4.1 Investigation of input parameters
In this study, using the Taguchi technique, the parameters of current intensity, welding speed, and the amount of siliconized Zn-graphene oxide complex nanoparticles in three levels were coded as input parameters, and the variables geometrical of weld bead, including penetration depth, bead height, and bead width were considered as output process parameters were measured. The S/N and analysis of variance (ANOVA) analyses were done to establish the relationship between input parameters and output variables. ANOVA aims to evaluate the impact of input parameters on penetration depth, bead height, and bead width. It gives a clear image of how far the input parameter affects the output characteristics and the impact rate of each factor. Furthermore, ANOVA can be performed to see which process parameter is statistically significant for each quality characteristic. In addition, the S/N approach is utilized by the Taguchi technique to measure the deviations of quality characteristics from the desired value. The S/N ratio is the ratio of “Signal,” representing the desirable value and the mean of output characteristics, and the “noise,” representing the undesirable value and squared deviation of the output characteristics [29–32]. The penetration depth values measured based on the test design matrix are shown in Table 3. Figure 2 also shows the effect of the three parameters of current intensity, welding speed, and the nanoparticle amount on the penetration depth. According to the results shown in the graphs of Fig. 2, if the current intensity was in level one with the value of 240 amps, the welding speed should be in level three with a value of 36 cm.min− 1, and the nanoparticle amount in level one with the value of 0.25 g, it will be possible to obtain a weld bead with the highest penetration depth. On the one hand, as shown in the analysis of variance table (Table 4), the current intensity parameter had the slightest effect, and the nanoparticle amount had the most significant impact on the penetration depth. In such a way that level 1 of the nanoparticle's parameter, with the amount of 0.25 g, will have the highest penetration depth, and in the second level, with the amount of 0.5 g, the penetration depth will be reduced, and finally, in level 3, with the amount of 0.75 g of nanoparticles, the penetration depth will be the lowest. Also, regarding the welding speed parameter in terms of cm.min-1, it can be stated that with the increase in the welding speed of the electrode, the penetration depth first decreased and then increased. Finally, at a rate of 36 cm/min, the maximum value of the penetration depth was reached. The degree of influence this parameter was about 17% on the penetration depth in the welding samples.
Table 3
Orthogonal array for L9 with response for penetration depth (mean value and S/N ratio)
Array type: L-9 | Current intensity (amp) | Welding speed (cm/min) | Nanoparticle (g) | Penetration depth |
Mean value | S/N ratio |
L1 | 1 | 1 | 1 | 8.9 | 18.987 |
L2 | 1 | 2 | 2 | 7.8 | 17.841 |
L3 | 1 | 3 | 3 | 8.2 | 18.276 |
L4 | 2 | 1 | 2 | 7.8 | 17.841 |
L5 | 2 | 2 | 3 | 7.4 | 17.384 |
L6 | 2 | 3 | 1 | 8.9 | 18.987 |
L7 | 3 | 1 | 3 | 7.4 | 17.384 |
L8 | 3 | 2 | 1 | 8.3 | 18.381 |
L9 | 3 | 3 | 2 | 8 | 18.061 |
Table 4
Analysis of Variance table for penetration depth
ANOVA for penetration depth (means) |
Factors | Degrees of freedom | Sums of squares | Adjusted mean square | Fisher ratio | Pure sum of squares | Percentage of Contribution |
Current intensity | 2 | 0.248 | 0.124 | 28.075 | 0.240 | 9.619 |
Welding speed | 2 | 0.435 | 0.217 | 49.123 | 0.426 | 17.097 |
Nanoparticle | 2 | 1.802 | 0.901 | 203.265 | 1.793 | 71.861 |
Other/Error | 2 | 0.008 | 0.004 | - | - | 1.423 |
Total: | 8 | 2.495 | | | | 100% |
ANOVA for penetration depth (S/N ratio) |
Current intensity | 2 | 0.286 | 0.143 | 18.705 | 0.271 | 9.553 |
Welding speed | 2 | 0.505 | 0.252 | 33.017 | 0.490 | 17.276 |
Nanoparticle | 2 | 2.030 | 1.015 | 132.600 | 2.015 | 71.011 |
Other/Error | 2 | 0.015 | 0.007 | - | - | 2.160 |
Total: | 8 | 2.838 | | | | 100% |
Table 5 shows the measured values of the bead height in the welded samples based on Taguchi's design and the mean value and S/N ratio. Figure 3 shows the effects of current intensity, welding speed, and nanoparticle amount on the bead height. According to the results, if the current intensity and the nanoparticle amount are at level one and the welding speed is equal to 36 cm.min-1, we will have a weld with the lowest bead height value. According to the variance analysis table (Table 6), the nanoparticle amount was the most influential factor on the bead height, in such a way that the bead height decreased by changing the nanoparticle amount from 0.25 g to 0.75 g. The nanoparticle amount in level one with 0.25 g will have the lowest bead height; in level three, with 0.75 g of nanoparticles, it will be the highest. On the other hand, by electing higher values of current intensity, the bead height increases, while increasing the welding speed has the opposite result, and increasing the welding speed decreases the bead height.
