Proceedings of the Twenty-ninth (2019) International Ocean and Polar Engineering Conference www.isope.

org
Honolulu, Hawaii, USA, June 16-21, 2019
Copyright © 2019 by the International Society of Offshore and Polar Engineers (ISOPE)
ISBN 978-1 880653 85-2; ISSN 1098-6189

Downloaded from http://onepetro.org/ISOPEIOPEC/proceedings-pdf/ISOPE19/All-ISOPE19/ISOPE-I-19-625/1129787/isope-i-19-625.pdf/1 by Bandung Inst. of Tech. user on 02 November 2023

Advanced Pipeline Crossing Analysis
Abdul Rahman El-Chayeb, Don Xiaodong Wang, Faris Ragheb Kamal, Oussama Takieddine
National Petroleum and Construction Company (NPCC)
Mussfah, Abu Dhabi, United Arab Emirates

ABSTRACT

In the Arabian Gulf Region the design and construction of pipeline

crossings poses a number of challenges due to the shallow water
conditions. Typical examples are on-bottom hydrodynamic stability and
Vortex Induced Vibration (VIV). For both of these design aspects, there
are no specific design criteria or methodology dealing with crossings as
is the case for pipelines resting on the seabed. Consequently, the
industry relies on interpretations of existing codes which do not capture
the particularities of a crossing. This paper addresses crossing VIV and
demonstrates that, in the presence of axial compressive loads, advanced
analyses are required for an assessment of the crossing behavior and
possible optimized crossing design.

KEY WORDS: Crossing; Free span; VIV; Pipeline; Crosswalking; Fig. 1: Pre-cast concrete sleepers used in subsea pipeline crossings.
Subsea

INTRODUCTION

A number of fields are developed in phases which span many years and
this phased development can include a large number of pipeline
PIPELINE
crossings. For each crossing, sleepers are installed on both sides of the
existing pipeline to achieve a minimum clearance of 300mm between SLEEPER
the new and existing pipeline. The number of crossings and sleepers
can be large and in one particular UAE offshore project, one phase of
the development included more than 400 crossings and required the SLEEPER
installation of some 3000 sleepers. Sleepers are made of precast SHOULDER
concrete and have a width which can range from 6m to 10m to
accommodate installation tolerances and in-service loads. Typical
Fig. 2: ROV image of pipeline laid on top of a crossing sleeper.
sleepers are shown in Fig. 1. These sleepers will be pre-installed along
the pipeline route and have different heights to ensure a profile which
will not overstress the pipe. The sleepers’ will be installed at a spacing Under operating conditions pipelines develop compressive axial loads
defined according to the calculated span length. The sleepers will due to temperature and pressure effects. This is because the expansion
graduate in height from the extremities near the touchdown region to of the pipelines is restrained by the seabed frictional resistance. These
the center of the crossing. Generally, the pipeline will be laid flat on top axial loads are significant and exceed hundreds of tons for most
of the sleepers as shown in Fig. 2 provided that the frictional resistance pipelines under typical operating temperatures and pressures.
exceeds the hydrodynamic loads. If this is not the case, the pipeline
will be tied to the sleeper using chain assembly, clamps or restraining In the presence of compressive axial loads, a particular challenge for
blocks. the design of crossings is that the allowable span lengths tend to be too
short. This, together with tolerances in the seabed bathymetric survey,

