Journal of Materials Science & Technology  2019 , 35 (7): 1261-1269 https://doi.org/10.1016/j.jmst.2019.01.016

Orginal Article

Joint formation mechanism of high depth-to-width ratio friction stir welding

Yongxian Huang*, Yuming Xie, Xiangchen Meng, Junchen Li, Li Zhou*

State Key Laboratory of Advanced Welding and Joining, Harbin Institute of Technology, Harbin 150001, China

Corresponding authors:   *Corresponding authors.E-mail addresses: yxhuang@hit.edu.cn (Y. Huang), zhouli@hitwh.edu.cn(L. Zhou).*Corresponding authors.E-mail addresses: yxhuang@hit.edu.cn (Y. Huang), zhouli@hitwh.edu.cn(L. Zhou).

Received: 2018-05-7

Revised:  2018-06-11

Accepted:  2018-07-9

Online:  2019-07-20

Copyright:  2019 Editorial board of Journal of Materials Science & Technology Copyright reserved, Editorial board of Journal of Materials Science & Technology

More

Abstract

High depth-to-width ratio friction stir welding is an attractive method for the joining demands of aluminum profiles, which is sparked with its extremely low heat input and high mechanical performance. In this study, the joint formation mechanism was studied by a numerical model of plastic flow combined with experimental approaches. A fluid-solid-interaction algorithm was proposed to establish the coupling model, and the material to be welded was treated as non-Newtonian fluid. The thread structure and the milling facets on tool pin promoted drastic turbulence of material. The thread structure converged the plasticized material by its inclined plane, and then drove the attached material to refill the welds. The milling facets brought about the periodic dynamic material flow. The thread structure and the milling facets increased the strain rate greatly under the extremely low heat input, which avoided the welding defects. The condition of the peak temperature of 648 K and the strain rate of 151 s-1 attributed to the lowest coarsening degree of precipitate. The tensile strength of the joint reached 265 MPa, equivalent to 86% of base material. The amelioration via the material flow model inhibits the welding defects and optimizes the parameter intervals, providing references to extracting process-structure-property linkages for friction stir welding.

Keywords: Friction stir welding ; Joint formation ; Material flow ; Al-Mg-Si alloy ; Microstructure ; Mechanical properties

0

PDF (5512KB) Metadata Metrics Related articles

Cite this article Export EndNote Ris Bibtex

Yongxian Huang, Yuming Xie, Xiangchen Meng, Junchen Li, Li Zhou. Joint formation mechanism of high depth-to-width ratio friction stir welding[J]. Journal of Materials Science & Technology, 2019, 35(7): 1261-1269 https://doi.org/10.1016/j.jmst.2019.01.016

1. Introduction

Extensive attentions in friction stir welding (FSW) are now focused on the tool design for complex structural applications [[1], [2], [3], [4]]. FSW is based on severe plastic deformation and low temperature cycling inside welding nugget zone (WNZ), which is implemented by the tool shoulder and pin [5]. Liu et al. [6] stated that the ratio of the pin length (PL) to the shoulder diameter (SD) should be modified to obtain sufficient material flow and heat input. Unrestricted increase in the PL with a certain SD led to the pin fracture or caused the welding defects. For over a decade, an empirical cognition has formed that the optimum ratio of the PL to the SD is about 0.33, as shown in Fig. 1(a). For specific applications of FSW of hollow aluminum profiles with a large thickness-to-width ratio (Fig. 1(b)), the conventional welding tools are incapable of achieving the high-quality joining. An important strategy to realize this challenging goal is the design of the pin based on numerical analysis [[7], [8], [9]]. Indeed, significant efforts in recent years have been aimed at extracting process-structure-property linkages using a variety of numerical approaches [[10], [11], [12], [13]].

Fig. 1.   (a) Depth-to-width ratio data of welding tools (The green points and the green blocks were the depth-to-width ratio data in the published open literature) and (b) illustration of hollow aluminum profiles with a large thickness-to-width ratio.

