International Journal of Fatigue 59 (2014) 114–128

Contents lists available at ScienceDirect

International Journal of Fatigue

journal homepage: www.elsevier.com/locate/ijfatigue

Finite element analysis of the fatigue strength of copper power

conductors exposed to tension and bending loads
Fachri P. Nasution a,⇑, Svein Sævik a, Janne K.Ø. Gjøsteen b,1
a
Department of Marine Technology, NTNU, Otto Nielsens vei 10, 7491 Trondheim, Norway
b
Norwegian Marine Technology Research Institute, MARINTEK, Otto Nielsens vei 10, 7052 Trondheim, Norway

a r t i c l e i n f o a b s t r a c t

Article history: The paper presents FE analyses for predicting longitudinal stresses from tension–tension and tension–
Received 29 April 2013 bending fatigue tests of a 95 mm2 stranded copper power conductor. As fatigue test results indicated that
Received in revised form 27 August 2013 the fatigue performance was dominated by longitudinal stresses, the models were formulated by a com-
Accepted 18 September 2013
bination of elastic beam and elastic–plastic beam-contact elements that included the friction. Two con-
Available online 28 September 2013
tact conditions were investigated: the point (trellis) contact between adjacent layers and the inline
contact within each layer and between centre wire and inner layer. Due to the plastic deformations of
Keywords:
the wires obtained from the manufacturing procedure, a simplified description of the contact behaviour
Copper conductor
Point (trellis) contact
was adopted and calibrated by axial tension testing. The FE models were further validated by calibration
Inline contact testing and mesh sensitivity checks. The simulated stresses were applied to attempt bridging the gap
Plastic deformations between the SN data obtained from full cross-section tension–tension and tension–bending testing
Mesh sensitivity and SN data obtained from individual wires testing.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction
Floating vessels like ships and semisubmersibles are exten-
sively used for production of oil and gas from offshore fields. Also
offshore wind turbine concepts based on floating units have been
developed. Electrical power cables are used in conjunction with
floating units for provision of energy to installations on the sea
bed, power from land to the floater, or export of power from a wind
turbine to land. Power cables that are linked to a floating unit are
subjected to fatigue loading from the waves and due to the move-
ment of the vessel in the waves. Therefore, fatigue strength needs
to be verified for design.
A power cable consists in general of multiple conductors each
representing an assembly of individual wires usually made of cop-
per or aluminium. Fig. 1a shows a typical offshore power cable for
an alternating current (AC) three-phase system. One cable usually
has three conductors, one for each phase. Each conductor consists
of copper wires helically wound in layers around a centre wire.
During operation, a cable will be exposed to gravity, environ-
mental loading from the sea, and to forces due to movements of
the vessel. The gravity will induce a mean global tension (T) and
mean global torque moment (M T ). The forces due to movements
of the vessel induce a dynamic tensile load (DT) and torque
(DMT) (due to heave and surge motion) and dynamic curvatures
⇑ Corresponding author. Tel.: +47 735 955 64.
E-mail address: fachri.nasution@ntnu.no (F.P. Nasution).
1
Presently employed at Reinertsen AS, 7010 Trondheim, Norway.
(Db) (from pitch and roll motion) acting on the cables. The most
heavily loaded section is close to the attachment point (at section
A–A) to the vessel (Fig. 2). Mean and dynamic tension will be trans-
ferred to the wires as tension and shear forces while the dynamic
bending moment will induce local bending as well as axial friction
forces into each individual wire.
As shown in Fig. 1b, the mean axial force (F x ) in each wire is a
function of the mean global tension (T) and the mean global torque
moment (M T ). The dynamic axial force (DFx) in each wire is a func-
tion of these, the corresponding dynamic quantities DT and DMT,
the dynamic curvature (Db) and the coefficient of friction (l) be-
tween the contact surfaces. The dynamic curvature results in local
bending in each wire where DMx is the dynamic torque moment
about the helix tangential x-direction, DMy is the dynamic bending
moment about the helix bi-normal y-direction and DMz is the dy-
namic bending moment about principal normal vector of the helix
curve.
The wires in a stranded conductor are stranded helically in lay-
ers which leads to contact longitudinally both within and between
each layer, which is illustrated in Fig. 3b. Contact within a layer
and between centre and first layer is denoted inline contact. The
global axial force will result in both longitudinal and transverse
forces within the layer where the transverse forces will cause
diameter reduction in the inline (hoop) direction of each wire. If
cylindrical bodies with diameter D1 and D2 are pressed together
with a certain load per unit length, p, a small continuous contact
surface area will occur as shown in Fig. 4.

