Materialization of the Heat-Affected Zone with Laser Tailor-Welded HPF 22MnB5 Steel Using FLD and the Fracture Displacement Method in FE Simulation

Abstract

Using a tailor-welded blank (TWB) and hot-press forming (HPF), a 22MnB5 blank was surface-treated under four conditions. The penetration rates of the Fe_xAl_y compounds under the four surface-treatment conditions were investigated, and the hardness values were measured. A finite element (FE) simulation was performed for the characteristics of the heat-affected zone (HAZ), using the hardness value and results of previous researchers. In particular, the mechanical property settings of the mesh were designed to realize the conditions for the Fe_xAl_y compounds in the HAZ. Fine meshing was performed by partitioning the HAZ sections. For the mechanical properties of the HAZ with the Fe_xAl_y compounds, the strength was predicted from the hardness value, and the elongation values investigated by other researchers were used. The forming limit diagram, which was proportional to the elongation, was predicted. Specific elements were defined as the areas with Fe_xAl_y compounds, which played the same role as impurities. Tensile TWB–HPF specimens with different HAZ characteristics under four surface-treatment conditions were fabricated. Experiments and FE simulations were performed and compared. Details are as follows: For loads, a minimum error rate of 3% and a maximum error rate of 6% were obtained. For displacement, a minimum error rate of 9% and a maximum error of 25% were obtained. The feasibility of the simulation was verified by comparing the simulation and experimental results. A match of more than 75% was obtained.

1. Introduction

In the tailor-welded blank and hot-press forming (TWB–HPF) method, materials of different thicknesses are welded and heated in an austenite state, and simultaneously pressed and quenched. HPF steel showed different mechanical properties according to the heat treatment conditions [1,2,3,4,5]. The water-quenching method showed the best results and the lowest value in the air-cooling conditions [1]. The martensite phase generated by the aforementioned process has a strength of approximately 1500 MPa with an excellent tensile ratio; however, it has the disadvantage of requiring the phase to be welded to a thick material to compensate for its weakness [6]. In addition, studies have been performed to further improve the collision toughness of vehicles by attaching soft materials at major collision sites [7,8,9,10,11]. However, these advanced methods have limitations. In particular, in the TWB–HPF method, an Al–Si coating needs to be applied to the component surface to prevent the oxidation of the 22MnB5 steel. When welding without removing the coating layer, metal compounds such as FeAl₃ and Fe₂Al₅ permeate the base material and molten part of the corresponding component. During hot pressing, this metal interferes with the transition of the martensite phase into the ferrite–pearlite phase [12,13,14,15,16,17], thereby reducing the strength and elongation of the component. Consequently, the molded material has a vulnerable section in the heat-affected zone (HAZ), which is at risk during vehicle collisions.

Accordingly, several studies have been carried out on the removal techniques of the Al–Si coating layer [12]. A comparative study was conducted by examining separate conditions, such as the removal of the coating from a single surface and from both sides using sandblasting and laser ablation methods [12]. The removal of the coating layers from both sides of the material by laser ablation demonstrated optimal results. Meanwhile, under other conditions, the material did not exhibit sufficient hardness or strength for ultra-high-strength steel.

HAZ, which represents the weak area during the TWB–HPF method, has been investigated [18,19,20,21,22]. In a previous study, a model for the collision simulation was developed by producing a TWB via resistance welding of hot-pressed 22MnB5. The high hardness of the base metal implies increased mechanical property degradation of the HAZ. Tapered tensile specimens with different neck thicknesses were used for verification. The displacement corresponding to the fracture was reduced when the HAZ was located at a thinner part of the neck than that when it was located at a thick part [23]. In another study, a TWB was produced using a distinct laser welding method. Additional reinforcement using a filler wire increased the joint strength [24,25,26,27,28,29]. In these aforementioned studies, welding was performed without removing the Al–Si coating layer, resulting in a severe reduction in the strength of the HPF steel. These degraded mechanical properties are key factors in situations with sudden changes, such as automobile crashes.

