You are on page 1of 18

OTC 5294

Buckling Considerations in the Design of the Gravel Cover for


a High.:remperature Oil Line
by S. Boer and C.H. Hulsbergen, Delft Hydraulics Laboratory; D.M. Richards,
Lloyd's Register of Shipping; A. Klok, Pipeline Engineering Services; and J.P. Biaggi,
Total Oil Marine Engineering & Construction Ltd.
Copyright 1986 Offshore Technology Conference
This paper was presented at the 18th Annual OTC in Houston, Texas, May 5-8, 1986. The material is subject to correction by the author. Permission
to copy is restricted to an abstract of not more than 300 words.
ABSTRACT
Total Oil Marine have recently installed a 15.3
km long 12" pipeline in the UK sector of the North
Sea as part of the development of the Alwyn Field.
The pipeline is completely covered by gravel for
physical protection.
This paper discusses design engineering problems
related to anticipated buckling phenomena
associated with high oil temperatures during
operation of the pipeline. Special studies were
performed by LLOYD'S REGISTER OF SHIPPING and
DELFT HYDRAULICS LABORATORY concerned,
respectively, with buckling and heat transfer
aspects of the pipeline design. In addition, full
scale pull-out tests were carried out.
INTRODUCTION
The Alwyn North field is located in the
northern part of the UK sector of the North Sea,
some 160 kms east of the Shetlands and 110 kms
north of the Frigg field and has been developed by
Elf UK (66.7%) and Total Oil Marine (33.3%), TOM
being the Operator.
The development plan consists of the following
f acilities:
one drilling and accommodation steel platform
(NAA)
one production steel platform (NAB)
two export subsea pipelines:
a 12" pipeline for the transport of the oil
production from Alwyn NAB to Ninian
Central.
a 24" pipeline for the transport of the gas
production from Alwyn NAB to Frigg TP-1.
References and illustrations at end of paper.
9
The Alwyn-Ninian oil pipeline is 15.3 kms long
and made up of 12" X65 (API 5LX) -0.406" W.T. -
seamless, epoxy and 2" concrete coated pipes.
For physical protection purposes the Alwyn-
Ninian pipeline is covered by a continuous gravel
embankment over its entire length. As such it is
amongst the longest pipelines completely covered
with gravel in the North Sea. The gravel
embankment has a triangular shape of which the
height relative to the seabed varies from 1.4 m to
1.0 m and a maximum embankment slope of 20 to
meet the requirements of the fishing industry.
Due to high oil temperatures to be sustained in
operation (design temperature 84C at riser sea
bottom extremity), a potential bar buckling
problem induced by large thermal stresses was
investigated as part of the normal pipeline
engineering design.
Lloyd's Register of Shipping - Ocean
Engineering Department (LRS) were commissioned to
perform a specific study to analyse all the
aspects of this potential bar buckling.
LRS first findings were that, assuming the
pipeline to be an extremely long beam "snaking" on
the sea bottom, there was no risk from bar
buckling as such. However an increase of the
initial lateral deflection would occur which did
not induce dangerous additional bending stresses.
Such a model was particularly suitable for an
uncovered pipeline simply laid on the seabed. This
concept, however, did not allow the proper
analysis of the local phenomenon, i.e. a straight
section of 100-200 m long completely laterally
restrained by gravel. LRS therefore developed an
alternative model which indicated that the
pipeline might buckle locally and emerge
vertically from the gravel embankment if the depth
of cover was not sufficient.
Stability curves giving the minimum required
depth of cover in terms of pipeline wall
temperature were established and as a result, a
2 BUCKLING OF HIGH TEMPERATURE OIL LINE COVERED BY GRAVEL OTC 5294
depth of cover profile along the pipeline
corresponding to the temperature drop could be
defined. Thus, a minimum depth of cover of 1.0 m
is required near NAB where the temperature is
maximum whereas only 0.6 m is sufficient near
Ninian Central. This also represents the minimum
requirement in terms of protection against
dragging cables.
In parallel with the above studies, the
possibility of thermal convection inside the
gravel embankment was identified and had to be
investigated. A conventional heat loss model based
on pure conduction predicted an arrival
temperature at Ninian CentraLin the order of 60'C
for an inlet temperature of 84c at NAB whereas a
first calculation taking into account thermal
convection had indicated that a lower arrival
temperature of about 30C could be reached.
Delft Hydraulics Laboratory (DHL) were
commissioned to perform a study to develop a heat
loss model. The model enables the computation of
heat transfer coefficients from which the
resulting temperature drop along the pipeline was
derived. The computation method, in which the
pipe, the gravel embankment and the surrounding
sea water constitute a single system, takes into
account the forced convection through the gravel
embankment as a result of bottom currents.
