Вы находитесь на странице: 1из 10

Proceedings of IPC2010

8th International Pipeline Conference (IPC 2010)


September 27- October 1, 2010, Calgary Canada

IPC2010-31622

EFFECT OF INTERNAL PRESSURE ON FREE SPANNING PIPELINES

Olav Fyrileiv
Det Norske Veritas,
Veritasveien 1,
N-1322 Høvik, Norway

ABSTRACT were raised and some experimental data presented that claimed
Free   span   assessment   has   more   and   more   become   an to prove that the way the internal pressure was handled in the
important part of modern pipeline design. The reason for this is DNV­RP­F105 “Free Spanning Pipelines” (2006) [2] is wrong.
partly that the remaining hydrocarbon reservoirs are located in The   intention   of  this  paper   is  to   show  how   the   internal
more challenging places, e.g. with very uneven seabed. Another pressure influences on the structural response of free spans, and
explanation is that the pipeline design codes a few decades ago that the DNV codes and standard non­linear FE software, e.g.
did not allow for vibrating free spans, while the modern, state­ Abaqus, handle this effect in an adequate manner.
of­the­art pipeline codes, such as DNV­OS­F101 “Submarine
Pipeline Systems” (2007) [1] and its Recommended Practices, Keywords:   Pipelines,   Effective   axial   force,   Pressure   effects,
opens for long spans that are allowed to vibrate as long as the Pipeline codes.
structural integrity is ensured. By opening for longer free spans
significant seabed intervention costs associated with trenching, INTRODUCTION
rock dumping and supporting spans by other means are saved. As   the   easiest   accessible   offshore   petroleum   reservoirs
One of the governing parameters to ensure the structural have been discovered and developed over the last five decades,
integrity of free spans is the natural frequency of the span. This the trend in the offshore industry points towards deeper waters
is a parameter that the designer can to some degree control by and harsher environment. This trend also calls for increased use
means   of   moderate   seabed   intervention,   e.g.   span   support. of subsea installations instead of platforms and use of pipelines
Since the natural frequency of the span together with the water to transport the hydrocarbons to shore or into a pipeline grid.
flow velocity normal to the span determine the vibrations and The pipelines may be installed at a very rough, uneven seabed
the cyclic loading it is of vital importance to be able to estimate causing a high number of free spans that can be difficult and/or
a realistic value of this frequency.  very costly to intervene. 
The natural frequency is influenced by several effects. One Free   span   design   is   challenging   and   important   –   some
of them is the effect of the internal pressure. This may represent decades   ago  only  short  spans  were   allowed,  no  vibrations   –
a challenge since the effect of the pressure is the opposite of now, DNV­OS­F101 “Submarine Pipeline Systems” (2007) [1]
what one instantaneously thinks is correct. Quite recently some and   the   associated   DNV­RP­F105   (2006)   [2]   open   for   long
discussion about the effect of internal pressure on free spans spans and allowing vibration as long as fatigue is controlled. In

