Process Parameters Optimization for Friction Stir Welding of Pure Aluminium to Brass (CuZn30) using Taguchi Technique

In this research, the friction stir welding of dissimilar commercial pure aluminium and brass (CuZn30) plates was investigated and the process parameters were optimized using Taguchi L9 orthogonal array. The considered process parameters were the rotational speed, traverse speed and pin offset. The optimum setting was determined with reference to ultimate tensile strength of the joint. The predicted optimum value of ultimate tensile strength was confirmed by experimental run using optimum parameters. Analysis of variance revealed that traverse speed is the most significant factor in controlling the joint tensile strength and pin offset also plays a significant role. In this investigation, the optimum tensile strength is 50% of aluminium base metal. Metallographic examination revealed that intermetallic compounds were formed in the interface of the optimum joint where the tensile failure was observed to take place.


Introduction
Many applications in auto-mobile, aero-space, power generations and military purposes need to join dissimilar metals. Using conventional fusion methods are not preferred for dissimilar jo ining because of solidificat ion defects and intermetallic co mpounds (IM Cs) formations. Therefore, the solid-state joining methods have received much attention for this purpose. Friction stir weld ing (FSW) is a revolutionary solid-state joining process patented at The Welding Institute (UK) in 1991. The joining process proceeds in a solid-state where temperature during welding is relatively less than the melting point of base metals. Friction Stir weld ing process is based on a quite simple concept: a nonconsumable rotating weld ing tool is plunged into adjoining parent metals. The heat generated by the friction between the tool shoulder and weld metal during this process makes the surrounding metals around the tool soft and causes local p lastic deformat ion. The softened metals are stirred together by the rotating tool pin resulting in a solid state bond. FSW min imizes or eliminates defects caused by melting, such as, impurit ies, porosity, solidification cracks and (IMCs) formation. [1]

FSW parameters
FSW imp roper process parameters cause abnormal metal flow or imp roper heat input that lead to defective weld ing. Therefore, the effect of process parameters on weld mechanical properties and microstructure evolution is attractive topics for researchers [2]. The fo llo wing is list of FSW process parameters: 1. Rotational speed 2. Traverse speed 3. Pin offset 4. Axial force 5. Tool geometry A. Es maeili et al. [3] investigated the effect of rotational speed and pin offset on the dissimilar FSW joint of alu miniu m 1050 to brass and a sound joint was successfully performed at rotational speed and pin offset of 450 rp m and 1.6 mm to alu min iu m respectively. The sound joint structure at the nugget zone of alu miniu m was made of a co mposite structure, consisting of brass particles and intermetallic, especially at the upper reg ion of the weld cross section. Also P. Xue et al. [4] investigated the effect of rotational speed and pin offset on mechanical p roperties and microstructure of the dissimilar FSW joint of alu miniu m 1060 to pure copper. The sound joint with 110 M Pa ult imate tensile strength (UTS) (84.6% of A l base metal) was produced at 600 rpm rotational speed and 2 mm pin offset to aluminium. C.W. Tan et al. [5] studied the effect of the traverse speed on appearance, microstructure and format ion mechanis m, and mechanical propert ies of 5A 02 FSW alu min iu m alloy -pure copper joint and the results revealed that At traverse speed of 40 mm/ min, cav ity defect was observed in the cross -sectional macrograph indicating incomp lete mixing between Al and Cu. Th is defect was usually associated with insufficient metal flow caused by insufficient heat input. At traverse speed of 20 mm/ min , the proper mixing occurred and void-free jo int was obtained with tensile strength of 130 MPa representing joint efficiency of 75.6% of the A l alloy base metal.

5
Aluminiu m and brass widely in industries, especially in sheet metal forming processes. Fusion welding of these metals is not possible because of solidificat ion defects and intermetallic co mpounds (IMCs) formations . Practically, researches in FSW of these metals are rare especially those used Taguchi optimization technique. In this investigation, an attempt is made to investigate process parameters such as rotational speed, traverse speed and pin offset on FSW of aluminiu m and brass. The levels and values of process parameters are listed in Table 1.

