Strain variations on rolling condition in accumulative roll-bonding by inite element analysis 589

24
x

Strain variations on rolling condition

in accumulative roll-bonding by
finite element analysis
Tadanobu INOUE
National Institute for Materials Science,
Japan

1. Introduction
Bulk ultrafine-grained (UFG) materials with grain sizes of tens to hundreds of nanometers
showing improved mechanical properties without the addition of alloying elements have
attracted the attention of researchers in materials science (Inoue et al., 2010a; Kimura et al.,
2008), and microstructural evolution and hardness variation in the UFG materials fabricated
by a plastic deformation process such as equal-channel angular pressing (ECAP) (Horita et
al., 2000; Segal, 1995), accumulative roll bonding (ARB) (Saito et al., 1998), caliber rolling
(Inoue et al., 2007b; 2009c; Mukai et al., 2010), and high-pressure torsion (HPT) shown in Fig.
1, have been studied in detail (Inoue et al., 2009a; 2010b; Todaka et al., 2008). Since the
microstructural evolution of plastically deformed materials is directly related to the
magnitude of plastic deformation, the understanding of the phenomenon associated with
the strain development is very important (Inoue et al., 2001; 2007a; 2008). For example, in
ARB which is a severe plastic deformation process for realizing UFG microstructures in
metals and alloys, the microstructure and texture in a sheet processed by one ARB cycle
without lubricant dramatically change depending on the thickness location of the sheet
(Kamikawa et al., 2007). In a rolling process, including ARB, it is reported that these
changes are caused by the redundant shear strain imposed by large friction between rolls
and sheet.
Rolling is an excellent plastic deformation process in the mass production of a metallic sheet,
and many reports have been published regarding the rolling characteristics, shape control,
and microstructure control through theory, numerical simulations, and many experiments.
Figure 2 shows a typical illustration of a rolling process. The classical rolling theory
(Underwood, 1952) has been used as a method to qualitatively understand variations of the
rolling characteristics of a rolling force and torque against the processing parameters such as
the roll diameter, reduction, rolling speed, and friction condition. On the other hand,
deformation in sheet metals by rolling has been studied in detail through many experiments
and finite element simulations. Flow of metals in rolling experiments makes the presence of
the shear deformation clear and qualitative relation between equivalent strain including
shear deformation and microstructure through sheet thickness has been reported in the
literature (Lee et al., 2002; Sakai et al., 1988; Matsuoka et al., 1997). However, only a few

www.intechopen.com
590 Finite Element Analysis

studies have been carried out on quantitative correlation between microstructure and strain
through a combination of experiments and finite element simulations (Um et al., 2000;
Mukhopadhyay et al., 2007). The embedded-pin method (see Fig. 3) is often employed to
measure the shear strain through thickness experimentally (Cui & Ohori, 2000; Hashimoto
et al., 1998), but magnitude of shear strain and equivalent strain obtained by this method do
not exhibit the exact value (Inoue & Tsuji, 2009b). Therefore, for controlling the
microstructures, it is essential to understand the deformation behavior in the sheet
accurately and quantitatively through a combination of rolling experiments and finite
element simulations.

(a) ECAP (b) ARB (c) Caliber rolling (d) HPT

upper roll

Rolling
Torsion
sample
pressure pressure

sample
Cutting
lower roll
90° 90°
Bonding

sample
Initial After Before After
sample 1st-pass 2nd-pass 2nd-pass

Fig. 1. Major severe plastic deformation processes.

upper Roll
roll diameter: d

angle : 
Contact

 d (t0  t1 )
Projected contact length : Ld 
pin
f from y axis : 
NP:neutral plane Angle of inclination
f  2
 t0  t1
Average thickness : td 
RD
t0 t1
sheet
f=0  2
center in RD : L  t0  t1
Distance from
y  Nominal reduction : r 
Ld  t0

lower
roll

Fig. 2. Schematic illustration showing the geometry of the rolling process.

The shear strain is caused by not only the friction between the rolls and the material surface
but also the roll bite geometry, Ld/td (Backofen, 1972; Dieter, 1988). Here, Ld denotes the
projected length of the contact arc to the horizontal plane, and td is the average sheet
thickness shown in Fig. 2. In interstitial free (IF) steel sheets rolled by 50% with and without
lubrication, Um et al., 2000 investigated the variations of the strain distribution and the
flection of an embedded pin against the roll bite geometry by changing the initial sheet
thickness and roll diameter. However, the distribution of the shear strain is not shown and

www.intechopen.com
Strain variations on rolling condition in accumulative roll-bonding by inite element analysis 591

the effect of the roll bite geometry on strain distribution in sheet rolling is not systematically
studied. Furthermore, it is not verified that the roll bite geometry is a universal parameter
on the strain distribution under unlubricated condition. Hence, it is important to
systematically explore the effect of the Ld/td ratio on the magnitude and distribution of
strains under various friction conditions using numerical simulations. Moreover, if the
magnitude and distribution of strains through the thickness in a rolled sheet can be
quantitatively estimated by using experimental data (Kamikawa et al., 2007; Sakai et al.,
1988), L and , measured from the embedded-pin method, as shown in Fig. 3, a
simulation result would provide useful guidelines for analyzing the evolution of
microstructures in the ARB process as well as designing the microstructure of the sheet
metal by conventional rolling.

(a) Without lubricant (b) With lubricant

L 
surface wire

center

surface
Rolling direction 1 mm

Fig. 3. Flection of stainless wire in sheet after rolling observed by the embedded pin method
(Hashimoto et al., 1998). Here, interstitial free steel of 10 mm in initial thickness was rolled
by a reduction of 40% per pass at 973 K using a two-high mill with a roll diameter of 300
mm at a rolling speed of 300 mm s-1. The Ld/td indicates 3.1 under this rolling condition.

This study aims to exhibit a quantitative correlation between strains and Ld/td in metal sheet
rolled under various friction coefficients using finite element analysis (FEA), which is a
powerful tool for understanding deformation behaviors during a plastic deformation
process. The each strain component and equivalent strain at various thickness locations in
the sheet during and after rolling were studied in detail, including L and  measured
from the embedded-pin method in rolling experiments. Furthermore, the problem
associated with universality of the Ld/td parameter on the magnitude and distribution of
strains is discussed.

2. Finite element model

The elastic-plastic FE simulation was carried out using the FE-code ABAQUS/Explicit
ver.6.5-4. A 4-node linear element in a plane strain model, element type CPE4R, was used
for sheets of 2 mm and 5.3 mm in initial thickness, t0, and the rolls were regarded as the
rigid body. No remeshing was carried out in the analysis because the deformed mesh by
rolling corresponds to the flection of the pin in the embedded-pin method. The Coulomb
condition was used as the frictional condition between the rolls and the sheet, f =p, where

www.intechopen.com
592 Finite Element Analysis

f denotes the shear stress,  is the friction coefficient, and p is the contact pressure.
Assuming the Coulomb law, a condition to pull a sheet into rolls bite is given by > tan ,
where  denotes the contact angle shown in Fig. 2.
The flection of the pin in the embedded-pin method makes the presence of the shear
deformation clear. The pin flection in pure aluminum (Lee et al., 2002), interstitial free (IF)
steel (Matsuoka et al., 1997; Sakai et al., 1988; Um et al., 2000; Kamikawa et al., 2007), and
Type 304 stainless steel (Zhang et al., 1996; Sakai & Saito, 1999) sheets rolled at various
temperatures has been observed using this method. In the present study, the condition of a
commercial 1100 Al sheet rolled at ambient temperature without lubricant and without front
and back tensions reported by Lee et al., 2002 was referred to as the main rolling condition:
initial thickness, t0=2 mm; nominal reduction per pass, r=50%; roll diameter, d=255 mm;
and rolling speed, 170 mm s-1. The Ld/td indicates 7.5 under this rolling condition, and its
value becomes smaller with decreasing d. Nine rolling conditions, Numbers 1-9, employed
in the present analysis are listed in Table 1. Numbers 1-6 denote the rolling conditions in
which the 1100 Al of t0=2 mm is rolled by r=50% using a rolling simulator with d=310, 255,
201, 118, 40, and 15 mm, resulting in Ld/td=8.3, 7.5, 6.7, 5.1, 3.0, and 1.8, respectively. In
order to study the universality of the Ld/td parameter on the strains through sheet thickness,
Nos. 7-9 at Ld/td = 5.1 are also analyzed under various combinations of t0, r, and d. The
stress strain  relationships of the 1100 Al at 301 K employed in the analysis were
described by =28+105.67 0.32  0.017 MPa depending on the strain rate,  (Ataka, 2006), and
a Young’s modulus of 70 GPa and a Poisson’s ratio of 0.35 were used as the elastic modulus.

