See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/301418596

Reeling Pipeline Material Characterization: Testing, Material Modeling and

Offshore Measurement Validation

Conference Paper · May 2015

DOI: 10.1115/OMAE2015-41919

CITATIONS READS

5 592

4 authors, including:

Erwan Karjadi Philippe Thibaux

Heerema Marine Contractors OCAS
21 PUBLICATIONS   36 CITATIONS    66 PUBLICATIONS   328 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Erwan Karjadi on 08 January 2018.

The user has requested enhancement of the downloaded file.

 Proceedings of the ASME 2015 34th International Conference on Ocean, Offshore and Arctic Engineering
OMAE2015
May 31-June 5, 2015, St. John's, Newfoundland, Canada

OMAE2015-41919

REELING PIPELINE MATERIAL CHARACTERIZATION – TESTING, MATERIAL

MODELING AND OFFSHORE MEASUREMENT VALIDATION

Erwan Karjadi Helen Boyd

Heerema Marine Contractors Heerema Marine Contractors
Leiden, The Netherlands Leiden, The Netherlands

Harm Demmink Philippe Thibaux

Heerema Marine Contractors OCAS
Leiden, The Netherlands Zelzate, Belgium

ABSTRACT in 2013. The material testing of a sample cut out from

Spoolbase test and trial, undergoes spool and un-spool 5
It is a fact that when a pipeline is installed by the reeling cycles, has been performed to confirm the distortional plasticity
method, it will undergo cyclic plastic straining and the pipe will hardening behavior obtained from the small scale Bauschinger
plastically deform. Due to the applied plastic bending moment, and perpendicular loading tests.
the residual deformation in terms of residual pipe ovality after
reeling is difficult to predict by Finite Element Analyses (FEA) INTRODUCTION
without a thorough understanding of the material
characterization and changes under cyclic plastic straining. Reel-lay is a well-established installation method for
offshore rigid pipelines and its rapidity and efficiency makes it
The paper describes how the material behavior of seamless particularly cost attractive for infield flowlines/risers and
pipe under plastic strain reeling cycle has been characterized smaller diameter export lines. Heerema new vessel the Aegir,
by a comprehensive material testing program including which has been built since 2011 and delivered in Q3 of 2013,
Bauschinger tests and perpendicular loading pre-straining was utilized for the first reeling project campaign in the Gulf of
tests. It turns out that for seamless pipe, by looking at the yield Mexico at the end of 2013. Along with the vessel & equipment
stress locus of the material after plastic straining, the reeled development, a reeling pipeline knowledge development
pipe material which initially shows isotropic behavior in the project has been commenced in 2011 [1] where a series of full-
un-strained condition will change and evolve to show scale bend tests, and small scale material tests are performed
anisotropic behavior. The material in the hoop direction of the and complimented with the development of finite element
pipe will become more hardened than the material in the models to perform analyses simulation of the bend rig tests and
longitudinal direction of the pipe. The cross hardening full reeling cycles on the spool yard and on board of the Aegir.
characteristics of material under cyclic plastic deformation The purpose is to obtain accurate prediction models that can be
have been modeled using the “distortional plasticity” principle used to guarantee safe reeling of pipelines for the Aegir.
and implemented in a user subroutine of an FEA software
package. Reference [2] presents the overview of the development of
the reeling pipeline analyses simulation with the required
This paper includes the validation of the ovality prediction by validation test program. The ultimate goal is to develop FEA
FEA model using the developed material model against the models which can be used to accurately predict the maximum
ovality measurement from full scale bend tests at Heriot-Watt and residual strain/ovality in the pipeline exposed to the loading
University as well as ovality measurements taken during the experienced during the installation steps of reeling. To obtain
spooling test and trial of 16”OD pipeline at Carlyss spool base an accurate prediction model one important aspect is

1 Copyright © 2015 by ASME

understanding the mechanical behavior of material under
plastic strain cycles.

This paper presents part of the scope of HMC’s reeling

analyses development and test program for the DCV Aegir:
 Small scale material tests to characterize the evolution
of yield stresses under plastic strain cycles;
 Selection and implementation of distortional plasticity
material model and validation of user material model
implementation against ovality measurements during
full scale bend tests;
 Validation of the material model implementation by
offshore spool base ovality measurements;
 Validation of the material characterization small scale
testing by offshore spool base material testing.