1727
gives rise to hyper-static conditions where the pipeline will not sit on
all the supports (Fig. 3). Remediation of this anomaly requires
installation of grout bags which are tailor made for the specific height
of the gap between the pipeline and the support.
SLEEPER PIPELINE
GAP
Fig. 3: ROV survey results showing gap between pipeline and support.
The analysis results discussed in this paper show that for crossings
without any restraints, the presence of compressive axial loads will
cause the pipeline to move to a new equilibrium position whenever the
pipeline undergoes significant vibration. The movement occurs over a
number of cycles and grows in amplitude with each loading cycle until
the final equilibrium position is reached. For short crossing spans, the
movement will occur for very small vibration amplitudes which can be
insignificant for fatigue considerations. In this respect, it will be shown
that taking into account this particularity of a crossing will help in
optimizing subsea crossing configurations and minimize the
construction issues related to gaps which can arise between the pipeline
and the supports as shown in ROV survey results of Fig. 3.
CROSSING DESIGN
Crossings design requires careful consideration of the spacing and
height of the supports to ensure that the resulting stresses are below the
allowable limits and the spans will not be susceptible to VIV which can
cause fatigue failure. For most crossing configurations, the spacing of
the supports is governed by the latter case and the allowable span
length is calculated based on DNVGL-RP-F105. While this
recommended practice (RP) does not specifically deal with crossings,
the industry relies on interpretation of this RP due to the absence of any
other viable resource to deal with VIV of crossings. Using this RP, the
VIV amplitudes are calculated without taking into account the
nonlinearity related to fixity at the supports which is primarily frictional
and will slip when the vibration amplitudes exceed a certain threshold.
In this respect, crossing VIV analysis should consider the potential for
slip and ensure compatibility of the calculated vibration amplitudes
with the permissible constraints at the supports.
Vortex Induced Vibration
When water flows across a fixed bluff body, vortices are shed with a
frequency given by:
S .U
fs = (1)
D
Where fs is the vortex shedding frequency, U is the flow velocity, S is
Strouhal’s number and D is the diameter of the pipeline. As the flow
velocity increases, the vortex shedding frequency increases and if the
body is free to oscillate the vortex shedding frequency lock-on to the
natural frequency of the body leading to resonant vibrations. These
vibrations will be associated with dynamic stresses which can cause
fatigue failure. The fluid velocity ranges where the vortex shedding will
1728
lock-on to the natural frequency of the crossing are defined by the non-
dimensional reduced velocity parameter Vr given by:
U
Vr = (2)
fi D
A comprehensive design criteria dealing with VIV of pipelines is
provided in DNVGL-RP-F105. For simplicity, reference is made here
to DNVGL-RP-C205 which defines the fluid velocity ranges for the
lock-on region as:
Inline 1 < Vr < 4.5
(3)
Downloaded from http://onepetro.org/ISOPEIOPEC/proceedings-pdf/ISOPE19/All-ISOPE19/ISOPE-I-19-625/1129787/isope-i-19-625.pdf/1 by Bandung Inst. of Tech. user on 02 November 2023
Crossflow 3 < Vr < 16
Inline and crossflow VIV are vibrations parallel and perpendicular to
the flow respectively. For inline VIV, the maximum displacement
amplitudes are of the order of 0.1D whilst for crossflow VIV the
maximum amplitudes can be as high as 1.0D. Since the stress
amplitude is proportional to the displacement amplitude, crossflow VIV
will be associated with significant fatigue damage and common
practice is to allow inline but not crossflow VIV.
While inline VIV is associated with low stress amplitudes, the resulting
fatigue damage can be significant due to the high frequency type of
vibrations. Consequently, allowable inline VIV is limited to extreme
environmental events. There are very limited options to achieve this
with the most common being limiting the span length to ensure that the
natural frequencies of the crossing are outside the lock-on region for
the major part of the design life. Under compressive axial loads this
approach may require spans which are too short and will not sit on all
supports due to seabed bathymetric tolerances and other tolerances
related to pipeline installation.
Natural Frequency
The first natural frequency of the crossing f1 is approximately the first
natural frequency of the longest span with pin-pin end conditions and,
as per DNVGL RP F105, is given by:
π  2
f1 =
EI 1 − P + 4  δ   (4)
2 me L4  Pcr 5  D  
 
