1.5 [49]. The boron nitride was uniformly sprayed on both sides of the sheet before the experiment. The sheet was heated to 480 °C in the heating furnace and then kept there for 2 min. Then, the sheet was transferred to the die with the heat insulation clamp, and when the temperature dropped to 450 °C, the punch was moved down at a speed of 131 mm/s. Holding pressure lasted for 15 s after closing the die, and then the sheet was air-cooled to room temperature; the test part is shown in Figure 12b.

Optimization of the Process Parameters
In order to obtain the part that meets the thickness and shape requirements, the process parameters were optimized using the hill climbing technique based on the established regression model. To maximize the minimum thickness and minimize the springback, the optimized parameter combination was stamping speed of 131 mm/s, die clearance of 4.16 mm, and holding time of 15 s.

Experimental Verification
The forming experiment was conducted by a box-type resistance furnace with a big chamber size of 2500 × 2000 × 1000 mm and a maximum heating temperature of 960 • C. The sheet was formed at 450 • C since the plasticity was greatest at this temperature [46]. The tools were made of H13 steel with high hardenability and resistance to thermal cracking (Figure 12a). When machining tools, a clearance of 4.16 mm was kept between the punch and the die. The friction between the sheet and the tool has a significant impact on the service life of the tool and the forming quality of the part [47,48]. Boron nitride is a common lubricant employed in aluminum alloy hot stamping, and its Coulomb friction coefficient under this condition is 1-1.5 [49]. The boron nitride was uniformly sprayed on both sides of the sheet before the experiment. The sheet was heated to 480 • C in the heating furnace and then kept there for 2 min. Then, the sheet was transferred to the die with the heat insulation clamp, and when the temperature dropped to 450 • C, the punch was moved down at a speed of 131 mm/s. Holding pressure lasted for 15 s after closing the die, and then the sheet was air-cooled to room temperature; the test part is shown in Figure 12b.

Thickness and Shape Checking
As shown in Figure 13a, the test part was cut at 497 mm intervals and the cutting sections were defined as Section A-D. Each section was divided into 18 segments with 19 separation points, and the thickness at these points was measured (Figure 13b). A comparison between numerical and experimental thickness distributions of the skin in four sections is shown in Figure 14. Section-A and Section-D are located at the ends of the skin and show slight wrinkling at the bottom sharp corner and entrance fillet (Figure 14), for the shallow forming depth. However, sections B and D are located in the middle of the skin, and the fillets thinning obviously caused by the large deformation. It can be seen that the deviation of thickness for the four sections is less than 5% from the experimental result.

Thickness and Shape Checking
As shown in Figure 13a, the test part was cut at 497 mm intervals and the cutting sections were defined as Sections A-D. Each section was divided into 18 segments with 19 separation points, and the thickness at these points was measured (Figure 13b). A comparison between numerical and experimental thickness distributions of the skin in four sections is shown in Figure 14. Section-A and Section-D are located at the ends of the skin and show slight wrinkling at the bottom sharp corner and entrance fillet, for the shallow forming depth. However, sections B and D are located in the middle of the skin, and the fillets thinning obviously caused by the large deformation. It can be seen that the deviation of thickness for the four sections is less than 5% from the experimental result. The finite element model correctly predicted the quality of the part formed using the HFQ process. The thickness of the skin was uniformly distributed (the range is 3.86-4.00 mm). The max thinning was less than 10%, which meets the technical requirements.