HEEREMA REELING PIPELINE INSTALLATION Figure 2. Spooling 16” OD pipeline on a lifted reel

HMC’s DCV Aegir, as shown in Figure 1, has reel-lay and Depending on the layout of the spool yard a number of
J-lay pipeline installation equipment. Pipeline can be installed supports will be provided between the spooling tensioners and
with the Aegir in three modes: heavy J-lay with collars, light J- the level winder, which is equipped with horizontal guide
lay with friction clamps and reel-lay. References [1 to 3] give rollers. The location of the the level winder on the barge will be
an overview of DCV Aegir pipelay capability in more details. determined by the available space on the barge.

One of the unique reeling concepts of the DCV Aegir is the Reeling off on board of the Aegir
lifted reel-drums concept. Spooling of the pipeline will not be The reel/J-lay pipelay equipment, as shown in Figure 3,
performed directly onto the reel drum on DCV Aegir. Instead, will have tensioners with a maximum capacity of 800mT top
in the yard pipeline stalks will be spooled onto several reel tension which are capable of executing pipelay installation
drums which will have been placed on a transportation barge. projects in ultra-deep water for infield flowlines and risers. The
After spooling up the pipelines, the barge will sail to DCV pipelay tower in reeling/J-lay mode can be set to as low as 50
Aegir location and the reels will be transferred to the deck of degrees which allows the possibility to install pipelines with
the vessel by the 4000mT mast crane on board the DCV Aegir. reeling in relatively shallow water for smaller diameter export
lines.

The tower top module for reeling consists of an aligner

wheel with a radius of 9m, two straightener tracks to straighten
the pipe before lowering to the firing line, and A&R system for
abandoment and recovery operation of pipeline during
installation. The hang-off clamp will be used to hold the pipe
catenary when welding a new reel or in-line/pipe end
termination structure.

Figure 1. DCV Aegir vessel

Spool base
In a spool base yard, pipeline joints are fabricated into long
pipeline stalks before being spooled on to the reel drum on the
transportation barge as shown in Figure 2. The minimum radius
of the reel drum is 8m and the pipeline stalks can be loaded to
the reel up to the radius of 12m. The range of pipe diameter that
can be reeled is from 6” to 16” steel outside diameter of the Figure 3. Reeling off pipeline from the Aegir
pipe.

2 Copyright © 2015 by ASME

REELING PIPELINE INSTALLATION MODELLING IDmin) by the laser tool, and compared with the ovality that is
calculated with FEA. The definition of ovality as presented in
The development of Aegir reeling pipeline installation this paper is OOR/ODnominal.
analysis models involves on the one side continuous efforts in
FEA modeling of spooling-on pipeline on the yard and reeling- All of these test measurement results were compared with
off pipeline on board of the Aegir and on the other side a series the predictions based on the FEA model which is developed
bend rig tests/measurements and small scale material model according to the layout of HWU bend rig. Ref [2] showed when
tests. The full scale bend tests were performed at Heriot-Watt the test measurements were compared to the FEA using
University (HWU) in Edinburgh and the small scale material conventional isotropic or kinematic hardening models, there
model tests were performed at OCAS in Belgium (Figure 4). was a difference after the first cycle and the results significantly
diverging with the increasing number of cycles.

Samples of material (test rings) are cut out from the test
pipe section before bending and the samples are used for small
scale material tests to characterize the material behavior under
plastic deformation. The small scale test program includes
Bauschinger tests with different strain levels as well as
perpendicular loading tests with different cyclic pre-strain
levels, both in tension and compression mode.
HWU bend rig test measurement OCAS material test
The results of the material test program is used to develop
an accurate material model in order to get the FEA prediction
results for the pipe ovality, with a number of bend cycles, closer
to the test measurements.

