Process Parameter Optimization for Particles Reinforced Weld-Bonding Joints of DP780 Dual-Phase Steel

Abstract

Alumina particles were added in the adhesive layer to stabilize the welding process and enhance the load capacity for weld-bonding joints. In this study, the strength of alumina particles reinforced weld-bonding joints of DP780 dual-phase steel sheets was investigated by conducting experiments based on the Box–Behnken Design (BBD). The multivariate regression models between process parameters (welding current, welding time and electrode pressure) and response values (failure load, nugget diameter) were established, and the optimal combination of weld-bonding process parameters was obtained. The findings indicate that welding current has the biggest impact on the strength of joints. The optimum process parameters ranges are: welding current of 8.5 kA, welding time of 125 ms, and electrode pressure of 0.5 MPa. Developed regression models were validated by experiments.

Fig. 8 Effect of the interaction of welding current and electrode pressure on nugget diameter. (a) 3D response surface plot; (b) Contour plot.

1. Introduction

Dual-phase steel has the characteristics of high tensile strength and good yield performance, and it is one of the best materials which can achieve energy saving and meet the requirements of lightweight body and improve collision safety.^1,2) Weld bonding is the combination of adhesive bonding and resistance spot welding (RSW). It not only reduces the stress concentration of the spot welding joints, but also increases the static and fatigue load performance of joints. Therefore, it is widely used in industrial technology fields such as automobile manufacturing.^3–6) However, the most common adhesive used for weld bonding is epoxy resins, and it has some disadvantages, such as poor toughness and low peeling strength. In order to enhance the mechanical properties of the adhesive, alumina particles are used as the reinforced phase,^7,8) and then it mixed with high polymer materials such as epoxy resin in accordance with a certain process and proportion to form a new composite material and to improve the strength of the adhesive.⁹⁾

Khan et al.¹⁰⁾ studied the properties of the weld-bonding joints of 6061 aluminum alloy and T651 dissimilar materials, and found that the failure load value of the weld-bonding joints showed a single curve variation rule with the change of welding time and welding current. Muhammad et al.¹¹⁾ investigated an orthogonal experiment of the weld-bonding parameters and the nugget diameter. The results showed that the welding current has significant effect on the nugget diameter. Kaščák et al.¹²⁾ analyzed the influence of welding current on the failure load of dual-phase steel and found that within a certain welding current range, the failure load of RSW joints increases with the increase of welding current. Li et al.¹³⁾ analyzed the influence of the weld-bonding parameters on the nugget diameter of the joints. The results showed that the welding parameters have a significant effect on the nugget diameter, among which the welding current plays the most important role. Yi et al.¹⁴⁾ analyzed the influence of process parameters on the quality of DP780 high-strength steel weld bonded joints based on the Box-Behnken Design analysis method. The results showed that the failure load of the weld-bonded joints increases with the increase of welding current and welding time, while they decreases with the increase of electrode pressure. Khan et al.¹⁵⁾ established a response surface model to study the influence of welding process parameters on the tensile and shear strength of aluminum alloy 6061T651 sheets, and the results showed that the developed model for tensile shear strength could give a good prediction of tensile shear strength within the applied range of parameters. Boriwal et al.¹⁶⁾ analyzed the influence of welding process parameters on the welding joints quality of dissimilar plate by adopting the response surface method. The results showed that the mathematical model obtained by response surface method could effectively predict the tensile shear and peeling strength of the joints. Liu et al.¹⁷⁾ focused on the shortcomings of the weld-bonding technology. The conductivity of the adhesive was improved by adding aluminum powder, magnesium powder and other conductive metal powder into the adhesive. The results showed that with the increase of powder content, welding current and welding time, the nugget diameter and tensile shear strength of the joint first increased and then decreased. Literature survey revealed that the effect of welding parameters on the weld-bonding has been studied, but there are few studies relating to the particles reinforced weld-bonding, especially with the modeling of interaction effect of process parameters on failure load and nugget diameter.