Initial Average Rolling Minimum

sheet Exit Nominal Roll Projected sheet Roll bite speed friction
thickness thickness reduction diameter length thickness geometry v coefficient
No. t0 (mm) t1 (mm) r (%) d (mm) Ld (mm) td (mm) Ld/td (mm/s) min=tan
1 310 12.4 8.3 0.08
2 255 11.3 7.5 0.09
3 201 10.0 6.7 0.10
1 50 1.500
4 118 7.7 5.1 0.13
2
5 40 4.4 3.0 170 0.23
6 15 2.7 1.8 0.38
7 1.41 29.5 255 8.7 1.705 0.07
8 1.76 12 768 9.6 1.880 5.1 0.03
9 5.3 2.65 50 310 20.2 3.975 0.13
Table 1. Rolling conditions used in the present study.

In the analysis, the classical metal plasticity models with Mises yield surface, *PLASTIC,
HARDENING=ISOTROPIC as keywordin ABAQUS/Explicit, 2006, were employed. The
equivalent strain, eq, imposed by rolling is defined as follows:

d  eq
 eq  
t ( steady )
dt (1)
0 dt

where deq/dt denotes the incremental equivalent strain, and t(steady) is the rolling time.
Since an incremental strain in the x direction, dxx/dt, is equal to a minus incremental strain
in the y direction, dyy/dt, under the plane strain condition, the incremental equivalent
strain, i.e., the equivalent strain rate, deq/dt, is represented as below:

www.intechopen.com
Strain variations on rolling condition in accumulative roll-bonding by inite element analysis 593

d  eq 2  d  xx  1  d  xy 
   
2


2


(2)
dt 3  dt  4  dt 

In equation (2), dxy/dt is the incremental shear strain. Since the direction of the shear stress,
f, in a roll bite changes to opposite directions before and after a neutral plane NP, shown in
Fig. 2, the total shear strain, , must be expressed as follows:

d  xy d  xy
  dt  
t ( NP ) t ( steady )
dt (3)
0 dt t ( NP ) dt

Here, the first term in the above equation denotes a positive shear strain, +, induced by the
shear stress, f, before NP, and the second term is the negative shear strain, , by the f after
NP. In other words, the  represents the total magnitude of the shear strain xy taking into
account the deformation history during rolling. At the thickness center with no shear
deformation, the dxy/dt is always zero, and the dxx/dt is constant throughout rolling.
Hence, eq at the center agrees with the value 2 / 3 ln{1(1r)} calculated simply from a
reduction in thickness independent of the deformation history, where xx= yy= ln{1/(1r)}
and  = 0.

3. Simulation results
3.1 Mesh dependence of strain in rolled sheet
At first, the appropriate mesh division in the FEA was examined because the magnitude of
strain depends certainly on mesh size. Figure 4 shows the variation of the equivalent stain,
eq, at a surface against the initial element length in the thickness direction, tel, for FEA using
=0.3 under the rolling condition No. 2 in Table 1, where the initial element length in the
rolling direction, Lel , is 0.0231 mm. The figure inset describes the FE mesh near the sheet
surface. It is found that eq at the surface depends strongly on the tel as expected. The eq
increases with decreasing tel and tends to be almost constant at tel below 0.03 mm. On the
other hand, eq at the thickness center exhibited about 0.80 regardless of tel below 0.26 mm.
This magnitude corresponds to a value, 2 / 3 ln{1/(1r)}, of equivalent strain calculated
theoretically under the plane strain condition, where the reduction r is 0.5 in the present
condition. In the present analysis, the sheet thickness was divided into 66 elements, i.e.,
tel=0.03 mm was used throughout the sheet. Moreover, the sheet length was determined
through some simulations on a steady-state of deformation where the strains remains
constant along the RD. As a result, the finite element mesh in the sheet with dimensions of 2
mmt×15 mmL included 20167 nodes and 19800 elements as illustrated in Fig. 5. Here, a
minimum Lel is 0.0231 mm at the mid-length (center element), and Lel gradually increases
toward the front and back from the center. The Lel at the front and back edges of the sheet is
the same 0.0923 mm. Similar, for the rolling condition No. 9 of t0=5.3 mm in Table 1, tel=0.03
mm was used throughout the sheet. Thus, the finite element mesh in the sheet with
dimensions of 5.3 mmt×15 mmL included 26488 nodes and 26100 elements.

www.intechopen.com
594 Finite Element Analysis

Initial mesh near sheet surface

Lel=0.0231mm
surface Lel=0.0231mm
tel=0.03mm
5 tel=0.1mm

surface
Center line
4
Equivalent strain at
3
eq

ln 1/(1  r )
2
3
1

0
0 0.05 0.1 0.15 0.2
Element length in thickness direction tel (mm)
Fig. 4. Variation of eq at surface in the rolled Al with =0.3 against initial element length in
thickness direction. Here, rolling condition No.2 in Table 2 was used.

Lel=0.0231mm Lel=0.0923mm

tel=0.03mm

Center mesh
x Center line

Back area Front area

Lel variable (>0.0231) Lel variable (>0.0231)

RD

Fig. 5. Initial finite element mesh used in the analysis. Here, Lel denotes the mesh length in
rolling direction (RD).

3.2 Variations of Ls, s, and eq at the sheet surface with roll bite geometry
Figure 6 shows the effect of the Ld/td ratio on the distance from center to surface in RD, Ls,
the angle of inclination from the y axis, s, and the equivalent stain, eq, at the surface of a
50% rolled Al sheet in the friction coefficient range of tan  0.4. The slip areas in Fig. 6
denote a condition of < tan  on the basis of the Coulomb law, and the open symbol in Fig.
6(a) represents the observed result, Ls = 1.1, of the pin method reported by Lee et al., 2002.
Although the Ls have no large difference by Ld/td under a low friction of  0.2, this
difference becomes larger with increasing Ld/td under high friction. Similarly, in Fig. 6(b),
the s becomes larger with increasing Ld/td and . However, under Ld/td = 8.3 of the
largest roll bite geometry in the present study, the s decreases from = 0.35 to 0.4.

www.intechopen.com
Strain variations on rolling condition in accumulative roll-bonding by inite element analysis 595

(a) 2
(c) 6
Ld/td=8.3
(d=310mm)
Ld/td=8.3

to surface in RD, Ls (mm)

Equivalent strain at surface,  eq

(d=310mm)
7.5(255)
5
1.5 7.5(255)
Lee et al., 2002 6.7(201)
6.7(201) 4

Distance from center 1 3 5.1(118)

5.1(118)

2
0.5 3.0(40)
3.0(40)
1 1.8(15)
1.8(15) 2  1 
ln  
3  1 r 
slip area slip area
0 0
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

Friction coefficient,  Friction coefficient, 

(b)
Angle of inclinatioon at surface, s (deg.)

90 5.1(118)
6.7(201)
s
75
7.5(255)
60
surface
Ld/td=8.3 3.0(40)
(d=310mm)

Ls
45 1.8(15)
y
30 center
x(RD)
slip area
15
t0=2 mm → t1=1 mm
0
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

Friction coefficient, 

Fig. 6. (a) Ls, (b) s, and (c) eq at the surface of a 50% rolled Al sheet against  under
different roll bite geometries Ld/td (Nos.1-6 shown in Table 1).