The temperature profile determined by DHL was
used in combination with the LRS study results to
optimise gravel dumping requirements. It was thus
possible to reduce the originally specified requi-
rements from 145,000 T to 95,000 t, representing a
gravel dumping cost reduction of about 50%, these
costs being almost directly proportional to the
gravel quantity to be dumped. .
Tests
In addition, pipeline pull-out tests were
carried out by DilL to confirm the calculated
vertical restraining forces exerted by the gravel
embankment on the pipeline to resist vertical
buckling.
Full scale tests were performed underwater at the
DHL Delta Flume facilities, on a 3 m section of
pipeline. The tests were set up to minimise end
effects and to measure directly the relationship
between pull-out force and gravel depth of cover.
A good correlation between the observed results
and those predicted by theoretical calculations
was found.
Offshore Works
--------------
The gravel dumping workswere entrusted to ACZ
Marine Contractors [1] using their new D.P. gravel
dumping vessel "TROLLNES" (8,200 T capaCity, -
flexible fall-pipe technique). Already 90% of the
12" Alwyn-Ninian pipeline has been covered
according to the above requirements in 1985 during
the first phase. The remainder will be achieved in
1986 during a second phase after completion of
subsea tie-ins connecting the pipeline ends to
platform risers. A section of 300 m was left '
uncovered at both extremites to allow for
necessary pipe handling during 1986.
10
BUCKLING OF COVERED PIPELINES
General
-------
The development of marginal fields has required
the design of small diameter export lines to work
at high pressure and temperatures. Pipelines of
this type are highly stressed and VUlnerable to
accidental damage, and will usually be protected
by trenching or gravel dumping.
A pipeline laid on a comparatively flat sea bed
will occupy a path which is not precisely
straight, following an approximately sinusoidal
path as the pre-specified route is pursued by the
constructor. This shape is maintained in
equilibrium by transverse friction forces acting
at the point of contact. When such a pipeline is
pressurised and heated, the resulting thermal and
mechanical strains may be accommodated by an
accentuation of transverse displacements without
excessive wall stresses. However, if transverse
resistance is high, as in the case of trenched or
covered lines, these movements cannot easily take
place and large axial compressive forces will
build up. This situation leads in turn to the
possibility of upheaval buckling.
As with other forms of structural instability, the
upheaval buckling mode provides the possibility of
an alternative equilibrium configuration or bifur-
cation. Again, by analogy with other buckling
phenomena, the existence of geometric
imperfections allow eqUilibrium paths to be traced
between states by appropriate analysis.
For a pipeline, the imperfections concerned may
include a crest in the sea bed profile, or a prop,
as when an isolated rock is located immediately
below the line or another pipeline is to be
crossed. Other less obvious possibilities include
the free span gap or trench step, or an angularly
mismatched field joint.
The upheaval phenomenon has historically been of
concern to railroad engineers who have to consider
track buckling under solar heating conditions
[2,3]. More recently the pipeline problem has been
discussed by various authors including Hobbs [4],
Kyriakides [5] and F ~ i e d m a n [6].
Upheaval buckling must be given serious
consideration at the project stage for a number of
reasons. In the first instance, upheaval must be
avoided otherwise the pipeline will almost cer-
tainly be overstressed and probably permanently
damaged. The other aspect is essentially economic
since the temperature which may be withstood is
directly dependent on pipeline weight, or negative
buoyancy and effective download. This relates
directly to the cover depth or trench depth
necessary to eliminate the possibility of
upheaval. Hence the importance of the pull-out
tests described in this paper which were used to
confirm the relationship between effective
download or upheaval resistance, and cover depth.
In order to illustrate the analytical treatment
of upheaval, the case of a pipeline supported by
rigid object of height h above an otherwise flat
sea bed will be discussed.
3 BOER, HULSBERGEN, RICHARDS, KLOK & BIAGGI OTC 52.94
herein is
( 11)
(12)
o
o
/) -y
so that, in terms of span
*
y -h2
j4 EI
P '" j/
The subsequent post-upheaval behaviour is given by
the conditions:
Generalised plots of these functions are shown
in Figures 3 and 4, which may be used to study the
complete stress and deformation history for a
propped pipeline under increasing axial load P.
It now becomes necessary to interpret the axial
load P in terms of internal pressure p and wall
temperature increase T. The axial load is made up
of the axial pressure force Pc and the axial wall
force P
w
' so that assuming compression is
positive:

