Materials and Design 30 (2009) 2547–2555

Contents lists available at ScienceDirect

Materials and Design

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

Regression modeling and process analysis of resistance spot welding

on galvanized steel sheet
Luo Yi a,b,*, Liu Jinhe a, Xu Huibin b, Xiong Chengzhi b, Liu Lin b
a
School of Material Science and Engineering, Northwestern Polytechnical University, Xi’an, 710072 Shanxi, People’s Republic of China
b
School of Material Science and Engineering, Chongqing Institute of Technology, Chongqing 400050, People’s Republic of China

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

Article history: The resistance spot welding process of galvanized steel sheet used in the body manufacturing of family
Received 16 June 2008 car was investigated on the basis of mathematical models. Method of non-linear multiple orthogonal
Accepted 23 September 2008 regression assembling design was applied in experiment. The indexes studied in experiment were nugget
Available online 30 September 2008
geometry and tensile–shear strength of spot welds. Furthermore, four process parameters, namely weld-
ing current, electrode force, welding current duration and preheat current, and interactions among them
Keywords: were considered as the factors impacting indexes. The mathematical models developed were optimized
Galvanized steel sheet
by means of variance analysis. The experimental results showed that there was a more accurate predic-
Spot welding
Process parameters
tion on nugget size and mechanical properties of spot welds by the models optimized. With these predic-
Regression model tion results, the optimization of welding process also was realized by the analysis to the effect of
Nugget diameter parameters and interactions on the welding quality.
Tensile–shear strength Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction
Resistance spot welding (RSW) is one of the major welding
technologies used in the industry of conveyance manufacturing,
especially in automobile manufacturing. For example, there are
3000–6000 spot welds in a car. As existence of spatter generating
and electrode contamination during the spot welding of galvanized
steel sheet, the weldability of galvanized steel sheet is poorer than
that of ordinary steel sheet, which is a limitation to the appliance
of galvanized steel sheet and the large-scale automatic fabrication
of automotive products.
The tensile–shear strength of spot weld, which is often associ-
ated with the effective area of nugget to load, is an important index
to welding quality [1]. So, investigation on the relations between
the strength of spot weld or nugget geometry and process param-
eters is the key to solve the RSW problem of galvanized steel sheet.
To achieve this, an experimental program devised by the meth-
od of non-linear multiple orthogonal regression assembling design
was performed to reduce the cost and time, as well as to obtain the
required information about the effects and interactions of process
parameters on the nugget geometry and mechanical properties of
spot welds. And, the mathematical models were developed to pre-
dict spot weld quality for the given process parameters. These
* Corresponding author. Address: School of Material Science and Engineering,
Chongqing Institute of Technology, Chongqing 400050, People’s Republic of China.
E-mail address: luoyi@cqit.edu.cn (Y. Luo).
models will also help to evaluate the interaction effects of param-
eters and optimize the welding process design to obtain high-qual-
ity spot welds at a relatively low cost.
2. Welding process
2.1. RSW thermal effects
During the RSW, metal sheets are compressed together tightly
under the action of electrode force at the welding location. Nugget
is forming on the contact interface between specimens as a result
of the heat created by electrical resistance while the welding cur-
rent passing from the electrodes through the work [2,3]. And then
the effect of welding is realized. So the welding thermal effect
mainly relies on the three factors, namely welding current, current
duration and contact resistance. During welding, the welding ther-
mal effect Q is
Z
Q ¼ g
2
i ðtÞ  rðtÞ  dt ð1Þ
where g is welding thermal efficiency, i is welding current and r is
material resistance. So the Q is the cumulative effect of i and r to
duration of current (t).
A series of resistances constituted by the secondary circuit of
resistance welding machine and workpieces are shown in Fig. 1,
and the total resistance is
R ¼ Rc þ 2Rw þ 2Rew þ 2Re ð2Þ

0261-3069/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.matdes.2008.09.031
2548 Y. Luo et al. / Materials and Design 30 (2009) 2547–2555

Fig. 2. Specimen specification and nugget geometry, where c is indentation depth, h

h
is 1/2 nugget height, d is slab thickness, d is nugget diameter and A ¼ dc  100%.

