You are on page 1of 7

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

OMAE2015-41816

EXPERIMENTAL AND NUMERICAL STUDY OF THE BENDING BEHAVIOR OF A


FLEXIBLE RISER MODEL

MT Rahmati S. Norouzi
Department of Mechanical, Department of Mechanical,
Aerospace and Civil Engineering Aerospace and Civil Engineering
Brunel University London Brunel University London
Uxbridge Uxbridge
UB8 3PH, UK UB8 3PH, UK

H. Bahai G. Alfano
Department of Mechanical, Department of Mechanical,
Aerospace and Civil Engineering Aerospace and Civil Engineering
Brunel University London Brunel University London
Uxbridge Uxbridge
UB8 3PH, UK UB8 3PH, UK

ABSTRACT deep waters in areas such as Gulf of Mexico, southern shores of


Unbonded flexible risers have become the main means of Caspian Sea, Norwegian Sea and Greenland waters has become
extracting hydro carbonates from deep waters. So, inevitable. This competition has led to the rapid development of
understanding the complex structural integrity of flexible risers flexible risers capable of functioning in a harsh environment in
has become a crucial issue for the offshore industry. In this deeper waters often exceeding 3km below sea level. Flexible
paper, an experimental test and a detailed finite element risers consist of several polymer and steel layers that are, to a
analyses were carried out on a scaled down model of a flexible certain extent, free to move internally relative to each other.
riser pipe in order to understand its bending behavior. The This gives low bending stiffness and makes them highly
model used consists of four layers which include two valuable tools for subsea oil and gas companies. Their ability of
cylindrical polycarbonate tubes and two steel helical layers. withstanding both horizontal and vertical large displacements
One helical layer, wounded around the pipe assembly, made them ideal for floating platforms.
represents the carcass layer in an actual flexible riser whilst the
other represents the riser tendon armour layers. The model was Although the importance of using flexible risers for subsea
subjected to a three point bending load in order to study its technology in offshore oil and gas production is well known,
bending-curvature behavior. The test data was then compared the complex structural behavior of flexible risers is not
with the FE numerical results which predicted a similar non- sufficiently understood for many design and development
linear trend but under predicts the strain at tendon layers. purposes. On the one hand flexible risers must be reliable
enough to safeguard the environment. On the other hand, they
must make the exploitation of the subsea hydrocarbons
INTRODUCTION economically feasible. Despite the fact that many attempts have
been made to develop highly reliable and economical flexible
In recent decades, as global oil demand has been rising, oil risers, they are not meeting their recommended service life.
exploration and production has been progressing into deeper This is because the basic technology compared to steel pipes, is
water. As a result, industrial competition for designing efficient relatively new. It has not been evolved yet to support more
and reliable facilities to reach the hydro-carbonates in ultra-

1 Copyright 2015 by ASME

Downloaded From: http://asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/86132/ on 01/20/2017 Terms of Use: http://www.asme.org/about-asme/term