Furthermore, the s indicates a large value of 45° at = 0.4 even if the roll bite geometry is
very small, Ld/td =1.8. The eq increases with increasing Ld/td under the same  Although
the tendency is slight under 0.2, it becomes remarkable under 0.2 at Ld/td 5.1. At
= 0.4 of a high friction condition, the eq exhibits a very large value of 6.0, which is seven
times higher than that at the center, 2 / 3 ln{1/(1r)}, under Ld/td = 8.3. On the other hand,
under Ld/td = 1.8 the eq is 1.0, which is slightly higher than that at the center. The variation
of eq at the sheet surface as functions of  and Ld/td is very similar to that of Ls. Using Fig.
6(a,c), it is considered that the friction coefficient was =0.34 in the rolling experiment
under Ld/td = 7.5 reported by Lee et al., 2002 and the eq of 4.23, which is five times higher
than that at the center, was introduced to the sheet surface. Kamikawa et al., 2007 reported
that the microstructural parameters, fraction of high-angle grain boundaries and average
misorientation, at the surface in the IF steel processed by one ARB cycle at 773K without
lubricant correspond to those at the center in the IF steel processed by five ARB cycles with
lubricant. The FE result shown in Fig. 6 is consistent with their microstructural results,
although a material is different. The results of Fig. 6(c) indicate that the eq introduced to the
sheet surface by rolling under the unlubricated condition is significantly different by the roll
diameter even if the initial thickness, nominal reduction, and friction condition are the same.
It is clear that eq at the surface depends on not only the friction condition between the rolls
and the sheet but also the roll bite geometry. It is evident from Fig. 6(b,c) that the eq and
s at the surface after rolling with =0.4 under Ld/td = 5.1 agree with those with =0.25
under Ld/td = 8.3. However, in Fig. 6(a) the value of Ls does not agree at these conditions.
Furthermore, although the s for =0.4 under Ld/td  6.7 indicates almost the same value,

www.intechopen.com
596 Finite Element Analysis

the Ls and eq are different under these conditions. These differences are attributed to the
complicated deformation history near the surface, as described in Subsections 3.3.3 below.

3.3 Deformation behavior at various thickness locations in a rolled sheet

3.3.1 Flection of pin after rolling
Figure 7 shows deformed meshes of a center part of the FE mesh (center mesh in Fig. 5) after
rolling with different , corresponding to the pin described in Fig. 2. Here, the rolling
condition No.2 in Table 1 was used. The flection through the thickness becomes larger with
increasing . This feature is consistent with the experimental results in which the flection of
the embedded pin is larger in a sheet rolled without lubricant than that with lubricant, as
shown in Fig. 3. Although the shear strain and equivalent strain are always calculated from
the inclination of the pin, based on equations proposed by Sakai et al., 1988, the embedded-
pin method does not exhibit the exact value of strains, because these equations are derived
under three assumptions; I) The ratio of incremental shear strain to incremental compressive
strain is constant during rolling; II) The incremental compressive strain is uniform through
the thickness; III) Plane strain conditions prevail in the deformation zone. And, the effect of
reverse shear deformation after the neutral plane, NP, shown in Fig. 2, is not considered in
these assumptions. As a result, it is clarified from strain histories during rolling shown in
Fig. 9 below that two assumptions I) and II) are not proper.

Before rolling
surface After rolling

=0.1 =0.2 =0.25 =0.3 =0.4

surface

Rolling y
by 50%
center center
x(RD)

surface

surface
Fig. 7. Initial mesh and deformed mesh after rolling with various friction coefficients
under rolling condition No.2 in Table 1.

3.3.2 Variations of L and  through sheet thickness after rolling

Figure 8 shows the distributions of the distance from the center in RD, L, and the angle of
inclination from the y axis, , through the sheet thickness after rolling with various  under
Ld/td =7.5 (d=255 mm) and Ld/td = 5.1 (d=118 mm), respectively. Here, these data were
obtained from the values of the integration point for each element divided into 33 through
the sheet thickness. In Fig. 8(a), the L increases with increasing  throughout thickness
locations, becomes gradually larger toward the surface from the center, and takes a
maximum at the surface. For the same , the L under Ld/td =7.5 is larger than that under
Ld/td = 5.1 throughout thickness locations. In Fig. 8(b), the  similarly increases with

www.intechopen.com
Strain variations on rolling condition in accumulative roll-bonding by inite element analysis 597

increasing , but the distributions of  through thickness are different by . Under Ld/td
=7.5, the  for  = 0.1 slightly increases from y=0 (center) to 0.45 and there is a steep
increase toward the surface thereafter. The  for  = 0.2 monotonously increases toward
the surface. The  for  0.25 increases with the distance y, takes a maximum at a
thickness location near the surface, and decreases toward the surface from the location. The
feature can be seen clearly in the flection of pin embedded in IF-steel sheets rolled at 973 K
under the unlubricated condition, as shown in Fig. 3 (Matsuoka et al., 1997; Hashimoto et al.,
1998; Kawabe et al., 1996). The location where the  takes a maximum is slightly away
from the surface with increasing . The distributions of  for  = 0.2 and 0.4 under Ld/td
=5.1 are similar to those for  = 0.1 and  0.25 under Ld/td = 7.5. Furthermore, the
magnitude of  near the surface for  = 0.4 under Ld/td =5.1 agrees with that for  = 0.3
under Ld/td = 7.5. At this time, the  at the surface for these conditions was almost the same
value of ~ 4.7, but the eq was different, where eq =3.4 for  = 0.3 under Ld/td = 7.5 and eq
=2.9 for  = 0.4 under Ld/td =5.1 (Fig. 6(c)).

(a) (b)
=0.4
center 1/4t surface center 1/4t surface

=0.35
1.5 90

=0.4
=0.4
solid symbols: Ld/td=7.5
=0.35
Distance from center in RD, L

Angle of inclination,  (deg.)

=0.25
75
open symbols: Ld/td=5.1
=0.3
=0.4 =0.3 =0.2
1 60

45 =0.2
=0.25
=0.1
0.5 30

=0.2
=0.2 15

=0.1
0 0
0 0.25 0.5 0 0.25 0.5

Distance from midthickness, y (mm) Distnace from midthickness, y (mm)

Fig. 8. Variations of (a) L and (b)  through sheet thickness after rolling with friction
coefficient  under Ld/td =7.5 (No. 2 in Table 1) and Ld/td =5.1 (No. 4 in Table 1).

3.3.3 Strain histories during rolling

The histories of the total strain in the x direction, xx, the total shear strain, , and the
equivalent strain, eq, at five elements, i.e., e1, e4, e7, e17 and e33, through sheet thickness
during rolling with = 0.4 under Ld/td = 7.5 are shown in Fig. 9, including deformed meshes
at times of 0.0875, 0.1, and 0.14 s illustrated in the inset. Here, the NP denotes the time at
which the shear stress f for each element equals zero and the stress in the y direction
becomes a maximum (Underwood, 1952). The e1 corresponds to the element located in the
sheet surface, e4, to the element for which the  indicated a maximum value for = 0.4
(Fig. 8(b)), e33, to the element located in the center, e17, to the element at 1/4 thickness
located between e1 and e33, and e7, to the element between e4 and e17. From Fig. 9, all
strains at e17 monotonously increase with increasing time. This means that each strain rate,
dxx/dt, d/dt, and deq/dt, at thickness locations from the center to the 1/4 thickness is
almost constant during rolling. Furthermore, the strain rates at e17 increases as a result of

www.intechopen.com
598 Finite Element Analysis

the shear deformation more than that at e33 without . Hence, the eq at the 1/4 thickness
(e17) becomes larger than that at the center (e33) which agrees with the value 2 / 3
ln{1/(1r)} calculated simply from reduction in thickness due to =0 constantly. On the
other hand, all strain rates at e7, e4, and e1 near the surface vary during rolling by the 
effect, and the xx and eq increase as a thickness location approaches the surface (Fig. 9(a,c)).
However, the  at e4 corresponding to the thickness location for which the  indicated a
maximum value in Fig. 8(b) is smaller than that at e7 (Fig. 9(b)). The shear strain rate, d/dt,
at e4 is initially faster than that at e7, but the  indicates the same value, += ~2.9, before the
NP, because the time of d/dt0 during rolling is longer in e4 than in e7. After the NP, the
d/dt at e7 inversely becomes faster than that at e4. Namely, although the positive shear
strain, +, is the same at these two locations, the negative shear strain, , at e4 is smaller
than that at e7.

