A STUDY ON FRICTION STIR WELDING OF AlSi1MgMn ALUMINIUM ALLOY PLATES

Original scientific paper Presented in this paper are an experimental investigation and numerical simulations of the Friction Stir Welding (FSW) process. The aim was to determine thermo-mechanical and kinematic parameters of the process of joining two AlSi1MgMn aluminium alloy plates with the thickness of 7,8 mm. Experimental study was designed based on the 2-level multifactorial design of experiment (DoE), with centre points. Parameters varied in the experiment were: welding speed, rotation speed, angle of pin slope, pin diameter and shoulder diameter. Based on the adopted geometric parameters, the family of tools was designed and manufactured for the experiment. The experiment was performed in laboratory conditions, which were similar to production environment. Numerical simulation of the FSW process was carried out using DEFORM-3D software. Detailed analysis of the process as well as comparison between experimental and numerical results is presented.


Introduction
During the nineties of the last century a new method of joining similar and dissimilar materials in the solid state without melting of material, known as Friction Stir Welding -FSW was developed. The process was patented by The Welding Institute -TWI in England in 1991, and invented by Wayne M. Thomas who successfully joined plates of aluminium alloys [1÷5, 9,10]. The method is primarily used to join plates of larger thickness. Tools that are used in the process of welding are cylindrical and consist of two concentric parts, which are rotating at high speed. The larger diameter part of the tool is called the shoulder, while the smaller diameter part is called the pin (Fig. 1). Rotating tool slowly approaches the joint line and plunges into material, generating heat. Due to that, the temperature increases to the heat metal forming where mechanical mixing and joining of materials is performed, enabling the tool to move in the longitudinal direction or along the joint lines (Fig. 2). After passing of the tool along the joint lines the solid phase of weld (joint) remains, where the upper plane remains smooth and flat thanks to the tool shoulder, while the lower plane of the workpiece is formed from the basis on which the workpiece is standing and is also smooth and flat [1÷4].  For the friction stir welding process the vertical milling machine "Gorki" 6P13 was used. The material employed in the experiment was aluminium alloy sheet AlSi1MgMn, numerical label 6082-T6, whose chemical composition is given in Tab. 1, and the mechanical properties of the basic material in Tab. 2. The dimension of the workpiece was 200 × 50 × 7,8 mm (L × W × H).  Material of welding tool was 1.2343 steel according to the standard EN 10027-2. The tool was designed in Pro/Engineer, using a program for parametric modelling (Fig. 3a).

Definition of experiment
The aim of experimental investigation was to determine the influence of the tool geometric parameters on thermo-mechanical and kinematic characteristics of the stir welding process. For that purpose a family of tools with different geometrical parameters was developed. The axisymmetrical tool consists of the working part and tool body. The body of the tool is adjusted to the jaws of machine that is used in the experiment. Tool geometry is presented in Fig. 3b.
The tool with adopted dimensions and parameters D, d and α

Design of experiment (DoE)
Based upon preliminary investigation, the 2-level factorial orthogonal plan with four enter points (n 0 = 4) was adopted. As the input values, the elements of welding regime and geometric factors of the tool were chosen: X 1 = v mm/min (welding speed), X 2 = ω rpm (rotation speed of tool), X 3 = α (angle of pin slope), X 4 = d mm (pin diameter) and X 5 = D mm (diameter of the shoulder). The adopted levels of variation of input factors are given in Tab. 3 [5]. Based on adopted values X 3 , X 4 and X 5 the set of nine tools was designed and manufactured, Fig. 4.

The performance of experiment
Experimental investigation was performed in two phases. Phase I: measurement of welding forces (downward force -Fz, longitudinal force -Fx and side force -Fy) and workpiece temperature at six locations. Phase II: mechanical testing of the welded joint.
Research site is presented in Fig. 5 and Fig. 6 provides an overview of welded workpieces for 36 th point of experimental plan.  During the FSW process, forces were measured using precise analog-digital equipment connected to the information measurement system. The sensors, measuring tapes (HBN 6/120LY11), were cemented on special holders which were adjusted to experiment requirements. Fig. 7 shows force-time diagram at the centre point of DoE. When the rotating tool plunges into the material (workpiece), force Fz equals 6,61 kN. The workpiece is heated because of the friction between the pin and the workpiece. After the plunge, force Fz begins to decline until the shoulder contacts the workpiece surface. At this moment the force reaches its maximum value (8,25 kN), because a large area of the shoulder begins to plunge into the workpiece. Due to the friction between the shoulder and the pin with the material, intensive heating of the workpiece takes place.
During execution of the welding process the force Fz begins to decline until the moment when the milling machine is programmed to start longitudinal movement of the table. After that, force Fz retains its constant value, until the tool exits the workpiece.
The values of Fx force depend mainly on the welding speed but tool geometry also influences this force to some extent. During the welding process, force Fy has lowest values.   Fig. 8, allow the measurement of temperature in micro structural zones. In this way we obtain a complete picture of the generation and distribution of heat through the material. Fig. 9 shows the measured values of temperature in the 2 nd (v 2 = 80 mm/min; ω 2 = 1000 rpm; α 2 = 5°; d 2 = 7 mm; D 2 = 28 mm) point of DoE.

Figure 9
The temperature diagram at the 2 nd point of DoE

Mechanical testing
During the FSW process, in the stirred zone significant micro structural evolution takes place, which leads to the substantial changes of mechanical properties within and around the welded joint.
Tensile and toughness properties of the welded workpiece were also investigated. Specimens for tensile test were machined in two different directions relative to the weld: longitudinal and transverse. The arrangement of specimens in welded workpieces was made according to standard for testing of welded joints.
For the Charpy impact test one specimen was machined from the "nugget" zone in lateral direction, where the V-notches were positioned at the opposite sides of the welding direction. Fig. 10 shows position of the specimens for the tensile and impact tests within the welded workpiece.

