You are on page 1of 13

j ournal of materi als processi ng technology 2 0 7 ( 2 0 0 8 ) 222234

j our nal homepage: www. el sevi er . com/ l ocat e/ j mat pr ot ec


Dynamic friction model and its application in at rolling
Xincai Tan
a,b,
, Xiu-Tian Yan
a
, Neal P. Juster
a
, Srinivasan Raghunathan
b
, Jian Wang
b
a
Department of Design, Manufacture & Engineering Management, University of Strathclyde, 75 Montrose Street,
Glasgow, Scotland G1 1XJ, UK
b
School of Mechanical and Aerospace Engineering, Queens University Belfast, Ashby Building, Stranmillis Road, Belfast,
North Ireland BT9 5AH, UK
a r t i c l e i n f o
Article history:
Received 25 March 2007
Received in revised form
3 November 2007
Accepted 20 December 2007
Keywords:
Contact stresses
Friction model
Metal forming
Plasticity
Rolling
a b s t r a c t
There have not been any friction models applied to successfully predict distributions of con-
tact stresses in at rolling yet, in particular for the neutral plane. In this paper, the dynamic
friction model (DFM) is expressed as a combination of both denitions of the viscosity and
the friction, and is employed to derive underlying mathematical expressions of forces in
at rolling. The model is validated through experimental results obtained by Lenard et al. in
the literature for various rolling processes, hot rolling, warm rolling and cold rolling of alu-
minium. By comparisons of the experimental data withthe results predicted by the dynamic
friction model, Amontons-Coulombs friction model and the constant friction model, it is
found that the application of the dynamic friction model leads to a better solution to predic-
tion of contact stresses at the neutral plane. It is believed that the dynamic friction model
could extensively be used to resolve dynamic plasticity problems of solids.
Crown Copyright 2007 Published by Elsevier B.V. All rights reserved.
1. Introduction
Friction does always prevail at the interface of contact bod-
ies, elements or things. There are a number of friction models
used in plasticity, such as the Amontons-Coulomb friction
model (ACM) (e.g., applications by von K arm an (1925) and
Kudo (1960)), the constant friction model (e.g., reviewed by
Schey, 1983), the general friction model by Wanheim and Bay
(Wanheim, 1973; Wanheim et al., 1974; Wanheim and Bay,
1978; Bay, 1987), the absolute constant friction stress model by
e.g., Orowan (1946), Alexander (1955), Tan et al. (1998), Levanov
friction model (Levanov, 1997), Anand friction model (Anand,
1993), the empirical friction model by Bay et al. (2002), and Tan
et al. (1999). For some rigid contact bodies with elastic defor-
mation, Amontons-Coulomb friction model can be modelled

Corresponding author at: School of Mechanical and Aerospace Engineering, Queens University Belfast, Ashby Building, Stranmillis Road,
Belfast, North Ireland BT9 5AH, UK. Tel.: +44 28 90974179; fax: +44 28 90975598.
E-mail address: Xincai Tan@hotmail.com (X. Tan).
with extreme accuracy (reviewed by Dowson, 1979; Bowden
and Tabor, 1954). For plasticity problems, some friction mod-
els have been used to successfully evaluate applied loads,
material ows and deformation, as reviewed by such as Schey
(1983), Bay and Gerved (1984), Ginzburg (1985), Tan (2002).
However, none of these friction models has been applied to
satisedly predict distribution of local contact stresses in at
rolling, especially for the neutral point/plane.
Material forming through at rolling is one of the typi-
cal dynamic plasticity problems. In production, the contact
stresses at the interface of roll and workpiece during rolling
do signicantly inuence on productivity, product cost, prod-
uct quality, and tool life cycle. To properly analyse a process
of rolling, correct evaluation of the contact stresses in the roll
gap is essential for process design and product development.
0924-0136/$ see front matter. Crown Copyright 2007 Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.jmatprotec.2007.12.080
j ournal of materi als processi ng technology 2 0 7 ( 2 0 0 8 ) 222234 223
A great deal of effort has been made to model and measure
distributions of the contact stresses. For example, Lenard et
al. (Hum et al., 1996; Lenard and Malinowski, 1993; Lenard,
1992) measured distributions of both the normal pressure and
the friction stress in various rolling processes, hot rolling,
warm rolling and cold rolling of Al 1100. Until now, distribu-
tions of the contact stresses at the interface between roll and
workpiece were experimentally measured with success. How-
ever, theoretical prediction in distribution of the local contact
stresses has never been satisfactorily achieved due to the lack
of suitable friction model.
The inaccuracy of the conventional theory of rolling partly
results from ignorance of the material velocity. Roll speed
affecting roll forces has been experimentally observed by
a number of researchers. Azushima (1978) examined coef-
cient of friction, surface qualities, and oil lm thickness
using a high-speed test mill, and observed that oil lm thick-
ness decreases with increasing reduction, but increases with
increasing rolling speed; and the order of the coefcient of
friction decreases with the magnitude of the rolling speed.
Matsui et al. (1984) conducted high-speed rolling tests up to
1500m/min and pendulum type of friction test and showed
that with increasing rolling speed, mean rolling pressure and
reduction in height decreases, oil lm thickness increases.
Azushima and Miyagawa (1984) experimentally investigated
lubricant behaviours in cold sheet rolling and showed that
the inuence of the roll speed on the coefcient of friction is
signicant: values of the coefcient of friction decrease with
increasing roll speed. Lin et al. (1991) conducted experiments
in cold strip rolling and showed that with increasing rolling
speed, the values of the forward slip decreases, the coef-
cient of friction decreases, whereas torque increases slightly.
The effects of the reduction, roll speed, lubricant-type and
viscosity on the roll separating forces, forward slip and result-
ing specimen surface roughness during the cold rolling of
aluminium have been studied by Zhang and Lenard (1992).
Increasing speed and/or viscosity was found to lower the for-
ward slip and the coefcient of friction.
Most recently, Tan (2007) has developed a new friction
modelthe dynamic friction model and has successfully
applied it to establish solution to the plane-strain com-
pression. The basis of this model was combination of solid
mechanics with uid mechanics. In this hybrid model (a)
Amontons-Coulomb friction model and the denition expres-
sion of viscosity have been jointly taken into account and (b)
the friction stress is related to not only the ow stress and
dimension of material, but also material velocity.
In this paper, the dynamic friction model is applied to
establish a new solution for the at rolling process, to pre-
dict the contact stresses in particular friction stress. After
briey reviewing the prediction and measurement of the local
contact stress in at rolling in this section, the dynamic fric-
tion model is put forward. Following establishment of the
mathematical model for at rolling applying the dynamic fric-
tion model, comparisons of the predicted results with the
experimental data of Lenard et al. in the literature are pre-
sented. Through comparison of the experimental data with
theoretical results predicted by the dynamic friction model,
Amontons-Coulomb friction model and the constant fric-
tion model in cold rolling, and Alexanders solution for both
warm rolling and hot rolling, conclusions are nally drawn
out.
2. The dynamic friction model
The dynamic friction model was suggested by Tan (2007) and
its formation can be seen as follows: in uid mechanics, the
viscosity, (eta), is dened as division of the shear stress,
(tau), by the velocity gradient, dV