In the present study, we focus on the process parameter optimization for particles reinforced weld-bonding of DP780 dual-phase steel. Based on the Box-Behnken Design (BBD) analysis method, the regression models were developed to study single parameter effect and interactions between various significant process parameters on the strength of the particles reinforced weld-bonding joints.

2. Experimental Procedure

2.1 Experiment materials and equipment

The substrate material used in the particles reinforced weld-bonding test was DP780 dual-phase steel. The DP460 epoxy resin adhesive was used in the weld-bonding joints, and alumina particles were used as the reinforced phase material in the adhesive. Based on the previous orthogonal experiment studies, the optimal parameters of the structural parameters of the adhesive layer were determined as follows: the adhesive layer thickness 0.4 mm, the alumina particles diameter 0.18 mm and the alumina particles mass fraction 10%. The thickness of the adhesive layer is achieved by copper gasket with a thickness of 0.4 mm. The dimension and lap length of the test specimen used in the present work are shown in Fig. 1. The determination of the test specimen size is based on the national standard GB2651-89. The test specimen was welded by using an intermediate-frequency inverter spot-welding machine MD-60. Before welding, the test specimen was sanded with a sand paper, and then anhydrous ethanol was used to scrub the polished test specimen to ensure the neatness of the test specimen. After the test specimen was polished, 0.3 g of alumina particles and 2.7 g of epoxy resin adhesive were taken and mixed evenly in a container. The fabricated adhesive was coated on the test specimen’s lap area. Finally, the test specimen was placed in a thermostat box to cure the adhesive layer. Curing time is 24 h and curing temperature is 25°C.

Fig. 1

Geometry of the particles reinforced weld-bonding joints.

2.2 Experimental design

The BBD response surface methodology was used to design the particles reinforced weld-bonding experiment. The process parameters which include welding time, welding current and electrode pressure were test factors. The mathematical regression model was established with the failure load value and the nugget diameter of the joints as the response value of the test. The range of process parameters was determined after conducting pre-experiments on spot welding and particles reinforced weld-bonding. The test factors and experimental levels are shown in Table 1. In order to reduce the test error, the specimens were prepared under each set of parameters for tensile shear test, and the final value of the failure load and nugget diameter of each group was the mean value of the specimens.

Table 1 Response surface experimental parameter.

3. Experimental Results and Analysis

3.1 The failure mode of joints

The tensile shear test results showed that there are two failure modes for the particles reinforced weld-bonding joints, that is, interface failure and plug failure. Figure 2 shows the fractures of the particles reinforced weld-bonding joints. In the weld-bonding process, it creates a large amount of heat that causes the bonding around the weld nugget to gasify and burn. Eventually a cavity region is formed, known as the adhesive layer burning zone. The reason for the interface failure is that the welding parameters are small and the heat input is lower, which leads to small nugget diameter and insufficient shear strength of weld nugget. Interface failure is an undesirable mode of fracture, while plug failure is a desired failure mode.

Fig. 2

Macroscopic fracture of particles reinforced weld-bonding joints. (a) Interface failure; (b) Plug failure.

3.2 Establishment of regression model

The failure load value and the nugget diameter of the joints are important indexes to evaluate the quality of joints. It is very important to analyze the relationship between failure load value and each process parameter. Therefore, the multivariate regression models between process parameters (welding current, welding time, and electrode pressure) and response values (failure load, nugget diameter) were established. The C-scan image of particles reinforced weld-bonding joints was obtained by ultrasonic scanning technique, which can be used to directly measure the nugget diameter (as shown in Fig. 3). The design factors and experimental results are shown in Table 2.

Fig. 3

C-scan image of the nugget diameter.

Table 2 Box-Behnken Design test design and the experimental results.

