You are on page 1of 21

International Journal of Machine Tools & Manufacture 91 (2015) 1 –

11

Contents lists available at ScienceDirect

International Journal of Machine Tools & Manufacture


journal homepage: www.elsevier.com/locate/ijmactool

The effect of the welding parameters and tool size on the thermal
process and tool torque in reverse dual-rotation friction stir welding
L. Shi a, C.S. Wu a,b,n
, H.J. Liu b
a
MOE Key Lab for Liquid-Solid Structure Evolution and Materials Processing, and Institute of Materials Joining, Shandong University, Jinan 250061, China
b
State Key Laboratory of Advanced Welding and Joining, Harbin Institute of Technology, Harbin 150001, China

a r t i c l e i n f o ab st r act

Article history: Reverse dual-rotation friction stir welding (RDR-FSW) is a novel variant of conventional friction stir
Received 5 October 2014
welding (FSW) process. The key feature is that the tool pin and the assisted shoulder are separated and
Received in revised form
rotate reversely and independently during welding process, thus it has great potential to improve the
13 January 2015
weld quality and lower the welding loads through adjusting the rotation speeds of the tool pin and the
Accepted 13 January 2015
Available online 16 January 2015 assisted shoulder independently. A 3D model of RDR-FSW process is developed to analyze the effect of
welding parameters and tool size on the thermal process and the tool torque quantitatively. The model
Keywords:
considers the effect of the welding parameters on the dimensionless slip rate and the friction
Reverse dual-rotation friction stir welding
coefficient between the tool-workpiece contact interfaces. It is found that with an increase of the radial
(RDR-FSW)
distance, the locations of peak and valley values of heat generation rate at the shoulder-workpiece
Numerical modeling
Thermal process contact interfaces vary from the retreating side (RS) to the advancing side (AS) and from the AS to
Material flow the RS, respectively. Although the reverse rotation of the tool pin and the assisted shoulder has little
Torque effect on the total heat generation, the corresponding material flow pattern and the distribution of heat
generation rate lead to a more homogeneous temperature distribution and a much lower torque
exerted on the workpiece in RDR-FSW process. The model is experimentally validated by comparing
the measured thermal cycles
with the calculated data.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction torque and plunge force are needed for the purpose of adequately
softening the material to form a good weld [7–9]. The relatively
Friction stir welding (FSW) process has been proved to be a high stress subjected by the tool causes severe tool wear and
successful solid-state joining process for aluminum alloys [1,2]. premature tool failure [9]. In addition, the relatively high tan-
During the process, a specially designed rotating tool is inserted gential speed at the periphery of the shoulder may cause over-
into the adjoining edges of the workpieces and then moved all heating or even incipient melting along the shoulder edge when
along the line of joint [3]. The heat generated by both friction and thick plates are joined [10–12], which leads to the mechanical
plastic deformation softens the material near the tool, and severe property degradation of the joint, especially for welding pre-
plastic deformation and flow of this plasticized metal occurs as the cipitation-hardened aluminum alloys [3,13,14].
tool moves. Material is transported from the front of the tool to the To overcome the above shortcomings of the conventional FSW,
trailing edge where it is forged into a joint [4,5]. The simultaneous the reverse dual-rotation friction stir welding (RDR-FSW) process
rotation and transverse motion of the tool create asymmetrical has been proposed as a variant technique [10–12]. In RDR-FSW, the
temperature distribution and material flow between the two sides tool pin and assisted shoulder are separated and rotate with op-
of the weld, which leads to different microstructures and me- posite direction independently, as shown in Fig. 1. Thus, the tool
chanical properties between the advancing side (AS) and the re- pin can rotate in a relatively high speed while the assisted
treating side (RS) of the weld [6]. The asymmetry of the welded shoulder can rotate in an appropriate matching speed. In this way,
joint is a unique characteristic of the FSW method, which leads to the tendency towards overheating or incipient melting can be
the deterioration of microstructures and mechanical properties of avoided through optimizing rotation speeds of both the tool pin
the weld [6]. During the conventional FSW, relatively high and the assisted shoulder [11,12]. In RDR-FSW, the welding
rotation torque exerted on the workpiece by the tool pin is partly offset
by the reversely rotating assisted shoulder. As a result, the total
n
Corresponding author. Fax: þ 86 531 8839 torque
2711. exerted on the workpiece by the tool is reduced. Thus, the
E-mail address: wucs@sdu.edu.cn (C.S. Wu).

http://dx.doi.org/10.1016/j.ijmachtools.2015.01.004
0890-6955/& 2015 Elsevier Ltd. All rights reserved.
2 L. Shi et al. / International Journal of Machine Tools & Manufacture 91 (2015) 1–
11

