Advances in Nonlinear Dynamics and Vibrations on Mechanical Systems - Research Article

Advances in Mechanical Engineering

2019, Vol. 11(5) 1–8
Ó The Author(s) 2019
Optimal design of side roller to DOI: 10.1177/1687814019846766
journals.sagepub.com/home/ade
minimize the roundness errors of lead
wire sides

Se Jong Kim and Jin Ho Kim

Abstract
This study improved the roundness error of the side portion of a lead wire produced by rolling operations. In the con-
ventional method, the side portion is cut after rolling. As an improvement, a side roller was designed for use after flat
rolling to directly form the side portion of the lead wire to reduce the roundness error. A rolling analysis of lead wires
was performed using LS-DYNA, which is a simulation program capable of analyzing the dynamic behavior of the material.
Finite element modeling of the lead wire was carried out, and the change of the cross section was observed after the
wire rolling process. The shape of the side roller was optimized to reduce the roundness error. The angle of the side
roller, the installation position, and the shape radius were selected as the design variables. PIAnO (Process Integration,
Automation, and Optimization), a commercial PIDO (Process Integration and Design Optimization) tool, was used for
the optimum design of the side roller, and the results confirmed that the roundness error was reduced.

Keywords
Side roller, optimal design, roundness error, lead wire, finite element modeling

Date received: 10 October 2018; accepted: 3 April 2019

Handling Editor: Yunn-Lin Hwang

Introduction Figure 1 shows the five-step continuous process for

The demand for high-quality fibers such as functional
fibers and micro-fibers is increasing. As a result, the
demand for flat wires with a round shape of the side
portion is greatly increasing compared to conventional
lead wires. Thus, the curvature of the side portion is
important. The conventional method for manufactur-
ing a lead wire with round sides is to cut the side part
after flattening the wire. However, this method does
not improve the sides of the wire to obtain a round
shape. The poor manufacturing environment of lead
wires has dust, noise, and smoke due to metal cutting,
so a processing method is needed to improve the pro-
duction process. Therefore, in this study, the cutting
process was removed, and a side roller was designed to
directly form the side shape of the lead wire. The side
roller was designed to make the side portion round dur-
ing the cold rolling process.1–4
obtaining the lead wire. The payoff reel on the left
serves to release the lead wire, which passes through
the roll. The tension-capstan maintains the tension of
the material to prevent warping or twisting. Finally,
the winder winds the modified lead wire. In order to
minimize the roundness error of the side of the lead
wire, a side roller was added, as shown in Figure 2.
Figure 3 shows a diagram of flat wire drawing, which
shows the lead wires passing between the upper and
lower rollers in a straight direction (Z-axis direction).5–8
School of Mechanical Engineering, Yeungnam University, Gyeongsan,
Republic of Korea
Corresponding author:
Jin Ho Kim, School of Mechanical Engineering, Yeungnam University,
Gyeongsan 38541, Republic of Korea.
Email: jinho@yu.ac.kr

Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License
(http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/
open-access-at-sage).
2 Advances in Mechanical Engineering

Figure 1. Continuous rolling pre-process line.

Figure 3. The direction of straightness of the wire and the

rolling components.

Figure 2. The cross section of a side roller.