/dz, (reviewed by, e.g.,


Roberson and Crowe, 1997; Papanastasiou, 1994; Schey, 1983):
=

dV

/dz
(1)
Or the shear stress (normal viscous stress) of a uid near a
wall is expressed by
=
dV

dz
(2)
where V

is the velocity parallel to the friction stress, and z


is the coordinate in the direction normal to the velocity. The
viscosity as a measure of the resistance of a uid to deform
under shear stress has been extensively observed in various
uids.
In solid mechanics, experimental results of plastic defor-
mation in the compression twist tests by Tan (1999) and the
forward rod extrusion tests by Tan et al. (2003) show that the
mean friction stress is proportional to the mean normal pres-
sure, i.e., following Amontons-Coulomb friction model which
is actually the denition expression of friction:
= p (3)
where (mu) is the coefcient of friction and p is the normal
pressure.
In a plasticity problem such as at rolling, the material
ows through the roll gap resulting in its plastic deformation.
It is obvious that the local frictionstress characterizes not only
solidmechanics, but alsouidmechanics. This means that the
local friction stress should be considered as a combination of
Eq. (2) with Eq. (3), i.e., the friction stress should be propor-
tional to both the time rate of strain, dV

/dz, and the normal


pressure, p. Therefore, a DFM is formed that the friction stress
at the interface between roll and workpiece is proportional
to both the time rate of strain and the normal pressure. The
friction stress can be then given by
=
dV

dz
p (4)
where (beta) is called the coefcient of dynamic friction
which is combined with the characters of the viscosity, , in
Eq. (2) and the friction coefcient, , in Eq. (3). Similar to and
, the value of for a given plasticity problem has to be deter-
mined by experiment. Since dV

is measured in mm/s; dz in
mm; and and p are measured inMPa; canbe thenexpressed
in s, the same as the unit of time.
224 j ournal of materi als processi ng technology 2 0 7 ( 2 0 0 8 ) 222234
3. Application of the dynamic friction
Model
Before applying the model, the following assumptions are
made for the at rolling: plane strain compression prevails
the whole deforming region; roll attening under load occurs,
and the effective radius (known as Hitchcock radius) is larger
thanthe nominal radius; andthe material deformationstrictly
follows the volume constancy.
3.1. Velocities
According to the friction denition in physics (e.g., Tipler,
1999), friction is the resistance of moment at the interface
between two bodies, or two elements, or two particles, or two
things. Material moment in at rolling is always in a single
direction, in the rolling direction. If friction resists the mate-
rial moving, the friction direction has to be opposite to the
material moment direction, or rolling direction, for the whole
deforming region. Both forward slip and backward slip zones
will be determined by the material acceleration relative to
the roll surface during rolling. The element (the trapezoid
with solid lines) used in the elementequilibrium approach is
shown in Fig. 1. The deforming region is symmetric about the
x-axis. This implies that the state of force, velocity and defor-
mation is a mirror image to the x-axis. Material velocities of
the element at both the entry and the exit, and the velocity
of the roll surface are shown in the bottom half (B) of the g-
Fig. 1 Equilibrium of forces in the deformation zone and
various velocities in the roll gap during at rolling. A:
Equilibrium of forces and B: pattern of velocities.
Table 1 Components of the increment, pressure,
friction stress, and velocity in the deformation zone in
Fig. 1
Component Increment Pressure Friction
stress
Velocity
Interface dl =Rd p V
R
x dx =Rcos d p
x
=psin
x
= cos V
x
=V
R
cos
z dz =Rsind p
z
=pcos
z
= sin V
z
=V
R
sin
ure. The material movement in at rolling exerts a resistance
to motion at the interface between roll and workpiece, and
the resistance results from two types of forces acting on the
workpiece: shear force and pressure force. The corresponding
stresses for the element can be seen in the top half A of Fig. 1.
Components of the increment, pressure, friction stress, and
velocities in the deforming region for the interface between
roll and workpiece, x-coordinate, and z-coordinate presented
in Fig. 1 are shown in Table 1.
As shown in half B of Fig. 1, the volume rate of material
ow due to volume constancy can be written as
Wh
0
V
0
= W(2z)V
x
= Wh
1
V
1
(5)
where W is the width of the strip; h
0
and h
1
are the initial
thickness at the entry side and the nal thickness at the exit
side, respectively; z is half height (coordinate) in the deform-
ing region corresponding to x-coordinate; V
0
, V
x
, and V
1
are
the velocities in x direction corresponding to h
0
, z, and h
1
,
respectively. The material velocity in x direction is then given
by
V
x
=
h
0
V
0
2z
=
h
1
V
1
2z
(6)
dV
x
dz
=
h
0
V
0
2z
2
=
h
1
V
1
2z
2
(7)
It is noted that forward slip has been observed in experi-
ments, for example, by Lenard et al. (Hum et al., 1996; Lenard
and Malinowski, 1993; Lenard, 1992). From the denition of
forward slip, S
f
=(V
1
V
R
)/V
R
, so V
1
can be given as a function
of S
f
:
V
1
= V
R
(1 +S
f
) (8)
where V
R
is the linear velocity of the roll surface.
In theory, the nominal contact length of the deforming
region is L
c
=
_
Rh h
2
/4, where R is the nominal radius
of the roll; h=h
0
h
1
is the draft during rolling. In fact, due
to the effect of the rolls attening, the effective contact length
of the deforming region, L
e
, is larger than the nominal con-
tact length (L
c
). Lenard et al. (Hum et al., 1996; Lenard and
Malinowski, 1993; Lenard, 1992) measured the values of the
forward slip by using S
f
=(L
e
L
c
)/L
c
, for example, thus actual
values for the effective contact length could be given by
L
e
= (1 +S
f
)L
c
(9)
Corresponding to the forward slip, the effective radius (or
Hitchcock radius), R
e
, is larger than the nominal radius R, and
j ournal of materi als processi ng technology 2 0 7 ( 2 0 0 8 ) 222234 225
can be expressed as
R
e
=
L
2
e
h
+
h
4
= R +R (10)
where R is the difference of the radii between the deformed
arc and the nominal arc, and R = L
2
c
(2S
f
+S
2
f
)/h. An approx-
imate value of the R will be estimatedfor eachrolling process
during determination of the contact stresses.
For a given deforming region, the curve of the interface
between roll and the strip is the same as the deformed roll
prole, x
2
+(z z
0
)
2
= R
2
e
, hence half height of the strip corre-
sponding to x is given by
z = z
0

_
R
2
e
x
2
(11)
where z is the half height of the strip in the deforming region;
z
0
is the distance between the centre of a deformed roll and
the centre of the exit plane; as shown in Fig. 1. The z
0
will be
a constant for a given rolling process and it can be expressed
as
z
0
= R
e
+
h
1
2
(12)
Differentiating Eq. (11) with respect to x leads to
dz
dx
=
x
_
R
2
e
x
2
(13)
The material velocity in z direction can be obtained as
V
z
=
dz
dt
=
dz
dx
dx
dt
=
x
_
R
2
e
x
2
V
x
(14)
The element in Fig. 1 is of a trapezoid of differential length,
dx, at an arbitrary length, x, and with a shear stress, , at each
decline surface (parallel to the surface of the roll). Assum-
ing that the stresses in the metal ow direction (parallel to
the x direction),
x
, and in the z direction,
z
, are principal
stresses, i.e.,
z
=
1
,
y
=
2
and
x
=
3
. The Tresca ow rule
is employed:
1

