Welding Parameters Optimization in Plunging and Dwelling Phase of FSW Medium Thickness 2219 Aluminum Alloy


 The influence of welding parameters on temperature distribution in plunging and dwelling phase of friction stir welding (FSW) medium thickness 2219 aluminum alloy is blank. Improper selection of welding parameters will result in uneven temperature distribution along the thickness of the weldment, which will lead to welding defects and ultimately affect the mechanical properties of the weldment. To realize the prediction of temperature distribution and achieve the optimization of welding parameters, a simulation model of FSW 18mm thick 2219 aluminum alloy is built based on DEFORM. The validity of the simulation model is verified by temperature measurement experiments. With the minimum temperature difference in the core area of the weldment as target value, and weldable temperature range of 2219 aluminum alloy as constraint conditions, orthogonal experiments are conducted considering the rotational speed, the press amount, the tool tilt angle, the plunging traverse speed and the dwelling time. The results of variance analysis show that the rotational speed and the dwelling time are significant factors affecting temperature field during plunging and dwelling phase. Through single factor simulation, the welding parameters in plunging and dwelling phase are optimized. This study provides a reference for realizing high-quality welding of a heavy rocket fuel tank.


Introduction
A heavy-lift launch vehicle is the basic launch vehicle for the long-term development plans for space ight and deep space exploration. A rocket fuel tank is the main load-bearing structure of a rocket, which has extremely high requirements for manufacturing quality and reliability. The advanced rocket fuel tank is made of 2219 high-strength aluminum alloy with a thickness of 18mm. Friction stir welding (FSW) has the advantages of no need to add welding wire, no shielding gas, no pollution, no smoke, no radiation, high welding e ciency, small product deformation after welding, and excellent weld mechanical properties and has become the main connection process for the rocket fuel tanks [1].
The whole process of FSW includes three phases: plunging and dwelling, welding, and tool withdrawal phase. The plunging and dwelling phase is the initial phase, which will affect the subsequent welding phase. In this phase, the pin is slowly pressed into the weldment. The plasticization and ow behavior of the weldment material near the pin is the basis for the formation of the joint. Improper selection of welding parameters during plunging and dwelling phase of FSW 18mm thick 2219 aluminum alloy will result in high temperature gradient and uneven temperature distribution along the thickness of the weldment, and thus leads to defects in welded joints, such as ashes, holes, incomplete penetration, etc..
Welding defects directly affects the mechanical properties of the weldment. The thermo mechanical affect zone and weld nugget zone of the weldment are de ned as the welding core area. The material ow in this area is intense, which directly affects the weld morphology and welding quality. It is di cult to measure the temperature distribution in the welding core area due to the tool rotation, shielding of the shoulder, material ow, and severe plastic deformation in the welding zone. Therefore, nite element method (FEM) has become an important means to study the temperature distribution in the welding core area. At present, software such as ANSYS, MSC. Marc, FLUENT, ABAQUS and DEFORM are widely adopted to simulate the FSW process.
Scholars have simulated temperature distribution in the welding phase of FSW based on heat source model. McClure et al. [2], Jiang et al. [3] and Liu et al. [4] used the Rosenthal analytical method to analyze the transient temperature distribution in FSW process, and obtained the thermal cycle curve of each feature point in the welding zone. He et al. [5] and Wan et al. [6] used MSC. Marc established FSW simulation models, analyzed the ultrasonic-assisted FSW and FSW, and studied the temperature distribution during the welding process. Ren et al. [7] and Xu et al. [8] established FSW heat source model based on the torque heat source model, and used ANSYS to study the temperature distribution of the welding process and the residual stress after welding. Most scholars simulated FSW process based on heat source model, and studied the in uence of welding parameters on temperature distribution. Complex nonlinear friction heat and plastic deformation heat in FSW process make it di cult to use analytical methods to describe temperature distribution.