And the mechanical properties would be varying with the ac-

Fig. 1. The resistances in RSW, where F is electrode force, Re is electrode resistance, tions. Here, the tensile–shear strength (Ft) was introduced to de-
Rew is specimen-electrode contact resistance, Rw is specimen resistance and Rc is
scribe the mechanical properties of spot weld. Referring to Fig. 3,
specimen contact resistance.
there is an eccentricity D between tensile axes of overlap joints.
The tensile stress and shear stress are all playing a role during ten-
sion test to spot weld because of the eccentricity. So, the mechan-
where the value of Re could be ignored because of the electrodes
ical properties of spot welds are described by the tensile–shear
made of copper alloy with low resistance. So, the resistance in weld-
strength.
ing circuit shown as Eq. (2) could be simplified as
The effects of welding process parameters on the values of d, A
R ¼ Rc þ 2Rw þ 2Rew ð3Þ and Ft of spot welds have a changing characterized by linearity
without regarding to the spatter and phase transition during the
where Rew, Rw and Rc are the resistance variation with the material
welding. Otherwise, it will reflect non-linear changing.
of specimens, which determines the temperature distribution and
the effect of metal melting in a spot weld during welding, as well
as the weldability [4,5]. Based on this fact, the galvanized steel 3. Experimental work
sheet has a poorer weldability against to ordinary steel sheet due
to the zinc coat with smaller contact resistance and higher conduc- 3.1. Test specimens
tive properties. In fact, the resistance heat generated on the inter-
face between workpieces has an intimate relationship with the The materials used in experiment were commercially available
contact resistance Rc. More heat generation and better weldability galvanized steel sheet widely used in car fabrication. The specimen
are produced with larger value of Rc. The Rc is not a static resistance was made as the specification of 80 mm  25 mm  1 mm refer-
but a time-varying parameter under the actions of welding current
and electrode force during welding. So the variations of quality
information are recorded in Rc varying, which is influenced by the Fτ Fτ
contact status variation with welding current and electrode force,
as well as the material property and the surface status of workpiec-
es. In view of all these mentioned above, the actions of interactions
among welding current, electrode force and duration of them on
Δ
quality of spot weld could not be ignored during the welding. So,
the nugget geometry relying on the resistance heat and mechanical
properties would depend on a certain marching between these
three factors.

2.2. Nugget geometry and mechanical property of spot weld

The same to metallurgical structure, the nugget geometry, δ

which relates to the effective area to load, has an intimate relation-
ship with mechanical properties of spot weld. Here, the parameters
of nugget diameter (d) and penetration ratio (A) were introduced to
describe the nugget geometry (Fig. 2).
Obviously, the resistance heat generated on the interface be-
Fτ Fτ
tween workspieces is influenced by the variation of welding cur-
rent, electrode force and current duration [6–8]. The values of d Fig. 3. The tensile–shear strength definition and calculation from laboratory test. D
and A would depend on the resistance heat during welding. is eccentricity and d is thickness.
 Y. Luo et al. / Materials and Design 30 (2009) 2547–2555 2549

encing Fig. 2. The two specimens were welded overlap by 20 mm
lengths. The surfaces of specimens were washed and dried before
welding.
3.2. Spot welding machine
A micro-processor controlled resistance spot welding machine
having 35 kVA capacity was used in experiment. The electrodes
were cooled by circulating water during the welding.
3.3. Welding cycle
Fig. 4 is double pulses circulation designed as welding process
in view of the existence of zinc coat with a higher electrical con-
ductivity. The first current pulse is preheating pulse (Ip) with dura-
tion of Tp. It was used to preheat workpieces, break zinc coat to a
certain extent and prepare for the following welding process. The
experiment (m0 = 2). The variation range of process parameters is
shown as follows: (1) welding current: 9000–12,000 A, (2) elec-
trode force: 0.2–0.5 MPa, (3) current duration: 8–18 cycles and
(4) preheating current: 4000–7000 A. Codes of element levels were
calculated on the basis of variation range, which is shown in Table
1. Where the value of c in asterisk experiment was calculated by
the equation:
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
sp ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðmc þ 2m þ m0 Þmc  mc
c¼ ð4Þ
2
And the value is 1.483.
The secondary level term of orthogonal regression assembling
design is Z 2ji ðj ¼ 1; 2; . . . ; 26; i ¼ 1; 2; . . . ; 26Þ which code is ex-
pressed by the symbol of z0ji . And the disposal of centralization
was calculated by the equation:
1X n

