Optimization of Dissimilar aluminum alloy by Friction stir welding Process Control variables with Multiple Objectives

The progress of FSW has offered a different approach for fabricating superior class weld. This paper deals with optimization of process control variables influencing weld features in customized friction stir butt welding having multiple objectives of 6 mm thick dissimilar plates of AA7075 and AA6101 using Taguchi grey relational approach. The L27 orthogonal array has been employed to Obtain the experimental trails and the joints have been fabricated in a laboratory stage friction stir welding equipment by changing tool rotation speed, worktable translational speed, tool plunge force and tool pin shape. After welding, tensile and impact strength of the weld have been evaluated and found that optimum ultimate tensile strength of 129 Mpa., maximum yield strength of 119.21Mpa, maximum % of elongation 24.64% and impact strength 23Mpa simultaneously in FSW process variables of Tool rotation speed is 1200 rpm, transvers speed of 30 mm/min and axial force is at 5 kN. Based on the experimental results optimum levels of process control variables were obtained using grey relation rating and compared with confirmation test.


Introduction
In technology growth, the scientists and technologists are handling extremely challenging problems in the area of metal joining technology. The problem with the conventional joining methods mainly be attributed to novel metals with low welding strength. Scientists in the field of material science are novel frontier metals with a high strength, hardness, toughness and other diverse behaviors. The welding of metals with these properties is more difficult and affect their basic properties by conventional welding methods. In last two decades, an innovative solid-state joining method generally called as FSW was invented and patented by The Welding Institute (TWI) in United Kingdom in the year 1991 [1] and its need was addressed in the first paragraph as FSW was environmentally friendly and accessible to materials with a high strength to weight ratio as well as with no welding defects such as hot cracks and porosity [2,3]. Genetic programming (GP) is a comparatively new approach to advanced computation, with the key benefit of this method being the estimation of efficient predictive mathematical models or equations without any assumption as to the potential type of functional relationship [4]. Owing to low precipitate dispersal, and/or rather than grain size in the weld, the hardness decreased with increased tool traverse speed. The R2 values for the projected model of all the properties were obtained nearly 90%, it showed a good committment between the independent variables and the response data [5]. Many combinations of process control variables were formed using L18 orthogonal array. By using Principal Component Analysis (PCA) it was noticed that weightage of 45.36 %, 44.51% and 10.11% hardness tensile strength and power consumptions respectively. Optimal process parameters were obtained using Multi-Objective Ratio Analysis (MOORA) optimization, which results in the increasing of tensile strength and hardness lower the power consumption [6]. The difference between the higher and lower value of the gray relation grade of the variables of the FSW operation is as follows: 0.2756 for tool rotational speed, 0.14171 for welding speed and 0.08436 for axial force. By comparing these values, the most active variable influencing quality attributes is ascertained. The correlation will provide the degree of importance over the various quality characteristics of the input process control variables. Here, the maximum value of 0.2756 indicates that the tool rotational speed has the biggest impact among the other process control variables on the quality characteristics [7]. The plot graphically assesses the impact of each input welding parameter on the efficiency of the weld. On the Basis of main impact plot, input parameters are found to be important as they reach their center point, while their lower and higher levels have not affected the consistency of the FSW specimens significantly. But the tilt angle of the FSW tool is considered important when it is between the middle and the higher level [8]. The interactions between the rotational speed of the tool and the shoulder base angle closely, it becomes clear that the impact of the shoulder base angle on the UTS depends on the tool rotational speed. If it rotates at 1250 rpm, the strength is minimized at 5°, however, at 1000 rpm the shoulder base angle yields the maximum strength at 5°. The shoulder base angle and rotational speed have a more significant impact on the surface hardness compared to other interactions [9]. Among the numerous common evolutionary algorithms, the Non-Dominated Sorting Genetic Algorithm-II (NSGA-II) is widely adopted as an effective method for improving product quality in all manufacturing activities such as machining, shaping, and welding. NSGA -II uses random genetic operations to scan a whole design space for global optimum operation through different design points [10]. The mean absolute percentage error (MAPE) is applied to conduct error analysis of the implemented decision-making processes used in this report. To do this, 30 separate computer code runs were performed, and a final solution was obtained for each trial using the Shannon entropy and decision-making methods of TOPSIS. The value of each target (ultimate tensile power, elongation, and minimum hardness in the HAZ zone) was then equated with the better solutions obtained by each method after 30 runs [11]. The optimal setting of process parameter represents the relationship between the reference sequence and objective sequence, therefore greater fuzzy grey relational grade reveals the objective sequence has a stronger relationship than the reference sequence. Accordingly, the optimal setting of process parameters is larger fuzzy -GRG is desirable for obtaining larger 3 UTS and TE of fabricated FSW specimen. The FSW joints made from square pin had the strongest Fuzzy GRG. Since flat-faced Pin profiles are correlated with eccentric material flow. This eccentricity of material flow allows for the movement of an incompressible material flow across the pin profile. The interaction between the static volume and the dynamic volume determines the direction from the leading edge to the trailing edge of the revolving tool for the distribution of fleshy material [12]. It is observed that increase in tool rotational speed increases frictional heat generation and lower heat input condition prevails at lower tool rotational speed like 700 rpm which are also associated with lack of stirring. The net result is poor due to consolidation of materials and it lead to poor wear resistance at lower tool rotational speeds. Higher tool rotational speed like 1200 rpm led to higher heat generation than required and release excessive stirred materials. Micro voids appear at higher tool rotational speed led to poor wear resistance [13]. It is evident from the graphical data that mean values of S/N ratio increase with increase in 'Tool rotational speed'. It means that the weld joint efficiency increases with increase in 'Tool rotational speed [14]. Alloys of Aluminum and magnesium are majorly finding their applications in industries like automobile, aerospace and sports applications. The joining of these dissimilar alloys is very difficult by using traditional fusion welding technique inherently because of the "brittle Inter Metallic Compounds (IMC)" like Mg2Al3 and Mg17Al12. In order to enhance the joint strength, the formation of inter metallic compounds due to friction stir welding has to be characterized first. In the current paper some important welding parameters and their effects on weld quality are discussed along with the mechanical properties of the weld joint" and to identify the optimized process variables by multi-objective optimization techniques.