(a) 2.5

e1

Total strain in RD,  xx

time=0.0875s time=0.1s time=0.14s 2
surface NP
e1 1.5 e4
e4 e7
1/4t 1
e7 e17
e17 y e33
0.5
e33 x(RD)
NP
center
0
0.06 0.08 0.1 0.12 0.14
Time (sec)
(b)
8
e1 (c) 5
7
e1
Equivalent strain,  eq

4
Total shear strain, 

6
NP NP e4
5
3
4 e7 e7
e4 2
3
2
NP e17
1
e17 1
e33
0 e33 NP
0
0.06 0.08 0.1 0.12 0.14 0.06 0.08 0.1 0.12 0.14
Time (sec) Time (sec)

Fig. 9. Histories of (a) xx, (b) , and (c) eq at five thickness locations (as inset) during rolling
with =0.4 under Ld/td=7.5 (No. 2 in Table 1).

Such shear strain history was not seen in 0.13  0.4 under Ld/td = 5.1 (d=118 mm). This
feature is further exhibited in Fig. 10, where the variations of , +, and  through thickness
are shown. Here, the thickness locations corresponding to five elements illustrated in the
inset of Fig. 9 are displayed. In Fig. 10(a), the + initially increases with the distance y. Its
incremental rate becomes larger from e17 (y=0.25) to e7 (y=0.4), becomes constant until e4
(y=0.45) thereafter, and shows a steep increase from e4 to e1 (y=0.5). On the other hand, the
 gradually increases from e33 (y=0) to e7 (y=0.4), decreases from e7 to e4 (y=0.45), and
shows a sharp increase thereafter. The magnitude of  is smaller than that of +
throughout thickness locations. As a result, the  which expresses these sums shows a
distribution with a sudden dip at y=0.45 near the surface, where the  indicated a

www.intechopen.com
Strain variations on rolling condition in accumulative roll-bonding by inite element analysis 599

maximum value (Fig. 8(b)). Although a similar variation of  with a maximum was seen in
 = 0.4 under Ld/td = 5.1, the  distribution with the sudden dip in Fig. 10(a) does not appear
in Fig. 10(b), where the variations of , +, and  through thickness after rolling with  = 0.4
under Ld/td = 5.1 are shown. In Fig. 10(b), the magnitude of  is smaller than that of +
throughout thickness locations except the surface, and the  exhibits the same magnitude
as the + at the surface. The feature which + equals  at the surface was the same in the
friction coefficient range of 0.13 0.4 under Ld/td = 5.1. This feature was also observed
in 0.10.3 under Ld/td = 7.5, as shown in Fig. 6(b) of T.Inoue & N.Tsuji, 2009.
Consequently, the magnitude of the shear strain + before the NP is larger than that of the
reverse shear strain  after the NP through thickness, but these magnitudes exhibit the
same value at the sheet surface. Provided that a sheet is rolled with a high friction condition
under a large roll bite geometry, the magnitude of + becomes larger than that of
throughout thickness due to the sudden dip of  near the surface.

(a) Ld/td=7.5(d=255mm) (b) Ld/td=5.1(d=118mm)

center 1/4t surface center 1/4t surface
8 8

=0.4 =0.4
e1


7 7
Shear strain, ,  
Shear strain, ,  

6 6


5 5 e1
e7
4 e4 4
e4

+ +
3 3
e7


2 e17 2


e17
1 1
e33 e33
0 0
0 0.25 0.5 0 0.25 0.5
Distance from midthickness, y (mm) Distance from midthickness, y (mm)

Fig. 10. Variations of the total shear strain  , positive shear stain +, and reverse shear strain
 through sheet thickness after rolling with =0.4. Here, e1, e4, e7, e17 and e33 are
displayed in the inset of Fig. 9.

3.4 Strain distributions through sheet thickness after rolling

According to the embedded-pin method (Sakai et al., 1988; Matsuoka et al., 1997), the
“apparent” shear strain, (pin), shows a maximum at the thickness location of 0.10.2 from
surface, and, hence, the “apparent” equivalent strain, eq(pin), also takes the maximum there.
Here, in Sakai et al., 1988, a large roll bite geometry of 9.1 had been empoyed. This is clear
from the variation of  through sheet thickness in Fig. 8(b). The (pin) is calculated from the
inclination of the pin after rolling. Provided the direction of shear deformation remains
unchanged during rolling, the (pin) and eq(pin) must take their maximum at the surface.
However, since the direction of shear deformation changes to the opposite direction at the
NP, the inclination of mesh at the surface becomes smaller in the surface layer under high
friction and large roll bite geometry conditions, as shown in Fig. 8(b). Therefore, the eq(pin)

www.intechopen.com
 600 Finite Element Analysis

measured by the embedded-pin method would be underestimated in the surface layer

compared with the “substantial” equivalent strain obtained in the present study.

(a) Total strain in x-direction xx (b) Total shear strain  (c) Equivalent strain eq
=0.4 =0.4
surface center surface surface center surface surface center surface

=0.4
8
2 =0.35 5

=0.35
7
Total strain in RD  xx

Equivalent strain  eq
=0.35
6 4

Total shear strain 

=0.3
1.5
=0.3 5
=0.3 3

=0.25 =0.25
4
1

=0.2 =0.25
=0.1
3 2

0.5 2 =0.2
=0.2 =0.1
1
1

0 0
=0.1 0
-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5

Distance from thickness center y (mm) Distance from thickness center y (mm) Distance from thickness center y (mm)

Fig. 11. Distributions of xx, , and eq, through sheet thickness for various friction
coefficients  under Ld/td =7.5 (No. 2 in Table 1).

Figure 11 represents the distributions of the “substantial”  and eq through the sheet
thickness after rolling with various  under the rolling condition No. 2 in Table 1. Similarly,
the distribution of xx (=yy) is shown in Fig. 11 because all the strains are associated as
shown in Fig. 9. The  at the thickness center is always zero in Fig. 11(b). The  for =0.1
takes a maximum at y=±0.25mm (1/4t) and decreases toward the surface thereafter. It is
likely that this distribution results from the roll bite geometry Ld/td, because the
corresponding strain xx is constant through the sheet thickness as shown in Fig. 11(a). The
xx for friction coefficient other than =0.1 gradually increases toward the surface from the
center, but for =0.35 and 0.4 there is a sharp raise at the thickness location of 0.050.15 from
surface. In Fig. 11(b),  for =0.2 to 0.3 increases toward the surface from the center.
However, the distributions of  for  0.35 are different from those for =0.2 to 0.3.
Especially for =0.4 there is a sudden dip at the thickness location near surface. It is
considered that this behavior is attributed to the decrease in the reverse shear strain  as
shown in Fig. 10(a). If the + and  are the same magnitude regardless of the change in ,
the total shear strain  might gradually increase toward the surface from the center such as 
for =0.2 to 0.3, and, furthermore, larger total shear strain might occur at the surface. On the
other hand, in Fig. 11(c), eq at the thickness center is constant regardless of , and its
magnitude indicates 0.80 because no shear strain is imposed at the center. Although eq for
=0.1 looks almost constant throughout the thickness, the eq showed a distribution in the
strain range of 0.8 eq0.83; it increases toward the surface from zero (y=0) at the center,
takes a maximum at y ±0.45 mm, and decreases thereafter. It is considered that this
distribution is the effect of Ld/td. The eq for all other  gradually increases toward the
surface from the center and shows a distribution with the maximum at the surface. The
maximum eq increases with increasing , and eq for =0.4 reaches 5.33, which is six times
higher than that at the center. The eq=5.33 corresponds to a 99% reduction in plane strain
compression. In the distribution of eq, a sharp raise or a sudden dip, as seen near the
surface of xx and  are not observed. This means that the equivalent strain eq that denotes

www.intechopen.com
Strain variations on rolling condition in accumulative roll-bonding by inite element analysis 601

scalar amount varies continuously throughout sheet thickness regardless of the roll bite
geometry and the friction between the rolls and the material surface. On the other hand,
strains in each component that denote vector amount vary complicatedly in the sheet
thickness. Hence, it is found that a much larger eq can be introduced to the sheet surface
through a complicated deformation by the shear strain effect.