Figure 4(a) shows 13th group of specimen load-displacement curves with particles reinforced weld-bonding joints. The average failure load of the 13th group of specimen is 20669 N, and the average nugget diameter is 4.57 mm. The load-displacement curves of the weld-bonding joints without particles at same weld parameters are shown in Fig. 4(b). The average failure load value of weld-bonding joints without particles is 17886 N and the average nugget diameter is 5.13 mm. The test results show that the difference of the nugget diameter between the weld-bonding joints and the particles reinforced weld-bonding joints is not obvious. Because the adhesive in the weld nugget area is squeezed out by welding pressure in the welding process, particles in the adhesive layer have little influence on the forming of weld nugget.

Fig. 4

The load-displacement curves of the joints. (a) The particles reinforced weld-bonding joints; (b) The weld-bonding joints.

As can be observed, under identical welding conditions, the failure load of the particles reinforced weld-bonding joints is 13.5% higher than that of the weld-bonding joints. Meanwhile, the failure displacement of weld-bonding joints is half of the failure displacement of particles reinforced weld-bonding joints. It indicates that the addition of particles can transform the crack propagation pattern, which leads to a change in the stress distribution of the adhesive layer and the fracture process of joints. Therefore, the fracture resistance of the adhesive layer is enhanced, so that the mechanical properties of the particles reinforced weld-bonding joints are effectively improved.

The welding current, welding time and electrode pressure are taken as independent variables, which are represented by x₁, x₂ and x₃ respectively. The failure load value and nugget diameter of the joints are represented by F and D respectively. The least square method is used to fit the experimental data, and the obtained response surface equation is not significant when the standard quadratic model is adopted. In order to obtain a more significant model, the model is optimized by removing the insignificant factors. The optimized response surface regression models are shown in eqs. (1) and (2).

Failure load:   

\begin{align} F &= 581.071 + 2187.375x_{1} + 103.625x_{2} \\ &\quad + 4278.750x_{3} - 13.250x_{1}x_{2} \end{align} (1)

Nugget diameter:   

\begin{align} D &= 29.130 - 6.590x_{1} - 0.135x_{2} + 22.125x_{3} \\ &\quad + 0.020x_{1}x_{2} - 3.050x_{1}x_{3} + 0.400x_{1}^{2} \end{align} (2)

The analysis of variance (ANOVA) results obtained from the optimized model are shown in Table 3, and A, B and C in the table represent welding current, welding time and electrode pressure respectively. The objective of using analysis of variance ANOVA is to investigate whether the process parameters have significant effects on the quality of joints and the established model is significant.

Table 3 Variance analysis after model optimization.

The variance table of the response surface model is analyzed to evaluate the significance value. The P-value represents the significance level of the sample results. If P < α (α = 0.05), it indicates that the difference between the sample observation value and the overall hypothesis value is significant at the confidence level of 1 − α, so the null hypothesis cannot be accepted, which means the null hypothesis is refused. Thus, When the P value is less than 0.01, the factor is highly significant. When the P value is greater than 0.01 and less than 0.05, the factor is significant. On the contrary, when the P value is greater than 0.05, it means that the factor is not significant. It is found from Table 3 that the regression models of the failure load and nugget diameter are highly significant, indicating that the response surface model is highly reliable. As shown in Table 3, the ANOVA results indicate that three factors: welding current, electrode pressure and welding time, affected the failure load of the particles reinforced weld-bonding joints significantly according to the degree of influence. Furthermore, the interaction of welding time and welding current has a significant effect on the failure load of the joints. Three factors are welding current, welding time and electrode pressure, affected the nugget diameter of the joints significantly according to the degree of influence.

3.3 Interaction effect of welding process parameters on the failure load and nugget diameter

The results of ANOVA show that the welding time has the least influence on the failure load among the three process parameters. Thus, the welding current and electrode pressure are analyzed. The other two factors are kept as the median value to evaluate the influence of a single factor on the failure load. As shown in Fig. 5, the curve in the middle of the graph represents the average. The upper and lower curves are the least significant difference (LSD) curves determined by design, model, and confidence level. The failure load value increases with the increase of welding current and electrode pressure. The comparison of Figs. 5(a) and (b) shows that the influence curve slope of welding current on failure load is larger than that of electrode pressure, which indicates that the change of welding current has more obvious effect on failure load. When the additional two factors are kept unchanged, a larger welding current parameter is required to get a higher failure load within a certain range. For example, groups 2 and 14 in Table 2 show that a larger welding current can obtain a larger failure load. It is because the higher welding current can increase the diameter of the weld nugget.

