Journal of the Chinese Institute of Engineers, Vol. 31, No. 5, pp.

757-767 (2008) 757

FRICTION STIR WELDING OF AL 6061-T6 THICK PLATES:

PART I – EXPERIMENTAL ANALYSES OF THERMAL
AND MECHANICAL PHENOMENA

Yuh J. Chao, Shu Liu and Chi-Hui Chien *

ABSTRACT
In friction stir welding (FSW), heat is generated by two mechanisms: (1) the
friction between the rotating tool and the workpiece at the interface, and (2) the plas-
tic shear deformation of the material in the vicinity of the pin tool. The amount of
heat generated is crucial to the quality of the weld. In addition, the mechanical force
and torque developed in the process dictates the life of the tool as well as the design
of the fixture. In this paper, a butt joint was made by FSW on two thick plates (thickness
= 37.6 mm) of aluminum alloy 6061-T6. The joining was conducted by load control
mode (applying a controlled vertical plunging force). Two different tests were
performed, one using a long probe pin (24.3 mm in length) and the other using a very
short pin (3.5 mm in length). The objectives of the research were to study the heat
generation process from the shoulder, the pin, and the mechanical process parameters.
In this Part I, we report on the welding process parameters, i.e. tool RPM, weld-
ing speed, tool plunge-in depth, force, torque, and power history, and the transient
temperature history measured from thermocouples. It was found that (a) a quasi-
steady state existed after the tool moved about 82.2 mm from the starting point, dur-
ing which the tool forces, torque and generated heat reached an equilibrium state in
the neighborhood of the tool, (b) the highest temperature measured in the weld seam
was less than the melting temperature of the workpiece material, (c) the tool pin plays
an important role in the heat input to the process, and (d) the heat generated from the
pin strongly depends on its length. The comprehensive test data could be used by
researchers interested in numerical modeling of the FSW process without duplicating
the complicated tests.

Key Words: friction stir welding, thick plates, transient temperature, pin length.

I. INTRODUCTION of the most promising solid-state joining processes

in recent years.
Friction Stir Welding (FSW) is a relatively new FSW is a complicated process involving com-
material joining technique patented by the Welding bined thermal and mechanical actions. In the ther-
Institute (TWI) in 1991. Due to the superior mechani- mal part, heat is generated by friction between the
cal properties of the resulting joint relative to a con- rotating tool and the workpiece as well as the plastic
ventional fusion weld, FSW has been regarded as one shearing of materials. Sufficient heat is required to
warm and soften the workpiece material and then
*Corresponding author. (Tel: 886-7-5254223; Fax: 886-7- ensure smooth tool movement. Yet, heat input can-
5254299; Email: [email protected]) not be so high that it melts the material leading to
Y. J. Chao and S. Liu are with the Department of Mechanical defective welds. During the process mechanical
Engineering, University of South Carolina, USA.
C. H. Chien is with the Department of Mechanical and Electro-
forces are applied by the welding machine through
Mechanical Engineering, National Sun Yat-Sen University, Kaohsiung the tool to the workpiece to “stir” the material and to
80424, Taiwan. form the joint. Process parameters such as force,
758 Journal of the Chinese Institute of Engineers, Vol. 31, No. 5 (2008)

