Three Dimensional Analysis of

High Frequency Induction

Hyun-Jung Kim
e-mail: kim1392@kaist.ac.kr
Sung-Kie Youn
e-mail: skyoun@kaist.ac.kr
Department of Mechanical Engineering,
Korea Advanced Institute of Science and
Technology,
373-1 Guseong-Dong, Yuseong-Gu,
Daejeon 305-701, Republic of Korea
Welding of Steel Pipes With
Impeder
High frequency induction welding is widely employed for longitudinal seam welding of
small scale tubes and pipes due to its relatively high processing speed and efficiency. This
research is aimed at understanding the variables that affect the quality of the high
frequency induction welding. The welding variables include the welding frequency, weld
speed, vee angle, and tube thickness. Temperature distribution of the tube is calculated
through three dimensional coupled electromagnetic and thermal finite element analysis.
The skin and proximity effects are considered in the electromagnetic analysis. The influ-
ence of the impeder is also analyzed. The effects of the operating welding variables on
the temperature distribution are investigated quantitatively by exhibiting the heat affected
zone. The results explain the mechanism of significant enhancement of welding efficiency
when the impeder is used. Not only good weld state can be obtained but also overheated
edge can be avoided by understating the effect of welding variables. Suggestions are
made for the better induction welding conditions. 关DOI: 10.1115/1.2844586兴

Keywords: high frequency induction welding, electromagnetic analysis, welding

variables, eddy current, impeder, travel speed

1 Introduction Although attempts were made to analyze the tube heating by

High frequency induction welding 共HFIW兲 melts joining sur-
faces using resistance heating by induced eddy current. The strip
is formed into a circular shape in a continuous roll-forming mill
and the seam is brought together at the weld point under a small
amount of pressure at the upsetting roll stand. High frequency
current has certain characteristics such as skin and proximity ef-
fects that make it useful for welding. High frequency current tends
to flow only near the surface of the conductor at a relatively shal-
low depth 共skin effect兲. Magnetic field produced by a high fre-
quency current flowing in one direction attracts a high frequency
current flowing adjacent to the parallel surfaces in the opposite
direction by the effect of electromagnetic induction 共proximity
effect兲. Since the induced eddy current heats only a small volume
of the metal near the weld, the tubes can be welded with enhanced
welding efficiency and high speed. Although HFIW process has
been widely used in manufacturing pipes, no unified theory is
established for the prediction of performances of various param-
eters on the HFIW process. Previous works were based on empiri-
cal data and experiment.
Scott 关1,2兴 suggested that high frequency welding process can
be classified into electric mode and thermal mode based on the
mill speed, welding frequency, and vee length. Relative tempera-
ture distribution and electric power were computed at the strip
edge from the 2D mathematical model 关3,4兴. Asperheim 关5–7兴
performed the parametric study on vee angle, springback, weld
speed, and frequency with relative temperature distribution in the
cross section at the weld point.
Impeder placed inside the tube plays important roles in improv-
ing welding efficiency. Since impeder increases the reactance of
the current which flows on the inner surfaces of the tube, un-
wanted current can be restrained 关8兴.
Contributed by the Manufacturing Engineering Division of ASME for publication
in the JOURNAL OF MANUFACTURING SCIENCE AND ENGINEERING. Manuscript received
February 9, 2006; final manuscript received November 16, 2007; published online
May 6, 2008. Review conducted by Dariusz 共Darek兲 Ceglarek.
the induced eddy current, systematic three dimensional analysis
has not been conducted and the effect of the impeder has never
been thoroughly investigated. In the present work, temperature
distribution of the tube which moves with a certain travel speed is
calculated through coupled three dimensional electromagnetic and
thermal finite element 共FE兲 analysis and the effects of the welding
variables during HFIW process are investigated.
Since residual stress which causes premature failures of the
welded structure is generated at the heat affected zone 共HAZ兲,
HAZ width is a critical factor in determining good weld state.
With this viewpoint, the characteristic of the weld state is dis-
cussed based on maximum temperature, temperature distribution
at the weld point, and HAZ width.
2 Analysis Procedure
Since low-carbon steel and aluminum are widely used as tube
materials, calculations are based on AISI 1008 steel, which is
classified into low-carbon steel. The impeder is made of materials,
which are compounds with very low conductivities such as ferrite.
Since the temperature of ferrite plays a key role in determining the
life of the impeder, the impeder should be water cooled below
Curie temperature 关9兴.
As shown in Fig. 1, commercial code, EM STUDIO, is used to
solve the steady state magnetic part of the problem. The steady
state magnetic field computation gives eddy current, finally, the
dissipated power distribution, which is a used as source for a
transient-thermal computation. Skin and proximity effects are
considered in the electromagnetic analysis. The temperature field
is obtained from the heat transfer analysis by using the commer-
cial code, ABAQUS 关10兴. Heats will not only diffuse because of
conduction but also are transported 共or convected兲 down the tubes.
To handle this phenomenon, movement of heat through the mesh
is considered in the transient heat transfer analysis.
Because of inconsistency of node information between two
analysis models, eddy current density per node of the thermal

