Design Optimization of Roller Straightening Process for Steel Cord using Response Surface Methodology

Jong-Sup Lee, Hoon Huh, Jong-Gu Bae, Jun-Wu Lee and Deuk-Tae Kim
Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology, 373-1, Science Town, Daejeon, 305-701, Korea R&D center, KISWIRE, 533, Nam-Gu, Pohang, Kyungbuk, 790-840, Korea Abstract. A roller straightening process is a metal forming technique to improve the geometric quality of products such as straightness and flatness. The geometrical quality can be enhanced by eliminating unnecessary deformations produced during upstream manufacturing processes and minimizing any detrimental internal stress during the roller straightening process. The quality of steel cords can be achieved by the roller straightening depends the process parameters. Such process parameters are the roll intermesh, the roll pitch, the diameter of rolls, the number of rolls and the applied tension. This paper is concerned with the design optimization of the roller straightening process for steel cords with the aid of elasto-plastic finite element analysis. Effects of the process parameters on the straightness of the steel cord are investigated by the finite element analysis. Based on the analysis results, the optimization of the roller straightening process is performed by the response surface method. The roller straightening process using optimum design parameters is carried out in order to confirm the quality of the final products. Keywords: optimization, roller straightening process, steel cord, response surface methodology, finite element analysis PACS:47.11.fg

INTRODUCTION
One of the most important problems in the production of steel cords is assurance of straightness that is the basis for the successful manufacturing of final products since the straightness of the steel cord has a major contribution to the quality of final products. During the manufacturing process, the steel cord experiences severe plastic deformation. The ununiformly distributed plastic strain at the cross section of the wire makes the wire bent and deflected. On the other hand, the tensile residual stress at the outside of the wire along the longitudinal direction due to the drawing process has very detrimental effect on the fatigue life of wire products[l]. A roller straightening process is a metal forming technique to improve the geometric quality of products such as straightness and flatness. To reduce or remove the undesired the residual stress and plastic strain, the roller straightening process is indispensable for the production of the straight steel cord. During the roller straightening process, the wire is bended repeatedly and the outside of the wire is slightly stretched. This stretching of wire reduces the tensile residual stress around the outside of the wire and makes the wire straight[2].

The quality of steel cords can be achieved by the roller straightening depends the process parameters[3]. Such process parameters are the roll intermesh, the roll pitch, the diameter of rolls, the number of rolls and the applied tension[4-5]. It is well known that the design of an effective roller straightener becomes more difficult and nearly impossible by experience-based methodology as the strength of the material increases. It is indispensable to determine the process parameters systemically using successive simulations based on the optimization techniques such as the response surface method or the Taguchi method[6]. This paper is concerned with the design optimization of the roller straightening process for steel cords with the aid of elasto-plastic finite element analysis. Effects of the process parameters on the straightness of the steel cord are investigated by the finite element analysis. Based on the parameter study, the optimization of the roller straightening process is performed by the response surface method. Finally, the roller straightening process using optimum design parameters is carried out in order to confirm the quality of the final products.

CP908, NUMIFORM '07, Materials Processing and Design: Modeling, Simulation and Applications edited by J. M. A. Cesar de Sa and A. D. Santos 2007 American Institute of Physics 978-0-7354-0415-1/07/S23.00

581

4200 -, 4000-

_ ^ . -

1
ow Stre
1L

3800360034003200300028002600M1: a =2778.5 MPa M2: a =2974.7 MPa

y

M3: a =3276.4 MPa M4: a =3410.8 MPa

y

FIGURE 1. Cross sectional view of finite element model of the wire

0.000

0.002

0.004

0.006

0.008

0.0

Plastic Strain

FIGURE 3. This is the Style for Figure Captions. Center this if it doesn't run for more than one line the roll pitch, the diameter of rolls and applied tension are selected for the parameter study. FIGURE 2. Schematic diagram of the roller straightener and the wire Finite element analysis of roller straightening process For the parameter study of the roller straightening process, finite element analyses of the roller straightening process are carried out using a commercial implicit finite element code, ABAQUS/Standard. Fig. 1 and Fig. 2 show the cross sectional view of the finite element model for the analysis and the schematic diagram of the roller straightener and the wire respectively. Only half of wire is considered utilizing the symmetry. The diameter of wire is 0.30 mm and the length of finite element model for the wire is 38mm. In the real roller straightening process, the wire passes the roller straightener continuously and the length of wire is much higher than 38 mm. Since the diameter of wire is very small compared with the length of wire, extremely large number of finite elements are required to model the wire. In this study, only 38mm of wire is considered and the equivalent boundary conditions are imposed on the front and end of wire to reduce computational time. For the finite element analysis, wires with different carbon contents and area reduction ratios are used. Table 1 shows the conditions and Fig. 3 denotes the flow stress curves of the wires which are applied to the finite element analysis. TABLE 1. conditions of 4 wire materials
wire carbon contents material [wt %] Ml M2 M3 M4 0.8 0.9 0.8 0.9 area reduction ratio [%] 96.85 96.85 97.56 97.56 yield stress [MPa] 2778.5 2974.7 3276.4 3410.8

