Materials and Design 67 (2015) 145–153

Contents lists available at ScienceDirect

Materials and Design

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

Technical Report

Experimental optimization of the mechanical properties of friction stir

welded Acrylonitrile Butadiene Styrene sheets
Nasser Sadeghian ⇑, Mohammad Kazem Besharati Givi
School of Mechanical Engineering, College of Engineering, University of Tehran, North Kargar St., Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Article history: One of the joining methods of engineering thermoplastics is friction stir welding which is based on fric-
Received 27 June 2014 tional heat generated through contact between a rotating tool and the workpiece. In this article, Acrylo-
Accepted 22 November 2014 nitrile Butadiene Styrene sheets are friction stir welded using a metallic tool. Mechanical properties of the
Available online 29 November 2014
welded sheets are investigated under varying processing parameters. Empirical welds are performed
using design of experiment theorem and analyzed by statistical methods. The main factors are geometry
of the tools such as pin profile, diameters ratio, and welding process parameters include rotational and
linear speeds. The orientation of the tool is another important factor, named tilt angle. A fully central
composite design for welding process is assessed using analysis of variance and is optimized through
response surface methodology. Tensile yield stress is considered as response of designed process exper-
imentally by standard test method. Analytical and experimental results showed an optimized weld
strength up to 97% and 101% respectively for cylindrical and conical pinned tools compared to tensile
strength of the parent ABS sheet. Finally, significant factors and optimal conditions obtained to be 2°
of tilt angle, 900 rpm for rotational speed, a conical tool with diameters ratio of 20/6 and linear speed
of 25 mm/min.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction be produced (Fig. 1) [1]. Different parameters including rotational

speed, tilt angle, linear speed and the pin geometry affect the prop-
Plastics have numerous designing and forming advantages over erties and characteristics of the weld zone.
metals. Excellent resistance to chemicals and humid environments As shown in Fig. 1, the rotating tool, consisting of a shoulder and
is among the more liked points of their behavior. However, there is a pin, inserts in the seam between the sheets, which are con-
some problems for using the plastics. The main problem with the strained together. Friction generates the heat and the ‘‘key hole’’
application of plastics relates to their strength and temperature appears. The softened material traps in a limited space by the
stability restrictions. In order to overcome the problem, making boundaries of the shoulder, pin and solid edges of thermoplastic
one-piece of multi-parts and use of permanent joining methods and then weldment is produced [2,3].
are preferred instead of using non-permanent and temporary The concept of thermoplastic FSW is similar to that of the
joints. Development of welding techniques with low energy con- metallic material. Thermoplastics are very low thermal dissipated
sumption, environment compatibility, cost-effectiveness, availabil- materials compared to metals. In addition, their melt and softening
ity, ease of training, high production speed, and no- edge points are much lower than those of the metallic cases are. Besides,
preparation, is among the noticeable welding conditions. More- inter-molecular connections are weaker in thermoplastic struc-
over, mechanical strength and stability are very important in weld- tures. FSW of metallic materials are investigated in many of litera-
ing processes. tures [2,3], but there are few studies reported practices in
Friction stir welding (FSW) technique is a novel method, which evaluation of thermoplastics).
was invented by the Welding Institute (TWI) in 1991. In FSW, the Bagheri et al. employed the design of experiment in their
heat of welding is generated by friction between the tool and the research about the FSW of ABS sheets. They used a rotating tool
work-piece. Rotation of the tool causes mechanical stirring and with a hot shoe. They reported an optimizing research in mechan-
mixing in a solid-state; and then weldment of the materials will ical strength of welded joint that reached to 89% compared to the
base material at 100 °C of the hot shoe temperature. In FSW of a
thermoplastic, the melt flow is created by stretching of the melted
⇑ Corresponding author. Tel.: +98 9124397872; fax: +98 2156725560. material on the bottom surface of the shoulder at the advancing
E-mail address: [email protected] (N. Sadeghian).

http://dx.doi.org/10.1016/j.matdes.2014.11.032
0261-3069/Ó 2014 Elsevier Ltd. All rights reserved.
146 N. Sadeghian, M.K. Besharati Givi / Materials and Design 67 (2015) 145–153

