Key Engineering Materials Submitted: 2019-05-05

ISSN: 1662-9795, Vol. 833, pp 35-39 Accepted: 2019-05-13

doi:10.4028/www.scientific.net/KEM.833.35 Online: 2020-03-06
© 2020 Trans Tech Publications Ltd, Switzerland

Multi-Response Optimization of Dissimilar Al-Ti Alloy FSW Using

Taguchi - Grey Relational Analysis
Shival Patel1, Kishan Fuse2, Khushboo Gangvekar3, Vishvesh Badheka4
Pandit Deendayal Petroleum University, Gandhinagar, India
[email protected], [email protected], [email protected],
[email protected]

Keywords: Dissimilar friction stir welding, grey relational analysis, Taguchi, ANOVA

Abstract. This article presents multi-response optimization of friction stir welding of dissimilar Al
6061- Titanium alloy using Taguchi based grey relational analysis. Taguchi’s L9 orthogonal array
was used for designing the experiments. Process parameters considered for the experiments were
rotational speed, traverse speed and tilt angle. Ultimate tensile strength, yield strength, and %
elongation were the responses measured which all are larger-the-better characteristics. Based on grey
relational grade, optimum levels of process parameters were identified and further ANOVA analysis
was carried out to find most significant process parameter.

Introduction
Friction stir welding (FSW) is an advanced welding process invented at The Welding Institute,
London, UK, in 1991. This technique is having numerous advantages over conventional fusion
welding such as no arc, no fumes, no requirement of shielding gases or filler wires [1]. Joining of
dissimilar material is highly difficult with conventional fusion welding due to differences in
performances of material, melting point, heat conductivity, and coefficient of thermal expansion.
Friction stir welding is potential candidate solid state welding process to overcome these problems
[2]. There are several studies on dissimilar FSW of aluminium alloys such as AA7050-AA6061 [3],
AA6061-AA7075[4], AA7020-T651 and AA6060-T6[5]. Apart from dissimilar aluminium alloy
joining, FSW is successfully adopted for joining AA5052-AISI1030 steel [6], aluminium-titanium
[7-9], aluminium-brass[10], Al-Mg [11]. Multi-objective optimization of process parameters is one
of the requirements of the industries. Grey relational analysis (GRA) is widely used technique for
multi-response optimization. Joshi et al. [12] used grey relational analysis for optimizing EDM
process parameters. Kao et al. [13] also used GRA for EDM parameter optimization of Ti-6Al-4V
alloy. Kundu and Singh [14] performed optimization of FSW process parameters for ultimate tensile
strength, % elongation, and micro-hardness. Parida and Pal [15] used this technique for optimizing
the eight input process parameters during FSW of aluminium alloy. Also, Grey relational technique is
successfully used in turning [16], wire-drawing [17] operations. Literature review suggested that
very less attention is given to multi-response optimization of FSW process parameters for dissimilar
joining of Al-Ti alloy. In the present study, grey relational analysis is used for optimizing input
process parameters during dissimilar friction stir welding of Al6061-T6 & Ti alloy FSW.

Experimental Details
Al6061-Ti plates with thickness of 4 mm were utilized to prepare FSW butt joint. In the present
study, plates are machined with dimension 120 mm long and 50 mm wide. The surfaces of welding
samples were properly cleaned. The experiments were conducted on vertical milling machine. A
cylindrical taper tool was manufactured using tungsten carbide (WC) as tool material. Schematic of
tool is as shown in Fig. 1. The shoulder diameter of tool was 18mm, pin base diameter 6 mm, pin tip
diameter 4 mm and pin length was 3.8mm. Titanium plates were positioned on advancing side (AS)
and Al6061 was set on retreating side (RS). The tool offset was given on retreating side (RS) as 2mm.

All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of Trans
Tech Publications Ltd, www.scientific.net. (#541192425, University of Melbourne, Melbourne, Australia-25/07/20,22:57:30)
36 Material and Manufacturing Technology X

Guage Length
Fig. 1 Schematic of FSW tool Fig. 2 Sampled tensile specimens

Based on literature study[2], three process parameters as rotational speed, traverse speed, and tool
tilt angle were selected. The process parameters with their levels are given in Table 1.
Table 1 Process parameters and levels
A B C
Level Rotational Traverse Speed Tilt angle
speed (rpm) (mm/min) (degree)
1 765 20 1
2 1070 31.5 2
3 1500 50 3
Taguchi’s L9 orthogonal array [14] was used for systematically designing the experiments which is
shown in Table 2.
Table 2 Experimental layout – L9 orthogonal array and measured responses of UTS, YS, %E
A B C
Expt. Rotational Traverse Tilt UTS YS %E
No. speed Speed angle (MPa) (MPa) (%)
(rpm) (mm/min) degree)
1 1 1 1 142.97 59.48 9.82
2 1 2 2 31.27 20.84 4.32
3 1 3 3 94.51 56.71 5.27
4 2 1 2 103.98 65.13 7.80
5 2 2 3 18.91 14.21 5.78
6 2 3 1 60.94 45.46 6.37
7 3 1 3 68.70 54.51 5.23
8 3 2 1 61.80 38.67 4.11
9 3 3 2 35.84 25.68 6.01