Figure 10
Position of the specimens for tensile and impact tests which were cut out from the welded workpiece

Determination of tensile strength
Tensile test was carried out on the 200 kN tensile machine. Flat specimens, according to standard EN 10002-1:2008 [6] were used. Testing was conducted at room temperature. For the purpose of tensile testing, 72 specimens were cut from welded workpieces to provide data for all DoE points, while 4 specimens were made from the basic material, which was not welded. Fig. 11 shows the specimen, and Fig. 12 and Fig. 13 present photo images of specimen at the centre points of DoE, in x and y directions.   The diagram shows that the tensile strength is higher for the specimens taken from the longitudinal (x) direction than for the specimens taken from the lateral (y) direction. The ratio between tensile strengths Rm x /Rm y is 1,158, while the elongation ε x /ε y ratio of the specimen from x and y -direction is 1,66.

Determination of impact toughness
Charpy impact tests of welded material were performed with a maximum available initial energy of 150 J.
When the impact occurs, the specimen is exposed to bending. Toughness of the welded material or impact strength to bending, equals the work applied to break the specimen, reduced to a unit of the specimen cross-section: The specimens were made according to the standards EN 10045-1:1993 [7], (Fig. 15). A narrow specimen with a V-notch was adopted.

Numerical simulation
DEFORM-3D was used for finite element analysis [8,9]. Workpieces were designed in Pro/ENGINEER. In order to be able to compare the results obtained by numerical simulation and experiment, 36 simulations of FSW were performed, which correspond to DoE points. Numerical simulation consisted of two stages. The first stage simulated the tool plunging into the material to the depth of 6,9 mm. In fact, the shoulder also plunges 0,1 mm into the workpiece. In the second stage, the process of welding was simulated, with the longitudinal tool travel of 166 mm.
After entering input data and generating finite element mesh, boundary conditions were entered in order to carry out simulation of the FSW process. Simulation results were interpreted in the Post Processor module, in graphical and numerical form. Fig. 16 shows the 3D model used in the first stage of numerical simulations with generated mesh of finite elements and tool. Simulation of the first stage ended with the 117 th step. The downward welding force, Fz, obtained in this stage, is given in Fig. 17. The final distribution of temperature fields in the workpiece (in the 117 th step) as the result of tool action, is given in Fig. 18.
The second stage of the simulation was completed in the 1805 th step. Fig. 19 shows the downward force Fztime diagram for the second stage of simulation.  For the predefined measurement points (Fig. 8), temperature values were obtained and shown in Fig. 21, for numerical simulation number 36.   Analysis of obtained results reveals that the downward force Fz is largely affected by the tool geometry, especially the size of the shoulder. All the obtained values of force Fz can be divided into three areas. The first 16 points of DoE belong to the first area (shoulder diameter of 28 mm). The second area contains DoE points from 17 to 32, (shoulder diameter of 25 mm), while the third area contains the centre DoE points.
In numerical simulations, the values of force Fz can also be divided into three areas, depending on the size of the shoulder.
Dependence of Fz on shoulder diameter, D, obtained by numerical analysis and experimentally is given in Fig.  26. As can be seen from Fig. 24, 25 and 26, numerical results for the Fz are slightly higher than experimental. This can be attributed to the large physical nonlinearity of FSW process.
Temperature measurement was conducted at six points of the workpiece, positioned in two sections, U and L, as shown in Fig. 27.
Temperature distribution is presented by contour diagrams for intersections U and L (Fig. 27) where the values of the measured temperature in °C are given.  It can be seen from a schematic view (Fig. 30) that the selected cross-section, ABCD, represents all the micro-structural zones within the workpiece.  Points T 1 to T 6 were considered for the analysis of numerically obtained temperature distribution. Temperature distribution for the simulation number 3 in the upper zone (intersection U) of the tool position (Fig.  27), is given in Fig. 32. The same diagram for the lower zone (intersection L) is given in Fig. 33.    By comparing temperature distribution obtained numerically and experimentally, one concludes that DEFORM gives slightly higher temperatures than measurement. This difference is about 10 % and it becomes smaller as the distance from the joint line increases.

Conclusion
An AlSi1MgMn Aluminium Alloy was joined by friction stir welding, applying various process parameters. The impact of different input welding parameters on welding forces in x, y and z direction, as well as the temperature fields distribution within the workpiece, was investigated numerically (FEM) and experimentally. Furthermore, mechanical properties (tensile strength and impact toughness of the joints) were determined and discussed.
For the purpose of this experiment, special tool and auxiliary equipment were developed.
The results thus obtained reveal that tool geometry (shoulder diameter, pin diameter and angle of pin slope) and regimes of welding (welding speed and rotation speed) exert major influence on the weld quality. This can be confirmed by the analysis of mechanical parameters (tensile strength and impact strength).
Welding forces obtained by numerical simulation are approximately 10 % higher than experimental forces. This can be explained by imperfections of the experiment and numerical simulation, which requires further analysis.
Comparison of experimentally obtained temperature values measured at six locations of the workpiece with the values obtained by numerical simulation shows a relatively good agreement, with the difference below 10 %. This allows the conclusion that heat generation and temperature distribution within the workpiece in FSW process can be successfully simulated by FEM. Further investigation will be directed towards establishing a generalized model of FWS which would enable a complex analysis of this process, including impact of all relevant parameters as well as their interrelation.