Professional Documents
Culture Documents
Abstract: Extensive study of stresses induced during and after winding of coils has led to signicant improvements to the processing of web-like materials for the packaging industries and
consumer goods. The current paper presents an inverse solution for the prediction of the winding
tension prole required to satisfy a specied residual stress distribution in the coil, for non-linear
material properties (in the radial direction), with allowance for large strain and large deformation
in the coil. The inverse solution approach has advantage over the forward solution approach by
directly determining the winding tension prole, which can be readily controlled in a processing
line, once the desired residual stresses in the coil are known. The current paper, together with
several recently published ones, completes the study of stresses in the winding of coils using the
inverse solution approach.
Keywords: winding stresses, winding, coil collapse, coiling, non-linear material, large
deformations
INTRODUCTION
daniel.yuen@bluescopesteel.com
Limited, Old Port Road, Port Kembla NSW 2505, Australia. email:
1522
PROBLEM FORMULATION
Because winding of the coils is under tension (prestressed), the denition of coordinates requires some
care. First, a computational domain is dened Rc
r R R0 where Rc is the undeformed radius of the
core (the former and/or mandrel) and the radial coordinate r denes a particular wrap (Figs 1 and 2).
Specically, r denotes the [(r Rc )/h]th wrap, where
h is the wrap thickness in the unstressed state. Note,
however, that r is not the radial position in the physical coil, which is deformed. In fact, the radius of
the [(r Rc )/h]-th wrap is r + u(r, R) and this may be
signicantly different to r for large deformations. Similarly, the current coil consists of (R Rc )/h wraps while
the completed coil consists of (Ro Rc )/h wraps.
Fig. 1
1523
at r = R
(6)
Here
w (R) = (R, R)
(7)
As described previously [2224] constitutive equations of the following form are considered
r = g ( r , ),
= f ( r , )
(1)
u
,
r
r + u (r)
(r)
(2)
= (1 + r )(1 + z )
= F (r , )
(3)
(4)
(5)
(8)
1524
at r = Rc
(10)
The stresses r (Ro , r) and (Ro , r) can now be calculated by solving the differential equation (9) subject to
the boundary condition (10). Since (r) is now known,
this is just a simple two-point boundary value problem for a pair of non-linear differential equations, for
which solution techniques are well known [28].
r + u (n+1) (Ro , r)
1 + f ( r(n+1) (Ro , r), (Ro , r))
(Ro , )d = 0
Rc
(n+1) (r) =
(11)
and
r (r)
r(n+1) Ro , Rc
E c
Rc
r (Ro , r) = 0 at r = Ro
(12)
r (Ro , r) = 0 at r = Ro
These can be viewed as a standard two-point boundary
value problem for the unknowns (r) and r (Ro , r) [28].
Alternatively, an iterative scheme such as the following
could be used
(0) (r) = r
For n = 0, 1, 2, . . .
Ro
d
(n+1)
r
(Ro , r) =
(R, ) (n)
(r)
r
r + u (n+1) (Ro , r)
1 + f ( r(n+1) (Ro , r), (Ro , r))
The case when a target radial stress r (Ro , r) is specied is technically a little different. Although equation
(11) yields a differential equation in the unknowns (r)
and (Ro , r), the boundary conditions (12) only give
a single constraint on the unknowns. Furthermore,
the interwrap pressure is zero on the outer boundary
and thus the differential equations become singular.
An additional boundary condition is not required if
E c > 0. Imposing the condition that the solution is
bounded is sufcient. Reference [27] provides further
information on singular boundary value problems and
their numerical solution. On the other hand, if E c = 0,
the winding stress at the bore needs to be specied. Further discussion of this case is deferred until
section 3, where a simple approximation is given.
Having obtained (r) and the residual stresses
r (Ro , r) and (Ro , r), the calculation of the winding
stress is now considered. From equations (4), (6), and
(r) = r
=
((r)f ( r (Ro , r), (Ro , r)))
r
g ( r (Ro , r), (Ro , r))
(13)
and boundary conditions
r (R, r) = Ec (f ( r (R, r), (R, r)) f (0, w (r)))
at r = Rc
(14)
(r) =
at r = R
(17)
Equations (16) and (17) form a system of differential equations subject to boundary conditions when
(Ro , r) is given or a system of algebraic differential
equations when r (Ro , r) is given. In either case the
solution can be found by standard methods.
For the solution of (16) and (17), one requires the
winding stress at the bore w (Rc ), which can be found
for E c > 0 by solving
E c f (0, w (Rc )) = E c f ( r (Ro , Rc ), (Ro , Rc )) r (Ro , Rc )
(18)
When E c = 0, an arbitrary winding stress may be
specied at the bore.