Although many researchers have conducted studies on TWB–HPF parts, there have been few studies of finite element (FE) simulation. In particular, FE modeling that considers the deteriorated mechanical properties of the HAZ of the TWB–HPF remains insufficient because of the difficulty in experimentally obtaining the tensile strength, elongation, and forming limit diagram (FLD) of the HAZ. Y. Liu performed the quantitative measurement for inclusions and others to investigate the influence of Cu addition on microstructure and impact toughness in the simulated coarse-grained HAZ of high-strength low-alloy steels [30]. Therefore, limited FE simulation studies describe the actual properties of the HAZ. As fractures occur in the region of discontinuity and have the weakest physical properties, the HAZ containing Fe_xAl_y compounds, which are impurities, should not be ignored in the FE simulation.

In this study, a 22MnB5 steel was surface treated under four different conditions of the TWB–HPF method. In order to study the mechanical characteristics of TWB under these conditions, the welding area was divided into fine elements. The penetration rates of the Fe_xAl_y compounds according to the surface-treatment conditions were investigated, and their hardness values were measured. The Fe_xAl_y compound was assumed to be an impurity that causes cracking, and detailed results of each condition could be obtained by entering mechanical properties individually in an FE simulation. The characteristics of the HAZ were described in the FE simulation. The setting of the material properties and the mesh were designed to realize the conditions for the Fe_xAl_y compounds in the HAZ. Tensile specimens under four surface-treatment conditions were fabricated and tensile tests were performed. The feasibility of the simulation was verified by comparing the simulation and experimental results. As a result of this paper, through analyzing TWB–HPF materials used in automobiles and various industries, a new method is proposed to input the properties of HAZ, which is a weak part of collisions and forming.

2. Experimental Section

To examine the mechanical properties, a tensile specimen was machined from TWB–HPF 22MnB5 steel. The blanks prepared under the four surface-treatment conditions were subjected to the TWB–HPF processes, and then tensile and Vickers hardness tests were performed. Each condition was performed three times for the tensile test, and an ASTM-E8 standard specimen, illustrated in Figure 1, was tested at a crosshead speed of 2 mm/min.

2.1. Experimental Method

2.1.1. Material

For the TWB and HPF analyses, 22MnB5 steel was used. The chemical composition of the 22MnB5 sheet is listed in

Table 1. Chemical composition of 22MnB5 steel.

C	Si	Mn	P	S	Cr	Ti	B	N	Al	Fe
0.21	0.26	1.24	0.016	0.002	0.2	0.025	0.0024	0.002	0.033	Bal.

. Figure 2 illustrates the stress–strain curve of hot press formed 22MnB5 (22MnB5-HPF). The tensile strength and elongation were 1480 MPa and 4.8%, respectively. Figure 3 illustrates the macrostructures of the laser tailor-welded 22MnB5 (22MnB5-TWB) with and without a coating layer. Figure 3 presents the results before and after HPF, respectively. For the TWB with a coating layer, an Al-rich phase was observed in the fusion zone (weld metal). Moreover, Fe_xAl_y compounds, such as FeAl₂ and Fe₂Al₃, were formed. After HPF, Fe_xAl_y compounds were clearly identifiable. However, for the TWB, from which the coating layer was removed, no singularity was observed because of the similar structures of the base metal and the fusion zone after HPF.

2.1.2. Surface Treatment of TWB–HPF 22MnB5 Steel

The coating layer of the 22MnB5 blank was mechanically removed using sandpaper. The 22MnB5 blank was surface-treated to remove the Al–Si coating layer under different conditions, as listed in

Table 2. Conditions of surface treatment of 22MnB5 blank for removing Al–Si coating layer.

No.	Removal	Position of Treatment	Number of Repetitions
1	X	-	-
2	O	One side	One time
3	O	One side	Three times
4	O	Both side	Three times

. The blank surfaces of sample no. 1 were not treated. For sample no. 2, one surface of the blank was ground once using sandpaper. In sample no. 3, one surface of the blank was ground several times using sandpaper. For sample no. 4, both surfaces of the blank were ground several times using sandpaper.

