Efficient Assessment of Subsea Pipelines and Flowlines for Complex Spans

James Wang, Paul Jukes, Steven Wang, Gang Duan

JP Kenny, Inc
Houston, TX, USA



ABSTRACT
Pipeline spans occur when a flowline is laid on a rough seabed and/or
when upheaval buckling of the flowline is generated due to thermal
expansion. This not only results in static and dynamic loads on the
flowline at the span section, but also generates Vortex Induced Vibration
(VIV) responses. The phenomenon, if not predicted and controlled
properly, will result in significant damage to the pipeline integrity.

The span issues can be very complicated to analyze due to the long span
lengths, a rough seabed, the large number of spans, and multi-span
interactions. In addition, the complexity can be more onerous and
challenging when soil uncertainty, unknown residual lay tension, and
variation of spans from year to year is considered in the analysis.

The methodology discussed in this paper will not only highlight the
most important areas in the assessment of the complex spans but also
provide many technical details. The new methodology presented in this
paper includes the following: The initial data assessment for seabed and
wave/current is discussed and certain assumptions are made for a
conservative design; Understanding of DNV design code and
implementing it using advanced numerical FE tools is an evaluation
basis for span analysis; In the FEA modeling, many details are discussed
such as model length and concrete induced SCF (Stress Concentration
Factor) at field joints; Certain sensitivity studies for concrete
degradation, survey accuracy, and soil stiffness are also discussed to
ensure the most conservative cases are captured. Special cares are
mentioned in the ULS check for wave/current data (extreme or
significant) and wave load application to interacting spans; In addition,
an example of fatigue calculation for an interacting span is provided.

The approach used in the methodology brings a useful guideline to the
span analysis, especially in the complex span conditions.

KEYWORDS: Boundary Condition (BC); DNV (Det Norske Veritas);
FLS (Fatigue Limit State); FM (Force Model); KP (Kilometer Post);
Mode Shape; Natural Frequency; RM (Response Model); VIV (Vortex-
Induced Vibration); ULS (Ultimate Limit State); and Unit Stress.
NOMENCLATURE
Boundary Condition Coefficients: c1 c3
Concrete Stiffness Enhancement Factor: CSF
Critical Buckling Load: P
cr

Effective Axial Force: S
eff

Effective Mass: m
e

Maximum Current Velocity U
Moment of Inertia for Steel: I
Pipe OD: D
Static Deflection:
Strouhal number S
t

Vortex Shedding Frequency f
s

Youngs Modulus: E
Natural Frequency f
INTRODUCTION
When a pipeline experiences potential damage due to a span, it is
important to adopt an appropriate methodology to identify any potential
damage. The integrity of the pipeline can then be evaluated with
confidence in order to make a decision for the future service of the
pipeline.

The analysis for a free spanning pipeline can be very complicated due to
variations of the current/wave data within different pipeline sections,
soil complexity, multimode vibrations and a high number of spans that
can either be very long or interacting. Even more complexities will be
introduced when both response and force models are considered in the
analysis. The response model is an empirical model providing the
maximum VIV response in both in-line and cross-flow directions. A
force model is based on Morisons equation for direct in-line loading.
The response model has been commonly used in many analyses in JPK
projects.

The span evaluation of new or existing pipelines is compliant with the
design principles in DNV-RP-F105 (DNV, 2006). Based on the DNV
code, the study of free spanning pipeline includes both response and
force models. The response models are based on a Vortex Induced
Vibration (VIV) amplitude response where the VIV is caused by vortex
shedding across the pipeline. There are two types of oscillations to
consider: in-line and cross-flow oscillations. The in-line oscillation
occurs when the pipeline vibrates in a lateral motion. The cross-flow
oscillation occurs when the pipeline vibrates in a vertical motion. The
influencing factors in VIV design are:
Pipe size, weight, and geometry;
Any additional weight such as content, insulation, and fittings;
Current and wave parameters;
Residual lay tension within the pipeline;
Operational conditions such as temperature and internal pressure.
VIV analysis includes natural frequency calculations, which are
normally performed using a Finite-Element (FE) method for the
following reasons:
The seabed geometry is complicated;
Proceedings of the Eighteenth (2008) International Offshore and Polar Engineering Conference
Vancouver, BC, Canada, July 6-11, 2008
Copyright 2008 by The International Society of Offshore and Polar Engineers (ISOPE)
ISBN 978-1-880653-70-8 (Set); ISBN 1-880653-68-0 (Set)
237

The boundary conditions are not easy to be identified;
Interacting spans are occurring.
In some limited cases, such as a single span with a well-defined
boundary condition and a small static vertical defection, the natural
frequency is calculated using an analytical beam method indicated in
Equation (1). Refer to DNV-RP-F105 (DNV, 2006).