The ultimate tensile strength, yield strength, and % elongation are most important output measures
which determine mechanical properties of butt welded joint are considered as response variables. The
wire cut EDM is used to cut tensile specimens as per ASTM–E8. Three coupons are cut from each
specimen. Tensile samples of such one specimen is as shown in Fig. 2. Tensile testing was carried out
at cross head speed of 1mm/min. The experimental layout of L9 orthogonal array for the process
parameters & tensile properties measured in terms of selected response variables is shown in Table 2.
 Key Engineering Materials Vol. 833 37

Methodology
In grey relational analysis, the steps to optimise the quality characteristics are as follows:
Calculating and normalizing of S/N ratio. The first step is to calculate S/N ratio for the
measured response characteristics. Response characteristics are of three types - a) smaller-the-better
b) larger-the-better or c) nominal-the-best. Irrespective of response category, large S/N ratio
corroesponds to better process response. But, this straightforward conclusion is true only for single
response consideration. When response considered during analysis are two or more, higher S/N ratio
of one response may corroespond to lower S/N ratio for other response. Further, to avoid the problem
of different scales and units, response data are normalized in the range between zero and one. In the
present study, all three responses that is ultimate tensile strength, yield strength, and % elongation is
higher-the-better type. Linear normalization of such characteristics can be done using equation (1).
∗ 𝜂𝜂1𝑖𝑖 −𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚1𝑖𝑖
𝜂𝜂1𝑖𝑖 = , i=1,2,…,9 (1)
𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚1𝑖𝑖 −𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚1𝑖𝑖
∗
Where 𝜂𝜂1𝑖𝑖 is normalized S/N ratio.
Calculate grey-relational coefficient. It is calculated to express the relationship between the
ideal (best) and the actual experimental data. The grey relational coefficient can be calculated as
follows:
𝛥𝛥𝑚𝑚𝑚𝑚𝑚𝑚 +𝜁𝜁𝛥𝛥𝑚𝑚𝑚𝑚𝑚𝑚
𝜉𝜉𝑖𝑖 (𝑘𝑘) = (2)
𝛥𝛥0𝑖𝑖 (𝑘𝑘)+𝜁𝜁𝛥𝛥𝑚𝑚𝑚𝑚𝑚𝑚
Where,
𝛥𝛥0𝑖𝑖 (𝑘𝑘)= deviation
𝜁𝜁= distinguishing coefficient, range {0, 1}. It helps to make better distinction between the reference
sequence and the comparability sequence. 𝜁𝜁= 0.5 is mostly used as it offers moderate distinguishing
effect between the sequences
Generate grey relational grade. Grey relational grade is computed by averaging the grey
relational coefficients corresponding to each performance characteristic. The experimental data of the
multi-response characteristics is evaluated by using this grey relational grade. The optimum level of
the process parameters is the level with the highest grey relational grade. The grey relational grade is
obtained as:
1
𝛾𝛾𝑖𝑖 = ∑𝑛𝑛𝑘𝑘=1 𝜉𝜉𝑙𝑙̇ (𝑘𝑘) (3)
𝑛𝑛
𝛾𝛾𝑖𝑖 = Grey relational grade, correlation between the reference sequence and the comparability
sequence.

Result and Discussion

The calculated S/N ratio, grey-relational coefficient and grey relational grade using equation (1),
(2), and (3) are as shown in Table 3. The grey relational grade is the deciding parameter for optimized
process parameters. The highest grey relational grade indicates the optimized combination of the
parameters. Table 3 indicated that first experiment is the optimized experiment as its grey relational
grade is highest that is 0.30. This study revealed that, Rotational speed at level 1 (750 rpm), feed rate
at level 1 (20 mm/min) & tool tilt angle at level 1 (10) is the optimized setting. Figure 3 shows mean
effect plot of means of grey relational grade which also represents same optimized results as A1B1C1.
Further, analysis of variance (ANOVA) is used to check most significant process parameter. The
results of ANOVA analysis using Stepwise method is as shown in Table 4. P-value from the table 4
shows that feed rate is having P-value as 0.009 which is less than 0.05 (considering 95% confidence).
This indicates to reject the null hypothesis (H0) corresponding to feed rate which can be stated as
there is no significant difference between feed rate on response. Rejecting null hypothesis for feed
rate suggest that feed rate is the most significant process parameter affecting multiple response
characteristics.
38 Material and Manufacturing Technology X

Table 3 Result of S/N ratio, grey relational coefficient, grey relational grade and rank
Grey relational Grey
Exp S/N ratio
coefficient relational Rank
No.
UTS YS %E UTS YS %E grade
1 43.10 35.49 19.76 1.00 1.00 1.00 0.30 1
2 28.57 25.78 10.01 0.47 0.51 0.33 0.13 8
3 35.51 32.02 13.87 0.63 0.75 0.45 0.18 5
4 40.02 34.64 15.28 0.81 0.92 0.52 0.23 2
5 17.23 15.08 15.11 0.33 0.33 0.51 0.12 9
6 35.40 32.84 16.00 0.63 0.79 0.57 0.20 3
7 36.39 34.34 12.74 0.66 0.90 0.41 0.20 4
8 35.43 31.47 12.16 0.63 0.72 0.39 0.17 6
9 29.72 27.16 15.56 0.49 0.55 0.54 0.16 7
Table 4 ANOVA results of GRG
Source DF Adj SS Adj MS F-Value P-Value
B 2 0.016289 0.008144 19.04 0.009
C 2 0.005756 0.002878 6.73 0.053
Error 4 0.001711 0.000428
Total 8 0.023756

