Professional Documents
Culture Documents
Strain Basedformulasforstressesinprofiledcenter Woundrolls
Strain Basedformulasforstressesinprofiledcenter Woundrolls
net/publication/291822790
CITATIONS READS
21 316
3 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Stephen J. Burns on 23 August 2016.
C
stress–strain relation, i.e., the modu- differences between these calculated
convenient, compact stor- lus was proportional to the stress. stresses and previous solutions. The so-
age scheme for web prod- Boutaghou and Chase (2) noted that lution here rigorously satisfies two
ucts. However, the stresses Maxwell’s equations (from elastic en- boundary conditions and is in closed
in rolls can cause damage, especially ergy function arguments) reduce the form. The proposed approach contains
in tightly wound rolls that are stored number of independent terms in Alt- four stress terms: Two are general
for a long time. The objective in cre- mann’s elastic constants. They also stresses from the homogeneous part of
ating a roll is to select parameters in found some general properties of the solution, while two new solutions
the web, hub, and roll that profiles center-wound rolls with anisotropic come from the residual strains in the
the stresses for minimum damage to linear elastic media. Yagoda (3) es- roll generated by web tension stresses.
the medium while avoiding cinch- tablished the core compliance as an The radial dependence of the residual
ing, spoking, telescoping, or soft-hub inner boundary condition on center- strains (i.e., the roll profile) is very im-
roll buckling. wound rolls. The constitutive rela- portant, since the residual roll stress de-
The stresses in center-wound tions for stacked sheets were exam- pend on the winding conditions. All
rolls are essential to the integrity of ined by Pfeiffer (4) and Olive (9), winding profiles lead to complete, al-
the roll. Consequently, these stresses who suggested empirical forms for though algebraically complicated, elastic
have been the subject of theoretical stress vs. strain. A general nonlinear solutions in analytic form. Nonlinear
and experimental studies for at least power-series constitutive law was al- elastic solutions are also investigated
the last 30 years (1–8). They were ternatively used by Hakiel (5). The using the effective residual web stress
initially modeled by assuming a cylin- first term of Hakiel’s series is related and a stress-dependent radial elastic
drically symmetric geometry for the to one of Pfeiffer’s (4) material con- modulus. Numerical nonlinear equa-
center-wound roll that is mechani- stants. Hakiel incorporated nonlinear tions developed from expressions de-
cally loaded on the outermost wrap. material properties into his expres- rived here have straightforward solu-
This layer is considered a hoop sions of the basic mechanics and nu- tions. A selection of material
stress, at the web tension, on the merical solutions of wound-roll constitutive laws and roll profiles are
outer wrap as the roll is constructed stresses. evaluated for comparison. Finally, the
layer by layer. Early work by Altmann Good et al. (6) compared results modeled roll stresses are compared
(1) provided a general solution for from Hakiel’s model with interlayer with pull-tab-measured stresses from
an anisotropic linear elastic roll ma- pressure measurements obtained instrumented wound rolls. The radial
terial while using a nonlinear consti- using pull tabs. They noted that the plateau stresses in the midroll are fur-
tutive relation to find the radial and model typically predicted stresses ther approximated with several simple
hoop stresses for successive wraps. that were twice as large as their mea- closed-form algebraic expressions.
The nonlinear constitutive laws at- sured values. However, they were Application:
tempt to account for interlayer con- able to bring predicted and mea-
A new approach for modeling the
tacts, material discontinuities, and sured values into better agreement
elastic stresses in center-wound rolls.
anisotropic intralayer wraps in the by modifying the outer hoop-stress
roll. boundary condition to relax relative
Altmann (1) solved a second- to the outer-layer tensile stresses by
order differential equation for the their model of “wound on tension”
anisotropic linear elastic material in loss. Benson (8) later showed that a
a center-wound roll. He introduced large-strain, nonlinear constitutive the predicted interlayer pressures
to his solution an outer wrap law that accounted for hoop-stress and measurements without addi-
described by a parabolic radial relaxation gave agreement between tional assumptions about wound-on-
σW
∫
+ 1 r –β t βσ*(t)dt – r β t –β σ*(t)dt
R
]
(12)
(no strain) less the state of strain in stress, σrr.
