You are on page 1of 6

A Study of Tube Deformation under Bend and Tension

Ji Lei Levandovskiy Andrey


Corning Optical Fiber and Cable (China) Corning Modeling & Simulation (Russia)
BeiJing, P.R.C St. Petersburg, Russia
+86-132-6968-7731 +7-812-329-20-98
jil3@corning.com levandovan@corning.com

Abstract 2. Problem Description


Tube deformation due to bend and tension in the cabling process A simplified configuration of the problem is used to build finite
directly affects the free space inside buffer tube, and subsequently element model. As illustrated in Figure 2, the model has a section of
impacts on the tensile and temperature performance of the cable. single-layer tube formed from a uniform plastic material with given
Finite element method is used to evaluate the impact of bending cross-sectional dimensions wrapped around a mandrel with given
with certain amount of tension on the tube deformation. Linear least radius and a certain amount of tensile force is applied at one end of
square polynomial fitting is then applied to establish empirical the tube while the other end is fixed. The tube would deform from
equations from the finite element analysis data for the purpose of the initial round shape. Omitting the friction of the tubes on mandrel,
predicting the minimum tensile force to make a tight strand and the the stable deformation of the tube cross section is supposed to fall at
tube deformation. the location of the symmetrical axis, which is indicated by a red
dash line.
Keywords: optical fiber cable; tube deformation; bend; tension;
finite element; linear least square; multivariate polynomial fitting. RBend
OD
OD’x
1. Introduction ID
ID’x
In optical fiber cable design theory, the amount of free space inside
the buffer tube is, among other factors, one of the most important
parameters, closely related to the tensile and temperature

OD’y
ID’y
performance of the cable. For stranded loose tube cable, under
compression, the fiber bundle moves away from the center of cable
to consume the excess fiber length generated by the reduction of
cable length. When the fiber bundle approaches its critical position,
namely it touches the inner wall of buffer tube, it starts buckling. F
Under tension state, the fiber bundle moves towards the center of
a) Simplified b) Initial c) Deformed cross
cable to compensate the increase of cable length. Once the fiber model of tube cross section of tube
bundle touches the inner wall of buffer tube, fiber strain will occur. bending section
of tube
Besides crush, impact, the tension of binder yarn and any other side
pressure against tubes, when bending a tube to a certain radius Figure 2. Simplified model of tube bending
under certain tensile force, e.g. a tube stranded on central strength A typical non-linear plasticity model is used for the tube material.
member, will lead to tube deformation and its transverse dimensions The corresponding stress-strain dependency, in terms of von Mises
will change accordingly. This deformation can effectively suppress norm of the tensors, is derived by multiply a scale factor to the
the cross-section area of the tube and limit the movement of fiber original curve of a material with initial modulus E = 0.798GPa and
bundle away from cable center. For stranded loose tube cable, a yielding at a strain of ~0.03. The original curve, as well as those of
small amount of deformation could impact the performances of E = 1GPa and E = 2.1GPa, are given in Figure 3.
cable significantly, especially for the low temperature window of
the cable, as illustrated in Figure 1. However, whether the effect is
significant or not has not been demonstrated so far. 0.10 E = 0.798GPa (Original)
E = 1.0GPa
E = 2.1GPa
0.08
Stress (GPa)
ΔODy