Taguchi Method
Taguchi method is a good problem solving tool, which can upgrade/improve the performance o f the product, process, design and system with a significant minimizing in experimental time and cost. Further, by using this technique, the most influential parameters can be determined in the overall performance. The optimu m process parameters obtained fro m the Taguchi method are not sensitive to the variat ion in environ mental condition and other noise factors. The Taguchi method uses a special design of orthogonal array to study the entire process parameter space with a small nu mber of experiments only. With the increase of process parameters, the nu mber o f experiments increases . Taguchi proposed a three-step Approach [6]: 1. System design 2. Parameter design 3. Tolerance design In system design, the engineer applies scientific and engineering knowledge to produce a basic functional prototype design. The objective of parameter design is to optimize the settings of the process parameter to improve quality characteristics. In addition, it is expected that the optimal process parameter values obtained from parameter design are insensitive to variation and other noise factors. Tolerance design is used to determine and analyse tolerances around the optimal settings recommend by the parameter design. Tolerance design is required if the reduced variation obtained by the parameter design does not meet the required performance, and involves tightening tolerances on the product or process parameters fo r which variat ions result in a large negative influence on the required product performance.
Steps of Taguchi parameter design are following [2,6]: Step 1: Selection of factors to be evaluated.
Step 2: Selection of number of levels for each factor.
Step 3: Selection of the appropriate orthogonal array.
Step 4: Conduct the experiments.
Step 5: Analyse the data and determine optimu m levels for control factors.
Step 6: Confirmation experiment. The selection of which orthogonal array to use depends on the following [2, 6]: 1. The number of factors and interactions of interest. 2. The number of levels for the factors of interest.
As three factors and three levels are selected, L 9 orthogonal array (OA) is used in this investigation. Only the main factor are taken into consideration and not the interactions. The chemical co mposition and mechanical properties of the plate are given in Tables 2 and 3.

Experimental Procedure
The materials used in this investigation are commercial pure aluminiu m and brass (CuZn30). Both metals are received as fully rolled plate with 5 mm in thickness. The received base metals were cut to square plates of 100 mm in width. Tradit ional milling machine was used to produce the FSW jo ints. Brass plate was fixed in the advanced side. FSW tool was made of heat treated Bohler K340 cold work tool steel with shoulder diameter of 20 mm, tapered pin with 5-4 mm in diameter and 4.85 mm in length and hardness of 57 HR C . The experiments were performed using L 9 orthogonal array (OA). The selected process parameters are rotational speed, traverse speed and pin offset. The tensile tests are conducted on three traverse flat tension test specimens for each condition according to ASTM E 8M Standard [7] on a universal testing machine. Fig. 1 represents the flat tensile test specimen d imensions. The ultimate tensile strengths were determined for each specimen. Table 4

Analysis of tensile test results
For each experimental condit ion, three tensile test trials are conducted and three ultimate tensile strength values (responses) are determined. For analysis of the results, S/N ratios and mean values are calculated for each condition and are illustrated in table 4. The S/N ratio is calculated based on the quality of the characteristics intended. The higher the better S/N rat io is calcu lated because the objective function of this investigation is to maximize the tensile strength.
The average mean and S/N ratio values for each level of each parameter are calculated and illustrated in tables 5 and 6. The main effects for mean and S/N ratio are plotted in Fig. 2 and 3. Based on mean values the optimal level setting is A 2 B 1 C 2 and based on S/N ratio the optimal level setting is A 1 B 1 C 2 .
Analysis of Variance (A NOVA ) is performed for mean and S/N ratio. The aim of ANOVA is to determine the significant process parameters and level of confidence of them. The ANOVA table for mean and S/N ratio are calculated and listed in tables 7 and 8. Tabulated F values are listed at table 9. In this investigation, the analysis based on the mean revealed that the traverse speed and the pin offset are significant factors at confidence levels of 95% and 90% respectively, and the analysis based on S/N ratio revealed that the traverse speed and the pin offset are significant factors at confidence levels of 99% and 95% respectively. As the rotational speed is not significant factor as revealed fro m the both analyses; the analysis based on S/N ratio and the analysis based on mean, the optimu m setting revealed fro m analysis of mean is selected and the level 2 of rotational speed factor (350 rpm) is considered the optimal setting in this work.      tool to the brass interface, forging force and brassshoulder contact area will be increased resulting in more softening of brass that will produce continuous large frag ments of brass in the stirring zone acco mpanied by cavities and cracks which reduce the UTS of the joint as shown in Fig. 4. But the all fractures of joints of 0.5 mm pin offset occurred in the interface. So the UTS might be decreased as a result of formation of IM Cs layers at interface of the joints. Increasing forging force and brass -shoulder contact area leads to excessive heat generation resulting in fo rming brittle thick IM Cs layer in the interface. Increasing the pin offset to 1.5 mm, the jo int UTS will reduce as a result of low forged b rass through alu min iu m side especially in the bottom of the joint. Consequently, the bottom of the joint interface is poor welded resulting in low UTS of the jo ints as shown in Fig.  5. According to the results, pin offset of 1 mm scored highest UTS as a result of providing sufficient forging force and optimu m heat generation that decreased formation of IMCs layer at joint interface. According to Fig. 2, at traverse speed of 11 mm/ min, the UTS of the jo int reached its highest value as a res ult of sufficient of two factors accompanied with traverse speed; heat input rate and stirring time. Heat input rate is responsible for softening the metals and the stirring time is responsible for giv ing the metals a chance to mix together. The UTS of the joint decreased at the traverse speeds of 14 mm/ min and 18 mm/ min as a result of the heat input rate and stirring time are insufficient to softening the brass and make a good diffusion between brass and aluminium The predicted value of response for the optimu m setting (Run no. 4 setting) is calculated by the below formula: Where, Y Mean is the total mean of response, Y i is the mean of response at optimal level of the factor, and i is the factor number.
This predicted values is co mpared with experimental values (run no. 4) in table 10 and that indicates a high accuracy; 99.4% and 97.6%.