In the above equation, EI is the bending stiffness of the pipeline, me is
the effective mass, L is the span length, P and Pcr are the compressive
axial and Euler buckling loads respectively, δ is deviation from straight
configuration and D is the diameter of the pipeline.
The above equation shows that when the span is straight, i.e δ is zero, f1
reduces to zero when the compression ratio P/Pcr is 1.00. This
theoretical scenario has been discussed by Bokaian (1988). In reality, a
pipeline crossing will have a number of inherent imperfections and the
vibration modes of compressed beams with initial slight imperfections
have been the subject of extensive research. The work of Carpinetri et
al. (2014), Hallauer et al. (2010), Zhao et al. (2018) and Dickinson
(1980) is a sample of the experimental and analytical studies on this
subject.
Typical normalized experimental results of Carpinetri et al. (2014)
showing the influence of compression and initial imperfections on the
natural frequency are shown in Fig. 4. The natural frequency is
normalized with the natural frequency of the perfectly straight beam
under zero compression. The figure also shows a plot of the analytical
expression of equation (4) for the straight condition.
The experimental results (Carpinteri et al, 2014) show that for all
values of initial imperfections considered, the frequency decreases as
the axial compression increases. At a certain point this pattern reverses
and the frequency increases with an increase in compression. For the
case with the smallest initial imperfection of 0.2mm, the pattern
reverses at compression ratio of 1.1. Further, for this case and for
compression ratio of 1.00, the theoretical results give a normalized
frequency of zero whereas even for this very small initial imperfection
the experimental results show a normalized frequency of 0.5. These
differences in calculating the natural frequency have a direct
consequence on crossing configurations.
It is worth noting that the experimental results of Carpinteri et al.
(2014), relate to a bar with a rectangular cross section which is 15 mm
deep, 30 mm thick and has a length of 1000 mm. Three different initial
imperfections were considered with values of 0.2mm, 1mm and 2mm.
In terms of out of straightness (OOS), these values equate to 0.02%,
0.1% and 0.2% which are small in comparison with the OOS expected
for a crossing.
Fig. 4: Experimental results of Carpinteri et al. (2014) for natural
frequency of compressed bars
While it is not possible to determine the level of initial imperfections at
the design stage, VIV is always associated with environmental loads
which are expected to have a similar effect to initial lateral
imperfections. VIV is also associated with motion amplitudes which in
the presence of compressive axial loads can cause slip at the supports.
With this premise, the sections below examine the dynamic response of
a crossing under axial compression and initial lateral loads.
DYNAMIC SIMULATION
Dynamic simulations were considered to assess the impact of
compressive axial loads and vibration amplitudes on the natural
frequencies of a crossing. The dynamic simulations were performed for
a typical crossing configuration with six supports. The two central
supports have a height of 1.0m, the two edge supports had height of
0.6m and the supports nearest to the touchdown point had a height of
0.1m. The crossing configuration is shown in Fig. 5.
ABAQUS CAE 2018 was used for the finite element (FE) simulations.
The length of pipeline considered is 2km to minimize boundary effects
with the crossing located at the center of the model. The supports were
modeled as rigid surface with stoppers at both ends to replicate the
shoulders of a sleeper intended to stop the pipeline from falling on an
existing pipeline in case of unforeseen events (Fig. 6). The seabed was
also modeled as flat rigid surface. Interaction between the pipe and the
supports or the seabed was modeled as hard rigid contact with isotropic
friction coefficient of 0.5. The pipe was modeled with PIPE31H
elements with fixed boundary conditions at both ends. The element
length was 1.0m in the crossing region and 10m at the extremities of
the model. ABAQUS PIPE31H elements are 2-node linear pipe
1729
elements with hybrid formulation which include additional variables
relating to the axial force and transverse shear force (ABAQUS, 2018).
The analysis steps considered are a laydown analysis where the initially
straight pipeline is brought into contact with seabed and the crossing
supports. A number of intermediate analysis steps were then carried out
to activate the friction and constrain axially the end nodes. When the
pipeline is in the as-laid position, the pipeline is heated up to induce
compressive axial loads before a static lateral force is applied on the
pipeline. It is considered that the onset of VIV will occur at a reduced
velocity Vr=1 and in this case the static force applied to the pipeline is
Downloaded from http://onepetro.org/ISOPEIOPEC/proceedings-pdf/ISOPE19/All-ISOPE19/ISOPE-I-19-625/1129787/isope-i-19-625.pdf/1 by Bandung Inst. of Tech. user on 02 November 2023
equivalent to the hydrodynamic drag force Fd which is given by:
1
Fd = ρ . C d . f i D f i D (5)
2
Where ρ is the water density and Cd is the drag coefficient. This force
is applied in the lateral direction and will induce an initial imperfection
which is expected to have a similar effect to the initial imperfections