Colegrove and Shercliff [14] established a 3-dimensional computational fluid dynamics (CFD) model in FSW. The framework gained a preliminary understanding of the plasticized flow around a complex FSW tool. Hoyos et al. [15] presented a phenomenological based semi-physical model to produce an interpretable flow model near the tool surface. Liu et al. [16] evaluated the volume of the deformed materials around the pin, the material flow velocity, and the strain/stress rate based on numerical methods. Non-continuous low and high speed regions around a cylinder pin were revealed. Fagan et al. [17] proposed a friction stir forming model to analyze the complex physical mechanisms, which visualized the surface flow on an analytical circle tool using a particle streamline tracing approach. Tongne et al. [18] studied the material flow induced by the tool pin based on a 2-dimensional Coupled Eulerian-Lagrangian (CEL) formulation. Nevertheless, the comprehensive analysis about the 3-dimensional flow near the tool with the complex topological structure is seldom. A credible plasticized material flow model corresponding to the exploration of the process-structure-property linkages is necessary [[19], [20], [21], [22]].

In the present paper, high depth-to-width ratio FSW, a method suitable for the specific application (Fig. 1(b)), was considered. A fluid-solid-interaction (FSI) algorithm was proposed to establish the coupling model consisting of CFD and computational solid mechanics (CSM). The plasticized material flow near a designed tool with a tapered thread triple-facet pin was studied. Joint formation mechanisms including flow feature, defect analysis, precipitation evolution and joint characterization were discussed in detail.

2. Materials and experimental procedure

The base material (BM) was Al-Mg-Si precipitation hardenable alloy, whose sheet dimensions were 300 mm × 75 mm × 5 mm. The chemical composition and mechanical properties are listed in Table 1. Schematic of the high depth-to-width ratio FSW is shown in Fig. 2. The welding tool is made of H13 high speed steel and has the topologies of the thread and the three milling facets. It has the sizes of 8 mm, 6 mm and 4.8 mm in the SD, the pin root diameter (PRD) and the PL, of which the depth-to-width ratio is 0.6. In contrast, the ratio of conventional welding tool is about 0.33, as shown in Fig. 1(a). A rotational velocity of 800 rpm, a shoulder plunge depth of 0.15 mm and a tool tilt angle of 2.5°, were identical in the high depth-to-width ratio FSW.

Table 1   Chemical composition and mechanical properties of Al-Mg-Si alloy sheets.

Chemical composition (wt%)Mechanical properties
MgSiCuFeMnZnCrTiAlTensile strength (MPa)Elongation (%)
1.100.810.230.750.150.310.070.19Bal.31015.6

New window

Fig. 2.   Schematic diagram of high depth-to-width ratio FSW.

Weld surface appearances were observed by a surface reconstruction method. Metallographic and mechanical specimens were cut perpendicular to the welding line using an electrical discharge cutting equipment. Metallographic specimens were etched by anodic coating with 43 ml H2SO4, 38 ml H3PO4 and 19 ml H2O. Tensile tests were carried out at room temperature at a crosshead speed of 2.0 mm/min using a computer-controlled testing machine. Tensile properties were evaluated with three tensile specimens. Electron backscattered diffraction (EBSD) pattern and fractographs were analyzed by scanning electron microscopy (SEM). X-ray diffraction (XRD) using a Cu- radiation was made from 30° to 80° at a scan rate of 2°/min, to investigate the precipitates on the fracture surface.

3. Numerical modeling

Fig. 3shows the components of the high depth-to-width ratio FSW model. The model consists of three key parts: a CFD method of the fluid domain to simulate the heat generation and the plasticized material flow, a CSM method of the solid structure to describe the deformation and the heat dissipation of the welding tools, and a coupling FSI algorithm to integrate the front two parts and transfer the flux data. Full details of three parts and other essential points are described here.

Fig. 3.   Schematic view of high depth-to-width ratio FSW numerical model including solid structure and fluid domain.

3.1. CFD model

A computational domain with the dimensions of 5 mm × 50 mm × 50 mm was used in the numerical model. The dynamic mesh technique was adopted to achieve the FSI coupling. Only the dwelling procedure during the FSW was calculated in this study. The calculation time lasted for more than 15 s to achieve a quasi-steady state. The solver was set as a pressure-based mode. The Al alloy to be welded was assumed as an incompressible, continuous fluid with Navier-Stokes descriptions as:

∇⋅v=0 (1)

$\rho\frac{dv}{dt}=-\bigtriangledown p+\bigtriangledown ·σ$ (2)

$\rho c_{p}\frac{dT}{dt}=-\bigtriangledown(K\bigtriangledown T)+(σ:\bigtriangledown v)-\phi$ (3)

where ρ is the density of the Al-Mg-Si alloy, t is the flow time, v is flow velocity, σ=μ(∇v+∇vt) is the deviatoric stress tensor, μ is the non-Newtonian viscosity, P is the relative pressure, T is the absolute temperature, ϕ is the sink term which represents the heat loss of convection and radiation, cP is the specific heat capacity and k is the thermal conductivity.