Journal of Manufacturing Science and Engineering JUNE 2008, Vol. 130 / 031005-1
Copyright © 2008 by ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 08/20/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Fig. 2 Geometric information of the tube and impeder
Fig. 1 Flowchart of the analysis procedure

dT
model is interpolated from the eddy current density of the electro- Kⵜ2T − ␳C p +q=0 共4兲
magnetic model using three dimensional shape function. The eddy dt
current density per element is represented as where K is the thermal conductivity, ␳ the density of the material,

Uel =
1
Vel
冕 unodedVel =
1
8

Vel i=1
冉 int.pt

兺 兺 N 共␰ , ␩ , ␨ 兲兩J兩w
l=1
i l l l l 冊 ui 共1兲
and C p the heat capacity. In order to deal with weld speed, time
derivative should be replaced with the spatial derivative in z to
convert the problem from coordinates relative to the tube 共La-
grangian coordinate兲 to coordinates relative to the mill 共Eulerian
where ui is the eddy current density per node, Ni the shape func- coordinate兲, which induces modified governing equation as
tion, and Uel the eddy current density per element. In the case of
the high frequency current, electrical energy 共Pec兲 in elements is
written as Kⵜ2T − ␳C P 冉 ⳵T
⳵t
+ vz 冊
⳵T
⳵z
+q=0 共5兲

Pec =
1
2
Re共J* · E兲 =
1 J* · J
2 ␴el
冉 冊 共2兲 where vz is the travel speed. If the rate of temperature change of
all nodal points is less than 0.5° C / s, the computation is stopped.
The diameter of the tube which can be produced using HFIW
where J represents the eddy current density, E the electrical field
process ranged from ␾13 mm to ␾610 mm. A 34-mm-outer diam-
intensity, * the complex conjugate operator, and ␴el the electrical
eter of the tube is used. Figure 2 shows geometric information and
conductivity. The thermal energy is calculated from the electrical
Table 1 shows material properties used in electromagnetic and
energy multiplied by the conversion factor 共␩v兲.
thermal analysis. It is assumed that a rate of welding is the same
q = ␩v Pec 共3兲 as the rate at which the edge surfaces approach each other. Thus,
the vee length is kept constant. Convection due to the temperature
where q denotes the thermal energy. Biot–Fourier equation which difference occurred through the inner and outer surfaces of the
is the governing equation of the heat transfer problem is written as tube. The conversion factor is assumed to be 1 and the melting
follows 关11兴: temperature is assumed to be 1500° C.

(a) (b)

Fig. 3 Results of electromagnetic and thermal analysis: „a… eddy current distribution and „b… temperature distribution

031005-2 / Vol. 130, JUNE 2008 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 08/20/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Table 1 Material properties

Material property
Electromagnetic analysis Thermal analysis

Tube Density 共kg/ m3兲 7,872 Room temperature 共°C兲 27

共mild Electrical conductivity 共S/m兲 850,000 Thermal conductivity 共W / m ° C兲 27
steel兲 Relative permeability 共H/m兲 1 Specific heat 共J / kg ° C兲 708

Impeder Density 共kg/ m3兲 5,000 Convection coeff. Inner surface 5.0
共ferrite兲 Electrical conductivity 共S/m兲 1 共W / m2 ° C兲 Outer surface 4.5
Relative permeability 共H/m兲 2,000