(1)
w[}r
(2)
h2
W
[Tr
(3)
and central
( 4)
6lEI h
M
l[1] + M2[1] + F8f +
62
EI
+ M2[1] + F[1f +
/) EI hO
M
2[1f + \
Where 6
1
,6
2
and /) are boundary slopes
displacement, respectively, and .
The equilibrium of a simply supported slender
beam (Figure 1) of span L, loaded by a uniform
lateral load w, an axial force P, a central
vertical force F, and boundary moments M
l
and M
2
,
is conveniently expressed in the following form:
EI pipeline bending stiffness
The coefficients to are functions of j
only and may be derived directly from standard
beam column theory [7]. The algebraic form of
these functions depends on the nature of the axial
load P. Thus if P is compressive, the are
circular functions, if P is tensile the are
hyperbolic and when the are algebraic.
The interpretation of axial force in terms of
temperature increment for any given internal
pressure is decided by the requirement for axial
strain compatibility which gives at all stages:
For the propped pipeline (Figure 2), before
upheaval when the pipeline is still in contact
with the rock, the following conditions apply:
6
1

6
2
0
M
l
M
2
'" 0
F -V
/)
-h
where
1T
Pc '" 4" Dr p
D
i
'" internal diameter
(13)
Upheaval begins when the rock reaction V becomes
zero, i.e when
(14)
-vpD
i
-P
w
AE
axial wall strain '"
steel area
Poissons ratio
'" thermal strain coefficient
'" axial friction coefficient
'" Poisson pressure strain
1T /) 2
'" curvature strain '" - 4" (i)
'" boundary displacement strain '" 2
L
(P _ P ) 2
woo w
'" second moment of area
'" thermal strain '" aT
2EA lJW
'" external steel diameter
E + + + + Ed constant
apt c
Individual strain components are:

p
t

c
d
6.
Do
A
I
v
a
II
11
(9)
( 5)
(10)
8.9868, for which (5),
4.5148
p '" P* '" 3.9621
'" tan (.1) _.1.;. 0 (8)
.. 2 2
L L*
this occurs when j '" j*
(7) give:
4 BUCKLING OF HIGH TEMPERATURE OIL LINE COVERED BY GRAVEL OTC 5294
P
w-
= aETA -
V~P Di(Do+ Di) - T
o
To = residual laying tension
By comparing current with initial laying
conditions, (14) may be used to find either the
internal pressure for any given temperature and
total axial load, or for fixed pressure, to
calculate pipeline temperature. In this way the
complete pressure/temperature history may be
traced through the upheaval process, together with
corresponding stresses, spans and displacements.
Qplication
--------
The above procedure has been applied to the 12
Alwyn-Ninian oil pipeline configuration to
demonstrate the nature of the phenomenon, and to
show in detail the effects of vertical constraint
force and prop height.
Figure 5 shows a typical complete upheaval history
including pipeline temperature (T), vertical dis-
placement (y), maximum equivalent stress (u/uy),
internal pressure (p) and prop reaction (V).
The feature of particular interest 2s the equili-
brium temperature T, which rises to the upheaval
value T1 after internal design pressure has been
reached. After upheaval, the temperature required
for equilibrium falls to a minimum T2, before
rising again continuously. Thus if pipeline tempe-
rature is increased beyond the upheaval value Tl,
the configuration will experience a catastrophe;
jump to a greatly increased deformation as indica-
ted in Figure 5.
The effects of prop height h and download w on
equilibrium temperature history are illustrated in
more detail in Figures 6 to 9. Figure 6 indicates
how all equilibrium temperatures are increased by
increasing download, i.e. gravel cover. A curious
feature of the phenomenon is illustrated in Figure
7, which shows that although upheaval temperature
is increased for smaller prop heights, the post-
upheaval minimum temperature is virtually
unchanged. Upheaval, and post-upheaval minimum
temperatures T1 and T2 are plotted in_Figures 8
and 9 against prop height and download.
Because of uncertainties in predicting or
measuring prop height, and the possibility of
transient loss of cover effectiveness due to
hydrodynamic or other actions, T2 is recommended
as the design value for cover depth choice
together with a margin to cover construction and
operational variations.
The damaging effects of upheaval are emphasised by
the nominal stress levels achieved at upheaval as
shown in Figure 10.
HEAT TRANSFER STUDY
General
-------
The heat loss of the oil pipe and the
subsequent temperature drop along the line depends
on the initial oil temperature, the masa flow of
the oil, the thermal conductivity of the composite
pipe and the subsequent heat transport in the
gravel surrounding the pipe. The heat released
from the oil pipe can be transported through the
gravel both by conduction and convection. The
conductive heat transport is controlled by the
thermal conductivity of the gravel material and
contained pore water. It occurs even in the case
of static pore water. Convective heat transport
occurs only in the case of pore water flow through
the gravel embankment. Whether conduction or
convection dominates the heat transport strongly
depends on the pore water flow characteristics.
The flow through the gravel embankment is con-
trolled by the permeability of the gravel material
and hydraulic pressure gradients resulting from
bottom currents, however buoyancy effects may also
play a part.
The heat loss model determines the 2-D flow and
temperature field around the pipe. In the model,
turbulent as well as laminar flow characteristics
can be taken into account. From the resulting
temperature field the corresponding heat transfer
coefficient can be derived which is used to
compute the temperature drop along the line. To
that end a model has been applied which solves the
1-D heat transport equation for the oil. In the
model a non linear relation between heat transfer
coefficient and oil temperature is taken into
account, as well as effects of variation of
density and specific heat.
Computation of Flow and Temperature Field
--- ----------------------- -------------
To determine the flow and temperature field
around the pipe the full Navier-Stokes equations
representing flow of an incompressible fluid in
two dimensions, together with the heat transport
equation were used. The computation method is
based on the computer programme ODYSSEE [8], which
is a joint product of Delft Hydraulics Laboratory
and Laboratoire National dtHydraulique in Chatou,
France.
In the model the pipe, the gravel embankment and
the surrounding sea water are handled as a single
system for which the flow and temperature field is
computed. To schematize the fluid resistance of
the gravel material additional source terms have
been added to the momentum equations. The flow
through the gravel has been described in a
microscopical way, applying a general formulation
of
f
in
Uf
k
I
a
of
Darcys law (1856):
I/a
=k (1) (15)
which:
= filter velocity
= permeability coefficient
= hydraulic gradient
.
coefficient
In doing so the permeability is the counterpart
the momentum losses encountered by the flow.
For gravel the permeability usually varies between
0.001 and 1 m/s. Figure 11 [9] shows the
permeability for various porous materials. There
is a gradual transition from laminar (a=l) to
fully developed turbulent flow (a=2). The
permeability of the gravel material used for the
construction of the embankment varies between 0.14
and 0.22 m/s. Therefore it is liable that
turbulent flow characteristics will occur inside
the embankment. Rowever, the embankment will act
5 BOER, HULSBERGEN, RICHARDS, KLOK & BIAGGI
OTC 5294
.
as a sandtrap for the sediment which is transported 2 and 11. The corresponding dispersion coefficient
along the sea bottom. Therefore the permeability
varies between 0.0005 and 0.0025 m2/s. For the
of the embankment may change by sand infill. After initial permeability of the gravel embankment with
complete infill a permeability between 0.0002 and turbulent pore water flow a thermal conductivity
0.008 m/s is found. For these circumstances laminar
flow conditions will exist inside the gravel
coefficient of 2262 W/mC has been applied for the
gravel. In case of laminar flow the thexmal
embankment. conductivity of the fluid is 0.6 W/mC. The
corresponding thermal conductivity coefficient of
The influence of buoyancy is left out of the gravel after complete sand infill is 2.5 W/mC.
consideration. Although, especially for laminar
flow conditions the omitting of buoyancy is ODYSSEE computations have been carried out for
questionable, this simplification seems to be a number of situations in which the dominating
justified as a first approximation of the flow properties, such as: bottom current velocity,
field through the gravel embankment. By comparing embankment configuration, permeability of the
the order of magnitude of individual terms of the gravel embankment and oil temperature have been
momentum equations it may be demonstrated that the varied. Thus heat transfer coefficients were
flow in the gravel embankment is mainly governed computed which were used for the computation of
by hydraulic pressure gradients and fluid the resulting temperature profile along the line.