torque, power input, and plunge-in depth are impor-

tant in the ultimate quality of the weld and they also
dictate the service life and potential failure of the tool.
After welding, residual stresses in the weldment and
distortions of the workpiece are inherent due to the
heating/cooling and mechanical deformation during
welding.
About previous experimental studies in FSW,
measurement of the temperature history in FSW for a
butt-welded workpiece was first performed by McClure
et al. (1998) and Tang et al. (1998) and later by Khankar
et al. (2002), Chao et al. (2003), Song and Kovacevic
(2002), etc. Stewart et al. (1998) used dynamom-
eters on a milling machine to measure the history of
forces from the welding tool. Cook et al. (2003), Smith
(2000) and Strombeck et al. (2000) monitored the force Fig. 1 FSW PDS (Friction Stir Welding Process Development
System)
during the welding process and found that feedback
control is the key to providing an automated control
in robotic FSW. Most recently, Chen et al. (2003),
Jandric et al. (2003), Smith et al. (2003) and Packer heat treatable aluminum alloys), without voids and
et al. (2003) measured force components and tran- no tool damage, etc. In our current experiments,
sient temperatures in a displacement-controlled FSW. however, both pin lengths are much less than the plate
In summary, while it is extremely valuable to thickness, i.e. 24.3 mm and 3.5 mm vs. 37.6 mm.
experimentally investigate both thermal and mechani- These shorter pins were purposely selected so that a
cal phenomena in FSW, there are no comprehensive proper heat distribution model could be developed to
experimental results reported in open literature in- study the relative or percentage contributions of the
cluding both process parameters and temperature his- heat generated from the shoulder and the pin.
tory of the process. Most experimental data focused Furthermore, welding parameters were carefully se-
solely on temperature history, while a few research- lected so that no voids, or tool damage occurred and
ers reported incomplete thermal and mechanical data. good visible welds were made in these experiments.
A lack of a completely instrumented FSW system Since it would be too lengthy to include both experi-
might be the reason. mental results and numerical analysis in one paper,
In the current paper, we report the test and ex- the experimental results are reported in the current
perimental results, both thermal and mechanical, from paper and the numerical modeling/results are pre-
a well-controlled FSW process. A butt joint was made sented in Part-II (Chao et al, in press).
by FSW with two thick plates (thickness = 37.6 mm)
of aluminum alloy 6061-T6. Experiments were car- II. EXPERIMENTAL SET-UP AND
ried out using an FSW PDS (Friction Stir Welding PROCEDURE
Process Development System) equipped with sensors
and controllers for achieving a vertical force-controlled FSW was performed using a FSW PDS (Friction
process and the recording of process parameters. Stir Welding Process Development System) designed
Transient temperatures were measured by 25 embed- by MTS Systems Corporation. The FSW PDS (Fig.
ded thermocouples in the workpiece. Two different 1) is a robust system composed of a weld head as-
welds were made in this study. One used a rotation sembly and manipulator, as well as a digital control
tool with a long cylindrical probe pin (24.3 mm in system that has the capability for either force or dis-
length) and one used a very short cylindrical pin (3.5 placement control of the process. The FSW PDS is
mm in length). They are designated as the long-pin instrumented with high-resolution digital MTS
weld and the short-pin weld respectively throughout Temposonic linear transducers to measure displacement.
the paper. Vertical plunge force is measured using a custom MTS
Note that in a typical FSW, the length of the pin load cell. Traverse load is measured using precision
is slightly less than the thickness of the workpiece pressure transducers. The rotational speed of the pin
(McClure et al., 1998; Tang et al., 1998; Khankar et tool is measured using a precision tachometer. Torque
al., 2002; and Chao et al., 2003) in order to achieve a on the spindle is measured using pressure transduc-
good quality joint. “Good” means that the weld has ers mounted on the inlet and outlet of the motor.
the expected quality from an FSW process, such as a During the FSW process, the two flat plates were
tensile strength higher than 2/3 of the base metal (for clamped tightly along the edges to a thick backing
 Y. J. Chao et al.: Friction Stir Welding of Al 6061-T6 Thick Plates: Part I – Experimental Analyses 759

(a) Dimensions of the workpiece and 3D layout of thermocouples

Fig. 2 Close-up view of the FSW process

plate made of stainless steel and restrained to the sup-

port of the system (Fig. 2). The distance between
any two clamps in the X direction was about 23 mm.
For current experiments, constant plunge (downward)
force (F Z ) was used as the control to maintain weld
stability and constant pressure on the tool. Other pro- (b) Dimensions of the workpiece (YZ plane) and 2D layout of
cess parameters, such as traverse forces (F X), travel thermocouples (unit: mm)
speed, tool spindle speed, torque (T), and tool dis-
placement were monitored and recorded by the
system. Meanwhile, transient temperatures were mea-
sured by 25 thermocouples embedded in the
workpiece
Note that two tests, one with the pin and one (c) Dimensions of the workpiece (XY plane) and 2D layout of
without the pin, were initially designed to study the thermocouples (unit: mm)
contribution of the heat input from the pin and the
shoulder. However, it was soon realized from trial
tests that the tool wobbled during the welding due to
the lateral flexibility of the spindle head assembly
(d) Dimensions of the workpiece (XZ plane) and 2D layout of
when no pin was used. Therefore, a short pin, 3.5 thermocouples (unit: mm)
mm long, which was in stock, was chosen to approxi-
mate the no-pin condition. Fig. 3 Dimensions of the workpiece and layout of thermocouples
The dimensions of one workpiece (Length × Width
× Thickness = L × W × H = 609.6 × 76.5 × 37.6 mm =
24 × 3.01 × 1.48 inches) and locations of the thermo- the centerline of the workpiece. Each hole was filled
couples (circular symbols) are shown in Fig. 3(a). In with one gauge-36 K-type thermocouple beaded at the
the welding, the tool started at (X = 17.8 mm, Y = 0 tip and stuck at the end of the hole with high thermal
mm) and finished near the end of the workpiece. The conductivity Omegabond epoxies (OB-200) or chemi-
tools were operated at 350 rpm rotational speed and cal set cements (OB-600 powder) as labeled, C1-C25,
2.5 mm/sec linear speed in both cases. in Figs. 3 (a)~(d). The location of each thermocouple
Note that previously in our lab, we have suc- can be found from Figs. 3(b) through 3(d) where the
cessfully accomplished “nominal weld” for Al 6061- thermocouples are depicted as circular symbols. As
T6 with linear welding speed = 2.5 mm/sec, rotational shown in the figures, thermocouples were positioned
speed = 350 rpm, load-control mode of 32 KN z-force at different distances from the weld centerline and at
for workpiece of 25.4 mm thick. Therefore, in our different depths in the Z or the thickness direction.
current long-pin welding experiment, these same Obviously, the layout of the thermocouples was used
welding speeds and z-force were used again. to facilitate the measurement of the temperature field
Twenty-five holes in the workpiece were pre- inside/outside the weld tool “stirring” zone.
drilled using long drill-bits from the opposite side to Thermocouples C10 and C18 have the same Y
760 Journal of the Chinese Institute of Engineers, Vol. 31, No. 5 (2008)