side to the retreating side downward to the weld root. They divide and comparing to parent material is presented. Improving process
the weldment into four segments: stirring zone, advancing bound- are not clear in past researches. Since there are more parameters
ary, retreading boundary and heat affected zone. This division can that affect the welding process it is important to design a system-
be obtained by experimental observations. They mentioned that atic procedure to investigate the process accurately and traceabil-
melt flow does not stretch from the retreating side to the pin side ity. The system will be introduced the significant parameters, we
[4]. can name them as factors of the process inputs. Besides, morpho-
Strand et al. conducted a feasibility study from technical and logical studies of the welded joint lead one to give expression
economical viewpoint about the FSW of polymers and pointed that about mechanical and physical changing at welding process. In
there are some advantages to other welding methods [5]. Sorensen current article, a systematic process shall designed to implementa-
and co-workers invented a specific tool for the FSW of polymers, tion of FSW in an amorphous thermoplastic. Significant factors are
named ‘‘hot shoe’’. In this tool, a hot plate was used for improving selected experimentally using statistical methods. Their effects on
the quality of the weld bead [6]. At the same time, Scialpi et al. response of system is studied. Tensile strength of the weld joint
introduced a reciprocating blade for thermoplastic FSW [7]. Nelson defined as response of system. Empirical equations will derived
et al. evaluated the effects of FSW on the polymer’s structure [8]. from experimental tensile tests and analyzed using mathematical
They reported zero-defect welds and introduced four zones in techniques. Optimization of weld joint efficiency is the main goal
welding of thermoplastic using a special tool in their research. of the article. Then the optimal conditions of welded joints will dis-
The zones are boundary of advancing side, retreating side, central cussed and the effects of welding input variables onto mechanical
zone, and distributed subzone. Also, some defects such as voids, properties of welded joints will determined.
root defects and flow lines of matters were observed in their exper-
imental observations.
Arici and Sinmazçelýk presented an investigation on double 2. Experimental methods
side welding of polyethylene sheets without root defects. The
welded samples were prepared according to standard tensile and Acrylonitrile Butadiene Styrene (ABS) sheets with 8 mm thick-
bending tests. Weld joint strength was reported to be about 80– ness cut into the dimension of 150 ⁄ 100 mm. typical properties
100% of the base material [9]. Kiss and Czigány conducted FSW of ABS sheet are tabulated in Table 1 [12]. ABS is a triple copolymer
process for polypropylene plates [10] and analyzed the welded thermoplastic that has a unique structure with polymerization
joint strength according to the variations of rotational and linear point of 75 °C and a highest glass transition temperature of about
speeds of the welding tool. In addition, crystallization of the weld- 80 °C. ABS acts as an amorphous thermoplastic and has low mois-
ing zone and boundaries were evaluated by calorimetric tests and ture absorption content that prevents from growing up of micro
the melt flow observed via scanning electronic microscopy. Fur- bio-organs. Main applications of ABS sheet is in automobile indus-
thermore, they reported Brittleness of weldment due to welding tries due to its capabilities in impact energy absorption, and
process in their results. They also reported that weld strength is dimensional stability. In addition, good wear resistance in room
a fraction of the base material. Authors of the present study con- temperature make ABS as a suitable choice for home appliances.
ducted the feasibility study of the FSW of polyethylene sheets. A FSW of thermoplastics may have problems related to void cre-
Taguchi design method is employed to study the effect of process- ation, weld root and top defects, as well as through thickness dis-
ing parameters on the yield strength of welded sheets. Results indi- continuities. In this research, the core of weldment free from top
cated that Maximum strength found about 62% of the base and root of weld defects is used. Based on the experiences, two
material. Welding tool was a conventional FSW tool with cylindri- types of FSW tools are selected: (a) cylindrical shoulder with a
cal shoulder and pin. Important parameters were rotational and cylindrical pin, and (b) cylindrical shoulder with a conical pin.
linear speeds and tilt angle [11]. These two type of tools are shown in Figs. 2 and 3, respectively.
Past researches proposed FSW of thermoplastics because of its The tools are made of tool steel, and the bottom surface of the
benefits and suitability. In addition, technical challenges about shoulders is machined with a concave (about) from edge to center
execution and implementation of welding process are reported of the tool. This is a technical methodology can be employed to
as case studies. Design considerations related to the tooling of increase welding efficiency by trapping generated heat of friction
the welding was explained less. Evaluation of welded joint studied in a restricted space to transform the material state from solid to
molten matter. Because this technique prevents of rotating cooled
rings of softened materials rings around the pin during welding
process. An adjoined blade also is used to produce a good quality
weld top in FSW of ABS sheets. This blade differs from hot shoe,
and is called ‘‘statically hot’’ tool. The blade is made of stainless
non-stick steel. The term statically refers to the worming up of
the blade, because the blade does not have rotational motion.

Table 1
Specifications of ABS sheets [12].