3
=
f
, where
f
is the ow stress. At the
interface,
z
=p
z
=pcos as shown in Table 1, where (theta)
is the angle of the contact arc for the considered point (x, z) as
shown in Fig. 1. Since the angle is very small, and cos 1
(in the present work, the smallest value cos 0.991, where
the angle of whole contact arc), so it is assumed
z
=p. Thus
the relationship between p and
x
is
p
x
=
f
(15)
also
dp = d
x
(16)
For the stresses shown in half A of Fig. 1, there are
2[
x
z (
x
+d
x
)(z +dz) 2(zd
x
+
x
dz)] due to longitudinal
stress, 2(pdx/cos ) sin =2pdz due to radial pressure on both
rolls, and 2(dx/cos ) cos =2dx due to friction against both
rolls. With plane strain compression, equilibrium between
the stresses acting on the metal element will prevail. Conse-
quently, the equilibrium of forces in the x direction (

F
x
= 0)
yields:
2(zd
x

x
dz +dx +pdz) = 0 (17)
The friction stress Eq. (4) at the considered point (x, z) in the
deforming region can be expressed by
=
dV
x
dz
p (18)
By the substitution of Eqs. (15), (16) and (18), Eq. (17) then
becomes
dV
x
dz
=

f
p
dz
dx
+
z
p
dp
dx
(19)
By the substitution of Eq. (13), Eq. (19) is
dV
x
dz
=

f
p
x
_
R
2
e
x
2
+
z
p
dp
dx
(20)
This equation can be integrated directly
V
x
=

f
p
xz
_
R
2
e
x
2
+
z
2
2p
dp
dx
+C
1
(21)
The constant C
1
is determined from the boundary condition
at exit. V
x
=V
1
when z =0, and x =0, so C
1
=V
1
, where V
1
is the
material velocity at the exit side. Thus, integration of Eq. (20)
gives
V
x
=

f
p
xz
_
R
2
e
x
2
+
z
2
2p
dp
dx
+V
1
(22)
3.2. Acceleration vs. friction direction
Dividing both the numerator dV
x
and the denominator dz in
Eq. (18) by the differential of time dt, the relationship between
friction and acceleration of material can be obtained by
=
dV
x
/dt
dz/dt
p =
a
x
V
z
p (23)
where a
x
is the acceleration of material and its direction is the
same as the velocity in x direction, V
x
; and V
z
is the velocity
of material in z direction normal to the acceleration.
The local friction stress distributes along the rotating roll
at the interface between roll and workpiece, so the mov-
ing roll surface has to be taken as the frame of reference.
Consider a differential element of material moving from a
position x
i
with a velocity relative to the roll surface in x
direction, V
ri
, to another position x
i+1
with another veloc-
ity relative to the roll surface in x direction, V
ri+1
, in the
roll gap, as shown in Fig. 2(a). Their distance is d =|x
i
x
i+1
|
and the element deforms following the volume constancy.
The time for the element movement is dt. Correspondingly,
the material acceleration relative to the roll surface is then
a
ri+1
=dV
r
/dt =(V
ri+1
V
ri
)/dt. Since the average velocity rela-
tive to the roll surface is V
rav
=(V
ri+1
V
ri
)/2, the distance is
226 j ournal of materi als processi ng technology 2 0 7 ( 2 0 0 8 ) 222234
Fig. 2 Various velocities in the deforming region. (a)
Differential element of material moving from x
i
to x
i+1
in
the roll gap, and velocities for the neutral point, the forward
slip zone and the backward slip zone, V
n
, V
f
, and V
b
at the
interface between roll and workpiece during at rolling. (b)
Distributions of the material velocities in both x and z
directions, V
x
and V
z
, and the material velocity relative to
the roll surface in x direction, V
rx
, along the contact length
for an exemplar rolling process.
then given by d = V
rav
dt = (V
2
ri+1
V
2
ri
)/(2a
ri+1
). The accelera-
tion relative to the roll surface is thus
a
ri+1
=
V
2
ri+1
V
2
ri
2d
=
V
2
ri+1
V
2
ri
2|x
i
x
i+1
|
(24)
FromEq. (23), the directionandmagnitude of the frictionstress
at the interface between roll and workpiece is dependent on
the material acceleration relative to the roll surface.
Fig. 2(a) shows various material velocities for the neutral
point, the forward slip zone and the backward slip zone, V
n
, V
f
,
and V
b
, respectively. Fig. 2(b) shows example material veloci-
ties in both x and z directions, V
x
and V
z
, and material velocity
relative to the roll surface in x direction, V
rx
, along the contact
length in the deforming region for a typical rolling process.
Whenmaterial moves fromentry to exit, V
x
increases fromthe
material velocity at entry, V
0
, to the material velocity at exit,
V
1
, whereas V
z
decreases from its initial magnitude to zero.
The material velocity relative to the roll surface in x direction,
V
rx
, however, fromnegative in the backward slip zone through
zero at the neutral point and then to positive in the forward
slip zone.
Variations of material acceleration relative to the roll sur-
face will result in varying local friction stress in both direction
and magnitude.
(1) At the neutral point N, x =L
n
(where L
n
is the contact length
of neutral plane, as showninFig. 2(a)), the material velocity
is equal to the roll surface velocity, i.e., V
n
=V
R
, so corre-
spondingly, the material velocity relative to the roll surface
is equal to zero, i.e., V
rn
=V
n
V
R
=0, though both V
n
and
V
R
are moving with a signicant velocity relative to the
Cartesiancoordinates seeninFig. 2. The material instanta-
neous acceleration relative to the roll surface is also equal
to zero, i.e., a
rn
=0, so the friction stress is zero according
to Eq. (23). Friction stress of zero at the neutral point has
been observed by a number of experiments (Hum et al.,
1996; Lenard and Malinowski, 1993; Lenard, 1992). Exam-
ples can be seen in Figs. 35.
(2) For the backward slip zone (NE in Fig. 2(a)), L
n
<x L
e
, the
material velocity is slower than the roll surface velocity,
i.e., V
b
<V
R
, thus the material velocity relative to the roll
surface V
rb
=V
R
V
b
<0. Let V
ri+1
=V
rn
=0, and V
ri
=V
rb
, the
material acceleration relative to the roll surface in the
backward slip zone is then given by a
rb
= (V
2
rn
V
2
rb
)/2|x
L
n
| = V
2
rb
/2|x L
n
| < 0. In this case, the friction stress
will have a direction resisting the deceleration of mate-
rial moment opposite to the suggested direction shown
in Fig. 1, or as the same as the rolling velocity direction.
Therefore, friction stress in the backward slip zone will be
of a different mathematical sign fromthe normal pressure
indicating different fromthe suggested direction. This fact
has been observed by experiments in hot, warm, and cold
rollings (Hum et al., 1996; Lenard and Malinowski, 1993;
Lenard, 1992) as shown in Figs. 35.
(3) For the forward slip zone (FN in Fig. 2(a)), 0x <L
n
, the
material velocity is faster than the roll surface velocity,
i.e., V
f
>V
R
, so the material velocity relative to the roll
surface V
rf
=V
f
V
R
>0. Let V
ri+1
=V
rf
and V
ri
=V
rn
=0, the
material acceleration relative to the roll surface in the for-
ward slip zone is then given by a
rf
= (V
2
rf
V
2
rn
)/2|L
n
x| =
V
2
rf
/2|L
n
x| > 0. The friction stress will be in the nor-
mal way resisting the acceleration of material moment, or
opposite tothe rolling velocitydirection, as the same as the
suggested direction shown in Fig. 1. Therefore, the friction
stress will be of the same mathematical sign as the nor-
mal pressure indicating both the stresses with the same
as the suggested direction in the forward slip zone. This
fact is also observed from different experiments (Hum et
al., 1996; Lenard and Malinowski, 1993; Lenard, 1992), as
seen in Figs. 35.
Moreover, the phenomenon on material relative accelera-
tion with different values and different directions has been
experimentally observed. Applying the digital image correla-
tion technique to measure the velocity distribution in roll gap
j ournal of materi als processi ng technology 2 0 7 ( 2 0 0 8 ) 222234 227
Fig. 3 Comparisons of both the normal pressure (p) and the friction stress (fs) between experimental data (ex) obtained by
Hum et al. (1996) and theoretical results predicted by present solution with the dynamic friction model (DFM), and
Alexanders solution (AS) with the absolute constant friction model for hot rolling of Al 1100H14.
during cold rolling of aluminium alloy by Li et al. (2003), it
was found that the materials movement direction is always
parallel to the rolling direction, but the ow rate of the mate-
rial movement will be divided into two parts with opposite
directions. At the neutral point, the ow (deformation) rate is
zero, and it indicates that the roll surface velocity is identical
to the material velocity. This implies that different directions
of the ow rate of material will occur in the two deforming
zones, backward slip zone and forward slip zone. The ow
rate of material represents the material acceleration relative
to the roll surface. This experimental observation is identical
with the above discussion: the material acceleration relative
to the roll surface has different mathematical signs for both
the backward slip zone and the forward slip zone.
It should be pointed out that the alteration of friction stress
direction from the backward slip zone through the neutral
point to the forward slip zone is successfully represented in
mathematical expressions at this present work. There have
not been any reasonable explanations in theory yet although
experimental facts of such friction stress distribution have
been observed by researchers, as reviewed in Section 1.
3.3. Normal stresses
Eq. (22) can be rewritten as
dp
dx
= (V
x
V
1
)
2
z
2
p +
2
f
z
x
_
R
2
e
x
2
(25)
Let P(x) =(V
1
V
x
)(2/z
2
) and Q(x) =
2
f
z
x
_
R
2
e
x
2
, Eq. (25) then is
the linear differential equation:
dp
dx
+P(x)p = Q(x) (26)
The general solution can be given by the formula
p = e