challenging application for deeper waters and harsher M Bending moment
environment [1]. P Bending load
Ri Inner radius
Due to the complex geometry of flexible risers, the Ro Outer radius
conventional stress prediction and fatigue analysis tools based t Thickness
on analytical formulations and linear methods are not adequate. x Horizontal direction at bending
Instead, detailed non-linear finite element methods and k Bending curvature
experimental methods are required to take into account the y Vertical direction at bending
contact and friction between the layers, radial contraction and Strain
interlayer detachment. One of the first FE methods for Stress
analysing flexible risers was presented by Felippa and Chung y Yield stress
[2]. They described the formulation and algorithmic
implementation of a geometrically nonlinear beam model.
EXPERIMENT
Leroy et al. [3] used an explicit FE model to predict the stress
in flexible pipe layers. The model dealt with stress and A prototype of flexible pipe was tested by applying
displacement calculation in dynamically bent unbounded variable bending loads to check the nonlinear behaviour of the
flexible pipes. The movements and stresses of all helical layers specimen under bending. The prototype consists of one carcass,
were coupled. The model was cross validated on a dedicated one helical tendon layer, and two tubes as outer sheath and anti-
case study that consisted of a pressurized pipe cyclically bent wear layers. A schematic view of the model is shown in figure
with constant curvature. 1. Details of the material, dimensions and arrangement of
constituent layers are given in Tables 1 and 2. The outer tube
was designed similar to the outer sheath layer in real flexible
Svik [4] compared stresses of bending FE models with risers and covered all other layers of the prototype. The inner
experimental data for a flexible pipe. In this multi-layer tube was designed in such a way to seal the pressurized oil
method all components were modelled independently. Good inside it. This layer was placed inside the tendon and covered
agreement between the FE models and experiment in terms of the carcass layer. A single helical armour (or tendon) was
the curvatures were obtained. Svik [5] also presented a finite designed which wounded around the inner cylinder. The helical
element model to predict stresses under axisymmetric loading carcass layer is similar to the tendon layer but with a very short
as well as two equations to predict bending stresses in tendon pitch length.
layers in unbonded flexible risers. An experiment was also
carried out to compare the bending stresses. The axisymmetric Several other components have been designed in order to
model was used to analyze the boundary conditions with keep the layers in right place, to seal the prototype and to apply
regards to initial contact pressure in both formulations. the external load. These additional components have listed
below:
End-caps: were designed to keep the layers of the test
Many commercial software are now available for analyzing specimen in a fixed position and to apply load; they
flexible pipes, such as ABAQUS Adina, ALGORA, ANSA, were installed at each end of the specimen.
ANSYS, LS DYNA, LUSAS, and NASTRAN. It is essential to
determine the degree to which these numerical codes represent Threaded rod: was designed to keep the end-caps in a
the physical reality of the riser model by conducting an fixed position.
experiment. In this paper an experimental study on a simplified
prototype of a flexible pipe with only four layers under bending
moment is described. A simplified model is chosen to better Load component: It was designed to apply the external
understand the bending behavior and, in particular, the load. It was installed at the middle of outer sheath.
frictional contact between the layers. Then, the capability of a
finite element model developed in ANSYS for prediction of the
structural behavior of this model is examined. Bending bed: was designed to keep the end-caps in a
fixed position.

NOMENCLATURE
E Module of elasticity
I Area moment inertia
L Length of beam

2 Copyright 2015 by ASME

Downloaded From: http://asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/86132/ on 01/20/2017 Terms of Use: http://www.asme.org/about-asme/term


Table 2 Material data for each layer

Layer Material E (GPa) y (MPa) (kg/m3)


Carcass Steel AISI1018 200 0.3 740 7860
Inner
Polycarbonate 2 0.37 65 1500
tube
Tendon Steel AISI1018 200 0.3 740 7860
Outer
Polycarbonate 2 0.37 65 1500
tube

Figure 1 schematic assembly of prototype layers

Table 1 Prototype dimensions


Layer Ri Rm t (mm) No. of Length
(mm) (mm) Pitches (mm)
Carcass 44 45.5 3 15 651
Inner
47 49.5 3 - 667.8
tube
Tendon 50 52.5 5 6 651
Outer Figure 3 specimen for bending test
55 57.5 5 - 681
tube

The rig used to investigate the non-linear behaviour caused


by interlayer slip during the bending test is shown in Figures 2
and 3. Five strain gauges were used at the positions shown in
Figure 4. All of the strain gauges were calibrated before each
test.

Figure 4 strain gauge positions

The yield load corresponding to first yield in the


Figure 2 schematic assembly of bending rig specimen was calculated. This is to ensure that during the
(dimension in mm) experiment the materials of flexible riser remain elastic with no
plastic deformation. The yield load was found to be 16.88kN
so, in order to study the inter-layer non-linear effects without
reaching yield point, the maximum load used for bending test

3 Copyright 2015 by ASME

Downloaded From: http://asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/86132/ on 01/20/2017 Terms of Use: http://www.asme.org/about-asme/term


was chosen as 13KN. The value of curvature at various NUMERICAL SETUP
bending moment is calculated from:
The dimension of the model, the components and details of
2 the material are the same is the experimental model shown in
2
K=
3 (1) figure 1 and table 1 and 2. The numerical model used and the
[1+( )2 ]2
grid are shown in figure 6.