Figure 4 presents a schematic of the surface treatment, TWB and HPF processes. A 22MnB5 blank was placed on a grinding machine equipped with a rotary sandpaper to remove the coating layer on the surface. This method uses a certified device patented as a coating layer removal machine for TWB in Korea. The contact area between the sandpaper and the sample was 3 mm and the contact time was within the range of 1–1.5 s. The gap was set according to the thickness of the blank with a bolt and it was ground by applying a certain pressing force with the spring. Sandpaper with a grit size of 80 was used, and the rotational speed of the grinding machine was set to 1710 rpm. After removing the coating layer, two 22MnB5 blanks were butt-welded using a laser source. Subsequently, the TWB was heated to 920 °C for 5 min in an electric furnace. After heating, 22MnB5-TWB was simultaneously press-formed and quenched. In the HPF process, the microstructure of 22MnB5-TWB was transformed into a martensite structure, and its mechanical properties, such as tensile strength and hardness, dramatically improved.

2.2. Experimental Results

Figure 5 illustrates the macrostructures of the 22MnB5-TWB tensile specimens according to the surface-treatment conditions listed in Table 2. In sample no. 1, the coating layer was not removed. As shown in Figure 5a, the Fe_xAl_y compounds were mixed throughout the HAZ in sample no. 1. Sample no. 2 had one surface that had been blank ground once, whereas the same surface was ground several times in sample no. 3. Figure 5b,c confirm the partial mixing of the Fe_xAl_y compounds in the HAZ for these samples. Figure 5b shows a slightly larger mixed area than that shown in Figure 5c. In sample no. 4, both surfaces of the blank were ground several times. The coating layer of the sample was completely removed, and the Fe_xAl_y compounds in the HAZ could not be confirmed, as shown in Figure 5d.

Figure 6 illustrates the Vickers hardness of the TWB tensile specimens obtained under four surface-treatment conditions. The Vickers hardness of sample no. 1 was in the range of 360–500 HV. In addition, the Vickers hardness value in the HAZ was in the range of 360–425 HV. The average hardness value was about 444 HV. For sample no. 2, the Vickers hardness ranged from 425 to 520 HV, and its HAZ was in the range of 425 to 460 HV. The average hardness value was about 475 HV. The Vickers hardness values of sample nos. 3 and 4 were in the range of 450–480 HV. The average hardness value was about 470 HV. In samples no. 1 and 2, the difference in the Vickers hardness values could be confirmed in the base metal and HAZ, in which the hardness of the HAZ was lower than that of the base metal. In contrast, the difference in the Vickers hardness of the HAZ and base metal in sample nos. 3 and 4 could not be confirmed.

The Al coating layer (white) infiltrated the HAZ in Figure 6a,b, and could deteriorate the mechanical properties owing to chemical fusion with the Fe_xAl_y content in the HAZ. For sample no. 3, the removal of most of the coating layer is confirmed in Figure 6c. In sample no. 4, the coating layer was completely removed, as shown in Figure 6d. In sample nos. 3 and 4, full martensite phases are predicted in the HAZ because of their hardness of at least 450 HV.

Figure 7a shows the load–displacement curves from the tensile tests of the samples after four conditions for the blank surface treatments. For sample no. 1, the maximum load of the specimen was 66.23 N, and the maximum displacement was 0.43 mm. The maximum load and displacement of sample no. 2 were 64.67 N and 0.59 mm, respectively, and those of sample no. 3 were 76.05 N and 0.73 mm, respectively. For sample no. 4, the maximum load and displacement were 86.58 N and 1.64 mm, respectively, which were 1.3 times (20.35 N) and 2.8 times (1.21 mm) the values of sample no. 1, respectively. This implies that the mechanical properties are sequentially improved as the Fe_xAl_y content decreases.

Figure 7b shows the stress–strain curves obtained from the tensile tests for the samples after four types of blank surface treatments. In sample no. 1, the maximum stress and strain were 910.6 MPa and 0.84%, respectively. The maximum stress and strain of sample no. 2 were 1131.7 MPa and 1.19%, respectively, and those of sample no. 3 were 1330.9 MPa and 1.47%, respectively. For sample no. 4, the maximum stress and strain were 1515.9 MPa and 3.29%, respectively.

The strength and elongation of sample no. 4 were 605.3 MPa and 2.45% larger than those of sample no. 1, respectively. The stress–strain curves indicate that the mechanical properties improved as the Fe_xAl_y content decreased. In addition, if the Fe_xAl_y compounds were included in the HAZ, the mechanical properties degraded. Thus, the coating layer must be removed before welding the blank.