_
P(x)dx
[
_
Q(x)e
_
P(x)dx
dx +C] (27)
If the power series expansion of exponent
e
u
=1+(u/1!) +(u
2
/2!) +. . . is used, the general solution to
Eq. (27) is
p = exp
_
(V
x
V
1
)
2
z
2
x
__

2
f
z
_
R
2
e
x
2
+
4
f
z
3
(V
1
V
x
)
_

x
2
_
R
2
e
x
2
+
R
2
e
2
arcsin
x
R
e
_
+C
_
(28)
The particular solutionto the present problemcanbe obtained
fromp=
f
, and V
x
=V
1
whenz =h
1
/2, and x =0. The constant of
the integration, C, is givenby C=
f
(1+(4R
e
/h
1
)). By the substitu-
tion of Eq. (8), the normal pressure of Eq. (28) isthen expressed
228 j ournal of materi als processi ng technology 2 0 7 ( 2 0 0 8 ) 222234
Fig. 4 Comparisons of both the normal pressure (p) and the friction stress (fs) between experimental data (ex) obtained by
Lenard and Malinowski (1993) and theoretical results predicted by present solution with the DFM, and AS with the absolute
constant friction model for warm rolling of Al 1100H14.
by
p =
f
exp
_
[V
x
V
R
(1 +S
f
)]
2
z
2
x
__

2
z
_
R
2
e
x
2
+
4
z
3
[V
R
(1 +S
f
) V
x
]
_

x
2
_
R
2
e
x
2
+
R
2
e
2
arcsin
x
R
e
_
+
_
1 +
4R
e
h
1
__
(29)
It was realised that the normal pressure values calculated
by Eq. (29) have some differences in scale fromthe experimen-
tal data after determination of the coefcient of the dynamic
friction, . It is known that the material velocity relative to the
roll surface inuences on the material acceleration relative to
the roll surface and the friction stress. Moreover, the mate-
rial velocity in x direction, V
x
, should be taken the concept
of relative velocity into account, and it will be inuenced by
process conditions resulting in the theoretical value different
from actual value. Hence a linear factor is added to V
x
, i.e., V
x
is replaced by K
p
V
x
. K
p
is called the factor of relative velocity
due to the normal pressure. In this case, calculated value of
the coefcient of dynamic friction, , may differ fromthe orig-
inal value in Eq. (29). So
p
is used to replace and it is called
the coefcient of dynamic friction due to the normal pressure.
Similar to the viscosity, in Eq. (2) and the coefcient of fric-
tion, in Eq. (3), the values of K
p
and
p
have to be determined
by experiment. Therefore, Eq. (29) becomes
p =
f
exp
_
[K
p
V
x
V
R
(1 +S
f
)]
2
p
z
2
x
__

2
z
_
R
2
e
x
2
+
4
p
z
3
[V
R
(1 +S
f
) K
p
V
x
]
_

x
2
_
R
2
e
x
2
+
R
2
e
2
arcsin
x
R
e
_
+
_
1 +
4R
e
h
1
__
(30)
From Eq. (15), the normal stress in the x direction is given by

x
=
f
_
exp
_
[K
p
V
x
V
R
(1 +S
f
)]
2
p
z
2
x
_ _

2
z
_
R
2
e
x
2
+
4
p
z
3
[V
R
(1 +S
f
) K
p
V
x
]
_

x
2
_
R
2
e
x
2
+
R
2
e
2
arcsin
x
R
e
_
+
_
1 +
4R
e
h
1
__
1
_
(31)
3.4. Friction stress
From Eqs. (13) and (25), Eq. (19) can be rewritten as
dV
x
dz
=