Property Unit Value

Elastic modulus MPa 1108–1339a
Tensile strength MPa 39.35–42.87a
Ultimate elongation % 2.92–3.14a
Notched Izod impact strength at 23 °C J/m 30–450
Coefficient of friction (with steel) ASTM D1894 0.21–0.28
Coefficient of friction (with aluminum) ASTM D1894 0.40
Specific gravity – 1.03–1.09
Moisture absorption at 23 °C and R.H. 50% % 0.21–0.35
Rockwell hardness R 102–124
a
Fig. 1. Schematics of friction stir welding [1]. Experimentally assessed in this research.
 N. Sadeghian, M.K. Besharati Givi / Materials and Design 67 (2015) 145–153 147

Fig. 2. Cylindrical pinned tools for friction stir welding of themoplastcs.

Fig. 3. Conical pinned tools for friction stir welding of themoplastcs.

Table 2
Statically hot blade has some advantages include: (a) Preventing
Factors and of designed experiments for friction stir welding of ABS sheets.
from undercutting at the top of weld; (b) Flattening the weld bead;
(c) Applying a uniform compression on the extrudates from the Parameters Symbol/unit Levels Coded parameter value
weld zone and (d) Decreasing in die swelling intension by delaying 1 0 1
extrudates. Rotational speed n (rpm) 3 900 1400 1800
To keep the sheets together during the welding, a fixture made Linear speed V (mm/min) 3 6 16 25
of a structural steel prepared and mounted on the table of the Tilt angle a (°) 3 0 1 2
Shoulder diameter D (mm) 3 10 15 20
welding machine. FSW tool inserted into the collet of the welding
Pin diameter d (mm) 3 5 6 8
machine. The welding machine is a universal FRITZ WERNER mill- Diameters ratio D/d (mm/mm) 3 2 2.5 20
6
ing machine. Pin profile Block 2 Cylindrical – Conical
Design of experiment (DOE) theorem [13,14] proposed to exam-
ine the mechanical strength of welded samples. Significant factors
and their values intervals accrued from the pretest experiences. A
fully central composite of the experiments is designed, and type of
the pin profile is proposed as the block parameter for the experi-
mentation. Processing parameters including rotational (n) and lin-
ear (V) speeds, diameters ratio (D/d), and tilt angle of the tool (a)
are considered as the main factors. Levels and values of factors
are listed in Table 2. Acceptable welds are evaluated by standard
test method for mechanical strength properties. The samples of
welds are examined based on the ASTM: D638 method by using
an INSTRON 1095 tensile test machine. Test coupons are prepared
according to the type IV of standard test method as illustrated in
Fig. 4. In order to avoid defects in the root and top of the weld joint,
the samples are prepared from the core of the weld joint. Because
of strain-rate dependency of thermoplastic, the tensile test speed Fig. 4. Sample preparation sketch for tensile test according ASTM: D638.
148 N. Sadeghian, M.K. Besharati Givi / Materials and Design 67 (2015) 145–153

Table 3
Tensile test results of the experimentation in current study.

Run Block D/d (mm/mm) n (rpm) V (mm/min) Alpha (°) Max. tensile strength
True stress (MPa) Relative strength (%)
– Parent ABS – – – – 41.80 100
1 Cylindrical 2 1400 16 1 26.71 63.9
2 Cylindrical 20/6 1400 16 1 41.42 99.1
3 Cylindrical 2.5 1800 16 1 34.45 82.4
4 Cylindrical 2.5 1400 16 2 34.13 81.7
5 Cylindrical 2.5 1400 16 1 27.18 65
6 Cylindrical 2.5 1400 16 1 25.10 60
7 Cylindrical 2.5 1400 25 1 37.36 89.4
8 Cylindrical 2.5 900 16 1 23.93 57.2
9 Cylindrical 2.5 1400 6 1 26.91 64.4
10 Cylindrical 2.5 1400 16 0 27.76 66.4
11 Conical 20/6 900 25 0 5.79 13.9
12 Conical 2.5 1400 16 1 20.20 48.3
13 Conical 2 1800 25 0 – –
14 Conical 20/6 1800 25 2 33.75 80.7
15 Conical 2.5 1400 16 1 27.03 64.7
16 Conical 2 1800 25 2 24.93 59.6
17 Conical 20/6 900 6 0 6.82 16.3
18 Conical 2 900 6 0 37.14 88.9
19 Conical 20/6 900 6 2 34.14 81.7
20 Conical 20/6 1800 25 0 33.04 79
21 Conical 2 900 25 2 – –
22 Conical 2 1800 6 0 39.30 94
23 Conical 2 1800 6 2 32.56 77.9
24 Conical 2 900 25 0 16.65 39.8
25 Conical 2.5 1400 16 1 27.01 64.6
26 Conical 20/6 1800 6 0 27.77 66.4
27 Conical 2.5 1400 16 1 23.84 57
28 Conical 2 900 6 2 34.38 82.2
29 Conical 20/6 1800 6 2 33.91 81.1
30 Conical 20/6 900 25 2 41.95 100.4