+ + + =
2
3 4 1
) ( 1 1
D
c
P
S
L m
EI
CSF c f
cr
eff
eff e

Eq. 1
Refer to the nomenclature for the definition of the parameters in Eq1.

Similar to the response model, the natural frequency for the force model
is more appropriate if it is from the Finite Element Analysis (FEA). The
force model is for the direct in-line loading. The direct loading may not
be very influential for deep water but it might significantly impact the
fatigue life results of spans in shallow water scenarios.

This paper presents a methodology for the screening and fatigue
analyses of pipelines, and intends to be used as a practical guideline for
future pipeline span analysis.
METHODOLOGY FOR ASSESSING FREE SPANS
A span analysis is conducted based on DNV-RP-F105 (DNV, 2006) that
takes into account the complicated scenarios using the methodology
discussed in this paper. An initial assessment of the spans from the
survey data is performed, followed by static and dynamic Ultimate Limit
State (ULS) checks. A VIV screening analysis is then conducted to
determine maximum allowable span length limits for in-line and cross-
flow directions under both current and wave conditions. Finally, a
fatigue analysis is performed on the spans that exceed the allowable
span length limits. A Finite Element Analysis (FEA) model is used in
the analysis to determine natural frequencies, unit stresses and mode
shapes. The results of the analyses provide efficient solutions to the field
in terms of mitigation management for existing pipeline or new pipeline
design.

In the screening analysis, the onset screening criterion and fatigue
screening criterion are used to applicable code requirements regarding
acceptance criteria from DNV-RP-F105 (DNV, 2006). In the fatigue
analysis, the fatigue life is determined using equations for both response
and force models defined in DNV-RP-F105 (DNV, 2006). The proposed
methodology includes the following key areas: assessment of the field
data, ULS check, screening analysis, fatigue analysis, and FEA
modeling. The detailed discussion of the methodology is provided in the
following sections.
ASSESSMENT OF FIELD DATA
The current and wave field data are required in both the screening
determination and fatigue life calculation. The data is evaluated and
processed before use in the analysis.

There are different types of current and wave data normally provided in
the analysis:
Histrogram current/wave data;
Weibull distribution current data;
1, 10, and/or 100 year current/wave data.
The histrogram current data is discrete current velocities with the
associated number of occurrences. If this data is available, it is
categorized in various velocities and directions to be used in the fatigue
calculation. Similar to the current data, the histrogram wave data is also
categorized in various wave heights, wave periods and directions. It is
important that the pipeline and current/wave data are consistent with the
same coordinate system.

The following is key points about the current and wave data:
The current and wave data in the histrogram is collected for at least
a one-year period for more accurate fatigue calculation results. The
fatigue results are conservative if a hurricane is included in the
histrogram current and wave data since the hurricane introduces
much larger current velocity and wave heights;
The current data is normally measured in various heights, and the
data at the height closest to the actual pipeline height (gap) is used;
An assumption of the current/wave perpendicular to the pipeline is
made for conservative results if direction information is not
available;
The different types of current and wave data can be transformed
from one to the other.
Both seabed and pipe profiles are normally provided with raw data
directly from measurement. If applicable for existing pipeline, the
evaluation of the seabed and pipe profiles from a field survey is
performed before the data is used in the span analysis. During the data
processing, the following items are considered:
The seabed data is more appropriate to the analysis when it is
measured underneath the pipe rather than away from the pipe;
Pipe profiles and/or seabed profiles at span shoulders may be
falsely reported: Interpolation from their adjacent points
sometimes is used.
Unreported touchdown point: Due to the limitations of the survey,
touchdown points may be missed. The missing touchdown points
are determined when the seabed and pipe profiles are compared
together. The points are verified through a video, followed by a
confirmation with the surveyors. If no confirmation can be made,
the worst scenario in the span analysis is to assume two cases one
with suspected touchdowns and the other without.
Unreported false touchdown: Sometimes the surveyed seabed may
slightly exceed the pipe bottom, but may not be reported and is not
considered a touchdown. However, it may be taken as a real
touchdown if the seabed profile is converted into a FEA model.
The conservative approach is to lower the seabed below the pipe
bottom by slightly trimming the associated seabed peaks. This may
have to be re-visited if the calculated fatigue results are extremely
lower.
SCREENING ANALYSIS
The purpose of the screening analysis is to define a criterion that is
applied to the free spans. This ensures that any spans that may exceed
screening criteria are identified and analyzed for possible fatigue, and
the study of the most onerous spans is prioritized.