3.5 Strain distributions in ARB

There are numerous reports on microstructure evolution in various sheet materials by ARB
process (Hidalgo et al., 2010; Kolahi et al., 2009; Xing et al., 2002). In most papers, a rolling
by a 50% reduction in thickness is conducted without lubricant, and equivalent strain
calculated simply from reduction in thickness has been used regardless of a high friction
condition. However, it is found from previous results that equivalent strain introduced in a
rolled sheet is different by the effects of not only friction condition but also roll diameter
used.
From Figs. 6(a,c) and 11(b,c),  and eq at the surface imposed by the first cycle of ARB at
ambient temperature reported by Lee et al., 2002 show the values of 5.60 and 4.23,
respectively. That is, a much larger equivalent strain, which is five times higher than that at
the center, is introduced to the surface. Here, it is considered that the friction coefficient is
0.34. Moreover, assuming that the same deformation is repeated in the following cycles,
 and eq at the surface in the multi-cycle ARB-processed 1100 Al can be estimated by =5.60
N and eq =4.23 N, respectively, where N denotes the number of cycles. On the other hand,
the  and eq at the center are calculated by =5.60 (N1) and eq=0.8+4.23 (N1), respectively.
Utsunomiya et al., 1999 showed the flection of the pin in the 1100 Al processed by one ARB
cycle at 473 K with and without lubricant under the same roll bite geometry as Lee et al.,
2002. Here, Ls indicated 0.15 with lubricant and 1.5 without it. Since the effect of
temperature in the rolled Al or IF steel sheet on the relation among eq, , and Ls is small in
comparison with the friction effect (Um et al., 2000; Inoue & Tsuji, 2009), it is likely from Figs.
6(a,c) and 11(b,c) that the strain at the surface is 0.59 and eq 0.97 for rolling with
lubricant and 7.66 and eq 5.34 for rolling without it. Here, it is considered that the
friction coefficient is about 0.12 for rolling with lubricant and 0.40 for rolling without it.
That is, equivalent strain and shear strain at the surface in a sheet rolled without lubricant
are five times and twelve times higher than the strains with it. Namely, eq at the surface in
the 1100 Al processed by one ARB cycle without lubricant would correspond to eq in that
processed by five ARB cycles with lubricant. Consequently, since ARB process shown in Fig.
1(b) is usually conducted by a 50% reduction of thickness without lubricant to aid bonding,
a rolling condition with large roll bite geometry is desirable for fabricating ultrafine-gained
materials efficiently.

3.6 Variations of the total shear strain  and equivalent strain eq against L or 
The embedded-pin method has a limitation on the quantification of strains imposed by
rolling, as mentioned previously. In particular, the strains near the surface are determined
by very complicated histories during rolling, as shown in Fig. 9. However, if the magnitude
and distribution of strains in a rolled sheet can be quantitatively estimated by using
experimental measurements, such as L or  observed from the pin flection shown in Fig.
3, these quantitative strain analyses would be useful for designing the microstructure of the

www.intechopen.com
602 Finite Element Analysis

sheet metal by rolling. Figure 12(a,b) shows the variations of  and eq with L through
sheet thickness for different values of  under Ld/td =7.5 and Fig. 12(c,d) shows the
variations of  and eq with . Here, all data were obtained from the values of the
integration point for 33 elements in sheet thickness, as shown in Fig. 8. The L = 0 and  =
0 correspond to a location of the thickness center.

(a) center e1 (c) center e1

8 8
 

7 7
e1  e1
 

Total shear strain, 

Total shear strain, 

6 6
 e1  e1
5  5 
 
4 e1 4 e1
3 3
2 2 e1
e1
1 e1 1
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0 15 30 45 60 75 90
Distance from center in RD, L(mm) Angle of inclination, (deg.)
(b) center (d) center
7 7
=0.1 e1  e1
=0.2 
6 6

Equivalent strain,  eq
Equivalent strain,  eq

=0.25 e1 e1
=0.3 
5 5
=0.35 e1 
=0.4 
4 4

3 e1 3
1.9
2 1 2 e1
e1
1 2  1  1 2  1 
ln   ln  
e1 3  1 r  3  1 r 
0 e1 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0 15 30 45 60 75 90
Distnace from center in RD, L (mm) Angle of inclination, (deg.)

Fig. 12. Variations of (a)  and (b) eq with L and (c)  and (d) eq with  through thickness
in 50% rolled Al for various friction coefficients  under Ld/td =7.5 (No. 2 in Table 1). Here,
e1 denotes data at the sheet surface shown in the inset of Fig. 9.

In Fig. 12(a,c), the L and  relations for = 0.1 are significantly different from the
relations for other . The  for the  close to tan , which is a condition to pull a sheet into
the roll bite, took a maximum at y ±0.25 mm (1/4 thickness) and decreased toward the
surface thereafter, as shown in Fig. 11(b). This is because the distribution of  is dominated
by the roll bite geometry Ld/td due to the small effect of . The  for  0.2 increases as the
L or  becomes larger, i.e., the thickness location approaches a sheet surface. However,
since there is a decrease of the negative shear strain near the surface, as shown in Fig.
10(a), the  for  0.35 initially increases with increasing the L or , decreases near the
surface, and shows a steep increase thereafter. The  relations, except = 0.1, have a
good correlation, compared to the L relation. In particular, the variation of  with 
until 60° is almost the same despite . It is likely that the  depends strongly on the shear
deformation. It is evident from Fig. 12(b,d) that the eq, including all strain components as
defined in equation (2), has very good correlation with the L and . The eq
monotonically increases with increasing the L, and the relation is expressed by eq = 1.9 L

www.intechopen.com
Strain variations on rolling condition in accumulative roll-bonding by inite element analysis 603

+ 0.8, but the eq sharply increases near the surface by the effect of the shear strain (Fig.
12(b)). On the other hand, in Fig. 12(d), the eq remains 0.8 calculated by 2 / 3 ln{1/(1r)} in
the range of 0 30°, slightly increases with  thereafter, and sharply increases at
70° by the shear deformation effect. It is clear from Fig. 12 that the eq has better
correlation with the L and , than . This means that L and  are determined by a
combination of the strains in the x and y directions and the shear strain and their histories
during rolling. As shown in Figs. 6 and 8, the deformation behaviors in a rolled sheet
depend strongly on not only the friction but also the roll bite geometry, especially when
rolling with high friction. Figure 13 shows the variations of eq with L and , respectively,
in a 50% rolled Al with =0.4 under various Ld/td of five rolling conditions, Nos. 1-5, in
Table 1. These correlations agree with the eqL and eq  relations shown in Fig.
12(b,d). Consequently, the Ld/td ratio has a similar effect on  regarding the relation among
eq, L, and  through sheet thickness after rolling.

(a) (b)
7 7
e1
 
e1
Ld/td =3.0(d=40) Ld/td=3.0(d=40)
Ld/td=5.1(d=118)  Ld/td=5.1(d=118)
6 e1 6 e1
 Ld/td=6.7(d=201)
Equivalent strain,  eq

Ld/td=6.7(d=201)
Equivalent strain,  eq
e1

Ld/td=7.5(d=255)  Ld/td=7.5(d=255)
5 5

4 
Ld/td=8.3(d=310) 4  Ld/td=8.3(d=310)
e1
3 1.9 3

2 1 2 e1

1 2  1  2  1 
ln  
1
ln  
e1 3  1 r  3  1 r 
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0 15 30 45 60 75 90
Distance from center in RD, L(mm) Angle of inclination, (deg.)

Fig. 13. Variations of eq with (a) L and (b)  through thickness in 50% rolled Al with 
=0.4 under various Ld/td (Nos. 1-5 in Table 1). Here, e1 denotes data at the surface shown in
the inset of Fig. 9.