Scholars have simulated temperature distribution in the welding phase of FSW based on FLUENT and ABAQUS. Eyvazian et al. [9] and Feng et al. [10] used FLUENT to simulate FSW process of dissimilar and same materials, and studied the temperature distribution of weldments. Yang et al. [11] used FLUENT to establish a simulation model for friction stir lap welding of dissimilar materials of Q235 steel and 6061 aluminum alloy, and studied the in uence of different welding process parameters on the temperature distribution. Su et al. [12] and Yang et al. [13] established FSW simulation models of different stir pin shapes based on CFD method, and studied the effects of stir pin shape, shoulder radius, rotational speed and welding speed on temperature distribution in the welding process. The simulation method based on CFD has no strict restriction on the mesh size, but it cannot simulate the plunging and dwelling phase and the tool withdrawal phase, only the welding phase. Some scholars used ABAQUS to simulate FSW process, and studied the in uence of size parameters of the tool and the welding parameters on temperature distribution of the weldment. Zhang et al. [14], Iordache et al. [15] and Liu et al. [16] used the Arbitrary Lagrangian-Euler (ALE) method in ABAQUS to avoid mesh loss, established FSW simulation model, and studied the velocity distribution, temperature distribution and equivalent plastic strain distribution during the welding phase, and studied the in uence of different tool size parameters on the temperature distribution. ALE method usually ignores the simulation of plunging and dwelling phase to reduce the deformation of the mesh, and is di cult to simulate the whole process of FSW medium thickness weldment.
DEFORM attracted the attention of many scholars to simulate temperature distribution of FSW. Zhou et al. [17] and Han et al. [18] simulated the FSW process using local mesh re nement and adaptive following technology, obtained the temperature, strain distribution and material ow law of the welding process, and studied the in uence of the rotational speed, welding speed and down pressure on the temperature distribution. Asadi et al. [19] used DEFORM to study FSW process of magnesium alloy, and adopted point tracking method to study the temperature distribution and material ow in the welding process. DEFORM has strong mesh re nement capability, which reduces the total number of meshes and shortens the computation time. In the process of simulation, if the mesh deformation reaches a certain degree, DEFORM will automatically re-divide the global meshes to make the model easier to converge. Most notably, DEFORM can simulate the whole phase of FSW.
Scholars have carried out research on temperature distribution in the welding phase, and have drawn many meaningful conclusions. However, no research on the optimization of welding parameters in the plunging and dwelling phase has been done. There are many welding parameters in the plunging and dwelling phase, such as the rotational speed, the press amount, the tool tilt angle, the plunging traverse speed and the dwelling time. The in uence law of the contributing factors on temperature are still blank. Different welding parameters are suitable for weldments of different materials and sizes. Therefore, it is necessary to study the in uence of welding process parameters on the temperature eld of FSW 18mm thick 2219 aluminum alloy during the plunging and dwelling phase, and to optimize the welding parameters. To realize the prediction of temperature distribution of FSW 18mm thick 2219 aluminum alloy and the optimization of the welding parameters, a simulation model of FSW is built based on DEFORM. The validity of the simulation results is veri ed by experiments. With the minimum temperature difference in the core area of the weldment as the target value, and the weldable temperature range of 2219 aluminum alloy as the constraint conditions, orthogonal experiments and variance analyses are carried out. The research determines the signi cant in uence factors of the temperature difference and achieves the optimization of welding parameters in the plunging and dwelling phase.

Geometrical Model and Meshing
The tool is composed of a shoulder with a concave angle and a conical stir pin, as shown in Fig. 1a. The radius of the shoulder is 16mm. The root radius and length of the conical stir pin are 7.5mm and 17.8mm, respectively. The weldment with the dimensions of 100 × 150 × 18mm 3 is modeled. The threedimensional model after assembly is shown in Fig. 1b.
The mesh needs to be re ned in the interaction area between the weldment and the tool to improve the simulation accuracy and coarsen on the outside of the interaction area to short the computation time. The position of mesh windows is shown in Fig. 2a. The size ratio is set to 4 to avoid the mesh outside the mesh windows from being too large to affect simulation accuracy. The mesh window density is set as 1 mm. The simulation model after meshing is shown in Fig. 2b.