Fig. 5

Effect of single factor on failure load. (a) Welding current; (b) Electrode pressure.

It is very important to analyze the interaction of various factors for the optimized welding process parameters to obtain a joint with the best comprehensive performance. The influence of the interaction between welding current and welding time on the failure load value is analyzed when the electrode pressure is 0.4 MPa. The 3D response surface plot and contour plot are shown in Figs. 6(a) and (b). As illustrated, within the range of 6.5∼7 kA, it is not obvious that the tendency of the failure load value increases with the increase of the welding time. When the welding current is 7.5∼8.5 kA, the failure load increases as the welding time decreases, indicating that a larger failure load value can also be obtained with a shorter welding time. The reason is that with a welding current of 7.5∼8.5 kA, the lower welding time can avoid over-burn of the particles reinforced weld-bonding joints, which have a reasonable nugget diameter. Additionally, the failure load of the joints reaches the maximum value when the welding current takes the maximum value and the welding time takes the minimum value.

Fig. 6

Effect of the interaction of welding current and welding time on failure load. (a) 3D response surface plot; (b) Contour plot.

The results of ANOVA show that the electrode pressure has the least influence on the nugget diameter among the three process parameters, thus, only the welding current and welding time were analyzed. The effect of single factor on the nugget diameter is shown in Fig. 7, and the other two factors are kept in the median value of the value range. As shown in Figs. 7(a) and (b), the nugget diameter increases sharply when the welding current and welding time increase. Figure 7(a) shows that the slope of the curve increases gradually, which shows that the influence of changing welding current on the nugget diameter is more obvious when large current is used. Therefore, the change range of nugget diameter caused by changing welding current is larger when the other two factors are kept in the median value of the value range.

Fig. 7

Effect of single factor on nugget diameter. (a) Welding current; (b) Welding time.

The interaction between welding current and electrode pressure on the nugget diameter is analyzed when the welding time is 125 ms. The 3D response surface plot and contour plot are shown in Figs. 8(a) and (b). As shown in Fig. 8(a), when the welding current is small, the electrode pressure has little effect on the diameter. When the welding current is large, the decrease of the electrode pressure has distinct effect on the increase of the nugget diameter. This is because the higher electrode pressure increases the effective contact area between the steel sheets and reduces the contact resistance. When the contact resistance is reduced, the resistance heat generated is reduced and the nugget diameter is reduced. Figure 8(b) shows that the nugget diameter can be significantly increased by increasing welding current when the electrode pressure is low. When the welding current is large and the electrode pressure is low, changing these two parameters at the same time can make the nugget diameter increase sharply. However, according to the test process, when the welding current is large and the electrode pressure is low, the joints are prone to splash because of the excessive heat input during welding. Therefore, the larger welding current and lower electrode pressure are generally not used as welding process parameters.

Fig. 8

Effect of the interaction of welding current and electrode pressure on nugget diameter. (a) 3D response surface plot; (b) Contour plot.

3.4 Process parameter optimization and experimental verification

The optimization of the response surface model was performed by using failure load and nugget diameter as the primary objectives. However, obtaining a larger nugget diameter requires choosing a larger welding current and a smaller electrode pressure. And under these welding process parameters, the joints are prone to splashing due to excessive heat input in practices. The splashing of the welding joint can cause defects in the weld nugget and reduce the mechanical properties of the joints. Therefore, obtaining a larger failure load should be taken as the primary objective for the process parameter optimization of particles reinforced weld-bonding.