ΔODy

0.06

0.04

0.02

a) Compression State b) Tension State 0.00

Figure 1. The impact of tube deformation on cable 0.0 0.2 0.4 0.6 0.8
performances
Strain
Figure 3. The stress-strain dependency of a typical non- distributions of strain is not uniform on any cross sections, but on
linear plasticity model all planes perpendicular to tube axial direction they are assumed
Following values of variables and material properties are used for to be the same. This is supposed to be true for a relatively long
finite element analysis. 1.0GPa and 2.1GPa are used as the initial piece of tube with multiple turns of wraps.
modulus of tube material. Poisson’s ratio is 0.4. Tube outer Figure 4 gives an example of deformed shape for the 5N tensile
diameter, OD, ranges from 1.4mm to 3.5mm. Tube wall thickness, load for the weakest configuration, where E = 1GPa, OD = 3.5mm,
t, ranges from 0.1mm to 0.4mm. Bend radius, RBend, ranges from t = 0.1mm, and RBend = 60mm.
60mm to 100mm. The tensile forces, F, are up to 5N for E = 1GPa
and up to 10N for E = 2.1GPa.
3. Finite Element Analysis and Polynomial
Fit
3.1 Finite Element Model
Following models are developed for the analysis:
1) 2D beam model
A 2D beam model is used to calculate the minimum tensile force
Fmin needed to make a tight strand for given settings. Fmin is
treated as a function of material’s elastic modulus, area moment Figure 4. An example of 2D Generalized plain strain
of inertia and bend radius. model with symmetry expansion view
Analytical methods were demonstrated for this purpose in 3) 3D full model
previous papers [1-3]. According to [1-3] the following formula A 3D full model is used to check the results of the 2D simulations
should be used to calculate Fmin: and to track potential side effects which are observed in 2D
simulations. See later section for more detailed discussion.
1 EI
Fmin  2
, 3.2 Side effect
4 Rbend Quite an interesting effect was discovered when working with the
Where E is material’s Young’s modulus, I is the cross-sectional 2D beam model. When a moderate tensile force, somewhat larger
moment of inertia, RBend is the final bend radius. than the minimum, is applied to the tube, there is a constant
contact linear pressure along the whole tube section that is in
Fmin values calculated using the formula are approximately 1.7-1.9 contact with the central element cylinder. However, in the region
times lower than those obtained through the 2D beam model. The where the contact is lost, namely the point where tubes start
difference could not be attributed to the way boundary conditions contacting the central element, the contact force has its maximum.
are set in the numerical model because the boundary conditions The value of this maximum is significantly (2-5 times) larger than
were varied and for no configuration the resulting Fmin change the value of the constant.
was larger than 7-8%. Nor could the difference be yielded by the
nonlinear material properties used in the beam model: the switch In order to estimate the influence of this extremum on the final
from plastic to pure elastic material does not change the calculated shape of the cross-section, the full (quarter-symmetry) 3D model
Fmin for more than 2% (plasticity influences most when we has been created. The weakest configuration, namely the case
determine cross-section deformation under very large forces in the with the largest diameter, thinnest wall and smallest modulus and
next section). The only parameter that influences the Fmin in the smallest bend radius (OD = 3.5mm, t = 0.1mm, E = 1GPa, RBend =
beam model is the total bend angle, which is the angle between 60mm), was launched. The result is given in Figure 5. Side effects
tensile force and the tangent line at the point where the initial can be seen, but the tube deformation in the regions of potential
point of the cable is attached to the drum. Analytical models only side effects is not larger than the tight strand section. This
consider 180°whereas our numerical model has been tested for indicates that there are no issues in the contact region close to the
various angles and indicate convergence at 270°degrees or more free end of the tube.
(winds of up to 630° were tested). So 180° is not a converged
solution according to the beam model. For 180° angle case
analytical and numerical results for Fmin are the same. This means
that the Fmin obtained by the formula will only yield some initial
small contact spot at 180° back from the direction of force
application whereas the cable behind this 180°mark will be loose.
This is not acceptable in practice where a long, tight wind is
normally required. Thus the numerical solutions are to be used.
The difference between the analytical and numerical models is
due to no contact to the drum (reel) surface being considered
analytically.
Figure 5. Side effects in 3D full Model
2) 2D generalized plain strain model
A 2D generalized plain strain model is used to calculate the tube 3.3 Polynomial fitting
deformation (cross-section dimensions change) during tension. In practice, it is useful to derive some forms of empirical formulae
The tube is wound around the “cylinder” of the set radius first, from the finite element analysis data instead of conduct time-
and then the tensile force is applied. The final deformation would consuming FEA for every configuration. Linear least square
be a summation of the deformation due to bend and tension. The
polynomial fitting is used for this purpose, with a form given tension slightly higher than the minimum tensile assures the
below. tightness of cable core, i.e. the tubes are tightly stranded
around central element member. Setting the maximum lower
 
f  x,       

1 2 ... m  x  
i
i than a critical level (for example, ΔODy ≤ 0.05mm) which,
along with other sources of pressure, does not generate
i n im
significant tube deformation assures the cable performs as
where expected.
xi is the ith independent variable.
αi is the the integer exponent of the ith variable of a term. 0.6
Symbol Size Symbol Type & Color
  ... is the coefficient of the corresponding component.
1 2 m 0.5
t = 0.1
t = 0.15
RBend = 60
RBend = 80
t = 0.3
RBend = 100
n is the largest degree of a single independent variable. Note t = 0.35
0.4