considered in other work. After this step, a frequency analysis is
performed to calculate the first natural frequency at the start of the
dynamic simulation f1,S. A dynamic analysis was then performed at a
frequency which is 0.75 times the analytical natural frequency. The
magnitude of the dynamic force was calibrated to ensure dynamic stress
ranges are small (<10MPa) and of the same order of magnitude of
stress ranges of interest to inline VIV. The dynamic force was applied
as a uniformly distributed line load on the central spans. A frequency
analysis was performed at the end of the dynamic analysis step to
calculate the first natural frequency at the end of the simulation f1,E.
S1
S2
S3
S4
S5
S6
Fig. 5: Crossing configuration with six equally spaced supports.
Fig. 6: Model for Sleepers including idealization of sleeper shoulders.
BASE CASE RESULTS
The base case configuration considered was a 16” pipe with a wall
thickness of 17.5mm and concrete weight coating of 50mm. This pipe
has a specific gravity of 1.87. The span lengths for the crossing were
considered as 20m.
In the absence of any compressive loads, the FE calculated natural
frequency is 1.58Hz for the inline direction and 1.61Hz for the
crossflow direction. It is to be noted that the standard ABAQUS
approach to calculate the natural frequency in contact problems is to
assume that the contact nodes are tied together in the translational
degrees of freedom. The analytical natural frequency for the individual
span length assuming pin-pin conditions is 1.47Hz which is less that the
FE calculated frequency. The higher FE frequency is due to the multi-
span configuration of the crossing with some shorter spans at the
extremities of the crossing and the fact that the nodes at the touchdown
area will have a degree of constrained against rotation. Plots of the first
and second Eigen modes of the crossing are shown in Fig. 7 and Fig. 8
respectively.
Fig. 7: Inline direction: First Eigen mode of the crossing
Fig. 8: Crossflow direction: Second Eigen mode of the crossing
Different temperatures were considered and for any given temperature
the pipeline was subjected to dynamic loads at a frequency which is
0.75 times the analytical frequency. The load magnitude was calibrated
to ensure that the dynamic stress ranges are small. For a compression
ratio of 0.5, a history plot of the mid node displacement is shown in
Fig. 9. The plot shows that at the start of the simulation, the mid node
will undergo a cyclic motion superimposed on a steady motion which
will increase as the simulation progresses. The amplitude of the cyclic
motion is small and of the order of 10mm (0.02D). At a certain point,
the steady motion together with the imposed axial compression will
trigger buckling with a significant lateral movement of 0.8m. The mid
node will continue to undergo a steady motion before the response will
be purely cyclic at the end of the simulation which was performed for
250 seconds.
Fig. 9: History plot for mid node displacement for compression ratio of
0.5. Inserts showing zoom on displacement history at start and end of
simulation.
1730
The displacement of the crossing is associated with a reduction in the
compressive axial force and a history plot of the axial force is shown in
Fig. 10. The plot shows that the initial compressive axial force of
1.03MN reduces to 0.68MN. As for the stresses, a history plot of the
mid node longitudinal stress is shown in Fig. 11. The plot shows that
the stress range at the start of the simulation is ~4MPa and it is ~2MPa
at the end of the simulation.
Downloaded from http://onepetro.org/ISOPEIOPEC/proceedings-pdf/ISOPE19/All-ISOPE19/ISOPE-I-19-625/1129787/isope-i-19-625.pdf/1 by Bandung Inst. of Tech. user on 02 November 2023
Fig. 10: History plot for mid node effective axial force.
Fig. 11: History plot for mid node stress for compression ratio of 0.5.
Inserts showing zoom on stress history at start and end of simulation.
Based on the results discussed above, for any level of significant
dynamic vibrations the crossing and will undergo “crosswalking”
where the pipeline will walk across the sleeper to a new equilibrium
position. This new equilibrium position is associated with a lateral
displacement and a reduction in the effective compressive force. For
the case considered here the natural frequency at the start of the
simulation is 1.18Hz whereas at the end of the simulation it is 1.35Hz.
This increase in frequency should allow for an optimization of the
crossing span lengths as discussed below.
PARAMETRIC ANALYSIS
Compression ratio
Results for a compression ratio of 0.5 have shown the crossing will
undergo crosswalking for any level of vibration however small in terms
of amplitudes or stress ranges. Results of a parametric analysis for
different levels of compression are shown in Fig. 12. The figure shows
that, apart from the case of zero compression, crosswalking will occur
for all cases considered. For low compression, crosswalking will lead to
small global displacement and for high compression it will increase up
to a certain threshold which will trigger buckling. The figure shows
that for a compression ratio of 0.25, the total displacement at the end of
the simulation is less than 0.25m. For a compression ratio of 0.75, the
total displacement is 3.0m and this is the maximum possible value
before the displacement is stopped by the sleeper shoulder (Fig. 13).
 coefficient of 1.00 it is 15mm.