The thermal conductivity and the specific heat capacity of the alloy were given in our previous work [23]. The yield stress σe of the Al alloy equals to the flow stress approximately [24]. It is highly temperature and strain rate dependent according to the constitutive equation introduced by the Sellars-Tegart model [25]:

$σ_{e}=(1-\sqrt{\frac{T-273.15}{T_{melt}-273.15}})\frac{1}{\alpha}sinh^{-1}[(\frac{Z}{A})^{\frac{1}{n}}]+σ_{melt}$ (4)

Here, α, A and n are the material constants of 1.7 × 10-8 Pa-1, 8.081 × 1011 and 6.66, respectively, Tmelt is the melt temperature of the Al-Mg-Si alloy with 908 K, σmelt is the yield stress with 14.4 MPa above the melt temperature. Z is the Zener-Hollomon parameter, which is defined as:

$Z=\dot{ε}_{e}e^{\frac{Q}{RT}}$ (5)

where Q is the activation energy with 181.53 kJ/mol, R is the gas constant with 8.314 J/(mol K) and $ \dot{ε}_{e}$ is the effective strain rate.

Two parts of the heat input were taken into consideration. The surficial heat source term F represents the friction heat at the fluid-solid coupling interface:

$F=β_{1}R_{ijk}\omega·[\frac{δσ_{e}}{\sqrt{3}}+(1-δ)μP_{ijk}sinξ_{ijk}]·cosθ_{ijk}$ (6)

where subscript ijk corresponds to the mark of the elements.

The other plastic deformation heat source term ϕ is the volumetric heat source, which is given as:

$\phi=β_{2}·σ_{e}·\dot{ε}_{e}$ (7)

where β1 and β2 are the empirical coefficient which represents the thermal conversion rate, μ is the friction coefficient which is taken as 0.4.

Fig. 3shows the sketch of the numerical model. The material inlet and the material outlet are set at the front side and the back side of the model as the velocity boundaries. The thermal boundary condition for the heat exchange between the surfaces of the workpieces is given by:

$-k\frac{\partial T}{\partial z}=h(T-T_{∞})+σε(T^{4}-T^{4}_{∞})$ (8)

where σ is the Stefan-Boltzmann constant, ε is the emissivity which is taken as 0.4, h is the heat transfer coefficient which is taken as 15 W/(m2 K), 15 W/(m2 K) and 200 W/(m2 K) on the upper surface, the lateral surface and the lower surface, respectively.

3.2. CSM model

Johnson-Cook material model of the H13 steel was chosen as the constitutive equation for the elastic-plastic analysis. The equation can be written as:

$σ=(A+Bε^{n})[1+C(ln\frac{\dot{ε}}{\dot{ε}_{0}})][1-(\frac{T-T_{0}}{T_{meltSteel-T_{0}}})^{m}]$ (9)

where A, B, C, m and n are material constants, T0 is the reference temperature and TmeltSteel is the melting temperature of the H13 steel. The values of the Johnson-Cook model and the other physical properties are shown in Table 2.

Table 2   Material properties of H13 high speed steel.

Density (kg/m3)Young’s modulus (GPa)Poisson’s ratioSpecific heat [J/(kg K)]Thermal expansion coefficient (10-6 K-1)Thermal conductivity [W/(m K)]
78002110.28560379.1
A (MPa)B (MPa)CmnT0 (K)
908.5321.40.281.180.278300

New window

3.3. FSI algorithm

The coupling algorithm defines the interaction between the tool surface and the plasticized materials, which can be written as:

where f1,f2,f3 and f4 terms are the forcing terms, and the other nonzero entries of the equation can be written as:

$K_{11}=M_{u}+\frac{1}{\rho}K_{\tau}+C_{U}-ΔtK_{u}$ (11)

$K_{12}=G^{T}+\frac{Δt}{2}P$ (12)

K21=G (13)

$K_{22}=Δt[GM^{-1}_{u}(G^{T}+\frac{Δt}{2})P]-H$ (14)

K33=c3Md+c2Cd+c1Kd (15)

$K_{14}=M^{\Gamma,T}_{\lambda u}$ (16)

$K_{34}=M^{\Gamma,T}_{\lambda d}$ (17)

$K_{41}=M^{\Gamma}_{\lambda u}$ (18)

$K_{43}=M^{\Gamma}_{\lambda d}$ (19)

In addition, the thermodynamic properties of the Al-Mg-Si alloy used in this model were calculated using the JMatPro, a CALPHAD based thermodynamic software.