and Z coordinates and therefore are shown at the same is zero degree, i.e. the tool is operated normal to the
location in Fig. 3(b). The same is true for C11 and plate.
C19, C12 and C20, C13 and C21, C14 and C22, C15
and C23, C16 and C24, C17 and C25. To compare III. EXPERIMENTAL RESULTS AND
the effects of different glues on the measured DISCUSSIONS
temperature, cements (epoxies) are used to bond
C10~C17 (C18~C25) to the workpiece. Note that the 1. History of the Welding Process Parameters
maximum service temperature of the cement (epoxy)
is 1426°C = 2600°F (260°C = 500°F) from the catalog. Time histories of the welding process parameters
Cements are applied to C1~C9 that are close to the are presented in Fig. 4 (Fig. 5) for the long-pin weld
weld centerline because the anticipated temperature (short-pin weld).
there is higher than 260°C during welding. Figure 4(a) shows the time history of the longi-
All thermocouples were connected to a DC-pow- tudinal position (X) of the tool. Prior to 135 sec (X =
ered chassis (Ni SCXI-100DC), and then to a data 17.8 mm), the rotating tool was plunging into the
acquisition board in a computer. The transient tem- workpiece. After 141.2 sec, the tool began to move
peratures from thermocouples were digitally acquired at a constant speed of 2.54 mm/sec. At 356.8 sec (or
using National Instruments Labview program at about X = 559.0 mm), the tool started to withdraw from the
10Hz sampling rate. Note that the FSW process is workpiece and the weld process was completed. The
asymmetrical (Arbegast and Hartley, 1998; Shi et al., whole FSW process, therefore, can be divided into
2003; Lindner et al., 2003; Deng and Xu, 2004), and three periods, i.e. (I) tool plunge-in period (0 ~ 141.2
the thermal profile of the workpiece is also somewhat seconds), (II) welding period (141.2 ~ 356.8 seconds),
asymmetrical (Arbegast and Hartley, 1998; Shi et al., and (III) tool retreating period (356.8 ~ 368.8
2003; Khandkar and Khan 2001). However, the ther- seconds).
mocouples are only placed on the advancing side of Figure 4 (b) shows the history of the plunge or
the workpiece in our current experiments. The rea- tool downward force FZ. As the rotating tool plunged
sons are (1) the number of channels on the data ac- into the workpiece, FZ gradually augmented while the
quisition board for thermocouples is limited to only tool drilled a hole in the workpiece forming a region
25; (2) even though the temperature profile in the of severely deformed and softened material. The first
workpiece may be asymmetric, research has shown spike shown in Fig. 4(b) is apparently caused by overly
the advancing side is normally hotter than the retreating fast plummet of the tool into the workpiece. When
side (Arbegast and Hartley, 1998; Shi et al., 2003; the pitching speed of the tool was found to be excessive,
Khandkar and Khan 2001). And, the temperature it was adjusted to lower speed and F Z became lower
difference between the two sides was found to be small accordingly. When the shoulder of the tool was in
and non-significant (Tang et al., 1998; Shi et al., 2003; contact with the top surface of the workpiece, it caused
Khandkar and Khan 2001). In addition, in many a spike in F Z , marked as “the second spike” in Fig.
publications (McClure et al., 1998; Chao et al., 2003; 4(b). After that, more heat was generated due to friction
Song and Kovacevic, 2002; Chen and Kovacevic 2003) at the shoulder/workpiece interface. The workpiece
the thermal profile is assumed to be symmetric in order material then became softer and reduced FZ slightly.
to simplify the temperature measurement and numerical Shortly after the contact, the tool started to move in
modeling. For all these reasons, only one side the X direction while F Z was maintained constant.
(advancing side) of the workpiece was chosen for our In Fig. 4(b), the line represents system controlled
temperature measurement. FZ, and square-shaped symbols are the feedback data.
The pin tool was made of heat treated high car- They match quite well during the welding period
bon steel. Two shaded areas in Fig. 3(b) represent (141.2 ~ 356.8 seconds). The fluctuation of FZ dur-
one-half of the cross-section of the shoulder and the ing the welding period is less than 1 KN, which is
pin. The radius of the shoulder (pin) was r 0 = 16.5 about 3.1% of the average FZ (= 32.0 KN). This small
mm (r i = 6.4 mm) and the length of the long (short) variation in FZ indicates that the force control of the
pin was h1 = 24.4 mm (h2 = 3.5 mm). The shoulder is system is quite successful. The fluctuation in FZ could
designed as ascroll shoulder (Dawes and Thomas, 1999) be due to the formation of material flash and the damping
and the thread on the bottom of the shoulder is 1.5 effect of the highly plastically deformed weld zone.
mm/thread. There are also threads on the cylindrical Figure 4(c) depicts the history of F X , which
pin with a pitch 1.0 mm/thread. The function of the gradually increased as the tool moved forward before
shoulder is to add pressure to the weld material and it stabilized at 174 seconds. Comparing F Z and F X,
restrain it from flashing out of the surface (Dawes one can see that the average of F X (2.3 KN) is less
and Thomas, 1999) during welding. The head angle than 7.2% of F Z (32.0 KN) and FX also reveals oscil-
or lag angle between the pin tool and the workpiece lation with a small amplitude of about 0.6 KN. The
 Y. J. Chao et al.: Friction Stir Welding of Al 6061-T6 Thick Plates: Part I – Experimental Analyses 761