set to be 15 mm/min for ABS welded sheets with regard to the Analysis and calculations are carried out using Minitab software,
maximum strain of the parent material that is about 10%. In Table 3, version 16.
the tensile test results based on proposed designed welding pro- The penetration of the pin at the thickness of the sheet is lim-
cess are summarized. Test results presented for true values of the ited by contacting the rear edge of the shoulder onto top surface
factors. Ten welding runs design to accomplish by the cylindrical of the sheets. Thus, the length of the pin is limited to about 90%
pinned and the other twenty welding runs, conducted by the con- of the sheet’s thickness. Melt matter was created by the frictional
ical pinned tools. True yield stress and relative tensile strength of heat restricted in a local space between the bottom surface of
the welded parts are introduced as the response of the welding the shoulder, perimeter of the pin and solid walls of welded mate-
process. Relative strength is calculated compare to parent ABS ten- rial. Passing of the rotating tool along the seam line is led to
sile strength, which is presented in first row of Table 3. increase in the temperature of the welded sheets. Molten state of
Analysis of variance (ANOVA) proposed to determine the contri- a thermoplastic has lower mechanical strength than the solid
bution of factors on a specified response. In addition, regression
equation between the factors and the response is derived. ANOVA
results of the standard tensile tests for the cylindrical and the con- Table 5
ical pinned tools are presented in Tables 4 and 5, respectively. ANOVA table of tensile strength for the conical pinned tools.

Source DF Seq. SS Adj. SS Adj. MS F P

Regression 11 1455.88 1455.88 132.353 4.78 0.034
Linear 4 522.50 505.61 126.404 4.57 0.049
Table 4 D/d 1 38.93 41.2 41.2 1.49 0.268
ANOVA table of tensile strength for the cylindrical pinned tools. N 1 157.45 100.86 100.86 3.64 0.105
V 1 66.9 34.714 34.714 1.25 0.306
Source DF Seq. SS Adj. SS Adj. MS F P
a 1 259.21 171.094 171.094 6.18 0.047
Regression 8 277.753 277.753 34.7192 17.02 0.185 Square 1 44.39 52.522 52.522 1.9 0.218
Linear 4 213.123 213.123 53.2808 26.12 0.146 D/d ⁄ D/d 1 44.39 52.522 52.522 1.9 0.218
D/d 1 95.22 95.22 95.22 46.67 0.093 Interaction 6 888.99 888.99 148.166 5.35 0.030
N 1 50.501 50.501 50.5012 24.75 0.126 D/d ⁄ n 1 80.39 30.963 30.963 1.12 0.331
V 1 49.402 49.402 49.4018 24.21 0.128 D/d ⁄ V 1 231.61 109.234 109.234 3.95 0.094
a 1 18 18 18 8.82 0.207 D/d ⁄ a 1 262.38 204.05 204.050 7.37 0.035
Square 4 64.63 64.63 16.1576 7.92 0.26 n⁄V 1 8.37 2.381 2.381 0.09 0.779
D/d ⁄ D/d 1 27.44 55.354 55.3536 27.13 0.121 n⁄a 1 293.63 293.63 293.634 10.06 0.017
n⁄n 1 0.435 8.381 8.381 4.11 0.292 V⁄a 1 12.615 12.615 12.615 0.46 0.525
V⁄V 1 15.962 32.948 32.9476 16.15 0.155 Residual error 6 166.13 166.13 27.689
a⁄a 1 20.794 20.794 20.7936 10.19 0.193 Lack-of-fit 3 136.87 136.87 45.625 4.68 0.119
Residual error 1 2.04 2.04 2.0402 Pure error 3 29.26 29.26 9.753
Pure error 1 2.04 2.04 2.0402 Total 17 1622.02
Total 9 279.974
 N. Sadeghian, M.K. Besharati Givi / Materials and Design 67 (2015) 145–153 149

