### 1. Introduction

^{1)}. AA2014 is commonly used in heavy duty forgings, plates and sheets, aircraft components, wheels, major structural components, space booster tanks and heavy duty road and rail vehicles

^{2)}. It is widely used in the structures that require high strength to weight ratio, high fracture toughness, good fatigue resistance and durability.

^{3}

^{-}

^{5)}. The FSW process is a highly consistent joining process with high reproducibility and is effectively used in several high-tech industries such as automotive, railway, aerospace and aeronautics. Prominent features of this technique are absence of segregation, porosity, hot cracking, high temperature phase transformations and results in low distortion, low residual stresses, good surface finish and improved mechanical properties. In addition, this method yields better joint and energy efficiency

^{5}

^{-}

^{7)}.

^{8}

^{-}

^{10)}. Friction stir welding involves various process parameters, out of which the significant are tool rotation speed, tool travel speed and tool tilt angle. Weld quality and mechanical characteristics are highly influenced by these parameters. A slight maladjustment in these parameters can result in inappropriate softening of the material and causes defects in the weld. If the material softening is less than a critical value, defects like tunnel, kissing bond, root cracks, etc. may occur. Excessive flash, voids, surface cracks like defects may be observed when there is excessive softening of the material

^{11}

^{,}

^{12)}. Therefore, it is of fundamental importance to properly optimize the process parameters. In general, optimization is carried out by considering a set of pre-defined process parameters, such as tool rotational speed, tool linear travel speed, tool tilt angle, axial force, tool shoulder to pin diameter ratio, etc., keeping in mind some intended objective parameters such as maximum temperature generated, yield strength, ultimate tensile strength, hardness, impact strength, etc.

^{6}

^{,}

^{13}

^{,}

^{14)}. Most of the engineering processes have a large number of factors that influence the final product. Identification of their individual contributions and their intricate interrelationship is necessary to design a process

^{15)}. Classical experimental approach is generally employed for this purpose. This approach is not only time consuming, expensive but also inaccurate because it depends on trial-and-error based analyses. Now-a-days more robust statistical methods are being utilized for process optimizations. These approaches not only save time and cost but also give near-optimal results. Researchers have extensively used various statistical approaches for the optimization of performance characteristics in many engineering analyses. Various techniques have been used for optimization of Friction Stir Welding parameters including Response Surface Methodology (RSM), Taguchi, Full Factorial (FF), Genetic Algorithm (GA), Multiple Regression Analysis (MRA), Analysis of Variance (ANOVA), Regression Analysis (RA), Artificial Neural Networks (ANN), Finite Element Analysis (FEA)

^{6}

^{,}

^{13}

^{,}

^{16)}. Many of these Design of Experiment (DoE) practices need large number of experimentation. In many industrial problems, carrying out a large number of experiments is not feasible. In this respect, Taguchi orthogonal array approach has gained much popularity due to its simplicity, efficiency, low-cost and excellent quality. As an example, the Full Factorial process will need a total of 81 (3

^{4}

^{)}experiments for a process optimization with four factors and three levels each. With Taguchi Method, the same optimization will be accomplished with only 9 tests by utilizing L9 (3

^{2}

^{)}orthogonal array approach

^{17)}. It has been observed that, among AA2xxx series, the research works mainly focuses on friction stir welding of AA2024 and AA2219. The advent of Friction Stir Welding has opened the door for structural applications of AA2014 on large scale. Currently, most of the reported data regarding optimization of friction stir welding of AA2014 is limited to thick plates and the work on thin sheets is lacking.

### 2. Experimental Methods

### 2.1 Materials

### 2.2 Friction Stir welding

### 2.3 Selection of parameters and levels

### 2.4 Orthogonal array experiment

### 2.5 Tensile Testing

^{18)}. The uniaxial tensile testing was performed on Universal Testing Machine, with a load cell of 100 kN and high resolution extensometers. All the tensile tests were performed at cross head speed of 0.5mm/min. The testing was performed at room temperature.

### 3. Results and Discussion

### 3.1 Tensile Test Results of Taguchi Matrix

### 3.2 Signal to Noise Ratio (S/N)

##### Table 7

Experiment No. | Ultimate tensile strength, MPa | S/N ratio | ||||||
---|---|---|---|---|---|---|---|---|

1 | 337 | 50.58 | ||||||

2 | 332 | 50.77 | ||||||

3 | 338 | 50.32 | ||||||

4 | 382 | 51.68 | ||||||

5 | 399 | 52.01 | ||||||

6 | 383 | 51.45 | ||||||

7 | 394 | 51.94 | ||||||

8 | 408 | 52.19 | ||||||

9 | 411 | 52.27 |

_{i}’ is the observed result (response) of each experiment. The calculated S/N ratio values (mean) as per equation (1) are given in Table 8 and graphically presented in Fig. 2. The maximum effectiveness ranks were calculated on the basis of delta values which represent the variance (scatter) between the highest and lowest average response values for each factor.