Fmin/E (N/GPa)
t = 0.4
that this does not equal to the degree of a polynomial by
common definition.
0.3
m is the number of independent variables.
0.2
By this definition, a nth-order polynomial with m independent
variables would have mn+1 terms. Since the responses appear to be
0.1
nonlinear in many senses as shown in following sections, to avoid
Runge's phenomenon, 2nd-order polynomial (n = 2) is used for
0.0
curve fitting.
3.4 The minimum tensile force to make a tight 1.0 1.5 2.0 2.5 3.0 3.5 4.0

strand, Fmin Tube OD (mm)


Fmin is defined as the minimum tensile force to make a tight strand a) Fmin vs. tube OD
of given settings. In cable manufacturing practice, it is closely
related to the minimum tube tension in stranding process. 0.6
Symbol Size Symbol Type & Color
OD = 1.4 RBend = 60
The FEM data show that Fmin is proportional to initial modulus of 0.5 OD = 1.5 RBend = 80
tube material. This is intuitive that higher tension is needed for OD = 1.8
RBend = 100
OD = 2.0
stronger tube materials to make the same bend. OD = 3.5
Fmin/E (N/GPa)

0.4
The minimum tensile force is studied as a function of tube OD,
tube wall thickness, bend radius, and the modulus of tube material. 0.3
For simplification, the minimum tensile force could be normalized
by the modulus of tube material. Therefore Fmin/ E is a function of 0.2
three variables, i.e. tube OD, tube wall thickness, and bend radius.
The relations of normalized minimum tensile force with respect to
0.1
these three variables are given in Figure 6 respectively.
Fmin increases with bigger tube OD, thicker tube wall, and smaller 0.0
bend radius, but none of these relations are linear. 0.10 0.15 0.20 0.25 0.30 0.35 0.40

 Tube OD: Solely changing tube OD is equivalent to change Tube Wall Thickness (mm)
the tube OD and ID simultaneously, therefore the cross area b) Fmin vs. tube wall thickness
of tube wall changes linearly. However, the response is
0.6
nonlinear. Symbol Size Symbol Type & Color
 Tube wall thickness: Solely changing tube wall thickness is t = 0.1 OD = 1.4
0.5 t = 0.15 OD = 1.5
equivalent to change the tube ID only. Increasing the wall t = 0.3 OD = 1.8
thickness while maintain constant tube OD, the tension t = 0.35 OD = 2.0
t = 0.4 OD = 3.5
Fmin/E (N/GPa)

asymptotically approaches the limit where a tube becomes a 0.4


solid rod.
 Bend radius: When reducing the bend radius, Fmin increases 0.3
nonlinearly until the tube kinks before reaching the
theoretical extremum, namely zero bend radius. On the other 0.2
hand, increasing the bend radius makes the tension finally
approach 0N, because the infinite bend radius is actually a 0.1
straight tube.
 The minimum tensile force is generally low for standard 0.0
OSP cable, e.g. 0.1N and 0.2N is equivalent to 10g and 20g 60 70 80 90 100
respectively. An equivalent tension of a few tens of grams Bend Radiu (mm)
is usually enough to make a tight strand. Considering tube
c) Fmin vs. bend radius
tension fluctuates during stranding, especially for SZ
stranding, and the variations of tube dimension and material Figure 6. Normalized minimum tensile force with respect
property deviate the responses, setting the minimum tube to tube OD, tube wall thickness and bend radius
A 2-order polynomial fit is conducted on the normalized 0.00
minimum tensile force, Fmin/E. Because the normalized minimum
tensile force, Fmin/ E is a function of three independent variables, -0.02
there are totally 33 = 27 coefficients. The isosurfaces of Fmin/ E
for a range of parameter settings are plotted in Figure 7. The
minimum value is at the upper-left corner and has a value of 0.004 -0.04