Numerical optimization is carried out following the hill climbing technique. Firstly, let X, a vector of x_i for i = 1 ⋯ n, represent design variables over the optimization space, which is a subset of the design space. Secondly, let y_j, U_j, L_j for j = 1 ⋯ m be responses with upper and lower bounds to serve as constraints. Finally, let y(X) be the response to be optimized (failure load and nugget diameter value); then f(X) = y(X) for minimization, f(X) = −y(X) for maximization. Constraints are defined as g_j(X), a series of discontinuous functions (eq. (3)):   

\begin{align} g_{j}(X) & = y_{j}(X) - U_{j} \quad \text{for}\ y_{j} > U_{j}\\ g_{j}(X) & = 0 \qquad \qquad\quad\text{for}\ \text{L}_{j} \leq y_{j} \leq U_{j}\\ g_{j}(X) & = \text{L}_{j} - y_{j}(X) \quad \text{for}\ y_{j} < \text{L}_{j} \end{align} (3)

This produces a system (eq. (4)) of m constraints that can be solved as an unconstrained problem via a penalty function approach:   

\begin{equation} \textit{Minimize}\left(f(X) + p\sum_{j}g_{j}(X)\right) \end{equation} (4)

Where p is a penalty parameter > 0 for j = 1 to m.

The optimal process parameters based on the regression models are as follows: welding current 8.5 kA, welding time 150 ms and electrode pressure 0.5 MPa. However, over-burn occurs under these parameters. This is because the longer welding time and the larger welding current lead to the excessive heat input of the joints. So it is necessary to reduce the welding time or decrease the welding current to avoid over-burn. The response surface analysis shows that the influence of welding current on the failure load is greater than the welding time, and the failure load of the joints can be improved by choosing a larger welding current and a shorter welding time. Therefore, it is necessary to appropriately reduce the welding time of the calculated optimal process parameters. After experimentation, it is found that when the welding time is reduced to 125 ms, the joint without over-burn can be obtained. Finally, the optimal welding parameters are determined as follows: welding current 8.5 kA, welding time 125 ms and electrode pressure 0.5 MPa.

In order to verify the reliability of the response surface model, experiments were performed with three different groups of process parameters. Each group of experiments was repeated three times, and the experimental results are shown in Table 4. Substituting the first group of process parameters into the regression model, the calculated failure load was 19015 N and the nugget diameter was 3.82 mm. Then the failure load and the nugget diameter obtained from the experiment were 18138 N and 4.08 mm. The results show that the errors between the predicted values of the established model and the experimental values are 4.8% and 6.4% in turn and the errors are very small and within a reasonable range. Therefore, the response surface method in this experiment can be used to establish the corresponding mathematical regression model to predict the mechanical performance and optimize the process parameters of the particles reinforced weld-bonding joints.

Table 4 Validation experiment table.

4. Conclusion

1. (1)    Alumina particles, added in the adhesive for weld-bonding, can increase the failure load of the weld-bonding joints.
2. (2)    The optimal process parameters for alumina particles reinforced weld-bonding joints were found to be: 8.5 kA welding current, 125 ms welding time, and 0.5 MPa electrode pressure. The results show that the errors between prediction values and experimental values are within 7%. The developed model gives a good prediction of failure load and nugget diameter of particles reinforced weld-bonding joints within a certain range of parameters.
3. (3)    The interaction of welding current and welding time is the most significant effect on the failure load. The interaction analysis indicates that a higher failure load can be obtained by choosing a larger welding current and a shorter welding time. The results of the interaction between welding current and electrode pressure on the nugget diameter show that a larger welding current and a lower electrode pressure should be selected to obtain a larger nugget diameter value. While the electrode pressure should be reasonably selected to avoid welding defects in the particles reinforced weld-bonding process.

Acknowledgments

This study is supported by the National Natural Science Foundation of China (Grant No. 51565022).