Downloaded from http://onepetro.org/ISOPEIOPEC/proceedings-pdf/ISOPE19/All-ISOPE19/ISOPE-I-19-625/1129787/isope-i-19-625.pdf/1 by Bandung Inst. of Tech. user on 02 November 2023

Fig. 12: History plot for mid node displacement for different Fig. 14: History plot comparing mid node displacement for friction
compression ratios. coefficient of 0.5 (black line) and 1.0 (red line).

As for the stress range, which is of direct relevance for VIV induced
fatigue, the stress range for friction coefficient of 0.5 is 4MPa and it is
10MPa for a friction coefficient of 1.0. In this respect it is shown that
even for the upper bound friction of 1.0 crosswalking can be initiated
for small stress ranges. Sensitivity analyses were also performed on the
S1 distance for the full mobilization of the friction force at for values of
S2 1mm, 5mm and 10mm. The results show that the stress ranges for the
S3
S4 onset of crosswalking are small and less than 10MPa.
S5
S6

Fig. 13: Crossing displacement at end of dynamic simulation for case

with high compression ratio.

Results showing the influence of crosswalking on the first natural

frequency are summarized in Table 1. The table shows the first natural
frequency of the crossing at the start of the dynamic simulation f1,S and
the end of the simulation f1,E. The table shows that in the absence of
compressive axial forces the ratio of f1,E to f1,S is 1.00. As the
compressive load increases, this ratio increases to 1.14 at a compressive Fig. 15: History plot comparing mid node stress ranges for friction
load of 0.5 and 1.53 at a compressive load of 0.75. This increase in coefficient of 0.5 9 (black line) and 1.0 (red line).
natural frequency, which was calculated after a dynamic simulation was
performed, shows that these kinds of simulations are beneficial to Span length
optimizing the span lengths on a crossing. The analysis detailed above was performed for different span lengths
and summary of results is shown in Table 2. The table shows the first
Table 1: Natural frequency for the crossing of 20m spans for different natural frequency for different compression ratios normalized with the
compressive axial loads. natural frequency at zero compression. For span length of 15m, the
minimum ratio is 0.91 while it is 0.88 for a span of 20m and 0.62 for a
P Stress Range Frequency (Hz) span of 25m. As compared with the analytical results, the frequency is
Pcr (MPa) f1, S f1, E Ratio higher for all cases considered.
0 N/A 1.58 1.58 1.00
Table 2: Normalized Frequency for different span lengths. The
0.25 ~4 1.39 1.41 1.01 calculated frequency is normalized with the frequency for zero axial
0.50 ~4 1.18 1.35 1.14 compressive loads.
0.75 ~4 0.92 1.41 1.53 P Frequency (Hz)
Pcr L = 15m L = 20m L = 25m Analytical
0 1.00 1.00 1.00 1
Friction Coefficient 0.25 0.92 0.89 0.89 0.87
For a compression ratio of 0.5, the friction coefficient between the
pipeline and the seabed or supports was taken as 0.5 which is a typical 0.50 0.91 0.88 0.77 0.71
value. Sensitivity analysis for the upper bound friction coefficient of 0.75 0.92 0.89 0.62 0.5
1.0 shows that the pipeline is also susceptible for crosswalking but
higher vibration amplitudes are required to initiate crosswalking. A plot
of the vibration amplitudes for different friction coefficients is shown in
Fig. 14. The figure shows that the displacement amplitudes to initiate Plot of the results for different span lengths is shown in Fig. 16. The
crosswalking for a friction coefficient of 0.5 is 10mm and for a friction figure shows that for the span of 25m, the frequency will reduce with
increasing compression but remains higher than the analytical

1731
expression shown in equation (4) and also shown in the plot. For the
other span lengths, the frequency will initially reduce before it
increases again. The reduction in frequency is due to crosswalking.
which will generate an out of straightness in the lateral direction and a
release in the axial compression. For short spans, the plot shows a
similar trend to the experimental results shown in Fig. 4 but here the
analysis was considered for initial drag forces at the onset of VIV rather
than initial deflection imperfections. For the span of 25m, the onset of
VIV can occur at considerably lower flow velocity (i.e lower drag
forces) and the plot shows that for the compression ratios considered,
3, a span of 15m can only accommodate a tolerance of ±0.04m and
+0.04m/-0.08m respectively. Obviously, for these tight tolerances the
likelihood of gaps between the pipeline and the support is high given
that these tolerances are well below typical survey tolerances and also
because these tolerances can occur between the time of survey and the
time of installation due, for example, to sand movement.
The likelihood of gaps is reduced considerably when the span length is
increased to 20m. For this case the crossing configuration can
accommodate tolerances well above the typical +/100mm survey

Downloaded from http://onepetro.org/ISOPEIOPEC/proceedings-pdf/ISOPE19/All-ISOPE19/ISOPE-I-19-625/1129787/isope-i-19-625.pdf/1 by Bandung Inst. of Tech. user on 02 November 2023

these drag forces are not sufficient to trigger crosswalking.
tolerance. To note that for support location 2 and as shown in Table 3,
the tolerance has a range of 0.32m and the initial crossing configuration
can be adjusted to give a symmetrical tolerance of ±0.16m. For typical
pipeline construction tolerances, gaps between the pipeline and
supports are unlikely for a span of 25m.