Table 5
Orthogonal array for L9 with response for bead height (mean value and S/N ratio)
Array type: L-9 | Current intensity (amp) | Welding speed (cm/min) | Nanoparticle (g) | Bead height |
Mean value | S/N ratio |
L1 | 1 | 1 | 1 | 1.6 | 4.082 |
L2 | 1 | 2 | 2 | 1.5 | 3.521 |
L3 | 1 | 3 | 3 | 2.1 | 6.444 |
L4 | 2 | 1 | 2 | 1.9 | 5.575 |
L5 | 2 | 2 | 3 | 2.4 | 7.604 |
L6 | 2 | 3 | 1 | 1.5 | 3.521 |
L7 | 3 | 1 | 3 | 3 | 9.542 |
L8 | 3 | 2 | 1 | 1.8 | 5.105 |
L9 | 3 | 3 | 2 | 1.85 | 5.343 |
Table 6
Analysis of Variance table for bead height
ANOVA for bead height (means) |
Factors | Degrees of freedom | Sums of squares | Adjusted mean square | Fisher ratio | Pure sum of squares | Percentage of Contribution |
Current intensity | 2 | 0.353 | 0.176 | 48.994 | 0.346 | 18.352 |
Welding speed | 2 | 0.200 | 0.100 | 27.765 | 0.193 | 10.235 |
Nanoparticle | 2 | 1.327 | 0.663 | 183.747 | 1.319 | 69.882 |
Other/Error | 2 | 0.006 | 0.003 | - | - | 1.531 |
Total: | 8 | 1.888 | | | | 100% |
ANOVA for bead height (S/N ratio) |
Current intensity | 2 | 5.908 | 2.954 | 107.172 | 5.853 | 18.575 |
Welding speed | 2 | 2.755 | 1.377 | 49.971 | 2.699 | 8.567 |
Nanoparticle | 2 | 22.792 | 11.396 | 413.427 | 22.737 | 72.156 |
Other/Error | 2 | 0.054 | 0.027 | - | - | 0.702 |
Total: | 8 | 31.511 | | | | 100% |
Table 7 shows the mean value and S/N ratio measured for the weld bead in the samples. Figure 4 also shows the graphs related to the effects of input parameters on the bead width as an output variable. The results show that increasing the current intensity from 240 amps at level one to 280 amps at level three has reduced the bead width. However, the results obtained regarding the welding speed were the opposite. According to the variance analysis table (Table 8), the nanoparticle amount in level three with 0.75 g has a smaller bead width than 0.25 g and 0.5 g, so to have a weld bead with the smallest width, the current intensity on level three with a value of 280 amp, the value of welding speed should be placed at level one with a value of 32 cm.min-1 and the nanoparticle amount at level three with 0.75 g.