40
600 Shoulder
contact
First
500 I II III
30 spike
X Position (mm)

Z force (kN)
400 Quasi-steady state

300 20
Stable period of FZ
200
10 Z Force Feedback
100
Z Force Command
0 0
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
Welding time (s) Welding time (s)
(a) History of tool longitudinal position (b) History of Z force (FZ)
3.0 30
Shoulder
contact
2.5 25

Plunge-in depth (mm)

X force (KN)

2.0 20
Quasi-steady state
1.5 15
Quasi-steady state
1.0 10 Stable period of DP
0.5 Stable period of Fx 5

0.0 0
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
Welding time (s) Welding time (s)
(c) History of X force (FX) (d) History of tool plunge-in depth (DP)
350 12000
Shoulder Shoulder
300 contact 10000 contact
Torque (KN*mm)

250
8000
Power (W)

200 Quasi-steady state Quasi-steady state

6000
150
100 Stable period of MT 4000 Stable period of PT

50 2000
0 0
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
Welding time (s) Welding time (s)
(e) History of total spindle torque (MT) (f) History of total system power (PT)

Fig. 4 History of the welding process parameters (long-pin weld)

fluctuation of the force in the longitudinal direction Figure 4(e) (Fig. 4(f)) illustrates the time his-
F X could be due to (1) the discrete linear movement tory of the total spindle torque MT (total system power
of the tool; (2) flash formation in the plate during the PT ), where M T is related to P T by
welding, and/or (3) the flexibility of the tool.
Figure 4(d) shows the time history of the tool PT = M T ⋅ 2πω , (1)
60
plunge-in depth D P . It is shown that after the tool
was forced into the workpiece, D P reached a peak where ω is the rotational speed of the tool in rpm.
value and then deceased slightly as the tool Both figures show the same trend.
progressed. The data implies that the applied load Note that the mechanical and heat transfer pro-
control mode also properly kept the plunge-in depth cess in the entire weld process is transient and never
to a nearly constant value that is important to a good in steady state due to the variation of the welding
quality weld. The minimum of the nearly constant parameters with time and X position. However, in
value of DP is 24.38 mm, which is slightly longer than the welding period 174 ~350 seconds (or 100 < X <
the pin length of 24.3 mm. It means that the shoul- 547.1 mm), all welding parameters are nearly
der of the tool is pushed down into the workpiece a constant. This period can be defined as the quasi-
small amount. A small amount of material was there- steady state of the FSW process, i.e. during which
fore flashed out along the weld line. the force, torque and heat reached an equilibrium
762 Journal of the Chinese Institute of Engineers, Vol. 31, No. 5 (2008)

Shoulder contact
600 30

500 I II III 25
X Position (mm)

20 Quasi-steady state

Z force (KN)
400

300 15 Stable period of FZ

200 10
Z Force Feedback
100 5
Z Force Command
0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Welding time (s) Welding time (s)
(a) History of tool longitudinal position (b) History of Z force (FZ)
2 5 Shoulder contact

Plunge-in depth (mm)

4
1.5 Quasi-steady state
X force (KN)

3 Quasi-steady state
1 Stable period of Fx
2
Stable period of DP
0.5
1

0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Welding time (s) Welding time (s)
(c) History of X force (FX) (d) History of tool plunge-in depth (DP)
300 300
Shoulder contact Shoulder contact
250 250
Torque (KN*mm)

Torque (KN*mm)

200 200
150 150

100 Quasi-steady state 100 Quasi-steady state

50 Stable period of MT 50 Stable period of MT

0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Welding time (s) Welding time (s)
(e) History of total spindle torque (MT) (f) History of total spindle torque (MT)

Fig. 5 History of the welding process parameters (short-pin weld)