f
p
x
_
R
2
e
x
2
+
z
p
_
(V
x
V
1
)
2
z
2
p +
2
f
z
x
_
R
2
e
x
2
_
(32)
j ournal of materi als processi ng technology 2 0 7 ( 2 0 0 8 ) 222234 229
Fig. 5 Comparisons of both the normal pressure (p) and the friction stress (fs) between experimental data (ex) obtained by
Lenard (1992) and theoretical results predicted by the DFM, Amontons-Coulomb friction model (ACM), and the constant
friction model (CFM) for cold rolling of Al 1100H14.
By substitution of Eqs. (8) and (32) into Eq. (18), the friction
stress is then given by
=

f
x
_
R
2
e
x
2
+
2[V
x
V
R
(1 +S
f
)]p
z
(33)
Again, the concept of the material relative velocity and the
actual process conditions different fromthe theoretical should
be considered for the material velocity in x direction within
the deforming region in Eq. (33). Similar to the modication of
the normal pressure expressed in Eq. (30), [V
x
V
R
(1+S
f
)] are
replaced by [K
f
V
x
V
R
(1+S
f
)]
f
, and then Eq. (33) becomes
=

f
x
_
R
2
e
x
2
+
2[K
f
V
x
V
R
(1 +S
f
)]
f
p
z
(34)
where K
f
is called the factor of relative velocity due to the fric-
tion stress; and
f
is called the coefcient of dynamic friction
due to the friction stress. Similar to the viscosity, , and the
coefcient of friction, , the values of K
f
and
f
have to be
determined by experiment.
It is worth noting that the contact stresses expressed in
the Eqs. (30) and (34) do not directly include the temperature
parameter which has been proven to have a signicant inu-
ence on the contact stresses. The effect of the temperature,
however, is indirectly involved from the material ow stress,

f
, the coefcients of dynamic friction due to both the normal
pressure and the friction stress,
p
and
f
, and the factors of
relative velocity due to the normal pressure and due to the
friction stress, K
p
and K
f
. It has been observed that variation
of temperature will signicantly result in varying ow stress
of material, for example, by Tan et al. (2005).
230 j ournal of materi als processi ng technology 2 0 7 ( 2 0 0 8 ) 222234
Table 2 Flow curves of the samples at different temperatures
Temperature (

C) Flow curves (MPa) Data source


504
f
=10+9
0.06
Regressed from experimental data
by Anand and Zavaliangos (1990)
487
f
=15+9.2
0.11
Regressed from data by Anand and
Zavaliangos (1990)
300
f
=111.29+25.623 for 0.1;
f
=113.85(10)
0.363
for <0.1 Experimental data by Lenard and
Malinowski (1993)
Room temperature
f
=36.75(1+726.7)
0.2099
Experimental data by Lenard (1992)
4. Validation of the dynamic friction model
To validate the above mathematical expressions of the con-
tact stresses, it is necessary to compare the predicted results
by the DFM with experimentally measured data, as well as
with the predicted curves by other prevailing friction models,
e.g., Amontons-Coulomb friction model and the constant fric-
tion model. Because none of experimental data on the normal
stress in x direction,
x
, is available in the literature, and it
is extremely difcult to be measured, comparisons of predic-
tions and experiments for both the normal pressure and the
friction stress, p and , are carried out.
The experimental data were chosen from the work of
Lenard et al. at the University of Waterloo in the literature.
Various rolling types were included, hot rolling by Hum et al.
(1996), warmrolling by Lenard and Malinowski (1993), and cold
rolling by Lenard (1992) with a similar material, aluminium
alloy 1100-H14. The material temperatures for hot rolling were
up to 504

C and 487

C; for warm rolling 300

C; and for cold


rolling room temperature. Flow curves for warm rolling and
cold rolling were available from the papers by Lenard et al.,
whereas the ow curves for hot rolling were regressed from
the experimental data of a similar material with the same
alloy (1100 aluminium) measured by Anand and Zavaliangos
(1990). Table 2 lists the ow curves of the materials for the
three types of rolling. A 250100mm two high-rolling mill
was usedfor all the three types of rollings. Different roll speeds
were applied, from3 to 100 rpmand values of the reduction in
height, , rangedfrom27%to 39%. Table 3 summarises detailed
experimental parameters used in the present work.
When the DFM was used, rst, experimental data for a
given process were plotted out. Secondly, the values for both
factors due to the normal pressure,
p
and K
p
, in Eq. (30) were
chosen to determine the normal pressure to t with the exper-
imental data. Finally, the values for both factors due to the
friction stress,
f
and K
f
, in Eq. (34) were chosen to obtain
the friction stress to t with the experimental data. Through
adjustment of the tting parameters in prediction, it was real-
ized that:
(a) For a predicted curve of the normal pressure, the curve
level increases with decreasing
p
, and the peak shape
appears sharper with increasing K
p
.
(b) For a predicted curve of the friction stress, the magnitude
of the curve increases with decreasing
f
, and the contact
length of the neutral plane increases with increasing K
f
.
Traditionally, when Amontons-Coulomb friction model or
the constant friction model is applied to a rolling process, the
value of the friction factor/coefcient is normally determined
by adjustment of a considered curve of the predicted normal
pressure to match with the experimental data. How to t the
corresponding experimental data of the friction stress is not
consideredat all. Thus the tting curves of bothprevailing fric-
tion models here are obtained in the same way, just adjusting
values of the coefcient of friction or the friction factor to t
with the experimental data of the normal pressure, and pre-
dicted curves of the friction stress were then plotted adapting
these tting friction parameters.
Validations of the DFM are shown in Figs. 35. The stresses
are plotted on the ordinates, while the distances along the roll
gap are shown along the abscissa and it starts from the exit
plane to the entry plane as the same as x shown in Fig. 1. For
each graph, the top set of curves is for the normal pressure, p,
and the bottom set for the friction stress, .
4.1. Hot rolling
In the conventional theory of hot rolling, one of the well-
known approaches is the Alexander solution (AS) (Alexander,
1972) in which the absolute constant friction stress model is
used. For the backward slip zone, the normal pressure is given
by
p
0
= 2K
_
1 +ln
h
h
0

1
2
tan +
1
2A
ln
tan(/2 +/4)
tan(/2 +/4)
+
1
A
_
A
2
1
_
arctan
_
_
A+1
A1
tan
_

2
_
_
arctan
_
_
A+1
A1
tan
_

2
_
___
(35)
= K (36)
For the forward slip zone,
p
1
= 2K
_
1 +ln
h
h
1
+
1
2
tan +
1
2a
ln
_
tan
_

2
+

4
__
+
1
A
_
A
2
1
arctan
_
_
A+1
A1
tan
_

2
_
__
(37)
= +K (38)
j ournal of materi als processi ng technology 2 0 7 ( 2 0 0 8 ) 222234 231
Table 3 The experimental parameters for the processes of hot rolling (Hum et al., 1996), warm rolling (Lenard and
Malinowski, 1993), and cold rolling (Lenard, 1992) used in the present work
Hot rolling Warm rolling Cold rolling
Samples: (1100-H14)
Chemical composition (wt%) Mn: 0.05; Si: 1.0; Zn: 0.1; Cu: 0.05; Al: remainder
Rolling temperature (

