You are on page 1of 12

Global to Local Model Interface for Deepwater Top

Tension Risers
Mateusz Podskarbi
Karan Kakar
2H Offshore Inc, Houston, TX
Abstract
The water depths from which oil and gas are being produced are reaching several thousand feet. Deep
water risers are pipe structures that extend from the floating vessels to the bottom of the sea. In order to
capture the time dependent, geometrically non-linear response of the global system with the complex multicomponent response of local details and properly analyze the riser and its interaction with subsea
equipment and vessels it is necessary to use several different models. A global model is used to analyze the
global response of the riser to various environmental conditions and local models to analyze the interface
between the riser components. Global models usually employ finite elements such as nonlinear (large
deflection) beam or pipe elements with highly simplified boundary conditions (fixed, pinned, springs, etc.)
Local models are usually built of 3D solid elements with much more detailed boundary conditions, often
using contact elements, to realistically predict local response of the particular components. In this paper
the methodology of analytical interfacing and data transmission between the local and global models is
described. The example of the interface between the global and local model for the deep water top tension
risers is included.

Introduction
The water depths from which oil and gas is being produced are reaching several thousands feet. Deep
water risers convey fluids such as oil, gas or water between the seabed and the surface vessel. There are
many different configurations that are used, depending on application, vessel type and environmental
conditions. The most typical are SCRs (Simple Catenary Risers), TTRs (Top Tension Risers) and free
standing risers connected to the vessel with flexible jumpers (Reference 1). At the first sight, the riser
appears to be a simple structure, but in reality, its response is very complicated. The global response is
time dependant and geometrically nonlinear. To capture the local response of the numerous interfaces,
detailed modeling of these areas is necessary. The complexity of both global response and detailed
components leads to the approach that employs several models of different level of details for various steps
of analysis. Global models are usually modeled using pipe elements with simplified boundary conditions.
Global analysis results are then used as the input data for further analysis of the sub-components. The
paper describes the multiple steps of the analysis and interfacing between the various models and provides
the means to verify the correctness of each step. As an example, the modeling and the analysis of TTRs
and their interface with the floating vessel is described. The tension in the TTRs is usually provided by the
aircan/stem assembly that is located around the upper portion of the riser string. The tension is transferred
through the stem head to the riser casing. The bending moment in the riser due to vessel motions and
environmental loading is transferred to the riser primarily by the keel joint that is located at the lower end
of the stem and set of centralizers between the stem and the riser. This paper focuses on the analysis
methods of the keel joint to stem interface as well as on the analysis of the upper stem section.

Procedure
Risers are subject to severe loading coming from different sources. The types of loading could by
described as:

Loading caused by the vessel motions


Pressure of contained fluids
Environmental loading

All of the above loads act together so they need to be considered jointly during any type of analysis.
The main objectives of the riser analysis are (Reference 1):

To ensure that the maximum stress is below the design allowable


To ensure that the fatigue life induced by cycling loading meets the design life of the structure

These objectives are achieved by conducting global analysis with riser modeled using pipe or beam
elements (Reference 2). The results from the analysis utilizing this model are used to assess the global
performance of the structure. Often, there is a need to analyze the structure in a more detailed manner,
focusing on local behavior of particular components. Thus, local models are developed using shell and
solid finite elements. Simplified global boundary conditions such as pinned or fixed nodes of the beam
elements can be modeled more realistically using contact elements on the surfaces of the 3D solid or shell
local models.
The loading that is applied to the global model are the vessel motions that are transferred by the appropriate
boundary conditions and environmental loading in form of the current and waves. However, as the
application of the pressure of contained fluids is fairly straightforward and is done virtually the same way
for the global and local model, the pressure modeling is excluded from the scope of this study.
The application of the boundary conditions and loading to the local model is not that straightforward as the
local model is only the cutout from the larger global model. It is necessary to extract the information from
the global model in order to appropriately constrain and apply the loads to the local model.
The following data can be extracted at each node of the pipe or beam model from the global analysis:

Displacements in 6 DOFs for each timestep


Internal forces in 6 DOFs for each timestep

If the displacements are known for each timestep of the global dynamic analysis, then the following
approach can be implemented to conduct the analysis using local model:
1.
2.
3.
4.
5.
6.
7.

Build the local beam model that will represent a cut-off from the global model.
Apply the boundary conditions extracted from the corresponding nodes of the global model to the
local model.
Compare the response of the local model with the response of the global model as a quality check.
Build the full 3D local model using the solid or shell elements.
Apply the same boundary conditions as applied to the local beam model to the 3D local model.
Compare the response of the 3D local model with the response of the local beam model as a
quality check.
Perform the required analysis on the 3D local model, using ANSYS (Reference 2) built-in submodeling procedures if required.