resistance. The influence of time dependent terms,
convective terms and viscous terms is only small. ODYSSEE Results
In Figure 12 the numerical grid of ODYSSEE is
---------------
shown. To reproduce strong temperature gradients Figure 14 illustrates the results of the initial
around the pipe, local grid refinements have been permeability k = 0.16 mjs, a bottom current
applied. In the horizontal x- direction grid sizes velocity u = 0.3 m/s, and an oil temperature of 60
vary between 0.05 and 0.5 m, and in the vertical C. In Figure 14a streamlines have been plotted
Y- direction between 0.025 and 1.5 m. for a range of discharge rates. The discharge
For the computations the usual type of flow through the gravel embankment is approximately
boundary conditions have been applied with 0.02 m2/s . At the lee side of the embankment a
velocity components specified at the inflow, no- recirculation zone exists as a result of negative
slip conditions at the bottom and free conditions hydraulic pressure gradients. In Figure 14b cor-
at the outflow. For the upper boundary, which was responding current velocities and directions are
selected at a level where the disturbances of the shown. Average filter velocities are 0.01 to 0.02
gravel embankment are subdued, th_erigid lid m/s. The temperature distribution in Figure 14c
assumption has been made. This means that the
boundary is treated as a streamline with free-slip
shows that heat is transported by convection in
the main current direction. No significant heating
conditions. These conditions were also imposed of the gravel material occurred. The heat loss is
along the boundary of the pipe. 1585 W/m. Figure 15 illustrates the results of the
At the inflow the temperature of the seawater is permeability after complete sand infill k = 0.002
specified. At the outflow, the bottom and the mjs. Discharge through the gravel embankment
upper boundary free conditions for the temperature (Figure 15a) is approximately 0.0005 m2/s. In
are imposed. For the temperature along the pipe a Figure 15b the flow pattern is shown. Filter velo-
gradient type of boundary condition has been used
cities are less than 0.001 m/s. In Figure 15c the
in which the heat released by the pipe is in computed temperature distribution is shown. The
equilibrium with the heat transport to the gravel absence of large filter velocities implies that
surrounding the pipe (Figure 13). the heat transport is dominated by conduction. The
For the thermal conductivity coefficient of the temperature of the gravel surrounding the pipe is
gravel material the assumption was made that it
behaves like a set of alternate strata of fluid
approximately 55C, with a heat loss of 250 W/m.
Figure 16, in which the temperature field for both
and solid in which the heat flux is parallel to cases are reproduced only for a small section of
the strata [10]: the gravel embankment, emphasizes once again the
fundamental difference between convection and
a
- Afluid gravel -
+ (1-n) \soud
(16) conduction dominated heat transport.
in which:
Computation of Temperature Profile
--- -------------- ---------------
A thermal conductivity of the gravel To determine the temperature profile along the
gravel =
A thermal conductivity of the pore water
line a computation method has been used which
fluid =
a
enables the solution of the one dimensional heat
= thermal conductivity of the solid transport equation for the oil. The computation
solid
n = porosity of the gravel
method is based on the computer programme PHOENICS
[12]. Because of
The thermal conductivity of the pore water
$he high Reynolds number of the
oil flow (Re = 10 ) turbulent mixing will
depends among other things on the thermal
establish an almost uniform velocity and tempera-
diffusion coefficient which can be related to
ture profile over the cross section of the pipe.
longitudinal and transverse dispersion. Most
Therefore the presence of a non-uniform
investigators have reported dispersion coeffi-
temperature distribution with temperature low at
cients in terms of the Peclet number. According
the wall and high in the core is disregarded.
to Perkins et al. [11] Peclet numbers for fully
By comparing the order of magnitude of the advec-
developed turbulent pore water flow vary between
tive and dispersive terms of the heat transport
equation it could be demonstrated that the
.-
13
6 BUCKLING OF HIGH TF@ERATURE OIL LINE COVERED BY GILA~L
OTC 5294
dispersive transport of heat is small compared to
the advective transport. The dispersive transport
was therefore neglected.
In essence the temperature drop along the line
depends on the initial temperature of the oil, the
thermal conductivity of the composite pipe, the
heat transport in the gravel surrounding the pipe
and the mass flow, density and specific heat of
the oil.
The density and specific heat are funct~ons of the
temperature and pressure. Since the actual pressure
variation was not computed the magnitude of the
density and specific heat were selected on the
basis of the average pressure of the oil flow
along the line. To that end the pressure drop was
estimated. This pressure drop is converted via the
wall shear stress and the internal shear stress of
the oil into heat. It Is estimated that a tempera-
ture rise of approximately 2C results from this
phenomenon.
For the computation of temperature profile the
line was divided into 8 sections. For each section
a representative thermal conductivity coefficient
of the pipe was established.
The initial oil temperature at NAB is 84c. For
the sections which preceed the section covered by
gravel a temperature drop of 2.7C has been compu-
ted. This temperature drop refers to a heat loss
of an unprotected pipe in open air or sea water of
5C. In Figure 17 the computed temperature profiles
for the pipeline covered by gravel are shown. The
temperature drop is strongly influenced by heat
transport through the gravel. For the initial
condition an oil temperature of 24C occurs at the
end of the section whereas after complete sand
infill this temperature will be 60c.
PULL-OUT TESTS
Aim of the Tests
------.-.--.--
The aim of the tests was the assessment of the
pull-out force P as function of depth of gravel
cover h under realistic conditions, simtlar to
those of the gravel covered pipeline offshore,
subjected to upheaval forces, and to provide an
experimental check on the computations made by LRS.
Geotechnical literature on pull-out force P
mainly refers to horizontal anchor plates in fine
granular soils with horizontal upper boundary.
Although the basic internal mechanisms are similar
for cylinders, the computed solutions or experi-
mental data of anchor plate pull-out forces cannot
be applied directly to a pipeline covered with
gravel. This limitation, and the lack of verified
methods to calculate pipeline pull-out force, were
important reasons for the full scale pipeline
pull-out tests which were carried out by Delft
Hydraulics Laboratory in close collaboration with
DeIft Soil Mechanics Laboratory.
Distinction is made between shallow and deep
anchors with respect to failure mechanisms. During
failure of shallow anchors continuous displacement
occurs in the entire zone between plate and
surface. On the other hand failure of deep anchors
is characterized as a deformation in a limited
zone around the plate, hardly affecting the
surface. In the present case only shallow be-