C) 504 487 300 300 Room temperature


Initial thickness (mm) 6.32 6.30 6.28 6.28 3.17 3.17
Initial width (mm) 50 50 50 50 50 50
Initial length (mm) 200 200 200 200 200 200
Reduction (%) 30.4 39.21 28.46 37.86 27.44 36.91
Two high-rolling mill
Roll diameter (mm) 250 250 250 250 250 250
Roll length (mm) 100 100 100 100 100 100
Constant roll speed (rpm) 80 100 12 12 3 3
Roll surface velocity (mm/s) 1046.67 1308.33 157 157 39.90 39.90
Motor dc dc dc dc dc dc
Constant torque (kW) 42 42 42 42 25 25
Roll surface roughness Ra (m) 0.18 0.18 0.2 0.2
Lubricant Emulsion (2% oil +water)
Forward slip, S
f
0.03456 0.04558 0.077 0.106 0.0548 0.0694
where A is the constant and A=1+(h
1
/2R
e
); h is the local thick-
ness and h=2z; and K is the maximum shear ow stress. At
this present work, a predicted curve adopting the shear ow
stress value could not t the experimental data of the normal
pressure at all from Eqs. (35) and (37). Therefore, an absolute
constant frictionstress value is chosenas anapproximate half
of the order of the normal stress measured at the entry end of
the deforming region for a given hot rolling process.
Fig. 3 shows comparisons of both the normal pressure
and the friction stress between experimental data obtained
by Hum et al. (1996) and theoretical results predicted by the
present solution with the DFM Eqs. (30) and (34), and the AS
for hot rolling of Al 1100H14. For a normal pressure curve,
the starting point at x =0 depends on the magnitude of the
ow stress (
f
) for the DFM, and on the value of the maxi-
mum shear ow stress (K) for the AS. The higher the value of
K, the greater the magnitude of the normal pressure at both
the ends of deforming region, i.e., the exit plane and the entry
plane. It can be seen that the pressure peak is smooth round
for the DFM, but is quite sharpfor the AS. At the neutral point, a
curve of the friction stress for the DFMtransfers smoothly and
has two signicant peaks with different mathematical signs (+
and ). But for the AS, the magnitude of a friction stress curve
is a horizontal constant and transfers sharply from positive
to negative at the same point, the neutral point; correspond-
ingly, the friction hill for the normal pressure coincides at the
neutral point.
4.2. Warm rolling
Similar to hot rolling, the AS Eqs. (35)(38) were used to predict
the contact stresses for the warmrolling. Fig. 4 shows compar-
isons of experimental data for both the normal pressure and
the friction stress obtained by Lenard and Malinowski (1993)
with theoretical results predicted by the present solution with
the DFM Eqs. (30) and (34), and the AS for warm rolling of Al
1100H14. It can be seen that the predicted curves by the DFM
t the experimental data quite well and better than those by
the AS, in particular for the neutral plane.
Similar topredictions tothe hot rolling, at the neutral point,
a curve of the friction stress for the DFM transfers smoothly
and has two signicant peaks with different signs (+ and ).
But for the AS, the magnitude of a friction stress curve is a
horizontal constant with a bit of underestimation and trans-
fers sharply from positive to negative at the same point. The
location of the neutral point predicted by the AS appears quite
a big difference from the experimental results.
4.3. Cold rolling
For cold rolling, distributions of contact stresses can be pre-
dicted by application of Amontons-Coulomb friction model or
the constant friction model (CFM). Bland and Ford (1948) used
Amontons-Coulomb friction model to resolve the problem in
cold rolling. For the backward slip zone,
p
0
=
f
2z
h
1
exp[(H
0
H)] (39)
= p (40)
and for the forward slip zone,
p
1
=
f
2z
h
1
exp(H) (41)
= +p (42)
where H
0
and H are functions which are dened as H
0
=
2
_
R/h
1
arctan
_
_
R/h
i

_
andH = 2
_
R/h
1
arctan
_
_
R/h
1

_
.
To t the experimental curves, values of 0.13 and 0.09 was
chosen for the reductions of 27.44% and 36.91%, respectively.
When the CFM is employed, it is assumed that the neutral
plane is located at the middle of the contact length. Similar
to the plane strain compression, the variation in pressure is
available in the literature (e.g., book by Mielnik (1991)). For the
232 j ournal of materi als processi ng technology 2 0 7 ( 2 0 0 8 ) 222234
Table 4 Parameters used for the dynamic friction model to calculate the normal pressure and the friction stress in Figs.
35
Rolling type (%)
p
(s) K
p

f
(s) K
f
z
0
(mm) L
e
(mm) L
c
(mm) R (mm) K
x0
K
x0.5
Fig. 3 Hot 30.40 0.0019 0.87 0.0028 1.13 177.20 18.31 15.47 50 0.1 1.0
39.21 0.0014 0.92 0.0032 1.20 134.91 18.05 17.50 8 0.1 1.0
Fig. 4 Warm 28.46 0.0084 0.56 0.002 1.22 187.25 18.16 14.92 60 0.1 1.0
37.86 0.010 0.82 0.013 1.30 154.95 19.04 17.20 28 0.1 1.0
Fig. 5 Cold 27.44 0.011 0.26 0.032 1.13 184.15 12.61 10.42 58 0.1 0.5
36.91 0.012 0.46 0.022 1.20 181.0 14.50 12.08 55 0.01 0.5
backward slip zone
p
0
=
f
_
1 +
4m(L
c
x)
h
1
+h
0
_
(43)
= mk (44)
and for the forward slip zone
p
1
=
f
_
1 +
4mx
h
1
+h
0
_
(45)
= +mk (46)
where m is the friction factor; k is the shear ow stress. In
the present work, values of the friction factor were chosen as
0.26 and 0.18 for both the reductions of 27.44% and 36.91%,
respectively.
By comparisons of the experimental data with the curves
for distributions of both the friction stress and the normal
pressure predicted by the DFM (Row I), the ACM (Row II), and
the CFM (Row III) in Fig. 5, it can be seen that there are dif-
ferences among the theoretical results predicted by the three
models compared with experimental data at and around the
neutral point.
Referring to the normal pressure, a curve for the DFMhas a
smooth peak. At the centre of the sample, or the neutral point,
acurve is spine-like for the twoprevailing frictionmodelsthe
ACM and the CFM. There are no such sharp spine-like peaks
observed in experiment.
As to the frictionstress, a predictedcurve basedonthe DFM
does always pass through the neutral point (the friction stress
being zero), and the curve varies with the length position of
the sample in x direction, as shown in Row I of Fig. 5. For the
ACM, the absolute values of the sharp peak points are right
located at the theoretical neutral point and then decreases
outwards. The trend of the predicted friction stress curve is
quite similar to its predicted normal pressure distribution, as
seen in Row II of Fig. 5, and it is denitely opposite to the
trend of the experimental curves. For the CFM, the curve is a
constant horizontal line although at both ends of the entry
and the exit there are some slight declines due to inuence
of reduction of the ow stress. The predicted absolute value is
almost the same, no matter where the position is, as plotted in
RowIII of Fig. 5. The magnitudes of the frictionstress predicted
by both the ACM and the CFM are generally underestimated,
i.e., their theoretical curves are all lower thanthe experimental
data.
4.4. Overview
Overall, the curves predicted by the three friction models
approximately agree with the experimental data for both the
normal pressure and the friction stress. But at the neutral
plane, application of the DFM leads to a better solution. To
model trends of the friction stress distribution, Tselikov (1958)
divided the deforming region into three zones in which differ-
ent friction models had to be employed. Due to the complexity
of friction stress distribution in at rolling, few researches on
modelling friction stress distribution have been reported. The
predicted trends of the friction stress distributions by the DFM
are similar to the experimental data, although there are some
discrepancies between the predictions and the experiments.
These discrepancies, however, seem to be tolerable for engi-
neering application.
The inuence of the ow stress on the normal pressure is
seen in Figs. 35. With similar sample dimensions, the magni-
tudes of the applied normal pressure for hot rolling are lower
than those for warmrolling since the owstress for hot rolling
is lower than that for warmrolling. The inuence of reduction
on the normal pressure can also be seen. For a given rolling
process, the higher the reduction, the higher the normal pres-
sure will be.
It should be noted that a pressure peak (known as friction
hill) is not always located at the neutral point where the value
of the friction stress is zero. Some evidence can be seen in
Figs. 35, and more are in the literature. For example, Jeswiet
(1995) concluded that the peak normal stress did not coin-
cide with the theoretical friction hill for all reductions, based
on experiments in the cold rolling of aluminium with 15%
reduction and 2.86.8 aspect ratios. Experimental data from
cold rolling aluminium and lead strips with various drafts
by