PARAMETER STUDY OF ROLLER STRIGHTENING PROCESS

The general process parameters of the roller straightening process are the roll intermesh, the roll pitch, the diameter of rolls, the number of rolls and the applied tension. Also, there is additional parameter such as groove shape of rolls and which has some effects on the geometric quality of the wire. In this study, the response surface methodology is used for the optimization of the roller straightening process. To construct the response surface of the process, finite element simulations for obtaining the response at the each design points should be performed. In general, the number of finite element analysis required for constructing the RSM increases as the number of design variables increases. It is almost impossible to optimize the roller straightening process considering the all process parameters. Prior to the optimization stage, it is necessary to investigate the effects of the process parameters on the geometrical quality of the wire and to exclude the process parameters which have very small effects from the design variables of the optimization for an efficient design process. It is easily to expect that decreasing the number of rolls keeping the quality of wire will improve efficiency of the roller straightener especially the fatigue life of rolls. In this study, the roller straightener with 11 rollers is considered and number of rolls is not changed. Among the process parameters, the roll intermesh,

582

4.0

" ---M1 3.5 - - i - M 2

0.025 _==X /Cyi X X

r
- 2.5 _. 2.0

--M3 - -T-M4

0.020

-M1,1=1.5mm M2,1=1.5mm M3,1=1.5mm M4, l=3.0mm

0.015
o

0.010
0.005

I"
i

0.000 0.5 1.0 1.5 2.0 2.5 3.0 200 400 600 800

10
0.5 0.0 Intermesh [mm]

1000

Applied tension [MPa]

FIGURE 4. maximum stress along the longitudinal direction with respect to the intermesh
0.0025

FIGURE 6. maximum stress along the longitudinal direction with respect to the applied tension
4.25

"
_______ -T

4.00 0.0020 =; 3.75

^^^\M^^^

^^v

0.0015 0.0010
0.0005

3.50
A~^^ ^ m -

-""^

85 3 "
| 2.75 0.25 0.00 1.0 1.5 2.0 2.5 200 400

M1,1=1.5mm AM2,1=1.5mm M3,1=1.5mm M4, l=3.0mm 600 800

0.0000 Intermesh [mm]

1000

Applied tension [MPa]

FIGURE 5. maximum plastic strain along the longitudinal direction with respect to the intermesh

FIGURE 7. maximum plastic strain along the longitudinal direction with respect to the applied tension

Young's modulus of the wire is 210GPa and the rolls are modeled by the analytic rigid surface option of ABAQUS/standard. The friction between the wire and rolls is neglected and the rotational degrees of freedom of rolls are removed.

Results of parameter study

The process parameters which are estimated in this study are the diameter of rolls, roll pith, intermesh of rolls and applied tension. The reference design of roller straightener is diameter of 8mm, pitch of 19 mm, intermesh of 1.5 mm and applied tension of 150MPa. At first, the effect of the intermesh of rolls is investigated and the intermesh of rolls is change from 0.5 mm to 3.0mm. Fig. 4 and Fig. 5 show the maximum stress and the maximum plastic strain along longitudinal direction of the wire as a function of the intermesh with respect to the material of wire during the roller straightening process.

From the graphs, it is easily known that as the intermesh increases, the stress and plastic strain also increase and the stress and plastic strain do not increase any more above some intermesh. It is because the curvature of wire during bending deformation determines the maximum stress at the outside of the wire and the curvature radius of the wire can not be smaller than the radius of the roller. Secondly, while the applied tension is changed from 200MPa to lOOOMPa with interval of 200MPa, finite element analyses are carried out to estimate the effect of the applied tension on the straightening quality. Fig. 6 and Fig. 7 show the maximum stress and the maximum plastic strain along longitudinal direction of the wire as a function of the applied tension with respect to the material of wire during the roller straightening process. As expected, increase of the applied tension leads the higher maximum residual stress and plastic strain. Finally, the influences of the roller pitch and roller radius are investigated.

