Available online at www.sciencedirect.

com -

Transactions of
Nonferrous Metals
Society of China
Science Trans. Nonfcrrous Met. Soc. China I7(2007) 600-605
Press wwLv.csu.edu.ctiiysxbi

Analysis of rolls deflection of Sendzimir mill by 3D FEM

LIU Xiang-hua(;(I]&lI'F)', LEE Gyoo Taek'

YU lHai-liang($~$$ Q)',
I . State Key Laboratory of Rolling and Automation, Northeastern University, Shenyang I 10004, China;
2. POSCO Technical Research Laboratories, Pohang 790-785, Korea
Received 28 June 2006; accepted 16 February 2007

Abstract: The deflection of rolls of Sendziinir mill with double AS-U-Roll was simulated by finite element niethod(FEM). The
intlucticcs of rolling pressure, strip width and rolls-assignment on rolls deflection were analyzed. The results show that thc work roll
dctlectioii increases with the increase of rolling pressure and the reduction ofwork roll radius, but the rigid displacemcnt of work roll
slightly changes; the work roll end might appear negative displacement for the narrow strip width and high rolling prcssure that
might ':;Lux the contact of work rolls. The research results are significant for guiding production and theoretical analysis of the rolls
system of Scndzimir mill.

Kei words: roll deflection; Sendzimir mill; tinitc element method; double AS-U-roll

deflection is significant for analyzing strip shapc. FEM

1 Introduction
Sendziinir inill is made up of work rolls, 1st
intcrmediate rolls ( 1 st IMR), 2nd intermediate rolls (2nd
IMK) and backup rolls, which is widely used for rolling
aluminum foil, stainless steel, Si-steel, high degree of
accuracy sheet, etc, for sinall diameter of work roll, high
rigidity and high precision of mill. Till now there have
bccn more than 400 20-high mills in the world.
Nowadays, there arc two kinds of AS-U-roll control
nicthods in Sendzimir mill: single AS-U-roll and double
AS-U-rolls. and the latter is shown in Fig. 1. Because the
rolls s,ystein of Sendzimir mill is complex, researchers
have clone many works on it, such as KIM et al[l],
ASANO et a1[2], RlNGWOOD[3], RAY et a1[4],
SCHEIDER ct a1[5], HARA[6], BERGER et a1[7],
FOREI-ILING[S], PAN[9]. They gencrally research the
strip shape control method or the rolls assignment or the
friction and failure of rolls, etc. However, there are still
not any report about the research by finite element
method( FEM).
Strip shapc control and material macrostructure
property control are two key topics of strip rolling
research. Roll deflection and roll flattening are the direct
intluence factors of strip shape. Analysis of the roll
__
Foundation item: Projectt50534020) supported by the National Natural Science Foundation of China
Corresponding author: Y lJ Ilai-liang: Tel: +86-24-8368 1803; Fax: +86-24.23906472; E-mail: yuhailiang I980@;tom.co111
[lo-131 is one of the methods used to analyze the roll
deflection. But there are no research reports about
Sendziinir inill by FEM.
Explicit dynamic FEM is widely uscd to analyze
metal deformation problems[ 14- 171 for quick
calculating speed and easily dealing with contact
problem. In this study, this method was used to simulate
the rolls deflection of Sendziinir mill, and the influences
of rolling pressure, strip width and rolls-assignment on
rolls deflection were analyzed.
2 FE simulation
During simulation, 7 schemes were employcd, and
scheme 1 was considered the criterion of comparison.
Schemes 1-3 were used to analyze the influcnccs of
rolling force on rolls deflection; scheincs I , 4 and 5 for
analyzing thc influences of strip width on rolls deflection;
and schemes I , 6 and 7 for analyzing the influences of
rolls assignment on rolls deflection.
The backup rolls A-D are AS-U-rolls that are made
up of a series of parts, such as backing bearing, mandrel,
etc. For simplifying simulation model, the backup rolls
were divided into 13 parts (7 necks, 6 barrels), as shown
in Fig.2.
 YU Hai-hang, et a1:Trdns. Nonferrous Met. SOC.China 17(2007) 60 1

LAD *

AS-U LBC

roll

Fig.1 Schematic diagram of Sendziniir mill with double AS-U-rolls

which was assumed invariablc during rolling process)

was loaded on nodes at the bottom of the work roll. The
geometrical model and meshing of rolls system were
shown in Fig.3.