havi.ouris considered since the cover depth/pipe


diameter ratio h/D < 2.6 whereas deep behaviour
occures if h/D > 5 [13,14].
Pull-out of an infinitely long shallow horizontal
anchor plate with width D and burrial depth h is
illustrated in Figure 18. The pull out force P is
usually written as:
P = yDh(l +} q f) (17)
where P = pull-out force per unit length of
anchor plate
y = effective soil weight
D = width of anchor plate
h = burial depth
f = empirical factor incorporating
specific geotechnical characteristics.
On the right hand side of equation (17) the
first term represents the weight of the soil right
above the plate, whereas the second term with f
accounts for all other vertical force components
including the weight of both lateral triangles.
The literature for shallow anchors under a
horizontal surface shows a wide variation in
experimental f-values, ranging from f = 0.36
(loose sand, h/D = 1, [14]) to f = 0.68 (loose
sand, h/D = 1 [13]). In [15] full scale pull-out
tests with pipelines in sand produced f-values
ranging from f = 0.3 (loose sand) to f = 0.7
(density packed sand).
In the present case some important deviating
aspects are the non-horizontal upper surface and
the large gravel grains. In terms of equation
(17), the best set-up aims at finding P- and f-
values as a function of h, for a specific set of
Y-
and D values.
Test Set-u~ and Procedure
---------- --------------
In [16] a complete description is given of all
test set-up requirements and of the method how
these were met. The main aspects are briefly
mentioned here.
- Plane strain conditions must be obtained, i.e.
no gradients are allowed in stresses and
strains in a direction parallel to the pipe
axis, and no shear stress should occur in a
plane perpendicular to the pipe. The test set-
up designed to meet these requirements is
illustrated in Figure 19. In the 5 m wide Delta
Flume [171 the 2 m long central pipe serves as
the proper test section, while on both sides
dummy sections accommodatethe wall effects.
Moreover, steel plates on both ends further
reduce wall effects. During pull-out the three
pipe sections and end plates, which all are
separately suspended from a stiff steel lifting
beam, are lifted without mutual transfer of
vertical forces due to a special construction.
The complete history of pull-out force and
vertical displacement is measured by trans-
ducers incorporated in both suspension nodes
which pull the central pipe section.
- For the tests a smooth steel pipe was used with
outer diameter D = 422 mm. All pull-out forces
in this paper are nett values, i.e. after
eliminating the proper pipe weight.
7 BOER, HULSBERGEN, RICHARDS, KLOK & BIAGGI OTC 5294
The gravel has mass density p = 3090 kg/m3,
pore volume of some 42%, and immersed effective
soil weight y! = 11.4 kN/m3. Median grain size
is about 70 mm.
The cross section of the gravel cover has a
symmetric triangular form with a slope of 20.
For each test a fresh gravel cover was carefully
constructed in the dry flume, but the actual
pull-out tests were carried out in submerged
conditions in still water.
For the pull-out procedure electronic
controlled hydraulic jacks were used in two
subsequent phases: a force-control phase and a
displacement-control phase. During the first
phase the applied force was increased in small
steps, until the maximum load was achieved
after a lift of some 1 to 2 cm. Thereafter the
load was released and the continuous pull-out
phase started, applying a lift speed of 1 cm/
minute. In both phases a maximum pull-out force
was obtained by both suspension rod transducers,
giving four pull-out force values per test.
est Results
-----------
typical force/displacement curve ia presented in
igure 20, which was obtained during the
isplacement-control phase with depth of cover h =
.80 m. The curve is jagged due tothe
iscontinuous rearrangement of the covering
tones. After reaching a maximum value, the pull-
ut force gradually decreases as the pipe is
ulled towards the surface. The pull-out force
esults are summarised below.
depth of cover max. pull-out force
(average value)
h (m) P (kN/m)
0.60 6.32
0.80 9.38
~ooo---
11.95
1.15 15.05
h
Ising equation (17) the following f-values result:
depth of cover relative depth f-value
h(m) h/D eq. (17)
0.60 1.36 0.79
0.80 1.81 0.71
1.00 2.26 0.59
1.15 2.60 0.60
This apparent negative correlation of f and h/D
!ay tentatively be described as:
= 0.96 - 0.14 h/D (18)
~ith (18), equation (17) changes finally to the
ollowing empirical equation:
= y (Dh+ 0.96 h2 - 0.14 h3/D) (19)
which is presented in Figure 21 together with
experimental results. The gravel cover design
confirmed using these results.
CONCLUSIONS
The final gravel cover profile which was
eventually specified (Figure 22) was based on
the
was
both
the LRS and DHL studies and tests. Compared with
the profile suggested by pipeline engineering
studies previously performed, the overall gravel
requirements could be reduced from 145,000 T down
to 95,000 T by diminishing the depth of cover by
0.2 m in average along the line. Such a reduction
comes mainly from the DHL findings with regard to
the temperature drop to be considered along the
pipeline.
The LRS study, although leading to results quite
similar to those obtained by using previous
pipeline upheaval phenomenon analyses and models,
in terms of critical temperatures at which the
buckling occurs [4, 5 and 6], included a complete