1 Draft Copy
shallow   waters   stress   cycling   due   to   direct   wave   force   may As will be shown later in this paper,  the effective  axial
cause fatigue failure while in deeper waters VIV may do the force   governs   the   structural   response   of   the   pipeline   in   an
same. overall  perspective,  influencing on lateral  buckling, upheaval
VIV   is   governed   by   the   so­called   reduced   velocity,   an buckling, anchor forces, end expansion and natural frequencies
expression found by setting the shedding frequency given by of   free   spans.   For   this   reason,   it   of   utmost   importance   to
the   well­known   Strouhal’s   number   equal   to   the   natural understand its effects and be able to estimate it accurately in
vibration   frequency   of   the   span.   In   this   way   the   natural order to end up with a safe and reliable design.
frequency becomes one of the most important parameters along Some   non­linear   finite   element   analysis   results   will   be
with the flow velocity and the outer diameter of the pipe. given   and   compared   with   recent   experimental   results.   These
The structural response parameters of a free span, such as cases   will   prove   that   recent   claims   about   the   way   pressure
natural frequencies and associated modal stress, are often based effects are treated by the industry and in the DNV codes are
on   simplified   beam   theory   expressions.   The   response   is based   on   a   wrong   interpretation   of   the   experimental   results.
significantly influenced by factors such as the soil stiffness, the Further  it  will   be  shown  that   the standard   way  of  analysing
axial force and the static deflection of the span. For this reason, pipelines   using   standard   pipe   elements   will   yield   realistic
ordinary beam theory expressions will often fail to estimate the estimates   for   the   structural   response   as   long   as   materials,
structural   response   with   a   required   accuracy   for   free   span boundary conditions and loads are represented realistically.
assessment. 
One alternative to beam theory expressions is clearly using PRESSURE EFFECTS AND EFFECTIVE AXIAL
non­linear   finite   element   (FE)   methods.   Provided   that   the FORCE
modeling of span support conditions and other details are done The   effective   axial   force   is   explained   in   detail   in   many
properly,   this   method   provides   reliable   and   accurate   results. papers,   e.g.   Sparks   [7]   and   Fyrileiv   and   Collberg   [11].   The
The method also provides large flexibility in modeling different effect   of   the   external   pressure   is   most   easily   understood   by
aspects like span geometry, span support etc, see Fyrileiv et al. considering the law of Archimedes: 
[3]   and  Kristiansen  et  al.   [4].  However,   FE  analysis   is  time
consuming and not very efficient if a large number of spans are “The effect of the water pressure on a submerged body is
to be screened.  an upward directed  force  equal  in size to the  weight  of  the
According  to  the   Recommended   Practice   DNV­RP­F105 water displaced by the body”.
“Free Spanning Pipelines” (2006) [2], the internal pressure will
have   a   negative   effect   on   the   natural   frequency.   This   is Now, consider a section of a pipeline exposed to external
normally accounted for by the so­called effective axial force, pressure as illustrated in fig. 1. For clarity the only sectional
and as this force increases towards the critical global buckling force included is the axial force, N, the so­called true wall force
value,  the natural  frequency  will  asymptotical  move towards found by integrating the steel stress over the steel cross­section
zero. Even though the effective axial force has been used in area. 
pipeline codes for several decades, see for example DNV’76
(1976) [5], it is still misunderstood and misinterpreted when it
N
comes   to   the   effect   of   pressures.   The   effect   of   the   internal
pressure is treated by for example Palmer and Baldry [6] and pe
Sparks [7] who seemed  to introduce  the term  effective  axial
force.  N
However, some years ago a discussion about this concept = pe
peAe
was initiated and doubt on how to treat the internal pressure
raised,  ref.  Galgoul  et al. [8, 9]. Recently, Massa et  al. [10]
+
conducted an experiment in order to prove that the frequency is
governed by the true axial force rather than the effective axial
force.   The   main   challenge   with   the   internal   and   external
pressures is that the effect of these is often the opposite of what
one instantaneously thinks is correct. Therefore, these effects
have been sources to misunderstandings and wrong designs. 
Figure 1 - Equivalent physical systems – external pressure

2 Draft Copy
As shown in the figure, the section with the axial force and R
v q
the   external   pressure,  pe,   (left   figure)   can   be   replaced   by   a
section where the external pressure acts over a closed surface
M+dM
and   gives   the   resulting   force   equal   to   the   weight   of   the
displaced   water,   the   buoyancy   of   the   pipe   section   (middle M N
figure), and an axial force equal to  N + peAe. Considering the
effect of the external pressure in the way as shown in fig. 1 N Q+dQ
Q
does   not   change   the   physics   or   add   any   forces   to   the   pipe
section. However, it significantly simplifies the calculation as
the   alternative   would   be   to   integrate   the   pressure   over   the x
double curved pipe surface.  dx
In this way it becomes clear that the effect of the external Figure 2 – Forces acting on an infinitesimal length of a pipe.
and internal pressures may be accounted for by the so­called
effective axial force:
By   considering   equilibrium   of   lateral   forces   and   the
S  N  p i Ai  p e Ae (1) bending moment acting on the pipe section shown in fig. 6, the
dynamic equation of motion for the pipe section exposed to an
axial tensile force, N, and a distributed lateral force, q, reads:

NATURAL FREQUENCY AND INTERNAL  4v  2v  2v (2)