The permissible pipeline freespan length is normally evaluated in the
screening analysis. When doing so, the pipeline route is divided into
several sections corresponding to changes in the concrete coating
thickness, water depth, environmental data, and pipeline orientation.

The maximum allowable span lengths are determined in accordance
with DNV-RP-F105 (DNV, 2006). It indicates that the maximum
allowable span lengths are calculated based on the following criteria:
Onset of vibration due to in-line vortex shedding;
238

Onset of vibration due to cross-flow vortex shedding;
Fatigue screening due to in-line VIV;
Fatigue screening due to cross-flow VIV;
Static and dynamic load ULS.
For each of these criteria, the permissible span length is calculated for
the applicable pipeline sections. The screening analysis is performed
using user-developed spreadsheets with the following features:
Static load, VIV load, and direct wave load are considered in the
calculation;
Boundary condition: single span on seabed is used for a single
span, and the pinned-pinned is used for interacting spans. Based
on DNV RP-F105, (DNV, 2006) the effective span length is used
only for the single span case. For the interacting span case with
pinned-pinned, the apparent length is used. The effective span
length is defined as the equivalent length considering soil stiffness
and boundary conditions. The apparent span length is the original
span length identified based on the seabed profile and pipeline
information;
The maximum gap is used in the screening analysis for
conservative purposes, as the feature may vary from span to span.
Note the assumption made for the fatigue screening as per DNV RP-
F105 (DNV, 2006). When the assumption is not clear, full fatigue
calculations are performed. Also, cases are taken when static or dynamic
stiffness is used in the natural frequency calculations. Refer to Section
6.5 in DNV RP-F105 (DNV, 2006). Note that ULS allowable span
lengths are pipe design code related.

An alternative method to determine VIV, is to compare the vortex
shedding frequency with the natural frequency of the structure. If the
vortex shedding frequency is close to the frequency of any mode, VIV
approximately occurs within that mode. Refer to DNV 30.5,
Environmental Conditions and Environmental Loads, the Vortex
Shedding Frequency is defined by the Strouhal number so it is also
named as a Strouhal frequency. Refer to equation (2).
D
U
S f
t s
= Eq. 2
Refer to the nomenclature for the definition of the parameters in Eq. 2.

The behavior of the Strouhal number is stable for a wide range of
Reynolds numbers except a transition region. For the range of Reynolds
numbers up to 105, the Strouhal number remains at 0.2.
The determination of the single and interacting spans is based on the
initial survey data and engineering experience. In most cases, a single
span occurs when the pipe is fully embedded into the soil at the span
shoulders. The interacting span barely touches the soil (less than 0.5
meters) at one end of the pipe. The final decision is made with the help
of engineering judgment. Alternative ways to distinguish between the
single and interacting spans is to use the free span classification defined
in Section 6.2 of DNV RP-F105 (DNV, 2006). In some cases, a
screening analysis is based on the assumption that all spans are
interacting spans, thus the calculated allowable lengths are conservative.
ULS CHECK
The ULS analysis is performed separately for each span corresponding
to the applicable changes in concrete coating thickness, water depth,
span height, and environmental data. This ULS check is different from
the ULS calculation in the screening analysis in that it computes the
allowable span lengths based on the ULS unity of 1. In the screening
analysis, the ULS loads are calculated based on the natural frequency,
the unit stress from the analytical method, and the maximum span gap at
each section.