Further simplication is possible if additional
assumptions are made about the constitutive relation (1). Specically, almost all cases considered in
the literature assume that r = g ( r , ) = g ( r ), from
which it follows that the rst and second equations in
(16) uncouple when r (Ro , r) is specied. It should be
noted, however, that
( r , )
r
( r , )
f ( r , ) =
g ( r , ) =
(19)
(15)
at r = Rc
r (R, r) = 0
r (R, r) = 0 at r = R
1525
(20)
= f ( ) =
E
(21)
Proc. IMechE Vol. 221 Part C: J. Mechanical Engineering Science
1526
Table 1
Rc
Ro
Ec
E
25 mm
100 mm
6 GPa
4 GPa
w (Rc )
40 kPa
25
20 MPa
to estimate the interwrap pressure, a substantial deviation as shown in Fig. 5 would be obtained, where the
radial pressures calculated from large and small deformation analyses are compared for the hoop/tangential
stress prole shown in Fig. 3(b). On the other hand,
when the interwrap pressure is specied, the hoop
stresses calculated on the basis of large deformation
or small deformations are quite similar.
Figure 4 also shows the deviation of (r) from its
nominal value for material properties typical of steel.
(For this calculation the constitutive relations for steel
described in reference [23] were used.) As can be seen
the deviation for steel is very small, indicating that the
small deformation theory is largely valid for materials such as steel. Furthermore, corresponding results
for aluminium and polymer lm are also included
in Fig. 4. In these calculations, an E of 72 GPa and
a constant winding stress of 50 MPa were used for
aluminium, while material properties and winding
conditions similar to those assumed by Hakiel [15]
were adopted for polymer lm. It can be seen that the
deviations of (r) from their nominal values are very
small for both aluminium and polymer lm, indicating
small deformation analysis is suitable for these materials. The somewhat different behaviour of the polymer
lm from paper material is due to its higher value of Er
(around one to two orders of magnitude higher than
that of paper), despite similar values of E for the two
materials.
Having analysed the target residual stresses, one
can now nd the required winding stress that will
result in the target residual stresses. As explained
previously, this is achieved by the numerical solution of the differential equations (13) subject to the
boundary conditions (14). The required winding stress
Fig. 4
Table 2
1527
Comparison of nominal and actual radius after removal from the mandrel for a variety of
materials. The large deformation winding stress of Fig. 6 and parameters listed in Table 1
were used for the comparison for material properties typical of paper. For material properties typical of steel a constant winding stress of 100 MPa and the parameters listed in Table 2
were used. For aluminium, a constant winding stress of 50 MPa and an E of 72 GPa were
used, while for polymer lm, those assumed by Hakiel [15] were adopted
Rc
Ro
h
Ec
E
250 mm
750 mm
0.3 mm
100 GPa
200 GPa
l0
Eg
7.485 m
10 MPa
Fig. 6
Comparison between the winding stress calculated from the inverse solution for large deformations and those for small deformations for
material properties typical of paper. The target
tangential stress used for the calculations and
corresponding radial pressure are shown in Fig. 3.
All other parameters are given in Table 1
Fig. 5
Comparison between the radial pressure calculated for large and small deformations for material properties typical of paper. The same target
tangential stress, shown in Fig. 3(b), was used
for both solutions. All other parameters are given
in Table 1
1528
a smaller area than that with respect to the undeformed system, i.e. the radial pressure will be higher
for the deformed coordinate system, compared with
the undeformed coordinate system. This explains the
higher radial pressure calculated for Bensons model
compared to the model described in the current paper.
It is straightforward, however, to transform either
result into the other form by taking into account the
strain/deformation eld in the coil after winding.
In addition, in the model of this paper (as well as
in Hakiels model [15]), it is assumed that the tension
of the outermost wrap is equal to the applied tension,
whereas Bensons model allows for a slight reduction
in the tension of the outermost wrap [9]. This difference partially offsets the effect of the difference in the
coordinate systems.
Now that models for all four regimes have been
developed (small deformation theory for both linear
and non-linear material properties, as well as large
deformation for linear and non-linear material properties), the applicability of these regimes for different
types of materials shall be investigated. Those studied include metals, such as steel and aluminium, as
well as polymer lms and paper. Typical properties of
these materials can be found in references [9], [15],
[17], [23], and [30]. It is found that for all materials
considered, non-linear material properties need to be
employed for accurate stress predictions. To illustrate
this, a comparison of the stresses calculated from the
small deformation theory using non-linear and linear material properties for both paper and steel is
shown in Figs 8 and 9, respectively. In order to make
these comparisons, the equivalent material parameters for the linear case have to be deduced from
those of the non-linear material parameters listed in
Tables 1 (for paper material) and 2 (for steel). The
adopted method for the material properties of paper
is described in reference [24] and a similar method is
used for the material properties of steel. The material properties for paper listed in Table 1 and the
calculated winding stress (from small deformation
theory) of Fig. 6 are used for the comparison shown
in Fig. 8. As can been seen, there is a very large difference in the radial pressure calculated using linear
material properties as well as a signicant difference
in the tangential stress. A constant winding stress of
30 MPa and the material properties listed in Table 2
have been used for the comparison for steel. As can
be seen from Fig. 9(a), the radial pressure calculated
using linear material properties differs from that for
non-linear material properties even for low winding
stresses.