PRESSURE EI  N  m  q ( x, t )
x 4 x 2 t 2
As   mentioned   above   free   spanning   pipelines   may   be
subjected to fatigue failure due to direct wave loading and/or where v is the lateral displacement, x the axial co­ordinate, t is
vortex   induced   vibrations   if   the   combined   wave   and   current time and EI and m the bending stiffness and dynamic mass per
flow   velocity   is   sufficiently   high.   Therefore,   the   design   and unit length, respectively.
follow­up   during   operation   need   to   ensure   that   the   fatigue The classical solution to this differential equation without
criterion is not exceeded. The most important parameter when any   lateral   force   (free   vibrations)   gives   the   following
it comes to fatigue of free spans is the natural frequency. By fundamental vibration frequency:
changing the natural frequency, e.g. by introducing additional (3)
span   supports   or   reducing   the   span   length   by   seabed
intervention, the cyclic response may be mitigated.
EI  N 
f0 C1 4 1C2 
The   natural   frequency   of   a   span   is   determined   by   the
following more or less well­known parameters/effects:
 Bending stiffness
 Axial force
 Geometrical effect from initial deformation – arc effect
 Effect of soil stiffness on span shoulders
mL  PE 
Basic textbooks like Clough and Penzien [12] describes in where  L  is   the   span   length   and  PE  =  2EI/L2  (positive   in
detail   how   the   axial   force   influence   on   the   dynamics   and compression) is the Euler buckling force for a pinned­pinned
frequencies of a free spanning pipeline. However, there have span. Note that  N  and  Pcr  are positive in tension, and that  Pcr
been a lot of mistakes when it comes to the effect of the internal and   that   the   constants  C1  and  C2  depends   on   the   boundary
pressure   in   various   papers   and   reports,   e.g.   the   paper   by conditions of the span, for more specific guidance see DNV­
Galgoul et al. [8] where it is claimed that the internal pressure RP­F105 [2] or Fyrileiv and Mørk [13].
will increase the frequency.
To get the details correct it may be worthwhile to consider What happens when the pipe is put in water and subjected
the   governing   differential   equation   of   the   dynamics   for   the to   external   and   internal   pressures   in   addition   to   the   axial
pipe.  force?

3 Draft Copy
and the fundamental frequency:
Looking at the governing differential equation it is obvious (6)
that   most   of   the   terms   remain   the   same;   the   effects   of   the

EI  (N  piAi)
bending   and   axial   force   do   not   change.   Note   that   the   axial
force, N, is strictly speaking the so­called true axial force given
by the integrated axial stresses over the steel cross­section and
contains terms from residual lay tension, temperature expansion f0  C1 4 1C2 
effects,   Poisson   effects   from   hoop   stresses   and   any   axial
expansion/sliding effects.
  The external  and internal  pressure will have an indirect
meL  PE 
effect   on   this   equation   through   the   hoop   expansion   and   the
Poisson’s effect on the true axial force. However, as shown by In the same way the effect of the external pressure can be
Galgoul et al. [8], the effect of the internal pressure can also be incorporated.   It   is   seen   that   by   using   the   definition   of   the
seen as a lateral force,  q. Consider the infinitesimal length of effective axial force:
the pipe as shown in fig. 6 which is deformed with a radius R =
S  N  p i Ai  p e Ae (7)
∂2v/∂x2.   Due   to   the   deformation,   the   pipe   fibres   at   the
compression side become shorter than the ones at the tensile both the governing differential equation and the expression for
side. The relative shortening becomes (R­r∙sin)/R  where    is the frequency are simplified. 
defined in fig. 7. Of course, the same expressions would have been deducted
if the concept of effective axial forces has been applied instead
of the axial force, N, from the beginning. Then the cumbersome
R=1/d2v/dx2
integration of double­curves surfaces would have been avoided.
The effect of the pressures can also be explained by a more
physical model. During vibration of the span the pipe is being
bent, the pipe surface at the compression side becomes smaller
q than the surface at the tensile side. Thus, the internal pressure
will cause a resultant force acting away from the static/mean
dynamic position. Hence, the internal pressure will act in the
same   way   as   increased   mass   or   increased   compressive   axial
force and the frequency will drop. 
 Solving the differential equation and considering just the
pi
static part, i.e. neglecting the dynamic, inertia term, will give a
solution where the span buckles when the effective axial force
has reached a certain critical value depending on the boundary
Figure 3  –   Deformed   pipe   cross   section   with   internal conditions   and   any   significant   static   span   deflection.   This
shows that there is a clear relationship between how the internal
pressure and with a bending radius R.
(and   external)   pressure   affects   both   global   buckling   and   the
natural frequencies of a span. 
The net effect of the internal pressure will be similar to a
It   is   also   quite   obvious   that   the   natural   frequency   will
distributed load, q. The integrated effect becomes:
decrease   as   the   internal   pressure   increases.   Finally,   as   the
2
R  ri sin  (4)  2 v effective axial force approaches the critical buckling load, the
q  pi 
0
R
ri sin d   p i Ai / R   p i Ai
x 2 frequency approaches zero, 0. This occurs for the theoretical
case   with   pinned­pinned   support   conditions.   In   reality,   a
where  ri  denotes   the   internal   pipe   radius.   The   governing pipeline span will experience a gradual deflection or sagging
equation of motion now becomes: that will give raise to non­linear effects not accounted for in the
(5) expressions above. By this reason, the natural frequency of free
 4v  2v  2v
EI  ( N  pi Ai )  me 0 spans should be estimated by use of other tools, e.g. non­linear
4 2 2
x x t FE analysis, when the effective  axial  force  reaches  a certain