3. Method of Finite Element Simulation

3.1. Defining the Method for the Mechanical Properties of the HAZ

Pavlina et al. [21] predicted the yield and tensile strengths of TWB based on the Vickers hardness, as defined in Equations (1) and (2), respectively:

$${\sigma }_Y=-90.7+2.876 \mathrm{HV}$$

(1)

$${\sigma }_T=-99.8+3.734 \mathrm{HV}$$

(2)

where σ_Y and σ_T denote the yield and tensile stresses, respectively, and HV denotes the Vickers hardness. The calculated yields and tensile strengths of the HAZs in the presence of the Fe_xAl_y compounds are listed in

Table 3. Engineering data used in the FE simulations.

Type	Hardness (HV)	Yield Strength, σY (MPa)	Tensile Strength, σT (MPa)	Elongation
HAZ without FexAly	450	1203.5	1580.5	1.80 [20]
HAZ with FexAly	360	944.6	1244.4	1.61 [20]
22MnB5—HPF	455	1217.88	1599.17	8.05

. The Vickers hardness values are those obtained from the experiments (Figure 6). The lowest hardness value in the HAZ was applied to the calculation. The calculated yield and tensile strength of the HAZ, including the Fe_xAl_y compounds, were 944.6 and 1244.4 MPa, respectively. The calculated yield and tensile strength of the HAZ excluding the Fe_xAl_y compounds were 1203.5 and 1580.5 MPa, respectively.

Figure 8 illustrates the FE modeling method for determining the HAZ of a TWB tensile specimen. The FE model was of the same size as that of an actual ASTM E8 tensile specimen. As shown in Figure 8, the base metal region and HAZ were divided by partitioning. Fine meshes (0.5 mm) were used for the areas where fractures were expected, whereas rough meshes (2 mm) were used for the areas in which fractures were not expected. The element type was S4, which was applied to reduce the hourglass effect. In particular, the HAZ was meshed with 100 elements (size of 0.5 mm × 0.5 mm) in an area with a length and width of 2.0 and 12.5 mm, respectively.

Figure 9 shows the mechanical property settings of the mesh in the HAZ obtained with four conditions of blank surface-treatment conditions. The mechanical properties of the red element sections were assigned to the HAZ, excluding Fe_xAl_y. The mechanical properties of the blue-element section were assigned to the HAZ, including Fe_xAl_y. Sample no. 1 was designed to be the blue-element section, which included the Fe_xAl_y compounds (blue 100%). Sample nos. 2 and 3 were designed so that several blue elements were in the red-element section. Sample no. 2 had six blue elements in the center (blue 6% and red 94%), and sample no. 3 had four blue elements in the center (blue 4% and red 96%). Similar to the role of the impurities, the Fe_xAl_y compounds were assumed to play a role in partially weakening the mechanical properties of the HAZ. Sample no. 4 designed the HAZ as a red-element section, excluding the Fe_xAl_y compounds (red 100%). The reverse engineering method based on the yield and tensile stresses specified in Table 3 was used. Comparing the experimental and simulation results, the proper elongation and FLD data were entered. Three material cards within ABAQUS, 22MnB5-HPF, HAZ without Fe_xAl_y, and HAZ with Fe_xAl_y, were set up. First, the cases of no. 1 and no. 4 were matched with the experimental results using the reverse engineering method. The mechanical properties of no. 1 were considered to be mechanical properties of HAZ with Fe_xAl_y for all 100 elements, which were welding areas. The mechanical properties of no. 4 were considered, due to the welding area, to be the mechanical properties of all 100 elements of HAZ without Fe_xAl_y. In other words, the properties of the blue area (HAZ with Fe_xAl_y) were assumed to be HAZ mixed with Fe_xAl_y composite because the coating layer was not removed. That is, the component with the lowest mechanical properties and the red area (HAZ without Fe_xAl_y) was assumed to be HAZ without the Fe_xAl_y compound due to surface grinding. Then, the process was performed that matched the HAZ without Fe_xAl_y and the HAZ with Fe_xAl_y properties, defined as the result of no. 1 and no. 4, above, with the experimental results of no. 2, and no. 3. In the process, the center that received the most deformation of the welding part was set as the vulnerable part, and it was set as a factor that causes the start of fracture. Under all four conditions, the properties of the base metal were designated as 22MnB5-HPF.