MATERIAL CHARACTERIZATION UNDER CYCLIC

PLASTIC STRAIN

Material testing
FE Analysis - HWU bend rig test Finite element simulations in ABAQUS 6.12 [4] of the
bend rig experiments using isotropic hardening model predicts
Figure 4. Development process Aegir reeling FEA simulation
model a larger ovalization than observed during the experiments [1].
Therefore, the development of a more advanced material model
A series of bend rig tests on 16” OD carbon steel seamless was started. The material characterization testing program aims
pipes were performed at HWU including the ovality and strain at describing the evolution of the material behavior during the
gauge measurements of test pipe section which was bent to the reeling process. The loading was schematized as a succession
reel and straightener former up to 5 bend cycles. At one pipe of tensile / compression cycles, eventually combined with a
end, the 12m length test pipe is attached by a pin connection to biaxial condition at the point where the pipe is in contact with
the anchor end while the other pipe end is attached to a pull- the reel.
head which is connected by a wire to the crane. The pipe was
bent to the former during the bend test by pulling the crane Bauschinger tests
wire. Cyclic loading tests have been performed at OCAS N.V.,
Belgium, to investigate the effect of reversed plastic straining
Ovality measurements have shown to be a good tool to on the material behavior. Material samples were loaded
correlate the deformation of the pipe during full-scale bending subsequently in tension and compression and the stress-strain
tests with the corresponding deformation from a FEA curves were recorded. These tests are known as Bauschinger
prediction model. Ovality measurements have been performed tests. Bauschinger tests have been performed for 1%, 2% and
using the 2D laser tool from OMS (Optical Metrology Services 3.2% strain.
Ltd), to determine the pipe out-of-roundness at several
locations in the pipe and at different stages of the full-scale When applying the load in compression mode, depending
bending tests. The inside diameter specific tool was mounted on on the specimen size and the compression strain level, OCAS
wheels and pushed through the pipe using push rods to has developed an optimized design and test setup with anti-
predetermined locations as illustrated in Figure 4 (HWU bend buckle device system to prevent buckle of the specimen (see
rig test measurement). The ovality is calculated from the out-of- Figure 5). The shape of the grip inserts is optimized and the
roundness of the pipe (OOR=ODmax - ODmin), which was specimen was machined out to fit the anti-buckling plates to
derived from the inner pipe diameter measurements (IDmax and

3 Copyright © 2015 by ASME

improve alignment and avoid buckling in the thickness A guiding/anti-buckling system was applied during
direction of the sample. mechanical testing in order to decrease the risk of buckling in
the compressive part of the cycles. Due to the risk of
inhomogeneous deformation if the pre-strain would be of the
magnitude of the Lüders plateau, a pattern was applied prior to
the pre-strain and a deformation mapping was measured prior
to sampling for the second part of the testing as shown in
Figure 6.

Tensile testing of the un-strained (initial) material indeed

demonstrated that the Lüders plateau was of around 2% as
shown in Figure 7.
Heerema

700
(a) Standard test without ABS (b) with ABS 600
A2763-0°
A2760-0°

Engineering stress (MPa)

Figure 5. Compression tests with and without Anti Buckle System 500
(ABS)
400

Perpendicular loading tests 300

In addition to the Bauschinger tests, perpendicular loading 200

tests were performed. For these tests a larger sample was pre-
100
strained in the longitudinal direction, from which smaller
samples were machined and a tensile or compressive load was 0
applied in the transverse direction. The objective of these 0 5 10 15 20 25 30

perpendicular loading tests is to determine the effect on the Engineering strain (%)

yield stress of the material considering a pre-strain in the Figure 7. Stress-strain curve from tensile test un-strained material
perpendicular direction. The pre-strains were performed in
tension or compression, or in cycles of +2/0% deformation. The Yield locus of un-strained material
initial dimension of the pre-strained panels was 100mm width Tensile and compression tests were performed on in the
and approximately 300mm length and 12mm thickness. longitudinal and the transverse / hoop direction of an X65 pipe.
The yield stresses measured for the different specimens are
reported in Table 1. The larger dispersion observed in the
longitudinal specimen is due to the sampling at different
positions through the thickness (inner diameter, middle
thickness or full thickness). As a consequence, the dispersion of
the results due to anisotropy is considered smaller than the
dispersion due to inhomogeneity of the material in a pipe.

Table 1. Overview of the yield stress on base material pipe in

longitudinal and transvers (hoop) direction
Average Standard Deviation
Tensile Compression Tensile Compression
Direction [MPa] [MPa] [MPa] [MPa]
Longitudinal 562 582 18 13
Transverse 591 571 6 3

Yield locus under different single pre-strain level

The plastic deformation applied during the reeling process
is expected to change the shape of the yield locus. Isotropic
hardening would predict a dilatation of the yield surface, while
kinematic hardening would predict a displacement of the yield
surface. The experimental results of the Bauschinger tests
Figure 6. Pre-straining samples and digital mapping to measure indicate a decrease of the yield stress in the direction opposite
the strain/deformation

4 Copyright © 2015 by ASME

to loading, and an increase in the loading direction with the pre-
deformation level.