2. Numerical modeling

2.1. Geometric model description

A schematic drawing of RDR-FSW is shown in Fig. 1. In RDR-


FSW, the tool pin with the sub-size concave shoulder is mounted
on the spindle of a conventional FSW machine, while the assisted
shoulder is mounted on the fixed frame of the conventional FSW
machine through a self-designed support frame [11,12]. In this
way, the tool pin with sub-size concave shoulder rotates with the
same rotation speed and direction of the spindle. However, the
assisted shoulder is driven by two servo motors mounted on the
self-designed support frame so that it rotates independently and
reversely. During the RDR-FSW process, the side of the weld
where the pin rotation and welding direction are the same is
Fig. 1. Schematic drawing of reverse dual-rotation friction stir welding (RDR-FSW).
defined as advancing side (AS), and the opposite side of the weld
is defined as retreating side (RS). Detailed description of the RDR-
FSW process can be found in literatures [11,12].
2024 Aluminum alloy plates (500 mm in length, 300 mm in
clamping equipment can be simplified, and the size and mass of width, and 5 mm in thickness) are welded by RDR-FSW. The di-
the welding equipments can be lowered [10–12]. In addition, the mensions of the tool is given in Table 1. Table 2 lists the nominal
differences of the temperature distribution and material flow be- chemical composition of the 2024 aluminum alloy. The yield
strength of 2024 aluminum alloy is obtained from literature [32].
tween the AS and the RS can be reduced by the reverse rotation
The specific heat capacity and the thermal conductivity are ob-
of the tool pin and the assisted shoulder. Thus, the homogeneity
tained from literature [33], which is given in Table 3.
of temperature fields and microstructures can be
The geometric model is shown in Fig. 2a. A 3D Cartesian co-
improved [12,15,16]. In a word, the RDR-FSW process has great ordinate system is established on the plate. The origin of co-
potential to reduce the tool torque and improve the properties ordinate system is located at the intersection between the bottom
of the joints through adjusting the rotating speed of the tool pin surface of the workpiece and the tool axis. The welding direction is
and the as- sisted shoulder independently [10–12,15,16]. identical to the positive x-axis, and the z-axis is along the thick-
However, the reverse and independent rotating of the tool pin and ness of the plate (towards to the top surface of the workpiece).
the assisted shoulder increases the complexity of the process The commercial CFD code FLUENT was used to conduct the
and the number of the process parameters. The complicated numerical simulation [4]. The sensitivity analyses of the size of
interactions among the welding parameters and physical the Eulerian calculation domain were conducted. In order to
variables simultaneously affect the thermal process and the final compare the cal- culated results with the experimentally
properties of the joints [17–19]. In order to optimize the process measured data, the cal- culation domain equal to the real
workpiece dimension was used to conduct the numerical
and the relevant properties of the joints, a fundamental
simulation. Non-uniform grid system was used to discretize the
knowledge of the complex thermal process should be required,
calculation domain. Finer grids were used near the tool to
and a rigorous numerical model coupled with experimental
resolve the severe plasticized material flow and heat convection.
validation is suitable. The design and number of mesh elements were determined by
In the past decades, numerous thermal models [20–23], ther- considering both the efficiency and accuracy of the computation.
mo-mechanical or thermo-flow coupled models [24–30] have The grid dependence study was conducted, and the calculation
been developed to analyze the thermal process in conventional accuracy was found independent on the grid system. Top views
FSW. Furthermore, a model to evaluate the tool torque in con- of mesh system for finite volume calculation was shown in Fig.
ventional FSW has been established, and the effect of welding 2b.
parameters on tool torque has been analyzed [31]. All those pro-
vide significant progress in quantitative understanding of the 2.2. Governing equations
conventional FSW process. Recently, Shi et al. [15,16] developed a
model to analyze the heat transfer and material flow in RDR-FSW. RDR-FSW process involves fully-coupled heat generation, heat
However, the effect of the welding parameters on the di- transfer, material flow and microstructure evolution. Only the
mensionless slip rate and friction coefficient between the tool- quasi-steady state (the welding period) is dealt with in this study.
In the quasi-steady state, the material near the tool is heated to a
workpiece contact interfaces has not been considered in the
relatively high temperature, and only the plastic deformation is
model, which leads to an over predicted value of the temperature
considered. The plasticized material during the RDR-FSW process
near the tool. In addition, there is no quantitative analysis of the
is assumed to behave as an in-compressible and single-phase
tool torque during RDR-FSW.
In this study, a numerical model is developed to analyze the Table 1
heat generation, temperature distribution, material flow and tool Tool dimensions for RDR-FSW (Unit: mm).
torque for different welding parameters and tool size in RDR-FSW
process. The dimensionless slip rate and friction coefficient at the Assisted shoulder Sub-size concave Tool pin
tool-workpiece contact interfaces are determined under different shoulder

welding conditions. The effect of the welding parameters and tool


Outer Inner Outer Inner Diameter Diameter Pin length
size on the thermal process and tool torque during RDR-FSW diameter diameter diameter diameter at the at the tip
process are quantitative analysis. To validate the model, the cal- root

culated thermal cycles are compared with the corresponding


14.0 10.0 10.0 6.0 6.0 4.0 4.8
measured data.
L. Shi et al. / International Journal of Machine Tools & Manufacture 91 (2015) 1–11 3
1⎫
Table 2

Chemical composition of 2024 aluminum alloy (wt%).
⎡ ⎤
1 ⎛ Z ⎞n
2

⎪ ⎛ Z ⎞n
1 2
σ¯
μ
ln ⎨⎜ ⎟ ⎢
⎜ ⎟


1
= 3ε = 3εα
⎪⎪⎝ A⎠ + ⎢ +⎝ A⎠ ⎥ ⎪
Element Si Fe Cu Mn Mg Ni Zn Ti Al ⎦
⎩ ⎭ (3)

Wt.% 0.15 0.25 4.58 0.63 1.59 o 0.10 0.20 o 0.10
Balance
⎛Q ⎞
Z = ε exp ⎜ ⎟ = A (sinh
(4)
Table 3 ασ¯) n
⎝ RT ⎠
Specific heat capacity and thermal conductivity of 2024 aluminum alloy.
where σ¯ is the flow stress, ε is the effective strain rate [4,34,35],
Temperature (K) 290 373 473 573 673 Q is the temperature-independent activation energy [34], R is
1 1
Specific heat capacity/(J kg K ) 864.0 921 1047 1130 the gas constant, T is the temperature, and α, A, n are the
1172
1 1 material constants which represent the characteristics of material
Thermal conductivity/(W m K ) 120.0 134.4 151.2 172.2
176.4 [34].
The thermal energy conservation equation,
non-Newtonian fluid. For simplification, the tilt angle of the tool is
∂ (uT
i ) ∂ ⎞
p ⎛ v
taken as zero, and the tool shoulder is assumed to be flat [4,5]. The ∂ T
ρc p (T) = − ρc (T) U + ⎜λ (T) ⎟+S
∂T