The purpose of the ULS check is to ensure the pipeline is within the
corresponding pipeline specification limit by using the more accurate
natural frequency and unit stress obtained from the FEA. The ULS
analysis is performed based on the DNV-OS-F101 (DNV, 2000) Section
5, as well as other pipeline specifications. The loads used in the ULS
calculation include the following:
Static load weight;
VIV loads from both in-line and cross-flow vibrations;
Direct wave load direct drag and inertia effects.
It is also important that the ULS analysis include both static and
dynamic loads. Specifically, the analysis takes into account the drag and
inertia effects, as well as the in-line and cross-flow VIV effects. The
following are noted in the ULS calculation:
The direct wave loads are calculated for both case I (100-year
wave) and case II (1-year wave and 100-year current) defined in
DNV-RP-105.
Either Design Storm Approach or Design Wave Approach is
used for bending stress due to direct wave load per DNV RP-F105.
For design Wave Approach with regular wave analysis in time
domain, the maximum (extreme) wave data instead of significant
wave data, is used.
For interacting spans, the bending moment (stress) due to the direct
wave load is calculated when the load is applied to one of the spans
or/and two of the spans, and maximum bending moments between
the two is used for the ULS check. Note that the wave and current
might come from different angles, therefore it is possible that the
load meets only one span first when the wave/current direction is
not perpendicular to the span length.
Using FEA provides more accurate results for unit stress and
natural frequency, as well as the bending moments due to pipeline
weight and direct wave. When the pipeline is within a bar buckling
condition defined in DNV RP F105, the bending moment
calculation using the analytical equations is not appropriate per
DNV.
It is project dependent whether the full or reduced safety factors are
needed for the function and environmental loads in the ULS check.
FATIGUE ANALYSIS
Similar to the span screening analysis, the fatigue analysis is conducted
for each span corresponding to the applicable changes in concrete
coating thickness, seabed topography, water depth, span gap, and
environmental data. In the fatigue analysis, static load, VIV load, and
the direct wave loads are considered in the calculation. The span gap is
calculated as the average value over the central third of the span based
on the suggestions from DNV-RP-F105 (DNV, 2006).
Preliminary Fatigue Analysis
The preliminary fatigue analysis is performed for a quick fatigue life
assessment for the most onerous spans. In the preliminary analysis, the
natural frequencies and unit stresses of the flowline are calculated using
an analytical beam method based on DNV-RP-F105 (DNV, 2006). In
the calculation, it is important to understand the following items:
When static or dynamic soil stiffness should be used.
When an effective span length or apparent span length should be
used.
239

When the calculation results are invalid due to buckling or
exceeding certain limits.
The results of the preliminary fatigue analysis are normally conservative
due to the analytical results of the natural frequency and unit stress
calculations. However, the fatigue results provide a comparison among
the spans in different sections as well as the spans of different lengths.
This allows the spans to be prioritized for the final fatigue analysis.
Analysis Model
For the span with no weld defects, the final fatigue life is calculated
based on DNV-RP-F105 (DNV, 2006). Both the Response Model and
Force Model are normally included in the calculation. In the
Response Model, both the in-line and cross-flow are considered in the
current and wave induced VIV. For the Force Model, the calculation
is based on Morisons Equation, which is detailed in DNV-RP-F105
(DNV, 2006).

For the span with weld defects, the resulting stress ranges are used as
input variables in a fracture assessment according to the corresponding
design codes, such as the British Standard BS7910 (BS, 1999), to
determine the fatigue life of the girth welds with the presence of flaws.
If available, the initial flaw sizes are found from the associated welding
inspection records or from the more conservative initial flaw size
acceptable criteria.
Natural Frequency Determination
For the final fatigue calculation, the natural frequencies are determined
from the FEA modeling. The input parameters for the modeling include
soil stiffness, residual lay tension and concrete stiffness etc. If the
project is for existing pipelines, the residual lay tension is sometimes not
available. During the FEA modeling for this scenario, the tension is
determined such that it results in a static span profile that matches the
flowline profile measured by the infield survey. Sometimes it is difficult
to match the pipeline profile at the span shoulders, due to local pipe
embedment processes that are not modeled in the analysis. The analysis
only considers the initial contact process and resulting mobilization of
friction forces. In contrast, the real pipeline may vibrate due to VIV and
embed into the soil, or soil movement may cause a build-up over the
pipeline at the shoulder locations. These time-dependent mechanisms
are not captured in the analysis. It is expected that the numerical pipeline
profile may not match the actual profile at the shoulder locations due to
the local soil and pipe movement that occurs during operation.
Therefore, the match of the pipeline profile is selected at the mid-point
of a span. A match between the numerically predicted profile and the
surveyed profile is determined by the following procedure (Refer to Fig.
1):
Determine the static profile of flowline and identify span shoulders;
Determine the mid-point between span shoulders;
Calculate the difference between the predicted span height and the
surveyed span height at the mid-point;
The numerical profile matches the surveyed profile, if the mid-
point height difference is less than a certain limit.

Alternatively, the averaged differences between the predicted span
height and the surveyed span height along the span length are also used
as the match criteria. For this method, the seabed shoulders are trimmed
off such that the height of the seabed at the shoulders is the same height
as the surveyed pipe.