After a pre-strain in a given direction, the yield stress is

increased in the perpendicular directions. The yield stress in the
perpendicular directions differs from what a classical kinematic
hardening theory would predict, i.e. the shape of the yield
surface is changed, and it is necessary to include “distortional
hardening” to describe the evolution of the yield locus with the
pre-deformation. This is illustrated in Figure 8, where a tensile
curve after pre-strain in the perpendicular direction is compared
to the results of a Bauschinger test. The stress level achieved
after perpendicular pre-strain is much higher, which indicate the
presence of a cross hardening phenomenon.

800
Engineering Stress (MPa)

600
400
200
0 Figure 9. Yield locus from Bauschinger and various pre-strain
0% 5% 10% 15% 20% level perpendicular loading tests
-200
-400 Yield locus under cyclic pre-strain
Perpendicular
-600 Experiments were performed to evaluate the influence of
Bauschinger cyclic pre-strain on the yield locus. Measurements indicate that
-800 cyclic pre-strains introduce a small dilatation of the yield locus,
Engineering Strain (%) but the main effect on the yield stress in a given direction is
dominated by the direction of the last pre-strain.
Figure 8. Stress-strain curves from Bauschinger and
perpendicular loading pre-strain tests

Increasing the magnitude of the pre-strain decreases further

the yield stress in the opposite direction, and increases the yield
stress in the three other directions. After a pre-deformation in
tensile direction, the yield stress in tensile perpendicular
direction is slightly lower than the yield stress in compression.
This influence of the second loading direction on the evolution
of the yield stress is compatible with the kinematic hardening
theory, but not its magnitude. Similarly, pre-strain in
compression leads to a slightly higher yield stress in tensile
perpendicular direction compared to the compression mode.

Figure 9 shows the yield locus for various pre-strain levels:

2%, 4%, 6% (tension) and -1.7% (compression). In general it is
observed that there is a clear displacement of the yield surface
along the axial direction, indicated by blue arrow for tension
pre-strain and red arrow for compression pre-strain, according
to kinematic hardening rule while in the hoop direction
dilatation of yield surface (isotropic hardening rule), indicated
by green arrow, is more pronounced.
Figure 10. Yield locus from Bauschinger and cyclic pre-strain
perpendicular loading tests

5 Copyright © 2015 by ASME

 Figure 10 shows the yield locus for various cyclic pre- deformation. In brief this material model requires an elasticity
strains: 2%, 2%/0, 2%/0/2% and 2%/0/2%/0. Again, in general law and a plasticity law. The plasticity law is given through a
it is observed that yield surface the axial direction is shifting yield condition and associated flow rule. The change of the
according to kinematic hardening rule while in the hoop shape of the yield surface is introduced by the hardening law.
direction there is a dilatation of yield surface according to
isotropic hardening rule. Theory background of the implemented material model
The elasticity law employed in the material model is the
DISTORTIONAL PLASTICITY MATERIAL MODEL usual isotropic elasticity law. This means that for one has for
the covariant increment in stress:
One is looking for a material model which is able to cope
with cross hardening effects in the material as results from an (1)
arbitrary increase of the applied plastic strain level as well as
the number of cycles of plastic loading. When the material is In this equation is the fourth order elasticity tensor,
plastically loaded in the main loading direction (in the case of which is completely defined through the Young's modulus E
reeling pipeline, the longitudinal/length direction of pipeline) and the Poisson ration ν. The strain increment is provided by
the material will harden (distortion of yield stress) in the and the plastic strain increment is given through . The
perpendicular direction of the main loading direction (in the yield function compares the equivalent stress with the yield
case of reeling pipeline, the transverse/hoop direction of stress.
pipeline).
(2)
Distortional plasticity has been identified already more
than 50 years ago in the experiments of Philips and Gray [5], This equivalent stress is given through the following formula:
and is still under investigation today. In the past, many authors
focused more on the initial shape of the yield function than on
the evolution of the shape. In the first category of models, one √ (3)
will find the classical models of Von Mises, Hill [6], and their
further refinements like the ones of Barlat [7]. Because the In these equations is the deviatoric stress, which is given
associated plasticity does not always allow the anisotropy of through:
strains and stresses to be predicted accurately, a new category
of models was also developed of so-called non-associated (4)
plasticity, see for example [8]. Other categories of model
considers crystal plasticity in order to predict the evolution of where is the fourth order identity tensor and is the
the shape of the yield locus [9-10].The change of shape hydrostatic pressure, given as the average stress:
predicted by crystal plasticity is linked to the rotation of the
crystal axes, and requires therefore large rotations/deformation (5)
to have a significant effect. Korkolis and Kyriakides proposed
an extension of the model of Barlat 2000 where the anisotropy where tr is the trace (sum of the diagonal terms) of the second
parameters are a function of the plastic strain [11]. order .