the web before it is spun onto the The homogeneous solution con-
roll. Substituting Eq. 2 into Eq. 1 pro- [D 2
]
+ 2D + (1 – β 2 ) σrr = σ*(r) (8) tained in Eq. 12 incorporates two ar-
vides compatibility for the strains εij bitrary constants, A and B. The par-
in the two principal stress direc- where ticular solution to the differential
tions. 2 equation is found using Lagrange’s
s11s33 – s13
β2 = (9) method of variation of parameters
∂εθθ ∂(r ε )* 2
s22s33 – s23
r + εθθ – εrr = ε*rr – θθ
(3) (11). The lower limit in the integra-
∂r ∂r and tion was used for r, since this choice
Initially, the strains will be as- facilitates direct comparison with
sumed to be linear in stress. Nonlin- {
s33 1
E
d
dr [rσw(r)]+ Eν σw(r) } Altmann’s and subsequent work on
ear elastic stress solutions are ad- σ*(r) = 22
2
22
(10) stresses in center-wound rolls (1–9).
s22 s33 – s23
dressed later. The linear elastic The first boundary condition is
strains in the roll are related to the Note that for most wound flexi- that the outside of the roll is stress
stresses by Hooke’s law: ble media, β2 > 1.To find the residual free, or σrr = 0 when r = R. Thus Eq.
stress in the roll σ*(r), use the right 12 can be recast as Eq. 13.
εrr = s11σrr + s12σθθ + s13σzz (4) side of Eq. 3 with Eqs. 11a or 11b:
r ∫ r
R
∫
+ 1 r –β t βσ*(t)dt – r β t –β σ*(t)dt ]}
E22 (13)
The hoop stress follows from Eq. that B as given here depends on R. 12, 13). Here, s12 is unspecified, since
7 and is given by Eq. 14. This implies that the solutions in the it does not appear in Eqs. 8–10. Note
literature (1–8) should include ad- that the stresses in the roll are from
r{[ (
σθθ = 1 βB r β + R β
r
2β
)] justments to the inner boundary con-
dition as laps are wound onto the roll.
the differences in the principal resid-
ual strains in the roll. Thus, both
2 [ r ∫ r
R
∫
– 1 r –β t βσ*(t)dt + r β t –β σ*(t)dt
R
( ) ν
E22 + E22
2
s33 s22 – s23
(17)
0.80 0.60
Constant web stress
0.60 Profiled roll
0.40 Constant torque
0.40
Constant web stress 0.20
0.20 Profiled roll
Constant torque
0 0
1.5 2.0 2.5 3.0 3.5 4.0 1.5 2.0 2.5 3.0 3.5 4.0
RADIUS RATIO (r) RADIUS RATIO (r)
EFFECTIVE WEB STRESS (σ*/σ0)
1.50 0.50
b d
0.25 0.10
0
1.5 2.0 2.5 3.0 3.5 4.0 1.5 2.0 2.5 3.0 3.5 4.0
RADIUS RATIO (r) RADIUS RATIO (r)
2. Stresses—a: web stress, b: effective web stress, c: radial compressive stress, and d: hoop stress—vs. roll radius in profiled linear
elastic center-wound rolls (experimental details in Table I)
3. Stresses—a: radial compressive stress and b: hoop stress—vs. roll radius in a center-wound roll at constant web tension for linear
elastic media with outer-roll radii ratio ranging from 2 to 5 (experimental details in Table I)
layers of the roll. Thus, the radial This nonlinear modulus must be previously. The stresses developed in
modulus for center-wound-roll media used in compression. Such a modu- the roll are quite large because the
is widely represented by a noncon- lus, of course, would be unstable effective web stress σ*n–R is large. The
stant, nonlinear expression. The con- were the stresses to change sign. value of 1/(α s22) is 12.1 and 15.6
stitutive law is typically obtained by The solution to Eq. 20—after times the web stress for R = 2 and 3
curve-fitting the load per unit area eliminating one arbitrary constant rolls, respectively. The very large
vs. strain for a stack of sheets (4, (by satisfying the boundary condi- value of σ*n–R elevates all the stresses
12–14). The derivative of this stress tion σrr = 0 at r = R)—is shown in Eq. in the roll, with the stresses being
with respect to the strain is the radial 21. high compared with pull tabs (6).