In order to use equation 1 for measuring curvature at each load


increment, seven displacement transducers were placed under
the specimen, each with accuracy of one hundredth of
millimetre. The spacing between the displacement transducers
was 65.1 mm (Figure 5). The bending load was increased (for
unloading decreased) in 100 steps with the load increment of
250N.
The results of the experiments are plotted in bending-
moment curvature diagrams.

Figure 6 Mesh of the model (6500 elements)

An eight-noded structural element, SOLID185, for solid


load component was used. The element has three degrees of
freedom at each node [6]. For the riser layers, a four-node shell
element SHELL181 with full integration was used. In this
element there are two assumptions:
The normal vectors to the centre-plane remain straight after
deformation, yet they do not have to remain perpendicular
to centre-plane.
Every integration points set of any layer in radial direction
are similar in terms of element (or material) orientation.
Figure 5 displacement transducers positions
To keep the load component in a vertical direction for
either loading or unloading tests a frictionless support was
The calculation of curvature for each load is carried out in required. So, SURF154 element [6], a three dimensional
four steps: structural surface with four nodes were used. In total, 6500
First, for each load, the seven deflection values shown elements were used in the meshing process for the riser model
by displacement transducers were measured and in ANSYS.
plotted.
A three dimensional surface-to-surface contact algorithm
Then, their approximate second order polynomial was used both between layers and between the outer tube and
trend-line curve of the plotted data is estimated. loading part. So, a contact modelling technique is required to
take into account the contact effects. A penalty method with
The first and second derivation of the displacement Gauss integration points was used to model contact in ANSYS
curve were calculated, Workbench [6]. Coefficient of friction between layers is
assumed as 0.2. For contact modelling two types of elements
Finally equation 1 is used to calculate the curvature in have to be defined contact and target. These elements, which
the middle of specimen. are defined automatically by ANSYS, are CONTA174 and
TARGET 170. As shown in Figure 7, deformable solid
The results for bending are compared with numerical materials are normally defined as contact whereas rigid bodies
models in the next section. are defined as targets. In the riser model, tubes were considered
as contact, while helical carcass, helical tendon, and load part
modelled as target.

4 Copyright 2015 by ASME

Downloaded From: http://asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/86132/ on 01/20/2017 Terms of Use: http://www.asme.org/about-asme/term


during loading. However, the trend is not captured well during
unloading. By increasing the load, friction is overcome and
inter-layer slip occurs, resulting in the non-linear behaviour in
bending-moment curvature. There is no material non-linearity
as the maximum load used is less than the first yield load. So,
the non-linear trend shown in Figure 10 is only caused by inter-
layer frictional slipping. The numerical model over predicts the
bending moment during loading but under predicts it during
unloading. However, the trend captured by the numerical model
is very close to the experimental results. The numerical and
experimental local strain in axial direction and transverse
direction at three strain gauges on the tendon is shown in
Figures 11 and 12. There is a very good agreement between the
numerical and experimental axial strain. There is a difference in
transverse strain although the trend is well captured by the
Figure 7 Contact detection point at Gauss point [6] numerical model. This could be due to the numerical
inaccuracies related to the contact modelling used in this study.
The strain-load relationship show that at some points when
For the analysing of the riser, a loading force was applied loads surpass some values, the strains decrease. Also, when the
at the support which is increased by 250N increment value to loads reduce to zero, there is a non-zero residual strain. This
reach 13000N (Figure 8). For unloading test analysis this value phenomenon observed in both experimental and numerical
decreased to zero with the same increment. To keep the loaded results could be due to the frictional slipping between the
part in its position, a frictionless support was modelled through layers.
four surfaces surrounding the load component (blue surfaces in
figure 8). As for the boundary conditions, one side of the riser
was modelled as fixed support whilst the other side is allowed
to slide slightly in axial direction (see D, E, F, and G signs in
Figure 8). The nonlinear solution scheme was fully implicit.