0.1 mm gap between the assisted shoulder and the sub-size con- ∂xi ∂x1 ∂xi ⎝ ∂xi ⎠ (5)
cave shoulder is ignored. The governing equations are given as
where λ (T) is the thermal conductivity, c p (T) is the specific heat of
follows:
the material, and Sv is the viscous dissipation heat source due to
The continuity equation,
plastic deformation near the tool in the shear zone which can be
∂ui calculated as [4,5,35]
=0
∂xi (1) Sv = fm μ (ε, T) (6)
Φ
where u is the velocity of material flow, and index notation for
i ¼ 1, 2, and 3 represents x, y, and z directions as shown in Fig. where fm is an arbitrary constant that indicates the extent of
2a, respectively. atomic mixing in the system, and Φ can be given by [4,5,35]
The momentum equations, ⎛ ∂ 2 2 2⎞
⎛ u1 ⎞ ⎛ ∂u 2 ⎞ ⎛ ∂u 3 ⎞
∂ui uj ⎞ Φ = 2 ⎜⎜ ⎜ ⎟ +⎜ ⎟ +⎜ ⎟
∂p ∂ ∂u j ui ∂u j ⎟
⎛ ∂ ⎝ ∂x1 ⎠ ⎝ ∂x 3 ⎠ 2
ρ ⎝ ⎝ ∂x 2 ⎠
u⎠
=− + μ +μ − ρU 2
∂xi ∂x j ∂xi ⎜⎝ ∂xi ∂x j ⎟⎠ ∂x j (2)
u
2
⎛∂ u1 ∂ u2 ⎞ ⎛ ∂u1 ∂3⎞ ⎛ ∂u3 ∂2⎞
+⎜ + ⎟ +⎜ + +⎜ + ⎟
where ρ is the density, P is the pressure, U is the welding speed, ⎟
⎝ ∂x2 ∂x1 ⎠ ⎝ ∂x 3 ∂x1 ⎠ ⎝ ∂x2 ∂x 3 ⎠ (7)
and μ is the non-Newtonian viscosity which can be express as
[33,34]
Fig. 2. (a) Geometric model in the simulation, (b) top views of mesh system for finite volume calculation (the red frame area magnified the mesh elements near the tool).
(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
4 L. Shi et al. / International Journal of Machine Tools & Manufacture 91 (2015) 1–11
Fig. 3. A schematic illustration of tool-workpiece contact interfaces.

Fig. 5. Calculated heat generation rate at the shoulder-workpiece contact


interfaces for sample 10-800-800-60 (the black line in the middle shows the
boundary of the sub-size concave shoulder).

The heat generation rate at Ω1 region can be express as


⎛ U 2πω1 ⎞
q a = ⎣⎡η (1 − δ) τc + δμf P1 ⎦⎤ ⎜ sin θ − r⎟
⎝ 60 60 ⎠ (8)

The heat generation rate at Ω2 region can be written as [4,5,35]


2
q i= ⎣⎡η (1 − δ) τ c + δμ f P2 ⎤⎦⎛ 2πω r − U sin θ⎞
⎜⎝ ⎟⎠ (9)
60 60

The heat generation rate at Ω3 region can be written as [7]


2
q = ⎡η (1 − δ) τ c + δμ f σ y ⎤⎦⎛ 2πω r − U sin θ⎞
ps ⎣ ⎜⎝ ⎟⎠
60 60 (10)

Fig. 4. Variations of dimensionless slip rate and friction coefficient for different Similarly, the heat generation rate at Ω4 region can be express
rotation speeds. as

Table 4
The typical welding parameters used in this study.

Samples Designation Inner Diameter of assisted shoulder Di Assisted shoulder rotation speed ω1 Tool pin rotation speed ω2 Welding speed U (mm/
(mm) (rpm) (rpm) min)

10-800-800-60 10 800 800 60


10-800-800-120 10 800 800 120
10-800-800-180 10 800 800 180
10-600-1000-120 10 600 1000 120
*
6-800-800-120 6 800 800 120
8-800-800-120 8 800 800 120
12-800-800-120 12 800 800 120
10-800-1000-100 10 800 1000 100
10-600-1200-200 10 600 1200 200

n
The inner diameter is equal to pin diameter at the root, there is no sub-size concave shoulder at this condition.