0142-1123/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijfatigue.2013.09.009
 F.P. Nasution et al. / International Journal of Fatigue 59 (2014) 114–128
Fig. 1. (a) Offshore power cable for three-phase AC system using helically stranded copper conductor inside. (b) Helically stranded copper conductor with lay angle a exposed
to dynamic and static loads.
Contact between wires in different layers is called point contact
(Figs. 3a and 5).
A comprehensive literature study has been conducted with re-
spect to models for describing fatigue and contact stresses in ca-
bled structures. Mindlin [1] described the contact surfaces of two
isotropic bodies subjected to normal and tangential loading with-
out friction. Dong and Steidel [2] studied the contact stress condi-
tions between layers of strands using a photoelastic technique.
Hobbs and Raoof [3] have reported fatigue test results for socketed
structural strands. They concluded that the fatigue behaviour was
governed by several failure mechanisms related to the contact con-
ditions close to the socket. Johnson [4] described the Hertzian con-
tact stresses between two solid bodies under a certain normal load.
A theory of contact was developed for predicting the shape of the
contact area and the growth in size with increasing load; the mag-
nitude and distribution of surface tractions, normal and possibly
tangential forces, transmitted across the interface. Raoof and
Hobbs [5] proposed an analytical model for relevant multi-layered
stranded structures for determining the point and inline contact
forces and associated relative displacements with its ends fixed
against rotation. Raoof [6] developed from first principles a theo-
retical model using axial single wire data for predicting the axial
fatigue of the full cross-section at constant load amplitude, and
was able to correlate the theoretical predictions to observations
from experimental testing. Raoof [7] concluded that his theoretical
model provides useful upper bounds to the fatigue life of cables
failing at the end termination and that the termination type signif-
icantly affects the observed fatigue life. Raoof and Huang [8] pro-
posed simple methods for estimating strand plane section
bending stiffness when imposed to cyclic bending and external
hydrostatic pressures.
As opposed to the aforementioned authors, the present study
focus on copper conductors applied in dynamic cables where the
design is based on using protective steel armours and bending
stiffeners to resist the external forces, thus limiting the tensile
forces and curvatures in the copper conductor.
115
Zhou et al. [9] studied the plastic flow, local wear and fretting
failure for overhead electrical conductors imposed by bending fati-
gue loads. Papailiou [10] presented a new model of conductors un-
der simultaneous tensile and bending loads. During the bending
loads, the model takes into account the interlayer friction and slip
in the conductor. A mesoscale model has been developed by Hong
et al. [11] in order to describe the bending behaviour of helically
wrapped cable under tension. The model accounts for the nonlin-
ear behaviour of the cable due to friction forces. Karlsen [12] sim-
ulated the fatigue mechanism in dynamic power cables by
dynamic testing in tension and bending. He concluded that the ef-
fect of fretting on the fatigue properties was less dominant for cop-
per conductors than for steel wires and ropes. However, there is no
detailed information regarding the failure positions on the speci-
mens. Lévesque et al. [13] presented investigations related to fret-
ting fatigue of an overhead electrical conductor focusing on the
contact conditions at the trellis contact points.
The copper power conductor investigated here further pro-
duced by a compacting procedure where the following features
were noted:
1. Introduction of irregularities in wire geometries, specially for
the outer layer due to irregular supporting point contact
conditions.
2. An increase in the contact area, thus reducing the contact
stresses.
The effect of irregularities on the fatigue performance of indi-
vidual wires were investigated by Nasution et al. [14–16] where
a FE model was presented that allowed transforming the nominal
SN data into SN data of actual longitudinal stresses that included
the effect of irregularities. Full cross-section SN data obtained from
fatigue tension–bending tests results were also presented using a
simple analytical stress model. Significant deviations were noted
between the full cross-section SN data and the individual wire data
which were believed to be due to:
116 F.P. Nasution et al. / International Journal of Fatigue 59 (2014) 114–128

Fig. 2. Flexible power cable attached to a floating offshore structure.

Fig. 3. Point and inline contact in a stranded copper conductor. (a) Point contacts between inner and outer layer. (b) Plan view of conductor cross section. The black arrows
show point contact between wires in adjacent layers (outer and inner layer), the red arrows show inline contact between wires within one layer and between inner layer and
core wire. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
 F.P. Nasution et al. / International Journal of Fatigue 59 (2014) 114–128 117

Fig. 4. Inline contact between two wires within layers and between inner and centre wires.

1. The effect of localised bending between layers in the ten-

sion–bending mode.
2. The effect of friction between layers.
3. The effect of the irregularities in terms of stress
concentration.

The specimens tested in [16] were inspected by Scanning Elec-

tron Microscopy (SEM). Although, it was not possible to see any
signs of fretting, neither from the SEM results nor by visual inspec-
tion, it was not possible to completely rule out that fretting con-
tributed to the crack initiation process. Compared to steel rope
and aluminium conductors the contact stresses in copper conduc-
tors are lower, and the displacements larger and it is at least rea-
sonable to assume that fretting plays a less dominant role in the
fatigue performance of compacted copper conductions exposed
to the loading conditions investigated here. This is also inline with
the observations by Karlsen [12]. One reason for this may be the in-
creased contact area and reduced contact stresses from the com-
pacting procedure. If the assumption of negligible influence from
fretting is correct it should be possible to bridge the gap noted be-
tween the full cross-section SN data and the individual wire SN
data obtained from tension testing based on beam and beam con-
tact theory. Fig. 5. Point contact between two cylinder bodies with elliptical contact surface
The purpose of the present paper is therefore to perform beam and angle in the plane.