!LA
Fig.2 Simplified model of backup roll

In the simulation, the main parameters are as

follows: barrel length of roll 1 346 mm, density of roll
7 850 kg/m', elastic modulus of roll 220 GPa, Poisson
ratio of roll 0.3, LAD1 086 mm, LBC415 mm, HAH241.7
mm, W 64 mm, L 171 mm. And other parameters are
listed in Table 1.
Owing to the symmetry of rolls system, the above
Fig.3 Geometry and meshing of Sendzimir mill
10 rolls (rolls A-D, I-K, 0, P, S) were included in the
geometric model. During simulation, the necks of backup
3 Results and discussion
rolls were considered to be rigid, which were constrained
to Ux,y, ~0 by AS-U racks. Other parts were considered
Owing to the symmetry of rolls A and D, B and C, 1
elastic parts. The rolls were meshed by the elements with
and J, 0 and P, the roll deflection of rolls A, B, I, J, 0
8 nodes and hexahedron, and the elements near contact and S were analyzed as follows.
positions between rolls were refined (in the simulation, Fig.4 shows the rolls deflection under variable
there were 42 contact pairs). The rolling pressure rolling forces. From Fig.4(a), the rigid displacement of
(obtained by Stone's equation according to the roll S is 0.49 mm and the maximum deflection of roll S
deformation resistance of strip and the rolling schedules, is 1.27 mm for scheme 1; similarly, 0.60 mm and 1.06
602 Y U Hai-liang. et allTrans. Nonferrous Met. Soc. China 17(2007)
Table I <icometrical Daraincters and comnutational conditions
Scheme
Paramctcr
- 1 2 3 4 5 6 I
fiadius of rolls A-Dlmm 203.2 203.2 203.2 203.2 203.2 203.2 203.2
itad i LIs of ro IIs I Kimm ~ 1 ls.u 115.0 115.0 115.0 115.0 115.0 117.2
Radius o f rolls 0-Phiin 66.5 66.5 66.5 66.5 66.5 72.0 69.2
Radius of roll S/mm 32.5 32.5 32.5 32.5 32.5 37.5 37.5
Rolling forccikN 5 x4s 5 468 4 152 S 146 4 447 5 845 5 845
Strip width/mm I 121 1 121 1 121 981 853 1 121 I 121

LI )
-~

. - Schemt. 1
- Schelnc 2
'- Scheme I
- Scheme 2
A - \theme 3 A - Scheme 3

I
0 200 400 600 xoo I 000 1 200 I 401
Barrel length ol'roll Olinin

0.22

0.20
E
.g
d
0.1x
c
E
"J
0.16
n
0

0.14

.- Schen1c I
- Schelne 2
6
0.I2

A - Scheme 3 0.10

0.08 i

' - 0 200 400 600 x00 I 000 I 200 I 400 0 200 400 600 xoo I 000 I 200 1 40
Barrel length o f roll I/mm Barrel length of' roll J/mm

c.
r:
5-
- Scheme 2

E
2.5

2.0 -
Y A A
'- Schcme
A - Schcme 3
I
- Schc111c 2

2 4- 1
3
fl
IJ 5 1.5-
EC J 3 - f?d
c: 0
--
0
2 1.0-
..-
E-2- .-9
n n
0.5 -
I-

0- 0-
- 8 d I
 Y U Hai-liang, et aliTrans. Nonferrous Met. SOC.China 17(2007)
mtn for scheme 2, 0.57 inm and 0.77 m m for scheme
3. From Fig.4(b), the rigid displacement of roll 0 is
0.41 min and the maximum deflection o f roll 0 is 0.44
mm for scheme 1 ; similarly, 0.38 m m and 0.43 tnm
for scheme 2, 0.30 mm and 0.33 mm for scheme 3. It
can be found that the roll deflection increases and the
rigid dislocation changes slightly with the increase of
rolling force. From Figs.4(c) and (d), it can be found that
the maximum deflection of roll J is smaller than that of
roll I. Similarly, the maximum roll deflection of roll B is
smaller than that of roll A.
Fig5 shows the rolls deflection under various strip
widths. When the strip width reduces from 1 121 to 987
603
mm, the rigid displacement and deflection of roll S
decrease. However, when +he strip width reduces from
987 to 8.53 mm, the displacement of cnd of roll S
becomes negative. Meanwhile, thc maximum of roll
deflection increases, which also appears in rolls 0 and I.
From Figs.S(e) and (f), the deflection of parts 2-4 of
rolls A and B changes little, but that of the parts 1 and 6
changes obviously.
Fig.6 shows the rolls deflection under variable
rolls-assignment. From Fig.6(a), the maximum of roll S
deflection increases with the reduction of roll radius, and
it increascs with the reduction of the radius of 1st IMR
when the work roll radii are equal. From Fig.6(b), the