2.3. Heat generation model


2
q = ⎡η (1 − δ) τ c + δμ P2 ⎤⎦⎛ 2πω r − U sin θ⎞
pb ⎣ f ⎜⎝ ⎟⎠
Fig. 3 shows the tool-workpiece contact interfaces. The 60 60 (11)
assisted
shoulder-workpiece contact interface is defined as Ω1 region
where η is the conversion efficiency of plastic deformation, δ is the
(Ri ≤ r ≤ Ro, z = hmax ). The sub-size concaver shoulder-workpiece dimensionless slip rate, τc is the contact shear stress, μ f is the
contact interface is defined as Ω2 region (R po < r < Ri , z = hmax ),
friction coefficient, P1 is the plunge force of assisted shoulder, P2 is
while Ω3 (R pi ≤ r ≤ R po, hmin ≤ z ≤ hmax ) and region the plunge force of tool pin, ω 1 is the assisted shoulder rotation
Ω4
(0 ≤ r ≤ R pi , z = hmin ) represent the pin side-workpiece contact speed, ω 2 is the tool pin rotation speed, r is the length between an
interface and pin bottom-workpiece contact interface, elemental area and the tool axis, and θ is the angle between the
respectively. welding direction and the r radius vector direction.
L. Shi et al. / International Journal of Machine Tools & Manufacture 91 (2015) 1–11 5

Fig. 6. Variations of heat generation rate with angle θ for different radial distance from the tool axis. (a)r = 3.5 mm, (b)r = 4.5 mm, (c)r = 5.5 mm, (d)r = 6.5 mm (sample 10-
800-800-60).
Table 5
Values and locations of θp and θ v under different radial distance. Q total = ∮Ω 1
qa dA + ∮Ω 2
qi dA + ∮Ω 3
qps dA + ∮Ω 4
qpb dA

Radial distance (mm) θp θv + ∮Ω S dV (14)


v
Value Location Value Location
where dA is an element area at the tool-workpiece contact inter-
3.5 45° RS 225° AS faces, Ω is the shear zone near the tool where viscous dissipation
4.5 25° RS 200° AS heat generation occurs, and dV is an elemental volume inside the
5.5 355° AS 175° RS
shear zone.
6.5 345° AS 165° RS
And the pin heat generation fraction is written as,

In this study, partial sticking condition was assumed at the ∮Ω3 qps dS + ∮Ω4 qpb dS
tool-workpiece contact interfaces. The effect of the welding φp =
Q total (15)
parameters on the dimensionless slip rate and the friction coeffi-
cient can be expressed by the following relations [7,36]:
During the RDR-FSW process, the torque exerted on the
workpiece by the tool pin with sub-size concave shoulder is in
⎛ ωr ⎟⎞
δ = 0.31 × exp ⎜⎝ − 0.026 positive z-direction while the torque exerted on the workpiece by
1.87 ⎠ (12)
the assisted shoulder is in negative z-direction. Based on the tor-
μ f = 0.5 × exp ( − δωr) (13) que at an element area [7,36], the total torque exerted on the
workpiece by the tool during RDR-FSW can be expressed as

where ω in Eqs. (12) and (13) can be substituted by ω 1 and ω 2 for


assisted shoulder and tool pin with sub-size concave shoulder, MR = ∫Ω2 r × ⎡⎣ (1 − δ) τc + δμ f P⎦2 ⎤ dA + ∫Ω3 r ×⎣ ⎡ (1 − δ) τc + δμ f⎦σ y ⎤ dA
respectively. Eqs. (12) and (13) were based on the reported
experimental data of the tool-workpiece interfacial slip in a
+ ∫Ω4 r × ⎡⎣ (1 − δ) τc + δμ f P⎤2 ⎦ dA − ∫ r ×⎡ ⎣ (1 − δ) τc + δμ f⎤P1⎦
(16)
dA
cross-wedge rolling process [37]. Eqs. (12) and (13) have been
used in conventional FSW process and proved to be appropriate where the first, second, third and fourth terms on the right-hand
[7,36]. Fig. 4 shows the variations of dimensionless slip rate and side of Eq. (16) represent the torque exerted on the workpiece by
friction coefficient for different rotation speeds. the sub-size concave shoulder, by the side surface and the bottom
Then, the total heat generation is obtained as follows: surface of the tool pin, and by the assisted shoulder, respectively.
6 L. Shi et al. / International Journal of Machine Tools & Manufacture 91 (2015) 1–11
∂T (λρc p ) Work
λ = q
∂n (λρc p ) Work + (λρc p ) Tool (18)

where the subscripts Work and Tool denote the workpiece and
the tool, respectively. And q can be substituted by qa , qi , qps and
qpb for the assisted shoulder, the sub-size concave shoulder, the
pin side surface and bottom surface, respectively.
Both convective and radiative heat losses were considered for
the heat exchange between the top surface of the workpiece and
the surroundings [5]
∂T
−λ = σ ε (T 4 − T 4 ) + h (T − T
) r r a t a
∂z (19)

where σr is the Stefan–Boltzmann constant, εr is the external


emissivity, h t is the convective heat transfer coefficient at the
top of the workpiece, and Ta is the ambient temperature.
At the bottom and side surfaces of the workpiece, the boundary
condition for heat exchange involves only convective heat transfer
∂T
λ = hb (T − Ta ) (20)
∂n
where hb is the heat transfer coefficient at the bottom and side
surfaces of the workpiece.