Swi atoniowski et al. (2004) also supported this fact. This
is quite different from the conventional theory of rolling. By
the DFM, the neutral point of the friction stress curve can be
properly located and at this point friction magnitude is zero.
Table 4 shows parameters used for the DFM to calculate
both the normal pressure and the friction stress in Figs. 35.
Fromhot rolling, to warmrolling, and to cold rolling, the tting
values of
p
and
f
are fromlow, to medium, and to high; while
the values of K
p
are from high, to medium, and to low. This
indicates that these parameters,
p
,
f
and K
p
, are dependent
on material mechanical properties which indicates that the
indirect effect of temperature can be seen in the predictions.
The values of the factor of relative velocity due to the pressure,
K
p
, are in the range between 0.26 and 0.92, i.e., less than 1.
This indicates that the backward slip zone might have more
remarkable inuence on the pressure than the forward slip
j ournal of materi als processi ng technology 2 0 7 ( 2 0 0 8 ) 222234 233
zone. Contrary, the K
f
values are all in the range between 1.13
and 1.30, i.e., more than 1. This indicates that the forward slip
might have more signicant inuence on the friction stress
than the backward slip zone.
It should be pointed out that only a fraction of the theo-
retical value of the ow stress is taken into account when the
contact stresses at/close to the exit plane (x 0) are calculated.
The reduction at the exit plane will always be the maximum
and the corresponding ow stress should also reach the max-
imum in theory. In actual prediction, however, input of the
maximum ow stress resulted from the maximum reduction
leads to an additional small peak for the normal pressure
curve at the exit plane during application of the DFM. There-
fore, the calculated values of the strain at the origin point
(x =0) were given as a fraction of the maximum magnitude of
theoretical true strain, i.e.,
x0
=K
x0
ln(h
0
/h
1
), where
x0
is the
calculated true strain at x =0; and K
x0
the fractional factor and
0<K
x0
1. Similarly, at the adjacent origin point (x =0.5mm),