Figure 4 shows the roller and lead wire from the direc-
tion opposite to the straight line of the lead wire. The
distance between the upper and lower rollers is defined
as the roll gap. The lead wire is subjected to compres-
sive load while passing between the rollers. The defor-
mation of the thickness is reduced, and the shape of the
cross section is changed. Figure 5 shows a diagram of
the side roller and auxiliary roller. The auxiliary roller
is designed to maintain the direction when the lead wire
is rolled in the straight direction (Z). When the lead
wire passes through the side roller at a fixed position,
the side roller rotates about the center in the Y-axis
direction to form the side portion of the wire.
Rolling was simulated using LS-DYNA, which is a
program capable of analyzing the dynamic material
behavior.9–12 Optimum design was carried out to mini-
mize the roundness error of the lead wire using PIAnO
(Process Integration, Automation, and Optimization),
which is a common tool for PIDO (Process Integration
and Design Optimization). For the optimum design,
the distance from the center of the lead wire to the side Figure 4. Definition of the roll gap.
Kim and Kim
Figure 5. Side roller and auxiliary roller.
Figure 6. Symmetric conditions of the wire.
roller was selected as a design variable, and the round-
ness error of the side of the lead wire after the rolling
simulation was selected as the objective function.
Design target
Finite element model selection and boundary
condition
The size and number of grids should be selected for the
model. The results of the simulation depend on the size
and number of grids. In order to improve the accuracy
of the simulation results, the size of the mesh is reduced,
the number of meshes is increased, and the user must
select the appropriate values. The rolling process was
carried out using a one-quarter model and applying
symmetric conditions. The analysis was performed
using a quadratic element because the deformation that
occurs during rolling of the lead wire is nonlinear.
Figure 6 shows the method of modeling a whole lead
wire under symmetric conditions. The size of the object
3
Figure 7. Lead wire with symmetry.
Figure 8. Rollers and lead wire with symmetrical conditions.
was reduced by applying symmetry in the directions of
the Y- and X-axis in the order of the green part in the
figure. Figure 7 shows the symmetric division into three
equal parts vertically and horizontally based on the
center of the finite element model of the lead wire. The
size of the grid was reduced, and the number of grids
was increased by dividing it into four parts vertically
and five parts horizontally. As shown by the red dotted
line of Figure 5, half of the upper and left side rollers
were modelled by removing the lower roller on the X-
axis and removing half of the upper roller about the Y-
axis. The side roller is removed to the right about the
Y-axis and the lower roller is removed around the X-
axis. Finally, the modeling was performed as shown in
Figure 8. In the rolling process, it was assumed that the
deformation of the roller is not large, and the analysis
time was shortened by treating it as a rigid body.
The lead wire used in the analysis had an initial dia-
meter of 1.05 mm, and 430 stainless steel material was
used in the actual rolling process. The material proper-
ties are shown in Table 1. The length of the lead wire in
the Z-axis direction is several kilometers in the real
world. However, modeling this length significantly
increases the mesh count and the simulation analysis
time. Therefore, the length was limited to 300 mm to
shorten the analysis time. Even if the length of the
model is limited, the inlet and outlet portions of the lead
wire that deform heterogeneously are excluded from the
analysis, so the analysis results are not affected.
The moving speed of the wire was set to 700 mm/s.
The coefficient of friction was selected to be 0.2 to
4 Advances in Mechanical Engineering

Table 1. Material properties.

Material 430 stainless steel

Density 7750 kg/m3
Poisson’s ratio 0.31
Yield stress 205 MPa
Tensile stress 450 MPa
Initial diameter 1.05 mm

Table 2. Analysis conditions.

Velocity 700 mm/s

Length 300 mm
Coefficient of friction 0.2
Tension 60 N Figure 9. Rolling process of the lead wire and roller modelled
at the 1/4 size.

produce the same process and conditions as the actual

rolling. When the lead wire passes between the rollers, a Table 3. Pass schedule.
tension of 60 N is applied to its surface to hold it so that
it does not warp or twist during deformation (Table 2). Roll gap (mm) Thickness reduction ratio (%)

Initial diameter 0 0
Selection of the pass schedule First rolling 0.68 35.23
Second rolling 0.5 26.47
The pass schedule shows the order of the installed roll- Third rolling 0.4 20
ers through which the wire passes when it passes the Fourth rolling 0.28 30
Fifth rolling 0.22 21.42
roll.13–15 If the thickness reduction rate is large when
Sixth rolling 0.22 0
the wire is straightened between the rollers, unbalanced
plastic deformation may occur because the load acting
on the wire is greatly affected. In order to solve this
problem, several rolling processes are carried out.
Table 3 shows the roll gaps, the distance between roll-
ers, and the reduction rate of the thickness in the pass
schedule. Six consecutive rolling operations were per-
formed. As shown in Figure 9, the rolling simulation
was performed through five flat rolling and side rolling
processes. Table 3 shows the roll gap and the thickness
reduction rate for the auxiliary roller and the sixth
rolling.