The FEA model normally includes up to a few hundred meters of the
seabed topography adjacent to the spans of interest. It is included to
accurately capture the boundary effects and possible influence of smaller
spans adjacent to the spans of interest. Seabed unevenness and changes
in the seabed over the operation life are sources of uncertainty that make
it difficult to determine the effective axial force in the pipeline from the
reported lay tension. However, the pipeline profile and soil conditions
are characterized by in-field measurements. The finite element analysis
is used to predict the residual tension through an iterative process where
the applied tension is increased until the model prediction matches the
pipeline profile. The soil stiffness is assessed based on the DNV codes.
Depending on the pipe size and the total weight, the dynamic soil
stiffness can vary with one location to another, mostly due to the
variation in concrete thickness.


Fig. 1: Pipe Profile Comparison

To ensure that a conservative estimate of the residual lay tension is
calculated, sensitivity studies are performed to determine a range of
residual tension that produces a matching pipeline profile. The
sensitivity study accounts for the uncertainty due to soil properties,
concrete conditions (intact or damaged), and survey accuracy. The range
of the parameters is initially taken as the following:
The soil static stiffness: nominal +/- tolerance in a certain
percentage;
Survey accuracy (height) +/- tolerance;
Concrete condition Youngs modulus: nominal +/- tolerance.
The tolerance for each item varies from case to case, thereby varying the
inputs. As a result, a range of tension and frequency values are
calculated and is used in the subsequent fatigue analysis.
Fatigue Software
DNV FATFree Software (Version 10) or a user-developed
software/spreadsheet is used in the final fatigue life calculation for the
span without weld defects. In the Version 10 Software, the following
features are included:
Current and wave modules accounting for the current and wave
effects in VIV;
Force model calculations for direct wave load response;
Multimode response for VIV calculations;
Fully compatible with the 2006 Version of DNV-RP-F105 (DNV,
2006).
In the FATFree calculation for fatigue life, the pipe and environmental
data are all inputs. The natural frequencies of the flowline in this
analysis are input obtained from the FEA results.

When considering the spans with weld defects, user-developed fatigue
240

spreadsheets are developed to obtain the stress ranges for each sea state.
Fracture mechanics software such as CrackWise (TWI) is then used to
determine the fatigue life through the fracture assessment based on
corresponding codes.
CALCULATION PROCEDURE
Spans can vary from year to year, and the effect will influence the
fatigue life. The procedure defined below for the overall fatigue
calculation captures this phenomenon:
Outer Loop: For each pipeline section;
Inner Loop: For each survey year where an associated seabed
profile is used;
Step 1: The FEA model is set up. Refer to Section FEA Modeling;
Step 2: Calculate the natural frequency, mode shape, and unit
stress;
Step 3: For the spans without weld defects, input the above results
into the fatigue software for each span, and calculate the fatigue life
based on an S-N curve. For the spans with weld defects, input the
FEA results into a spreadsheet to obtain the stress range. The
ranges are taken as input into a fracture mechanics software for a
fatigue life assessment;
Step 4: Segments are used within a span to capture the yearly
variation. The number and length of the segments totally depend
on the yearly span variation. Based on the calculated fatigue
damage (related to the fatigue life) of each segment during the
years with survey results (steps 1 - 3), the damage during the years
with no surveys is estimated. The cumulated fatigue damage is
then calculated and the fatigue life is determined when the
cumulative fatigue damage is equal to 1. Refer to Fig. 2 for an
example.
Fig. 2: Yearly Variation of Span Length
There are two options to calculate the cumulative fatigue life for each
segment (Refer to Table 1). For example, consider a pipeline where the
fatigue damage for the span is 0.2 and 0.3 for 2006 and 2007,
respectively. Option 1 is used if the fatigue damage for each segment
within the span is assumed to be a maximum of that associated with the
corresponding span length. Option 2 is used if the fatigue damage for
each segment within the span can be determined separately by using
fatigue software. If available, Option 2 is more accurate than Option 1.