The procedure described above is possible only, if all the necessary global analysis data is available for the
local analysis. This is not always the case - often the party conducting the detailed FE analysis is provided
only limited data. Usually the available data for the analysis is snapshots of the maximum internal forces
along the riser. Thus, the methodology specified above needs to be adapted to account for these
circumstances. Step two of the approach described above is modified to the following:
2A. Apply the fixed boundary conditions at one end of the model and internal forces resulting from the
local model at the other end.
Due to differences in the boundary conditions and disregarding of the dynamic forces, the local to global
model comparison, usually shows different response. The key issue with this approach is to be able to
assess the impact of these differences on the final conclusions. Otherwise, this method is very effective, if
used properly.

Analysis
In-house Global and Local Dynamic Analysis of the Stem
A schematic drawing of the structure analyzed is shown in Figure 1 below. The stem that is modeled as the
pipe in the global model is in reality a complex structure made up of several different sections. The
objective of the detailed FE analysis is to determine the stress distribution in the tapered section of the
stem, where the wall thickness changes from 0.3 to 0.5in. The global pipe model is shown in Figure 2 and
Figure 3 below. The stem and the riser are modeled with pipe elements. The interaction between the riser
and stem as well as between the stem and the vessel is modeled using friction guides modeled using
CONTA178 contact element with cylindrical gap option.

Figure 1. Coupled Riser - Stem/Air Can Sketch

Figure 2. Coupled Riser-Stem/Air Can Global Model

Figure 3. Coupled Riser-Stem/Air Can Global Model Upper Section


The objective of the analysis is to assess the stress distribution in a section of the stem under the dynamic
loading. As the first step, the global model is analyzed by applying the vessel motions and the current

loading to the model. Analysis is performed with 10 year hurricane regular wave loading. The timetraces
of the 6 DOFs are extracted at two locations. Further to this, the local beam model is developed.
The timetraces of the displacements from the global model are applied to the top and bottom nodes of the
local beam model as the boundary conditions. The results comparison between the local and global model
is shown in Figure 4.

Figure 4. Results Comparison - Global and Local Beam Model


The next step is to build the detailed local model using shell elements. In this case, the local model is a
simple cylinder meshed with shell elements as shown in Figure 5. Different colors of the elements
represent different wall thicknesses.

Figure 5. Local Shell Model FE Mesh


The top and bottom of the cylinder are constrained using the master-slave constrains as shown in Figure 6.
The boundary conditions are applied the same way to the local shell model as to the local beam model. The
dynamic analysis is performed for 40 sec. The timetrace of stresses for the local beam model and the
detailed shell local model are shown in Figure 7. The comparison of the response of both models indicates
the correctness of the analysis process.

Figure 6. Master - Slave Boundary Conditions

Figure 7. Stress in Local Beam and 3D Shell Model


Further steps may be taken using Ansys built-in sub-modeling functions. In this particular case the detailed
analysis of the transitions between two wall thicknesses is conducted using a 3D solid sub-model. The
section along the 3D solid sub-model is shown in Figure 8. The boundaries of the sub-model are defined as
specified in Figure 9.

Figure 8. FE Mesh along the Mid Taper Section of Stem 3D Solid Sub-Model

Figure 9. Stress in 3D Solid Model with Sub-modeling Boundary Conditions


Ansys sub-modeling allows transferring the resultant deformation along the boundary line from one model
and applying it to a different model as boundary conditions. This enables further more detail analysis of
the particular components. This function is very efficient with optimization analysis. However, it is
necessary to proceed with caution when using sub-modeling and ensure that the changes made to the submodel do not impact the global model response. In this case it is necessary to rerun the analysis with the
updated model.

Detailed Finite Element Analysis of Keel Joint


This section discusses another example that illustrates the interface between a global and local model. The
problem involves using ANSYS to perform the detailed local finite element analysis of a keel joint in a top
tensioned riser (TTR). The global results for the TTR are obtained using a pipe element model.
The details of a top tensioned riser are shown in Figure 1. The top tensioned riser in this example is a pipein-pipe type i.e. it has an outer casing and inner casing. The keel joint is used to locally stiffen the fatigue
critical region of the outer casing. The keel joint is centralized within the stem using a wheel/ball, which is
shrunk-fit around the keel joint. The keel joint is attached to the riser pipe using flanges.
The objective of the local analysis is to verify the strength of the keel joint under extreme hurricane loads.
The stress concentration factors (SCFs) (Reference 3) along the keel joint length are also required to
confirm that the geometry of the joint is suitable for the fatigue requirements of the riser system.

Model Details
The entire riser is modeled using pipe elements. This problem differs from the previous example as the
entire global analysis results are not available for the local analysis. The maximum bending moment,
effective tension and shear force distribution snapshots obtained from the global dynamic analysis are
provided along the critical region in the riser.