one. At the end of weld line, a vacant hole was created already strength of weldments is supposed as the response of systemati-
because of pulling out of the pin. cally designed FSW process. Calculation of the ANOVA is carried
out by Minitab software version 16. The ANOVA results of the
experimentations for cylindrical pinned tools that are presented
3. Results and discussion in Table 4 indicate that a quadratic equation between significant
factors affect the tensile strength of weld joint. Among the param-
3.1. Results eters, the diameters ratio has strongly affect mechanical strength
of the weldment. Rotational speed and linear speed of welding
FSW experimentation proposed to investigate the effect of pro- are other significant factors. Graphical illustration of the ANOVA
cessing parameters and geometric factors on the mechanical representation is shown in Fig. 5. As can be seen from the figure,
strength of the weld joint. Processing parameters include, rota- a curvature can be observed in the response diagrams. Curvature
tional and linear speeds while geometric factors include, diameters on tensile strength manner illustrated by quadratic terms of diam-
ratio and tilt angle that are summarized in Table 2. By increasing in eters ratio, rotational and linear speeds of welding process. The
initial stop-time, the stability and desirability of welding mathematical statement of the regression model for welded sheets
decreases, especially on the weld bead. Suitable values are by the cylindrical pinned tools is obtained as bellow by Eq. (1):
obtained in the range of 10–15 s for the ABS sheets welding process  
D
time. TS ¼ 25:59 þ 6:9 þ 5:025ðnÞ þ 4:97ðV Þ þ 3ðaÞ
d
Designed welding experiments are accomplished accurately by  2
selecting values through setting up a universal milling machine. D
þ 7:44 þ 2:895ðnÞ2 þ 5:74ðV Þ2 ð1Þ
Standard tensile tests are used to evaluate the test coupons. Test d
coupons made from a welded ABS sheets in triple positions
where TS is the response of tensile strength of the welded joint and
(including first, mid and end section) of a weld line. The goal of
significant factors are diameters ratio, rotational speed, linear
the investigation is to maximize the welded tensile strength up
speed, and tilt angle.
to parent material and improving the weld joint mechanical behav-
In the case of the welded parts using the conical pinned tools,
ior. Welding efficiency is defined as the ratio of yield strength of
the curvature of response represented by mathematical represen-
the welded ABS sheets to yield strength of the base ABS.
tation of Eq. (2).
Experimentation of the study is conducted using the DOE theo-
rem. Observations obtained from standard tensile tests are ana- TS ¼ 27:151 þ 4:005ðaÞ þ 3:075ðnÞ  4:685ðn  aÞ
lyzed through Statistical techniques. ANOVA determines the    
D D
main effects, contribution of factors on the response, and defining þ 4:374  a þ 3:279 V ð2Þ
d d
an empirical equation between the factors and the response, math-
ematically. The results of standard tensile tests and mechanical

Fig. 5. Contour plots of tensile strength in friction stir welded ABS using cylindrical tools.
150 N. Sadeghian, M.K. Besharati Givi / Materials and Design 67 (2015) 145–153

As presented in Eq. (2), interaction terms between factors. Com- amount of generated energy. By increasing in (the) energy level,
bination of linear terms also have significant effect on the weld- it is expected to create a suitable weld joint. Heat increases at high
ment tensile strength. ANOVA analysis is presented in Table 5. As rates of rotational speeds and increased friction surfaces. It is
summarized in Table 5, interaction of rotational speed and tilt worth mentioning that the amount of linear speed should be in a
angle has the most affect the response variation, significantly. Eq. range that the melted matter can fluidized and formed a good weld
(2) is more complicated than the Eq. (1). The graphical representa- bead. The form of weld bead is controlled by adjoin blade, and the
tion of the Eq. (2) is illustrated in Figs. 6 and 7. Fig. 6 shows the pressure from the shoulder on the melt extrudate by tilt angle.
response variation with main effects of factors variations. While Welding parameters have some effects on the weld joint expresses
in Fig. 7, the interaction effects on response variation is depicted. as following.
The mathematical expression statement of tensile strength of the
weld joint produced by conical pinned tools is presented by Eq.
(2) which is derived from ANOVA results in Table 5.
3.1.1. Pin geometry
Experimental observation results showed that welds with high
In this research, geometry of the pin proposed as a significant
rotational speed, low linear speed and deep penetrated pin cause
factor affecting the mechanical strength of the weld joint. Pin
to undercutting the weld bead and welding spotters on the sheets.
geometry varies into two levels: cylindrical and conical shape.
Moreover, low rotational speed and high linear speed produce no
Investigation of the welding and evaluation tests revealed that
joining lines at the stir zone and the traverse bond width, which
the welds created by conical pin have the highest strength, but
in turn causes to the fall of weld joint strength. Failure in generat-
the mean value of the maximal tensile strength according to the
ing heat of friction make pasty state and voids formation in the
cylindrical pin is higher than the conical one.
weld zone.
As shown in Fig. 8, there are 10 welding runs and then 30 test
As a rule, the amount of input heat must overcome to bonding
coupons are evaluated by standard tensile test. The runs produced
energy in thermoplastic structural chains and move the macromol-
using the cylindrical pinned tools as stated in Table 3. The average
ecules in microscopic scale to fluidize the melted matter. Stirring
yield strength of welded ABS sheets achieved to be about 28 MPa.
process causes to dissipate the friction heat as generated by itself.
Regression model predictor mainly has higher values than the
However, in a successful welding, generated heat is more than the
experimental results.
dissipated heat though in such a manner that in some cases, deg-
In addition, 18 welding runs and consequently 54 test coupons
radation of thermoplastic occurs by overheating. This defect
are produced using the conical pinned tools and evaluated simi-
appears by losing of materials and concave surfaces at the weld
larly. Two runs were missed because of unstable welding operation
bead and spotter dissipation onto sheets to be welded.
or weld defects generation (include welding runs 13 and 26 of
Experimental observations indicate that there is a close relation
designed experimentation in Table 3). The mean value of yield
between weld joint strength and quality improvement with the
strength of these samples is determined about 27 MPa. The regres-