3. Results and discussions

In this study, the plunge force of the assisted shoulder and the
tool pin are kept constant (29.5 MPa). The outer diameter of the
Fig. 7. Volumetric heat flux (viscous dissipation) (a) distribution at transverse assisted shoulder is kept as 14.0 mm, and the dimension of the
cross-section and (b) distribution at different horizontal planes (sample 10-800- tool pin remains unchanged. The inner diameter of the assisted
800-60).
shoulder (and the corresponding outer diameter of the sub-size
concave shoulder) is changed to consider its effect on the heat
2.4. Boundary conditions generation and tool torque. The RDR-FSW samples in the study are
designated in brief forms [22]. For example, 10-800-600-120 de-
The initial temperature of the material flow and the ambient notes the sample welded by an assisted shoulder of 10 mm inner
diameter with a rotation speed of 800 rpm in clockwise direction,
temperature were assumed to be 300 K, as shown in Fig. 2a. In-
while the tool pin with sub-size concave shoulder rotates at
itially, the plastic material flows in and out of the inlet and outlet
600 rpm in counterclockwise direction, and the welding speed is
boundaries at the welding speed, respectively. The moving wall
120 mm/min. Table 4 lists the typical welding parameters used in
boundary condition was applied on the other boundaries away this study.
from the tool-workpiece contact interfaces. The velocity boundary
conditions at the tool-workpiece contact interfaces were ex- 3.1. Heat generation rate
pressed in terms of angular velocity of the tool, the dimensionless
slip rate and the radial distance from the tool axis [35]. Fig. 5 shows the calculated heat generation rate at the
Geometrically modeling the effect of threaded pin is time- shoulder-workpiece contact interfaces. It illustrates that the heat
consuming and may lead to large mesh distortion. For simplifica- generation rate increases with an increase of the radial distance
tion, the effect of the threaded pin on material flow is considered from the tool axis, which is consistent with the results of Nandan
as a boundary condition in this study. The thread leads to down- et al. [4]. However, the heat generation pattern at the tool shoulder
ward material flow velocity according to the rotation direction of is clearly non-axisymmetrical. For the points with the same dis-
the tool pin and the thread direction. Therefore, the velocity tance away from the tool axis, the heat generation rate in front of
boundary condition at the side surface of the tool pin can be ex- the tool is higher than that behind it. This is because the
interfacial temperature in front of the tool is lower than that
pressed as
behind it, which results in relatively higher interfacial contact
shear stress in front
2πω ⎫ of the tool and causes higher heat generation rate there.
2
u x2 = (1 − δ) r sin θ ⎪ Fig. 6 shows the variations of heat generation rate at the
60 ⎪ shoulder interfaces with angle θ for different radial distance from

u y2 ⎬ > 0; R pi ≤ r < R po ) the tool axis. Let θv and θp represent the angle θ when the valley
2πω 2
= (1 − δ) r cos θ (ω2
60 ⎪ and peak values of heat generation rate are reached, respectively.
2πω 2 ⎪
uz2 =−Ψ ⎪ Table 5 lists the values and locations of θv and θp under different
60 ⎭ (17) radial distance. It can be seen that with an increase of the radial
distance, the location of the maximum heat generation rate
moves
where Ψ is the thread pitch of the threaded pin. from the RS to the AS while the location of the minimum heat
The heat flux at the tool-workpiece contact interfaces are de- generation rate moves from the AS to the RS.
scribed by [22] Fig. 7 illustrates the viscous dissipation heat generation rate for
L. Shi et al. / International Journal of Machine Tools & Manufacture 91 (2015) 1–11 7

Fig. 8. Effect of (a) the welding speed, (b) the assisted shoulder rotation speed, (c) the tool pin rotation speed, and (d) the inner diameter of assisted shoulder on the total
heat generation, pin heat generation, and pin heat generation fraction.
sample 10-800-800-60. As shown in Fig. 7a, the volumetric heat
flux due to the viscous dissipation is high at the periphery of the
tool where the strain rate is high [15]. Fig. 7b demonstrates the
viscous dissipation heat generation rate at different depths (i.e., z
values) along the transverse cross-section. At the plane z ¼ 5 mm
(the top surface of workpiece), the heat generation rate first goes
up with an increase of the distance from the tool axis, and reaches
a relatively high value at the periphery of sub-size concave
shoulder. As the distance continues to increase, the heat genera-
3
tion rate drops to about 800 MW/m , and then it increases again
and reaches its peak value at the periphery of assisted shoulder.
Outside the action zone of the assisted shoulder, the strain rate
decreases to zero which results in a rapid decrease of the viscous
dissipation heat generation rate to zero. At planes with low z va-
lue, the strain rate is high at the periphery of the tool pin, which
results in relatively high viscous dissipation heat generation rate
near the pin edge. At the plane z ¼ 0 mm (bottom surface of
workpiece), the viscous dissipation heat generation rate increases
as the distance from the tool axis increases and reaches its peak
value at the periphery of the tool pin edge, and then it rapidly
decreases with an increase of the distance away from the tool
axis.
Fig. 8 presents the total heat generation, pin heat generation,
and pin heat generation fraction for various welding parameters
and tool size. As the welding speed increases, the total heat gen-
eration, pin heat generation and pin heat generation fraction rise
slightly (Fig. 8a), because the lowered interfacial temperature at
Fig. 9. Temperature field (a) on the top surface of the workpiece, (b) on the
transverse cross-section, (c) on the longitudinal cross-section (sample 10-800-
the tool-workpiece contact interfaces results in an increase of the
800- interfacial contact shear stress and plastic deformation heat gen-
180). eration. As the assisted shoulder rotation speed increases (Fig. 8b),
the total heat generation increases slightly, while the pin heat
generation and pin heat generation fraction decrease. This is
8 L. Shi et al. / International Journal of Machine Tools & Manufacture 91 (2015) 1–11
Fig. 11. Comparison of the measured and calculated thermal cycles. (a) The mon-
itoring locations on transverse cross-section, (b) sample 10-600-1200-200 and
(c) sample 10-800-1000-100.

Consequently, the pin heat generation and its fraction decrease.