4 Draft Copy
value   and   non­linear   effects   become   significant,   as
recommended in the DNV­RP­F105 (2006)  [2].
For   most   real   spans,   the   span   deflection   will   gradually
increase as the effective axial force tends towards the critical
buckling value as correctly stated by Palmer and Kaye [14] and
Galgoul et al [8]. This deflection will release some of the axial
force and may also cause the span to vanish as the span height
is limited.   Anyway, as the effective axial force increases in
compression and approaches the theoretical buckling limit, the
pipeline response becomes complicated and highly non­linear.
Therefore, expressions based on linear beam theory can not be
applied   and  must   be  replaced  by  for   example   non­linear   FE
analysis.
Due to this, the range of applicability for the expressions Figure 4 - Experimental setup used by Massa et al. [10].
based on beam theory in DNV RP­F105 (2006)  [2] is limited
to   short   and   moderate   span   lengths   L/D   <   140,   moderate
deflections /D < 2.5 and moderate geometrical stiffness effects Based   on   the   experimental   results   and   FE   analysis
C2Seff/PE < 0.5­ This is explained in more detail in the paper by performed for this single test case, Massa et al. suggested the
Fyrileiv and Mørk [13]. following modified expression for the natural frequency of a
span:
EXPERIMENTAL RESULTS
Massa et  al.   [10] conducted  some experiments where  a  EI  NL2   
2
f 0  B1    1  2  B2    (8)
free span was supported in a test rig and instrumented, see fig. 2 me L4   EI  D  

Figure 4. Then the internal pressure was gradually increased,
and  the  natural  frequency   measured.   Massa  et  al.  support   in Here N is the true axial force (resultant of the pipe wall stress).
general the concept of effective axial force being a simplified The   stiffness   contribution   from   concrete   coating   is   not
solution   to   global   buckling   and   some   other   cumbersome included, neither is the effective span length as this equation is
problems but claim that that the concept has its limitations as assumed   valid   for   a   pinned­pinned   case.   The   boundary
well. One of these is the application to the structural response condition   coefficients  B1  and  B2  are   set   to   0.906   and   1.580
of   free   spanning   pipelines   and   the   natural   frequency   in respectively.
particular. Based on the test observations they claim that the Comparing   this   with   the   corresponding   expression   in
pipe wall stresses (the true axial force) will govern the natural DNV­RP­F105 (2006) [2] shows that the main difference is the
frequencies for the span. Thus, increased internal pressure will change from effective axial force to true axial force together
cause  pipe wall tension and increased  frequency.  This is the with the boundary condition coefficient of the sagging term.
direct opposite effect than predicted by DNV RP­F105 (2006) Figure   5  gives   the   main   results   of   Massa   et   al.’s
[2]. comparison between their experimental  results, their FE ana­
lysis and their use of the expression for the frequency in DNV­
RP­F105 (2006) [2] (same as Eq. (2)). Three observations can
be made looking at this figure:
 The analytical expression and the FE solution agree for
zero pressure whereas they move in opposite directions
with increasing internal pressure. However, note that the
calculations according to DNV-RP-F105 are probably
incorrect. This claim will be explained later.
 The equation from DNV-RP-F105 predicts zero frequency
(or infinite period) when buckling occurs at approximately
20 MPa internal pressure which is in accordance with also
Massa et al.’s understanding.