583

4.0 3.5 i:design points

I"
& 2.5

-%

<H 2.0

I"
i

10
0.5 0.0 0.0

- R=8mm, P=19mm - R=5mm, P=13mm 0.5 1.0 1.5 2.0 2.5 3.0

^H7
L
#

k'

FIGURE 10. design points of the Box-Behnken design

Intermesh [mm]

FIGURE 8. maximum stress along the longitudinal direction with respect to the roller radius
4.0 n
3.5 a 3.0 2.5

FIGURE 10. design variables of the optimization Box-Behnken design is used to generate higher order response surfaces using fewer required runs than a normal factorial. This method essentially suppresses selected runs in an attempt to maintain the higher order surface definition. The Box-Behnken design uses the twelve middle edge nodes and three center nodes to fit a 2nd order equation. Using the responses the design points, a second order response surface can be constructed using Equation 1:

<2. 2.0

I"
1-
0.5 0 0.0 1 2 3 4 5 6 Intermesh [mm]

P=19.00mm P=23.75 mm P=28.50 mm 7 8

FIGURE 9. maximum stress along the longitudinal direction with respect to the roller pitch Fig. 8 and Fig. 9 show variation of the maximum stress along the longitudinal direction for the wire material Ml with respect to the roller pitch and the roller radius respectively. The bending stress of wire is determined by the radius of curvature and the roller radius constrains the radius of curvature. So, the increase of roller radius induces the increase of maximum stress at same intermesh. From Fig. 9, it is obvious that the roller pitch has no significant effect on the stress state of the wire.

j = / U l M + /U 2 + I ifi^xj + s (i)
i=\ i=\ i<j=2

where y and xi are response and design variables. Optimization process of roller straightener In the real manufacturing process, there are few choices about the value of roller pitch, roller radius and applied tension. But, in general the control and adjustment of the intermeshes of the each roller is very easy. In this study, the design variables are selected as the intermeshes of the each roller for the efficient optimization. Fig. 10 shows the selected intermesh of rollers. The intermesh of first and second upper rollers are fixed as 1.4 mm which is maximum value of the roller straightener with roller pitch of 13 mm and roller radius of 5mm. The basic idea is that the first and second intermeshes impose sufficient plastic deformation on the outside of the wire so that the ununiform plastic strain which originated from the upstream process can be regularized and the wire is bent consistently. After passing the first and second upper rollers, the wire experiences a planed bending deformation by the remained upper rollers and becomes straight.

OPTIMIZATION OF ROLLER STRAIGHTENER Construction of 2 nd order response surface using Box-Behnken design
For the construction of the 2nd order response surface, the Box-Behnken design is used. Fig. 10 illustrates the design points of that for 3-design variable case. The total design points are 13.

584

(a) (b) FIGURE 12. deformed shapes after elastic recovery for: a) run 6; b)runl3 Optimization results FIGURE 11. moment and stress of wire along longitudinal direction To accomplish this purpose, the sum of moment of wire should be minimized or removed by the third, forth and fifth upper rollers. Fig. 11 shows the schematic diagram of wire. The constructed objective function and constraint as follows: ize < D = <D(/ 3 ,h,h) = \h minimize JL s.t. where i
wire

Finite element analyses of the roller straightening process for each design points are carried out and the stress state of wire after elastic-recovery is used to calculate the response by Equation (2).Using the results, the 2nd order regression model is constructed as follows: / , O = 10884.58 - 469.57/3 + 7881.1614 + 23028.37/5
(3)

wlre

( JA o^ydA J dx \JA
W

(2)