Table 1: Cumulative Damage Ratio Calculation Example
Option 1 -Damage
Segment:
Option 2 - Damage
Segment: Year
a b c a b c
2006 0.2 0.2 N/A 0.2 0.1 N/A
2007 N/A 0.3 0.3 N/A 0.2 0.3
Total 0.2 0.5 0.3 0.2 0.3 0.3
Note:
N/A indicates there is no damage to the particular segment
at its associated year
FEA MODELING
As previously discussed, finite element analysis (FEA) is necessary to
accurately calculate natural frequency and unit stresses under various
vibration modes. From the FEA, mode shapes are also obtained in order
to determine combined unit stress under various modes for the response
model. The first mode shape is also used for the force model calculation.
Pipe Element
The 2-node pipe element is used as they are simple and normally
provide enough accuracy. The element size used in the model can be 1D
as a start based on DNV-RP-F105 (DNV, 2006). The DNV code also
states that short elements may be required, especially for short
span/higher order modes. However, the span length and its shoulders
may not be clear before the pipe is laid on the seabed in the FEA
modeling, and equal size elements are used for the entire pipeline. This
may result in several iterations before an optimum element size is
identified.
Boundary Condition
Based on DNV RP-F105 (DNV, 2006), the boundary condition applied
at the ends of the flowline section model should represent the continuity
of the pipeline. If applicable, sufficient lengths of the pipeline at both
sides of the span are included to account for the effects of the side spans.
The length of the FEA model depends on the number of critical spans,
span interactions, and span isolation of the pipeline region. Normally, a
few hundred meters to a few thousand meters is adequate, as span
lengths rarely exceed 100 meters. If identified, virtual anchor points are
used as the alternative start/end points in the FEA model. This can be
performed by verifying the variation of the axial effective force along
the model length.
Concrete Modeling
A concrete is modeled as an outer pipe relative to the steel pipe. The
inner and outer pipes of a bonded PIP share the same FEA nodes and are
used for the intact concrete case. The inner and outer pipe of an
unbonded PIP will have relative axial displacement between the pipes
used to capture concrete slippage conditions.

The single pipe model is also used when an equivalent weight and
stiffness for both pipe and concrete are calculated as inputs. The single
pipe model does not capture the slippage phenomenon.

At the span areas of the pipeline, bending strains are generated on the
241

concrete. If the tensile bending is too large, it may generate cracks
within the concrete that reduces the concrete stiffness. In this case, an
equivalent concrete stiffness is used in the sensitivity study.
Field Joint
Modeling field joints in FEA may not be necessary, as they are too short
to impact the overall curvature. However, the stress concentration due
to discontinuity of concrete coating at field joints is needed. Normally,
this is not modeled in each analysis cases. Instead, the stress
concentration factor (SCF) is separately calculated in a simple FEA
model and is then applied to each analysis case for local buckle check
and fatigue assessment. The SCF normally increases the bending
moments which affects the ULS and fatigue results. Therefore, it is
important to identify how the SCF is applied to the ULS and fatigue
calculation for both response and force models.
Load Step
An example of ABAQUS load step is presented in Table 2.
Table 2: Loading Sequence
STEP NO. STEP DESCRIPTION
1 Activate pipe and flat (fake) seabed contact
2 Initial (Weight and External Pressure)
3 Activate pipe and actual seabed contact
4 Apply tension load
5 Activate axial friction
6 Move pipe to actual seabed
7 Remove pipe and flat (fake) seabed contact
8 Release tension at pipe ends
9 Operational pressure on
10 Operational temperature on
11 Dynamic spring stiffness on
12 Calculate natural frequency
Soil and Pipe Interactions
A contact model is used in the static phase where the pipe is being laid
onto the seabed, and the spring model is used in the calculation of
natural frequency and unit stress.
A contact model includes two contact surfaces, the pipe surface and the
seabed surface. The pipe surface is generated from pipe elements and the
seabed surface is defined as a rigid body with the stiffness between the
pipe and soil.
In the contact model, the pipe and soil interaction in the vertical
direction is considered using the stiffness defined as Linear
pressure-overclosure relationship with an upper limit based on the
soil bearing capacity.
In the contact model for axial and lateral directions, the constant
friction factors are used to account for the pipe and soil static
interaction. The soil stiffness and friction factor are obtained or
calculated using DNV-RP-F105 (DNV, 2006). The maximum
friction force is also defined based on the maximum contact force.
The elastic slip feature of the friction is also modeled in the FEA.
In the spring model, the nonlinear springs are defined to capture the
maximum contact forces.
In-line versus Cross-flow
In the dynamic analysis of the natural frequency calculation, the in-line
and cross-flow are modeled separately and are defined as the re-start
program that follows the static program when the spans are determined.
In the in-line model, the pipeline is fixed in the vertical direction, while
the in cross-flow model, the pipeline is fixed in the horizontal direction.
The spring elements are defined in the static phase along the entire
pipeline and are activated at the shoulders in the dynamic phase based
on the contact results from the static program. If the original seabed is
used, the contact feature is identified through the contact force between
the pipeline and seabed. If the modified seabed is used to capture the
pipe settlement results from the yearly variation, the springs are
activated at the entire shoulder as the pipeline is fully in contact with the
original seabed.
Spring versus Contact
The spring model is also used in the static phase depending on
preference. The spring model may not be straight-forward, as some
manual interactions are necessary in the determination of the contact
between the pipe and seabed around the shoulders.

