4 GENERAL PURPOSE FINITE ELEMENT ANALYSIS

4.1 Static Analysis

The modal analysis carried out by the SPFEA solver is based on a static equilibrium
configuration determined by a preceding static analysis. The static analysis is often referred to as
a bottom roughness analysis in the pipeline engineering community, and the analysis
methodology was outlined in Section 2.2. In the present work, the bottom roughness analyses are
performed using the general purpose finite element analysis (GPFEA) tool Abaqus [2012]. The
Abaqus modeling will be described in more detail in the following.
The pipeline is modeled using PIPE31H elements. PIPE31H elements are first order shear
deformable 3D hybrid beam elements based on Timoshenko beam theory [Shames and Dym,
1991], with an additional formulation to distinguish between effective and true wall axial forces.
The hybrid property (indicated by the H in the element name) implies that the elements use a
formulation in which the axial and transverse shear forces in the elements are included as primary
variables, in addition to the nodal displacements and rotations. The hybrid beam element
formulation is advantageous in geometrically nonlinear analysis for beams that undergo very
large rotations, but are quite rigid in axial and transverse shear deformation (e.g., for offshore
pipelines and cables). Although hybrid beam elements are computationally more expensive, they
generally converge much faster when the beam's rotations are large, thereby being more efficient
overall in such cases [Abaqus, 2012].
The PIPE-element library in Abaqus is assigned a beam section of type pipe by default,
implying that the cross-section is modeled as a single-layer hollow cylinder. Coating layers, i.e.
thermal insulation, paint, adhesives, corrosion protection, concrete coating or combinations
thereof, are assumed not to contribute with any stiffness. However, their impact on the dry mass
and buoyancy are included in the models.
The seabed is modeled using the surface elements R3D4. The seabed elements are
generated based on the survey data input, and they are assigned nodal coordinates corresponding
to their actual locations. A contact pair is generated between the pipe elements and the seabed
surface, and the contact is modeled using normal and tangential stiffness based on a static vertical
soil stiffness coefficient and friction coefficients for the lateral and axial directions. Element
resolution is taken equal to the survey data resolution (typically ~1 m in axial direction) for both

24
the pipeline and seabed elements. Having equal resolution enhances analysis stability since initial
contact between the surfaces is node-to-element boundary. Typical model lengths are between 6
and 24 km. Pipelines that are longer than the indicated model length interval are sectioned. In
order to ensure proper boundary behavior between pipe sections, the sections overlap. Suitable
overlap lengths are determined by requiring convergence of the effective axial force between the
overlapping sections.
In Section 2.2, it was described that the outcome of the static analysis depends on the load
history and that the static equilibrium configuration should be determined for each distinct phase
of the pipelines design life (e.g., for as-laid conditions, water-filled conditions and operational
conditions). For this reason, the analysis is typically divided into a series of load steps that are
meant to represent the various phases that a pipeline goes through. A typical bottom roughness
analysis comprises the following steps:

1. Gravity and buoyancy are introduced.

2. The pipe is laid down on the seabed.
3. The effective lay tension is applied.
4. Seabed friction is activated.
5. The pipe is filled with water.
6. Pressure is increased to model the pressure test.
7. Pressure is removed to model the end of the pressure test.
8. Operational temperature, pressure and content are introduced.
9. A number of shut-down cycles may be modeled, in which operational temperature and
pressure are removed and re-introduced.

Initially, the pipe elements are generated in a straight line with z-coordinate equal to the
highest point on the seabed, as visualized in Figure 5.

25
Figure 5 Seabed and initial pipeline configuration prior to introduction of gravity and buoyancy.

Subsequently, gravity and buoyancy are introduced to the model, and the pipe is laid down
on the seabed making use of stabilizing measures as detailed by Aamlid and Rneid [2008]. After
contact has been established between the pipe and the seabed, friction behavior between the pipe
and seabed is deactivated. One end node is fixated, and the effective lay tension is applied at the
other. When the effective lay tension is approximately constant over the length of the pipe, the
seabed friction behavior is reactivated and the pipe is regarded as in the as-laid condition. An
example of the nearly uniform effective axial force distribution in the as-laid condition is shown
in Figure 6.
In Figure 6, the pipeline is shown resting on the seabed. It is observed from the figure that
the effective axial force varies approximately 2.5% about the mean value of 200 kN. The value
of 200 kN was the input effective lay tension to the analysis, which in this case illustrates that the
methodology is fairly accurate in establishing the as-laid condition.