3.2. Damage Initiation

Figure 10 illustrates the FLDs of the materials used in the FE simulations. The theory of FLD damage was applied to the damage initiation in the material model for the tensile test simulation. The major and minor strain values were entered based on the 22MnB5-HPF FLD experiments. However, FLD experiments were difficult to perform for the materials of the HAZs with and without the Fe_xAl_y compounds. Therefore, the major strains of the HAZ with and without the Fe_xAl_y compounds were defined in proportion to the elongation value based on the 22MnB5-HPF FLD data. For the elongation, values investigated by Tom et al. were used. HAZ with Fe_xAl_y and HAZ without Fe_xAl_y were 1.61% and 1.80%, respectively [20].

For a minor strain of 12.0%, the major strain of 22MnB5-HPF was 10.0% and the major strains of the HAZs with and without the Fe_xAl_y compounds were 2.0% and 5.5%, respectively.

3.3. Damage Evolution

Necking began after the flow stress reached the ultimate tensile stress. Once necking had begun, damage evolution occurred and the flow stress began to decrease in the local region. The damage evolution values were derived using the following equations:

$${\epsilon }_{damage}={\epsilon }_f-{\epsilon }_o$$

(3)

where ε_damage is the damage strain, ε_f is the strain at the point of failure and ε_o is the strain during necking (ultimate tensile stress).

The ε_damage value indicates that each element undergoes additional deformation after necking, resulting in eventual failure. The failure displacement is obtained by multiplying the length of each element to define the damage evolution dependent on the mesh, as shown in Equation (4).

$$u_f=L_e{\epsilon }_{damage}=L_e\left({\epsilon }_f-{\epsilon }_o\right)$$

(4)

where u_f is the failure displacement and L_e is the element length. As shown in Figure 1, the ε_damage of 22MnB5-HPF was 0.0334. Because the element length of the fine mesh region at the HAZ was 0.5 mm, the u_f value was 0.0167 mm. Because the stress–strain curve data of the HAZ with and without the Fe_xAl_y compounds could not be obtained, the failure displacements were defined proportional to each elongation based on the 22MnB5-HPF data. The detailed FLD and damage evolution data are listed in

Table 4. Strain values obtained from the FLD curve.

22MnB5-HPF		HAZ without FexAly		HAZ with FexAly
Major, ε1	Minor, ε2	Major, ε1	Minor, ε2	Major, ε1	Minor, ε2
0.257	−0.15133	0.0514	−0.15133	0.14135	−0.15133
0.135	−0.055	0.027	−0.055	0.07425	−0.055
0.045	0	0.009	0	0.02475	0
0.06	0.0274	0.012	0.0274	0.033	0.0274
0.07	0.045	0.014	0.045	0.0385	0.045
0.08067	0.07	0.016134	0.07	0.0443685	0.07
0.100	0.120	0.055	0.120	0.020	0.120
Displacement at Failure
0.01670929		0.00334186		0.00461500

.

4. Comparison of the Experimental and FE Simulation Results

Figure 11 shows the load–displacement curves obtained from the tensile tests and FE simulations of the samples using four types of blank surface treatments. The values obtained from the experiments and FE simulations exhibited similar tendencies. These results indicate that the method of partitioning the HAZs used in the FE simulation and the setting method of the property assignment in the HAZ according to the four conditions for the blank surface treatment matched well with the experiment. It was necessary to define the mechanical properties for sample nos. 1–4 as the criteria for the four types of blank surface-treatment conditions. The mechanical properties of the HPF were tested experimentally, and those of the HAZ were determined based on the results described above.

The obtained results were as follows: the experimental and FE simulation maximum loads for sample no. 1 were 66.23 and 62.45 N, respectively, as shown in Figure 11a. The load error presented an extremely small difference of 0.30 N. The displacements obtained by the experimental and FE simulation methods were 0.43 and 0.51 mm, respectively, which were similar with an error of 0.08 mm. Sample no. 4 presented similar trends in the experiments and FE simulation, as shown in Figure 11d, in which the load values were similar at 86.60 and 83.90 N, respectively, indicating a small load error of 2.70 N. The experimental and simulated displacement values were 1.64 and 1.21 mm, respectively, which exhibited an error of 0.43 mm. The results of sample nos. 1 and 4 confirmed that the criteria determined based on the data collected in a prior study were set correctly [20,21]. The following results were obtained by reverse-tracking the Fe_xAl_y distributions in sample nos. 2 and 3, as shown in Figure 11b,c, respectively. The experimental and simulated load values obtained for sample no. 2 were 64.67 and 62.50 N, respectively, which presented a small difference of 2.17 N. The experimental and FE-simulated displacement values were 0.60 and 0.66 mm, respectively, which exhibited an error of 0.06 mm. The experimental and FE-simulated load values obtained under sample no. 3 were 76.05 and 70.76 N, respectively, which exhibited a difference of 5.29 N. The experimental and FE simulated displacement values were 0.74 and 0.81 mm, respectively, which exhibited an error of 0.07 mm.