state, except for a small oscillation, in the neighbor- Fig. 5 shows that the later has less mechanical and
hood of the tool. In this quasi-steady state, similar heat input. Note that identical F Z for both welds was
thermal profile and history at each cross-section of initially planned. However, if the FZ for the long-pin
the workpiece are anticipated. The temperature pro- weld was applied to welding with the short-pin, tre-
files during this state are shown later in Sections III. mendous material would have flashed out during the
2 and III.3. The quasi-steady state is the most im- process and would guarantee an unsuccessful weld.
portant period in the FSW process. Proper design Therefore, after several trial runs, the maximum value
and control of welding parameters in this period dic- of F Z (= 27KN) without substantial amount of metal
tate a consistent welding process and weld quality. flash was identified and chosen for the short pin FSW.
Similar to Fig. 4, Fig. 5 illustrates the history of The reason why a “colder weld” for short-pin weld
process parameters in the short-pin weld. In Fig. 5, caused “flash” while a “hotter weld” for long pin weld
the three periods are (I) tool plunge-in period (0 ~ 50 did not is difficult to explain. One possible reason
seconds), (II) welding period (50 ~ 260 seconds), and might be that the material flows from short-pin weld
(III) tool retreating period (260 ~ 272.8 seconds). The and long-pin weld are significantly different. The func-
quasi- steady state is within 70 ~ 258 seconds (or tion of the shoulder is known to add pressure to the
48.7mm < X < 524mm). Comparison of Fig. 4 and weld material and restrain it from flashing out of the
 Y. J. Chao et al.: Friction Stir Welding of Al 6061-T6 Thick Plates: Part I – Experimental Analyses 763

Table 1 Torque and Heat Input of the FSW process

Long-pin weld Short-pin weld
Total spindle torque (M T ), KN*mm 284.42 183.87
Total system power (PT ), Watt 10424.7 6739.2
Free rotation torque (M FR),KN*mm 36.65 36.65
Machine efficiency ( η) 0.8882 0.8627
Welding torque (M W), KN*mm 220.07 127.01
Welding power (PW), Watt 8066.0 4655.0

surface. The length of the pin probably also helped to thermocouples although it might have changed their
restrain the material flashing by altering the material positions slightly. The tool/thermocouple interaction
flow. Further research may be needed to answer this introduced oscillations in the temperature data, as shown
interesting issue. in the magnified view in the upper right corner of Fig.
In Figs. 4 (e) and 5 (e), M T includes three 6(a). Temperatures from C6~C9 that are very close to
components: (1) torque due to friction of the spindle the pin also showed small fluctuations and it is likely
(M FR) itself, which is equal to the machine’s free ro- due to the material plastic flow. The fact that all data
tation torque, (2) torque converting to heat transferred from other thermocouples are very smooth indicates
to motor (M MOT), and (3) torque (M W) converting to that there is no effect from the interaction of the tool/
heat in the workpiece/tool. Component (1) can be thermocouples or material-flow/thermocouples.
obtained by reading MT in free rotation of the spindle, Note that the maximum temperature from C10
components (2) and (3) can be related by the motor in Fig. 6(b) is 390°C which exceeds the recommended
efficiency ( η ): M MOT = ( η
1 – 1) . M . The average
W service temperature of the epoxy glue, i.e. 260°C.
value of M W at the quasi-steady state can therefore However, the smoothness of the temperature reading
be determined as suggests that (1) either the epoxy glue might still func-
tion well during the process without debonding; or
M W = (M T – M FR) . η . (2) (2) the hole collapsed against the thermocouples due
to the material flow during welding.
In addition, P W at the quasi-steady state is In addition, Fig. 6 (a) demonstrates that the peak
temperature closer to the weld centerline is higher.
PW = M W ⋅ 2πω . (3) The maximum peak temperature from C3 is 578°C
60
which is slightly lower than the material solidus point
The torque and power inputs (M W and P W ) to (582°C).
the FSW for both welds, as well as the average total The temperature histories from C18~C25 (Fig. 6
spindle torque(M T), average total system power (P T ) (c)) are very close to those from C10~C17 (Fig. 6(b))
during the quasi-steady state, machine’s free rotation and the only major difference is the welding time which
power (P FR), and machine efficiency (η ) are listed in is due to different X locations of thermocouples. Since
Table 1. Obviously, the welding toque and power C18~C25 are bonded by cements, while C10~C17 are
input in long-pin weld is larger than that of the short- glued by epoxies, and C10/C18, C11/C19, C12/C20,
pin weld. C13/C21, C14/C22, C15/C23, C16/C24, C17/C25,
respectively, are located at the same distance from the
2. Transient Temperature History weld seam (in the thickness and width directions), and,
since both groups of thermocouples are within the quasi-
The transient temperatures recorded from the steady state region, i.e. from 100 < X < 547.1 mm,
thermocouples are shown in Figs. 6 (long-pin weld) similar temperature histories are obtained. Or, in other
and 7 (short-pin weld) in which each curve corre- words, in quasi-steady state the temperature profiles
sponds to a specific thermocouple as indicated. near the tool become steady state as well. Since the
As shown in Fig. 6, temperature histories gener- material flow path, microstructure and texture distri-
ally are continuous during the process. Thermocouple bution in the weld are strongly dependent on the welding
C3 is located exactly on the path of the moving pin, process parameters (force, torque and total heat) and
and C1, C2 and C4 are on the peripheries of the pin. the temperature profiles, similar mechanical proper-
The tool was expected to pass through these ties at each cross-section of the workpiece in this quasi-
thermocouples. From the temperature readings, it ap- steady state period can thus be anticipated (Tang et
pears that the passing of the tool did not destroy the al., 1998; Chao et al., 2003; Lindner et al., 2003; Park
764 Journal of the Chinese Institute of Engineers, Vol. 31, No. 5 (2008)