5 Draft Copy
 The experiment is approximately 60% off at zero pressure Figure 6 – Comparison of natural frequencies
and shows poor agreement with both the analytical given by FE analysis and by DNV-RP-F105
approaches for increasing internal pressure. with experimental values by Massa et al.
 All curves spread out significantly and the experimental [10], long span, L/D = 151.
data does not really show that the DNV-RP- F105 (2006) [2]
approach using the effective axial force is wrong. As the tested span is classified as a quite long span it is
expected that 2nd order stiffening effect from the span deflection
(sagging term) will be significant for the natural frequency in
the vertical plane. As the internal pressure increases, the span
will release  the potential  build up of effective  axial  force  in
compression   by   deflecting   even   more.   The   effect   on   the
frequency   will   be  that  the  significance  of  the  effective  axial
force   term   that   tend   to   decrease   the   frequency   more   or   less
vanish and the effect of the sagging term that tend to increase
the frequency increases.
To check  the effect  of the internal  pressure  without any
significant influence from the sagging term, a moderate length
Figure 5 – Comparison of natural frequencies span of L/D = 50 has also been analysed. All input parameters
made by Massa et al. [10]. and boundary conditions were  kept the same as for the long
span case of L/D = 151. The results are shown in Figure 7 with
In order to investigate why the results reported by Massa et
frequencies from FE analysis and DNV­RP­F105 (2006) [2].
al. [10] do not fit with the estimates from DNV­RP­F105, the
As   seen   the   results   compare   very   well,   and   the   natural
test pipe was later analysed by DNV using a FE model with
frequency decreases as the internal pressure increases. For this
shell elements in the commercial FE program Abaqus.
span length the sagging term is negligible, and the effect of the
The   results   of   the   FE   analysis   for   different   internal
internal pressure is stemming from the effective axial force and
pressures are shown in Figure 6. As seen the frequency matches
the buckling term.
more or less perfectly the frequencies measured by Massa et al.
This   is   in  contrast   to   the   FE   results  shown   by   Massa   et   al.
which showed a more rapid increase with the pressure. Also the
frequencies predicted by the DNV­RP­F105 increases with the
pressure increase in  DNV’s calculations, also shown in Figure
6. However, as the range of applicability for both span length
(L/D = 151 > 140), static deflection or sag (d/D < 2.5) and axial
force  (Seff/Pcr  > ­0.5) are  exceeded;  only the start  value  with
zero internal pressure is regarded as reliable. This value does
however match the experimental value very well in contrast to
the comparison showed in Figure 5 made by Massa et al.

Figure 7 – Comparison of natural frequencies


given by FE analysis and by DNV-RP-F105 for
medium span length, L/D = 50.

The expression for the natural frequency as given in DNV­
RP­F105, Eq. (2) consists of three different terms, the bending
stiffness contribution, the contribution from the effective axial
force,   due   to   non­linear   effects   or   buckling,   and   the
contribution from the deformed shape, the arc stiffening effect.

6 Draft Copy
Figure 8 – Relative importance of different Figure 9 – Comparison of natural frequencies
terms in expression for natural frequency given by FE analysis, by DNV-RP-F105 and by
according to DNV-RP-F105, L/D = 151. modified expression by Massa et al. [10] for
medium span length, L/D = 50.
These three terms are plotted with relative importance in
Figure 8 for the tested span of L/D = 151. Here it is seen that
the sagging totally dominates the frequency when the internal DNV-RP-F105 SPAN RESPONSE QUANTITIES
pressure is increased. The direct contribution from the effective The  following  approximate  response   quantities based   on
axial force is more or less negligible. However, as the internal the effective span length concept are given in the new DNV­
pressure increases, the effective axial force causes the sagging RP­F105 (2002) [2].
to increase and thereby the frequency to increase. This means The expression for the lowest natural frequency reads:
that the indirect effect of the increase in internal pressure is the (9)
EI  S eff   
2
opposite of the expected effect from the effective axial force f 0  C1  1  CSF  1  C 2   C3   
me L4eff  PE  D  
term   and   dominates   the   change   in   frequency.   Please,   note 
however   that   these   relative   importance   numbers   should   be
  where    is the static/steady­state deflection at span midpoint.
taken as indications rather than exact numbers for this case as
Hence,   the   geometrical   stiffening   effect   due   to   sagging   is
the range of applicability for Eq. (2) is exceeded.  It is also of
accounted for by the  C3  term. For the horizontal direction  C3
interest to check the proposed expression for the frequency for
normally equals zero. Note that  PE  is now based on effective
another   case   than   the   one   used   in   the   experiment.   Using   a
span length, i.e. PE=2EI/Leff2
moderate   span   length   more   typical   for   most   free   spanning
The   modal   (dynamic)   stress   for   one   diameter   maximum
pipeline   cases   with  L/D  of   50,   some   interesting   results   are
deflection   amplitude   is   derived   from   the   mode   shape   and   is
revealed, see Figure 9. Here the FE analysis is assumed to give
given by: 
reliable results as no experimental data exists. The frequencies
predicted by DNV­RP­F105 is very close to the FE results and D  ( Ds  t )  E (10)
shows   a   decreasing   trend   as   the   internal   pressure   increases.
AIL / CF  C 4 1  CSF 
L2eff
This   is   as   expected   as   the   sagging   term   for   this   case   is
negligible. The modified expression by Massa et al. shows the where  Ds  is  the  steel  pipe  diameter   and  D  is the  outer  pipe
opposite effect. Starting below the value of the FE analysis, it diameter   (including   any   coating).  t  is   the   steel   pipe   wall
shows an increasing trend with the internal pressure as claimed thickness and C4 is a boundary condition coefficient. 
by the originators of this modified expression. This is simply The static bending moment may be estimated by: 
not correct. (11)
q  L2eff
M static  C5
 S 
1  C 2  eff 
 PE 
 