##### Table 8

^{19)}.

^{20)}. Since it is unreasonable to check the existence of non-linear element in the relationship between process parameters and the UTS of the weld joints, the analyses are based only on center points.

### 3.3 Interaction Plots for response analysis

^{21)}. Interaction plots in the form of 2D contours were plotted to analyze the relation of the chosen process parameters with the selected response i.e. UTS and are shown in Fig. 4. It is evident from the plots that the maximum tensile strength is achieved when ‘Tool rotation speed’ is around 950 - 1000 rpm and ‘Linear travel speed’ is about 400 to 600 mm/min.

### 3.4 Estimation of optimum parameters

_{I}), ‘Linear travel speed’ of 400 mm/min (B

_{II}) and ‘Tilt angle’ of 2° (C

_{II}). However, interaction contour plots show that the maximum properties can also be achieved at ‘Tool rotation speed’ of 960 to 1000 rpm and ‘Linear travel speed’ of 400 to 600 mm/min also.

### 3.5 Prediction of optimum parameters by Regres- sion Equation

^{21)}. Prediction in regression refers to estimating the value of one variable using assumed values of other input variables that are related to it. A regression fit equation was developed by analyzing the mean UTS against the chosen process parameters. Equation 2 describes the regression model for the present case.

### 3.6 Prediction of optimum parameters by Theoretical and Numerical modeling

^{22}

^{-}

^{26)}due to the dissolution of strengthening precipitates and grain coarsening. The estimation was done by theoretical calculations as well as by numerical modeling. For the estimation of maximum temperature attained during friction stir welding of aluminum, the expression as proposed by Arbegast et. al.

^{27)}was used and is reproduced as Equation 3.

_{m}(℃) are maximum and melting temperature, respectively, ω is ‘Tool rotation speed’, ν is ‘Linear travel speed’ and K and α are constants (taken as 0.7 and 0.05 respectively). The results obtained for maximum temperature are reported in Table 10. The results show that for combination 1000/600/2, maximum temperature attained has the minimum value.

##### Table 10

FSW Process parameters | Max. Temp. (°C) | |||
---|---|---|---|---|

Tool rotation speed, (rpm) | Linear travel speed, (mm/min) | Tilt angle, (°) | Theoretical | Simulated |

1000 | 400 | 2 | 579 | 626 |

1000 | 500 | 2 | 573 | 617 |

1000 | 600 | 2 | 568 | 608 |

### 3.7 Selection of optimum parameters

### 3.8 Validation Test

### 3.9 Analysis of Variance (ANOVA)

^{2}. This coefficient is a statistical number which shows how much the data is fitted to the regression line. The measured value of R

^{2}is 97.46% as shown in Table 10. As this value is almost equal to one, it can be concluded that the developed model has high accuracy

^{8}

^{,}

^{22)}.

##### Table 11

### 4. Conclusions

1) Taguchi Design of Experiment technique can be efficiently used to optimize the process parameters of friction stir welding process of thin sheets of aluminum.

2) S/N analysis showed that the maximum tensile strength can be attained when values of ‘Tool rotation speed’, ‘Linear travel speed’ and ‘Tilt angle’ were 1000 rpm, 400 mm/min and 2°, respectively. The attainable UTS was 409 MPa.

3) Since heat generated during FSW has great influence on the tensile strength of the joint, Interaction Contour plots and regression analysis were used to refine the optimum parameters. The results showed that the same tensile strength can be achieved by utilizing a higher ‘Linear travel speed’ of 600 mm/min.

4) The theoretical analysis and FE modeling showed that welding at higher ‘Linear travel speed’ can offer same results at lower peak temperature.

5) A maximum tensile strength of 412 MPa was achieved by the FSW joint made by the optimized parameters of 1000 rpm ‘Tool rotation speed’, 600 mm/min ‘Linear travel speed’ and 2° ‘Tilt angle’.

6) ANOVA analysis demonstrated that ‘Tool rotation speed’ has the most significant effect on the tensile strength of the joint with 92.3% contribution followed by ‘Linear travel speed’ and ‘Tilt angle’ having 4% and 1.16% contribution respectively.