Roundness error measurement of the side part

Roundness error is the measure of how closely the
shape of an object approaches that of a mathematically
perfect circle.16–18 As shown in the last image of Figure
10, the roundness error of the semicircular shape of the
side portion was measured when the lead wire was com-
pletely rolled. A radial method was used to determine
Figure 10. Initial shape of the lead wire and expected shape
the shape of the circular part theoretically by measuring
after rolling.
the roundness error. This method was considered to be
applicable to the section of the lead wire derived from
the rolling simulation. As shown in Figure 11, the least-
areas of the inside and outside it in equal amounts. The
squares circle (LSC) method was used to obtain the
roundness error can then be estimated as the difference
roundness error. It is a circle which separates the round-
between the maximum and minimum distances from
ness profile of an object by separating the sum of total
Kim and Kim
Figure 11. Least-squares circle.
Figure 12. The 1/4 size lead wire after rolling.
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
this reference circle. Therefore, if r1 = a21 + b21 and
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r2 = a22 + b22 , then f = r2  r1 .
Figure 12 shows the cross section of the lead wire
after finishing the rolling simulation. The thickness t/2
of the lead wire was measured and the center of the
mean circle at a point t/2 away from node1 was found.
The wire section of Figure 12 is symmetrical in the hor-
izontal and vertical directions. When the LSC is
obtained from the measurement points (node1 to
node5), the difference between the maximum distance
and the minimum distance was selected as the round-
ness error.
Formulation of the design problem
Design variables and objective function selection
In order to reduce the roundness error of the lead wire,
a side roller was designed, and the side portion of the
wire was directly molded. This changes the shape of the
side portion depending on the shape of the side roller.
The design variables were the distance from the center
of the lead wire (Figure 13) to the side roller and the
radius and angle of the side roller (Figure 14). As the
side roller becomes closer to the center at the distance
(the position of the side roller, X) from the center of the
lead wire to the side roller, much deformation occurs at
the side portion of the lead wire. However, there is not
enough space for deformation, which causes distortion
or cracking. Therefore, it is necessary to adjust the
5
Figure 13. Definition of the position of the side rolling.
position of the side roller by measuring the width of the
shape of the lead wire after the fifth round rolling.
The radius and angle of the shape roller also influ-
ence the portion of the lead wire in contact with the
roller. If the radius is too small, the side portion of the
wire will be engaged with the side wire, and there will
be insufficient space for the deformation. If it is too
large, the wire will not be engaged with the deforma-
tion. Thus, the angle was selected as a variable because
the side portion of the lead wire is deformed according
to the shape of the side roller. The roundness error was
selected as the objective function. As the roundness
error becomes lower, the side portion becomes round,
and thus a lead wire with excellent performance can be
produced. Therefore, optimal design was performed to
derive the result that minimizes the roundness error
Find X , R, Degree ð1Þ
To minimize out of roundness error ð2Þ
Subject to 1:15 mm\X \1:25 mm
0:12 mm\R\0:15 mm ð3Þ
08\Degree\208
Equations (1)–(3) show the design variables, objec-
tive function, and constraints, respectively. The design
variables are X, R, and Degree, which represent the
position of the side roller, the radius of the side roller,
and the angle of the side roller shape, respectively.
Constraints were designated for each design variable.
The minimum and maximum values of the variables
were selected by analyzing the shapes of the lead wires
passing through the fifth flat rolling. The reason for
this selection is that if the value of the variable is small,
there will not be enough space for deformation to
occur. If the value is too large, the side roller and the
lead wire do not contact each other, so deformation
does not occur.
The upper and lower boundary values and the initial
values of the design parameters were selected as the
constraints. Using the optimum design, we analyzed a
6 Advances in Mechanical Engineering

Table 4. Lower limit, initial value, and upper limit of design

variables.

Design variables Lower boundary Initial Upper boundary

X (mm) 1.15 1.2 1.25

R (mm) 0.12 0.135 0.15
Degree (°) 0 10 20

Table 5. Experimental points of design variables.

R (mm) X (mm) Degree (°)