26
Figure 6 Pipeline in as-laid condition, showing a nearly uniform effective axial force distribution.

After the as-laid condition, the pipe is filled with water. The water-filling process is
conducted in order to perform the system pressure test, which is the next step in the analysis. As
indicated in Section 2.2, it is not necessary to perform modal analysis for the system pressure test
condition (because its duration is very short and may thus be disregarded with respect to fatigue
utilization). However, the pipe-soil contact and axial feed-in into free spans are non-linear
effects, which means the system pressure test condition may influence on the static configuration
in the operational condition. Therefore, it is included in the static analyses even if not directly
relevant as a separate stage in the dynamic analyses.
Finally, the operational condition is modeled by introducing operational temperature,
content weight and pressure. The area previously shown in Figure 6 is shown again in Figure 7
for the operational condition. As expected from Eq. (2), the effective axial force becomes
compressive in this phase.

27
Figure 7 Pipeline in operational condition, showing compressive effective axial forces.

In all the steps of the static analysis, the NLGEOM option in Abaqus is turned on, implying
that geometric non-linearities are accounted for. Thus, the elements are, in each increment of
each step of the analysis, based on the current deformed configuration using current nodal
positions. Consequently, element lengths are continuously updated, and changes in the effective
axial force as a result of static deflections are accounted for.

4.2 Modal Analysis

The SPFEA solver presented in Section 3 will be validated by comparison to modal

analyses performed with the GPFEA software Abaqus [2012]. The modal analysis procedure in
Abaqus will be described in the following.
When the static configuration has been determined according to the analysis methodology
described in the preceding section, contact information is extracted from Abaqus for every
pipeline node, along with nodal coordinates and the effective axial force in each pipe element. In
Abaqus, whenever a contact pair has been established (such as the contact between the pipe
surface and the seabed in the present context), the nodal variable COPEN contains the contact
opening at the surface nodes. Thus, by extracting the value of COPEN for each pipeline node, the
gap between the pipe and the seabed is obtained. A negative value for COPEN implies that the

28
pipe has penetrated the seabed, while a positive value indicates that the pipe is in a span. By
inspection of the static pipe configuration and the distribution of corresponding gaps, the
modeled pipe is divided into relevant multi-span sections, as described in Section 2.4. An
example of a multi-span section and a corresponding gap distribution is shown in Figure 8.

-101.5

Seabed
-102
pipeline
Water depth (m)

-102.5

-103

-103.5

-104

-104.5

0.5

0.4

0.3
Gap (m)

0.2

0.1

KP position (m)
-0.1
15400 15500 15600 15700 15800 15900 16000 16100 16200 16300

Figure 8 Static pipeline configuration (above) and gap between the seabed and the pipe (below).

29
 For each multi-span section, a new contact pair is defined between the pipe and the seabed
surface elements, and intermediate pipe sections between multi-span sections are also assigned
individual contact pair definitions. The pipe elements are sectioned into new element sets,
corresponding to the new contact pair definitions. Sets of dynamic springs to model axial, lateral
and vertical pipe-soil stiffness are defined, with stiffness coefficients taken according to DNV-
RP-F105 [2006]. In vertical direction, the static vertical stiffness between the pipe and the seabed
must be subtracted from the vertical dynamic spring stiffness. The new dynamic spring nodes and
contact pair definitions, which are introduced based on the contact results from the static
analyses, are not permissible to introduce using the restart functionality in Abaqus. Hence, the
full static analysis must be re-run including the new dynamic soil stiffness springs and contact
pair definitions. Once the dynamic springs have been defined in the second static analysis, they
are deactivated until they are reintroduced prior to performing the relevant modal analysis.
After completing the second static analysis, modal analyses may be conducted for each
multi-span section, in each phase, by restarting the analysis in the relevant phase. There are two
individual steps in the modal analysis:

1. Vertical, axial and lateral dynamic springs are reactivated in the model (strain free). At
each end of the pipeline element set corresponding to the relevant multi-span section, the
nodes are fixed in translational degrees of freedom (pinned) and in rotation around the
pipe axis. Friction is deactivated in axial and lateral directions. All pipe element sets and
contact pairs which are not associated with the relevant multi-span section are removed
from the model.
2. Pipe added mass is included and modal analysis is conducted.

30