Materials and Design 45 (2013) 206211

Contents lists available at SciVerse ScienceDirect

Materials and Design

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

Mathematical model and optimization for underwater friction stir welding

of a heat-treatable aluminum alloy
Huijie Zhang a,b, Huijie Liu b,
a
b

State Key Laboratory of Robotics, Shenyang Institute of Automation, Chinese Academy of Sciences, Shenyang 110016, PR China
State Key Laboratory of Advanced Welding and Joining, Harbin Institute of Technology, Harbin 150001, PR China

a r t i c l e

i n f o

Article history:
Received 16 April 2012
Accepted 11 September 2012
Available online 21 September 2012
Keywords:
A. Non-ferrous metals and alloys
D. Welding
E. Mechanical

a b s t r a c t
During the friction stir welding (FSW) of heat-treatable aluminum alloys, the welding thermal cycles tend
to cause a local softening in the joints and thus lead to a degradation in joint properties. Underwater FSW
has been demonstrated to be available for the strength improvement of normal joints. In order to obtain
the optimum welding condition for underwater FSW, a 2219-T6 aluminum alloy was underwater friction
stir welded and a mathematical model was developed to optimize the welding parameters for maximum
tensile strength in the present study. The results indicate that a maximum tensile strength of 360 MPa
can be achieved through underwater FSW, higher than the maximum tensile strength obtained in normal
condition.
2012 Elsevier Ltd. All rights reserved.

1. Introduction
Heat-treatable aluminum alloys with favorable strength to
weight ratios are widely used in aircraft and aerospace engineering
[1,2]. Friction stir welding (FSW) is a solid state joining process and
has been widely utilized to weld various aluminum alloys that
were difcult to fusion weld [3,4]. Regarding the FSW of heattreatable aluminum alloys, the thermal cycles generally deteriorate the local mechanical properties of the joints due to coarsening
or dissolution of the strengthening precipitates [57]. In recent
years, of particular interest is to improve the joint properties by
controlling the temperature level. In order to do so, water has been
chosen by several researchers to exert an in-process cooling effect
on the samples during FSW because of its widespread distribution
and excellent heat absorption capacity. Fratini et al. [8,9] performed FSW of 7075-T6 aluminum alloy during which the water
owed across the top surfaces of the samples. The tensile strength
of the joints was found to be improved to some extent. In order to
take full advantage of the heat absorption capacity of water, the
present authors [10,11] conducted underwater FSW of 2219-T6
aluminum alloy, in which the whole workpiece was immersed in
the water environment during the welding. The results indicated
that the tensile strength of the underwater joint was higher than
that of the corresponding normal joint.
Previous investigations have highlighted the advantages of
external water cooling for strength improvement of normal FSW
joints. From the viewpoint of application, it would be more signif Corresponding author. Tel.: +86 451 8641 3951; fax: +86 451 8641 6186.
E-mail address: liuhj@hit.edu.cn (H. Liu).
0261-3069/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.matdes.2012.09.022

icant to optimize the in-process cooling FSW for maximum

mechanical properties of the joints. However, this work has not
been conducted until now. Hence, a 2219-T6 aluminum alloy
was underwater friction stir welded in the present study, and a
mathematical model was developed to predict the tensile strength
of the joints. Using this model, the optimization of welding parameters was conducted to maximize the tensile strength of the underwater joints.
2. Experimental procedure
The base metal (BM) used for the experiment was a 7.5-mmthick 2219-T6 aluminum alloy, whose chemical compositions
and mechanical properties are listed in Table 1. The plate was machined into rectangular welding samples with dimension of
300 mm long by 100 mm wide. After cleaned by acetone, the samples were clamped to the backing plate in a vessel, and then the
water at room temperature was poured into the vessel to immerse
the top surfaces of the samples. Fig. 1 shows the schematic view of
underwater FSW. Butt welds were made under water using an FSW
machine (FSW-3LM-003) along the longitudinal direction of the
welding samples.
It has been stated that the tool geometry, rotation speed, welding speed and shoulder plunge depth are main factors that affect
the mechanical properties of FSW joints [1215], however, in regard to the present BM, our previous work demonstrated that
the tool with a 22.5-mm-diameter shoulder and a 7.4-mm-length
pin was suitable for a wide range of parameters to produce defect-free FSW joints. Therefore, this tool was xed and used
throughout the whole experiment. The rotation speed, welding

207

H.J. Zhang, H.J. Liu / Materials and Design 45 (2013) 206211

Table 1
Chemical compositions and mechanical properties of 2219-T6 aluminum alloy.
Chemical compositions (wt.%)

Mechanical properties

Al

Cu

Mn

Fe

Ti

Zn

Si

Zr

Tensile strength

Elongation

Bal.