where q represents the loading, i.e. the submerged weight of the
pipe in the vertical (cross­flow) direction or the drag loading in

7 Draft Copy
the horizontal (in­line) direction. This expression is based on The   approximate   response   quantities   specified   in   this
the static moment of a span with an approximate correction due section may be applied for free span assessment provided: 
to the effective axial force. Such a correction is discussed by  Conservative assumptions are applied with respect to span
Timoshenko & Gere [15] and is found to give an error less than lengths, soil stiffness and effective axial force.;
 The span is a single span on a relatively flat seabed, i.e. the
2% for C2Seff/PE < ­0.6. Limiting the effective axial force in this
span shoulders are almost horizontal and at the same level;
way means that other sources such as the effective axial force  The symmetrical mode shape dominates the dynamic
level and the soil stiffness will be the dominating sources of response (normally relevant for the vertical, cross-flow
uncertainty and not this approximation itself. response only). Here the following limits apply:
In   case   the   static   deflection   is   not   given   by   direct L/Ds < 140
/D < 2.5
measurement (survey) or estimated by more accurate analytical
Note that these are not absolute limits; the shift in cross-
tools, it may be estimated by:  flow response from the symmetrical to the unsymmetrical
(12) mode will depend on the sagging and the
q  L4eff 1 levelling/inclination of the span shoulders. In cases where
  C6 a shift in the cross-flow response is considered as likely,
EI  (1  CSF )  S 
1  C 2  eff  the structural response of the span should be assessed by
 PE  using FE analysis including all important aspects;
 
 Bar buckling is not influencing on the response, i.e.
C2Seff/PE > -0.5;
The boundary condition coefficients  C1 to C6 are given in
Table   1  for   different   support   conditions.   The   present
INDUSTRY PRACTICE
formulation for the effective span length implies that the values
The industry practice for the last decades when designing a
for single spans on elastic supports (seabed) approach the fixed­
new   pipeline   with   free   spans   or   assessing   free   spans   during
fixed values, as the effective span length becomes equal to the
operation of the pipeline can be summarised as use of:
real   span   length   (very   long   free   spans   and   very   high   soil
 Simplified response quantities; and
stiffness).
 Non-linear finite element analysis.
In case of single spans on relatively even seabed and with
Table 1 ­ Boundary conditions coefficients moderate   operational   temperature   and   internal   pressure,
Pinned­ Fixed­ Single span on seabed simplified response quantities e.g. as given by DNV­RP­F105
Pinned 2) Fixed 3) can   be   an   efficient   and   quite   accurate   way   of   assessing   the
spans.
C 1.57 3.56                     3.56
In   case   of   uneven   seabed,   multi­spanning   sections,   high
1
temperature  and/or  high internal  pressure  (HTHP  conditions)
C 1.00 0.25                     0.25
and   very   long   spans,   more   accurate   methods   than   the
2
expressions based on beam theory are required. The industry
C 0.8 1) 0.2 1) 0.4 1)
practice   is   then   to   apply   common   non­linear   FE   analysis
3
software   such   as   Abaqus   and   Ansys.   The   span   frequency   is
C 4.93 14.1 Shoulder:    14.1(L/Leff)2
hard to estimate for several reasons like boundary conditions
4 Mid­span:    8.6
(where does the span start and end, what is the soil stiffness
C 1/8 1/12 Shoulder: etc),   effective   axial   force   and   effect   of   geometrical   shape
1
5
(sagging).   Therefore,   the   most   reliable   analytical/numerical
18( Leff / L) 2  6 estimate of the frequency is found by applying non­linear FE
Mid­span:   1/24 analysis. Then the survey results may be utilised as input to the
C 5/384 1/384                    1/384 analysis and/or as data to check the realism in the FE results.
6
FE   analysis   of   pipeline   on   uneven   seabed   is   normally
based on beam/pipe elements. All the most significant effects
Note   that   different   effective   span   lengths   apply   in   the with respect to span frequency are accounted for such as:
horizontal and vertical directions and for the static and dynamic  Axial and bending stiffness
response quantities in case of different soil stiffness.  Boundary conditions like soil stiffness and resistance