3.7 Universality of the roll bite geometry on strains through thickness in a rolled sheet
As a condition to impose a large equivalent strain by rolling, the roll bite geometry is as
important a processing parameter as the friction condition. The roll bite geometry Ld/td is
determined by the thickness before and after rolling, t0 and t1, and the roll diameter, d, as
shown in Fig. 2. However, even if the Ld/td is the same value, the deformation behaviors
through sheet thickness may be different by a combination of t0, t1, and d because the
deformation histories during rolling are very complicated under high friction conditions.
Figure 14 shows the distributions of  and equivalent strain added by the shear
deformation, eq(shear), through thickness in the rolled Al with =0.3 under the rolling
conditions Nos. 4, 7-9 of Ld/td = 5.1 in Table 1. Here, the horizontal axis was normalized by
the half of sheet thickness after rolling, i.e., 2y/t1 = 0, 0.5, and 1 indicate the center, 1/4
thickness, and surface, respectively, in the sheet. The eq(shear) represents the equivalent
strain added by shear deformation as defined by eq  2 / 3 ln{1/(1r)} and equals zero at the
center of =0. It is found from Fig. 14(a) that, for all conditions, the  through thickness

www.intechopen.com
604 Finite Element Analysis

does not exhibit exactly the same distribution and magnitude. In particular, the magnitudes
for Nos. 4 and 9 at the same r are significantly different from those for Nos. 7 and 8.

Rolling conditions
No.4:t0=2, t1=1mm (r=50%), d=118mm, No.7:t0=2, t1=1.41mm (r=29.5%), d=255mm
No.8:t0=2, t1=1.76mm (r=12%), d=768mm, No.9:t0=5.3, t1=2.65mm (r=50%), d=310mm

(a) (b)
surface 1/4t center 1/4t surface surface 1/4t center 1/4t surface

=0.3
90

=0.3
1

by shear deformation,  eq(shear)

No. 4 No.4
80 No. 7 No.7
No. 8
Angle of inclination, (deg.)

No.8
70 0.8
No. 9 No.9

60
0.6
50

Equivalent strain added

40
0.4
30

20
0.2
10

0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1

Relative distance from midthickness, 2y/t 1 Relative distance from midthickness, 2y/t 1

Fig. 14. Distributions of (a)  and (b) equivalent strain added by shear deformation, eq(shear),
through sheet thickness in rolled Al with =0.3 under Ld/td=5.1. Here, full details of the
rolling conditions are shown in Table 1.

Numbers 4 and 9 indicate the same magnitude at all thickness locations except near y/t1
=0.9. The magnitude of  decreases with decreasing r under the same t0 (Nos. 4, 7, 8).
The friction coefficient range of 0.13  0.4 showed a similar feature. This result indicates
that the  measured from the embedded-pin method is different by r even if the Ld/td has
the same value under the same t0. Therefore, we should use a reduction of the thickness, r,
as a variable of the Ld/td. In Fig. 14(b), the eq(shear) also exhibits the same tendency. At the
surface, the eq(shear) for No. 4 of r=50% is 0.88, which is twice as high as that for No. 7 of
r=29.5% and five times higher than that for No. 8 of r=12%. These values are similar to the
relations of the equivalent strain calculated simply from the r, i.e., eq at the center; eq = 0.8,
0.4, 0.15 for r=50%, 29.5%, 12%, respectively. Furthermore, although the  at the surface in
Nos. 4 and 9 indicated the same value in Fig. 14(a), the eq(shear) is larger in No. 4 than in No.
9. This feature was the same in the range of  0.3, and the difference of eq(shear) at the
surface increased with increasing . However, the eq(shear) at the surface in the Nos. 4 and 9
indicated the same magnitude in the range of  0.25. On the other hand, the eq(shear) from
the center to the 1/4 thickness in No. 4 is in good agreement with that in No. 9 (Fig. 14(b)).
A similar feature was seen in the friction coefficient range of 0.13  0.4. As shown in Fig.
9, at thickness locations from the center (e33) to the 1/4 thickness (e17), the strain rates are
almost constant during rolling. In other words, the magnitude of eq(shear) at the thickness
locations which take a constant strain rate during rolling agrees under the L d/td ratio with
the same r. However, the eq(shear) from the 1/4 thickness to the surface is not the same
magnitude in the range of  0.3, and the eq(shear) at the locations becomes larger with

www.intechopen.com
Strain variations on rolling condition in accumulative roll-bonding by inite element analysis 605

decreasing t0 for the same r. Figure 15 shows the distributions of  (1  2y/t1  0) and
eq(shear) (0  2y/t1  1) for =0.25. It can be seen that the eq(shear) and  in No. 4 are almost
the same magnitude and distribution as those in No. 9. Consequently, even if the Ld/td is
the same value, the strain imposed by rolling does not exhibit exactly the same magnitude
and distribution through sheet thickness because the deformation behaviors during rolling
are not simple due to the effect of the shear deformation. Under the Ld/td =5.1, the eq(shear)
increases with increasing r for the same t0. However, provided that the Ld/td is employed
under the same r, the eq through sheet thickness agrees regardless of the combination of t0
and d in the friction coefficient range of  0.25. For high friction of  0.3, the eq at
locations from 1/4 thickness to the surface becomes larger as t0 decreases, and this tendency
becomes more remarkable with increasing .

surface 1/4t center 1/4t surface

90

=0.25

by shear deformation, eq(shear)

No. 4
80 No. 7 0.8
Angle of inclination, (deg.)

No. 8
70
No. 9


60 0.6

50

Equivalent strain added

 eq(shear)
40 0.4

30

20 0.2

10

0 0
-1 -0.5 0 0.5 1
Relative distance from midthickness, 2y/t 1

Fig. 15. Distributions of  (lefthand side) and eq(shear) (righthand side) through sheet
thickness in rolled Al with =0.25 under Ld/td =5.1.

As a result, the Ld/td parameter shown in Fig. 1 should be expressed by the following
equation including t0, d and r, using the relation of t1=t0/(1r).


Ld 1 2d r
td 2  r
(4)
t0

The relation among the roll bite geometry Ld/td, roll diameter d, nominal reduction r, and
initial thickness t0 is plotted in Fig. 16 for various combinations of d, r, and t0. Here, the
open symbols indicate the rolling conditions employed in the literature related to the
microstructure change through sheet thickness in a rolling process, including ARB. The
Ld/td gradually increases with increasing d under constant t0 and r (Fig. 16(a)) and
monotonically increases with increasing r under constant d  and t0 (Fig. 16(b)). Under d
and r constant (Fig. 16(c)), the Ld/td gradually increases with decreasing t0, and there is a
steep increase at t0 < 5 mm. Compared to the increases of d or r, the decrease of t0 has a
large influence on the increase of Ld/td. It is found from Fig. 16 that the Ld/td is different
with regard to t0, d, and r. Even if the friction condition is the same, this causes the

www.intechopen.com
606 Finite Element Analysis

difference in the magnitude and distribution of strains, as shown in the present study,
which results in the changes of the microstructure and texture depending on the thickness
location of a rolled sheet.

(a) (c)
dr=50%
14 14
Kamikawa et al., 2007 Sakai et al., 1988

dr=30%
12 t0=2,r=50% 12

Roll bite geometry, Ld/td

Roll bite geometry, Ld/td

Lee et al., 2002

10 t0=2,r=40% 10
dr=10%
Lee et al., 2002
t0=2,r=30%
dr=50%
8 8

dr=30%
6 6
Um et al., 2000 t0=12,r=50% Um et al., 2000
4 t0=2,r=10% 4
t0=12,r=30%
2
dr=10%
2
t0=12,r=10%
0 0
0 100 200 300 400 500 600 0 5 10 15 20
Roll diameter, d (mm) Initial thickness, t0 (mm)
(b)
dt0=2
14
Kamikawa et al., 2007
12
Roll bite geometry, Ld/td

10 dt0=2
Um et al., 2000
dt0=2
8

6
dt0=12
dt0=12
dt0=12
4

0
0 0.1 0.2 0.3 0.4 0.5 0.6
Nominal reduction, r (%)
Fig. 16. Variations of the roll bite geometry Ld/td with (a) roll diameter d, (b) nominal
reduction r, and (c) initial thickness t0. Here, the open symbols indicate the rolling
conditions employed in the literature.