Different models were developed to describe the effect of Further is the back stress, which evolves as a function of
non-proportional loading. Teodosiu et al developed a model increasing plastic strains. is the initial weighting tensor,
oriented towards sheet metal forming [12]. François presented which for it to be the von Mises stress should start as
a model based on an egg shape for the yield surface of
aluminium [13]. Barlat [14] proposed recently a different
(6)
approach to kinematic hardening, where a pre-deformation in a
given direction trunks the yield surface in the direction opposite
to the loading, introducing an egg-shape similar to the one of with the Kronecker delta function. Finally, is the
François, but not specifically moving the center of the yield evolution of the weighting tensor due to plastic strains.
surface. Levkovitch [15] on the contrary proposed a model
based on the evolution of the anisotropy coefficients, which In order to determine whether yield as occurred, the value
rotates, displaces and modifies the size of the yield surface. of the equivalent stress is compared to the yield stress. The
yield stress is defined as
The last model [15] was chosen for implementation in
ABAQUS 6.12, due its limited number of parameters. The ( ̅ ) (7)
model is based on a “distortional plasticity” which means the
shape of the yield surface is changed by the plastic

6 Copyright © 2015 by ASME

here is a parameter which indicates the influence of pressure
on the yield function, is the initial isotropic yield, is the With the proper parameters, the UMAT was included in a
maximum increment of isotropic yield stress, and indicates full scale FEA model of the Heriot-Watt University (HWU)
how quickly this maximum is reached as a function of bend rig test. A general lay-out of the FEA model is shown in
equivalent plastic strain ̅ . Figure 11.

The flow rule dictates in which direction the plastic strain

evolves:
Straightener former
(8)

where is the plastic multiplier which is solved on a plastic

increment to be coherent with the yield condition.
Test Pipe, 16”x21.4 mm
The hardening rules dictate how the state variables evolve Pinned
as there is plasticity. One has 3 different state variables to take Anchor
into account, these are ̅ , and . These state variables are Reel former end
updated according to the following rules:

̅ √ (9)
Pull end
) (10)
Figure 11. FEA model of the HWU bend rig
) (11) A bend rig test entails of a number of bending cycles that
each consist of four steps:
The parameters in are indicating the rate at which - Bend the pipe against the reel former
the back stress moves into the direction of the plastic strain and - Relax the pipe
which is the equivalent magnitude of the saturated back - Reverse bend the pipe against the straightener former
stress. For the parameters in one has which gives the - Relax the pipe
rate in which the anisotropy develops in the direction of the
plastic strain rate, whereas is that rate in a direction For the verification of the UMAT with the HWU bend rig
perpendicular to it. and give the saturation values for FEA model two variants of the material model were compared:
these two respectively. The other parameters come from the with cross hardening parameters CL, LSAT and kinematic
calculation. hardening (degenerated model). Moreover, the results were
compared with the results of the OMS ovality measurements
FEA verification and results of implementation during actual bend rig testing. The results up to 2 bend cycles
The distortional plasticity hardening material model was are summarized in Figure 12. Note that the initial pipe ovality
implemented as a user material (UMAT) in ABAQUS 6-12. The has been subtracted from the measurement values in order to
UMAT was first tested with a single element FEA model. By compare the results from FEA model without initial pipe
choosing only a selection of the available parameters, the ovality before bending.
model can be “degenerated” to a conventional model with
isotropic or kinematic hardening. The model showed the same The results in Figure 12 show that ovality is considerably
behavior as the corresponding conventional material models reduced by including cross hardening in the material model. In
available in ABAQUS. Also, the results were verified with the the material model, cross hardening is induced by plastic
analytical solutions for the same load cases. deformation. Therefore the effect of reduced ovality is observed
from the end of the first bending cycle and throughout the
The second step was to fit the parameters to the results of second cycle. During the first bending steps the effect is
the Bauschinger and perpendicular loading material tests, using negligible, as expected.
the stress-strain curves and the yield locus at various single
load pre-strain levels as shown in Figure 9. For this, the The FEA results for the material model with cross
parameters CD CL, DSAT and LSAT were added to the set of used hardening match the measured ovality much more closely than
parameters. the (conventional) kinematic hardening model. This is

7 Copyright © 2015 by ASME

especially the case in the second bending cycle, where the error
for the residual ovality is reduced by more than 50%.