8 Draft Copy
 Axial force – release by sagging into spans and lateral In  Figure 11  below an example of a pipeline installed on
buckling uneven   seabed   is   shown.   Due   to   the   internal   pressure   and
 Geometrical shape – 2nd order stiffness effects
temperature, the pipeline will lift off the seabed at high points
 Added hydrodynamic mass
and buckle sideways. The effect on the free span lengths and
their fatigue lives in this area  are strongly influenced  by the
In additional the effect of different operational conditions
buckling response.
like   as­laid   (empty),   hydro   test,   operational   and   design
Figure   12  shows   the   fatigue   lives   as   functions   of   span
conditions may be simulated.
length   for   the   case   without   buckling  and   full   effective   axial
As   spans   are   supported   by   the   seabed   with   a   certain   soil
force and the case after release of axial force by buckling. The
stiffness the exact boundary conditions are hard to determine.
fatigue assessment is based on the S­N approach. It can easily
Uneven   seabed   and   multi­spanning   span   section   further
be   seen   that   the   maximum   acceptable   span   length   (given   a
complicates   the   boundary   conditions   when   applying
design life of 50 years) increases significantly from around 35m
approximate   expressions   for   the   frequency   et   cetera.  An
to 72m when the release of effective axial force due to lateral
example  of   this  is  shown  in  Figure  9,  where   the   lower   part
buckling is accounted for. 
illustrates the pipeline and seabed with its spans and the upper
part   shows   the   three   most   relevant   vertical   (cross­flow)
vibration modes.

Figure 11 – Release of effective axial force


by lateral buckling on uneven seabed
Figure 10 – Typical example of multi-
(vertical plan with uneven seabed in upper
spanning pipeline section with multiple
figure and horizontal plan with lateral buckle
modes that may be activated by VIV.
in lower figure).

Another complicating factor  is the axial  force  effect. As


proven in the previous section, the global response, including
free   vibrations   and   natural   frequency,   is   governed   by   the
effective axial force and not the true axial force. In case of very
uneven   seabed   and   multiple   long   spans,   a   part   of   the
compressive effective axial force will be released by deflection
of the pipeline into spans. This has been observed for HTHP
flowlines   in   the   Norwegian   Sea   where   the   seabed   is   very
uneven due to iceberg scars. 
In case of HTHP conditions, the pipeline will also tend to
buckle   globally,   e.g.   by   snaking.   Due   to   this   significant
Figure 12 – Fatigue life estimate (based on
deformation of the pipeline, most of the effective axial force is
S-N approach) as function of span length,
released at the apex of the buckle. By this reason the global single spans.
buckling needs to be addressed in order to establish the correct
effective axial force as an input to the free span design. This is Both   for   the   uncertainty   with   respect   to   span   kinematic
in line with the recommendations in DNV­RP­F105 (2006) [2]. boundary   conditions   and   with   respect   to   the   effective   axial