stress analysis to verify whether this is the case. The following pro-
cedure is adopted:
stresses obtained in the full cross-section tension–bending
tests presented in [16] and the tension–tension tests pre-
1. Build a FE model based on beam and beam-contact
sented here.
elements.
7. Use the simulated stresses and plot the results into the
2. Formulate a simplified contact stiffness and friction model
individual wire actual longitudinal stress range SN-curve.
representative for large contact areas (plasticised contact
surfaces).
3. Calibrate the normal contact stiffness model by full cross- 2. Experimental tests
section axial stiffness testing.
4. Calibrate the friction model by measured friction 2.1. General
coefficients.
5. Present SN data from full cross-section tension–tension The specimens used in this work were taken from a 95 mm2
tests. copper power conductor (ETP copper), designated by the UNS
6. Perform FE stress analyses, include the stress concentration C11000 series. The definition of ETP copper is related to copper al-
factors (SCFs) obtained from measured geometric irregular- loy purity of at least 99.95% and characterised by a very high elec-
ities obtained [15,16] and estimate the actual longitudinal trical conductivity and ductility. The conductor cross section
118 F.P. Nasution et al. / International Journal of Fatigue 59 (2014) 114–128

consisted of 19 wires, each with a nominal diameter of 2.5 mm. A
centre wire is followed by 6 and 12 helically wound wires in two
layers. The pitch lengths were measured to be 220 mm for both
layers. When using the right hand lay rule the obtained lay angles
were 4.1° and +7.3° for the inner and outer, respectively. The
modulus of elasticity (E) for copper is 115 GPa [12].
2.2. Previous fatigue tests
As previously reported elsewhere [15,16], SN curves based on
nominal stress range (Dr) and actual longitudinal stress range
(Drxx) have been constructed based on individual outer wires test-
ing in tension–tension mode. The actual longitudinal stresses (rxx)
of individual wires were assessed based on the measured SCFs.
The maximum SCF in the axial direction, SCF(a), occurred at thin-
nest section as the observed irregularities and was calculated as:
rpeak
SCF ðaÞ ¼
rnominal
where rpeak represents the maximum stress occurring and rnominal is
the nominal applied axial stress. The mean SCFs based on Eq. (1)
were found to be 1.238 and 1.058, for the outer and inner wires,
respectively.
The results showed that the fatigue lifes for the outer wires were
consistently below the fatigue lifes for the inner layer and centre
wire. All the observed fatigue failures occurred at the thinnest sec-
tion. This indicates that in these tests the fatigue failures were gov-
erned by the irregularities resulting from additional bending stresses.
Fatigue bending tests of full cross-section conductors were also
conducted. All fatigue failures occurred in the inner layer, in the
section that had been bent over a bellmouth. The observed fatigue
data obtained from full cross-section testing and the stress range
estimated by only considering wire bending showed that the fati-
gue strength of the full cross-section conductors was significantly
lower than the fatigue strength found for individual wires. This
may be explained by the fatigue strength being governed by local-
ised bending and friction effects between layers, which is the main
topic of this work.
was more severe for the outer layer, as noted for the individual
wires fatigue data reported in [15,16]. The detailed listing of the
tension–tension test results is found in Table 1.
Fig. 8 displays the full cross-section fatigue data in tension–ten-
sion mode compared to the individual outer wires fatigue data
based on the nominal stress ranges Dr. Noting that the black line
shows average of fatigue data taken from individual outer wires in
tension–tension mode. The upper and lower bounds of the SN
curve are based on nominal stress range with the plus/minus
two standard deviations scatter band.
The results show that by using the same level of longitudinal
stress ranges as the individual wires, the fatigue strength obtained
from full cross-section testing is very close to the fatigue strength
obtained from individual wire testing in tension–tension mode for
the outer layer.
2.5. Friction tests
ð1Þ
A series of tests on friction performance of copper wire against
copper wire were conducted. The purpose of these tests was to
investigate the coefficient of friction between the copper wires.
Several constant normal force conditions were investigated in
the tests. The tangential and normal loads were monitored through-
out the tests and the associated coefficient of friction l calculated.
Small variations were noted between the tests and the coeffi-
cient of friction was found to be 0.2.
3. Layer contact formulations
3.1. General
In the present study, the layer contact formulation in terms of
copper wire interaction between adjacent or within layers are split

2.3. Calibration test

The calibration test used to tune the FE model of the 95 mm2

copper power conductor was performed by investigating the mea-
sured axial stiffness vs. FE and analytical results.
A copper power conductor with free span length 1000 mm was
mounted in a servo-hydraulic test machine. On each individual
wire of the outer layer strain gages were attached for measuring
the axial strain. In addition, the specimen was equipped with a
extensometer for monitoring the axial strain of conductor.

2.4. The conductor tension–tension fatigue test

Six specimens with free span length 1000 mm and cross section
area 95 mm2 were subjected to nominal stress ranges (Dr) of
160 MPa and 190 MPa with stress ratio, R = 0.1. All the specimens were
terminated at the ends using standard sockets filled with high strength
epoxy resin. The intention of the tests was to provide full scale bench-
mark data on fatigue performance in tension–tension mode.
The tension–tension fatigue testing was carried out in a stan-
dard two-column servohydraulic test machine with hydraulic
clamp at both ends (Fig. 6). The loading was sinusoidal with fre-
quency 2 Hz, in load control.
All the fatigue failures consistently developed at the outer layer
in the thinnest section of the wire (at point contact area), as shown
in Fig. 7. This complies with the observation that the irregularities
resulting from plastic deformation from the compacting process Fig. 6. Standard fatigue test rig of a full cross-section by mounting the both ends.
 F.P. Nasution et al. / International Journal of Fatigue 59 (2014) 114–128 119

Fig. 9. Inline contact between two wires within a layer and between inner and
central wires by introducing total equivalent width beq.

into two distinct conditions based on the area of contact. The con-
dition includes inline contact governing the contact between wires
within layers and between inner layer and centre wire. The second
condition is point contact between outer and inner layers.
The Hertzian contact theory governs when the contact zone is
limited compared to the surrounding material volume (elastic
range and small strains). In the present case, the contact condition
between wires (inline and point contact) have developed from the
Fig. 7. A failed full cross-section copper power conductor at outer layer in tension–
manufacturing processes with associated large plastic deformation
tension mode.
between wires.