6.48

0.32

0.23

0.06

0.08

0.04

0.49

0.20

432 MPa

11%

Fig. 1. Schematic view of underwater FSW.

Table 2
Welding parameters and their levels.
Parameter (unit)

Symbol

Rotation speed (rpm)

Welding speed (mm/min)
Shoulder plunge depth (mm)

x
v
p

Table 3
BoxBehnken design matrix and experimental results.
Level

No.

Low (1)

Middle (0)

High (1)

900
150
0.2

1000
200
0.3

1100
250
0.4

speed and shoulder plunge depth were considered to be variables

for the optimization of underwater FSW process. Trial experiments
were performed to identify the working ranges of the parameters.
The feasible limits of these parameters should be determined on
the basis that neither external nor internal defects are formed in
the joints during FSW. The determined working ranges are listed
in Table 2.
In the present study, the three-level and three-factorial Box
Behnken experimental design was chosen for nding out the relationship between the response (tensile strength) and the variables
(welding parameters). The model has the advantage that it permits
the use of relatively few combinations of variables for determining
the complex response function [1618]. The levels of the variables
were coded as 1 (low), 0 (central point or middle) and 1 (high)
(see Table 2). A total of 15 experiments were required in the
BoxBehnken experimental design. The actual design of matrix is
shown in Table 3. Underwater FSW experiments were made
according to this design.
K-type thermocouples were used to measure the temperature
of the samples during FSW. The measured locations were started
at the heat affected zone (HAZ, 6 mm from the weld center) and
then extended to the BM at the weld mid-thickness. The space
was 3 mm between the adjacent measured locations.
After welding, the transverse tensile specimens were cut from
the joints. The tensile specimens were prepared with reference to
China National Standard GB/T2651-2008 (equivalent to ASTM
B557-02) [19]. The room temperature tensile test was carried out
at a crosshead speed of 1 mm/min using a computer-controlled
testing machine (Instron-1186). The tensile properties of each joint
were evaluated using three tensile specimens cut from the same
joint. After tensile test, the optical microscopy (OM, OlympusMPG3) and scanning electron microscopy (SEM, Hitachi-S4700)
were utilized to analyze the fracture features of the joints. The

Welding parameters

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

TS (MPa)

1
1
1
1
1
1
1
1
0
0
0
0
0
0
0

1
1
1
1
0
0
0
0
1
1
1
1
0
0
0

0
0
0
0
1
1
1
1
1
1
1
1
0
0
0

339
339
349
342
347
342
347
340
339
356
343
352
356
357
359

development of mathematical model and the subsequent statistical analysis were both performed by Design-Expert software.
Transmission electron microscopy (TEM) analyses were conducted on some specic joints. The foil disk specimens were cut
parallel to the welding direction from the HAZ (at a distance of
6 mm from the weld center) and the BM at weld mid-thickness.
The specimens were rst machined and manually polished down
to a thickness of 100 lm. The nal thickness reduction was acquired by electro-polishing with a HNO3 solution (HNO3 30% in
volume in methanol at 35 C under 18 V).
3. Results and discussion
3.1. Development of mathematical model
For three variables, the quadratic response y is described in the
form of the following equation [16]:

y b0

bi xi

bij xi xj

bii x2i

where b0 is the intercept term, bi is the linear term, bii is the quadratic term and bij is the interaction term.
In the present study, the response, i.e. tensile strength (TS), is a
function of rotation speed (x), welding speed (v) and shoulder
plunge depth (p), therefore the equation can be expressed as:

208

H.J. Zhang, H.J. Liu / Materials and Design 45 (2013) 206211

TS b0 b1 x b2 v b3 p b12 xv b13 xp b23 v p b11 x2

b22 v b33 p
2

The experimental results of tensile strength are given in Table 3.