second current pulse is welding pulse (I) with duration of T. It real-
izes the effect of welding.
3.4. Experiment design
The method of non-linear multiple orthogonal regression
assembling design was used to devise experiment program to
study the mathematical regression models between process
parameters and nugget diameter, penetration ratio and tensile–
shear strength of spot welds. Total nine elements were investi-
gated in the experimental work, including four process parame-
ters: (1) welding current (x1), (2) electrode force (x2), (3) current
duration (x3) and (4) preheating current (x4), and five interactions:
(1) welding current–electrode force (x1x2), (2) welding current–
current duration (x1x3), (3) welding current–preheating current
(x1x4), (4) electrode force–current duration (x2x3) and (5) current
duration–preheating current (x3x4). And the significance of the
interactions was seen as equivalent to that of the process parame-
ters during the experiment design.
The quantity of experiment is 26, including secondary level
experiment (mc = 16), asterisk experiment (mc = 8) and zero level
z0ji ¼ z2ji  z2 ð5Þ
n i¼1 ji
The final code table of quaternionic quadratic orthogonal regression
assembling design is shown in Table 2. The experimental scheme is
given in Table 3 on the basis of Tables 1 and 2. The experimental re-
sults are also shown in Table 3. And then the mathematical models
of regression between process parameters and nugget diameter,
penetration ratio and tensile–shear strength were developed by
the method of the regression analysis.
4. Results and discussion
4.1. Development of mathematical models
Mathematical relations between experimental indexes and pro-
cess parameters can be described by the equation of
y = f(x1, x2, x3, x4). The full equation of quaternionic quadratic
regression is expressed as
y ¼ a þ b1 x1 þ b2 x2 þ b3 x3 þ b4 x4 þ b12 x1 x2 þ b13 x1 x3
þ b14 x1 x4 þ b23 x2 x3 þ b34 x3 x4 þ b11 x21 þ b22 x22 þ b33 x23
þ b44 x24 ð6Þ
The final mathematical regression models developed by calculating
regression coefficients are given as Eqs. (7)–(9):
Nugget diameter:
D ¼ 6:05 þ 0:39z1  0:51z2 þ 0:43z3 þ 0:01z4 þ 0:31z1 z2
 0:39z1 z3 þ 0:01z1 z4 þ 0:36z2 z3  0:15z3 z4  0:06z21
 0:09z22  0:31z23  0:07z24 ð7Þ
Penetration ratio:
A ¼ 61:2 þ 2:39z1  2:88z2 þ 1:23z3 þ 1:42z4 þ 0:94z1 z2

Fig. 4. Welding cycle, where F is electrode force, I is welding current, T is welding  1:56z1 z3 þ 1:25z1 z4 þ 1:56z2 z3 þ 0:63z3 z4 þ 0:52z21
current duration, Ip is preheat current, Tp is preheat current duration and Dt is pulse þ 0:52z22  1:19z23 þ 0:52z24 ð8Þ
interval.
Tensile–shear strength:

Table 1 F t ¼ 6:664 þ 0:825z1  0:462z2 þ 0:97z3  0:03z4

Factor level encode
þ 0:682z1 z2  0:611z1 z3  0:051z1 z4 þ 0:551z2 z3
Code variable (zj) Natural variable (xj)
þ 0:004z3 z4  0:116z21 þ 0:134z22  0:449z23 þ 0:345z24 ð9Þ
x1 x2 x3 x4
Upper star arm level (c) 12,000 0.5 18 7000 where the process control variables are in their coded forms: z1, z2,
Upper level (1) 11,511 0.45 16 6511 z3 and z4. The relationships of transforming between codes and pro-
Zero level (0) 10,500 0.35 13 5500
cess control variables are given as by the following equations:
Bottom level (1) 9489 0.25 10 4489
Bottom star arm level (c) 9000 0.2 8 4000 x1  10; 500
Variable spacing (Dj) 1011 0.1 3 1011 z1 ¼ ð10Þ
1011
2550 Y. Luo et al. / Materials and Design 30 (2009) 2547–2555

Table 2
Quaternionic quadratic orthogonal regression assemble design encode

Number z1 z2 z3 z4 z1z2 z1z3 z1z4 z2z3 z3z4 z01 z02 z03 z04
1 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
2 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
3 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
4 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
5 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
6 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
7 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
8 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
9 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
10 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
11 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
12 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
13 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
14 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
15 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
16 1 1 1 1 1 1 1 1 1 0.215 0.215 0.215 0.215
17 1.483 0 0 0 0 0 0 0 0 1.415 0.785 0.785 0.785
18 1.483 0 0 0 0 0 0 0 0 1.415 0.785 0.785 0.785
19 0 1.483 0 0 0 0 0 0 0 0.785 1.415 0.785 0.785
20 0 1.483 0 0 0 0 0 0 0 0.785 1.415 0.785 0.785
21 0 0 1.483 0 0 0 0 0 0 0.785 0.785 1.415 0.785
22 0 0 1.483 0 0 0 0 0 0 0.785 0.785 1.415 0.785
23 0 0 0 1.483 0 0 0 0 0 0.785 0.785 0.785 1.415
24 0 0 0 1.483 0 0 0 0 0 0.785 0.785 0.785 1.415
25 0 0 0 0 0 0 0 0 0 0.785 0.785 0.785 0.785
26 0 0 0 0 0 0 0 0 0 0.785 0.785 0.785 0.785

x2  0:35 Nugget diameter:

z2 ¼ ð11Þ
0:1
x3  13 D ¼ 6:05 þ 0:39z1  0:51z2 þ 0:43z3 þ 0:31z1 z2  0:39z1 z3
z3 ¼ ð12Þ
3 þ 0:36z2 z3  0:15z3 z4  0:09z22  0:31z23 ð14Þ
x4  5500
z4 ¼ ð13Þ Penetration ratio:
1011
A ¼ 61:2 þ 2:39z1  2:88z2 þ 1:23z3 þ 1:42z4 þ 0:94z1 z2
The significance of the factors was tested by the analysis of variance
(ANOVA) technique with F ratio calculating to ensure the reliability  1:56z1 z3 þ 1:25z1 z4 þ 1:56z2 z3  1:19z23 ð15Þ
of models. According to the results of ANOVA, the mathematical
Tensile–shear strength:
models optimized with simplifying the indistinctive factors are gi-
ven as Eqs. (14)–(16): F t ¼ 6:664 þ 0:825z1  0:462z2 þ 0:97z3 þ 0:682z1 z2
 0:611z1 z3 þ 0:551z2 z3  0:449z23 þ 0:345z24 ð16Þ
Table 3 The technique of ANOVA was used to test the significance of
Experimental conditions and result models again with desired confidence level of 95% and 99%. The
Number x1 x2 x3 x4 d (mm) A (%) Ft (kN) model is inadequate if the calculated value of F ratio of the model
1 11,511 0.45 16 6511 5.70 62.5 7.572
2 11,511 0.45 16 4489 6.00 60.0 8.123
3 11,511 0.45 10 6511 6.00 62.5 7.354
4 11,511 0.45 10 4489 5.50 55.0 7.642
5 11,511 0.25 16 6511 6.30 72.5 7.710
6 11,511 0.25 16 4489 6.60 55.0 7.559
7 11,511 0.25 10 6511 6.40 67.5 6.919
8 11,511 0.25 10 4489 6.25 65.0 7.131
9 9489 0.45 16 6511 5.55 60.0 7.175
10 9489 0.45 16 4489 6.05 55.0 7.181
11 9489 0.45 10 6511 2.70 50.0 1.730
12 9489 0.45 10 4489 2.60 45.0 2.110
13 9489 0.25 16 6511 6.10 62.5 7.196
14 9489 0.25 16 4489 6.20 62.5 7.257
15 9489 0.25 10 6511 6.10 60.0 6.816
16 9489 0.25 10 4489 5.73 60.0 6.460
17 12,000 0.35 13 5500 6.20 65.0 7.131
18 9000 0.35 13 5500 6.00 62.5 5.284
19 10,500 0.5 13 5500 5.75 62.5 6.329
20 10,500 0.2 13 5500 6.30 65.0 7.181
21 10,500 0.35 18 5500 6.05 60.0 7.558
22 10,500 0.35 8 5500 5.07 60.0 3.392
23 10,500 0.35 13 7000 6.15 60.0 7.348
24 10,500 0.35 13 4000 6.00 67.5 7.093
25 10,500 0.35 13 5500 5.95 55.0 7.358
26 10,500 0.35 13 5500 6.05 55.0 6.558 Fig. 5. Comparison of nugget diameter between measured values and calculated
values.
 Y. Luo et al. / Materials and Design 30 (2009) 2547–2555
Fig. 6. Comparison of tensile–shear strength between measured values and
calculated values.
developed does not exceed the tabulated value of F ratio for a
confidence level of 95%. If the calculated value of the F ratio of
Fig. 7. Effect of process parameters on nugget diameter: (a) welding current duration is 12 cycles, (b) electrode force is 0.3 MPa, (c) welding current is 12,000 A and (d)
electrode force is 0.3 MPa.
2551
the model developed exceeds the standard tabulated value of
the F ratio for a confidence level of 99%, then the model is very
adequate. Or the model is adequate if the calculated value of F ra-
tio of the model developed changes between the confidence level
of 95% and 99%. The test results showed that the models of nug-
get diameter and tensile–shear strength were very adequate and
the model of penetration ratio was inadequate. So it is reasonable
to use the method of non-linear multiple orthogonal regression
assembling design to analyze nugget diameter and tensile–shear
strength. And the analysis of penetration ratio should use other
method like multiple linear regression design. However, this pa-
per would not discuss the mathematical model of penetration
ratio.
The comparison figures of measured values and calculated val-
ues with models of the nugget diameter and the tensile–shear
strength are shown in Figs. 5 and 6. According to these figures,
there is a certain correspondence between the data. At the same
time, the process parameters of welding current, electrode force
and current duration are the key factors to influence the value of
nugget diameter or tensile–shear strength of spot weld. And, inter-
actions of welding current and electrode force, that of welding cur-
rent and current duration and that of electrode force and current
duration have a significant influence on the values of nugget diam-
eter or tensile–shear strength of spot weld.
2552 Y. Luo et al. / Materials and Design 30 (2009) 2547–2555
4.2. Effects of processing parameter
According to the results of multiple regression analysis, there
were relationships among the welding process parameters, namely
the welding current, electrode force, and current duration, and the
quality indexes. In order to find them, and to obtain the desired
nugget geometry and mechanical properties of spot weld, the
regression models were used to study the effects of process param-
eters on the quality indexes, which were shown in Figs. 7 and 8.
Usually, the nugget diameter grows with the welding current
increasing, which can be explained by Eq. (1). A bigger nugget
would form in view of more resistance heat generated from higher
welding current. But limited by the influence of electrode diameter
fixed, the growing of nugget tends to an extreme value. This phe-
nomenon could be testified by Fig. 7a and b: the varying curves
of nugget diameter values tend to converge to an area as welding
current increasing under the actions of different electrode forces
or current durations. The bigger nugget diameter tends to bigger
effective area to load and higher tensile–shear strength of spot
weld, which could be supported by curves in Fig. 8a and b. So
the diameter changing in Fig. 7a and b and the strength changing
in Fig. 8a and b are responding synchronously.
In Fig. 7c, the varying of nugget diameter fluctuates with elec-
trode force increasing, which is important to galvanized steel sheet
welding. The effect of electrode force on nugget diameter is not
Fig. 8. Effect of process parameters on tensile–shear strength of weld: (a) welding current duration is 12 cycles, (b) electrode force is 0.3 MPa, (c) welding current is 12,000 A
and (d) electrode force is 0.3 MPa.
obeying linear discipline under the action of certain welding cur-
rent and current duration. On the one hand, to promote electrode
force properly is beneficial to compressing workpieces tightly
and nugget generation. But increasing electrode force excessively
leads to zinc coat with low melting point to melt and to accumu-
late around the electrode to enlarge the contact area between
workpieces and electrodes. So, the resistance heat decreases be-
cause of lower current density with this phenomenon, and then
smaller nugget diameter is obtained. On the other hand, excessive
extrusion force from electrodes crushes liquid metal to turn into
spatters during nugget forming. The nugget size decreases with
this effect, which is also disadvantageous to the quality of RSW be-
cause of the decreasing of effective area to load and mechanical
properties of spot weld. So, the proper electrode force is necessary
during resistance spot welding.
Non-linear effect of current duration varying on nugget diame-
ter and tensile–shear strength of spot weld is shown in Figs. 7d and
8d. On the one hand, more resistance heat is generated with the
increasing of current duration regarding to the description of Eq.
(1), and the bigger nugget size is obtained, which is beneficial to
reliable spot weld forming with higher tensile–shear strength. On
the other hand, liquid metal during nugget forming is crushed to
turn into spatter under the action of excessive resistance heat gen-
erated with larger current duration and electrode force, which de-
creases the nugget size. And there is so much resistance heat that
 Y. Luo et al. / Materials and Design 30 (2009) 2547–2555 2553