600 600 500

C3 500
C3 500 C3
C2 C2 450 C3 400
500 C4
400 C8 C1 C9 C2
C4 400 300
Temperature (°C)

C1 300 C9 C8
400 C8 C6 350 C1

Temperature (°C)
C9 200
C7 300 C9 C2 200 C4
C7
300 C6 100 C8 100
C5 250
0 C1 0 C6 C7 C5
200 300 320 340 360 200 C4
240 260 280 300
150 C6
100 C7
C5 100
C5
0 50
200 300 400 500 600 700 800 0
Welding time (s) 100 200 300 400 500 600 700 800
(a) Temperature history from thermocouples C1-C9 Welding time (s)
(glued by cements) – long-pin weld (a) Temperature history from thermocouples C1-C9
600 (glued by cements) – short-pin weld
400
C10
C11 500 300
500 300 C12 C10
250
C14 C11
Temperature (°C)

C10 200 400 200 C12

400 C16 C15 C14

Temperature (°C)
C17 C13 150
300 C11 100 300 100 C13
C12 C16 C10 50 C17 C15
C14 C17 0 C16
200 C15 240 260 280 300 200 0
C11 180 190 200 210 220 230 240
C13 C14 C12
100
100
C15 C16
0 C13 C17
100 200 300 400 500 600 700 800 0
Welding time (s) 100 200 300 400 500 600 700 800
(b) Temperature history from thermocouples C10-C17 Welding time (s)
(glued by epoxies) – long-pin weld (b) Temperature history from thermocouples C10-C17
400 (glued by epoxies) – short-pin weld
600 C18 500 300
C19 C18
300 C22 C19
500 C20 C20
200 400 200 C22
Temperature (°C)

Temperature (°C)

C18
400 C23 C21 C24 C25
100 300 C18 100
300 C19 C21 C24 C25
C20 0 C19 C23
C22 180 200 220 240 200 C23
200 C20 0
C23
C21 C22 110 120 130 140 150 160 170
C21 100
100 C24 C25
C25
C24
0 0
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
Welding time (s) Welding time (s)
(c) Temperature history from thermocouples C18-C25 (c) Temperature history from thermocouples C18-C25
(glued by cements) – long-pin weld (glued by cements) – short-pin weld

Fig. 6 Temperature history from thermocouples – long-pin weld Fig. 7 Temperature history from thermocouples – short-pin weld

et al., 2003; Allehaux et al., 2003; and Fonda et al., or/and the variation of thickness of glue. Nevertheless,
2004). Proper control of welding parameters in quasi- the experimental data proved that both glues give similar
steady state will clearly result in a consistent weld reliable results for temperatures close to or below 400°C.
quality. Figure 7 shows the temperature history obtained
Note on Fig. 6, the peak temperature from C18 by thermocouples for the short pin weld. The tem-
is a little higher than that from C10; the peak tem- perature history shows the same trend as that of Fig.
perature from C19 (C20) is a little less than that from 6. Similarly, it is found that the temperature history
C11 (C12); while the temperature profiles from all from thermocouples C18~C25 (Fig. 7(c)) are very close
other thermocouples are very close. The small dif- to those from thermocouples C10~C17 (Fig. 7(b)) in
ference between the two could be caused by either terms of shape and magnitude. The only major dif-
the small fluctuation of welding process parameters ference between them is the welding times. In addition,
 Y. J. Chao et al.: Friction Stir Welding of Al 6061-T6 Thick Plates: Part I – Experimental Analyses 765

the distinctions between Fig. 7 and Fig. 6 are (a) firstly,

the peak temperature in the short pin weld is less than
that in the long-pin weld, which is likely due to less
power input in the short-pin case, and (b) secondly,
the temperature history from all thermocouples (except
C3) is smooth in Fig. 7 because they are all far away
from the pin and therefore there is no pin-thermo-
couple interaction. Note there is no contact between
the probe pin and thermocouple C3, however, since
C3 is about 1.22 mm distance (shown in Fig. 7(a))
away from the pin in the thickness direction. The
oscillation of the data in C3 is very likely due to the (a) Peak temperature (data in parentheses, in °C) in the long-pin weld
plastic flow of the material.