Table 7
Orthogonal array for L9 with response for bead width (mean value and S/N ratio)
Array type: L-9 | Current intensity (amp) | Welding speed (cm/min) | Nanoparticle (g) | Bead width |
Mean value | S/N ratio |
L1 | 1 | 1 | 1 | 11.2 | 20.984 |
L2 | 1 | 2 | 2 | 11.6 | 21.289 |
L3 | 1 | 3 | 3 | 11.1 | 20.906 |
L4 | 2 | 1 | 2 | 11 | 20.827 |
L5 | 2 | 2 | 3 | 9.8 | 19.824 |
L6 | 2 | 3 | 1 | 12 | 21.583 |
L7 | 3 | 1 | 3 | 9 | 19.084 |
L8 | 3 | 2 | 1 | 10.2 | 20.172 |
L9 | 3 | 3 | 2 | 10.9 | 20.748 |
Table 8
Analysis of Variance table for bead width
ANOVA for bead width (means) |
Factors | Degrees of freedom | Sums of squares | Adjusted mean square | Fisher ratio | Pure sum of squares | Percentage of Contribution |
Current intensity | 2 | 2.548 | 1.274 | 31.068 | 2.466 | 35.432 |
Welding speed | 2 | 1.528 | 0.764 | 18.635 | 1.446 | 20.781 |
Nanoparticle | 2 | 2.802 | 1.401 | 34.157 | 2.720 | 39.072 |
Other/Error | 2 | 0.081 | 0.040 | - | - | 4.715 |
Total: | 8 | 6.962 | | | | 100% |
ANOVA for bead width (S/N ratio) |
Current intensity | 2 | 1.771 | 0.885 | 54.814 | 1.739 | 35.944 |
Welding speed | 2 | 1.049 | 0.524 | 32.476 | 1.017 | 21.024 |
Nanoparticle | 2 | 1.985 | 0.992 | 61.424 | 1.952 | 40.359 |
Other/Error | 2 | 0.032 | 0.016 | - | - | 2.673 |
Total: | 8 | 4.838 | | | | 100% |
4.2 Prediction of weld geometry
The S/N characteristics can be divided into three stages: nominal-the-better, smaller-the-better, and higher-the-better when the quality characteristics are continuous for engineering analysis. The higher-the-better and the smaller-the-better quality characteristics are employed in the welding process since this study aims to maximize the penetration depth and minimize the bead height and width through optimum process parameters. Therefore, in this research, the ideal weld was considered a weld with high penetration depth, low bead height, and width. Thus, in this regard, according to the results of testing the samples based on the test design matrix, as well as the results obtained from the statistical analyses of ANOVA and S/N, taking into account the objective function Bigger is better, to achieve the highest penetration depth, the input parameters of current intensity at level 1 with 240 amps, welding speed at level 3 with 36 cm.min− 1 and nanoparticle amount be placed at level 1 with 0.25 g (Fig. 2). Therefore, for the penetration depth, we can write:
Penetration depth = C\(\stackrel{-}{\text{S}}\)1+T\(\stackrel{-}{\text{S}}\)3+N\(\stackrel{-}{\text{S}}\)1-2\(\stackrel{-}{\text{T}}\) (1)
where \(\stackrel{-}{\text{T}}\) is the overall mean of penetration depth, 8.077 mm (Table 3); C\(\stackrel{-}{\text{S}}\)1 is the average penetration depth at first level of current intensity, 240 amp; T\(\stackrel{-}{\text{S}}\)3 is the average penetration depth at third level of welding speed, 36 cm.min− 1; N\(\stackrel{-}{\text{S}}\)1 is the average penetration depth at first level of nanoparticle, 0.25 g. Substituting the values of various terms in Eq. (1), then:
Penetration depth = 8.30 + 8.366 + 8.699-(2×8.077) = 9.211 mm
Also, considering the smaller is better function as the objective function to achieve the lowest bead height, the current intensity should be at level 1, the welding speed at level 3, and the nanoparticle amount at level 1 (Fig. 3), so we have:
Bead height = C\(\stackrel{-}{\text{S}}\)1+T\(\stackrel{-}{\text{S}}\)3+N\(\stackrel{-}{\text{S}}\)1-2\(\stackrel{-}{\text{T}}\) (2)
where \(\stackrel{-}{\text{T}}\) is the overall mean of bead height, 1.961 mm (Table 5); C\(\stackrel{-}{\text{S}}\)1 is the average bead height at first level of current welding, 240 amp; T\(\stackrel{-}{\text{S}}\)3 is the average bead height at third level of welding speed, 36 cm.min− 1; N\(\stackrel{-}{\text{S}}\)1 is the average bead height at first level of nanoparticle amount, 0.25 g. Substituting the values of various terms in Eq. (2), then