As discussed in reference [24], differences between
the large deformation and small deformation theory
are likely for cases for which (r) differs signicantly
from r. Hence for very compressible materials
(i.e. materials with very small values of Er ), such as
Fig. 8
Fig. 9
1529
CONCLUSION
1530
ACKNOWLEDGEMENT
The authors wish to thank the management of CSIRO
and BlueScope Steel Research for permission to publish the material contained in the current paper.
REFERENCES
1 Yuen, W. Y. D. and Cozijnsen, M. Optimum tension proles to prevent coil collapses. SEAISI Q., 2000, 29(3),
5059.
2 Altmann, H. C. Formulas for computing the stresses in
center-wound rolls. Tappi, 1968, 4, 176179.
3 Yagoda, H. P. Centrifugally induced stresses within
centre-wound rolls part I. Mech. Res. Commun., 1980,
7, 181193.
4 Yagoda, H. P. Centrifugally induced stresses within
centre-wound rolls part II. Mech. Res. Commun., 1980,
7, 233240.
5 Olsen, J. E. On the effect of centrifugal force on winding.
Tappi, 1995, 78(7), 191195.
6 Bouquerel, F. and Bourgin, P. Irreversible reduction of
tension due to aerodynamical effects. In Proceedings
of the 2nd International Conference on Web Handling,
StillWater, Oklahoma, June 1993, pp. 265285.
7 Jones, D. P. Air entrainment as a mechanism for low traction on rollers and poor stacking of polyester lm reels,
and its reduction. In Proceedings of Web Handling of the
ASME Winter Annual Meeting, Anaheim, CA, USA, 813
November 1992, pp. 123131.
8 Keshavan, M. B. and Wickert, J. A. Air entrainment during steady-state web winding. ASME, J. Appl. Mech., 1997,
64, 916922.
9 Benson, R. C. A nonlinear wound roll model allowing for
large deformation. Trans. ASME, E., J. Appl. Mech., 1995,
62, 853859.
10 Zabaras, N. A. Theory for small deformation analysis of
growing bodies with an application to the winding of
magnetic tape. Acta Mech., 1995, 111, 95110.
11 Qualls, W. R. and Good, J. K. Orthotropic viscoelastic
winding model including a nonlinear radial stiffness.
ASME, J. Appl. Mech., 1997, 64, 201208.
12 Tramposch, H. Relaxation of internal forces in a wound
reel of magnetic tape. ASME, J. Appl. Mech., 1965, 32,
865873.
13 Tramposch, H. Anisotropic relaxation of internal forces
in a wound reel of magnetic tape. ASME, J. Appl. Mech.,
1967, 34, 888894.
14 Lin, J. Y. and Westmann, R. A. Visco-elastic winding
mechanics. J. Appl. Mech., 1989, 56, 821827.
15 Hakiel, Z. Non linear model for wound roll stresses.
Tappi, 1987, 70(5), 113117.
16 Willett, M. S. and Poesch, W. L. Determining the stress
distributions in wound reels of magnetic tape using a
nonlinear nite difference approach. Trans. ASME, E.,
J. Appl. Mech., 1988, 55, 365371.
17 Zabaras, N., Liu, S., Koppuzha, J., and Donaldson, E.
A hypoelastic model for computing the stresses in centerwound rolls of magnetic tape. Trans. ASME, E., J. Appl.
Mech., 1994, 61, 290295.
APPENDIX
Notation
effective modulus of core
Ec
(former and mandrel)
modulus of gap
Eg
azimuthal modulus of coiled
E
material
f ( r , )
F (r , )
g ( r , )
G(r , )
h0
p
p
r
r +u
R
(r Rc )/h0
(R Rc )/h0
Rc
1531
Sm
u = u(R, r)
interlayer compression
modulus
interlayer springiness
radial strain
azimuthal strain
stress strain relation
stress strain relation
axial strain
radius of the (r Rc )/h0 wrap
if it were unconstrained
winding stress
target stress
target stress
r = u/r
= (r + u )/
r = g ( r , )
= f ( r , )
z
w (R)
r