Figure 13. Reeling of the 16” test and trial pipe.

Figure 12. Ovality resulting from FEA with and without cross
hardening, compared to measured results.

The FEA results were subsequently compared with the

ovality measurements performed on a similar pipe during the
Spoolbase test and trial as reported below. Small scale material
tests were also performed on a sample of pipe from the
Spoolbase test and trial, in order to compare with the tests
described above.

OFFSHORE SPOOLBASE OVALITY MEASUREMENTS

Figure 14. Test section before passing the level winder roller.
Spoolbase test and trial included spooling of two ~650 m
stalks of 16”x23.8mm pipe and a number of ovality A FEA model was created to simulate the test and trial test
measurements were done. During the test, a relatively short
sequence. A long pipe stalk consisting of beam-elements was
pipe section of around 40 m was repeatedly spooled on and off.
connected to a short (12 m) detailed shell-element section. The
A total of five (5) bend cycles was performed to simulate a full
pipe was spooled on and off 5 times on a reel, supported by
reeling cycle, including a contingency scenario. After final un-
rollers. Level winder rollers provide the required shear force
spooling, a 3 m section was cut out of the strain cycled pipe and resulting bending moment in the pipe while spooling on.
section and a material test program was performed to
An overview of the model is shown in Figure 15.
investigate the effect of cyclic plastic strain on the material
properties.

Pipe ovality measurements

First, the pipe was spooled onto the reel the (Figure 13)
and spooling off was done until a marked section just passed
the level winder rollers (Figure 14) and thereby a section of Reel
pipe was straightened.

Ovality measurements were done at the end of each bend

cycle after pipe straightening. An additional measurement was Level winder rollers Detailed pipe section
done after the cycled pipe section was cut out, before the
material was tested at the OCAS laboratory in Belgium. Note
that this last measurement was therefore done without back Figure 15. Spoolbase test and trial FEA model.
tension.

8 Copyright © 2015 by ASME

 In Figure 16 the spool base ovality measurements are
compared with the results of the spool base FEA model with the
cross hardening material model. Also the results of prediction
of bend rig ovality at HWU according to the HMC modified
DNV pure bending equation as described in Ref.[1] have been
added for the same pipe diameter and wall thickness.

Figure 17. distance between the level winder rollers and the reel.

On the HWU bend rig, the shear forces are introduced by

the touchdown-point of the pipe on the reel former and the pull
wire at the pull head (Figure 18). The distance between the two
varies during bending the pipe, but the approximate distance
between the pull wire and the area around the weld (where
ovality measurements are usually performed) is 8-9 m.

Figure 16. Residual ovality measured at the Spoolbase compared

with an analytical prediction based on ovality measured at HWU.

The results in Figure 16 show that for the first 2 cycles the
measured Spool base ovality is in a good agreement with both
the predicted ovality from spool base FEA model and the
predicted bend rig ovality at HWU.