9 Draft Copy
force the non­linear FE method provides a tool to realistically Configurations Subjected to Vortex Induced Vibrations”,
model and analyse all important effects. ISOPE’98, Montreal, Canada, May 24­29, 1998.
Present industry practice is to use non­linear finite element [5] DNV’76,   “Rules   for   the   Design,   Construction   and
tools based on beam/pipe elements. The geometric stiffness of Inspection of Submarine Pipelines and Pipeline Risers”,
these elements is normally based on the effective axial force Det Norske Veritas January 1976.
concept and should therefore give a reliable and quite accurate [6] Palmer, A.C. and Baldry, J.A.S., “In­Service Buckling of
estimate of the pipeline response. Heated Pipelines”, Journal of Petroleum Technology, pp.
1283­1284, November 1974.
CONCLUSIONS [7] Sparks,   C.P.,   “The   Influence   of   Tension,   Pressure   and
The   following   comments   can   be   used   to   summarise   the Weight   on   Pipe   and   Riser   Deformations   and   Stresses”,
discussion within this paper: Proceedings of OMAE, 1983.
 The expression for the natural frequency as given in the [8] Galgoul, N.S., de Barros, J.C.P. and Ferreira, R.P., “The
DNV code is correct and gives good estimates as long as Interaction of Free Span and Lateral Buckling Problems”,
used within its range of application. Proceedings of IPC 2004, Calgary, Canada, 2004.
 Short to moderate spans will show a drop in the natural [9] Galgoul,   N.S.,   Massa,   A.L.L.   and   Claro,   C.A.,   “A
frequency   as   the   internal   pressure   increases,   while   long Discussion   on   How   Internal   Pressure   is   Treated   in
spans may show an increase in the natural frequency as the Offshore   Pipeline   Design”,   Proceedings   of   IPC   2004,
internal   pressure   increases.   This   is   due   to   the   stiffening Calgary, Canada, 2004.
effect of increased deflection (arc effect). [10] Massa,   A.   L.   L.   ,   Galgoul,   N.   S.,   Guevara   Jr.,   N.   O.,
 Using realistic values of the effective axial force and the Fernandes,   A.   C.   ,   Coehlo,   F.   M.,   and   Neto,   S.   F.   da
static   deflection   cause   the   DNV­RP­F105   (2006) Silva.,   “The   influence   of   internal   pressure   on   pipeline
expression for the natural frequency to fit very well with natural frequency”. Proceedings of OMAE2009, Hawaii,
the frequencies measured by Massa et al. [8].  USA, 2009.
 The stiffening effect of the increased static deflection due [11] Fyrileiv, O. and Collberg, L., “ Influence   of
to increased internal pressure was wrongly interpreted as Pressure  in   Pipeline   Design   ­   Effective   Axial   Force”,
the effect of the true axial force by Massa et al. Proceedings of OMAE2005, Halkidiki, Greece, 2005.
 In many cases, and especially in case of uneven seabed, [12] Clough, R.W. and Penzien, J., “Dynamics of Structures”,
global buckling and/or long and multi­spanning sections, McGraw­Hill, 1975.
use of non­linear FE analysis is highly recommended  to [13] Fyrileiv,   O.   and   Mørk,   K.,   “Structural   Response   of
account   for   variations   in   effective   axial   force   and Pipeline Free Spans based on Beam Theory”, Proceedings
geometrical stiffness effects. of OMAE2002, Oslo, Norway, 2002.
 A common misinterpretation of DNV­RP­F105 is that the [14] Palmer, A.C. and Kaye, D., “Rational Assessment Criteria
limitations   in   range   of   application   for   the   response for Pipeline Spans”, OPT, Offshore Pipeline Technology
quantities   are   valid   for   the   code   in   general.   This   is   not conference, 1991.
correct. There is no known limitation in the span length as [15]   Timoshenko,   S.P.   and   Gere,   J.M.,   “Theory   of   Elastic
long as the span frequency and modal stress are estimated Stability”, McGraw­Hill 1961.
within acceptable accuracy e.g. by non­linear FE tools.

REFERENCES
[1] DNV­OS­F101,   DNV   Offshore   Standard   “Submarine
Pipeline Systems”, Det Norske Veritas, 2007.
[2] DNV­RP­F105,   DNV   Recommended   Practice,   ”Free
Spanning Pipelines”, Det Norske Veritas, 2006.
[3] Fyrileiv,  O., Mørk, K.J., Kathrud,  K., &  Sortland,  Lars
“Free   Span   Assessment   of   the   Zeepipe   IIA   Pipeline”,
OMAE’98, Lisboa, Portugal, July 6­9, 1998.
[4] Kristiansen,   N.Ø.,   Tørnes,   K.,   Nystrøm,   P.R.   and
Damsleth, P., “Structural  Modelling of Multi­Span Pipe

10 Draft Copy

Вам также может понравиться