Thickness and Shape Checking
As shown in Figure 13a, the test part was cut at 497 mm intervals and the cutting sections were defined as Section A-D. Each section was divided into 18 segments with 19 separation points, and the thickness at these points was measured (Figure 13b). A comparison between numerical and experimental thickness distributions of the skin in four sections is shown in Figure 14. Section-A and Section-D are located at the ends of the skin and show slight wrinkling at the bottom sharp corner and entrance fillet (Figure 14), for the shallow forming depth. However, sections B and D are located in the middle of the skin, and the fillets thinning obviously caused by the large deformation. It can be seen that the deviation of thickness for the four sections is less than 5% from the experimental result. The finite element model correctly predicted the quality of the part formed using the HFQ process. The thickness of the skin was uniformly distributed (the range is 3.86-4.00 mm). The max thinning was less than 10%, which meets the technical requirements.  The shape accuracy of the outer surface is important for the skin, in order to e the beauty of the high-speed train. The skin was placed on the solid inspection too positioned. As shown in the Figure 15a, the section templates are machined by sel 13 positions along its longitudinal direction using the laser cutting method, based shape of the outer surface of the standard part. The clearance between the test pa the template was measured using a feeler gauge and was defined as the section acc deviation. It can be seen from Figure 15b that the accuracy deviations of the 13 se meet the accuracy requirement of less than 1.1 mm and are nearly consistent wi simulation results. The shape accuracy of the outer surface is important for the skin, in order to ensure the beauty of the high-speed train. The skin was placed on the solid inspection tool to be positioned. As shown in the Figure 15a, the section templates are machined by selecting 13 positions along its longitudinal direction using the laser cutting method, based on the shape of the outer surface of the standard part. The clearance between the test part and the template was measured using a feeler gauge and was defined as the section accuracy deviation. It can be seen from Figure 15b that the accuracy deviations of the 13 sections meet the accuracy requirement of less than 1.1 mm and are nearly consistent with the simulation results.
positioned. As shown in the Figure 15a, the section templates are machined by selecting 13 positions along its longitudinal direction using the laser cutting method, based on the shape of the outer surface of the standard part. The clearance between the test part and the template was measured using a feeler gauge and was defined as the section accuracy deviation. It can be seen from Figure 15b that the accuracy deviations of the 13 sections meet the accuracy requirement of less than 1.1 mm and are nearly consistent with the simulation results.

Material Properties Checking
In order to test the material properties of the parts, a tensile test was performed. The geometry of the specimen for the tensile tests and the true stress-true strain curves of the initial and deformed sheet at room temperature are shown in Figure 16. After deformation, the yield strength increased by 18.5% and the elongation after fracture decreased significantly because of work-hardening of the material. The dislocation density increased as plastic deformation proceeded, resulting in dislocation entanglement and other obstacles. Therefore, E in dislocation resistance allows the strength of the alloy to be higher.

Material Properties Checking
In order to test the material properties of the parts, a tensile test was performed. The geometry of the specimen for the tensile tests and the true stress-true strain curves of the initial and deformed sheet at room temperature are shown in Figure 16. After deformation, the yield strength increased by 18.5% and the elongation after fracture decreased significantly because of work-hardening of the material. The dislocation density increased as plastic deformation proceeded, resulting in dislocation entanglement and other obstacles. Therefore, E in dislocation resistance allows the strength of the alloy to be higher.

Conclusions
In this paper, the HFQ process of Al5083 high-speed train skin was investigated by finite element simulation and forming experiments. The conclusions are as follows: 1. The tensile tests of Al5083 in directions that are 0°, 45°, and 90° from RD at different temperatures (200 °C, 300 °C, 400 °C, 450 °C) and strain rates (0.01 s −1 , 0.10 s −1 , 1.00 s −1 ) revealed that increasing the strain rate or decreasing the deformation temperature resulted in a monotonic increase in the flow stress. The elongation increased with the decreasing strain rate at the same temperature. For the same strain rate, the higher the temperature, the higher the elongation. Elongation could exceed 230% when the specimen was tensile at 450 °C with a strain rate of 0.01 s −1 . 2. The hot stamping process was proven by finite element simulation to be more suitable for the formation of skin than the hot drawing at 450 °C. The effects of stamping speed, die clearance, and holding time on forming quality during the HFQ process were investigated. Based on the RSM, the linear and quadratic regression models Figure 16. The geometry of the specimen for tensile tests and true stress-true strain curve of Al5083.

Conclusions
In this paper, the HFQ process of Al5083 high-speed train skin was investigated by finite element simulation and forming experiments. The conclusions are as follows: 1.
The tensile tests of Al5083 in directions that are 0 • , 45 • , and 90 • from RD at different temperatures (200 • C, 300 • C, 400 • C, 450 • C) and strain rates (0.01 s −1 , 0.10 s −1 , 1.00 s −1 ) revealed that increasing the strain rate or decreasing the deformation temperature resulted in a monotonic increase in the flow stress. The elongation increased with the decreasing strain rate at the same temperature. For the same strain rate, the higher the temperature, the higher the elongation. Elongation could exceed 230% when the specimen was tensile at 450 • C with a strain rate of 0.01 s −1 .