Fig. 8c shows that the total heat generation, pin heat generation
and pin heat generation fraction increase slightly with an increase
of the tool pin rotation speed. Fig. 8d shows that there is no dif-
ference of the heat generation with an increase of the inner dia-
meter of assisted shoulder, which indicates the rotation direction
has little effects on the total heat generation. However, the dif-
ferent rotation direction changes the material flow mode and the
distribution of heat generation rate, which has great effect on the
temperature distribution, and leads to more homogeneous tem-
perature distribution and microstructure during the RDR-FSW
Fig. 10. Effect of (a) welding speed, (b) assisted shoulder rotating speed, and
(c) tool pin rotating speed on thermal cycle. (To make the thermal cycles of dif-
process [11,12,15,16].
ferent welding parameters clear to see, the thermal cycles of the red dotted line
and the blue dot dash line are moving to the right by added 100 s and 200 s to
3.2. Calculated temperature profiles
the time coordinate in the figure.). (For interpretation of the references to color in
this figure legend, the reader is referred to the web version of this article.)
Fig. 9 shows the calculated temperature distribution for sample
because the heat generation rate and interfacial temperature at 10-800-800-180. The temperature distributions on the top surface
Ω1 region rise with an increase of the assisted shoulder (Fig. 9a) and the longitudinal cross-section (Fig. 9c) of the work-
rotation speed, which results in higher temperature near the tool piece are compressed in front of the tool and are expanded behind
pin and
lower heat generation rate at both Ω3 and Ω4 regions. the tool. Asymmetrical temperature distribution is observed near
L. Shi et al. / International Journal of Machine Tools & Manufacture 91 (2015) 1–11 9

S’(S) precipitates. Fig. 10b and c show that the variation of rotation
speed affects slightly on the ta and tc . However, the Δt increases
from 46.77 s to 56.47 s as the rotation speed of the assisted
shoulder increases from 600 rpm to 1000 rpm (Fig. 10b).
Fig. 11 shows the comparison of the calculated thermal cycles
at three monitoring locations with the experimentally
measured ones for two cases. Fig. 11a shows the monitoring
locations on transverse cross-section. The thermocouples were
placed at 8 mm (A), 12 mm (B), and 20 mm (C) away from the
weld line on the RS, respectively. All the temperature
measurements were made at
2 mm depth below the top surface of the workpiece. Fig. 11b and
c compare the calculated and measured thermal cycles for
sample
10-600-1200-200 and 10-800-1000-100, respectively. It shows
good agreement between the measured and the calculated results
at different monitoring locations for both cases.

3.3. Calculated material flow

Fig. 12 illustrates the predicted material flow field near the tool
on horizontal planes (xy-plane) with different z-coordinates for
sample 10-800-800-120. The material flow on the top surface
(z ¼ 5 mm) is the strongest because it is contacted tightly with the
shoulders. The highest material flow velocity locates at the per-
iphery of the assisted shoulder. Two reverse material flows are
clearly demonstrated at the top surface due to the reverse rotation
of assisted shoulder and tool pin. This flow pattern is beneficial to
the uniformity of both the temperature and microstructure at AS
and RS [11,12,15,16]. It can be seen that the material flow
velocity and the flow region decrease as the distance from the top
surface of workpiece increases. This is because the action due
to the shoulders gets lowered with an increase of the depth.

3.4. Analysis of tool torque


The welding parameters and tool size in the RDR-FSW process
have decisive effect on the total torque which determines the
needs for clamping equipments and the lifespan of the tool [3,31].
Fig. 13 shows the total torque and the torque components from
different parts of the tool (i.e., the assisted shoulder, the sub-size
concave shoulder and the tool pin) under different welding
para- meters and tool size. As shown in Fig. 13a, the torque
Fig. 12. Calculated velocity field at different horizontal plane (sample 10-800-800-
120). components from different parts of the tool vary slightly with
the welding speed, but the total torque remains constant ( 0.9
the tool at the transverse cross-section (Fig. 9b). The temperature Nm). If the rotation speed of the tool pin or the assisted shoulder
at the AS is slightly higher than that at the RS. The high tem- rises, there is tiny variation of the total torque and its components
perature region is located at the AS behind the tool. This is (Fig. 13b, c). Fig. 13d demonstrates that as the inner diameter of
because the assisted shoulder preheats that region before the the assisted shoulder increases, the total torque and the torque
tool pin touches there, and then the heat generated by the tool components due to the assisted shoulder and the sub-size concave
pin con- tinues to heat that region. shoulder get larger, while the torque due to the tool pin
The partial dissolution of the Guinier–Preston–Bagaryatsky remains a constant value. For conventional FSW with a
(GPB) zones and solute clusters as well as the coarsening of condition of 800 rpm and
S’(S) precipitates lead to a deterioration of the microstructures and 120 mm/min, the experimentally measured torque was about
mechanical properties during FSW of 2024 aluminum alloys 13.75 Nm [8]. But in the RDR-FSW process with similar condition,
[17,18]. Fig. 10 shows the effects of welding parameters on the reverse rotation of the tool pin or the assisted shoulder lower
thermal cycles at a special point (location M in Fig. 11a). The the torque to an absolute magnitude of 1.0 Nm.
dwelling time of the temperature higher than 400 K is defined as Table 6 summarizes the effects of the process parameters on
Δt , since above the heat generation and tool torque in the RDR-FSW process. The
400 K the GBP zones and the solute clusters start to dissolve, total heat generation rises with increasing of the assisted shoulder
which has significant effects on the properties of the joint [17,18]. rotation speed, the tool pin rotation speed and the welding speed.
And ta is the duration time of the temperate increasing from 300 The inner diameter of assisted shoulder has little effect on the total
K to its peak value and tc is the duration time of the heat generation. An increase of the assisted shoulder rotation
temperature speed causes the rising of the total tool torque.
decreasing from its peak value to 300 K. Fig. 10a shows that the
values of Δt , ta and tc decrease with an increase of the welding 4. Conclusions
speed. As the welding speed increases from 60 mm/min to
180 mm/min, the Δt decreases from 90.44 s to 37.23 s (Fig. (1) A numerical model is developed to quantitatively analyze the
10a). The lower value of Δt corresponds to less dissolution of the heat generation, temperature profiles, material flow and the
GPB
zones and solute clusters as well as less coarsening of
10 L. Shi et al. / International Journal of Machine Tools & Manufacture 91 (2015) 1–11