Figure 9 Deformed prototype (m)

Numerical Experiment
1100

Figure 8 ANSYS modelled load, frictionless support, 880


and boundary conditions
Moment (N.m)

660

RESULTS
440

220
The displacement contour plot predicted by the numerical
model is shown in Figure 9. The moment-curvature diagram is 0
0 0.02 0.04 0.06 0.08 0.1
shown in Figure 10. When the bending moment is small (less
than 200 N.m) a linear behaviour is predicted by the Curvature (1/m)
experimental model because there is no slip between specimen. Figure 10 Moment Curvature [Bending]
The same linear relationship is shown in the numerical model

5 Copyright 2015 by ASME

Downloaded From: http://asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/86132/ on 01/20/2017 Terms of Use: http://www.asme.org/about-asme/term


Numerical Experiment Numerical Experiment
80 80

26
Micro strain

Micro strain
-20
-28

-82
-120

-136

-190
-220
0 3250 6500 9750 13000
0 3250 6500 9750 13000

Load (N) Load (N)


a) a)

Numerical Experiment Numerical Experiment


200
100

-100 -200

Micro strain
Micro strain

-300 -600

-500 -1000

-700 -1400
0 3250 6500 9750 13000 0 3250 6500 9750 13000

Load (N) Load (N)


b) b)

Numerical Experiment Numerical Experiment


115 100

50
Micro strain

67
Micro strain

19 0

-29 -50

-77 -100

-125 -150
0 3250 6500 9750 13000 0 3250 6500 9750 13000

Load (N) Load (N)

c) c)

Figure 11 Strain-Load (axial) curves at: a)SG1 b) SG2 c) SG3 Figure 12 Strain-Load (transvers) curves at: a)SG1 b) SG2 c) SG3

6 Copyright 2015 by ASME

Downloaded From: http://asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/86132/ on 01/20/2017 Terms of Use: http://www.asme.org/about-asme/term


CONCLUSIONS

A simplified prototype of a flexible pipe was tested both


numerically and experimentally by applying loading and
unloading bending loads. The bending moment-curvature
relation predicted by the numerical model captures several
aspects of the experimentally measured one both qualitatively
and, to a certain extent, quantitatively. On the other hand, the
amount of hysteresis predicted by the numerical model is
somewhat lower than the measured one. Good agreement was
found between the numerically determined axial strains and the
measured ones at most locations. For the transversal strains
measured at the same locations, while good agreement is found
in some parts of the curves some discrepancy is found in others.
This may be attributable to some inaccuracy in the modelling
contact in the present model and suggests that further work is
needed. Finally, it should be noted that the current calculations
and experiments were carried out on a simplified riser
prototype with only four layers with a few strain sensors. For
future work, extra strain sensors on different locations would be
useful to better understand the non-linear effects of slip
between the layers.

ACKNOWLEDGMENTS
The authors gratefully acknowledge Mr. Keith Withers,
whose guidance and ideas helped to improve the experiments.

REFERENCES
[1] PSA, 2013, Unbounded Flexible Risers - Recent Field
Experience and Actions for Increased Robustness, Norway
PSA report.

[2] Felippa, C.A., Chung, J.S., 1981, Nonlinear static


analysis of deep ocean mining pipe Part I, Modeling and
Formulation, Journal of Energy Resources technology, Vol.
103, pp. 11-16.

[3] Leroy, J.M., Estrier, P., et al, T.P., 2010, Stress


assessment in armour layers of flexible risers, Proceedings of
the ASME 2010 29th International Conference on Offshore,
Ocean and Arctic Engineering, OMAE2010-20932.

[4] Svik, S., 2010, Comparison between theoretical and


experimental flexible pipe bending stresses, Proceedings of
the ASME 2010 29th International Conference on Offshore,
Ocean and Arctic Engineering, OMAE2010-20352.

[5] Svik, S., 2011, Theoretical and experimental studies


of stresses in flexible pipes, Computers and Structures, 89, pp.
2273-2291.

[6] [6] ANSYS Academic Research, Release 13.0, help


viewer, Static structural, ANSYS, Inc. 2012.

7 Copyright 2015 by ASME

Downloaded From: http://asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/asmep/86132/ on 01/20/2017 Terms of Use: http://www.asme.org/about-asme/term

You might also like