/ <I <I
mm l max

i = 3,4,5

+ 4461.93/4 +12939.621/ +3238.20/4 l5-342.89/4 I5 The values of R2 and * are 0.999 and 0.998

and A are the length and cross sectional

-"-wire '-'

area of the wire respectively. / denotes the intermesh of i-th rollers and /mjn is defined as the minimum intermesh which can induce the plastic deformation on the outside of the wire. / is the maximum
max

intermesh and is determined geometrically. In this case, / and / are 0.6mm and 1.4mm respectively.
mm

max

respectively. Table 3 denotes the observed and predicted responses. Fig. 12 shows the deformed shape after elastic recovery for the run 6 and run 13. The length of wire used for the elastic-recovery is 38 mm. The design point of the smaller response value shows much smaller the deflection after elastic-recovery. The optimum design calculated from Equation (3) is showed table 4. TABLE 3. responses at the each design points
Run 1 2 3 4 5 6 7 8 9 10 11 12 13 observed response 3756.43 3971.14 934.24 278.72 3780.78 4195.41 250.21 206.84 352.99 422.73 39.52 119.83 87.50 predicted response 3756.43 3971.14 934.24 278.71 3808.43 4167.76 277.85 179.19 325.34 450.39 11.87 147.48 87.50 error [%] -2.40e-05 -2.31e-05 -6.61e-05 -2.31e-04 -0.73 0.66 -11.05 13.37 7.83 -6.54 69.97 -23.08 -8.83e-04

Table 2 is the run table of the Box-Behnken design for the optimization of the roller straightener. For all of the design cases, the applied tension is 150MPa and the wire material is Ml. The analysis tool is a commercial explicit finite element cord, ABAQUS/Explicit. Full length of wire is modeled as finite elements and the length of finite element model is 170mm. The optimization software HyperStudy is used to construct the regression model with results of finite element analysis [7]. TABLE 2. run table for FEA , unit [mm] Run
1 2 3 4 5 6 7 1.00 1.00 1.00 1.00 1.40 0.60 1.40 1.40 0.60 1.40 0.60 1.00 1.00 1.00 1.40 1.40 0.60 0.60 1.40 1.40 0.60

Run
8 9 10 11 12 13 0.60 1.40 0.60 1.40 0.60 1.00 1.00 1.40 1.40 0.60 0.60 1.00 0.60 1.00 1.00 1.00 1.00 1.00

TABLE 4. optimum design design variable values 0.85957 h

h h

0.80322 0.80608

585

 before roller straightening after roller straightening

Residual stress along longitudinal direction [GPa]

FIGURE 14. comparison of residual stress of wire between before roller straightening and after roller straightening analyses of the roller straightening process. The investigations made following conclusions: - As increase of the intermesh and the applied tension, the maximum stress of wire during the process also increases. -the maximum stress is determined by the roller radius and the applied tension. The optimization of the roller straightening process using response surface methodology with aid of finite element analysis is performed. The intermeshes of each rollers are selected as design variables. The most dominant variable is the intermesh of the last rollers. The optimum design can reduce the residual stress of wire remarkably and make the bent wire straight.

-Qt

REFERENCES
(c) FIGURE 13. response surfaces and contour plots: a) I3=0.85957; b) I4=0.80322; c)I5=0.80608; Fig. 13 show the response surface and contour plots. Since there are three design variable, one of design variable is fixed at the optimum value. I5 is most important and effective variables among the three intermesh of rollers. To confirm the leveling effect of optimum design, the finite element analysis of the roller straightening process with wire which has un-uniform longitudinal stress distribution initially. In general, the cold-drawn wire has tensile residual stress along longitudinal direction of wire[8]. Fig. 14 shows the variation of residual stresses of wire after the roller straightening process with optimum design values. The optimum design can reduce the residual stress of wire remarkably. 1. M. Elices, "Influence of residual stresses in performance of cold-drawn pearlitic wires", Journal ofMaterials Science, 39, 3889-3899 (2004) 2. E. Albert and M. Schilling, "Quality improvements during wire processing", Wire, 49, 1/98, 44-46 (1998) 3. M. Paech, "Factors relating to the wire straightening process: process and peripherals", Wire journal international, 35, 12, 64-68 (2002) 4. H. Schneidereit and M. Schilling, "Determination of the minimum number of rollers on wire straightening units", Wire, 47, 5, 318-320(1996) 5. N. K. D. Talukder and W. Johnson, "On the arrangement of rolls in cross-roll straighteners", InternationalJournal ofMechanical Science, 23,213-220 (1981). 6. H. Huh, J. H. Heo and H. W. Lee, "Optimization of a roller levelling process for A17001T9 pipes with finite element analysis and Taguchi method", International journal of machine tools & manufacture, 43, 4, 345-350 (2003) 7. HyperStudy 7.0 User's Manual, Altair Engineering. 8. J. M. Atienza et al., "Residual stresses in cold drawn ferritic rods", ScriptaMaterialia, 52, 305-309 (2005)

CONCLUSION
In this study, the influences of process parameter of roller straightening are investigated by finite element

586