4. Outlook
It is commonly accepted that the distribution of the equivalent strain through thickness in a
rolled sheet depends on the presence of shear deformation due to friction between the rolls
and the sheet and brings about a change of microstructure (grain size, texture) through sheet
thickness. It is found from the present results that the roll bite geometry defined by equation
(4), as well as the friction coefficient, is an important parameter for studying microstructural
changes through thickness in metal sheet rolling including ARB process. The grain size
decreases with an increase of the equivalent strain, and, hence, a rolling condition with high
reduction using a mill with a large roll diameter without a lubricant is effective for the
refinement of crystal grains. Such rolling conditions can be seen in hot rolling (high friction
and large roller) rather than cold rolling (low friction condition and small roller) and in a
commercial rolling mill of product level (large roller) rather than a rolling mill of laboratory
level (small roller). Moreover, the microstructural evolution depends strongly on not only
equivalent strain but also shear strain. Inoue et al., 2002 reported that in the ferrite grain
transformed from deformed austenite, the grain size in the area with the shear strain is finer
than that in the area without the shear strain under the same equivalent plastic strain. Cho
et al., 2004; Kang et al., 2010 demonstrated significant advances in high angle grain

www.intechopen.com
Strain variations on rolling condition in accumulative roll-bonding by inite element analysis 607

boundaries and the subdivision of grains under the deformation with shear strain than
under without shear strain using Ni-30Fe alloy and low carbon steel. In other words, it is
noted that the shear deformation plays a important role for refining crystal grains.
Also, it is known that a shear texture different from a conventional rolling texture develops
at the locations where shear deformation is introduced by rolling. Control technology of
texture associated with the shear deformation is positively used to improve the deep
drawability (ND//<111>) and magnetic characteristic (Goss orientation: {110}<001>). The
texture of ND//<111>, which contributes to improve the deep drawability, is produced by a
large shear deformation in fcc metals, but its deformation is disadvantageous in bcc metals
(Matsuo, 1989). Therefore, for a study of deep drawability, using a large roller is important
for the case of aluminum alloys (fcc), and a small roller is recommended for the case of IF
steels (bcc). In study of deep drawability in ferritic rolling of extra low carbon steels,
Kawabe et al., 1996 reported that the r-value in the sheet improved more significantly when
using a small than a large roller under the same friction condition. They confirmed, using
the embedded-pin method, that the improvement in the r-value was the result of the
decrease of the shear deformation through sheet thickness. Consequently, in order to study
universal relation between microstructure (grain size, texture) and plastic deformation, it is
essential to understand the quantity (magnitude) and quality (component) of the strain into
materials introduced by a plastic deformation process through a combination of numerical
simulations and experimental measurements and observations. And a quantitative relation
among the strains, friction, and roll bite geometry obtained from numerical simulations
would provide useful guidelines for studying the microstructure design in a rolled sheet
and for understanding the quantitative correlation between microstructures and strain in
ARB process.

5. Conclusions
The effect of the roll bite geometry, Ld/td, on the magnitude and distribution of strains
imposed in a metal sheet by rolling under various friction conditions was studied using a
finite element simulation that takes the deformation history into account. The relation
among the strains, distance from center in RD, L, and angle of inclination from the y axis,
, was shown through thickness in a rolled 1100 Al. The present results will provide useful
guidelines for studying the correlation between microstructures and strain in accumulative
roll-bonding (ARB) process as well as the microstructure design in a rolled sheet. The main
results are as follows:

1. The deformation through sheet thickness during and after rolling depends strongly on
not only the friction condition between the rolls and the sheet but also the roll bite
geometry Ld/td. The Ld/td ratio has a similar effect on friction coefficient  regarding
the relation among eq, L, and  through sheet thickness after rolling. In other words,
the roll bite geometry is as important a processing parameter as the friction condition
for studying microstructural changes through thickness in metal sheet rolling and ARB
process.
2. The eq in a rolled sheet gradually increases toward the surface from the center and
shows a distribution with the maximum at the surface. The maximum eq increases with
increasing  or Ld/td. On the other hand, in the total shear strain , there is a sudden

www.intechopen.com
608 Finite Element Analysis

dip at the thickness location near surface under high friction condition. This behavior is
attributed to the decrease in the reverse shear strain  induced by the shear stress after
a neutral plane.
3. Under different values of the friction coefficient or roll bite geometry, the equivalent
strain eq had much better correlation with the L and  through sheet thickness,
except near the surface, than the total shear strain . This is attributed to the fact that L
and  are determined by a combination of strains in the x and y directions and the
shear strain and their histories during rolling.
4. The Ld/td is expressed by equation including initial thickness, t0, roll diameter, d, and
nominal reduction, r. The Ld/td at the same r can be employed as a universal parameter
on the equivalent strain in a rolled sheet, except for thickness locations near the surface
under high friction conditions.
5. In ARB process, which half of the sheet-surface regions comes to the center in the next
cycle and its procedure is repeated, a rolling condition with large roll bite geometry at a
high friction condition is desirable for fabricating ultrafine-gained materials efficiently.

6. References
ABAQUS/Explicit ver. 6.5 User’s manual, Theory manual; (2006). Providence, ABAQUS
Inc., CD-ROM.
Ataka, M., (2006). Handbook of Technology of Plasticity, Corona Publishing Co., Ltd., Tokyo,
pp. 11-131.
Backofen, W.A., 1972. Deformation Processing, Addison-Wesley Pub. Co., Massachusetts, pp.
94-115.
Cho, J.Y., Inoue, T., Yin, F. & Nagai, K. (2004). Effect of shear deformation on
microstructural evolution of Ni-30Fe alloy during hot deformaiton. Materials
Transactions, Vol. 45, pp. 2966-2973.
Cui, Q. & Ohori, K. (2000). Grain refinement of high purity aluminium by asymmetric
rolling. Materials Science and Technology, Vol. 16, pp. 1095-1101.
Dieter, G.E., 1988. Mechanical Metallurgy SI Metric Edition, McGraw-Hill Book Co., 0-07-
100406-8, Singapore, pp. 501-563.
Hashimoto, S., Tsukatani, I., Kashima, T. & Miyoshi, T. (1998). Development of hot-rolled
steel sheet with a high r-value by rolling in ferrite region with lubrication. Kobe Steel
Engineering Reports, Vol. 48, pp. 14-18.
Hidalgo, P., Cepeda-Jiménez, C.M., Ruano, O.A. & Carreño, F. (2010). Influence of the
Processing Temperature on the Microstructure, Texture, and Hardness of the 7075
Aluminum Alloy Fabricated by Accumulative Roll Bonding. Metallurgical and
Materials Transactions A, Vol. 41, pp. 758-767.
Horita, Z., Fujinami, T., Nemoto, M. & Langdon, T.G. (2000). Equal-channel angular
pressing of commercial aluminum alloys: grain refinement, thermal stability and
tensile properties. Metallurgical and Materials Transactions A, Vol. 31, pp. 691-701.
Inoue, T., Torizuka, S., Nagai, K., Tsuzaki, K. & Ohashi, T. (2001). Effect of plastic strain on
grain size of ferrite transformed from deformed austenite in Si-Mn steel. Materials
Science and Technology, Vol. 17, pp. 1580-1588.
Inoue, T., Torizuka, S. & Nagai, K. (2002). Effect of shear deformation on refinement of
crystal grains. Materials Science and Technology, Vol. 18, pp. 1007-1015.

www.intechopen.com
Strain variations on rolling condition in accumulative roll-bonding by inite element analysis 609