By applying multiple regression analysis on the design matrix and
the response values, the following second-order polynomial equation is established:

TS 804:88 1:92x 1:49v 370p  3:5  104 xv

 0:05xp  0:4v p  9:29  104 x2  2:32  103 v 2
 404:17p2

The variance was analyzed to test the signicance of t of the

equation (see Table 4). The model gives a highly signicant F-value,
demonstrating that the model adequately represents the actual
relationship between the response and the variables. Any model
term with p < 0.05 is signicant. Therefore x, v, x2, v2, p2 are signicant model terms and the Lack of Fit is not signicant.
The determination coefcient reects the goodness of t for the
model. The model presents a determination coefcient (R2) of 0.96
and an adjusted determination coefcient (adjusted-R2) of 0.89,
implying a high correlation between the experimental and the predicted results.
There are two functions of the developed model. First, the joint
properties obtained at any combination of parameters in the present working ranges can be predicted by the mathematical model;
second, the optimization of underwater FSW process can be conducted by analyzing the model.
3.2. Inuence of welding parameters on tensile strength
Prior to optimization, it is necessary to illuminate the inuence
of welding parameters on tensile strength. For this purpose, 3D response surface graphs and contour plots are formed based on the
developed model by considering one parameter in the middle level
and the other two parameters as variables (see Fig. 2).
With increasing the rotation speed at a given welding speed or
shoulder plunge depth, the tensile strength rst increases to a
maximum value and afterwards shows a decrease (see Fig. 2ad).
The poor tensile strength at lower rotation speeds is generally owing to the inadequate tool stirring action [20,21]. Increasing the
rotation speed in a certain range enhances the strain hardening effect induced by tool stirring and thus results in the increase in tensile strength. However, when the rotation speed increases to a
rather high value, the excess heat input plays the predominant role
and lowers the tensile strength signicantly. This is just the reason
why the rotation speed close to 1000 rpm tends to yield high
tensile strength.

At a given rotation speed or shoulder plunge depth, the initial increase of welding speed increases the tensile strength to a certain value, and further increase of welding speed makes the tensile strength
nearly constant (see Fig. 2ab and ef). The lower welding speed
produces larger heat input into the welding samples, which signicantly deteriorates the mechanical properties of joints. As the welding speed increases, the effect of thermal cycles on joint properties is
weakened, leading to an improvement in tensile strength. Therefore,
the relative high welding speeds (above 200 mm/min) are benet to
produce good-quality underwater joints.
When the shoulder plunge depth increases at a xed rotation
speed or welding speed, the evolution of tensile strength shows
similar trend to that with rotation speed, i.e. rst increases to a
maximum value and then presents a decrease (see Fig. 2cf). However, the tensile strength is less sensitive to change of shoulder
plunge depth than to change of rotation speed. Increasing the
shoulder plunge depth in a certain range can improve the forging
action on the plasticized metal and enhance the material mixing
and inter diffusion, which is benet to strength improvement. Nevertheless, if a rather large shoulder plunge depth is employed, the
heat input then leads to the decrease of tensile strength by deteriorating the strengthening precipitates. Accordingly, the shoulder
plunge depth close to 0.3 mm facilitates higher tensile strength.
3.3. Optimization of welding parameters
A maximum point is observed from the response surface graphs
between any two variables and the response (see Fig. 2a, c and e),
indicating that there is a maximum tensile strength value inside
the experimental region. By analyzing the response surface graphs
and contour plots, the optimum welding parameters and the maximum tensile strength are obtained, as listed in Table 5. To demonstrate the validity of the procedure, underwater FSW was
conducted using the optimum welding parameters. The experimental result exhibits highly agreement with the predicted value.
For the base metal used in this study, the maximum tensile
strength of the FSW joints was found to be 340 MPa in normal condition, which was obtained at the rotation speed of 600 rpm, welding speed of 200 mm/min and shoulder plunge depth of 0.3 mm
[22]. However, the present investigation demonstrates that a maximum tensile strength of 360 MPa can be achieved through underwater FSW, 6% higher than that obtained in normal FSW.
3.4. Intrinsic reason for strength improvement by underwater FSW
Under normal condition, the tensile strength of the FSW joints
of heat-treatable aluminum alloys cannot be further improved
due to the limitation of heat dissipation level, but the joint with

Table 4
Variance analysis of the second-order polynomial equation.
Source

Sum of squares

df

Mean square

F value

p-value prob > F

Remark

Model

712.31667
45.125
190.125
0.5
12.25
1
16
318.77564
123.85256
60.314103
29.416667
24.75
4.6666667
741.73333

9
1
1
1
1
1
1
1
1
1
5
3
2
14

79.146296
45.125
190.125
0.5
12.25
1
16
318.77564
123.85256
60.314103
5.8833333
8.25
2.3333333

13.452628
7.6699717
32.315864
0.0849858
2.082153
0.1699717
2.7195467
54.182829
21.051427
10.251689

0.0053
0.0394
0.0023
0.7824
0.2086
0.6972
0.1600
0.0007
0.0059
0.0239

**
**
**
*
*
*
*
**
**
**

3.5357143

0.2283

x
v
p

xv
xp
vp

x2

v2
p2
Residual
Lack of t
Pure error
Cor. total
**
*

Means signicant variables.