4. Results and discussion

4.1. Surface morphology

Surface appearances of the high depth-to-width ratio FSW joints at different welding speeds are shown in Fig. 4. The shoulder marks with concave troughs and convex crests are clearly visible. The distance of the two neighboring troughs or two neighboring crests is proportional to the advance per rotation at certain welding parameters. The thickness reductions decrease with the increase of welding speed, which are below 100 μm at the welding speeds higher than 300 mm/min. Specifically, the joints in Fig. 4 exhibit the regularly fluctuant surfaces with tiny flashes, indicating the adequate plasticized materials flow under the driving effect of high depth-to-width ratio tool.

Fig. 4.   Characteristics of weld surface appearances by high depth-to-width ratio FSW (The pentagons were the trough thickness reduction points at different welding speeds).

4.2. Metallography and flow feature

The typical macrostructure is shown in Fig. 5. A sound joint has been realized under a low heat input generated by a tiny SD. Apart from BM, WNZ and thermo-mechanically affected zone (TMAZ) can be identified in Fig. 5(a). No heat affected zone (HAZ) is observed, as the EBSD pattern depicted in Fig. 5(g). The shape of the WNZ differs from that of the conventional FSW [26] and appears to be lanky but non-axisymmetric between the advancing side and the retreating side. The interface between the TMAZ and the WNZ at the advancing side is clear, but the interface at the retreating side is obscure as shown in Fig. 5(b and f). The material domain at the leading side is driven by the shear force to refill the cavity behind the tool [27]. The converged materials exhibit a periodic fluctuant structure due to the deep thread feature. There is a great difference in the flow velocities between the crest and the root of the thread, which results in the drastic changes of the strain rate along the interface. A lamellar structure inside the WNZ is observed as shown in Fig. 5(c). According to Tongne et al. [28], the lamellar band is the result of the periodical deposited layers. The space between the two bands is equal to one third of the advance per tool rotation, which accords with the three milling facets. The milling facets increase the extent of the turbulent flow, promoting the joint formation. Fig. 5(d and e) shows the fluctuant interface between the WNZ and the TMAZ at the advancing side. The materials of the WNZ was converged to the interface according to these red arrows, while the materials of the TMAZ was converged to the intervals of the interface. Based on the tool topology in Fig. 2, the downward inclined plane of the thread transforms the materials to move downward, while the upward inclined plane pushes the materials to move upward. Subsequently, the downward and upward materials are accumulated at the converged domains, of which the number is equal to the rounds of the thread.

Fig. 5.   Macrostructure and microstructures of the typical joint: (a) joint macrostructure; microstructures of (b) “B”, (c) “C”, (d) “D”, (e) “E” and (f) “F” in Fig. 5(a); (g) EBSD pattern at advancing side.

4.3. Numerical analysis

Numerical modeling is applied to simulate the flow field near the tool surface for the precise description of the flow pattern. Octree contours of the equivalent strain rate, the velocity and their comparison are displayed in Fig. 6. The distributions of the strain rate and the velocity show the similar feature as the metallographic images (Fig. 5). The differences of the velocity at the thread crest and root are obvious with a fluctuant amplitude of about 2.5. There are opposite pushing directions between the downward and the upward inclined planes of the thread.

Fig. 6.   Distributions of equivalent strain rate, velocity and their comparison: (a) equivalent strain rate contour, (b) equivalent velocity contour and (c) comparison between simulation and experiment at advancing side.

A quantitative curve analysis is carried out by the simulation data from the sampling points, as shown in Fig. 7(a). The 200 sampling points distribute uniformly along the red edge of the pin at the advancing side, of which the sampling direction is from the bottom to the top. Undulation curves of the equivalent strain rate at different welding speeds can be attained in Fig. 7(b). The strain rate of the thread crest reaches 300 s-1, while that of the thread root is only 70 s-1. The drastic change of the strain rate promotes the plasticized turbulence near the pin surface. Additionally, the double-apex structure on single thread crest is an another feature on the strain rate curves. The strain rate gradients at the upward and downward inclined planes of the thread are extremely high, which lead to the relatively lower strain rate at the vertical plane of the thread. The average strain rate of the high depth-to-width ratio FSW is much higher than that of conventional FSW [29]. The plots of the equivalent velocity and the peak temperature are shown in Fig. 7(c and d). The same fluctuant feature induced by the thread structure is obtained. The velocity and the temperature near the tool surface decrease with the raise of welding speed. The low heat generation per length and the less nominal flow velocity at the high welding speed have almost no influence on the severe material flow. An effective promotion of the plasticized materials is obtained.