Table 1
Fatigue test data for full cross-section due to tension–tension loads.

No Dr MPa No. cycles to failures [cycles] Irregularities [mm] Failure position (Layer)
1 160 776,070 0.24 Outer
2 160 797,203 0.235 Outer
3 160 898,909 0.20 Outer
4 190 308,412 0.165 Outer
5 190 409,001 0.125 Outer
6 190 421,446 0.115 Outer

Fig. 8. SN fatigue data for full cross-section compared to nominal SN curve based on individual wires testing in tension–tension mode.
120 F.P. Nasution et al. / International Journal of Fatigue 59 (2014) 114–128

Fig. 10. Contact geometry area between inner and outer layer.

bodies. Since the strategy adopted here is based on beam elements,

believing that the fatigue performance is primarily governed by
longitudinal stresses, the contact conditions need only to be repre-
sented with sufficient accuracy with respect to the load sharing be-
tween layers, local bending at the trellis points and friction due to
the relative displacements between wires. Therefore, a simplified
procedure for representing the contact conditions has been
adopted applying the concept of elastic springs in the normal
direction and elastic–plastic springs in the tangential direction to
deal with friction.

3.2. Contact in normal direction

3.2.1. Inline contact

The total equivalent width, beq, of the contact zone between
wires within a layer and also between inner and central wire is
shown in Fig. 9, where A denotes the cross section area of the
wires, taken as 5 mm2 and r is the wire nominal radius of 1.25 mm.
The inline contact force finline due to the elastic deformation is
shown in:

Ebeq
finline ¼ C in dn ð2Þ
R

where Cin is a calibration factor for inline contact and dn denotes the
A
normal contact displacement. By substituting R = 2r and beq = 2r , Eq.
(2) can be rearranged as

EA
finline ¼ C in dn
4r2 ð3Þ
EA
Fig. 11. Stick–slip displacement on the wire interfaces. (a) Curvilinear beam kinline ¼ C in 2
deformations. (b) Stick–slip displacement quantities between helix wire and centre. 4r

The inline contact stiffness kinline represents the normal stiffness to

The surface areas are therefore enlarged with associated in- be applied for the inline contacts within inner and outer layers and
creased resistance against relative motion between contacting between inner and central wires.
 F.P. Nasution et al. / International Journal of Fatigue 59 (2014) 114–128 121

3.2.2. Point contact

Fig. 10 represents the contact geometry area, Ac, as a parallelo-
gram between outer and inner wires indicated by the shaded area.
The contact area Ac is the product of Leqbeq, given as
D
Ac ¼ D ð4Þ
jsinða2  a1 Þj
where D is the wire diameter, a2 and a1 are the lay angles of outer
and inner wires, respectively.
The point contact force, Fpoint, acting on the inner wires at point
O is calculated as,
Ac E
F point ¼ C po dn ð5Þ
2r

EAc
kpoint ¼ C po ð6Þ
2r
where Cpo is a calibration factor for point contact.

3.3. Stick–slip model

Fig. 13. Individual wire model. (a) Outer wire model [15,16]; (b) Ovalised wire
The contact elements applied are based on an ideal elastic–plas- cross section at section B–B (at the thinnest section).
tic relation where the stick stiffness is given by
kstick ¼ lfinline =D ð7Þ
where D represent the relative displacement at which slip occurs,
after which the stiffness is zero (Coulomb friction).
The D-parameter needs to be selected such that adequate stiff-
ness is obtained, i.e., Navier hypothesis of plane surfaces remain
plane applies before slip occurs. In order to ensure this, sandwich
beam theory is applied looking at the internal work performed
along the helix path using the approach proposed by Sævik
[17,18]. Noting that the free relative displacement in one layer is
proportional to the helix radius squared, a layer by layer approach
is assumed. This means that the stick stiffness setting at the inter-
faces belonging to one layer is based on the helix radius in this
layer. The stick stiffness can be studied by considering the internal
work done (Wi) along a quarter pitch between the wire and the
surrounding interfaces which for inline contacts can be written as:
Z p=2    
@ v @up @ v @up
Wi ¼ EA þ d þ
0 @x @x @x @x
X
n
R
þ kstick v dv dW ð8Þ
i¼1
sin a

where EA is axial stiffness in the layer, v = uw  up, is the relative

displacement at the interface between helix wire and the centre Fig. 14. A conductor model subjected to tension–tension loads (Dr) with stress
ratio R = 0.1.
along the helical path (see Fig. 11), uw denotes the longitudinal helix
wire displacement, n is the number of interfaces sharing the inter-
action process and up represents the wire displacement that would is the helix radius of wire, j represents the global curvature and W
occur if plane surfaces remained plane during the stick condition. R is the angular position, see Fig. 11a. The plane surfaces remain plane
strain is taken as:
dup
¼ R cos2 a cos Wj ð9Þ
dx
Furthermore, by assuming the relative displacement (v) as:
v ¼ v 0 sin W ð10Þ
The following expression is obtained for v0:
2a
R2 cos j
v0 ¼  sin a
kstick R2
 ð11Þ
1 þ n EA sin2 a

where it has been assumed that each interface carries equal load,
i.e., shared among n interfaces.
The upper term represents the free displacement which
Fig. 12. Complete 3D model of full cross-section power copper conductor. approaches zero when the shear interaction stiffness approaches
122 F.P. Nasution et al. / International Journal of Fatigue 59 (2014) 114–128