REFERENCES

• 1)   Q.L.  Cui,  D.  Parkes,  D.  Westerbaan,  S.S.  Nayak,  Y.  Zhou,  D.C.  Saha,  D.  Liu,  F.  Goodwin,  S.  Bhole and  D.L.  Chen: J. Mater. Eng. Perform. 26 (2017) 783–791. doi:10.1007/s11665-016-2454-0
• 2)   M.  Sabzi,  S.M.  Far and  S.M.  Dezfuli: Int. J. Mater. Prod. Technol. 59 (2019) 3–15.
• 3)   X.C.  He: Int. J. Adhes. Adhes. 48 (2014) 59–66. doi:10.1016/j.ijadhadh.2013.09.017
• 4)   A.  Kumar,  G.  Sharma and  D.K.  Dwivedi: Int. J. Adhes. Adhes. 84 (2018) 350–359. doi:10.1016/j.ijadhadh.2018.04.012
• 5)   X.Y.  Sun,  K.  Zeng,  X.C.  He and  L.  Zhang: Nondestruct. Test. Eval. 32 (2017) 242–254. doi:10.1080/10589759.2016.1241251
• 6)   A.M.  Pereira,  J.A.M.  Ferreira,  F.V.  Antunes and  P.J.  Bártoloet: J. Adhes. Sci. Technol. 28 (2014) 1432–1450. doi:10.1080/01694243.2012.698107
• 7)   A.C.  Cadore-Rodrigues,  C.  Prochnow,  T.A.L.  Burgo,  J.S.  Oliveira,  S.L.  Jahn,  E.L.  Foletto,  M.P.  Rippe,  G.K.R.  Pereira and  L.F.  Valandro: J. Adhes. Dent. 22 (2020) 149–159. doi:10.3290/j.jad.a44151
• 8)   A.  Drah,  T.  Kovačević,  J.  Rusmirović,  N.  Tomić,  S.  Brzić,  M.  Bogosavljavić and  A.  Marinković: J. Compos. Mater. 53 (2019) 2727–2742. doi:10.1177/0021998319839133
• 9)   Y.S.  Zhao,  Y.H.  He,  K.R.  Yang,  X.P.  Wang,  J.H.  Bai and  B.  Du: High Volt. 5 (2020) 472–481. doi:10.1049/hve.2019.0108
• 10)   M.F.  Khan,  G.  Sharma and  D.K.  Dwivedi: Int. J. Adv. Manuf. Technol. 78 (2015) 863–873. doi:10.1007/s00170-014-6670-1
• 11)   N.  Muhammad,  Y.H.P.  Manurung,  R.  Jaafar,  S.K.  Abas,  G.  Tham and  E.  Haruman: J. Intell. Manuf. 24 (2013) 1175–1183. doi:10.1007/s10845-012-0648-3
• 12)   Ľ.  Kaščák,  E.  Spišák and  I.  Gajdoš: Key Eng. Mater. 635 (2014) 143–146. doi:10.4028/www.scientific.net/KEM.635.143
• 13)   M.F.  Li,  Y.J.  Wang,  Z.  Niu and  S.L.  Yang: Automot. Innov. 3 (2020) 221–230. doi:10.1007/s42154-020-00106-0
• 14)   J.Q.  Yi,  K.  Zeng,  X.C.  He,  B.Y.  Xing and  Y.Y.  Feng: Mater. Trans. 61 (2020) 700–707. doi:10.2320/matertrans.MT-M2019295
• 15)   M.D.F.  Khan,  D.K.  Dwivedi and  S.  Sharma: Mater. Des. 34 (2012) 673–678. doi:10.1016/j.matdes.2011.05.025
• 16)   L.  Boriwal,  R.M.  Sarviya and  M.M.  Mahapatra: J. Adhes. Sci. Technol. 31 (2017) 1591–1616. doi:10.1080/01694243.2016.1266844
• 17)   L.M.  Liu and  D.X.  Ren: Mater. Des. 32 (2011) 3730–3735. doi:10.1016/j.matdes.2011.03.050