Fig. 13. Effect of (a) the welding speed, (b) the assisted shoulder rotation speed, (c) the tool pin rotation speed, (d) the inner diameter of assisted shoulder on the total
torque and the torque components from different part of the tool (i.e., by the assisted shoulder, the sub-size concave shoulder and the tool pin).
Table 6
little effect on the heat generation. However, the different
Summaries of the effects of process parameters on heat generation and tool
torquen.
rota- tion direction changes the material flow pattern, which
has great effect on the temperature distribution, and
leads to more
min)
Items Di (mm) homogeneous temperature distribution in the RDR-FSW
ω1 (rpm) ω2 (rpm) U (mm/
Total heat generation ¼ ↑ ↑ ↑ process. (4) With similar conditions, the separated and the reverse
Pin heat generation ¼ ↓ ↑ ↑ rotate of the tool pin and the assisted shoulder lower the total net
Pin heat generation fraction ¼ ↓ ↑ ↑
torque exerted on the workpiece from 13.75 Nm in conventional
Total torque ↑ ↑ ↓ ¼
Torque by the assisted shoulder ↑ ↑ ↑ ↓
FSW
Torque by the tool pin ¼ ↓ ↓ ↑ to an absolute magnitude of 1.0 Nm in the RDR-FSW.

n
↑The item increases with an increase of the process parameter; ↓The item
Acknowledgments
decreases with an increase of the process parameter; ¼ The process parameter has
little effects on the item. This research was supported by the State Key Laboratory of
Advanced Welding and Joining at Harbin Institute of Technology in
total torque for different welding parameters and tool size in China (Grant no. AWJ-Z13-02), and the Sino-German Center for the
reverse dual-rotation friction stir welding (RDR-FSW) process. Promotion of Science (Grant no. GZ-739). The authors were
The calculated thermal cycles agree well with the corre- grateful to the Editor, the Associate Editor and anonymous re-
sponding measured data. viewers for their insightful comments which have helped to im-
(2) The heat generation rate at the shoulder-workpiece contact prove the quality of the paper.
interfaces is non-axisymmetrical. For the points with the same
distance away from the tool axis, the heat generation rate in
front of the tool is higher than that behind it. With an increase
of the radial distance, the locations of peak heat generation References
rate move from the retreating side (RS) to the advancing side
[1] P.L. Threadgill, A.J. Leonard, H.R. Shercliff, P.J. Withers, Friction stir welding of
(AS), while the locations of the minimum heat generation
aluminium alloys, Int. Mater. Rev. 54 (2) (20 09) 49–93.
value move from the AS to the RS. [2] R. Nandan, T. Debroy, H.K.D.H. Bhadeshia, Recent advances in friction-stir
(3) The reverse rotation of the tool pin and the assisted shoulder welding-process, weldment structure and properties, Prog. Mater. Sci. 53
has
L. Shi et al. / International Journal of Machine Tools & Manufacture 91 (2015) 1–11 11
(20 05) 980–1023. [20] M. Song, R. Kovacevic, Thermal modeling of friction stir welding in a moving
[3] R.S. Mishra, Z.Y. Ma, Friction stir welding and processing, Mater. Sci. Eng.: coordinate system and its validation, Int. J. Mach. Tools Manuf. 43 (6) (20 03)
R: Rep. 50 (1–2) (20 05) 1–78. 605–615.
[4] R. Nandan, G.G. Roy, T. Debroy, Numerical simulation of three-dimensional [21] C. Hamilton, S. Dymek, A. Sommers, A thermal model of friction stir welding in
heat transfer and plastic flow during friction stir welding, Metall. Mater. aluminum alloys, Int. J. Mach. Tools Manuf. 48 (10) (20 08) 1120–1130.
Trans. A 37 (4) (20 06) 1247–1259. [22] X.X. Zhang, B.L. Xiao, Z.Y. Ma, A transient thermal model for friction stir weld.
[5] H.H. Cho, S.T. Hong, J.H. Roh, H.S. Choi, S.H. Kang, R.J. Steel, H.N. Han, Three- part I: the model, Metall. Mater. Trans. A 42 (10) (2011) 3218–3228.
dimensional numerical and experimental investigation on friction stir welding [23] C. Hamilton, A. Sommers, S. Dymek, A thermal model of friction stir welding
processes of ferritic stainless steel, Acta Mater. 61 (7) (2013) 2649–2661. applied to Sc-modified Al–Zn–Mg–Cu alloy extrusions, Int. J. Mach. Tools
[6] D. Rao, K. Huber, J. Heerens, J.F. dos Santos, N. Huber, Asymmetric mechanical Manuf. 49 (3) (20 09) 230–238.
properties and tensile behaviour prediction of aluminium alloy 5083 friction [24] P.A. Colegrove, H.R. Shercliff, R. Zettler, Model for predicting heat generation
stir welding joints, Mater. Sci. Eng. A 565 (2013) 4 4–50. and temperature in friction stir welding from the material properties, Sci.
[7] M. Mehta, K. Chatterjee, A. De, Monitoring torque and traverse force in Technol. of Weld. Join. 12 (4) (20 07) 284–297.