For the point contacts this is modified into:

3R2 F point l
Dpoint ¼  
2
ð14Þ
1
F tol
 1 Lp EA sin a

where Lp is the length between contact points (see Fig. 3).

4. Finite Element (FE) modelling

4.1. General

Uflex3D is a finite element tool for 3D stress analysis of umbili-

cals [19], however, only including beam stresses due to internal
pressure, external pressure, tension, torsion and bending loads.
The copper power conductor model was made using elastic
beam elements for all wires with a constant elastic bending, axial
and torsion stiffness.
The full cross-section copper conductor model was 1000 mm
long including one centre wire, six wires for inner layer and twelve
wires for outer layer, see Fig. 12. The model was meshed with 500
elements per wire in the longitudinal direction.
The contact elements include four parts, namely:

1. Point contact elements. Positioned between inner and outer

layers.
2. Inline contact elements, within layers.
Fig. 15. Tension–bending model at initial condition. 3. Inline contact between centre wire and inner wires.
4. Contact between the outer layer of the conductor and the bend
surface (i.e., bellmouth).
infinite. However, the shear interaction stiffness must be selected
based on a compromise between numerical stability and accuracy.
The bellmouth was modelled as a geometrical surface, where a
In this study, the shear interaction stiffness is selected by allowing
penalty stiffness is applied during wire-surface contact. No friction
<10% relative displacement in the stick phase. This gives the fol-
is included in this contact element.
lowing requirement:
The wire to wire contact elements were given an elastic stiff-
  2
1 1 EA sin a ness in the normal direction whereas friction was treated by an
kstick P 1 2
ð12Þ elastic–plastic model using the stick displacement parameter
n F tol R
established from the above.
where Ftol denotes the allowable tolerance for stick relative dis- In order to perform the fatigue calculation by FE analysis, this
placement. The stick–slip parameter for the inline contact is then has to be performed in two steps and based on the actual longitu-
established using Eq. (7) and selecting n = 3 as, dinal stresses (rxx) given as:
3R2 finline l rxx ¼ rxa SCF ðaÞ þ rxby SCF ðbyÞ þ rxbz SCF ðbzÞ ð15Þ
Dinline ¼  
2
ð13Þ
1
F tol
 1 EA sin a
where rxa represents the nominal axial stresses, rxby is the nominal
bending stresses in y-direction (weak axis), and rxbz is the nominal

Fig. 16. Comparison results of axial force vs. strain obtained from calibration tests, FE and theoretical model.
 F.P. Nasution et al. / International Journal of Fatigue 59 (2014) 114–128 123

Table 2
Detailed contact parameters of full cross-section copper conductor.

Interface Contact forces Nominal stiffness per interface Stick–slip per interface (Eq. (7)) Slip parameters [mm] (Eqs. (13) and (14))
2 2
Centre-inner layer 0.65 N/mm 200 N/mm 1411.5 N/mm 9.21  105
Inner layer–inner layer 0.65 N/mm 200 N/mm2 1411.5 N/mm2 9.21  105
Outer layer–outer layer 1.8 N/mm 200 N/mm2 1411.5 N/mm2 2.55  104
Outer layer–inner layer 12 N 930 N/mm 2.16  104 N/mm 1.11  104

Fig. 19. Axial force distributions of individual wires from different layer due to
Fig. 17. Point contact distribution between outer and inner layers due to tensile tensile load 9500 N.
load 9500 N.

Fig. 18. Inline contact distribution due to tensile load 9500 N (Red line is hidden
behind the blue line). (For interpretation of the references to color in this figure
legend, the reader is referred to the web version of this article.)

bending stresses in z-direction (strong axis) and according to the

following procedure:

1. Calculate the fatigue representative for the sections away from

the point contacts (away from the thinnest section) by using 1.0
for all SCFs in all layers.
2. Calculate the fatigue at the point contacts having the thinnest
sections by using the obtained data for the stress concentration
factors due to axial loads, SCF(a), as reported previously [15,16].

For the tension–bending model, the SCFs in the thinnest sec-

tions also included the weak (y-direction), SCF(by), and strong (z-
direction) axes, SCF(bz). Fig. 13 illustrates the associated wire cross Fig. 20. Longitudinal stresses due to tension–tension loads. (a) Dr = 160 MPa; (b)
Dr = 190 MPa.
section at the thinnest section (section B–B). The compacting pro-
cess lead to plastic deformation such as ovalising the wire cross
section at the thinnest section due to point contact. The SCF in Dnominal  t reduced
SCF ðbyÞ ¼ ð16Þ
bending about weak axis, SCF(by), at section B–B is given by Dnominal
124 F.P. Nasution et al. / International Journal of Fatigue 59 (2014) 114–128

Fig. 21. Fatigue damage distribution of full cross-section conductor model by

Dr = 160 MPa. (a) Outer layer; (b) Inner layer.