ΔODy (mm)
N/GPa. The maximum value is at the lower-right corner and has a OD = 1.4; t = 0.1; RBend = 100
value of 0.344 N/GPa. The curving isosurfaces again suggest the -0.06 OD = 1.4; t = 0.1; RBend = 80
nonlinearity of the effect. OD = 1.4; t = 0.1; RBend = 60

-0.08
OD = 1.5; t = 0.1; RBend = 60

-0.10
0 1 2 3 4 5
F/E (N/GPa)

b) ΔODy vs. F/E (enlarged small deformation)

Color Symbol Type Line Type


E = 1GPa t = 0.1 RBend = 60
1.0 E = 2.1GPa t = 0.15 RBend = 80
Symbol Size t = 0.3
RBend = 100
OD = 1.4 t = 0.35
OD = 1.5 t = 0.4
0.8 OD = 1.8
OD = 2.0
OD = 3.0

0.6

ΔODx (mm)
Figure 7. Isosurfaces of normalized minimum tensile
0.4
force Fmin/E
3.5 The change of tube dimensions under bend 0.2
and tension
The responses of main interest are the tube dimension changes in 0.0
x, y directions, i.e. ΔODy and ΔODx. The FEM results are plotted
0 1 2 3 4 5
in Figure 8, where same normalization theme is applied on the
F/E (N/GPa)
tension (divided by E), i.e. F/E. Similar to Fmin, the responses are
approximately proportional to the initial modulus, E. It is
suggested by the curves of different E fall on same lines. This c) ΔODx vs. F/E
implies that the resistance of tubes against deformation is
proportional to the modulus of tube material. 0.05 Color Symbol Type Line Type
E = 1GPa t = 0.1 RBend = 60
E = 2.1GPa t = 0.15 RBend = 80
ΔODy and ΔODx, are studied as functions of tube OD, tube wall 0.04
Symbol Size
OD = 1.4
t = 0.3
t = 0.35
RBend = 100
t = 0.4
thickness, the modulus of tube material, bend radius, and tensile OD = 1.5
OD = 1.8
OD = 2.0 OD = 1.5; t = 0.1; RBend = 60
force. When the tensile force is normalized by the modulus of OD = 3.0

tube material, those responses are reduced to functions of four 0.03


ΔODx (mm)

independent variables, i.e. tube OD, tube wall thickness, bend


OD = 1.4; t = 0.1; RBend = 60
radius, normalized tensile force (F/E). 0.02

OD = 1.4; t = 0.1; RBend = 80


0.01
0.0
OD = 1.4; t = 0.1; RBend = 100
-0.2
0.00
-0.4 0 1 2 3 4 5
F/E (N/GPa)
-0.6
ΔODy (mm)

-0.8 d) ΔODx vs. F/E (enlarged small deformation)


Color
-1.0 E = 1GPa
E = 2.1GPa Figure 8. Change of dimensions with respect to
-1.2 Symbol Size Symbol Type
normalized tensile force
OD = 1.4 t = 0.1 Line Type
OD = 1.5 t = 0.15 RBend = 60
-1.4 OD = 1.8 t = 0.3
RBend = 80 Under bend and tension, tube dimensions decrease in y direction
OD = 2.0 t = 0.35
OD = 3.0 t = 0.4 RBend = 100
-1.6 and increase in x direction, and the amount of changes in both
0 1 2 3 4 5 directions are not equal. Smaller tube with thin tube wall bent to
F/E (N/GPa) small radius is the most stringent situation.

a) ΔODy vs. F/E The responses are not linear with any of the independent variables.
The sources of nonlinearity could be geometrical nonlinearities,
stress-stiffening, material non-linearity (plasticity), possible large
deformation effects due to a large tension force, and etc.

 Tube OD: Under same tensile level and of same tube wall
thickness, bigger tube OD generally leads to greater
deformation.
 Tube wall thickness: For a given tube OD, the increase of
wall thickness leads to smaller deformation. It suggests that
there might be a critical wall thickness for each tube size
which could reduce ΔODy to a negligible level.
 Bend radius: Similar to Fmin, when reducing the bend radius,
the deformation increases nonlinearly until the tube kinks
before reaching the theoretical extremum, namely zero bend
radius. On the other hand, increasing the bend radius makes
no deformation, because the infinite bend radius is actually a
straight tube.
 Tension: Higher tension causes greater deformation. Along a) Tube OD = 2.0mm
with other sources of pressure, high tension can generate
significant tube deformation which in turn adversely
impacts the cable performance. Also, high nonlinearity due
to non-linear plasticity can be observed on small tubes with
small wall thickness. Due to the small cross area, higher
stress would take place under same tensile level. When the
stress exceeds the linear region of material and starts
exhibiting plasticity.