Fig. 7.   (a) Distributions of the 200 sampling points at the red line, (b) curves of equivalent strain rate, (c) curves of equivalent velocity and (d) curves of peak temperature at different welding speeds.

Fig. 8(a) exhibits the plasticized material flow calculated by the numerical model. In general, the material flow moves downward near the pin surface, while the other flow from the bottom of the joint moves upward near the sub pin surface and refills the cavity induced by the downward stream. In the local area near the thread, the plasticized flow is greatly enhanced by the converging effect of the thread and the dynamic flow by the milling facets. According to Fig. 7(d), the heat input becomes lower obviously compared to the conventional method [30]. The great plastic strain energy induced by the drastic plasticized turbulence compensates the extremely low heat input, which contributes to the sound joint formation during the relatively low-thermal process. The velocity distributions on three Cartesian axes as defined in Fig. 8(a) are shown in Fig. 8(b). The material flow mainly exhibits the characteristics of the downward movement along the thickness direction, and moves backward along the direction of the weld synchronously. However, there is also forward stream existing along the welding direction due to the thread pin, increasing the complexity of the material flow during the high depth-to-width FSW. In the lateral direction of the weld, the plasticized materials are mainly characterized by the flow from the advancing side to the retreating side at the leading side of the tool, while the materials flow from the retreating side to the advancing side at the treating side to fulfill the joining process. In conclusion, the flows on the three axes are jointly affected by the fluctuating effects of the deep-thread feature.

Fig. 8.   (a) Illustration of plasticized material flow via high depth-to-width ratio FSW and (b) velocity distributions on three Cartesian axes.

Particle tracing technique is applied to trace the material streamlines and predict the formation of the welds to validate the reliability. The injections are set as massless particles, which are superimposed on the inlet surface with a radius of 0.5 mm. The boundary conditions for injections are identical to the matrix metal. The welding formation predictions and their experimental contrasts are depicted in Fig. 9. The grey clusters of particles indicate the dense particle distribution, while the exposed red region means the occurrence of the defects. The defects appear when the welding speed reaches 400 mm/min, which mainly distribute in the WNZ at the advancing side and the root of the joints. From the comparison data, it can be inferred that the numerical modeling can simulate and predict the defects effectively, which proves the accuracy of the flow analyses powerfully [31].

Fig. 9.   Comparisons of defect distributions between experiments and simulations: (a) and (b) 30 mm/min; (c) and (d) 100 mm/min; (e) and (f) 300 mm/min; (g) and (h) 400 mm/min; (i) and (j) 600 mm/min.

4.4. Fractography

Fig. 10 depicts the tensile fracture morphologies of the joints at different welding speeds. The voids with an average diameter of 19 μm distribute at the root of the WNZ in 300 mm/min. Many voids appear and penetrate each other at the welding speeds of 400 mm/min and 600 mm/min. However, no welding defect can be observed from the metallographic image in Fig. 9(e), which implies that the nucleation of the voids occurs during the tensile process rather than the welding process. The heat generation per length increases with the decrease of welding speed, resulting in the increase of the peak temperature near the tool surface [21]. According to Fig. 7(b), the equivalent strain rate becomes bigger when getting closer to the shoulder along the thickness direction. Therefore, the diameters of the voids decrease gradually and diminish eventually with the increases of the temperature and the strain rate. Additionally, the distance between each row of the voids is approximately equal to one third of the advance per rotation, as depicted in Fig. 10(b and c).

Fig. 10.   Partly fracture surface morphologies of high depth-to-width ratio FSW: (a) 30 mm/min, (b) 300 mm/min, (c) 400 mm/min and (d) 600 mm/min.