Inoue, T., Ochiai, T., Yin, F. & Nagai, K. (2007a). Test production of ultrafine-grained steel
plate by large-scale forging press. Tetsu-to-Hagane, Vol. 93, pp. 693-702.
Inoue, T., Yin, F. & Kimura, Y. (2007b). Strain distribution and microstructural evolution in
multi-pass warm caliber rolling. Materials Science and Engineering A, Vol. 466,
pp.114-122.
Inoue, T., Horita, Z., Somekawa, H. & Ogawa, K. (2008). Effect of initial grain sizes on
hardness variation and strain distribution of pure aluminum severely deformed by
compression tests Acta Materialia, Vol. 56, pp. 6291-6303.
Inoue, T., Todaka, Y. & Horita, Z. (2009a). Special issue on severe plastic deformation for
production of ultrafine structures and unusual mechanics properties:
understanding mechanisms. Materials Transactions, Vol. 50, pp. 1-116.
Inoue, T. & Tsuji, N. (2009b). Quantification of strain in accumulative roll-bonding under
unlubricated condition by finite element analysis. Computational Materials Science,
Vol. 46, pp. 261-266.
Inoue, T., Somekawa, H. & Mukai, T. (2009c). Hardness variation and strain distribution in
magnesium alloy AZ31 processed by multi-pass caliber rolling. Advanced
Engineering Materials, Vol. 11-8, pp. 654-658.
Inoue, T., Yin, F., Kimura, Y., Tsuzaki, K. & Ochiai, S. (2010a). Delamination effect on impact
properties of ultrafine-grained low carbon steel processed by warm calibre rolling.
Metallurgical and Materials Transactions A, Vol. 41-2, pp. 341-355.
Inoue, T., Yin, F. & Kimura, Y. (2010b). Effect of deformation mode on texture of ultrafine-
grained low carbon steel processed by warm caliber rolling. Materials Science Forum,
Vols. 638-642, pp 2793-2798.
Kamikawa, N., Sakai, T. & Tsuji, N. (2007). Effect of redundant shear strain on
microstructure and texture evolution during accumulative roll-bonding in ultralow
carbon IF steel. Acta Materialia, Vol. 55, pp. 5873-5888.
Kang, J-.H., Inoue, T. & Torizuka, S. (2010). Effect of shear strain on the microstructural
evolution of a low carbon steel during warm deformation. Materials Transactions,
Vol. 51, pp. 27-35.
Kawabe, H., Matsuoka, S., Seto, K., Sakata, T., Furugimi, O. & Ohara, T. (1996). Effect of roll
diameter on deep drawability of extra low C sheet steels with hot rolling at ferrite
region. CAMP-ISIJ, Vol. 9, p. 1333.
Kimura, Y., Inoue, T., Yin, F. & Tsuzaki, K. (2008). Inverse temperature dependence of
toughness in an ultrafine grain structure steel. Science, Vol. 320, pp. 1057-1060.
Kolahi, A., Akbarzaden, A. & Barnett, M.R. (2009). Electron back scattered diffraction
(EBSD) characterization of warm rolled and accumulative roll bonding (ARB). J.
Materials Processing Technology, Vol. 209, pp. 1436-1444.
Lee, S.H., Saito, Y., Tsuji, N., Utsunomiya, H. & Sakai, T. (2002). Role of shear strain in
ultragrain refinement by accumulative roll-bonding (ARB) process. Scripta
Materialia, Vol. 46, pp. 281-285.
Matsuo, M. (1989). Texture control in the production of grain oriented silicon steels. ISIJ
International, Vol. 29, pp. 809-827.
Matsuoka, S., Morita, M., Furukumi, O. & Obara, T. (1997). Structural variation along
thickness direction of extra-low carbon sheet steels rolled in ferrite region. Tetsu-to-
Hagane, Vol. 83, pp. 127-132.

www.intechopen.com
610 Finite Element Analysis

Mukai, T., Somekawa, H., Inoue, T. & Singh, A. (2010). Strengthening Mg–Al–Zn alloy by
repetitive oblique shear strain with caliber roll. Scripta Materialia, Vol. 62,
pp. 113-115.
Mukhopadhyay, A, Higginson, R.L., Howard, I.C. & Sellars, C.M. (2007). Strain summation
in finite element modeling of multipass hot rolling. Materials Science and Technology.
Vol.23-1 pp.29-37.
Saito, Y., Tsuji, N., Utsunomiya, H., Sakai, T. & Hong, R.G. (1998). Ultra-fine grained bulk
aluminum produced by accumulative roll-bonding (ARB) process. Acta Materialia,
Vol. 39, pp. 1221-1227.
Sakai, T., Saito, Y., Hirano, K. & Kato, K. (1988). Recrystallization and texture formation in
high speed hot rolling of austenitic stainless steel. Trans ISIJ, Vol. 28, pp. 1028-1035.
Sakai, T. & Saito, Y. (1999). Effect of inhomogeneous deformation throgh the thickness on
microstructure and texture of hot rolled sheet. J. Japan Society Technology Plasticity,
Vol. 40, pp. 1158-1163.
Segal, V.M. (1995). Materials processing by simple shear. Materials Science and Engineering A,
Vol. 197, pp.157-164.
Todaka, Y., Inoue, T. & Horita, Z. (2008). Special issue on severe plastic deformation for
production of ultrafine structures and unusual mechanics properties: Investigating
role of high-density lattice defects. Materials Transactions, Vol. 49, pp. 1-106.
Um, K.-K., Jeong, H.-T., An, J.-K., Lee, D.N., Kim, G. & Kwon, O. (2000). Effect of initial
sheet thickness on shear deformation in ferritic rolling of IF-steel sheets. ISIJ
International, Vol. 40, pp. 58-64.
Underwood, L.R. (1952). The Rolling of Metals. Chapman&Hall Ltd., London, pp. 57-93,
203-268.
Utsunomiya, H., Tanda, K., Saito, Y., Sakai, T. & Tsuji, N. (1999). J. Japan Society Technology
Plasticity, Vol. 40, pp. 1187-1191.
Xing, Z.P., Kang, S. B. & Kim, H.W. (2002). Microstructural evolution and mechanical
properties of the AA8011 alloy during the accumulative roll-bonding process.
Metallurgical and Materials Transactions A, Vol. 33, pp. 1521-1530.
Zhang, X.J., Hodgson, P.D. & Thomson, P.F. (1996). The effect of through-thickness strain
distribution on the static recrystallization of hot rolled austenitic stainless steel
strip. Materials Processing Technology, Vol. 60, pp. 615-619.

www.intechopen.com
 Finite Element Analysis
Edited by David Moratal

ISBN 978-953-307-123-7
Hard cover, 688 pages
Publisher Sciyo
Published online 17, August, 2010
Published in print edition August, 2010

Finite element analysis is an engineering method for the numerical analysis of complex structures. This book
provides a bird's eye view on this very broad matter through 27 original and innovative research studies
exhibiting various investigation directions. Through its chapters the reader will have access to works related to
Biomedical Engineering, Materials Engineering, Process Analysis and Civil Engineering. The text is addressed
not only to researchers, but also to professional engineers, engineering lecturers and students seeking to gain
a better understanding of where Finite Element Analysis stands today.

How to reference
In order to correctly reference this scholarly work, feel free to copy and paste the following:

Tadanobu Inoue (2010). Strain Variations on Rolling Condition in Accumulative Roll-Bonding by Finite Element
Analysis, Finite Element Analysis, David Moratal (Ed.), ISBN: 978-953-307-123-7, InTech, Available from:
http://www.intechopen.com/books/finite-element-analysis/strain-variations-on-rolling-condition-in-accumulative-
roll-bonding-by-finite-element-analysis

InTech Europe InTech China

University Campus STeP Ri Unit 405, Office Block, Hotel Equatorial Shanghai
Slavka Krautzeka 83/A No.65, Yan An Road (West), Shanghai, 200040, China
51000 Rijeka, Croatia
Phone: +385 (51) 770 447 Phone: +86-21-62489820
Fax: +385 (51) 686 166 Fax: +86-21-62489821
www.intechopen.com
© 2010 The Author(s). Licensee IntechOpen. This chapter is distributed
under the terms of the Creative Commons Attribution-NonCommercial-
ShareAlike-3.0 License, which permits use, distribution and reproduction for
non-commercial purposes, provided the original is properly cited and
derivative works building on this content are distributed under the same
license.