3 Results and Discussion 4 Effect of Welding Variables

Figure 3 shows results of the electromagnetic and thermal
analysis, respectively, with conditions that the vee angle is 4 deg
and the frequency of the source current 400 kHz. Because of skin
and proximity effects, the eddy current is concentrated across the
full face width of the strip edges, skelp 共see Fig. 4兲, resulting in
controlled surface melting and subsequent welding. Since the heat
element moves at the travel speed, the maximum temperature
point appears at the weld point. Figure 5 shows the isothermal
lines in a cross section of the weld vee at the weld point in the XY
plane, which include molten zone and HAZ. The bold line indi-
cates the melting temperature line, i.e., 1500° C.
In order to validate that impeder enhanced welding efficiency in
HFIW process, analysis is performed for the two cases: tube with
an impeder and without an impeder. When the impeder is em-
ployed, the temperature as well as the eddy current density sig-
nificantly increase as compared with the tube without an impeder
共see Table 2兲. In this example, as shown in Fig. 6, the tube with an
impeder can be melted while the tube without an impeder cannot.
Therefore, when the impeder is employed, it is easier to melt the
weld vee without raising the electric power.
In order to produce a high-quality weld during HFIW process,
welding variables should be adjusted such that two edges are at
the welding temperature when they reach the weld point. The
welding variables involved in this study are the welding fre-
quency, weld speed, vee angle, and tube thickness.
4.1 Welding Frequency. In this section, a parametric study of
the influence of the welding frequency is carried out. The analysis
model is based on Fig. 2 and the temperature distribution is in-
vestigated at three different frequencies: 300 kHz, 400 kHz, and
500 kHz. Figure 7 shows the value of the maximum temperature
and average HAZ width with respect to frequency. Since the skin
depth is inversely proportional to the square root of the frequency,
the induced eddy current density on the surface increases with
frequency. As shown in Figs. 7 and 8, the concentration on the
surface of high frequency current results in more heating of the
tube, and the HAZ width is reduced as the frequency increases.
Therefore, with higher welding frequency, more molten metal
would be obtainable and a better welded structure can be pro-
duced due to reduced HAZ.
4.2 Weld Speed. The weld speed, or travel speed, also affects
the temperature profile of the weld vee. Transient-thermal analysis
is performed with the weld speed ranging from
16 m / min to 20 m / min. Figure 9 shows the temperature with re-
spect to time at the weld point. Since heat generated by the work
coil is transported quickly and the heating time for the melting

Table 2 Comparison of results

Without impeder With impeder

Fig. 4 Cross section of the skelp with an impeder
Max. eddy current 共A / m2兲 8.405⫻ 108 1.175⫻ 109
Max. temperature 共°C兲 779 1627

Fig. 6 Comparison of temperature distribution at the weld

Fig. 5 Temperature distribution of HAZ and molten zone at the point between the tube without an impeder and the tube with an
weld point impeder