0.12 1.15 0
Figure 14. Radius and angle of the side roller selected as the 0.12 1.15 10
design variables. 0.12 1.2 0
0.12 1.2 20
0.12 1.25 10
0.12 1.25 20
case with the optimum value of each design variable. 0.135 1.15 0
We started from an initial value that was obtained as 0.135 1.15 20
the average of the maximum value and the minimum 0.135 1.2 10
value (Table 4). 0.135 1.2 20
0.135 1.25 0
0.135 1.25 10
Design of experiments 0.15 1.15 10
0.15 1.15 20
It is important to effectively distribute the test points in 0.15 1.2 0
order to obtain more accurate results when conducting 0.15 1.2 10
optimum design. We used an experimental design that 0.15 1.25 0
0.15 1.25 20
can easily determine the impact of each variable in a
design method to obtain optimal results within a short
period of time. We selected an orthogonal array pro-
vided by PIAnO to achieve accurate results through an the local optimum, we performed the optimal design as
effective point distribution. We used the orthogonal follows. As shown in Figure 15, L1–L12 of optimum
array to obtain 18 experimental points by applying points and N1–N12 of measurement points are created,
L18 (33 ), which is suitable for three variables and three respectively. Since the measurement points are changed
levels. An orthogonal array table is an efficient experi- according to every analysis, the x and y coordinates of
mental design method that can derive the minimum the measurement points are calculated in real time as
experimental points with an appropriate combination an approximate model. Then, the distance between the
of variables considering the number and level of vari- measurement point and the optimum point is found.
ables. Then, finite element analysis of the lead wires Then, the distances between L and N are calculated.
was carried out using 18 experimental points, as shown After calculating the total of 12 distances, the round-
in Table 5. ness error was analyzed and the optimal design was
performed to minimize the largest value among them.
That is, the formulation of the objective function is
The approximation technique called Minimize Max[D1, D2, D3, ., D12]. In this
The difference between maximum inscribed circle way, we predict the change of points and maintain con-
(MIC) and minimum circumscribed circle (MCC) is sistency in obtaining MIC and MCC. In this study, an
assumed to be roundness error. Positions of MIC and approximate model is created using five nodes as shown
MCC are changed according to each experimental in Figure 13.
point. Accordingly, it is not predicted what kind of An approximate model was created using the
change will occur in the optimum design process if only Kriging method, which is among the various tech-
the roundness error value is utilized in the approximate niques supported by PIAnO. The Kriging technique
model of the optimization process. As a result, the consists of the sum of the global models and the
probability of convergence to the local optimal point regional deviations based on the information obtained
increases. To reduce the possibility of convergence to from a computerized experimental design.19 The
Kim and Kim 7

Table 6. Optimum values of design variables.

Lower Initial Optimal Upper

R (mm) 0.12 0.135 0.1222 0.15

X (mm) 1.15 1.2 1.15 1.25
Degree (°) 0 10 11.28 20
Roundness – 0.0325 0.0247 –
error (mm)

Figure 15. Measurement point and optimum point.
Figure 16. Flowchart of an evolutionary algorithm.
advantage of the Kriging technique is that a nonlinear
system is approximated, and there is no parameter that
the user has to input. The deformation due to rolling of
the lead wire has nonlinear characteristics due to elastic
recovery.
The optimization techniques
Optimal design was performed using an evolutionary
algorithm by a generated approximate method.20 An
2XWRIURXQGQHVVPP