Fig. 23. Longitudinal stresses for outer layer in tension–tension mode. (a)
Dr = 160 MPa; (b) Dr = 190 MPa.

Fig. 24. Curvature along the model at maximum deformation.

2.5 mm, resulting in SCFby equal to 0.92 and 0.96 for the outer
and inner wires, respectively. For the SCF in bending about strong
axis, SCF(bz), was taken to be 1.0 because it will not be important
for the maximum stress.
Fig. 22. Predicted fatigue damage distribution of full cross-section conductor model
by Dr = 190 MPa. (a) Outer layer; (b) Inner layer.
4.2. Tension–tension model

where the average treduced was found to be 0.21 and 0.1 mm for out- The principal model of full cross-section copper conductor ex-
er and inner wires [15,16], respectively. Dnominal was taken to be posed to the tension–tension loads with stress ratio, R = 0.1 is
 F.P. Nasution et al. / International Journal of Fatigue 59 (2014) 114–128
Fig. 25. Axial force histories acting at end beams due to reversed bending loads.
Fig. 26. Actual longitudinal stresses vs. curvature hysteresis loops due to reversed
bending motions at the location of maximum fatigue stress. (a) q1 = 1500 mm; (b)
q2 = 3000 mm.
shown in Fig. 14. The applied nominal stress ranges, Dr, were 160
and 190 MPa, for the fatigue load cases in accordance with the test
procedure.
4.3. Tension–bending model
FE analyses were also carried out for the full cross-section con-
ductor model subjected to reversed bending loads over the
125
Fig. 27. Predicted fatigue damage distribution of copper conductor model due to
reversed bending motion with radius of curvature q1 = 1500 mm. (a) Inner layer;
(b) Outer layer.
bellmouth where both ends were axially pretensioned by end
beams with a constant tensile force of 9500 N. The test setup in-
cluded 3000 mm model length. However, the full detail FE model
length was truncated to 1000 mm (five pitches) due to computing
limitations.
Two radii of bellmouth curvature were investigated
q1 = 1500 mm and q2 = 3000 mm at a constant tension of 9500 N
according to the test parameters.
Fig. 15 presents the complete FE model at the initial condition.
The end beams were connected to the outer external tubular
frames 1 and 2, where the cyclic bending is induced by prescribed
displacement of the beam actuator. This motion is then transferred
to the external frames by means of the transverse beam and the
attachment arms at 1 and 2.
5. FE results
5.1. Calibration
Fig. 16 shows the correlation results obtained from the proce-
dure described in Section 3. The results show good agreement be-
tween the calibration test, FE and theoretical results. Similar
correlation was also obtained by Papailiou [10]. The obtained con-
tact calibration factors for inline (Cin) and point contact (Cpo) were
2.2  106 and 2.1  102, respectively. The associated stick–slip
parameters are given in Table 2.
The point contact force distribution due to maximum tensile
load 9500 N is shown in Fig. 17. The point contact forces show that
the quantities are nearly similar over the model length unless close
to the end fittings, the average point contact forces indicating
12 N.
126 F.P. Nasution et al. / International Journal of Fatigue 59 (2014) 114–128
Fig. 28. Predicted fatigue damage distribution of copper conductor model due to
reversed bending motion with radius of curvature q2 = 3000 mm. (a) Inner layer; (b)
Outer layer.
The inline contact forces distribution within inner layer, be-
tween centre and inner layer and within outer layer are compared
in Fig. 18. The results demonstrate the inline contact force distribu-
tion within inner layer is similar to the inline contact force be-
tween centre and inner layer where the maximum inline contact
force is approximately 0.65 N/mm. For the outer layer the maxi-
mum inline contact force is close to 1.8 N/mm. The axial force
distributions of individual wires for the different layers is pre-
sented in Fig. 19. The highest axial force occurred in the centre wire
with 518 N. For the inner and outer wires, the axial forces are
505 N and 496 N, respectively, which demonstrates almost equal
load sharing between layers.
5.2. Tension–tension fatigue
The obtained actual longitudinal stress history of outer and in-
ner wires from full cross-section and individual wire models are
presented in Fig. 20. It is seen that the results compare well with
longitudinal stresses of individual wires for both load cases.
The fatigue damage distributions of the models for one load cy-
cle have been calculated. Thus the expected fatigue life can be as-
sessed as the inversion of the fatigue damage. The red spots
indicates the occurred maximum fatigue damage.
Fig. 21 shows the fatigue damage distributions for outer and in-
ner layers for nominal stress range Dr = 160 MPa. The predicted
number of cycles to failure are 1.17  106 and 2.13  106 cycles
for outer and inner layers, respectively. The first predicted fatigue
failure is located in the outer layer in accordance with the test.
For nominal stress range Dr = 190 MPa, the fatigue damage dis-
tributions are displayed in Fig. 22. The predicted number of cycles
to failure are 3.33  105 and 5.26  105 cycles for outer and inner
Fig. 29. Actual longitudinal stresses vs. curvature hysteresis loops due to reversed
bending motions for different number of elements at maximum deformation. (a)
q1 = 1500 mm; (b) q2 = 3000 mm.
layers, respectively. Similar to above results, the first predicted fa-
tigue failure occurred at outer layer as observed in the test.
In order to ensure sufficient accuracy, a mesh sensitivity check
was carried out with respect to actual stress range and predicted
fatigue life using 800 elements.
Fig. 23 shows the actual longitudinal stress results for outer
wires by varying number of elements. The outer layer longitudinal
stresses converged to 203 MPa and 235 MPa for applied nominal
stress ranges Dr = 160 and 190 MPa, respectively.
The first predicted fatigue failures took place at outer layer
where the predicted fatigue life converged to 1.17  106 and
3.33  105 cycles for applied nominal stress ranges Dr = 160 and
190 MPa, respectively.
5.3. Tension–bending fatigue
The curvature along the model at maximum deformation is
shown in Fig. 24 for q1 = 1500 and q2 = 3000 mm.
The end beams were applied to simulate the die springs by
associating the same axial stiffness and then introducing a preten-
sion of 9500 N as indicated at point A in Fig. 25. Next, the reversed
bending loads were exposed to the midspan of the model by the
bellmouth contact surface.
 F.P. Nasution et al. / International Journal of Fatigue 59 (2014) 114–128 127