Journal of Manufacturing Science and Engineering JUNE 2008, Vol. 130 / 031005-3

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 08/20/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Fig. 7 Maximum temperature and average HAZ width with re-
spect to frequency „weld speed: 18 m / min…
tube is short at higher speed, the time for convergence to steady
state is less and the value of the maximum temperature appears
lower.
4.3 Vee Angle. In order to investigate how the vee angle in-
fluences the eddy current and temperature distribution on the tube,
simulations are carried out by increasing the vee angle from
3 deg to 5.5 deg. Since the melting first occurs where the maxi-
mum temperature appears, it is necessary to find out temperature
distributions around the weld point and skelp. Figure 10 shows
maximum eddy current density and temperature with respect to
the vee angle ranging from 3 deg to 5.5 deg. From 3 to 5 deg, the
maximum eddy current density and temperature exhibit higher
values as the vee angle decreases. It is because the closer the
distance between the skelps is, the stronger the proximity effect is.
However, the value of the maximum eddy current density signifi-
cantly rises at 5.5deg vee angle. The eddy current distribution at
the skelp is shown in Fig. 11. There are no dramatic differences in
the shapes of eddy current distributions for the angle between
3 deg and 5 deg. However, eddy current is considerably concen-
trated at the inner surface of the skelp at 5.5 deg vee angle, which
makes the weld state worse. This phenomenon can be explained as
follows. The strength of eddy current on the conductor is propor-
tional to the magnitude of the magnetic flux whose direction is
normal to the surface of the conductor 关12兴. In the weld vee, the
magnitude of eddy current is proportional to the inner product of
the magnetic flux B and normal vector of the skelp n 共see Fig. 12兲.
The magnitude of the inner product of B and n is proportional to
the sine value of half of the vee angle. It makes sense that the
magnitude of eddy current increases as the vee angle becomes
extremely large. Therefore, if the vee angle goes over the point
that the effect of intensity of the normal vector of the magnetic
flux to the weld vee is more dominant than the weakness of the
proximity effect, the total magnitude of eddy current increases. At
5.5 deg vee angle, the shape of the isothermal line is very differ-
Fig. 8 Temperature distribution of HAZ and molten zone at the weld point with respect to frequency „723– 1670° C…
031005-4 / Vol. 130, JUNE 2008
Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 08/20/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Fig. 9 Result of thermal transient analysis with respect to
weld speed
ent from the other cases 共see Fig. 13兲. The phenomenon that the
outer surface of the tube begins to melt earlier than the inner
surface of the tube is a clear indication that the skelp would melt
nonuniformly and the edge is overheated. Overheated edges can
cause molten metal to drop onto the impeder, reducing the im-
peder lifetime as well as the deteriorating performance. The angle
that the edge becomes overheated is defined as the critical vee
angle. Adjusting the vee angle not to exceed the critical one is
important. Through analysis, it is examined that the critical vee
angle is related to the outer diameter of the tube and the vee
length, where the vee length is defined as the distance from the
downstream end of the work coil to the weld point. The critical
vee angle has nothing to do with the employment of the impeder,
the radius of the impeder, tube thickness, and strength of the cur-
rent. Nondimensionalization of the critical vee angle is repre-
sented in Fig. 14.
␪vee angle = C1 冉 共lvee/Do兲 + 1
冊
1 − 0.222共lvee/Do兲
− C 2D o 共6兲
where lvee represents the vee length 共m兲, Do the outer diameter of
the tube 共m兲, and ␪vee angle the critical vee angle 共rad兲. The con-
stant coefficients C1 and C2 are C1 = 0.27 rad and C2
= 0.69 rad/ m. Equation 共6兲 is the experimental equation of the
critical vee angle. It indicates that if the outer diameter of the tube
and the vee length are known, the vee angle that the welding state
starts severely deteriorating can be predicted. Also, when a welder
manufacturer designs the outer diameter and the vee angle not to
exceed the critical vee angle, overheated edges can be avoided.
Figures 15 and 16 show the temperature distribution toward the
circumferential direction and average HAZ width with respect to
the vee angle, respectively. A decrease in the vee angle produces a
decreased HAZ and increased molten zone. Therefore, if a welder
manufacturer sets the vee angle as small as possible, wider molten
zone is susceptible and a better welded structure is acquired be-
cause of reduced HAZ.
Transactions of the ASME