3. Peak Temperature in the Cross-Section and

Influence of the Length of the Pin

Peak temperature in the workpiece during FSW

is a key element that could dictate the formation of
material microstructure and consequently the local
mechanical properties of the welds. Figs. 8(a) and
8(b) show the peak temperature distributions in the
cross-section of the workpiece for the two welds.
It is shown, in Fig. 8, that the maximum tem-
perature is at the weld centerline and at the interface (b) Peak temperature (data in parentheses, in °C) in the short-pin weld
between the pin and the workpiece. The temperature
Fig. 8 Peak temperature in the weld
gradually decreases with increasing distance from the
centerline of the workpiece. Also, it is shown that
except at location C3, the gradient of the peak tem-
perature distributions in both thickness and width di- in the short pin case the lower end of the pin is closer
rection away from the pin are small. It suggests that to the shoulder/workpiece interface which is the pri-
grain size, material microstructure and mechanical mary heat source in the process and therefore a higher
behaviors in this area will not exhibit abrupt changes. temperature is expected.
Furthermore, it is found from Fig. 8 that when Numerical simulation and heat models based on
the pin is longer the isothermal region surrounding the the experimental results reported herein are presented
pin with elevated temperature is larger (shown in Figs. in Part-II of this study (Chao et al, in press) to fur-
8(a) and 8(b)) possibly due to more material in the ther study the process.
workpiece being plastically deformed and the resis-
tance of the material generating more internal heat in IV. CONCLUSIONS
the workpiece. It appears that in addition to the heat
generated from the friction at the interface of the tool Several important findings and derivatives, as
shoulder and the workpiece, there is considerable heat stated in the following, can be drawn from these care-
generated from the pin to the workpiece. The relative fully designed FSW experiments.
contribution of heat generated from the pin to the to-
tal heat generated in the process strongly depends on • There exists a quasi-steady state in the force-con-
the length of the pin. In addition, the rotation of the trolled FSW process, during which mechanical force,
pin facilitates the material plastic flow inside the heat input and temperature profiles remain nearly
workpiece, which assists the heat conduction and identical in the neighborhood of the tool. Consis-
convection, and thus there was more uniform distri- tent weld quality can therefore be anticipated dur-
bution of temperature in the thickness direction around ing the quasi-steady state period.
the pin. Note that from Fig. 8 the peak temperature • Under the welding conditions used in the experiments,
right below the long pin is 408°C, while the peak tem- maximum temperature in the workpiece is less than,
perature under the short pin is higher (500°C). The yet close to the melting temperature of the workpiece
difference may be from the difference in metal flow material.
path between the short pin and the long pin that would • It is conceivable that the relative amount of heat
influence the amount of heat generated. In addition, generated from the interface of the pin and the
766 Journal of the Chinese Institute of Engineers, Vol. 31, No. 5 (2008)

workpiece to the total heat in the process strongly of Engineers, Vol. 31, No. 5, pp. 769-779.
depends on the pin length. Chao, Y. J., Qi, X., and Tang, W., 2003, “Heat Trans-
• Thermocouples glued by both Omegabond epoxies fer in Friction Stir Welding – Experimental and
(OB-200) and chemical set cements (OB-600 Numerical Studies,” ASME Journal of Manufac-
powder) gave similar results for temperatures close turing Science and Engineering, Vol. 125, No. 1,
to or below 400°C. pp. 138-145.
Chen, C. M., Jandric, D., and Kovacevic, R., 2003,
ACKNOWLEDGEMENTS “Mechanical Force Analysis and Microstructure
Characterization in Friction Stir Welding of Lap
The authors are indebted to Mr. Daniel Wilhelm and Butt Welds,” 13 th International Conference
and Dr. Wei Tang of the University of South Caro- on Computer Technology in Welding, Grosvenor
lina for assistance in FSW operation and temperature Resort, Orlando, Florida, USA, in CDROM.
measurement. Chen, C. M., and Kovacevic, R., 2003, “Finite Element
Modeling of Friction Stir Welding –Thermal and
NOMENCLATURE Thermomechanical Analysis,” International Journal
of Machine Tools and Manufacture, Vol. 43, No.
DP plunge-in depth; mm 13, pp. 1319-1326.
FX traverse force; KN Cook, G. E,. Smartt, H. B., Nitchell, J. E., Strauss,
FZ downward force; KN A. M., and Crawford, R., 2003, “Controlling Ro-
H thickness of workpiece; mm botic Friction Stir Welding,” The Welding Journal,
h1 length of the long pin; mm pp. 28-34.
h2 length of the short pin; mm Dawes, C., and Thomas, W., 1999, “Development of
L length of workpiece; mm Improved Tool Designs for Friction Stir Weld-
M FR free rotation torque; KN*mm ing of Aluminum,” 1st International Symposium
M MOT torque converting to heat transferred to motor; on Friction Stir Welding, Thousand Oaks, California,
KN*mm USA, in CDROM.
MT total spindle torque; KN*mm Deng, X., and Xu, S., 2004, “Two-dimensional Fi-
MW welding torque; KN*mm nite Element Simulation of Material Flow in the
P FR free rotation power; Watt Friction Stir Welding Process,” Journal of Manu-
PT total system power; Watt facturing Processes, Vol. 6, No. 2, pp. 125-133.
PW welding Power; Watt Fonda,R. W., Bingert, J. F., and Colligan, K. J., 2004,
ri radius of the pin; mm “Development of Grain Structure during Friction
r0 radius of the shoulder; mm Stir Welding,” Scripta Materialia, Vol. 51, No.
T torque; KN*mm 3, pp. 243-248.
W width of workpiece; mm Jandric, D., Chen, C. M., and Kovacevic, R., 14-16
η motor efficiency; dimensionless May 2003, “Characterization of Weld Quality by
ω rotating speed of the tool; rpm Different Sensing Techniques in Friction Stir
Welding of Lap Joints,” 4th International Sympo-
REFERENCES sium on Friction Stir Welding, Park City, Utah,
USA, in CDROM.
Allehaux, D., Petit, G., Campagnac, M.-H., Lapasset, Khandkar, M. Z. H., and Khan, J. A., 2001, “Ther-
G., and Denquin, A., 2003,. “Microstructure and mal Modeling of Overlap Friction Welding for
Properties of a Friction Stir Welded 7349-T6 Alu- Al-Alloys,” Journal of Materials Processing &
minum Alloy,” 4 th International Symposium on Manufacturing Science, Vol. 10, No. 2, pp. 91-
Friction Stir Welding, Park City, Utah, USA, in 105.
CDROM. Khandkar, M. Z. H., Khan, J. A., and Reynolds, A. P.,
Arbegast, W. J., and Hartley, P. J., 1998, “Friction 2002, “A Thermal model of The Friction Stir Weld-
Stir Weld Technology Development at Lockheed ing Process,” Proceedings of IMECE’02, Interna-
Martin Michoud Space System – an Overview,” tional Mechanical Engineering Congress &
Proceedings of the 5th International Conference Exhibition, pp. 17-22.
on Trends in Welding Research, Pine Mountain, Lindner, K., Khandkar, Z., Khan, J. Tang, W., and
Georgia, USA, pp. 541-546. Reynolds, A. P., 2003, “Rationalization of Har-
Chao, Y. J., Liu, S., and Chien, C. H., 2008, “Friction ness Distributions in Alloy 7050 Friction Stir
Stir Selding of Al 6061-T6 Thick Plates: Part II – Welds Based on Weld Energy, Weld Power, and
Numerical Modeling of Thermal and Heat Trans- Time/Temperature History,” 4th Friction Stir
fer Phenomenon,” Journal of the Chinese Institute Welding International Symposium, Park City,
 Y. J. Chao et al.: Friction Stir Welding of Al 6061-T6 Thick Plates: Part I – Experimental Analyses 767