Material Properties
The chemical composition of 2219 aluminum alloy is shown in Table 1. Fig. 3 gives the temperaturedependent material properties of the 2219 aluminum alloy.  [20,21]. Johnson-Cook material model describes the effect of strain rate and temperature on ow stress and is used in FSW simulation [22]. Johnson-Cook constitutive equation is written as:

Boundary Conditions and Frictional Model
Boundary conditions are divided into mechanical and thermal boundary conditions. Mechanical boundary conditions limit the degree of freedom of the weldment to ensure that the weldment is xed during the simulation process. That is, limit the degree of freedom of movement in Z direction of the bottom surface of the weldment, and limit the degree of freedom of movement in X and Y directions of the side surface of the weldment.
Thermal boundary conditions setting mainly includes setting heat transfer mode and coe cient between the tool, weldment and backing plate. To shorten simulation time, heat convection coe cient of the bottom surface of the weldment and the air is used to simulate the heat conduction between the bottom surface of the weldment and the backing plate. Room temperature is set to 15°C, heat convection coe cient of the bottom surface of the weldment and the air is set to 5N/mm·s·°C, and heat convection coe cient of the remaining surface of the weldment, the surface of the tool and the air is set to 0.025N/mm·s·°C [24].
During the welding process, the temperature of the contact area between the weldment and the tool increases, and the surface of the weldment material with lower strength is partially sheared. Under the action of friction, part of the weldment material will be stick to the surface of the tool. To describe the state of the contact area between the weldment and tool during the welding process accurately, shear friction model that changes with temperature is adopted. The equation is expressed as: Where is the contact stress at the interface of the weldment and tool; m is the shear factor; k is the shear strength. The coe cients of friction are de ned as temperature-dependent, as shown in Table 3.

Implementation of simulation
Simulation of FSW 18mm thick 2219 aluminum alloy is realized, and the temperature distribution of the weldment in the three phases, i.e. plunging and dwelling, welding and tool withdrawal is obtained. The temperature distribution is shown in Fig. 4a, 4b, and 4c respectively.

Veri cation Of The Simulation Results
Temperature measurement experiments based on thermocouples are carried out in Capital Aerospace Machinery Company, and a temperature measurement system based on LabVIEW developed by our research group is used to collect the temperature data of the sampling points in the welding process. The thermocouples used for temperature measurement are K-type thermocouples, and the temperature measurement range is -200~1250°C. The FSW equipment is a large gantry FSW equipment, as shown in Fig. 5. The weldment with the thermocouples embedded is shown in Fig. 6.
The plunging traverse speed is 20mm/min, the welding speed is 100mm/min, the tool tilt angle is 2.5°, the press amount is 0.2mm, and the rotational speed is set as 350r/min, 400r/min, and 450r/min, respectively. Sampling points on the advancing side (AS) and the retreating side (RS) are obtained for temperature comparison to verify the validity of the built simulation model, as shown in Fig. 7. Figure 8 shows the comparison of experimental and simulated temperature curves of sampling points on the AS and RS when the rotational speed is 400r/min.
The measured peak temperature on the AS and RS are 458.3°C and 429.9°C, respectively, and the peak temperature on the AS and RS obtained by simulation are 464.1°C and 434.5°C, respectively. The relative errors between the simulation and experimental peak temperature of the sampling points on the AS and RS are 1.27% and 1.07%, respectively. Figure 9 shows the comparison of experimental and simulated temperature curves of feature points on the AS with rotational speed of 350r/min, 400r/min, and 450r/min. From Fig. 9, we can see that the experiment and simulation feature point temperature change law is basically the same, and both have experienced the process of heating-peak-cooling. The measured peak temperatures on the AS are 413.2°C, 424.2°C, and 417.9°C, respectively, and the simulated peak temperatures on the AS are 426.1°C, 429.9°C, and 431.7°C, respectively. The relative errors of the simulation and experimental peak temperature on the AS are 3.12%, 1.34%, and 3.30%, respectively. The validity of the established simulation model is veri ed.