modulus as a function of the strain. B1 r
The stress and modulus vs. strain are
parametric equations solved for mod-
σrr =
r
Rn
R () ESTIMATES OF PLATEAU STRESSES
IN THE ROLL
ulus vs. stress. Frequently, the modu-
lus is reported to be proportional to
the compressive stress, as assumed in
+1
r [∫r
R
Rn(t)σn–R
*
(t) dt – Rn(r) σn–R
*
(t)dt (21)
∫
R
r
σn–R
*
(t) dt
] (22) plateau of nearly constant stress over
most of the roll (5–8). If it is as-
dressed. If it is assumed in Eqs. 8–10 B1 is found using Eq. 15, which sumed that the radial stress is con-
that all Poisson ratios in the roll are was evaluated with the aid of Eqs. 2, stant throughout the roll, then Eq. 7
zero and that the modulus is propor- 5, 21, and 22. The expression for B1 is dictates that the hoop stress equals
tional to the pressure (i.e., the nega- given in Eq. 23. the radial stress. The web’s elastic
tive stress), then strain energy density for this case is
∫ ∫
R R then equal to the roll’s linear elastic
Er = – α σrr = s1 1
Rn(t)σn–R
*
[
(t)dt+Ecs22 σw(r=1)+ σn–R
*
(t)dt
1
] strain energy density. Equating these
11 B1= energies and solving for the radial
Rn(R) + Ecs22
so (23) stress in the roll gives Eq. 24.
σw
(D2 + 2D + 1)σrr = σn–R Thus the stresses σrr and σθθ are σrr, avg ≈
*
(r) (20)
uniquely obtained for any winding | | √2(1 + s11E22)
(24)
where the nonlinear characteristic profile. The stresses σrr and σθθ are
stress σ*n–R(r) is shown in Fig. 4 for the nonlinear The stresses seen in the pro-
web parameters listed in Table I posed model and in other models of
(r) = 1 + d [rσw(r)] + ν σw(r)
σn–R
*
αs22 dr E22s22 { } (Data Sets 4 and 5). The winding pro-
file is constant web tension, as noted
center-wound rolls are generally
larger at the hub and smaller on the
30 0.3
20 0.2
10 0.1
1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00 1.5 2.0 2.5 3.0
RADIUS RATIO (r) RADIUS RATIO (r)
4. Radial stresses vs. roll radius in a center-wound roll assuming 5. Comparison of calculated radial stresses and measured pull-
nonlinear elastic properties for the layered roll media (experi- tab data (16) that have been corrected for stress concentrations
mental details in Table I) (15) (experimental details in Table I)
outer laps. Equations 12–14 show representing the solutions of the Eq. 25a. In Fig. 3, where β ≈ 11, the
that these two boundary conditions inner and outer boundary condi- plateau stress is in agreement with
are represented by r β–1 and 1/r β+1, re- tions, respectively, as noted previ- Eq. 25b. Furthermore, a constant ra-
spectively. The large hub strain en- ously. If β is very large, then these dial plateau stress implies that the
ergy very roughly cancels with the two terms nearly become step func- hoop stress is equal to the radial
smaller strain energy on the outside tions. The constant term is the radial stress, as discussed previously and
of the roll in the approximation of plateau stress. This is seen in the mid- also seen in Fig. 3. Thus, a plateau
Eq. 24. Typically, β 2 >> 1, so the right section of the wound rolls in Fig. 3, stress implies that the hoop stress is
side _ of Eq. 24 is approximately which depicts compressive and in compression and is constant.
σw/√ 2β. Thus the stress in the roll is hoop stresses under constant web Good et al. (6) presented a figure
significantly reduced compared with tension. (Fig. 6 in their work) that predicts
the web stress in this rough but very Equating the constant stress term that plateau stresses from Hakiel’s
simple approximation. Pfeiffer (4) in Eq. 13, i.e., a stress that is indepen- model (5) are proportional to σw.
reported _ a similar expression aside dent of r, to a plateau stress repre- Equation 25 shows that –σrr is di-
from √2. sents the midstress shown in Fig. 3. rectly proportional to σ*(r) with a
Nonlinear elasticity in the roll Plateau stresses are widely measured slope of 1/( β2–1). Furthermore, –σrr
can be incorporated into Eq. 24 by in center-wound rolls (1, 5, 6). The is approximately proportional to
finding the area under the stress-vs.- lower limit of the two integrals on 1/s11, i.e., to Er , when β2 > 1, which is
strain or load-vs.-displacement curve the right in Eq. 13 are independent of typical for center-wound media.
for the media, i.e., the strain energy the roll radius ratio for constant web Thus, –σr r is proportional to Er ,
density in the layered roll media.This stress. The following algebraic ex- which itself may depend on σrr.
strain energy density would be pression is obtained after evaluating Equation 25 was suggested in the
found as a function of stress and these integrals. The roll’s plateau work of Good et al. (6) from numer-
equated to the web’s strain energy stress is given by Eq. 25a. ical solutions of Hakiel’s nonlinear
density. The approximation is that stress models (5) with a wide selec-
σrr, plateau = σ (r)2
*
the strain energy density in the roll is (25a) tion of Er and σw values for a specific
1–β
a constant. material. The slope from Eq. 25 vs.