The difference with the spool base FEA model is observed

after 2 bend cycles. One of the reason is because of the cross Figure 18. HWU bend rig lay out.
hardening parameters CL, LSAT used in the analysis are defined
by fitting the parameters to the results of the Bauschinger and The required pipe bending moment is defined by the
perpendicular loading material tests which are limited only up plastic moment of the pipe and this is a property of the cross
to 2 bend cycles. The cross hardening behavior between 3 and 5 section. When the distance between the shear force introduction
bend cycles is not included in the definition of hardening points is smaller, the moment arm is shorter and the shear force
parameters as no perpendicular loading test data is available in the pipe must be larger to generate the required bending
after 2 cycles. As the predicted ovality is further diverging from moment. Therefore a larger shear force is required for bending
the measured ovality, it gives an indication that the cross pipe on the HWU rig, which results in a larger contact pressure
hardening factor between cycle 3 and 5 is increasing and larger between the pipe and the bend rig formers and results in greater
than for the first two cycles. pipe ovality.
The HMC modified DNV pure bending equation, which is Removal of the back tension has a marginal effect on the
derived based on HWU bend rig test results [1], gives a good residual ovality after five bend cycles, as the last measured
prediction of the actual spool base ovality and is somewhat ovality in Figure 16 shows.
conservative. The difference may be caused by the relatively
large distance between the position of the level winder rollers OFFSHORE SPOOLBASE MATERIAL TESTING
and the touch down point of the pipe on the reel. At these two
points a shear force is introduced into the pipe. The touch down At the end of the 16” pipe test and trial, a 3 m section
point lies a bit beyond the centerline of the reel and the distance (Figure 19) was cut out of the ~40 m cycled (i.e. 5 times
to the level winder rollers is therefore greater than 14.5 m (see spooled on and off) pipe section for material testing. The
Figure 17). material test program was designed to find the yield stress in

9 Copyright © 2015 by ASME

tension and compression throughout the pipe cross section, both
in axial and in hoop direction.

Figure 19. Test section on the reel. Figure 21. Yield stress reeled pipe (5 bending cycles).

The material testing was performed at OCAS in Ghent, The results in figure 20 show an increase of the yield stress
Belgium. Sample plates were machined from a section of in tension in longitudinal direction at the 3 o’clock position and
cycled pipe, at four clock positions as defined in Figure 20. The a reduction at the 9 o’clock position, with respect to the values
3 o’clock position is the side of the pipe contacting the reel, at the 6 and 12 o’clock positions. For the longitudinal
while the 9 o’clock position contacts the level winder rollers. compression yield stress the results are the other way around.
The material shows mostly kinematic hardening in the
longitudinal direction. When the last loading of the material is
in tension, the material is hardened in tension and softened in
compression. When the last loading of the material is in
compression, it’s the other way around.

In transverse direction, the yield stress has increased both

in tension and in compression at clock positions 3 and 9. At
clock position 3 the increase of the compression yield stress is
more prominent, while at clock position 9 the increase of the
tension yield stress is greater. In transverse (hoop) direction the
material shows mostly isotropic hardening, independent of the
Figure 20. Definition of the clock positions. sign of the last loading in the longitudinal direction. However,
At each clock position a number of test specimens were when the last loading in the main loading direction was in
machined and tested. Test specimens were tested in tension, more hardening is observed in compression in hoop
compression and tension, both in longitudinal and transverse direction than in tension. When the last loading in the main
(hoop) direction. loading direction was in compression, it’s the other way around.
While isotropic hardening in hoop direction is dominant, some
The results are summarized in a graph, see Figure 21. The kinematic hardening is observed as well.
dotted lines represent the average values of the longitudinal and
transverse directions for clock positions 6 and 12. The material testing of samples cut out after spool and un-
spool cycles confirms the distortional plasticity hardening
The material at the 3 o’clock position was plastically behavior as given in Figure 10 from the small scale tests
deformed (strained) in tension during the last loading, and the Bauschinger and perpendicular loading with cyclic pre-
material at the 9 o’clock position in compression. The material straining.
at the 6 and 12 o’clock positions is located at the bending
neutral axis and was not plastically deformed (un-strained). For CONCLUSIONS
these clock positions the yield stress in the longitudinal and
transverse directions are practically the same, so the material is HMC has performed a large research program in order to
initially isotropic. understand the evolution of the pipe ovality during reeling.
Large scale tests were performed to measure the increase of the