Hardness test results
Vickers hardness test is performed on the traverse cross section of the run no. 4 welded joint (optimu m jo int). Various readings are gathered fro m different displacements from the interface on aluminiu m and brass sides along the middle o f the traverse cross section. Fig. 6 shows the average hardness profile p roceeded. Hardness profile reveals that hardness differs according to the weld ing zones on alumin iu m side; stirring zone (SZ) and base metal have appro ximately the same hardness number. The thermal mechanical affected zone (TMAZ) and the heat affected zone (HAZ) have lower hardness number as a result of exposed to rapidly elevated temperature. Hardness number of brass in the HAZ is lower than the base metal as a result of exposing to elevated temperature. p

Optical microscope examination
A specimen of the jo int produced by the optimu m condition (4th run) is investigated by the optical microscope. Fig. 7 shows the microstructure of alu min iu m side in stirring zone, thermal mechanical affected zone, heat affected zone and alu miniu m base metal. The grains size d iffers in the d ifferent zones. Fig.  7a shows large alu miniu m grains in the base metal zone. Fig. 7c shows the smallest grain size that located in the thermal mechanical affected zone because exposing to rapidly elevated temperature. As the grains located in the heat affected zone are exposed to lower temperature than those in TMAZ, their size is larger than those in TMAZ as shown in Fig. 7c and Fig. 7d. Fig. 7b shows moderate grains size in the stirring zone as the tool pin forged the grains which are deformed and elongated to be larger than those in TMAZ and HAZ.

SEM examination
A specimen of the jo int produced by the 4th run conditions is prepared for investigation under SEM. Fig.  8 shows that small brass particles exist in the stirring zone of alu miniu m but with very small ratio. A lu miniu m is the dark side and brass is the bright side. Fig. 9 shows large brass fragments in the end of aluminiu m stirring zone acco mpanied with cavities. These large frag ments are existed as a result of the brass being very hard and brittle. In tensile test, the fracture occurred in the interface at 52.5 MPa. That means the intermetallic compound that supposed to play the ro le o f the bonding is very brittle and has tensile strength lower than the tensile strength of the stirring zone and HAZ [ 8]. Fig. 10 shows the interface investigation that reveals format ion   of intermetallic co mpound layer along the interface with thickness of 0.887 μm.

Conclusions
Experiments of FSW process using different levels of various parameters were conducted to produce butt joints of 5 mm thickness aluminiu m and brass (CuZn30) plates.
In summary, conclusions can be illustrated as follows: x Within the selected ranges, the optimu m levels of rotational speed, traverse speed and pin offset are 350 rp m, 11 mm and 1 mm respectively with UTS = 52.55 MPa (50% of Al base metal).
x Traverse speed and pin offset have high significant effects on the ultimate tensile strength of the joint and contribute about 65% and 33% respectively to overall contributions. x Failure of optimu m setting occurs in the interface because of forming brittle intermetallic co mpound in the interface.