Methodology
The materials utilized for this examination are aluminum amalgams AA6101 and AA7075. With the help of a power hacksaw machine the rolled plates of 6mm in thickness were sliced into the essential size (100mm x 50 mm x 6 mm) and squaring the butting faces with the help of the milling process. Before FS welding process, butting edges of the weld specimens were cleaned by using a wire brush. Edges to be welded were arranged with the goals that are completely parallel to one another. This ensure that there is no uneven hole between the plates that may not give good properties to welded joints. In addition, surface arrangement was also performed in such a way that the surfaces of both plates are of same size and balance. The compound structure in terms of weight rate tabulated in Table 1 and Mechanical properties of the base metals employed in this investigation at atmospheric condition is recorded in Table.2.

Design of Experiments
Design of Experiments (DOE) is a formal, coordinated method for determining the relationship between factors that influence a process and performance of the process. Three operation control variables of FS welding are considered factors of control are tool TRS, WS, and AF. Each variable has 3 different pitches like high, medium and low represented by 3, 2 and 1 respectively. Due to the wide range of influential factors, it was decided to use three  Table 3.

Experimentation
The Friction Stir Welding (FSW) Machine setup is shown in Figure 1. In this work, Butt welding of AA7075 and AA6101 dissimilar alloy materials is carried out at different process parameters. For welding of 9 samples of AA7075 and AA6101, all specimens were prepared to a size of 100*50*6 mm. All specimens were positioned and firmly clamped with help of backing plates to avert separation of the attached butting edges. The forces are fairly large during the tool's initial plunge and additional alerts were needed to verify the plates were not separated in a butt arrangement. The tool mounted in tool holder with tilt angle of 1.5° and the tool pin was throwing to a predestined deepness at the edges of the butting surface of the plates to be joined. The tool was transversed forward after residing time at the end of which the joint was formed by a single pass. After the weld is finished, the tool is released from work piece and allowed to get cooled.

Process Response Measurement
The tensile behavior of FS welded joints were determined using the UTM (Make: FIE & Model: UTN 40). The trail samples were sliced from the fabricated joints and according to ASTM E8 dimension machined as in Figure 2. Three identical specimens were tested to acquire the average tensile strength. The camera image of the FS welded specimens after tensile fracture is shown in Figure 3. The impact toughness was measured using pendulum type impact test machine (Make: FIE & Model: IT 30 ASTM). The three specimens were extracted across the weld line from friction stir welded joints and machined as per ASTM E23 standard size displayed in Figure 4. FS welded joint specimens fractured realistic images after charpy test are exposed in Figure 5. Table 4

Grey-Taguchi Technique
In this work, Grey-Taguchi technique was employed for finding better combination of process control variables to join AA7075 and AA6101 alloy materials by friction stir welding method and it is one of the best practices for multi objective optimization problems. Generally, Taguchi method is supportive for planning of experiments and finding of optimal setting individually for each output response, but in the present research there are different output responses for tensile properties and impact strength. There is necessity to find out the supreme combination of process control variables for all the output responses simultaneously. The step-by-step procedure in Grey-Taguchi technique is shown in Figure 6.