The impact of ΔODy is considered more significant than ΔODx.


As illustrated in Figure 2, the low temperature window is
compressed directly by ΔODy. The expansion in x direction could
mitigate the effect indirectly by slightly reducing the effective
bundle size, but the effect is subtle. The impact of the decrease of
ΔODy cannot be neutralized the increase of ΔODx. Therefore
ΔODy dominates the overall impact.
b) Tube OD = 3.0mm
A 2-order polynomial fit is conducted on the tube deformation in
y direction, i.e. ΔODy. Because it’s a function of four independent Figure 9. Reduction of the tube size in y direction for
variables, there are 34 = 81 coefficients. For the purpose of getting various tube OD
a better fitting, data of several weakest cases, 3.5mm tubes with
0.1mm and 0.15mm tube wall, are excluded. Fmin/E (the minimum tensile force to make a tight strand
normalized by the modulus) is studied as a function of three
The isosurface of ΔODy for tube sizes of 2.0mm and 3.0mm are variables, i.e. tube OD, tube wall thickness, and bend radius. Fmin
given in Figure 9. If ΔODy ≤ 0.05mm is taken as an insignificant can be useful to find proper payoff tension setting in stranding
level, for 2.0mm tube, when tube wall thickness is larger than process. ΔODy dominates the overall impact. ΔODy, ΔODx (the
0.2mm, the tube deformation would be considered insignificant. tube dimension change in x, y directions) and ΔS (the reduction of
For 3.0mm tube, relatively thicker tube wall, approximately cross sectional area) are functions of four independent variables,
0.3mm, is needed. i.e. tube OD, tube wall thickness, bend radius, normalized tensile
force (F/E).
For small deformation situation, it is reasonable to assume the
shape of deformed tubes is elliptical and the wall thickness is This effect is not significant for typical tube dimensions of
assumed to be constant under deformation. By these assumptions, standard loose tube cables at normal process settings. Besides,
the change of cross area before and after deformation can be tubes in the cable could support each other to mitigate the impact
calculated. All of them have negative values. The bigger the tube of deformation. But caution should be exercised and process
OD is, the more amount is suppressed. This means that bending settings should be changed accordingly when parameters are
the tube actually suppresses the cross area, namely the free space. significantly changed, for example, tube wall thickness goes
smaller or softer material is used. The consideration of this effect
4. Conclusions could be a supplement of conventional cable design theory and a
potential improvement of the design quality and accuracy of
This paper evaluates the impact of tube deformation under bend conventional model.
and tension with finite element methods. A customized 2nd-order
polynomial fitting is proposed to establish empirical equations
from FEM data. Responses appear to be nonlinear in many senses.
5. References 6. Pictures of Authors
[1] R. C. Benson, “Stiff Elastic Tape Wrapped Onto a
Drum”, Journal of Applied Mechanics, Vol. 65, 870- Ji Lei is a Cable Design Subject Matter
874 (December, 1998). Expert in the technology group at Corning
Optical Fiber and Cable, BeiJing, China.
[2] C.-Y. Wang, “Winding a Long Elastic Sheet”, Acta He received his PhD Degree in Physical
Mech~nica 39, 297--301 (1981) Electronics from Tianjin University in
[3] Jianping Guo, “Three-Dimensional Wrapping Of Stiff 2006 and joined Corning in the same year.
Elastic Belts Around Rollers”, Dissertation in Mechanical
Engineering, The Pennsylvania State University, The Graduate
School, Department of Mechanical and Nuclear Engineering
Levandovskiy Andrey is a senior
researcher in the Corning Scientific Center,
St-Petersburg, Russia. He received his
Master Degree in Applied Mechanics from
St-Petersburg Polytechnical University in
2006 and joined Corning the same year.

You might also like