Fig. 6. Contour plots of tensile strength in friction stir welded ABS using conical tools.
 N. Sadeghian, M.K. Besharati Givi / Materials and Design 67 (2015) 145–153 151

Fig. 7. Interactions of parameters on tensile strength of weld joint at ABS sheets.

Fig. 8. Comparison of welded ABS sheets using cylindrical and conical tools.

sion model predictor’s value is not always higher than experimen- for cylindrical and conical pinned tools, respectively. The findings
tally results. illustrated in Fig. 8, indicates that, from mechanical point of view,
Comparing the results illustrated in Fig. 8, indicates that the the conical pinned profile of the tool has significant effect on
30th run in Table 3 has the best improving value which is created improving of the welded joint efficiency in ABS sheets.
using a conical pinned tool, while for the case of cylindrical pinned
tools, second run in Table 3 is improved to the highest strength of 3.1.2. Diameters ratio (D/d)
the runs produced using the current pin profile. In addition, there Friction generates the heat required for FSW, and increasing in
are 40% (i.e. 4 of 10) and 55% (i.e. 10 of 18) of evaluated samples the area directly causes to the increase the generated heat. Hence,
with higher yield strength greater than the accordance mean value for higher values of D/d, more frictional heating is expected.
152 N. Sadeghian, M.K. Besharati Givi / Materials and Design 67 (2015) 145–153

According to the experimental results, this correlation is shown 3.1.5. Linear speed of welding (V)
obviously in cylindrical pinned tool as illustrated in Fig. 5. Increas- High linear speeds cannot create the melt matter at the welding
ing in D/d value leads to improve the welding response. It can be zone of FSW; and consequently decrease the weld strength. This
obviously seen from Fig. 5, that the other factors have the same happens because there is no enough time to produce melt state
effect on response less significant than the D/d. Contour plots of of thermoplastic. With lower values of linear speed due to the
Fig. 5-bd, show the effect of coupled factors on response at a same overheating and instability of welding, no desired welds can
time. As shown in Fig. 5-a, increasing in D/d will cause to increase appear. The variation of linear speed of welding is in the direction
in the response value. Therefore, improved response can be opposite to the pin profiles. For the cylindrical pinned tools, mod-
obtained at high level (i.e. increased values) of D/d. erate to high-level values has maximal effect on yield strength of
In charts of Fig. 6, graphs related to the conical pinned tools are weldment (Fig. 5-a and -c) while for the conical pinned tools, mod-
presented. Fig. 6-a shows the main effects on the response varia- erate to low level values cause to maximum results for tensile
tion. Charts of 6-b, 6-c and 6-d in Fig. 6, represent the double fac- strength of welded joints (Fig. 6-a and 6-d).
tors effect on the response variation. As stated in regression Eq. (2),
there are interactions of significant factors, which affect the 3.2. Analysis of the optimization process
response. Interactions variation of the factors for conical pinned
tools is illustrated in Fig. 7, concisely. Mathematical Eq. (2), graph- Response surface methodology (RSM) based on studying about
ical charts of Figs. 6 and 7 indicates that D/d has a duality effect on mathematical relation between parameters effect on the welding
the response improvement; which means that in some situations response is employed. Response surface methodology focuses on
low level and some situations high-level values of D/d may to studying a mathematical relationship between parameters and
improve the response for welded parts by the conical pinned tools. the response of the studied system. Thus, it is necessary to develop
an approximating model for the true response surface [15,16]. An
empirical model relating the tensile yield strength of friction stir
3.1.3. Tilt angle (alpha) welded thermoplastic of ABS sheets to the welding process param-
A closed bounded space is required for creation of FSW in order eters and tool factors is developed as below:
to generate the melt state of thermoplastics. This situation is sim-
ilar to polymer extrusion process and can be formed in the weld X
k X
k XX
k