Figure 12 illustrates the post-fracture appearance of the TWB tensile specimen obtained using four conditions of blank surface treatment. The test results revealed different fracture patterns depending on the surface treatment conditions. The fracture steps in the HAZ for the FE simulation based on the penetration of Fe_xAl_y are shown in Figure 12b,c. TWB tensile sample nos. 2 and 3 with Fe_xAl_y contents of 6% and 4%, respectively, were first destroyed in the Al-rich region, after which fracture propagation occurred throughout the welding line. In sample no. 2, cracks appeared in the blue-element section when the tensile specimen was stretched by 1.12 mm, once the test had started. Moreover, the fracture propagated within the HAZ when the tensile specimen was stretched by 1.29 mm. For sample no. 3, cracking occurred when the tensile specimen was stretched by 1.18 mm, and fracture propagation occurred when the tensile specimen was further stretched by 0.4 mm. However, in sample no. 4, fracture occurred in the base material of the specimen, which was located away from the welding line. For sample no. 1, fracture occurred in the welding line, similar to sample nos. 2 and 3. These results indicate that the experiments and simulations presented similar fracture mechanisms. Consequently, the Fe_xAl_y compounds in sample nos. 2 and 3 of the FE simulations were partially destroyed by the deterioration of the mechanical properties and propagated within the HAZ.

Figure 13 illustrates the stress–strain curves obtained from the tensile tests and FE simulations of the samples that underwent four types of blank surface treatments. The values obtained from the experiments and FE simulations for all conditions exhibited similar tendencies.

Figure 14 shows the tensile strength and elongation values extracted from Figure 13. The experimental values represent the average values for the three specimens under each condition, and the standard deviation was calculated. The error in the tensile strength was in the range of 25.5–97.6 MPa, and that of the elongation was 0.008–0.001. This result indicates similar errors in the tensile strength and elongation within 6.9% and 24.9%, respectively.

Figure 15 shows the fracture surfaces of sample nos. 1 and 4 observed through SEM. In both no. 1 and no. 4, a tendency towards ductility failure due to dimpling was observed, and cleavage facets were observed in Figure 15a. In the case of sample no. 1, the size of the microvoids was larger than those of no. 4, and the size was not uniform. In addition, in the case of no. 4, the occurrence of slip could be inferred through the fracture surface.

5. Conclusions

In this study, an FE simulation of TWB–HPF 22MnB5 steel with different surface treatments was conducted. This study provides a basic database for evaluating the crash characteristics of vehicles with TWB–HPF components and a method of numerical approach for TWB mechanical properties such as tensile, hardness, damage initiation and damage evolution. The damage initiation with FLDs and damage evolution using failure displacement of the element were applied to FE simulation. For the validation of the simulation results, the simulation and experimental results for the tensile tests were compared. The main conclusions are as follows.

• In the FE simulation, the different mechanical properties were applied to material sections such as 22MnB5-HPF and HAZ with and without Fe_xAl_y. When the different damage evolutions were given using the displacement method with Equation (4), the fracture occurred first at the HAZ with Fe_xAl_y, and then crack propagation to HAZ was implemented.
• The displacement of occurrence for crack and fracture propagation could be predicted. Cracks appeared in the blue element section when the tensile specimen was stretched by 1.12 to 1.18 mm once the test started. The fracture propagated within the HAZ when the tensile specimen was stretched by 1.29 to 1.58 mm.
• The FLD and fracture displacement method were applied to the simulation as the damage initiation and propagation,
• This study demonstrates damage initiation and the evolution of the weld line for TWB parts. It is possible to evaluate the collision characteristics of automotive parts using TWB–HPF, such as bumpers and B-pillars.