Fig. 10 Maximum eddy current density and temperature with
respect to vee angle „weld speed: 18 m / min…
Fig. 11 Eddy current distribution at the skelp with respect to
vee angle
Fig. 13 Temperature distribution of HAZ and molten zone with respect to vee angle „723– 1750° C…
Journal of Manufacturing Science and Engineering
Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 08/20/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use
(a)
(b)
Fig. 12 Simplicity of weld vee: „a… small vee angle and „b…
large vee angle
4.4 Tube Thickness. In order to examine the temperature dis-
tribution according to tube thickness, four different sizes of the
tube are used in this research: 1.8 mm, 3.0 mm, 4.2 mm, and
5.4 mm. As shown in Fig. 17, the maximum eddy current density
and the temperature decrease as the thickness increases. Figure 18
shows eddy current distribution at the skelp. As the skelp facing
area is wider at the thick-wall tube, the proximity effect becomes
stronger at the skelp. Consequently, the maximum point of eddy
current density is obtained in the middle of the skelp. That is why
a maximum eddy current density of 4.2-mm-thick tube gets a
similar value as in the case of 3.0-mm-thick tube. The thickness
that the maximum eddy current density becomes to appear in the
middle of the skelp is defined as the critical thickness. Through
analysis, it is investigated that the critical thickness is a function
of the inner diameter of the tube, as shown in Fig. 19.
Lthick C3
= − 0.0375 共7兲
DI DI + 0.0342
where DI denotes the inner diameter of the tube 共m兲 and Lthick the
critical thickness 共m兲. The constant coefficient C3 is C3
JUNE 2008, Vol. 130 / 031005-5
 Fig. 14 Nondimensionalization of the critical vee angle
Fig. 15 Temperature toward the circumferential direction
Fig. 16 Average HAZ width with respect to vee angle
Fig. 17 Maximum eddy current density and temperature with
respect to thickness „weld speed: 18 m / min…
031005-6 / Vol. 130, JUNE 2008
Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 08/20/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Fig. 18 Eddy current distribution at the skelp with respect to
thickness
Fig. 19 Nondimensionalization of critical thickness
Fig. 20 Maximum temperature with respect to impeder radius
= 0.011 m. Equation 共7兲 is obtained from curve fitting of nondi-
mensionalization of critical thickness. In the case of thick-wall
tubes, 4.2-mm- and 5.4-mm-thick tubes, melting does not occur,
i.e., the maximum temperature could not reach the melting point.
In order to raise the temperature of these tubes, analysis is per-
formed by increasing the radius of the impeder from
12 mm to 13 mm. As shown in Figs. 20 and 21, the maximum
Transactions of the ASME
 Fig. 21 Temperature distribution of 4.2-mm- and 5.4-mm-thick tubes with respect to impeder radius
temperature can be raised by increasing the radius of the impeder.
In the HFIW process, the impeder not only enhances the welding
efficiency but also supports the magnetic field necessary to trans-
port the power from the work coil to the weld vee. As the radius
of the impeder increases, the eddy current is more concentrated on
the surface, raising the temperature. Therefore, when manufactur-
ing the thick-wall tube, the temperature can be raised by increas-
ing the radius of the impeder without increasing the extra electric
power. If the radius of the impeder becomes too large, however,
the molten metal could easily contact with the impeder. In that
case, the molten metal can drop onto the impeder, reducing the
impeder lifetime and deteriorating the performance of the im-
peder.
5 Summary and Conclusion
With an understanding of the key relationships between the
welding variables and product quality, weld defects can be signifi-
cantly reduced and more tonnage of high quality products would
be produced. In this research, HFIW process is simulated through
three dimensional coupled electromagnetic and thermal FE analy-
sis. Eddy current distribution is first computed through electro-
magnetic analysis. Temperature distribution is obtained by
transient-thermal FE analysis. Since the welding tube moves as a
travel speed, transporting heat element is considered in the simu-
lation.
When employing the impeder inside the tube, welding effi-
ciency can be enhanced and the temperature of the tube can be
raised to the required point with less electric power. When a vee
angle is as small as possible, not only a better welded structure
can be acquired because HAZ is reduced but also a wider molten
zone will be available because the proximity effect becomes stron-
ger. However, if the vee angle is increased beyond the critical one,
Journal of Manufacturing Science and Engineering
Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 08/20/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use
in the worst case, edges will be overheated and the skelp would
melt nonuniformly. To avoid the overheated edge, the vee angle
should be adjusted not to exceed the critical one by using the
empirical equation 共6兲. The tube which could not be melted can be
heated to higher temperature by increasing the radius of the im-
peder.
Acknowledgment
The authors deeply appreciate Professor Choong Don You at
the Korea Advanced Institute of Science and Technology for pro-
viding valuable comments.
References
关1兴 Scott, P. F., and Smith, W., 1995, “A Study of the Key Parameters of High
Frequency Welding,” Tube China ’95, pp. 161–181.
关2兴 Scott, P. F., 1996, “The Effects of Frequency in High Frequency Welding,”
Transactions of Tube 2000 Toronto, ITA Conference.
关3兴 Scott, P. F., 1996, “High Frequency Welding of Low Carbon Steel Tube,”
Thermatool Corporation Technical Papers.
关4兴 Scott, P. F., “Modern Methods of High Frequency Welding Used to Produce
Consistent Quality,” Thermatool Corporation Technical Papers.
关5兴 Asperheim, J. I., Grande, B., Markegaard, L., Induksjon, E., Buser, J. E., and
Lombard, P., 1998, “Temperature Distribution in the Weld Vee Cross Section,”
Tube International, 17共90兲, pp. 563–567.
关6兴 Asperheim, J. I., and Grande, B., 2000, “Temperature Evaluation of Weld Vee
Geometry and Performance,” Tube International, 19共110兲, pp. 497–502.
关7兴 Asperheim, J. I., and Grande, B., 2000, “Factors Influencing Heavy Wall Tube
Welding,” TubeNet Technical Papers.
关8兴 Davis, J. R., Ferjutz, K., and Wheaton, N. D., 1993, ASM Handbook 共Welding,
Brazing and Soldering Vol. 6兲, ASM International, Materials Park, OH.
关9兴 Scott, P. F., 1997, “Weld Area Design for High Frequency Induction Welding
Pipe and Tube,” Tube International, 16共78兲, pp. 185–190.
关10兴 Hibbitt, K., and Sorensen, Inc., 1998, ABAQUS Theory Manual, Version 5.8.
关11兴 Holman, J. P., 2000, Heat transfer, 8th ed., McGraw-Hill, New York.
关12兴 Cheng, D. K., 1992, Field and Wave Electromagnetics, 2nd ed., Addison-
Wesley, Reading, MA.
JUNE 2008, Vol. 130 / 031005-7