zone. Compression of the shoulder of the tool applied to the weld y ¼ b0 þ bj xj þ bjj x2j þ bij xi xj þ  ð3Þ
j¼1 j¼1 i<j¼2
zone caused to improve the mixing of the melt matter by mechan-
ical stirring. Duplicated pressure occurs by the edge of the shoulder where  is indicates the error value and includes some effects such
on the extrudates from the weld zone prior to exit via tilt angle of as measurement error on the response, other sources of variations
the tool. that are inherent in the process or system, and so on b the coeffi-
Increasing the tilt angle value leads to increase the heat gener- cients of independent variables (x) in coded representation. The
ation. However, increasing in the tilt angle may cause an instability response (y) depends on the controllable input variables.  as a sta-
of welding due to many occurrences such as: increase in die swell- tistical error, often is assumed to have a normal distribution with
ing intension, high fluidity of the melted matter due to frictional mean zero and variance r2 [16]. Then by defining g as true response
overheating, and scatter of the melted matter from the stirring function, it is concluded as:
zone.
Selecting of proportional values of tilt angle make the melt rings X
k X
k XX
k
g ¼ b0 þ bj xj þ bjj x2j þ bij xi xj ð4Þ
around the pin, and then advancing and retreating boundaries can
j¼1 j¼1 i<j¼2
be mixed efficiently to the base material. Tilt angle plays an impor-
tant role in improving of weld joint strength in the conical pinned The amount of coefficients indicates the importance of the fac-
tools at its high levels (Eq. (2)). The effect of tilt angle on yield tor. The relationship of the parameters and the response is
strength is illustrated in charts 6-a, 6-b and 6-c of Fig. 6. Mean- achieved from statistical consideration with predictor models pre-
while, tilt angle has similarly effect on response in samples pro- sented here as regression equations. One can use ANOVA study to
duced using the cylindrical pinned tools but in its mid-level introduce a mathematical expression of an equation mathemati-
values (i.e. 1° of tilt angle). cally. P-value test is employed to determine contribution of the sig-
nificant terms of the regression models. Illustrative presentation
includes surface of the response and contour plots are the graphi-
3.1.4. Rotational speed of the tool (n) cal explanation for each optimization paths. Wherever nonlinear
Low rotational speeds create locally pasty matter and conse- contours appear one must be aware about curvature of the
quently a little melt can be formed. Then discontinuities appear response. The steepest ascent method determines the optimization
in the form of voids on the weld appearance at the advancing process for FSW of the ABS sheets.
and retreating boundaries. Increase in rotational speed causes to Significant parameters are the factors are selected, chosen
improve the heat input in welding by two procedures; Increasing among numerous parameters by screening with ANOVA and Fisher
in density of the collisions of thermoplastic particles and Increas- test. The factors are rotational and linear speeds, diameters ratio
ing in intension of mechanical stirring. These two phenomena and tilt angle. Tensile yield strength set to be the response of the
occur frequently and then cause to increase the input energy even, designed system. The ANOVA results of the experiments for the
when the welding is unstable. The effect of rotational speed on the cylindrical pinned indicated that main effects, as a quadratic
strength of weld joint is presented in graphs 5-a, and 5-b of Fig. 5 expression, have significantly affect the tensile strength of welded
for the cylindrical pinned tools in which, increasing in rotational joint in ABS sheets. Among the parameters, the diameters ratio (D/
speed causes to maximize the yield strength of the welded joint. d) has a strong effect on the mechanical strength of weldment.
However, rotational speed of the conical pinned tool has a converse Rotational speed and linear speed of welding are other significant
effect on the response improvement, compared to tilt angle. This factors. Second order terms of regression model that is presented
means that, by decreasing the rotational speed associated to in Eq. (1) introduce curvature in response surface. ANOVA results
increasing in tilt angle and D/d will improving effect on the in Table 4 and the charts of Fig. 5 indicate that, diameters ratio,
response. The illustrative represent is shown in Figs. 6-a, -b and 7. rotational and linear speeds, respectively are the most important
 N. Sadeghian, M.K. Besharati Givi / Materials and Design 67 (2015) 145–153 153

Table 6
Optimized friction stir welded of ABS sheets.

Pin profile D/d (mm/mm) Rotational speed (rpm) Linear speed (mm/min) Tilt angle (°) Tensile strength (MPa) Tensile efficiency (%)
Predicted Experimental Predicted Experimental
Conical 20/6 900 25 2 40.42 41.95 96.7 100.6
Cylindrical 20/6 1400 16 1 39.93 41.42 95.5 99.1