friction stir welding from input electrical signatures of driving motors, Sci. [25] P. Ulysse, Three-dimensional modeling of the friction stir-welding process, Int.
Technol. Weld. Join. 18 (3) (2013) 191–197. J. Mach. Tools Manuf. 42 (20 02) 1549–1557.
[8] H. Su, C.S. Wu, A. Pittner, M. Rethmeier, Simultaneous measurement of tool [26] V. Soundararajan, S. Zekovic, R. Kovacevic, Thermo-mechanical model with
torque, traverse force and axial force in friction stir welding, J. Manuf. adaptive boundary conditions for friction stir welding of Al 6061, Int. J. Mach.
Process. Tools Manuf. 45 (20 05) 1577–1587.
15 (4) (2013) 495 –500. [27] P. Heurtier, M.J. Jones, C. Desrayaud, J.H. Driver, F. Montheillet, D. Allehaux,
[9] R. Rai, A. De, H.K.D.H. Bhadeshia, T. DebRoy, Review: friction stir welding tools, Mechanical and thermal modelling of Friction Stir Welding, J. Mater. Proces-
Sci. Technol. Weld. Join. 16 (4) (2011) 325–342. sing Technol. 171 (3) (20 06) 348–357.
[10] W.M. Thomas, I.M. Norris, D.G. Staines, E.R. Watts, Friction stir welding— [28] L. Fratini, G. Buffa, CDRX modelling in friction stir welding of aluminium al-
process developments and variant techniques, in The SME Summit, (2005) pp. loys, Int. J. Mach. Tools Manuf. 45 (20 05) 1188–1194.
1–4. [29] M. Assidi, L. Fourment, S. Guerdoux, T. Nelson, Friction model for friction stir
[11] J.Q. Li, H.J. Liu, Effects of welding speed on microstructures and mechanical welding process simulation: calibrations from welding experiments, Int. J.
properties of AA2219-T6 welded by the reverse dual-rotation friction stir Mach. Tools Manuf. 50 (2010) 143–155.
welding, Int. J. Adv. Manuf. Techno. 68 (9–12) (2013) 2071–2083. [30] K. Kuykendall, T. Nelson, C. Sorensen, On the selection of constitutive laws
[12] J.Q. Li, H.J. Liu, Characteristics of the reverse dual-rotation friction stir welding used in modeling friction stir welding, Int. J. Mach. Tools Manuf. 74 (2013)
conducted on 2219-T6 aluminum alloy, Mater. Des. 45 (2013) 148–154. 74–85.
[13] J.H. Yan, M.A. Sutton, A.P. Reynolds, Process–structure–property relationships [31] S. Cui, Z.W. Chen, J.D. Robson, A model relating tool torque and its associated
for nugget and heat affected zone regions of AA2524–T351 friction stir welds, power and specific energy to rotation and forward speeds during friction stir
Sci. Technol. Weld. Join. 10 (6) (20 05) 725–736. welding/processing, Int. J. Mach. Tools Manuf. 50 (2010) 1023–1030.
[14] T. Long, W. Tang, A.P. Reynolds, Process response parameter relationships in [32] T. Li, Q.Y. Shi, H.K. Li, Residual stresses simulation for friction stir welded joint,
aluminium alloy friction stir welds, Sci. Technol. Weld. Join. 12 (4) (20 07) Sci. Technol. Weld. Join. 12 (8) (20 07) 664–670.
311–317. [33] C.S. Wu, W.B. Zhang, L. Shi, M. Chen, Visualization and simulation of plastic
[15] L. Shi, C.S. Wu, H.J. Liu, Modeling the material flow and heat transfer in material flow in friction stir welding of 2024 aluminium alloy plates, Trans.
reverse dual-rotation friction stir welding, J. Mater. Eng. Perform. 23 (4) Nonferrous Metals Soc. China 22 (6) (2012) 14 45–1451.
(2014) [34] T. Sheppard, A. Jackson, Constituti ve equations for use in prediction of flow
2918–2929. stress during extrusion of aluminium alloys, Mater. Sci. Technol. 13 (3) (1997)
[16] L. Shi, C.S. Wu, H.J. Liu, Numerical analysis of heat generation and temperature 203–209.
field in reverse dual-rotation friction stir welding, Int. J. Adv. Manuf. Technol. [35] R. Nandan, G.G. Roy, T. DebRoy, Three-dimensional heat and material flow
74 (1–4) (2014) 319–334. during friction stir welding of mild steel, Acta Mater. 55 (3) (20 07) 883–895.
[17] C. Genevois, A. Deschamps, A. Denquin, B. Doisneau-Cottignies, Quantitative [36] A. Arora, A. De, T. DebRoy, Toward optimum friction stir welding tool shoulder
investigation of precipitation and mechanical behaviour for AA2024 friction diameter, Scripta Mater. 64 (1) (2011) 9–12.
stir welds, Acta Mater. 53 (8) (20 05) 2447–2458. [37] Q. Li, M. Lovell, On the critical interfacial friction of a two-roll CWR process, J.
[18] Z. Zhang, B.L. Xiao, Z.Y. Ma, Hardness recovery mechanism in the heat- Mater. Process, Technol. 160 (2) (20 05) 245–256.
affected zone during long-term natural aging and its influence on the
mechanical properties and fracture behavior of friction stir welded 2024Al–
T351 joints, Acta Mater. 73 (2014) 227–239.
[19] X. He, F. Gu, A. Ball, A review of numerical analysis of friction stir welding,
Prog. Mater. Sci. 65 (2014) 1–66.

You might also like