overheating microstructure generates in the heat affected zone
(HAZ). Then, the mechanical properties of spot weld decrease with
these effects.
To some extent, the influence of welding current on indexes is
shown in Figs. 7d and 8d, too. With reference to the first part of
these curves, larger welding current is beneficial to increasing
the nugget diameter and tensile–shear strength. But in view of
the spatter and overheating microstructure, the nugget diameter
and tensile–shear strength under larger welding current are infe-
rior to that under smaller welding current shown as the latter part
of these curves. That is the effects of interactions among the weld-
ing parameters to some extent.
4.3. Interaction effects between processing parameters
There is a disagreement tendency between nugget diameter and
tensile–shear strength in Figs. 7c and 8c, which are the results of
interactions. As it shown in Fig. 7c, the values of nugget diameter
varying with electrode force are steady under the conditions of
12,000 A welding current and 12 cycles current duration, which
is advisable in process design. The reasonable marching among
welding current, current duration and electrode force would be
benefit to tensile–shear strength by reducing welding flaws and
obtaining metallurgical structure with nice mechanical property,
which could be certificated by the curves in Fig. 8c under the same
welding conditions. This analysis result was proved in experiment
for tensile–shear strength of spot welds: the strength of spot welds
was nearly equal to base metal and spot welds almost were torn on
the side of base metal. When the values of nugget diameter de-
creased to some extent under the conditions of 10 cycles current
duration, the interactions of reasonable marching among welding
process parameters are weakened. So, the tensile–shear strength
of spot weld would not be changed much. And the varying ten-
dency of nugget diameter and tensile–shear strength resumes
decreasing simultaneously under the conditions of 8 cycles current
duration. The effects of interactions among parameters are compli-
cated, which could be illustrated by means of contour lines.
The effects of the three interactions among welding process
parameters on quality indexes could be illustrated distinctly by
contour line figures in Figs. 9 and 10, which are contour line vary-
ing of nugget diameter and tensile–shear strength by the actions of
two process variables. According to the information of (a), in order
to get nice combination between nugget diameter and tensile–
shear strength under a certain current duration (12 cycles), larger
welding current and middle or lower electrode force in this range
are advisable. According to the information of (b), nice welding
quality could be obtained by larger welding current with a certain
electrode force (0.3 MPa). The reasonable selecting of current dura-
tion is in the middle of this range. The interaction between current
duration and electrode force on indexes is shown in (c) under the