The static comparison study between the spring and the contact models
are performed and presented in this section. The dynamic comparison
between the spring + spring model and the contact + spring model is
also provided. The spring + spring model uses the spring model for
both static and dynamic phases. The contact + spring model uses the
contact model for the static phase, and the spring model for the dynamic
phase. Refer to Fig. 3 and Fig. 4 for model diagrams.










Fig. 3: Spring Model



Fig. 4: Contact Model
To allow for a direct comparison of the contact elements and spring
models, the results from the contact model are used to determine the
nodes that are in contact with the seabed profile. In the spring model,
spring elements were applied to the nodes that are in contact with the
seabed in the axial, lateral and vertical directions. In addition, two
spring + spring models are compared to the contact + spring model.
The first (lower bound) assumes that the maximum vertical reaction
force is equal to the submerged weight of the pipeline and the second
(upper bound) assumes that the maximum vertical force is equal to the
maximum vertical bearing resistance of the soils as defined in DNV RP-
F105 (DNV, 2006).

242

As an example, the calculations of the static and dynamic results are
performed for a 50-meter span. The calculation results include the
following:
Axial displacement;
Vertical deflection;
Contact force;
Static stress;
Effective axial force;
Natural frequency;
Mode shapes;
Unit stress amplitudes.
All of the above results show that the contact + spring results are
between the lower bound and upper bound from the spring + spring.
The pipeline vertical displacements and reaction force are presented in
Fig. 5 and Fig. 6. Therefore, the study shows that both the spring +
spring and contact + spring methods make little difference in the
determination of the natural frequency and mode shapes.

-0.45
-0.35
-0.25
-0.15
-0.05
0.05
0 50 100 150 200 250 300 350 400
KP (m)
S
t
a
t
i
c

D
e
f
l
e
c
t
i
o
n

(
m
)
Contact-Spring
Spring-Spring-Low B
Spring-Spring-High B
Span

Fig. 5: Defection Comparison between Spring and Contact Models

Fig. 6: Reaction Force Comparison between Spring and Contact
Models

The following are noted in the spring versus contact study:
An upper limit of the maximum reaction force is defined in both
methods; The upper limit should never exceed the soil bearing
capacity;
An upper limit of the friction force in the contact + spring
method is defined;
In the spring method for the static phase, care must be taken in
the area with no contact between the pipe and seabed at the edge of
the span shoulders, especially for the corners at the edge
extremities;
EXAMPLE
An example of a fatigue calculation is presented for a 24 x 0.625
pipeline with 2.36 concrete laid on a rough seabed. The span starts at
KP (Kilometer Post) 20.166 and ends at KP 20.242. This span is an
interacting span with lengths of 25 and 41 meters. Refer to Fig. 7 for the
pipeline and seabed profiles. The calculation determines the fatigue life
of the pipeline due to the VIV and direct wave loading.
Based on the methodology presented in the previous sections, the natural
frequency and mode shapes were calculated and the results are presented
in Table 3 and Fig. 8 and Fig. 9.
Table 3: Natural Frequency (Hz)
Natural
Frequency, Hz
1
st
Mode
2
nd

Mode
3
rd

Mode
4
th

Mode
Inline 0.790 1.190 2.700 5.530
Cross-Flow 1.130 1.440 3.100 6.150

-34
-33.5
-33
-32.5
-32
-31.5
-31
-30.5
-30
20000 20050 20100 20150 20200 20250 20300 20350 20400
Length along Flowline (m)
D
e
p
t
h

(
m
)
0
10000
20000
30000
40000
50000
60000
70000
FEA Pipeline
seabed
Spring Elements
Contact Pressure
25m Span
41 m Span

Fig. 7: Pipe and Seabed Profile
From the calculated natural frequencies and the mode shapes from the
FE modeling, the fatigue life was calculated using DNV code and the
results are presented in .
Table 4: Fatigue Life
Item Fatigue Life, Yr
In-line (Response Model) 5
Cross-Flow (Response Model) 214
In-line (Force Model) 146
In-line (Combined) 5
0
10000
20000
30000
40000
50000
60000
70000
0 50 100 150 200 250 300 350 400
KP (m)
V
e
r
t
i
c
a
l

R
e
a
c
t
i
o
n

F
o
r
c
e

(
N
)
Contact
Spring-Low B
Spring-High B
Span
Contact: 61295 N
SpringLowB: 61435 N
SpringHighB: 55948 N
243


