Friction stir welding (FSW) has many advantages rather than fusion welding, but details of internal phenomena during its processes have not yet been clarified. In this study, a thermo-mechanically coupled process model was developed to investigate FSW phenomena inside a tool and workpiece. As a workpiece, 6061-T6 aluminum alloy was employed. The system of FSW process model includes several thermal boundaries. Among heat flows through these boundaries, heat transfers into the exterior of the system become more sensitive to tool and workpiece temperatures than heat transfers within the system. This paper especially focused on a heat transfer coefficient at a workpiece bottom, and optimized it through experiments and finite element method (FEM) analyses. The tool temperatures during FSW were measured with a special tooling system with imbedded thermocouples within a tool. As a result, an analysis model that is able to investigate details at a wide range of traverse speeds was developed for practical high speed welding. Then, the accuracy of developed FEM model was validated with them. Finally, the temperatures and stress distribution around workpiece/tool interfaces were investigated with the developed model.
Friction stir welding (FSW) invented by the Welding Institute is a solid state joining technique initially intended to weld aluminum alloys [
It is of great importance to investigate these phenomena during FSW for sound joining of workpieces and high product qualities. Elucidation of the thermo-mechanical phenomena between tool and workpiece can lead to the optimization of welding parameters, effective tool design, and the application of FSW to new materials and products. For deep understanding of the FSW process, both experimental methods and numerical simulations can be adopted.
Welding parameters, material flows and tool/workpiece interface phenomena have direct influences on heat generation during FSW. Although much research has investigated the important aspect of welding phenomena based on temperatures measured at specific points of workpiece, only a few experimental studies have been reported about tool temperatures because very complicated settings are required for tool temperature measurements [
Temperature analyses of the tool and workpiece during FSW have been conducted in many research studies. One way to optimize the process is to utilize the thermo-mechanically coupled modeling of FSW, considering thermal boundaries and friction at tool/workpiece interfaces. Hence, numerical models to simulate many phenomena that occur inside the material and tool and at their interfaces have been increasingly required. However, it is very difficult to establish a model that exactly meets the experimental results over a wide range of FSW conditions. FSW models include some thermal boundaries, and temperatures of tool and workpiece are sensitive against heat flows through the boundaries, so that the optimum values of heat transfer coefficients are necessary for these thermal boundaries. However, almost no study has investigated the precise thermal boundary conditions for FSW modeling because relatively low traverse speeds have been adapted for the FSW simulations.
Relationships between process parameters, such as rotational and traverse speed, and inside temperatures are important information for production sites. Although some studies investigated the relationship between the traverse speed and FSW temperatures for aluminum alloys, they analyzed the temperatures at low traverse speed under around 3.5 mm/s [
In this study, a thermos-mechanically coupled process model was developed to investigate FSW phenomena inside a tool and workpiece and at their interface using the DEFORM-3D software. As a workpiece, 6061-T6 aluminum alloy was employed. The system of FSW process model includes several thermal boundaries. Among heat flows through these thermal boundaries, the heat transfers into the exterior of the system become more sensitive to tool and workpiece temperatures than the heat transfers within the system. This paper especially focused on a heat transfer coefficient at the workpiece bottom, and optimized it through experiments and finite element (FEM) analyses for a reasonable simulation model to be established. Although other FEM studies have investigated FSW process phenomena under low tool traverse speeds corresponding to laboratory scale experiments, this study developed an analysis model that is able to investigate material internal phenomena at a wide range of traverse speeds for practical high speed welding. Both as fundamental data and reference data for simulation modeling, the tool temperatures during FSW were measured with a special tooling system with imbedded thermocouples within a tool. Then, the heat transfer coefficient at the workpiece bottom was optimized and the accuracy of developed FEM model was validated. Finally, the temperatures and stress distribution around workpiece/tool interfaces were investigated with the developed simulation model.
Several experiments with 6061-T6 aluminum alloy were conducted to measure tool temperatures and welding forces while butt welding of the aluminum sheets. The obtained results served as the fundamental data to investigate FSW phenomena both inside the tool and workpiece and at their interface, and are used to develop an FEM analysis model which accurately represents internal phenomena around the tool/workpiece interface under a wide range of traverse speeds.
The experiments were conducted on a Hitachi Power Solution 2D-FSW machine and traverse and axial forces were measured with load cells equipped under the machine table. The geometry of each workpiece was 200 mm long, 70 mm wide and 4 mm thick and the sides of two workpieces were butt welded. The joining surfaces were machined previously to meet each other without small gaps. The tool had a shoulder 12 mm in diameter and 3˚ in taper angle, and a threaded probe 6 mm in diameter and 3.8 mm in height.
With respect to welding temperature, tool side measurement was adopted because tool temperatures are under nearly stable conditions during FSW process.
The tool inside temperatures during FSW were measured with MULTI INTELLIGENCE R tooling system for FSW.