post-upheaval and sensitivity analyses. These new
developments allowed a better understanding of the
pipeline behaviour throughout the buckling process
and hence a better evaluation of the margin of
safety required.
The DHL and LRS results were also used to
determine the gravel depth of cover required for
protection at the point where the 12 Alwyn-Ninian
pipeline crosses over the 36 FLAGS line on a
bridge made up of grout bags supports, by
considering the crossing as a large seabed
imperfection.
The DHL pull-out tests confirmed the assumptions
made in the LRS study regarding the upheaval
resistance provided by the gravel embankment.
The DHL theoretical heat transfer study was not
subjected to a specific test even at reduced
scale. However, it will be possible to verify this
analysis when the pipeline is put into operation
at the end of 1987. In any case for the first
year, the throughput of the 12 Alwyn-Ninian line
will not exceed 30% of the maximum design flow of
100,000 bbls/d which has been considered in DHL
temperature drop calculations.
As a result, should significant discrepancies
between the measured temperature values and those
predicted by the DHL model be observed at Ninian
Central, depth of cover could be increased in
1987/1988 by additional gravel-dumping to avoid
any risk of upheaval.
ACKNOWLEDGEMENTS
The Authors wish to thank Elf UK and Total Oil
Marine for permission to publish this paper.
REFERENCES
1. Groothuizen, A.G.M.: The Design and
Construction of Stabilization and Protection
of Subsea Pipelines and Cables up to 600 m
Water Depth. Paper OTC 5219, presented at the
1986 Offshore Technology Conference, Houston,
May 5-8, 1986.
8 BUCKLING OF HIGH TEMPEti~RE OIL LINE COVI?JU3D BY GWVEL OTC 5294
2. Kerr, A.D., On the Stability of the Railroad 10. Combarnous, M.A. and Bories, S.A.:
Track in the Vertical Plane. Rail Hydrothermal Convection in Saturated Porous
International, Vol. 5, Feb., 1974, pp 131-142. Media. Advances in Hydroscience, VO1. 10,
1975, Acadamic Press.
3. Kerr, A.D.: On Thermal Buckling of Straight
Railroad Tracks and the Effect of Track Length 11. Perkins, T.K. and Johnston, O.C.: 1A Review
on the Track Response. Rail International, of Diffusion and Dispersion in Porous Media.
Vol. 9, Septh., 1979, pp 759-768. Society of Petroleum Engineers Journal, March,
4.
1963.
Hobbs, R.E.: In-Service Buckling of Heated
Pipelines. Journal of Transportation 12. Rosten, H.I. and Spalding, D.B.: The Mathe-
Enigeering, ASCE, Vol. 110, No. 2, March,
1984.
matical Basis of the PHOENICS-EARTH Computer
Code. CRAM Ltd, London, January 1981.
5. Yun, H. and Kyriakides, S.: Model for Beam-
Mode Buckling of Buried Pipelines. Journal of 13. Das, B.M. and Seeley, G.R.: Uplift Capacity
Engineering Mechanics, ASCE, Vol. 112, No. 2, of Shallow Inclined Anchors.
Feb., 1985, pp. 235-253. Proc. 9 th Int. Conf. Soil Mech. Found. Engn,
6.
Vol. 1, pp. 463-466, Tokyo, 1977.
Friedman, Y.:
IfsomeAspects of the Design of
Buried Hot Pipelines. Offshore Oil and Gas 14. Rowe, R.K. and Davis, E.H.: The Behaviour of
Pipeline Technology Seminar, Paris, Jan., Anchor Plates in Sand.
1986. Geotechnique 32, no. 1, pp. 25-41, 1982.
7. Timoshenko, S.P., and Gere, J.M.: Theory of 15. Delft Soil Mechanics Laboratory: Site
Elastic Stability. Investigations and Behaviour of Pipelines.
2nd ed. Mc Graw-Hill, New York, 1961. Report CO-272040/75, 1985.
8. Officier, M.J., Vreugdenhil, C.B. and Wind, 16. Delft Hydraulics Laboratory: Full Scale Pull-
H.G.: Applications in Hydraulics of Numerical Out Tests for 12 Alwyn Oil Pipeline project.
Solutions of the Navier-Stokes Equations. Report M 2153, October 1985.
Recent Advances in Numerical Fluid Dynamics.
Taylor, C. (cd.), Swansea; Pineridge press,_
17. Kroezen, M., Vellinga, P., Lindenberg, J. and
1984. Burger, A.M.: Geotechnical and hydraulic
9.
aspects with regard to seabed and slope
Graauw, A. de, Van Der Meulen, T. and Van Der stability.
Does De Bye, M.: DeIft Hydr. Lab. publication series no. 272,
Design Criteria for Granular Filters. Delft June 1982.
Hydraulics Laboratory.
Publication no. 287, January 1983.
Ml
e,
F
M2
+~
e2
P
A w
6 A/p
- b
L
Fig. lSimply supported span.
---
/ N
0
A
/
Y
P
Y
P
/////////
(
/
4
Fig. 2Propped pipeline.
la
!
1.4
L/L
P/P
1.2
1.0
(h;w 3)I/~
4
3
2
1
0.8
0.6
0,4
0.2
0
j P
h F
100 1.0
0 2 4 6 8 10
,j
Fi g. 3Pr OP pr o-upheaval .
b
Jllw
&<
L I L*
4 k
Fi g. 5-Pr w ped pl pel l ne: upheaval hl st .r y.
80
60
40
20
0
0.8
0.6
0.4
0.2
0
m
T
(c)
I 1
-
L/L *
1.0 2.0 3.0
Fi g. 4-Pr OP post -upheaval .
5
:
s
.
(N~mm)
bar a pi pe
- * _.
1 2 3
*
j/j*
L
L/L*
*
Fi g. 6-Pr opp8d pl pal l n% t emper at ur ehi st or y r el at ed t o upheaval r edw .ame.
1
( (
-F
65
(;
m
40
20
/0[
k
j l )*
L(L*
-1-
.
q 0
Fi g. 7Pr .appedpl pel l ms t emper at ur eMst oq r el at ed t o pr op he!ght .
t
120-
loQ-
;()
m-
Go-
40-
20-
0.
0
,
5
t >
/0 /s-
U
Fi g. 9Pr opped pi pel i ne: pmt .uphew at ndnhnum t enmer at ur e
2s
A I
AL Ii-k
A
4.
3
$:
2-
/\/\200
Fi g. 8-Pr .spped pi pel l ns upheaval t er npar at .r e.
i-
& I r e P;(JC?
FI LI . i O-Pr Opped pi pel i ne: st r ess qt upheaval .
-?
g
I-2wK.l
w i f w m matarial
( unlass other P.% spacif iad)
turbulent flow
\\\ \\Y.\\\ \ \
\
\\\\\\ .N+N
uller 0,3-32 mm (P=15)
.-
D k (mI s)
====77
/
/
/
(i i-Wo kg
O(W-60 kg)
0.16
Fig. 11Permeability of porous media.
8.0(
6.4(
4.8C
3.20
1.60
0.00
1
-6.40 -4.8o -3.2o -1.60 0.00 1.60 3.2o 4.8o 6.40
Fig.
------+x
12Numerical grid ODYSSEE model.
AT= G-TU