Precipitates can be observed in the dimples of the fracture morphologies in Fig. 10(a). From the XRD results of Fig. 11, the phase is concluded to be Mg2Si and Al12(Fe,Mn)3Si. The needle-like β’’ phase (Alx+yMg5-xSi6-y) is unstable against the temperature increase in the WNZ and dissolves completely during the FSW process [32]. Thus, the β’’ phase cannot be detected in the fracture surface of the WNZ. The β phase (Mg2Si) dissolves at around 720-770 K, which is significantly affected by the thermal cycle [33]. However, due to the low heat input of the high depth-to-width ratio FSW, the holding time of elevated temperature between 720 K and 770 K is extremely short and even non-existing. A large amount of the β phases remain undissolved. Subsequently, the β phase is coarsened in accordance with the Wagner and Kampmann (KWN) model [34], which results in coarsened β phase existing in the WNZ and becomes the nuclei of the initial cracks. The rate law of KWN model is defined as:

$v=\frac{\partial r}{\partial t}=\frac{C_{ss}-C_{int}}{C_{p}-C_{int}}\frac{D}{r}$ (20)

$D=D_{0}exp(-\frac{Qd}{R_{g}T})$ (21)

where v is the growth rate of the β phase, Css is the concentration in the precipitation controlling element Mg within the matrix, Cint is the concentration of the Mg element at the matrix/precipitate interface given by the Gibbs-Thomson equation, Cp is the concentration in the precipitate of the Mg element, D is the diffusion coefficient, D0 is a constant, Qd is the activation energy, Rg is the gas constant and T is the temperature.

Fig. 11.   XRD pattern showing precipitations on fracture surfaces.

According to the formulae, the peak temperature and the holding time of elevated temperature drop with the increasing welding speeds, resulting in the decrease of the diffusion coefficient. The growth rate of the β phase reduces then, leading to the relatively small radii of the β phase. In Fig. 10, these hard precipitates become the nuclei of the cracks during the tensile process. The material necking occurs and the cracks form with the increasing plastic deformation [35]. The micro voids grow uniformly and eventually form the dimples, which result in the crack of the joint. Thus, the mechanical properties of the joints are improved effectively by hindering the coarsening of the precipitates, under the condition of extremely low heat input.

4.5. Mechanical properties

Increasing welding speed can improve the mechanical performances of the joint through reducing heat input, while it is also responsible for the welding defects. It is of vital importance to find the best welding parameters for high depth-to-width ratio FSW. A comprehensive evaluation based on the response surface method is applied to optimize the process window. As shown in Fig. 12, when the peak temperature is 648 K and the strain rate is 151 s-1, the void defects are eliminated. Compared to Fig. 7(b and d), the welding parameters of 800 rpm and 300 mm/min are the optimal parameters, which correspond to the best mechanical performances.

Fig. 12.   Response surface of the average void diameter based on the temperature and the equivalent strain rate.

Fig. 13 shows the tensile properties of the joints at different welding speeds. The tensile strength gradually increases firstly and then decreases as the welding speed increases. The elongation decreases with the increase of welding speeds. The tensile properties change drastically due to the strengthening effect of the precipitates [36]. The coarsening degree of the precipitates decreases with the increase of welding speed. Meanwhile, the degree of the dynamic grain refinement is promoted due to the low heat generation. However, the tensile properties decrease drastically due to the welding defects with the welding speeds exceeding 300 mm/min. The maximum tensile strength of 265 MPa is achieved at the welding speed of 300 mm/min and rotational speed of 800 rpm, which is obviously higher than that of the conventional FSW.

Fig. 13.   Tensile properties of the high depth-to-width ratio FSW joints at different welding speeds.

5. Conclusions

(1)The material flow model consisting of the CFD method, the CSM method and the FSI algorithm was proposed to investigate the joint formation mechanism. The numerical model was proved to be accurate and practical for high depth-to-width ratio FSW.

(2)The heat input decreased effectively and the plasticized material flow was enhanced compared to the conventional FSW. The tiny shoulder diameter reduced the heat generation. The thread structure and the milling facets increased the strain rate greatly under the extremely low heat generation, avoiding the welding defects.

(3)The process-structure-property linkage was achieved via the comprehensive evaluation. The peak temperature and the strain rate of 648 K and 151 s-1 eliminated the void defects and achieved the lowest coarsening degree of precipitate.

(4)The optimum tensile strength and elongation of the joint at the welding speed of 300 mm/min and rotational velocity of 800 rpm were 265 MPa and 8.1%, equivalent to 86% and 52% of the BM.

Acknowledgement

The work was supported by the National Natural Science Foundation of China (No. 51575132).

The authors have declared that no competing interests exist.


/