x0.5
=K
x0.5
ln(h
0
/h
x0.5
), where
x0.5
is the calculated true strain
at x =0.5mm; and K
x0.5
the corresponding fractional factor and
0<K
x0.5
1; h
x0.5
the strip height at x =0.5mm. Values of both
the K
x0
and K
x0.5
chosen for each rolling process are listed in
Table 4. In fact, due to recovery deformation and interwoven
deformationcharacteristics of bothzones of backwardslipand
forward slip, the ow stress at the exit plane would not have
a full scale of the theoretical maximum magnitude. Moreover,
the expansions of the contact lengthof the deforming region
and the roll radius will also signicantly inuence on the
material ow and characteristics at the exit plane. The values
of the difference between the effective radius and the nomi-
nal radius, and the corresponding effective contact length can
be seen in Table 4. Material ow and deformation at the exit
plane are complicated, and the effect of the Hitchcock radius
exists in all the rolling processes at the present work. Further
investigation would be needed.
Strictly speaking, the main different results predicted by
all the friction models are at the neutral plane. The curves
obtained by the DFM have an improvement around the neu-
tral plane. However, in the course of calculation, the DFMhave
four factors to be determined, whereas each of the other fric-
tion models has only one, or m or K. For either of the two
prevailing friction models, two different equations have to be
used to calculate a curve of either the normal pressure or the
friction stress, one for the forward slip zone, and the other for
the backward slip zone; whereas for the DFM, only one equa-
tion is used, Eq. (30) for the normal pressure and Eq. (34) for
the friction stress.
5. Conclusions
Based on the denition expressions of both viscosity of uid
mechanics and friction of solid mechanics, the dynamic fric-
tion model is formed. From fundamentals of friction and
movement dened in physics, a new rolling theory applying
the dynamic friction model to predict contact stresses in hot,
warmandcoldat rolling, has beensuggested. The theoretical
curves at the neutral plane predicted by the dynamic fric-
tion model t with the experimental data better than those
by both Amontons-Coulomb friction model and the constant
friction model, not only for the normal pressure but also for
the friction stress. Especially, proper prediction of friction
stress distributions at the interface between roll and work-
piece should be a breakthrough in the theory of rolling.
(1) Based on fundamentals of physics, the alteration of the
friction stress direction at the interface between roll and
workpiece has been successfully explained by the inu-
ence of material acceleration in theory. It is the rst time
to suggest that the friction stress is related to the material
acceleration and to establish mathematical expression for
their relationship.
(2) Parameters affecting contact stresses in at rolling are not
only the ow stress, the normal pressure, and the dimen-
sions of tool and material, but also the roll speed. All these
parameters are taken into account by application of the
dynamic friction model.
(3) For the dynamic frictionmodel, the coefcients of dynamic
friction due to the normal pressure and due to the friction
stress, and the factor of relative velocity due to the normal
pressure are signicantly dependent on the material prop-
erties and reduction. The factor of relative velocity due to
the friction stress is remarkably dependent on the neutral
point. By choosing these factors, contact stresses can be
properly predicted by using the dynamic friction model.
(4) Until now, the dynamic friction model has been success-
fully employed to resolve plasticity problems of the at
rolling and the plane strain compression. It is believed
that the model could be applied to the other processes of
solid deformation. It is recommended that the dynamic
friction model could be used to computing mechanics of
solid deformation like nite element methods.
r e f e r e nce s
Alexander, J.M., 1955. A slip line eld for the hot rolling process.
Proc. Inst. Mech. Eng. 169, 10211030.
Alexander, J.M., 1972. On the theory of rolling. Proceedings of the
Royal Society of London, Series A. Math. Phys. Sci. 326,
535563.
Anand, L., 1993. A constitutive model for interface friction.
Comput. Mech. 12, 197213.
Anand, L., Zavaliangos, A., 1990. Hot workingconstitutive
equations and computational procedures. Ann. CIRP 39 (1),
235238.
Azushima, A., 1978. Characteristics of lubrication in cold sheet
rolling, in lubrication challenges in metalworking and
processing. In: Proceedings of the First International
Conference, IIT Research Institute, Chicago, IL, USA, June 79,
pp. 17.
Azushima, A., Miyagawa, M., 1984. Evaluation of lubricity and
anti-seizure property of lubricant in cold sheet rolling: an
investigation into friction and lubrication in cold rolling IV. J.
JSTP 25 (285), 915920.
Bay, N., 1987. Friction stress and normal stress in bulk
metal-forming processes. J. Mech. Work. Technol. 14, 203223.
Bay, N., Gerved, G., 1984. Friction and pressure distribution in disk
forging. In: Presented at the 17th International Cold Forging
Group Plenary Meeting, Nagoya, Japan, September.
Bay, N., Eriksen, M., Tan, X., Wibom, O., 2002. An empirical model
for friction in cold forging, keynote paper. In: Proceedings of
the Euromech 435 Colloquium, FWMF (Friction and Wear in
234 j ournal of materi als processi ng technology 2 0 7 ( 2 0 0 8 ) 222234
Metal Forming), LAMIH, Valenciennes, France, June 2002, pp.
105124.
Bland, D.R., Ford, H., 1948. The calculation of roll force and torque
in cold strip rolling with tension. In: IMS, Proceedings,
Institution of Mechanical Engineers, 159, pp. 144153.
Bowden, F.P., Tabor, D., 1954. The Friction and Lubrication of
Solids, Part I. Clarendon Press, Oxford, Part II: 1964.
Dowson, D., 1979. History of Tribology. Longmans/Green,
London/New York.
Ginzburg, V.B., 1985. Basic principles of customized computer
models for cold and hot strip mills. Iron Steel Eng. 62, 2135.
Hum, B., Colquhoun, H.W., Lenard, J.G., 1996. Measurements of
friction during hot rolling of aluminium strips. J. Mater.
Process. Technol. 60, 331338.
Jeswiet, J., 1995. Aspect ratio, friction forces and normal forces in
strip rolling. J. Mater. Process. Technol. 53, 846856.
Kudo, H., 1960. Some analytical and experimental studies of
axi-symmetric cold forging and extrusion. Int. J. Mech. Sci.
Part I: 2, 102127, Part II: 3 (1961) 91117.
Lenard, J.G., 1992. Friction and forward slip in cold strip rolling.
Tribol. Trans. 35, 423428.
Lenard, J.G., Malinowski, Z., 1993. Measurements of friction
during warm rolling of aluminium. J. Mater. Process. Technol.
39, 357371.
Levanov, A.N., 1997. Improvement of metal forming processes by
means of useful effects of plastic friction. J. Mater. Process.
Technol. 72, 314316.
Li, E.B., Tieu, A.K., Yuen, W.Y.D., 2003. Application of digital image
correlation technique to dynamic measurement of the
velocity eld in the deformation zone in cold rolling. Opt.
Lasers Eng. 39, 479488.
Lin, J.F., Huang, T.K., Hsu, C.T., 1991. Evaluation of lubricants for
cold strip rolling. Wear 147, 7991.
Matsui, K., Matsushita, T., Takatsuka, K., Yamaguchi, Y., 1984.
Evaluation of lubricants for rolling of aluminium sheet and
foil. Adv. Technol. Plasticity 1, 247252.
Mielnik, E.M., 1991. Metalworking Science and Engineering.
McGraw-Hill, Inc., New York, pp. 353394.
Orowan, E., 1946, Section V. A simple method of calculating roll
pressure and power consumption in hot at rolling. Iron and
Steel Institute: Special Report, vol. 34, pp. 124146.
Papanastasiou, T.C., 1994. Applied Fluid Mechanics. P T R Prentice
Hall, Englewood Cliffs, NJ, 07632.
Roberson, J.A., Crowe, C.T., 1997. Engineering Fluid Mechanics,
6th edition. John Wiley and Sons, Inc., New York.
Schey, J.A., 1983. Tribology in Metalworking: Friction, Lubrication
and Wear. American Society for Metals, Metals Park.

Swi atoniowski, A., Sobkowiak, P., Gregorczyk, R., 2004.


Experimental investigation of friction phenomena on the
rollmaterial contact surface. J. Mater. Process. Technol.
155156, 15191525.
Tan, X., 1999. Friction modelling in connection with cold forming
processes. PhD Thesis. Department of Manufacturing
Engineering, Technical University of Denmark,
IPT.024.00(MM00.15).
Tan, X., 2002. Comparisons of friction models in bulk metal
forming. Tribol. Int. 35, 385393.
Tan, X., 2007. Friction of plasticity: application of the dynamic
friction model. Proceedings of the Institution of Mechanical
Engineers Part J. J. Eng. Tribol. 221, 115131.
Tan, X., Martins, P.A.F., Bay, N., Zhang, W., 1998. Friction studies at
different normal pressures with alternative ring-compression
tests. J. Mater. Process. Technol. 8081, 292297.
Tan, X., Bay, N., Zhang, W., 1999. A new friction test using simple
upsetting and ow analysis, advanced technology of
plasticity. In: Geiger, M. (Ed.), Proceedings of the sixth
International Conference of Technology of Plasticity (ICTP),
vol. 1. Nuremberg, Germany, September. SpringerVerlag, pp.
365370.
Tan, X., Bay, N., Zhang, W., 2003. Friction measurement and
modeling in forward rod extrusion test. Proceedings of the
Institution of Mechanical Engineers Part J. J. Eng. Tribol. 217,
7182.
Tan, X., Conway, P.P., Sarvar, F., 2005. Thermo-mechanical
properties and regression models of alloys: AISI 305, CK 60,
CuBe2 and Laiton MS 63. J. Mater. Process. Technol. 168,
152163.
Tipler, P.A., 1999. Physics for Scientists and Engineers, 4th
edition. W.H. Freeman/Worth, New York.
Tselikov, A.I., 1958. Present state of theory of metal pressure upon
rolls in longitudinal rolling. Stahl 18, 434441 (in Russian).
von K arm an, Th., 1925. Beitrag zur Theorie des Walzvorganges
(On the theory of rolling), Zeitschrift f ur angewandte
Mathematik und Mechanik, Band 5, Heft 2, pp.139141.
Wanheim, T., 1973. Friction at high normal pressures. Wear 25,
225244.
Wanheim, T., Bay, N., 1978. A model for friction in metal forming
processes. Ann. CIRP 27 (1), 189194.
Wanheim, T., Bay, N., Petersen, A.S., 1974. A theoretically
determined model for friction in metal working processes.
Wear 28, 251258.
Zhang, S., Lenard, J.G., 1992. The effects of the reduction, speed
and lubricant viscosity on friction in cold rolling. J. Mater.
Process. Technol. 30, 197209.

You might also like