Fig. 13Bounda~ conditions for temperature at pipe wall.


8.00
6.40
4.80
3.20
1.60
0.00
8.00
6.40
4.80
3.20
1.60
0.00
8.00
6.40
4.80
3.20
1.60
0.00
a. strcaml i ne pattarn
. . ...
0 -4.80 -3.20 -1 .60 0.00 1.60 3.20 4.80 6.40
.40 -4.80 -3.20 -1.60 0.00 1.60 X.20 4.80 6.40
c . t emper at ur e f i el d
----+=-~
2
-6.40 -4.80 -3.20 -1. S0 0.00 1.60 3.20 4.80 6.40
~x
Fi g. 14Resul t s 13DYSSEE. i ni t i al gr avel per meabi l i t y.
8.00
6.40
.
?
4.80
1.60
0.00
8.00
1
*
6.40
3.20
1.60
O.oc
8.OC
6.40
.
!
4.80
3.20
1.60
0.00
a. st r eam I i ne pat t er n

..=
6.40 4.80 -3.2o ~l . Ko 11.00 ,.60 S.20 , ~. 6.40
~x
c. t c smpemt um f i el d
-6.40 -<.80 -3.20 -1.60 0,00 ,.E.o 3.20 4.80 6.40
~x
Fi g. lSResults ODYSSEE permeability after sand I nf i l l .
h!
.
I
,:,,,,:..,:..,,,:
2:20 1:20 -0:20
I l-------d
0.80 1,80
x (m )
TEMPERATURE FIELD AROUND PIPE
Permeability coefficient cover material (D=O.08 1.2 mm): k= O.001 m/s
m
50 -65
m
35 50
m
20 - 35
m
5-20
m BELOW 5 :
.
0.75
0.45
0.15
-0.15
-2:20 - 1;20 -0.20 0.60
x (m )
l.ilo
TEMPERATURE FIELD AROUND PIPE
Permeability coefficient cover material (D= 10 127 mm): k=O.16 m\s
, 5.090-5.120
~ 5.060-5.090 -
m 5.030- 5.0s0
B 5. 000- 5. 030
Fig. 16Temperature field around pipe.
i ni t i al si t uat i on
- c ompl et e sand infi N
Fig. 17-Temper at ur e pr of i l e al ong pi pel i ne c over ed by gr avel .
m
n
D
w
Fig. 18-Pull-out of shallow anchor plate.
I
STIFF LIFTING 8EAM
I
hydr
j ac k
Fig. 19-Pull-out test setup.

10
P
(k Nl m)
8
6
7 %.
4
w
2
* *
00 0.1 0.2 0.3 0.4 0.5 0.6 0.7
ver t i c al di spl ac ement (m)
Fig. 20Typical force/displacement curve (h= 0.8 m).
16
14
12
10
(k Nt m)
t
8
6
4
2
0
f
ax par i mant al r esul t s
(over age ar x l sc at t er )
P= Y(Dh +0,96 h2-0.14h31D) q

(equat i on (19))
W-/
1 ,
4
~ h (m)
Fig. 21 Pul l .aut f or c el c ovar hei ght r esul t s.
t
12 ---mm-..---.
1
9
1.1
t
L.--. =..-%
I
GRAVSL
20 maximum
I
I
1.0i
b---------------
1
(145,000 T) :
g g -
I
I-----------
7
t
=0.8 -
9
I
L--------
L
:
(95,000 T)
g 0.7-
i
t
t
minimum h for mectxmical pr ot ec t i on
0.6
1
.___.._
c against dragging fishing gear and cables
z
G (25
H
0.4
I 1
0.3 --------
recommended profi k further to conventional engineering studies
I
mi l e r esul t i ng f r om LRS and DHL specific studies and tests
02
0.1
I
o
1
1 m 1 , #
o 1 2 3 4 5 6 7 8 9 10 11 12 13 14 ?5
~ flow direction
El
NAB
PLATFORM
1!
A
NINIAN
CENTRAL
PLATFORM
Fig. 22Final gravel cover profile.

You might also like