- 0.7
-
r
-
r

2 0.5

P
. - Scl1cmc I
2
d
-P
*
'- Scheme I
j
.- 0 - f
A
- Scheme 4
- Sc11eme 5 t .z 0.3-
c A
- Scllclllc 4
- Schcmc. 5
-0.4

-0.81 ,J
f 'il
, L - 1

0 200 400 600 so0 1 000 I 200 I 100 0 2 0 0 400 000 so0 1 000 I 100 1 400
t3arrc.l length of roll S ' m m Ha1.rel length 01'r011 O/mn1

- Schelne 1 1 - Schcme I
- Schcme 4
- Schen1e 4 A - Schcmc 5
- Schemt. 5
0.05 1 I L _
J 0.02 t , , , 1 L

0 200 400 600 800 I 000 1 100 1 100 0 200 400 600 so0 I 000 I 200 1 41 L
u,'illel
.. length o F r o l l I/mm B a r r e l length ol'roll .l/mm

1 - Schcme I
- Schcme 4 - Scheme 4
A- Scheme 5 .-5 chcme 5
1

5-
-I
c

> 4-
5
3-
-m
.-$ 2 -
n
I-

0 ~

I I
604

1.8

E 1.4
-E
1
c
w
E
-
:1.0
.-2
a
0.6

0.2 L' I I ,
0 200 400 600 800 1 000 1 200 I 300 0 200 400 600 800 I 000 I 200 I 400
Barrel length of roll Simm Barrel Icngth o f roll Oinini

0.30- (

0.22 -

0.20 ~

0.18-

0.16-

0.14-

0.12-
0.101 1

0 200 400 600 8 0 0 1 000 I 200 1 4c

(e) 9 - Schcme 1 3.0 - ( O .- Scheine I

0 400 800 1200 I600 0 400 800 I 200 I 600

Barrel length of roll Aimm Barrel length of roll Wnim
Fig.6 Roll deflection under variable rolls-assignments: (a) Roll S; (b) Roll 0; (c) Roll I; (d) Roll J; (e) Roll A; ( 0 Roll B