Utah, USA, in CD ROM. Welding International Symposium, Gothenburg,

Liu, S., Chao, Y.J. and Chien, C.H., 2004, “Study of Sweden, in CD ROM.
Thermal and Heat Transfer Phenomenon in Fric- Smith, C., Hinrichs, J., and Crusan, W., 2003. “Ro-
tion Stir Welding of Aluminum Alloy 6061-T6 botic Friction Stir Welding: The State of The
Thick Plates,” SAE paper 2004-01-1328, SAE Art,” 4th Friction Stir Welding International
World Congress, Detroit, Michigan, USA, in CD Symposium, Park City, Utah, USA, in CD ROM.
ROM. Song, M., and Kovacevic, R., 2002, “A New Heat
McClure, J. C., Feng, Z., Tang, W., Murr, L. E., and Transfer Model for Friction Stir Welding,” Trans-
Guo, X., June 1-5 1998, “A Thermal Model of actions of NAMRI/SME, Vol. XXX, No. MS02-
Friction Stir Welding,” Proceedings of the 5th In- 175, pp. 1-8.
ternational Conference on Trends in Welding Steward, M. B., Adams, G. P., Nunes Jr., A. C., and
Research, Pine Mountain, Georgia, USA, pp. 590- Romine, P., 1998, “A Combined Experimental
595. and Analytical Approach to Understanding Fric-
Packer, S., Nelson, T., Sorensen, C., Steel, R., and tion Stir Welding,” Proceedings of the 19th South-
Matsunaga, M.,2003, “Tool and Equipment for eastern Conference on Theoretical and Applied
Friction Stir Welding and Other High Melting Mechanics, Deerfield Beach, Florida, USA, pp.
Temperature Alloys, 4th Friction Stir Welding 472-484.
International Symposium, Park City, Utah, USA, Strombeck, A.V., Schilling, C., and J. Santos, F.D.,
in CD ROM. 2000, “Robotic Friction Stir Welding – Tool
Park, S. C., Sato, Y. S., and Kokawa, H., 2003, “Ef- Technology and Applications,” 2 nd Friction Stir
fect of Micro-Texture on Fracture Location in Welding International Symposium, Gothenburg,
Friction Stir Weld of Mg Alloy AZ61 during Ten- Sweden, in CD ROM.
sile Test,” Scripta Materialia, Vol. 49, No. 2, pp. Tang, W., Guo, X., McClure, J. C., Murr, L. E., and
161-166. Nunes, A., 1998, “Heat Input and Temperature
Shi, Q., Dickerson, T., and Shercliff, H. R., 2003, “Ther- Distribution in Friction Stir Welding,” Journal
mal-mechanical FE Modeling of Friction Stir of Materials Processing and Manufacturing
Welding of Al-2024 Including Tool Loads,” 4th Science, Vol. 7, No. 2, pp.163-172.
Friction Stir Welding International Symposium,
Park City, Utah, USA, in CD ROM. Manuscript Received: Nov. 23, 2007
Smith, C., 2000, “Robotic Friction Stir Welding Us- Revision Received: Apr. 07, 2008
ing A Standard Industrial Robot,” 2nd Friction Stir and Accepted: May 07, 2008