Indicates insignicant variables.

H.J. Zhang, H.J. Liu / Materials and Design 45 (2013) 206211

209

Fig. 2. (a, c, e) Response surface graphs, (b, d, f) contour plots corresponding to a, c and e, respectively.

Table 5
Optimum welding parameters and maximum tensile strength.
Optimum welding parameters

TS (MPa)

Parameter (unit)

Value

Predicted

Experimental

x (rpm)

983
223
0.3

360

358

v (mm/min)
p (mm)

relatively high tensile strength can be produced through underwater FSW. To clarify the intrinsic reason for the strength improvement by underwater FSW, the optimal joints obtained from
normal and underwater FSW were analyzed detailedly.

From the welding thermal cycles shown in Fig. 3 it can be seen

that the optimal underwater joint experiences lower peak temperature than the optimal normal joint (see Fig. 3a). Furthermore, the
optimal underwater joint is characterized by greater heating and
cooling rates, leading to a shorter dwelling time above a given temperature (see Fig. 3b). This means that the water cooling action
effectively controls the temperature levels during FSW. The diffusion of solute atoms is therefore restricted, which lowers the precipitate deterioration level during FSW (see Fig. 4).
Fig. 5 shows the fracture features of the both optimal joints. The
fracture of the optimal normal joint is observed near the interface
of thermal mechanically affected zone (TMAZ) and heat affected

210

H.J. Zhang, H.J. Liu / Materials and Design 45 (2013) 206211

Fig. 3. Temperature histories of the optimal joints: (a) peak temperature, (b) welding thermal cycles (at the distance of 9 mm from weld center).

zone (HAZ) (see Fig. 5a), as commonly occurred in other normal

specimens [5,7]. In contrast, the optimal underwater joint is fractured through the nugget zone, closer to the weld center. The
movement of fracture location further demonstrates the reduction
in material softening. The two fracture surfaces are signicantly
different because they occur in two different weld zones. The large
dimples in Fig. 5c are reections of the distorted and grown grains
near the TMAZ/HAZ interface; in turn, the small equiaxed dimples
in Fig. 5d characterize the typical nugget grain structures.
Several previous investigations have also found the positive
effect of external liquid cooling on strength improvement of
FSW joints [8,9,23,24]. However, these results were all obtained
under the condition that the welding parameters used for in-pro-

cess cooling FSW were the same as those used for normal FSW.
In comparison, the result of the present study is more signicant,
because it demonstrates that the joint properties can be further
improved on the basis of the optimal normal FSW. That is to
say, the mechanical properties of FSW joints of heat-treatable
aluminum alloys are able to be essentially improved by external
liquid cooling action. The water cooling conferred to the joint
during the FSW allows to limit as much as possible the heat ow
nearby the tool, and thus can further weaken the negative effect
of thermal cycles on the microstructures of joints when compared to the optimal normal FSW process. Finally, the joints with
superior tensile properties can be produced through underwater
FSW.

Fig. 4. Precipitate distributions in the BM and the HAZ: (a) BM, (b) HAZ of the optimal normal joint, (c) HAZ of the optimal underwater joint.

H.J. Zhang, H.J. Liu / Materials and Design 45 (2013) 206211

211

Fig. 5. Fracture features of the optimal FSW joints: (a and b) fracture locations of the optimal normal and underwater joints, (c and d) fracture surfaces of the optimal normal
and underwater joints. Note that the fracture surface micrographs were extracted from the locations marked by arrows in a and b.

4. Conclusions
Based on the present investigation, the results of signicance
are drawn as follows:
(1) A mathematical relationship between tensile strength and
welding parameters is developed to predict the tensile
strength of underwater friction stir welded 2219-T6 aluminum alloy. For the three investigated process parameters,
the variance analysis indicates that the rotation speed and
the welding speed are predominant factors that affect the
tensile strength.
(2) The effects of welding parameters on tensile strength of the
underwater joints are claried by analyzing the developed
model, and the intrinsic reasons for these effects are also
discussed.
(3) The optimization result indicates that a maximum tensile
strength of 360 MPa can be obtained through underwater
FSW, which is veried by the experimental result. This value
is 6% higher than the maximum tensile strength obtained in
normal FSW. The controlling of temperature histories and
microstructural evolutions is found to be the intrinsic reason
for the strength improvement by underwater FSW.