-1.2
-0.8
-0.4
0.0
0.4
0.8
1.2
20164 20174 20184 20194 20204 20214 20224 20234
KP, m
U
n
i
t

M
o
d
e

S
h
a
p
e

-
C
r
o
s
s

F
l
o
w

(
m
)
Mode1-CF
Mode2-CF
Mode3-CF
41 m Span 25m Span

Fig. 8: Mode Shape Cross-flow
-1.2
-0.8
-0.4
0.0
0.4
0.8
1.2
20164 20174 20184 20194 20204 20214 20224 20234
KP, m
U
n
i
t

M
o
d
e

S
h
a
p
e

-

I
n

L
i
n
e

(
m
)
Mode1-IL
Mode2-IL
Mode3-IL
Mode4-IL
41 m Span
25m Span

Fig. 9: Mode Shape In-line
The results for the in-line response model, cross-flow response model
and the in-line forces are separately reported. The results show that the
in-line response model is dominating in this case, primarily due to the
lower in-line natural frequency. The result from in-line force model is
higher than that from the in-line response model, primarily due to the
medium water depth. If the water depth decreases, the fatigue life
calculated from the in-line force model will be greatly reduced. Several
primary parameters that influence the results in general are the natural
frequencies, mode shapes, span lengths, current / wave data, water depth
and weld type etc. These parameters vary from project to project which
affect the fatigue results in each case. The in-line response model, the
cross-flow response model, or the in-line force model can be the
dominating model in terms of the fatigue life results.
If necessary, after the fatigue life is calculated against the design life,
further management is initiated. The plan depends on:
The fatigue results;
Whether it is a new pipeline or existing pipeline;
Long-term strategy for the pipeline.
If required, mitigation approaches such as the VIV suppression devices
are used to reduce the VIV response and fatigue damage. The most
commonly used devices are strakes and fairings. Some other methods,
such as using sand bags under the pipe are also used so that the span can
receive some vertical support. In this condition, a longer span would be
split into two or more shorter spans thus greatly enhancing the fatigue
life.
CONCLUSIONS
This paper presents a new methodology for analyzing free span
pipelines. The methodology with details and examples, highline key
factors in the span analysis. The methodology has been used on real
projects in various scenarios, yielding the following main conclusions:
The initial data assessment for seabed and wave/current is
important as the seabed data might come with false information and
wave/current data might not have some information such as
directions. Certain assumptions are required to be made and
detailed guidelines are presented in this paper,
Advanced numerical FE tools can adequately simulate the span of
pipelines in static and dynamic phases with a good understanding
of the DNV RP-105. In the FEA modeling, special cares are taken
in model length, element size, concrete induced SCF (Stress
Concentration Factor) at field joints, and etc.
Certain sensitivity studies for concrete degradation, survey
accuracy, and soil stiffness might be required to ensure the most
conservative cases are captured in the analysis.
Special cares are also taken in the ULS check for wave/current data
(extreme or significant) and wave load application to interacting
spans.
It is believed that this methodology could be used as a starting point for
projects with complicated spans.
ACKNOWLEDGEMENT
The author would like to thank all that have provided input into this
work.
REFERENCES
ASME (2003), Gas Transmission and Distribution Piping Systems,
B31.8.
BS (1999), Guide on Methods for Assessing the Acceptability of Flaws
in Metallic Structures, BS7910:1999.
DNV (2000), Submarine Pipeline System, DNV OS-F101.
DNV (2003), FATFree User Manual, 2003-0511.
DNV (2006), Free Spanning Pipeline, DNV-RP-F105.
DNV (2000), Environmental Conditions and Environmental Loads,
No. 30.5.
Fyrileiv, O., Mork K., and Chezhian, M., (2006) Ormen Lange Pipeline
Project Free Span Design in Deep Waters, 1st IOPF.
Fyrileiv, O., Mork K., and Chezhian, M., (2002) Structural Response of
Pipeline Free Spans based on Beam Theory, Proceedings of
OMAE2002, Osla, Norway.
Nielsen, F. G. A suggested procedure for estimating Vortex Induced
Vibrations for long free spanning pipelines, Norsk Hydro Report
2002a.
Nielsen, F. G., Kvarme, S.O. and Soreide, T., VIV response of long
free spanning pipelines, 21
st
International conference on offshore
mechanics and arctic engineering, (OMAE), Oslo, Norway, 2000b.
TWI, CRACKWISE Software.

Copyright 2008 The International Society of Offshore and Polar
Engineers. All rights reserved.
244