Finally, the stresses in the analytic so 1/s11, i.e., Er , is σ*/E22 , which is
equations with constant web stress smaller than suggested by the work
σrr, plateau ≈ – σ (r)
*
always reduce to a very simple form (25b) of Good et al. (6). Thus, Eq. 25 is in
s11E22
in numerical expressions for the agreement with the form of the non-
stresses. It has only three terms: a The plateau stress can also be ob- linear model but not with the indi-
constant, a term proportional to tained directly from the differential vidual calculated values from the
1/r β +1, and a third term proportional Eq. 8 by assuming the stress to be work of Good et al. (6). The more
to r β–1. The interpretation here is constant, thus converting all deriva- important comparison is to mea-
quite simple, with the two r terms tives to zero and reducing Eq. 8 to sured plateau stresses from pull-tab
data (15, 16).This comparison is pre- 13, 14, and 16 and the elastic con- rivatives to obtain the strains. Addi-
sented in Fig. 5 and is discussed in stants found in Step 2. tional terms contributing to the
the following section. stress σ*(r) come from the shape of
The nonlinear elastic constitutive the roll's profile. If the roll is profiled
DISCUSSION AND CONCLUSIONS law is explicitly included in the with a constant-torque winding, then
Elastic stress solutions in center- stress solution given above. The data most of the radial derivatives of the
wound rolls have been presented are from compression loading of residual strains are removed. The roll
using the concept of relaxing the stacked layers of media. The consti- stresses presented here show that
residual strains placed in the roll. The tutive law from load-vs.-displace- constant-torque rolls have a unique
basic model is that the wraps laid on ment curves typically emphasizes characteristic: In linear elastic
the roll do not slip and that the resid- the high part of the curve rather media, the web stress does not di-
ual strains from the web create the than low stresses. This can lead to rectly contribute to the roll’s
roll’s stresses. The homogeneous so- very large differences in the consti- stresses, and the Poisson effect in the
lution for these stresses is already tutive law. Our experience has led to web (not in the roll) is the only con-
known, while the particular solu- differences as large as 300% in the tributor to the roll’s stress. In gen-
tions found here are based on math- major terms in the nonlinear consti- eral, the residual strains in the roll
ematical solutions to differential tutive law. However, unloading may are from the differences in the prin-
equations. The particular solutions better represent the elastic compli- cipal web strains, so Poisson’s effect
are weighted averages of the residual ances necessary to describe stresses in the web adds to the roll’s stresses.
stresses when they are spooled onto in the roll. Comparison between the It also follows that the roll’s stresses
the roll from the web. Part of this so- model and the data in Fig. 5 shows scale with the maximum residual
lution agrees with the physically in- that the value of β used in the model shear strains in the web. Shear
tuitive stress model of the outer does not bring the analytic stresses stresses are the major contributor to
wraps placed on the roll. The partic- up as sharply on the outside of the plastic deformation and damage,
ular solution, however, is an average roll as the experimental data suggest. such as viscoelastic behavior in the
of weighted stress terms that are not Moreover, the stresses measured near roll. Thus these effects are directly
totally intuitive but are needed for the hub are smaller than predicted related to the differences in the
the complete mathematical solution by the analytic curve, again indicat- web’s principal strains. Interlayer ad-
of the second-order differential equa- ing that the value of β used in the an- hesion responds to the stresses σrr ,
tion. These solutions contribute to alytic expression is too small. The which for linear elastic materials also
the inner boundary condition. plateau stress depicted by the model scales with the difference in web
A final comparison is now made is in modest agreement with the ex- strains.