10 Copyright © 2015 by ASME

ovality with the number of cycles of bending / unbending. The REFERENCES
ovality measurements are taken during the full scale bend tests
at HWU and Spool base test and trial program. When these [1] Karjadi, E., Smienk, H., Boyd, H., Aamlid, O., July 2012,
tests were simulated with FEA model with conventional “Extend Reel-ability of HMC New Aegir Reeling Vessel
isotropic or kinematic hardening, both models predicted an Based on Reliability Based Assessment and Bending Tests
ovality larger than that measured experimentally. To improve Program”, OMAE2012-83061.
the prediction model, some efforts have been made to [2] Karjadi, E., Boyd, H., Van Rooijen, R., Demmink, H.,
understand the change of pipeline material characteristics Balder, T., June 2013, “Development on Aegir Reeling
during reeling by performing a series of small scale material Pipeline Analyses by Test Validation”, OMAE2013-
tests program including the Bauschinger and perpendicular 10365.
loading tests. [3] Smienk, H., Karjadi, E., Vasquez, G., Doherty, P., Dooley,
P., June 2013,” DCV Aegir Pipelay Installation Analyses
The test results show that the shape of the yield surface is and Capabilities”, OMAE2013-10262.
changed by plastic deformation for a different strain level and [4] ABAQUS 6.12 User’s Manual.
number of cyclic plastic strains. With the increase of strain [5] Philips, A., Gray, G.A., “Experimental investigation of
level and/or number of plastic straining cycles, the yield stress corners in the yield surface”, J of Basic Engineering Trans
in the hoop direction of the pipe is increasing and higher than ASME 1961 275-288
the yield stress in the longitudinal direction of the pipe. This so- [6] Hill, R., “A theory of the yielding and plastic flow of
called cross-hardening effect, which is not taken into account in anisotropic metals”, Proc. Roy. Soc. London, 193:281–297
conventional hardening model, gives a reduction of the increase [7] Barlat, F., Lege, D.J., Brem, J.C., “A six component yield
of ovality with a number of plastic cycles as measured during function for anisotropic metals”, Int J Plasticity (1991)
the full-scale bend tests and Spool-base test and trial program. 693-712
[8] Safaei, M., Zang, S., Lee, M.G., De Waele W.,
The cross hardening characteristics of material have been “Evaluation of anisotropic constitutive models: Mixed
modeled using the “distortional plasticity” principle and anisotropic hardening and non-associated flow rule
implemented in a user subroutine of an FE analyses software. approach”, Int J Mechnical Sciences (2013) 53-68
The implementation of the user subroutine has been validated [9] Holmedal, B., Van Houtte, P., An, Y., “A crystal plasticity
by the ovality measurements taken during the bend test at model for strain-path changes in metals”, Int J Plasticity
HWU. It is shown that by including the cross hardening (2008) 1360-1379
parameters the error of the residual ovality prediction is [10] Eyckens, P., Van Bael, A., Moerman,J., Vegter, H., Van
reduced by more than 50% after the second bending cycle. The Houtte, P., “Prediction of transient hardening after strain
cross hardening characteristics as identified by small scale path change by a multi-scale crystal plasticity model with
testing has been confirmed by the material testing of samples anisotropic grain substructure”, 11th Int Conf Techn of
cut-out from Spoolbase test and trial program after completing Plasticity ICTP 2014 Procedia Engineering (2014) p1318-
5 bend cycles. 1323
[11] Korkolis, Y., Kyriakides, S., “Inflation and burst of
The cross hardening user subroutine can at this time only aluminium tubes. Part II: An advanced yield function
be used to predict more accurate residual ovality after the including deformation-induced anisotropy”, Int J Plasticity
second bend cycle. More testing would be required to predict (2008) 1625-1637
ovality after further bend cycles. [12] Haddadi, H., Bouvier, S., Banu, M., Maier, C., Teodosiu,
C., “Towards an accurate description of the anisotropic
It is also shown that the modified DNV pure bending behavior of sheet metals under large plastic deformations:
equation as developed by HMC in Ref.[1] can give also a close modelling, numerical analysis and identification”, Int J
ovality predictions with respect to the predictions base on the Plasticity (2006) 2226-2271
FEA model with the cross hardening. This analytical tool can [13] François, M., “A plasticity model with yield surface
be used for a quick first prediction of the residual ovality of distortion for non-proportional loading”, Int J Plasticity
pipes with uniform properties. However, when a variation of 2001 703-717
the pipe stiffness due to wall thickness and yield stress variation [14] Barlat, F., Brem, J., Yoon, J.W., Chung, R.E., et.al., “Plane
has to be included in the analysis, the FEA model with the stress function for aluminium alloy sheets – part I: theory”,
developed cross hardening material model as presented in this Int J Plasticity 19 (2003) 1297-1319
paper is deemed necessary to give more accurate prediction of [15] Levkovitch, V., Svendsen, B., Aydin, M., Kessler, L.,
the ovality including its variation along the length of the pipe “Accurate Hardening Modeling as Basis for the Realistic
due to the pipe stiffness variation. Simulation of Sheet Forming Processes with Complex
Strain Path Changes”, LS-DYNA Anwendforum, 2007

11 Copyright © 2015 by ASME

View publication stats