roll deflection increases as the 1st IMR radius reduces.
From Figs.6(e) and ( f ) , the rolls have the same radius,
but they have different roll deflections. The main reason
is the change of acting force of rolls system.
The roll deflection and the rigid displacement of
work roll are important influence factors of strip exit
profile, which are affected by the 1st IMR, 2nd IMR and
backup roll radii, the acting force distribution in 20-high
Sendzimir mill, the position of AS-U racks and the
profile of roll J, etc. The AS-U racks position and the
profile of roll J are used to control the strip shape during
rolling processes. It will provide suitable theoretical
guide for adjusting AS-U-rolls to analyze the rolls
deflection without the change of AS-U racks position.
During rolling processes, there are many influence
factors of roll deflection, e.g. physical dimensions and
material attribution of rolls and strip, rolling schedules,
tensions, etc. It is undoubted that the rolling pressure, the
strip width and the rolls-assignment are the most
important factors.
There are many influencing factors of rolling force,
e.g. rolling schedule, tension, work roll radius, friction
 Y U Hai-liang, et aliTrans. Nonferrous Met. Soc. China 17(2007)
and material attribution etc. From the above results, the
rolls deflection reduces with decreasing rolling force.
When rolling the same physical dimension strip with
different material attributions, the strip with low
dcfortnation resistance should reduce the bend
deformation of backup roll by adjusting AS-U-rolls.
The strips with different widths are rolled for
variablc users. In this paper, the rolls deflection was
analyzcd with three kinds of strip widths. From the
results, the rigid displacement of the end of work roll is
not a positive value but a negative value when the strip
width i s narrow and the rolling force is big. Under this
condition. the main reason is that the roll flattening along
barrel length direction is not symmetrical. When the strip
width is narrow, the roll flattening in the roll middle part
is big and that in other parts are small, and the barrel
length i s long, so the end of work roll lowers, as shown
in Fig.7.
Fig.7 I<oll displacement with narrow strip width
For Sendzimir mill, there are hundreds of rolls-
assignment conditions. When the rolling force
distribution is identical, the acting force of rolls system
will change with the change of rolls-assignment, which
will cause variable rolls deflection, so it is needed to
adjust the AS-U-rolls. From the above results, the
increase of work roll radius will reduce the work roll
deflection. However, it might reduce the rolling power,
even can't roll some thin strip. So when the work roll
radius is not changed, it can suitably increase the radius
of I st IMR and reduce the radius of 2nd IMR.
4 Conclusions
1 ) The rolls deflection of Sendzimir mill under
different rolling forces, strip widths and rolls-
assignments was simulated by explicit dynamic FEM,
which supplies a method for dealing with complex
rolling problems.
2 ) The work roll deflection increases with the
increase of rolling force and the reduction of work roll
radius. but the rigid displacement of the work roll
605
changes little.
3) The work roll end might appear negative
displacement for narrow strip width and high rolling
pressure, which should be avoided for reducing the wear
of work roll end.
References
KIM J T, YI J J, HAN S Y. Shape control of alloy atcel rolled by
Sendziinir mill [J]. KSMt Journal. 1996. lO(3): 277-2x5.
ASANO K. TAKAHASHI H. MIYA'IA 'I.. KOIIIRO Y. Stabilization
o f tension control iii reverse rolling mills using statc feedback [J].
Tetsu To Hagane. 2004. 90( 1 I ) : 95X-963. ( i n Japanese)
RINGWOOD J V. Shape control systems for Scndmnir steel mills [J].
IEEE Transactions o n C'ontrol Systems Technology. 2000, X( I ):
70-86
RAY A, MUKHERJEE I). SARKAR t3. MISHRA S. Influence of
iiiicro~trueture on the premature failure of a 2nd intermediate
Sendziinir inill drive roll [J]. Journal o f Marcrials Engineering and
Performance. 1994. 3 ( 5 ) :049-656
SCHEIDER A, WERNERS R. Modeling and optimimtion lor a 20-h
cold rolling mill [J]. Iron and Steel Engineer. 1999. 76(3):44-47.
HARA K. YAMADA T. TAKAGI K. Shape controllability for quarter
buckles of strip in 20-high Scndmnir mills [J]. lSlJ International.
1991. 31(6). 607-613
BERGER B. SCHULTE H. t3ENFER M. RlNKli F W. Advanced
20-high inill technology [J]. Metallurgical Plant and Technology
Iiitcrnarional. 2000, 23(6): 70-72. 74.
FOREHLING P. RIECKMANN J , JATTA J . High tensile. light gage
rolling with 20-11 reversing cold mills [J]. Iron and Steel Engineer,
1099. 76(3). 41-43.
PAN C. 20-High Mill and High Precise Cold Strip Rolling [MI.
Bcijing: Metallurgical Industry Press, 2003. ( i n Chinese)
KNAPlNSKl M. FEM 3D analysia of roll deflection during plate
rolling [J]. Metalurgija, 2005. 44(3): 175-1 7X.
ZHANCi G M, XlAO H, WANG C H. 'Ihrec-dimensional model for
strip hot rolling [J]. Journal o f Iron and Steel Research International.
2006. I3( I ): 23-26.
NILSSON A. FF. simulations of camber i n hot strip rolling [J].
Journal of Materials Proccsaing Technology. I99X. X0'X I : 325-329.
XlAO W F. ZHANG L H. ZHU Z 11. FEM analysis ofcasting roller
intensity under aluniinuni strip high-speed casting condition [J]
Journal of Central South U n t \ crsity of Technology: Natural Science,
2000. 31(6): 54X-551.
YU H L, LIU X H. ZHAO X M. WU D. KUSARA Y. Explicit
dynamic FEM analysis of multipass vcnie~il-horirontal rolling [J].
Journal of Iron and Steel Research International, 2006. I3(3): 26- 30.
WANG M. YANG H, S l J N Z C, GUO L G, OU X Z. Dynamic
explicit FE modeling of hot ring rolling proceas [J]. Trans
Nonferrous Met Soc C'hina. 2006, I h ( 6 ) . 1274-1 2x0.
CHUNG W K. CHOI S K, THOMSON I' k-. I'lircc-dimensional
simulation of the edge rolling process by the explicit linite element
method [J]. Journal o f Vaterials Processing Technology. 1993.
3X( l/2): 85--l02.
LIU X B, TAN J P. YI Y P. Stress ticld analysis of extra-height
forging die using finite eleinent method [J]. Journal of Central South
University ofTechnology. 1999.6( I ) : 50-62
(Edited by CHEN Wei-ping)