,QLWLDO 2SWBPHWD 2SWBUHDO
Figure 17. Comparison of roundness error values.
evolutionary algorithm is one of the optimization meth-
ods. This method generates the next population by gen-
erating a random variable within a certain range from a
parent population derived through the approximation
method. The process of determining whether the next
population group generated converges closely to the
objective function is repeated to determine the optimum
design variable closest to the objective function. Figure
16 shows a flowchart of the evolutionary algorithm.
The optimal values of the design variables obtained
using the algorithm are shown in Table 6.
The optimum design result
Figure 17 shows the result of the optimal design using
the approximate model. The initial roundness error
decreased by 31.57% from 0.0325 mm. In order to
compare the optimum design result with the actual
analysis result, we analyzed the behavior of the mate-
rial using LS-DYNA. The lead wire analysis was per-
formed with the values obtained using the optimum
design. As a result of the analysis, the roundness error
was 0.0223 mm and the result obtained by the optimum
design had a 9.7% error.
8 Advances in Mechanical Engineering
Conclusion
In this study, we carried out optimal design to minimize
the roundness error of the side portion when rolling the
lead wire. Finite element analysis was performed using
LS-DYNA. The optimal design was accomplished
through PIAnO. The experimental design method was
completed by distributing 18 experimental points using
an orthogonal array table, and a Kriging model was
created by an approximation method. Next, we used an
evolutionary algorithm as an optimization technique to
derive the optimum value that minimizes the roundness
error of the side of the lead wire.
The initial roundness error value was 0.0325 mm.
The optimal roundness error value was 0.0247 mm,
which is 31.576% lower than the initial value. In order
to compare the optimum design results with the actual
analysis results, the behavior of the material was ana-
lyzed by applying optimal design variables. As a result
of the analysis, the roundness error was 0.0223 mm and
the optimum design result had an error of 9.7%.
In order to produce the round lead wire, the side part
was directly formed using a side roller, and the opti-
mum design was used to design the side rolling to mini-
mize the roundness error. As a result, the roundness
error was improved.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of this
article.
Funding
The author(s) disclosed receipt of the following financial sup-
port for the research, authorship, and/or publication of this
article: This research was supported by a Yeungnam
University Research Grant in 2018.
ORCID iD
Jin Ho Kim https://orcid.org/0000-0002-6371-2828
References
1. Komori K and Koumura K. Simulation of deformation
and temperature in multi-pass H-shape rolling. J Mater
Process Technol 2000; 105: 24–31.
2. Kusakabe T and Mihara Y. Analysis of residual stresses in
hot-rolled H-shapes. Tetsu-to-Hagane´ 1979; 65: 1375–1382.
3. Yoshida H, Sasaki T, Kondo N, et al. Analysis of resi-
dual stress in rolled H-beam. Tetsu-to-Hagane 1983; 69:
412–419.
4. Kim HJ, Kim TH and Hwang SM. A new free surface
scheme for analysis of plastic deformation in shape roll-
ing. J Mater Process Technol 2000; 104: 81–93.
5. Massé T, Chastel Y, Montmitonnent P, et al. Impact of
mechanical anisotropy on the geometry of flat-rolled
fully pearlitic steel wires. J Mater Process Technol 2011;
211: 103–112.
6. Carlsson B. The contact pressure distribution in flat roll-
ing of wire. J Mater Process Technol 1998; 73: 1–6.
7. Kazeminezhad M and Taheri AK. An experimental
investigation on the deformation behavior during wire
flat rolling process. J Mater Process Technol 2005; 160:
313–320.
8. Kazeminezhad M, Taheri AK and Tieu AK. A study on
the cross-sectional profile of flat rolled wire. J Mater
Process Technol 2008; 200: 325–330.
9. Komori K. Rigid-plastic finite-element method for analy-
sis of three-dimensional rolling that requires small mem-
ory capacity. Int J Mech Sci 1998; 40: 479–491.
10. Iankov R. Finite element simulation of profile rolling of
wire. J Mater Process Technol 2003; 142: 355–361.
11. Kim JM, Park CS and Woo KM. Rolling process and roll
stress analysis of CB100 using by 3D FEM. Trans Mater
Process 2010; 19: 101–106.
12. Liu X-H, Shi X, Li S-Q, et al. FEM analysis of rolling
pressure along strip width in cold rolling process. J Iron
Steel Res Int 2007; 14: 22–26.
13. Komori K. Simulation of deformation and temperature
in multi-pass caliber rolling. J Mater Process Technol
1997; 71: 329–336.
14. Komori K. Simulation of deformation and temperature
in multi-pass three-roll rolling. J Mater Process Technol
1999; 92: 450–457.
15. Lee S-H, Ok SY, Hwang IY, et al. Design of rolling pass
schedule in copper thin foil cold rolling according to roll
crown of 6 high mill. J Korea Soc Precis Eng 2008; 25:
66–72.
16. Lai J-Y and Chen H. Minimum zone evaluation of circles
and cylinders. Int J Mach Tool Manuf 1996; 36: 435–451.
17. Kim S-K and Jun J-U. A study for the roundness estima-
tion. J Korea Soc Manuf Process Eng 2011; 10: 38–45.
18. Rossi A, Antonetti M, Barloscio M, et al. Fast genetic
algorithm for roundness evaluation by the minimum zone
tolerance (MZT) method. Measurement 2011; 44:
1243–1252.
19. Kleijnen JPC. Kriging metamodeling in simulation: a
review. Eur J Operat Res 2009; 192: 707–716.
20. Mallipeddi R, Suganthan PN, Pan QK, et al. Differential
evolution algorithm with ensemble of parameters and muta-
tion strategies. Appl Soft Comput 2011; 11: 1679–1696.