Bead height = 1.733 + 1.816 + 1.633-(2×1.961) = 1.26 mm
On the other hand, considering the smaller is better function as the objective function to achieve the smallest bead width, the current intensity in level 3 is 280 amps, the welding speed in level 1 is 32 cm.min− 1, and the nanoparticle amount in level 3 with a value of 0.75 g should be placed (Fig. 4), so we have:
Bead width = C\(\stackrel{-}{\text{S}}\)1+T\(\stackrel{-}{\text{S}}\)3+N\(\stackrel{-}{\text{S}}\)1-2\(\stackrel{-}{\text{T}}\) (3)
where \(\stackrel{-}{\text{T}}\) is the overall mean of bead width, 10.755 mm (Table 7); C\(\stackrel{-}{\text{S}}\)1 is the average bead width at third level of current intensity, 280 amp; T\(\stackrel{-}{\text{S}}\)3 is the average bead width at first level of welding speed, 32 cm.min− 1; N\(\stackrel{-}{\text{S}}\)1 is the average bead width at third level of nanoparticle amount, 0.75 g. Substituting the values of various terms in Eq. (3), then
Bead width = 10.033 + 10.40 + 9.966-(2×10.755) = 8.889 mm
Therefore, according to the obtained results, to achieve the highest penetration depth and have a weld with the lowest bead height, the current intensity should be at level 1 with a value of 240 amps, the welding speed should be at level 3 with a value of 36 cm.min− 1 and nanoparticle amount should also be put in level 1 with the amount of 0.25 g. On the other hand, considering the importance of achieving a higher penetration depth in addition to having a low bead height, in the investigation of the bead width variable, the current intensity at level 1, the speed Welding was used in level 3, and the nanoparticle amount was also used in level 1. The predicted sample (L00) was welded by setting the parameters of the current intensity, welding speed, and the nanoparticle amount. After the initial confirmation of the correctness of the weld in terms of possible defects, the variables of penetration depth, bead height, and bead width of the weld were measured. The investigations conducted on the weld sample L00 showed that this sample had a weld with a high penetration depth and a wide and low bead height in terms of weld bead appearance. Figures 5 and 6 show the cut sections of the weld metal's surface and the weld metal's surface. The results show that the bead height and width in the L00 sample had 1.30 mm and 12 mm values, respectively. In comparison, the predicted values for these variables were 1.260 mm and 8.889 mm, respectively. Also, according to the predicted penetration depth value, which was equal to 9.211 mm, the weld penetration depth in the L00 sample was 9.22 mm. Therefore, the measured penetration depth and bead height were very clearly similar to the predicted values. Further examination shows that the weld geometry in the L00 sample, which has 0.25 g of nanoparticles, compared to the welded sample without nanoparticles shown in Figs. 7 and 8, has a broader width but a lower bead height. In addition, in L00, the penetration depth increased significantly. Also, the shape of the arc during the welding operation in the L0 sample (Reference sample) was almost similar to a circle. In contrast, the L00 sample has a concentrated and compressed arc. According to previous studies, the two factors of electric arc concentration and reversal of the Marangoni flow inside the molten pool due to the presence of surface-active elements such as oxygen [33, 34] can be among the main reasons for the increase in the penetration depth in the L00 sample.