Detailed FEA of the keel joint is performed in ANSYS. Two FE models - the stepped beam model and the
3D solid model - are developed for the local analysis.
A stepped beam model is used to make a comparison with the global model results and to obtain the
nominal stress distribution along the keel joint length. BEAM3 elements are used for this model. The upper
end of the model is fixed in all six degrees of freedom and the bending moment, tension and shear as well
as reaction between the stem and the keel ball are applied as concentrated forces. The center of the keel ball
is constrained only by the reaction force with no geometric boundary conditions in order to ensure that the
local model assumes the correct displaced shape.
A solid half model of the keel joint, the keel ball, the upper and lower crossover joints and nominal riser
joint is used for keel strength analysis and determining the SCF. SOLID95 3D elements (Reference 2) are
used for the model. Contact between the ball and the keel is modeled using CONTA174 and TARGE170
elements (Reference 2). The mesh at the keel ball is shown in Figure 10. The boundary conditions are
applied at a distance of 5 times the nominal diameter from the region of interest. This is done to ensure that
discontinuities due to the application of boundary conditions are at a sufficient distance from the regions of
interest and do not affect the results of the analysis. For determining the SCF and strength analysis, the
model is assumed fixed at the upper end. As shown in Figure 11, bending, tension and shear forces are
applied as equivalent nodal forces at the lower end of the model. Reaction from the stem is applied at the
center of the keel ball as a point load. Vertical friction between the stem and the ball is also included in the
3D model. The friction coefficient between the stem and the ball is assumed to be 0.5. The shrink fit is
represented by initial penetration of the ball into the keel. The flange is modeled as a solid block, resulting
in a slightly higher stiffness for this component.

Figure 10. 3D Mesh Details at the Keel Ball

Figure 11. Boundary Conditions Applied to 3D Model

Discussion of Results
The stepped beam model is used to verify whether the resulting force and stress distributions, obtained
using the intended method of applying the boundary conditions, match with the global analysis results and
to determine the discrepancies.
The bending moment distribution along the model length obtained using the two models is presented in
Figure 12. The bending moment from the ANSYS stepped beam model is higher than that predicted by the
global model by 7.3% at the keel joint center. All six degrees of freedom are constrained at the upper end
of the beam model. The tension, bending moment and shear loads are applied at the lower end. In the
global model, the displacements at the location of the upper end of the stepped beam model are limited by
the stiffness of the pipe above and by the arrangement of the stem-riser centralizers. Therefore, the fixed
end boundary condition in the beam model causes a higher bending moment distribution along the model
length. The exclusion of the inner casing in the ANSYS model also affects the bending moment. The
bending moment in the inner casing is approximately 10% of the bending moment in the outer casing for
the 1-year winter storm loading condition.

Figure 12. Comparison of Bending Moment from ANSYS Beam Model with Global Analysis
Results
The differences between the global analysis and beam model results do not affect the determination of the
SCF as the nominal stresses are derived by applying the loading and boundary conditions identical with the
3D model to the beam model. For the strength analysis, the higher bending moment obtained using the
beam model implies that applying the same load to the 3D model provides a conservative result.
The SCFs are calculated as the change in the first principal stress due to bending fluctuations divided by the
change in the nominal stress due to bending. The plot for the principal bending stress from the 3D model
and the bending stress from the beam model for a typical loading condition is shown in Figure 13.

Figure 13. Principal Stress (3D Model) and Nominal Bending Stress (Beam Model) along
Keel Joint

Conclusion
The paper presents a detailed methodology for interfacing between global and local finite element models
in order to capture the time dependent geometrically non-linear response of the global system with the
complex multi-component response of local details. Two examples illustrating the procedure are presented.
The first example shows how displacement timetraces can be used to interface between global and local
models. The second example shows how the snapshot of the maximum global loads can be used to perform
local component analysis.
The use of the timetrace approach is more straightforward and reliable. Performing both the global and
local analysis ensures that a one-to-one correspondence between the global and local models can be
maintained whilst avoiding undue complexity in both the global and local models. This facilitates the easy
transfer of boundary conditions from one model to another. However, the limited access to the results of
global analysis conducted by a different party can be a constraint and the second example demonstrates a
way to get around the problem. The results obtained in the second problem are shown to be conservative
and accurate within 10% of the expected results.

References
1) American Petroleum Institute, Recommended Practice for Design of Risers for Floating
Production Systems and TLP's, API-RP-2RD, API, June 1998.
2) Swanson Analysis Systems Inc, ANSYS/Structural Finite Element Analysis Software, Version
6.0, April 2002.
3) R. J. Roark, W. C. Young, Formulas for Stress and Strain, McGraw-Hill, 1984, Singapore.

You might also like