factors on improving of the tensile yield strength properties of strength. Finally 2° of tilt angle, 900 rpm of rotational speed, a tool
welded ABS sheets using the cylindrical pinned tools. with diameters ratio equal to 20/6 and linear speed of 25 mm/min
In the case of welded joints, which are produced using the con- generates a weld joint with approximately the same as the parent
ical pinned tools, the curvature of the response is represented by ABS sheet yield strength.
interaction terms between the significant factors. As summarized
in Table 5, charts of Figs. 6 and 7, interaction of rotational speed, Acknowledgments
tilt angle and diameters ratio, respectively has the most significant
effect on response optimization. Authors would like to thank Dr. T. Azdast to his helpful and
valuable notes on writing of manuscript. Also first author thanks
4. Conclusions Mr. M. Zamanpour at Machinery Workshop in the School of
Mechanical Engineering, College of Engineering at University of
In this article, FSW is conducted to optimally weld of ABS Tehran.
sheets. Experimental and statistical approach is used to prediction
of welding response. Significant factors are chosen using experi-
mental investigation via statistical methods. The main factors are References
diameters ratio, rotational and linear speeds and tilt angle. A cen- [1] Troughton MJ. Handbook of plastics joining a practical guide. 2nd ed. New
tral composite design of experimentation with four factors in three York: William Andrew Inc.; 2008.
levels and a two-level of block parameter (i.e. pin profile) is pro- [2] Mishra RS, Ma ZY. Friction stir welding and processing. Mater Sci Eng R
2005;50(1–2):1–78.
posed for welding of an amorphous type of thermoplastic (i.e.
[3] Mishra RS, Mahoney MW. Friction stir welding and processing. New York: ASM
ABS) sheets. Mechanical properties as evaluation criteria are stud- International; 2007.
ied for welded samples using standard tensile test according to [4] Bagheri A, Azdast T, Doniavi A. An experimental study on mechanical
ASTM: D638 method. The regression models are discussed statisti- properties of friction stir welded ABS sheets. Mater Des 2013;43:402–9.
[5] Strand SR, Sorensen CD, Nelson TW. Effects of friction stir welding on polymer
cally over experimental results, which are analyzed using ANOVA. microstructure. In: ANTEC, 2003. p. 1078–82.
Response surface methodology technique is employed to conduct [6] Sorensen CD, Nelson TW, Strand S, Johns C, Christensen J. Joining of
the optimization process over the tensile strength properties of thermoplastics with friction stir welding. In: ANTEC, 2001. p. 1031.
[7] Scialpi A, Troughton M, Andrews S, De Filippis L. In-line reciprocating friction
the welded joints. A second order model without interaction vari- stir welding of plastics. J Plas Mag 2007;1(1):42–51.
ables (Eq. (1)) presents a quadratic model of the main factors to [8] Nelson TW, Sorenson CD, Johns CJ. Friction stir welding of polymeric materials.
describe the behavior of FSW of ABS sheets using the cylindrical US 6,811,632, B2. United States Patent; November 2, 2004.
[9] Arici A, Sinmazçelýk T. Effects of double passes of the tool on friction stir
pinned tools. The fitted regression model for the welded parts welding of polyethylene. J Mater Sci 2005;40(12):3313–6.
using the conical pinned tools, (Eq. (2)) is rather complicated, [10] Kiss Z, Czigány T. Applicability of friction stir welding in polymeric materials.
because there are interaction effects terms and linear main effects Periodica Polytech Mech Eng 2007;51(No 1):15–8.
[11] Saeedy S, Besharati Givi MK, Sadeghian N. Design and evaluation of feasibility
on equation of model. study of friction stir welding of thermoplastic polypropylene sheets. In: ICME
By comparing the two models, the conical pinned tool corre- 2010. p Noshirvani University of Technology, Babol-Iran; 2010.
sponding model is more accurate than the cylindrical pinned tool [12] Idol JD, Lehman RL. Materials-polymers. In: Goswami DY, Kreith F, editors. The
CRC handbook of mechanical engineering. New York: CRC Press; 2005. p.
to predict the welding tensile strength. The response equation cor-
(12)18–40.
responded to the conical pinned predicts the optimum strength of [13] Montgomery DC. Design and analysis of experiments. New York: John Wiley &
the welded joint about 97% of experimental conditions, as summa- Sons; 2008.
rized in Table 6. The highest strength efficiency of experimentally [14] Mason RL, Gunst RF, Hess JL. Statistical design and analysis of experiments:
with applications to engineering and science. New York: John Wiley & Sons;
welded parts obtained to be about 101% and 99% in the conical and 2003.
the cylindrical pinned tools, respectively. [15] Myers RH, Montgomery DC, Anderson-Cook CM. Response surface
Tilt angle and diameters ratio in their high levels and rotational methodology: process and product optimization using designed
experiments. New York: John Wiley & Sons; 2009.
speed in its low-level conditions are optimal conditions for FSW of [16] Myers RH, Montgomery DC. Response surface methodology: process and
ABS sheets, based on the results of the present work. In addition, product in optimization using designed experiments. New York: John Wiley &
linear speed of welding at high level has significant effect on joint Sons; 2003.