Fig. 30. Comparison between the predicted fatigue life using individual wires SN data and full cross-section fatigue testing in tensiontension mode. The individual wires SN
curve is based on the actual longitudinal stresses.

Fig. 31. Comparison between the predicted fatigue life using individual wires SN data and full cross-section fatigue testing in tensionbending mode. The individual wires SN
curve is based on the actual longitudinal stresses.

Fig. 26 presents the longitudinal stresses vs. curvature hystere-
sis loops during the reversed bending motion. The hysteresis loops
corresponds to the frictional energy dissipated at the contact inter-
faces. The results exhibit that frictional energy dissipated in the
cyclic bending regime is larger for radius of curvature
q1 = 1500 mm than q2 = 3000 mm. The larger energy dissipated,
the larger longitudinal stress range. Therefore, the accumulated en-
ergy dissipated at the contact wire surfaces play an important role
in assessing fatigue damage of copper power conductors. The ac-
tual longitudinal stress ranges (Drxx) are 241 and 191 MPa for ra-
dius of curvature q1 = 1500 and q2 = 3000 mm, respectively, as
compared to 190 MPa and 95 MPa only considering pure bending.
The predicted fatigue damage distribution results for different
layers are displayed in Fig. 27 for radius of curvature 1500 mm.
The results show that the predicted first fatigue failure occurred
at inner layer within the bellmouth. This is in accordance with
the observed fatigue failures seen in the tests [16]. The predicted
number of cycles to failure are 2.33  105 and 4.0  105 cycles
for inner and outer layers, respectively.
For radius of curvature of 3000 mm, the predicted fatigue dam-
age distribution is shown in Fig. 28, the predicted first fatigue fail-
ure located at inner layer inside the bellmouth. The estimated
fatigue lifes are 1.88  106 and 5.0  106 cycles for the inner and
outer layers, respectively.
A mesh sensitivity check was performed using 800 elements.
Fig. 29 presents the hysteresis loops of actual longitudinal stresses
vs. curvature in the inner layer comparing to different number of
elements. For radius of curvature q1 = 1500 mm and
q2 = 3000 mm, the longitudinal stress ranges converged to 241
and 191 MPa, respectively.
6. Discussion of FE analyses and test results
FE analyses of a full cross-section copper power conductor ex-
posed to tension–tension and tension–bending loads have been
carried out including mesh sensitivity checks.
The predicted first fatigue failure of copper conductor model in
tension–tension fatigue mode is located at outer layer as shown in
Figs. 21 and 22. This is in accordance with the observed fatigue fail-
ure as shown in Fig. 7. It indicates that the fatigue failures were
governed by the local bending effects due to surface irregularities
on the wires where the most severe irregularities were found in
the outer layer [16].
128 F.P. Nasution et al. / International Journal of Fatigue 59 (2014) 114–128
In Fig. 30, the fatigue data of full copper cross-section in ten-
sion–tension mode is compared to the SN fatigue data of individual
outer wires in tension–tension mode based on the actual longitu-
dinal stress range. The predicted number of cycles to failure show
good correlation with the fatigue life data based on the actual lon-
gitudinal stresses. All the fatigue life results are within the fatigue
scatter band.
Moreover, the predicted first fatigue failure due to reversed
bending motions took place at inner layer within the bellmouth.
This is in accordance with the observed fatigue failures from the
testing. This indicates that the fatigue bending failures are caused
by friction and localised bending at the point contact between out-
er and inner layers. The friction effect created enlarged longitudi-
nal stresses at the point contact section as shown in Fig. 26. The
fact that friction forces will be largest in the inner layer has also
been noted previously by Papailiou [10] and Hong et al. [11].
The comparison between predicted fatigue life and fatigue life
data in tension–bending mode are presented in Fig. 31, noting that
the stress values used to present the test data only include the ana-
lytical bending term. The test data are seen to be within the scatter
band of the predicted fatigue lifes using FE analysis and individual
wires fatigue data.
7. Conclusions and future works
2
Fatigue strength investigations of a 95 mm copper power con-
ductor by FE analyses have been conducted in tension–tension and
tension–bending modes. FE models were established based on
beam and beam contact elements, where a simplified contact mod-
elling procedure was adopted. The FE model was calibrated by ax-
ial stiffness testing and validated by mesh sensitivity checks.
Relevant friction coefficients were obtained by testing. Two alter-