Acknowledgements
The authors are grateful to be supported by the National Basic
Research Program of China (973 Program, 2010CB731704) and by
the National Science and Technology Major Project of China
(2010ZX04007-011).
References
[1] Johnsen MR. Friction stir welding takes off at Boeing. Weld J 1999;78:359.
[2] Threadgill PL, Leonard AJ, Shercliff HR, Withers PJ. Friction stir welding of
aluminum alloys. Int Mater Rev 2009;54:4993.
[3] Joelj D. The friction stir welding advantage. Weld J 2001;80:304.
[4] Mishra RS, Ma ZY. Friction stir welding and processing. Mater Sci Eng R
2005;50:178.
[5] Fonda RW, Bingert JF. Microstructural evolution in the heat-affected zone of a
friction stir weld. Metall Mater Trans A 2004;35:148799.

[6] Simar A, Brchet Y, de Meester B, Denquin A, Pardoen T. Microstructure, local

and global mechanical properties of friction stir welds in aluminium alloy
6005A-T6. Mater Sci Eng A 2008;486:8595.
[7] Starink MJ, Seschamps A, Wang SC. The strength of friction stir welded and
friction stir processed aluminum alloys. Scr Mater 2008;58:37782.
[8] Fratini L, Buffa G, Shivpuri R. In-process heat treatments to improve FS-welded
butt joints. Int J Adv Manuf Technol 2009;43:66470.
[9] Fratini L, Buffa G, Shivpuri R. Mechanical and metallurgical effects of in process
cooling during friction stir welding of AA7075-T6 butt joints. Acta Mater
2010;58:205667.
[10] Liu HJ, Zhang HJ, Huang YX, Yu L. Mechanical properties of underwater friction
stir welded 2219 aluminum alloy. Trans Nonferrous Met Soc China
2010;20:138791.
[11] Liu HJ, Zhang HJ, Yu L. Homogeneity of mechanical properties of underwater
friction stir welded 2219-T6 aluminum alloy. J Mater Eng Perform
2011;20:141922.
[12] Cabibbo M, McQueen HJ, Evangelista E, Spigarelli S, Di Paola M, Falchero A.
Microstructure and mechanical property studies of AA6056 friction stir
welded plate. Mater Sci Eng A 2007;460461:8694.
[13] Cavaliere P, Squillace A, Panella F. Effect of welding parameters on mechanical
and microstructural properties of AA6082 joints produced by friction stir
welding. J Mater Proc Technol 2008;200:36472.
[14] Kumar K, Kailas SV. On the role of axial load and the effect of interface position
on the tensile strength of a friction stir welded aluminium alloy. Mater Des
2008;29:7917.
[15] Dixit V, Mishra RS, Lederich RJ, Talwar R. Inuence of process parameters on
microstructural evolution and mechanical properties in friction stirred Al2024 (T3) alloy. Sci Technol Weld Join 2009;14:34655.
[16] Ferreira SLC, Bruns RE, Ferreira HS, Matos GD, David JM, Brando GC, et al.
BoxBehnken design: an alternative for the optimization of analytical
methods. Anal Chim Acta 2007;597:17986.
[17] Nguyen N-K, Borkowski JJ. New 3-level response surface designs constructed
from incomplete block designs. J Stat Plan Infer 2008;138:294305.
[18] Ray S, Reaume SJ, Lalman JA. Developing a statistical model to predict
hydrogen production by a mixed anaerobic mesophilic culture. Int J Hydrogen
Energy 2010;35:533242.
[19] GB/T26512008/ISO4136: 2001. Tensile test method on welded joints.
Standardization Administration of the Peoples Republic of China; 2008.
[20] Xu WF, Liu JH, Luan GH, Dong CL. Microstructure and mechanical properties of
friction stir welded joints in 2219-T6 aluminum alloy. Mater Des
2009;30:34607.
[21] Zhang HJ, Liu HJ, Yu L. Microstructure and mechanical properties as a function
of rotation speed in underwater friction stir welded aluminum alloy joints.
Mater Des 2011;32:44027.
[22] Pan Q. Welding defects and mechanical properties of friction stir welded 2219
aluminum alloy joints. MS thesis. School of Material Science and Engineering,
Harbin Institute of Technology, Harbin, China; 2007.
[23] Hosseini M, Manesh HD. Immersed friction stir welding of ultrane grained
accumulative roll-bonded al alloy. Mater Des 2010;31:478691.
[24] Fu RD, Sun ZQ, Sun RC, Li Y, Liu HJ, Liu L. Improvement of weld temperature
distribution and mechanical properties of 7050 aluminum alloy butt joints by
submerged friction stir welding. Mater Des 2011;32:482531.