between recent data (16) and the perimental data. The simplest nonlinear radial
proposed strain-based model. The re- The stress solutions presented modulus has a term that is propor-
cent data (16) were corrected for have been constructed to be similar tional to the radial stress. This stress-
stress concentrations associated with to the elastic solutions in the litera- dependent modulus gives rise to an
the sandwiched pull tabs (15). The ture. However, the solutions would equivalent web strain, as seen in the
linear model in Fig. 5 was applied be simpler if the integrals in the par- first term on the right side of Eq. 20.
using a nonlinear modification that ticular solution were chosen to go This “nonlinear web stress”term is the
was found with the following proce- from 1 ⇒ r rather than from r ⇒ R. inverse of the proportional constant
dure: This change would significantly sim- from the radial modulus multiplied by
plify the inner boundary condition the hoop modulus. Thus the inverse
1. Plateau stress is obtained from while only slightly complicating the proportional constant can be consid-
Eq. 25a using the complete elastic outer one. As long as numerical solu- ered as a strain. Typically, this strain is
properties, including all the non- tions are obtained, then either very large, and the resultant stress
linear parts of radial elastic modu- method works. Simple analytic solu- dominates the web stress. The stress
lus, 1/s11. tions using the hub core as the lower from the nonlinear term is added to
2. The radial elastic modulus, 1/s11, limit in the particular solutions the web stress, so the radial stress in
is then determined at the ob- might provide some additional phys- the roll is not directly proportional to
tained value for plateau stress. ical insights. the web stress. The stress solution in
3. Roll stress vs. roll-radius ratio is The expressions for σ*(r) found Eq. 21 has assumed a radial modulus
then determined at the obtained here rely on compatibility relations exactly proportional to the stress.
value for plateau stress using Eqs. between displacements and their de- This means that log(stress) is propor-
NOMENCLATURE
a = hub radius R = outer roll radius ratio α = nonlinear radial coefficient
A = arbitrary constant sij = elastic compliances β = see Eq. 9
B = arbitrary constant s11 = radial elastic compliance εij = elastic strains
C = ratio of elastic properties s22 = hoop elastic compliance ε*ij = residual strains
D = r(d/dr) s23 = Poisson’s elastic compliance ν = Poisson’s ratio of web
Ec = hub core stiffness, GPa = –νs22 σij = stresses
Er = radial modulus s33 = z-axis elastic compliance σw = web stress
M = hub torque = s22 σrr = radial stress
r = roll radius ratio, dimension- t = integration variable σθθ = hoop stress
less T = tension σ* = effective residual web stress
u = radial displacement
LITERATURE CITED 7. Zabaras, N., Liu, S., Koppuzha, J., et al., J. 12. Pfeiffer, J. D., Tappi 49(8): 342(1966).
1. Altmann, H. C., Tappi 51(4): 176(1968). Appl. Mech. 61: 290(1994). 13. Pfeiffer, J. D., Tappi 51(8): 77(1968).
2. Boutaghou, Z-E. and Chase, T. R., J. Appl. 8. Benson, R. C., J. Appl. Mech. 62: 14. Qualls,W. R. and Good, J. K., J.Appl. Mech.
Mech. 58: 836(1991). 853(1995). 64: 871(1997).
3. Yagoda, H. P., J. Appl. Mech. 47: 9. Olive, G.A.,“A study of wound rolls dur- 15. Koppuzha, J. K., Adv. Info. Storage Systems
847(1980). ing winding and unwinding,” MS thesis, 4: 73(1992).
Dept. of Mechanical Engineering, Univer-
4. Pfeiffer, J. D., Tappi 62(10): 83(1979). 16. Fisher, L. W., Case Studies in Mechanical
sity of Rochester, Rochester, NY, 1994.
5. Hakiel, Z., Tappi J. 70(5): 113(1987). Engineering Design: Proceedings of 1998
10. Chang,Y. B., Chambers, F.W., and Shelton, ASME International Congress and Exposi-
6. Good, J. K., Pfeiffer, J. D., and Giachetto, J. J., J.Tribology 118: 623(1996). tion, ASME, New York, Paper No. 98-
R. M., 1992 Web Handling Symposium Pro-
11. Boyce, W. E. and DiPrima, R. C., Elemen- WA/DE-1.
ceedings, AMD-Vol. 149, ASME Applied
tary Differential Equations and Boundary 17. LaFleche, J., unpublished data, Materials
Mechanics Division, New York, pp. 1–12.
Value Problems, John Wiley & Sons, New Science Laboratory, University of
York, 1965, pp. 119–121. Rochester, NY, 1995.