native full cross-section fatigue tests of a 95 mm2 copper power
conductor were conducted in tension–tension and tension–bend-
ing modes, a total of twelve conductor specimens were tested
and the results were compared to FE analysis and applying SN fa-
tigue data obtained from individual wires tests. Several conclu-
sions can be drawn in this paper in terms of fatigue performance
of copper power conductors.
1. In order to perform accurately FE analysis calibration tests
are required to capture realistic load sharing between lay-
ers and in assessing the fatigue life.
2. For the tension–tension mode, all the predicted first fatigue
failures were initiated at the outer layer. The obtained
actual longitudinal stresses from full cross-section model
are close to the actual longitudinal stresses of individual
wires.
3. For the tension–tension test of full cross-section copper
conductors, all the first fatigue failures were initiated at
outer layer in the thinnest section. This indicates that the
fatigue failures were governed by the local stress concen-
tration factors (SCFs) due to the effect of surface irregular-
ities in the outer wires. Good correlation between tests and
FE analysis was obtained.
4. For the tension–bending mode, all the predicted fatigue
failures invariably occurred at inner layer inside the bell-
mouth. This is in accordance with the observed fatigue fail-
ures in the fatigue bending tests. This indicates that the
effect of friction between outer and inner layers at the
point contact area plays an important role in assessing
the fatigue life of full cross-section copper conductors.
5. As long as the longitudinal stresses governs the fatigue per-
formance FE analysis with beam elements can be used to
predict realistic fatigue lifes.
Work is on-going in terms of FE analyses and experimental
works for studying the fatigue strength of a 300 mm2 copper
power conductor subjected to tension–bending mode and individ-
ual wires tested in tension–tension mode applying similar
methodologies.
Acknowledgement
The authors acknowledge ABB, Karlskrona, Sweden for provid-
ing test materials applied in the test.
References
[1] Mindlin RD. Compliance of elastic bodies in contact. J Appl Mech
1949;71:259–68.
[2] Dong RG, Steidel Jr FR. Contact stress in stranded cable. Exp Mech
1965;5(5):142–7.
[3] Hobbs RE, Raoof M. Interwire slippage and fatigue prediction in stranded
cables for TLP tethers. In: Chryssostomidis C, Connor JJ, editors. Proceedings of
third international conference on behaviour of offshore structures, Boston, vol.
2; 1982. p. 77–99.
[4] Johnson KL. Contact mechanics. Cambridge University Press; 1985.
[5] Raoof M, Hobbs RE. Analysis of multilayered structural strands. J Eng Mech
1988;114:1166–82.
[6] Raoof M. Axial fatigue of multilayered strands. J Eng Mech 1989;116:2083–99.
[7] Raoof M. Axial fatigue life prediction of structural cables from first principles.
Proc Inst Civil Eng 1991;91:19–38.
[8] Raoof M, Huang Yu Ping. Cyclic bending characteristics of sheathed spiral
strands in deep water. Int J Offshore Polar Eng 1993;3:189–96.
[9] Zhou ZR, Cardou A, Fiset M, Goudreau S. Fretting fatigue in electrical
transmission lines.. Wear 1994;173:179–88.
[10] Papailiou KO. On the bending stiffness of transmission line conductors. IEEE
Trans Power Deliv 1997;12.
[11] Hong K-J, Der Kiureghian A, Sackman JL. Bending behavior of helically
wrapped cables. J Eng Mech 2005;131.
[12] Karlsen S. 2010. Fatigue of copper conductors for dynamic subsea power
cables. In: Proceedings of the OMAE 2010 29th international conference on
ocean, offshore and arctic engineering, Shanghai, China.
[13] L évesque F, Goudreau S, Cloutier L. Elastic–plastic microcontact model for
elliptical contact areas and its application to a treillis point in overhead
electrical conductors. J Tribol 2011;133.
[14] Nasution FP, Sævik S, Gjøsteen JKØ. Fatigue analysis of copper conductor for
offshore wind turbines by experimental and FE method. Energy Procedia. In:
Proceedings of 9th deep sea offshore wind R&D, Trondheim; 2012.
[15] Nasution FP, Sævik S, Gjøsteen JKØ. Study of fatigue strength of copper
conductor considering irregularities surfaces by experimental testing and FE-
analysis. In: The ASME 2012 31st international conference on ocean, offshore
and arctic engineering, Rio de Janeiro; 2012.
[16] Nasution FP, Sævik S, Gjøsteen JKØ, Berge S. Experimental and finite element
analysis of fatigue performance for copper power conductor. Int J Fatigue
2013;47:244–58.
[17] Sævik S. Theoretical and experimental studies of stresses in flexible pipes.
Comput Struct 2011;89:2272–91.
[18] Sævik S. On stresses and fatigue in flexible pipes. Ph.D thesis at Department of
Marine Structures. NTNU, Norway; 1992.
[19] UFLEX3D version 1.0.1 theory and user manual.