Fig. 9. Effect of interaction effects on nugget diameter: (a) interaction between welding current and electrode force, welding current duration is 12 cycles, (b) interaction
between welding current and current duration, electrode force is 0.3 MPa and (c) interaction between electrode force and current duration, welding current is 12,000 A.
2554 Y. Luo et al. / Materials and Design 30 (2009) 2547–2555
Fig. 10. Effect of interaction on tensile–shear strength: (a) interaction between welding current and electrode force, welding current duration is 12 cycles, (b) interaction
between welding current and current duration, electrode force is 0.3 MPa and (c) interaction between electrode force and current duration, welding current is 12,000 A.
condition of 12,000 A welding current. It can be inferred that the
combination of contour line figures under several welding condi-
tions is a reference to welding process design. So contour line fig-
ures like Figs. 9 and 10 could provide directions for welding craft as
an assist means.
5. Conclusions
The following conclusions were drawn from
investigation:
(1) The non-linear multiple orthogonal regression assembling
design was introduced to evaluate the effects of welding
process parameters and the interactions of them on quality
indexes of spot welds. It was an effective method used in
the experimental design and the investigation for regression
models between welding process and RSW quality of galva-
nized steel sheet.
(2) The welding process and the effects of interactions of param-
eters on the indexes of nugget diameter and tensile–shear
strength can be analyzed on the basis of mathematical
regression models of welding process on 1mm thickness gal-
vanized steel sheet. And it can provide profitable reference
to welding process devising as an assistant means.
(3) The results of analysis indicated that the effects of welding
process on nugget size and mechanical properties are not
varying with simple linear curve but with non-linear charac-
teristics. The effects of interactions must not be ignored. And
the means of contour line figures can be introduced to
describe the relationship between the interactions and the
quality indexes ambiguously. Due to the complicated effects
of interactions, it was important to pay more attention to the
marching of process variables during the welding process
above design.
Acknowledgements
We would like to thank Mr. Xiang Zhangbin, Mr. Li Wanyou and
Mr. Lu Yanbin for their help in experimental work, Ms. Pi Fengxian
from South Jinkang Auto Parts Ltd. for her help and materials
provided.
References
[1] Zhou M, Zhang H, Hu SJ. Relationships between quality and attributes of spot
welds. Weld J 2003;4:72s–7s.
[2] Na SJ, Park SW. A theoretical study on electrical and thermal response in
resistance spot welding. Weld J 1996;8:233s–41s.
[3] Agashe S, Zhang H. Selection of schedules based on heat balance in resistance
spot welding. Weld J 2003;7:179s–83s.
 Y. Luo et al. / Materials and Design 30 (2009) 2547–2555 2555

[4] Thornton PH, Krause AR, Davies RG. Contact resistances in spot welding. Weld J
1996;12:402s–12s.
[5] Livshits AG. Universal quality assurance method for resistance spot welding
based on dynamic resistance. Weld J 1997;9:383s–90s.
[6] Cho Y, Rhee S. Experimental study of nugget formation in resistance spot
welding. Weld J 2003;8:195s–200s.
[7] Aslanlar S, Ogur A, Ozsarac U, Ilan E. Welding time effect on mechanical
properties of automotive sheets in electrical resistance spot welding. Mater Des
2007;28:1–5.
[8] Aslanlar S. The effect of nucleus size on mechanical properties in electrical
resistance spot welding of sheets used in automotive industry. Mater Des
2006;27:125–31.