Optimization Of Welding Parameters In Plunging And Dwelling Phase
It has been proved that the friction heat generated by the friction between the shoulder and the weldment is higher than the friction heat generated by the stir pin [3,4,6]. In addition, the heat generated by the friction between the bottom surface of the tool and the weldment is transferred to the backing plate under the weldment, resulting in uneven heat input along the thickness of the weldment. The temperature difference in the core area of the weldment directly affects the joint structure, which ultimately affects the mechanical properties of the welded joint. Scholars have found that the part that hinders the overall tensile strength of the joint is the bottom joint of the core area [25][26][27]. If the bottom temperature increases, the dynamic recrystallization degree increases, the grain boundary angle increases, and the tensile strength of the joint increases. Therefore, the minimum temperature difference in the core area of the weldment is selected as the optimization goal.
When the temperature of the core area of the weldment is 80% of the melting temperature, the surface of the weld joint is smooth and the tensile strength of the joint is the highest [28]. The solidus temperature of 2219 aluminum alloy is 548℃, and the liquidus temperature is 649℃. 80% of the solidus and liquidus temperature are regarded as the upper and lower limits of the temperature in the core area of the weldment.

In uence of welding parameters on temperature difference of the weldment in plunging and dwelling phase
The welding parameters of the plunging and dwelling phase include the rotational speed, the press amount, the tool tilt angle, the plunging traverse speed and the dwelling time. All the welding parameters are selected as the factors of the orthogonal experiment, and the orthogonal experiment is carried out to study the in uence of the above parameters on the temperature difference of the weldment. The orthogonal table is shown in Table 4.
The sum of the freedom degrees of each factor is the number of factors × (number of levels -1) = 5 × (3 -1) = 10, which is less than the total freedom degree of L 18 (3 7 ). The orthogonal table is selected as L 18 (3 7 ).
The temperature difference between the peak and the minimum temperature in the core area is extracted through the post-processing module. The orthogonal experiment scheme and the temperature difference results are shown in Table 5.  Figure 10 shows that the peak temperature of the 18 groups of models does not exceed the upper limit of the temperature in the core area of the weldment. Only the minimum temperature of the 7th group model is within the weldable temperature range of 2219 aluminum alloy.
Then the mean square is calculated with the sum of squares and degrees of freedom to calculate the F test value. The analysis of variance table is shown in Table 6. It shows that the rotational speed and dwelling time have a signi cant effect on the temperature difference of the weldment. Among them, the dwelling time has the greatest in uence on the temperature difference, followed by the rotational speed.