The above illustrates that adopting a conservative approach for

assessing the crossing span lengths can be counterproductive. Short
span lengths will not only generate construction issues but also generate
other design challenges to ensure the integrity of the pipeline over the
design life. For the 15m span discussed above, internal corrosion, long
term settlement and variation in operating conditions are some of the
factors which need to be considered given the little tolerance to the
height of the supports. A gap which can arise during the design life will
effectively increase the span length from 15m to 30m which can lead to
excessive fatigue damage. Using a less conservative approach as
Fig. 16: Normalized plot of first natural frequency versus compression discussed in this paper can allow for an increase in the span length and
ratio for different span lengths. has a significant impact on reducing the likelihood of gaps between the
pipeline and the supports.
HYPER-STATIC ASSESSMENT

Results for different crossing span lengths showed that short spans have DISCUSSION
a higher tendency for crosswalking. The shorter the span the higher is
the tendency for gaps to arise between the pipeline and the supports due A typical crossing configuration was analysed under dynamic
to a number of uncertainties including the seabed survey tolerances. conditions using ABAQUS CAE 2018. The analysis considered
different levels of compression and different span lengths. For a friction
In this section, the tendency for gaps is considered for different span coefficient of 0.5, the model parameters for the cases considered are
lengths. Each support in moved vertically to simulate the seabed survey shown in Table 4.
tolerance. When the support is moved upward (positive tolerance) a gap
will be generated at an adjacent sleeper. When the support is moved The dynamic simulations show that short crossing spans exhibit a
downward (negative tolerance) the gap will be generated at the support phenomenon of crosswalking where the pipeline will move laterally on
location. The results of this assessment are shown in Table 3 for the the supports. Crosswalking occurs under combined axial compressive
installation condition, i.e no internal pressure and temperature loads. loads and dynamic vibrations and is distinct from global buckling as it
Only three supports are considered due to the symmetry of the can occur at axial compressive loads well below the threshold for
configuration. The number of supports is as shown in Fig. 5. global buckling. For short spans, the level of vibrations to trigger
crosswalking is very small and the associated stress ranges are
Table 3: Seabed height sensitivity for different span lengths insignificant for fatigue considerations
Support Height L = 15m L = 20m L = 25m
Number Tolerance Crosswalking will result in a release of the axial compressive loads due
1 (1) Positive +0.05 +0.32 +0.52 to the displacement of the pipeline on the crossing. The combined
effect of this behavior results in an increase in the first natural
Negative -0.02 -0.06 -0.09
frequency which has a beneficial effect on reducing VIV induced
2 Positive +0.04 +0.08 +0.3 fatigue. For example under a compressive axial load which is 75% the
Negative -0.04 -0.24 -0.5 buckling load and for a span length of 20m, the natural frequency
3 Positive +0.04 +0.24 +0.5 calculated analytically is expected to reduce to 50% as compared to the
Negative -0.08 -0.24 -0.5 natural frequency for zero compression. Taking into account
crosswalking, the frequency will reduce to 89%. What is more
Note 1: This support with a height of +100mm is often post installed
significant for design considerations is the fact that for short crossing
spans dynamic analysis shows there is a minimum frequency for all
levels of compression considered. For the 20m span considered here
At support location 1, Table 3 shows that for all span lengths the normalized minimum frequency is 0.88 and this value can be used
considered the crossing configuration is very sensitive to seabed to define the crossing configuration regardless of the operating pressure
tolerances. This support with a height of 0.1m is typically a grout bag and temperature conditions. This is helpful as these operating loads
which will be post installed as per site conditions. For supports 2 and cover a range of values which need not be well defined to perform the