Tool | Alloy tool steel SKD61, Thread pin |
---|---|
Workpiece | 6061-T6 |
Workpiece geometry | 200 × 70 × 4 mm |
Tool rotational speed [min−1] | 750, 1000, 2000 |
Traverse speed [mm/s] | 1, 5, 10, 20 |
Tool advancing angle [degree] | 3 |
study but also for practical welding, traverse speeds were set for a wide range from 1 mm/s to 20 mm/s, close to the practical traverse speed at production sites. In addition, the tool plunge speed was set to 0.5 mm/s, and then the tool was held for 5 s for the dwelling before traversing.
To clarify the thermal and mechanical phenomena of workpieces and tools during the FSW process, it is essential to understand temperature distributions, stress fields and strains inside tools and workpieces or at their interfaces. These inside and interface phenomena are difficult to investigate only by experiments. Hence, a numerical analyses model with FEM that is able to cover a wide range of welding speeds considering precise thermal boundary conditions was developed in this study.
The commercial software DEFORM-3D was used for the FEM simulation, and Lagrangian-code was employed for the analyses. The workpiece was set to a continuum model of 70 × 70 × 4 mm with meshing into about 32 thousand tetrahedral elements. The element sizes around the tool/workpiece interfaces were set to be about 0.1 - 0.3 mm in length, finer than those in the surroundings. The tool in this model had the same geometries as that used in experiments except that the probe height was 3.6 mm. That is, the probe height was 0.2 mm smaller in the FEM simulation than in experiments and mesh breaking at model’s bottom while re-meshing was avoided by an increase in the clearance between the tool tip and workpiece bottom. The probe part of the model was a cylinder pin without threads, giving priority to analysis stability during FSW simulations. There were 15 thousand tool elements, and tool/workpiece interface elements were set to be finer like the workpiece modeling. The workpiece model had a rigid plastic body and tool model a rigid body, and thermo-coupled viscous-plastic analyses were conducted.
σ = [ A + B ε n ] [ 1 + C ln ε ˙ * ] [ 1 − T ∗ m ] (1)
A: yield stress, B: strain hardening constant, ε: equivalent plastic strain,
n: hardening exponent, C: strain rate constant, ε ˙ * : dimensionless strain rate,
T * : homologous temperature, m: temperature exponent.
A system of FSW process model includes several thermal boundaries in itself. A thermal boundary within the system is at the tool/workpiece interface, where the heat flowing between them remains inside the system itself. As for thermal boundaries against the exterior of the system, there are heat convections from the tool surface and the workpiece-top surface to the air, and heat conductions from the workpiece-bottom surface to the back plate. The amount of heat transfer from a solid body to the air is relatively small. In contrast, the heat transfer between two solid bodies in contact is not negligibly large when there are large temperature differences between them. For this reason, the heat transfer coefficient between them has a large influence on the amount of heat removed from the system and corresponding temperature reduction. Therefore, its value could be sensitive to analysis results, so that the heat transfer coefficient between the workpiece bottom and back plate was optimized in this research.
The optimization of the workpiece-bottom heat transfer coefficient for the FEM analysis model was carried out by comparing temperatures numerically and experimentally obtained. The traverse speed and the tool rotation speed were set to 5 mm/s and 1000 min−1, respectively. Five different values of the heat transfer coefficient at the workpiece bottom surface were set between 1.0 to 5.0 kW/m2/K for optimization. For modeling a convective heat transfer as the thermal boundary condition, Equation (2) was used.
q = h × ( T w − T f ) (2)
q: heat flux, h: heat transfer coefficient, Tw: object1 surface temperature,
Tf :ambient temperature or object 2 temperature.
The constraint condition of the workpiece was that the bottom surface of the workpiece was in contact with the back plate without friction and the side surfaces are fixed in all directions. Workpiece and tool surfaces conduct heat transfer against the air of ambient temperature 20˚C. The friction model between the tool and workpiece was followed by a shear friction model shown in Equation (3), and the frictional coefficient m = 1.0 was employed [
τ = m × k (3)
τ : shear stress, m: friction coefficient, k: shear yield stress.
After the heat transfer coefficient was optimized for the workpiece bottom surface at a traverse speed of 5 mm/s, the validation analyses were carried out at different traverse speeds to compare the analysis and experiment results of tool temperatures and traverse force.
Figures 5-8 represent the data profile of tool temperature histories while welding at traverse speeds of 1, 5, 10 and 20 mm/s, respectively. The tool rotational speeds were 1000 min−1. In each graph, temp1 is the temperature at the probe tip, temp2 at the probe root and temp3 at the tool shoulder. Each graph also indicates the welding forces of the axial and traverse directions. It should be noted that no defect was visually detected for the above four welding conditions.