Normalization of Experimental Results
In the Grey-Taguchi test the initial step is to normalize the experimental results of tensile properties and impact strength. out once the signal-to-noise ratio of the required quality criteria is obtained as shown in Table 10. This process conveys the original sequence to a comparable sequence and the experimental results are normalized in the range between zero to one due to different measurement unit [15]. For normalizing tensile properties (ultimate tensile strength, yield strength, % of elongation) and impact strength 'Higher-thebetter' (Equ. 1) criterion is used. The normalized data for tensile properties and impact strength is given in Table 5. Where, Xj (k) = value after normalizing data/Grey relational generation value, min yj(k) = smallest value of yj(k) for kth response, max yj(k) = largest value of yj(k) for kth response.

Grey Relational Coefficient
Once calculate normalized tensile properties like ultimate tensile strength, yield strength, % of elongation and impact strength, the next step is estimation of grey relational coefficient values for tensile properties and impact strength. The grey relational coefficient j (k) can be estimated by using Equ. (2). the grey relational coefficient values for tensile properties and impact strength are given in Table 6. (2) Where, Δ ( ) = Xoj(v) -X j (v), Δmax = larger value of Δoj, Δmin = smaller value of Δoj, Xj (v) = value after normalizing data/Grey relational generation value, Xoj(v) = Ideal value = 1 and in general assumed φ=0.5.

Grey Relational Grade (GRG) and Order
Grey relational grade (GRG) for each investigational run is the normal of grey relational coefficient value for a particular investigational run. GRG can be calculated by using Equ. (3). Larger value of grey relational grade specifies the top value, so highest grade value provides the higher order. The GRG and their position are given in Table 7.

Multi Objective Optimization
In order to investigate the significant control variables on tensile properties and impact strength. ANOVA was performed. ANOVA table for GRG are shown in Table 8., it is shows that tool revolving speed has utmost dominating process control variables which is about 70.88% influence on grey relational grade and succeeding with transverse speed and axial force has effect on grey relational grade with contribution of 10.63% and 4.93%. The interaction effects of TRS x AF, TRS x WS, AF x WS on grey relational grade with contribution of 2.80, 4.70, 2.76 % respectively. Table 9 & Table 10 shows the response table for grey relational grade which gives the average of each process responses (Means and S/N ratio) for each level at each response. The ranks and delta values show that rotational speed have high effect on grey relational grade as well as it is followed by transverse speed and axial force in that order.  As grey relational grade 'higher the better' type response, it can be seen from Figure 7, that the third level of rotational speed, third level of force and first level of transvers speed offers extreme value of grey relational grade. Hence, Tool rotational speed is 1100 rpm, force is at 4 KN and transverse speed of 30 mm/min is the finest combination of process control variables for obtaining utmost tensile properties and maximum impact strength simultaneously in FSW process. Figure 8 also suggest the same combination of process control variables.

Anticipated optimum condition
Based on trials, the optimum level setting was found at tool rotational speed of 1100 rpm table transverse speed of 30 mm/min and axial force of 4 kN. So, the anticipated grey relation grade can be determined as:  (4) Where is the total mean GRG, is the mean GRG at the optimum level, and n is the number of process control variables that affect the quality characteristics.
The validation experiment is not necessary here, as the optimized experiment at factor level now available within the planned trialing. The actual grade of gray relation at optimal condition is 0.9735; while the grade of gray relation expected is 0.8067. So the gap is just 0.16 (approx.) and it arises due to ignoring the nonlinear effects in three factor three level Taguchi L27 orthogonal array.

Conclusions
In this investigation, to find the optimum combination of friction stir welding process control variables to join AA7075 & AA6101 alloy materials, Taguchi based grey analysis was applied. The important conclusions from the present research work are summarized as follows:  The finest combination of process control variables for obtaining optimum ultimate tensile strength of 129 Mpa., maximum yield strength of 119.21Mpa, maximum % of elongation 24.64% and impact strength 23Mpa simultaneously in FSW process is found at Tool revolving speed is 1200 rpm, transverse speed of 30 mm/min and axial force is at 5 kN.
 ANOVA results shows that process control variables too revolving speed shows major effect on the output responses in FSW process and axial force shows less influence on the output responses.
 The current Gray Relationship Grade at the optimum condition is 0.9735, while the predicted Gray Relationship Grade is 0.8067. Thus, the difference is only 0.16 (approx.). This variation occurs due to the neglect of the nonlinear effects in the orthogonal 3-level Taguchi L27 array.