1732
VIV analysis and it suffice to consider this minimum frequency for the
entire design life.
Results for different span lengths have shown that crosswalking is a
phenomenon which affects crossings with short spans. Crossing with
longer spans may not be susceptible to crosswalking. Reasons for that
is the fact that for these long spans the onset of VIV occurs at very
small flow velocities and consequently very low drag forces (Eq. 5).
As shown in Table 4 and for a compression ratio of 0.5, the drag force
at the onset of VIV is 450N for a span of 15m while it is only 60N for a
CONCLUSIONS
For phased field developments, pipeline crossings are an integral part
of the project execution and their design poses a number of technical
challenges. These challenges are more pronounced in shallow water
conditions and in the presence of high axial compressive loads where
the calculated allowable span lengths tend to produce short spans. The
short spans, lacking the flexibility to accommodate typical seabed
survey tolerances, generate hyper-static conditions where the pipeline
will not sit on all supports. These challenges can also be attributed to

Downloaded from http://onepetro.org/ISOPEIOPEC/proceedings-pdf/ISOPE19/All-ISOPE19/ISOPE-I-19-625/1129787/isope-i-19-625.pdf/1 by Bandung Inst. of Tech. user on 02 November 2023

span of 25m. These drag forces as considered in this paper are the absence of a clear and specific design criteria dealing with crossings as
lateral forces intended to replicate the effect of initial imperfections is the case for VIV of pipeline free spans. This paper has highlighted
considered by Carpinteri et al. (2014) and others. It is evident that for some certain features which have not been addressed in previous work.
very small initial lateral loads, the level of dynamic stress ranges The paper has shown that dynamic analyses of a crossing, rather than
required to initiate crosswalking in the 25m span is high. Consequently, the static approach usually employed, identify a phenomenon of
the crossing may undergo considerable fatigue damage even in benign crosswalking where the pipe will move to a new equilibrium position.
environmental conditions and well before the occurrence of a storm This new position is associated with a deflection in the horizontal plane
which can initiate crosswalking. together with a shedding in the axial compressive loads and an
associated increase in the natural frequencies. The use of the analysis
Assessment of hyper-static conditions has shown that short span techniques discussed in the paper will help in optimizing crossing
lengths have a very high likelihood to produce gaps between the configurations.
pipeline and the supports. For the 15m span, gaps can be produced
with a tolerance of ±0.04m on the support heights. This tolerance is
well below the typical survey tolerance of ±0.1m. Increasing the span REFERENCES
length to 20m, the crossing can accommodate support height tolerances
well above ±0.1m. This assessment shows that using a conservative ABAQUS CAE 2018 User Manual, 2018
approach for crossing span lengths can be counterproductive and any Bokaian A, (1988). “Natural frequencies of beams under compressive
design optimization should take into account constructability and any axial loads”, J. of Sound and Vibration, Volume 126, Issue 1, 49-65
related issues which can arise during the design life.. Carpinteri, A et al. (2014). “Fundamental Frequency Evaluation in
Slender Beams Subjected to Imposed Axial Displacements”, J. of
Table 4: Model Parameters Sound and Vibration, 2390-2403
Dickinson SM (1980). “The lateral vibration of slightly bent slender
P L = 15m L = 20m L = 25m beams subject to prescribed axial end displacement”, J. of Sound and
Pcr Vibration, Volume 68, Issue 4, 507-514
DNVGL RP F105 (2017). “Free Spanning Pipelines”.
0.25 Temperature (C) 18 10 6.3
DNVGL RP C203 (2017). “Fatigue design of offshore steel structures
Static force (N) 670 214 120
DNVGL RP C205 (2017). “Environmental Conditions and
Dynamic force (N) 100 400 100 Environmental Loads”.
Force Frequency (Hz) 1.70 0.96 0.35 Hallauer, W L, and Ma, C (2010). “Vibration Modes of a Compressed
0.50 Temperature (C) 35 20 12.7 Clamped-Clamped Beam With Slight Initial Curvature”, Virginia
Static force (N) 450 142 60 Polytechnic Institute and State University.
Dynamic force (N) 75 200 100 Zhao, Y, and Huang, C (2018). “Temperature Effects on Nonlinear
Force Frequency (Hz) 1.39 0.78 0.50 Vibration Behaviors of Euler-Bernoulli Beams with Different
0.75 Temperature (C) 53 30 19.1 Boundary Conditions” J. of Shock and Vibrations, 1-11
Static force (N) 220 70 30
Dynamic force (N) 20 25 100
Force Frequency (Hz) 0.98 0.55 0.35

1733