Heat transfer coefficient [kW/m2/K] | |
---|---|
Workpiece bottom (to back-plate) | 1, 2, 3, 4, 5 |
Workpiece top-surface | 0.01 |
Tool/workpiece interface | 5 |
Tool surface | 1 |
The above results confirmed that temperatures of each portion increased monotonously from the tool plunge stage to the dwelling (rotation holding)
stage. Here, traversing began after a 5-second tool rotation hold. After the traverse began, probe tip temperatures (temp1) became almost constant and this steady state continued until all processes finished. On the other hand, the temperatures of the probe root (temp2) and tool shoulder (temp3) increased continuously, but slightly throughout the traverse process. Generally, the measured temperatures of tool tips were about 100˚C to 150˚C higher than those of tool shoulders.
As for welding forces, they were almost constant during the traverse process at traverse speeds of 1, 5 and 10 mm/s, but at a traverse speed of 20 mm/s they kept increasing until the process finished.
In the experiment result, the tool-tip temperature was 462˚C, the shoulder temperature 313˚C and the traverse force 1113 N. As for the analysis results calculated for five different heat transfer coefficients, tool-tip temperature decreased by 67˚C with increasing heat transfer coefficient from 1.0 to 5.0 kW/ m2/K. On the other hand, the shoulder temperature changed little with heat transfer coefficient. These suggest that the amount of heat flows from workpiece
bottoms to the back plate affects tool-tip temperatures sensitively since the tool-tip is close to the workpiece bottom. In contrast, the traverse force increased with heat transfer coefficient. This is because the material flow stress increased with decreasing tool tip temperature.
Error = | T t , a − T t , m T t , m | + | T s , a − T s , m T s , m | + | F t , a − F t , m F t , m | (4)
where Tt,a and Tt,m are tool tip temperatures analyzed and measured, respectively, Ts,a and Ts,m are shoulder temperatures analyzed and measured, respectively, and Ft,a and Ft,m are traverse forces analyzed and measured, respectively. The error ratio was the minimum at 2.0 kW/m2/K, the optimum value of heat transfer coefficient. It was also confirmed from
the best agreement with the experiment ones for a heat transfer coefficient of 2.0 kW/m2/K.
This optimization process suggests that workpiece-bottom thermal boundary conditions remarkably affect tool-tip temperatures and welding forces, i.e. material inner flow stress. The optimized value of heat transfer coefficient for the workpiece bottom obtained in this study was quite different from values of 0.2 kW/m2/K [
Validation was conducted with the developed analysis model for a low traverse speed of 1 mm/s to a high speed of 20 mm/s, comparing the analysis data with experiment results.
Internal thermal-mechanical phenomena around the tool/workpiece interface were investigated with the developed analysis model. The analyses for a low traverse speed of 5 mm/s and a high speed of 20 mm/s were performed to investigate the temperatures and stress distributions around the tool/workpiece interfaces.
surface temperatures are higher than temperatures measured with thermocouples.
With regard to workpiece inner temperatures comparing the traverse speed of 5 and 20 mm/s, the tool front temperature distributions were definitely different, showing that the temperature was lower for 20 mm/s than for 5 mm/s. This is because cold workpiece rapidly come close to the tool at higher traverse speed before sufficient temperature increase by plastic and friction work. As for the tool front stresses, the maximum stress was about 70 MPa at 5 mm/s, whilst it exceeded 110 MPa at 20 mm/s. For this reason, a zone of higher stress appeared in front of the tool at 20 mm/s, showing that the flow stress of aluminum alloy 6061-T6 is critically temperature-dependent.
Tool temperatures were measured during FSW, and a thermal boundary condition was optimized with these experiment results to develop an analysis model. The following was learned through this study:
1) The heat transfer coefficient of workpiece bottom was optimized to h = 2.0 kW/m2/K. Slight differences of heat transfer coefficients affected the analysis results for tool temperatures and traverse force, indicating thermal boundary conditions while actual welding may affect the workpiece and tool internal mechanical phenomena.
2) The analysis modeling considered with an optimum value of workpiecebottom heat transfer coefficient led to the reasonable model that can analyze FSW phenomena through a wide range of tool traverse speeds.
3) Analysis results with the developed model show that temperature inside the probe was around 440˚C at the center of the top surface and 400˚C at the probe root. Calculated temperature distribution proved that the measured temperature with the thermocouple at the probe tip is almost the same as the surface temperature. On the other hand, the shoulder had a steep temperature gradient along the tool axis, and thus the measured temperatures inside the tool were different from the surface temperatures by about 40˚C to 50˚C. It showed that the shoulder surface temperatures are higher than temperatures measured with thermocouples.
The authors are grateful to YAMAMOTO METAL TECHNOS CO., LTD. for technical support with measurements and tool precision machining.
Nakamura, T., Obikawa, T., Yukutake, E., Ueda, S. and Nishizaki, I. (2018) Tool Temperature and Process Modeling of Friction Stir Welding. Modern Mechanical Engineering, 8, 78-94. https://doi.org/10.4236/mme.2018.81006