Main Effects Plot for Means of GRG

Data Means
A B C
0.250

0.225
Mean of Means

0.200

0.175

0.150

1 2 3 1 2 3 1 2 3

Fig. 3 Main effects plot for means of GRG

Conclusion
This article represents use of Taguchi based grey relational analysis for optimizing multi-response
characteristics. The friction stir welding of dissimilar Al6061-Titanium alloy was carried out using L9
orthogonal array considering input process parameters as rotational speed, traverse speed and tool tilt
angle and the response variables considered were ultimate tensile strength, yield strength and %
Elongation. The following conclusion has drawn from the study:
1. Higher grey relational grade implies better quality characteristic. On the basis of which the
optimum level was attained at A1B1C1 i.e. rotational speed of 750 rpm, feed rate 20 mm/min and
tool tilt angle 1°.
2. ANOVA results considering grey relational grade as factor indicated that feed rate is the most
significant process parameter affecting multiple response characteristics.
 Key Engineering Materials Vol. 833 39

References
[1] K. Fuse and V. Badheka, Bobbin tool friction stir welding: a review. Science and Technology
of Welding and Joining, 2018: p. 1-28.
[2] Y. Chen, C. Liu, G. Liu, Study on the joining of titanium and aluminum dissimilar alloys by
friction stir welding. The Open Materials Science Journal , 2011. 5: p. 256-261.
[3] R.I. Rodriguez, et al., Corrosion effects on fatigue behavior of dissimilar friction stir welding of
high-strength aluminum alloys. Materials Science and Engineering: A, 2019. 742: p. 255-268.
[4] M. Raturi, A. Garg, and A. Bhattacharya, Joint strength and failure studies of dissimilar
AA6061-AA7075 friction stir welds: Effects of tool pin, process parameters and preheating.
Engineering Failure Analysis, 2019. 96: p. 570-588.
[5] L. Giraud, et al., Investigation into the dissimilar friction stir welding of AA7020-T651 and
AA6060-T6. Journal of Materials Processing Technology, 2016. 235: p. 220-230.
[6] Abhinand, Numerical Simulation of Friction Stir Welding for Dissimilar Metal Welding.
Materials Today: Proceedings, 2017. 4(10): p. 11265-11269.
[7] J.-W. Choi, H. Liu, and H. Fujii, Dissimilar friction stir welding of pure Ti and pure Al.
Materials Science and Engineering: A, 2018. 730: p. 168-176.
[8] A. Kar, et al., Effect of niobium interlayer in dissimilar friction stir welding of aluminum to
titanium. Materials Characterization, 2018. 145: p. 402-412.
[9] A. Kar, S. Suwas, and S.V. Kailas, Two-pass friction stir welding of aluminum alloy to titanium
alloy: A simultaneous improvement in mechanical properties. Materials Science and
Engineering: A, 2018. 733: p. 199-210.
[10] A. Esmaeili, M.K.B. Givi, and H.R.Z. Rajani, Investigation of weld defects in dissimilar
friction stir welding of aluminium to brass by radiography. Science and Technology of Welding
and Joining, 2013. 17(7): p. 539-543.
[11] G. Buffa, et al., Friction stir welding of dissimilar aluminium–magnesium joints: sheet mutual
position effects. Science and Technology of Welding and Joining, 2015. 20(4): p. 271-279.
[12] R. Joshi, et al., Multi-Response Optimization of EDM for Ti-6Al-4V Using Taguchi - Grey
Relational Analysis. Solid State Phenomena, 2017. 266: p. 43-50.
[13] J.Y. Kao, et al., Optimization of the EDM parameters on machining Ti–6Al–4V with multiple
quality characteristics. The International Journal of Advanced Manufacturing Technology,
2009. 47(1-4): p. 395-402.
[14] J. Kundu, and H. Singh, Friction stir welding: multi-response optimization using
Taguchi-based GRA. Production & Manufacturing Research, 2016. 4(1): p. 228-241.
[15] B. Parida, and S. Pal, Fuzzy assisted grey Taguchi approach for optimization of multiple weld
quality properties in friction stir welding process. Science and Technology of Welding and
Joining, 2014. 20(1): p. 35-41.
[16] C. L. Lin, Use of the Taguchi Method and Grey Relational Analysis to Optimize Turning
Operations with Multiple Performance Characteristics. Materials and Manufacturing
Processes, 2004. 19(2): p. 209-220.
[17] Optimizing SUS 304 wire drawing process by grey relational analysis utilizing Taguchi
method. Journal of University of Science and Technology Beijing, 2008. 16(6): p. 714