Optimization of welding parameters in plunging and dwelling phase
The results of variance analysis show that the dwelling time and the rotational speed have signi cant in uence on the temperature difference. Next, comprehensively research on the in uence of the rotational speed and dwelling time on the temperature difference is conducted.
The plunging traverse speed is set as 20mm/min, the welding speed is set as 100mm/min, the tool tilt angle is set as 2.5°, the press amount is set as 0.2mm, the rotational speed is set as 300r/min, 400r/min, 500r/min, and 600r/min respectively, and the dwelling time is set as 0s, 2.5s, 5s, 7.5s, 10s, 12.5s, and 15s respectively. The peak temperature of weldment with different rotational speed and dwelling time is shown in Fig. 11. Figure 11 shows a certain upward trend in the peak temperature during the dwelling phase of the different rotational speeds. When the rotational speed is 600r/min, the peak temperature exceeds the upper limit of the temperature in the core area of the weldment after dwelling for 2.5s. When the rotational speed is less than or equal to 500r/min, the peak temperature within 15s is lower than the upper limit of the temperature in the core area of the weldment.
The peak and the minimum temperature in the core area of weldment with different rotational speeds and dwelling times are shown in Fig. 12. Figure 12 shows that when the rotational speeds at 300r/min and 400r/min, the minimum temperature in the core area of weldment is lower than the minimum value of the weldable temperature in the core area of the weldment. When the rotational speed is 500r/min and the dwelling time is from 5s to 15s, the peak and the minimum temperature in the core area of weldment are both within the weldable temperature range of 2219 aluminum alloy. When the rotational speed is 600r/min, the peak temperature in the core area of weldment is higher than the maximum value of the weldable temperature in the core area of the weldment.
The temperature differences in the core area of the weldment with different rotational speeds and dwelling times are shown in Fig. 13. Figure 13 shows that the higher the rotational speed, the smaller the temperature difference. There is no signi cant difference in the temperature difference when the dwelling time is in 5-15s range at the rotational speed of 500r/min and 600r/min. The reason may be that with the increase of rotational speed, the temperature of the contact area between the tool and the weldment increases, and the surface of the weldment material with lower strength is partially sheared. Under the action of friction, part of the weldment material will be stick to the surface of the tool, reducing the friction coe cient, which is manifested by the fact that there is no signi cant change in the temperature difference when the rotational speed increases to a certain extent.
When the dwelling time is 5s, the temperature difference of the weldment decreases sharply, and when the dwelling time is 15s, the temperature difference of the weldment decreases in signi cantly The reason may be that the tool and the weldment produce frictional heat during the FSW process after dwelling for 5s, and the plastic deformation heat generation and heat dissipation of the weldment reach a stable state.
From the above analyses, when the rotational speed is 500r/min, the dwelling time is 5s, the tool tilt angle is 2.5°, the plunging traverse speed is 20mm/min, and the press amount is 0.2mm, the temperature difference of the weldment is relatively small, so the optimal welding parameters for the plunging and dwelling phase is obtained.

Conclusion
To realize the effective prediction of temperature eld of FSW medium thickness 2219 aluminum alloy and the optimization of the welding parameters, a simulation model of FSW 18mm thick 2219 aluminum alloy is established based on DEFORM, and the simulation of the plunging and dwelling, welding, and tool withdrawal phases is realized. The temperature eld of weldment during the welding phase are obtained, and the effectiveness of the model is veri ed by experiment. The maximum relative error of the peak temperature of the sampling points is 3.30%, and the average relative error is 2.02%.
With the minimum temperature difference in the core area of the weldment as the target value, and the weldable temperature range of 2219 aluminum alloy as the constraint conditions, orthogonal experiments are carried out in consideration of the rotational speed, the press amount, the tool tilt angle, the plunging traverse speed and the dwelling time. The results of variance analysis show that the rotational speed and dwelling time exert great in uence on temperature eld during the plunging and dwelling phase.
The simulation results show that when the rotational speed is 600r/min, the peak temperature exceeds the maximum value of the weldable temperature of 2219 aluminum alloy after dwelling for 2.5s. When the rotational speeds is 300r/min and 400r/min, the minimum temperature in the core area of weldment is lower than the minimum value of the weldable temperature of 2219 aluminum alloy. When the dwelling time is 5s, the temperature difference of the weldment decreases sharply, and when the dwelling time is further extended, there is no signi cant change in the temperature difference of the weldment.
Considering the processing e ciency and temperature difference comprehensively, the optimal welding parameters combination of FSW 18mm thick 2219 aluminum alloy in the plunging and dwelling phase is obtained: the rotational speed is 500r/min, the dwelling time is 5s, the tool tilt angle is 2.5°, the plunging traverse speed is 20mm/min, and the press amount is 0.2mm.

Con icts of interest/Competing interests
Not applicable. The authors declare that they have no con ict of interest.

Availability of data and material
All data generated or analysed during this study are included in this manuscript.

Code availability
The code is available on request.
Authors' contributions Meshing: a diagrammatic drawing of the mesh window position, b simulation model after meshing   The weldment with the thermocouples embedded  Comparison of experimental and simulated temperature curves of sampling points on the AS Page 18/21

Figure 10
The peak and the minimum temperature in the core area of the weldment Page 19/21

Figure 11
The peak temperature of weldment with different rotational speed and dwelling time Figure 12 The peak and the minimum temperature in the core area of weldment with different rotational speeds and dwelling times