International Journal of Offshore and Polar Engineering (ISSN 1053-5381) http://www.isope.

org/publications
Copyright by The International Society of Offshore and Polar Engineers
Vol. 21, No. 4, December 2011, pp. 308315
Consideration for On-Bottom Stability of Unburied Pipelines Using a Dynamic
Fluid-Structure-Soil Simulation Program
Yinghui Tian, Mark J. Cassidy and Bassem S. Youssef
Centre for Offshore Foundation Systems, University of Western Australia, Perth, Western Australia, Australia
The modeling of pipe-soil interaction in current industry practice is still based on Coulomb friction models and passive
resistance. Analysis of the ultimate resistance capacity of the pipeline subjected to wave and current loading is taken as
the criterion for on-bottom stability design. This approach is widely used in industry, and considerable experience has been
accumulated. However, a sounder theoretical basis and more thorough understanding of pipe-soil behavior are still required.
This paper provides considerations and discussion regarding the analysis of unburied pipeline stability using an integrated
uid-structure-soil interaction program. It also emphasizes the advantages of stability design based on a displacement limit
state, a method encouraged in DNV (2007). The aim of this paper is to illustrate important issues for pipeline on-bottom
stability, and to highlight alternate results using this integrated modeling approach. The example analyses discussed include
optimizing the pipe weight for increased self-burial, the effect of a free span on pipe movements, and the inuence of
additional vertical load during a pipe installation.
NOMENCLATURE
A=wave amplitude
C
=(subscript) conjugate point
D=pipe diameter
D=stiffness matrix
D(w, 0) =spreading function
E =Youngs modulus of pipe steel
e
e
=(superscript/subscript) elastic
E =hydrodynamic force
F =force vector =]V, H]
T
E
l
=bounding surface
]
l
=bubble (yield) surface
g =plastic potential surface
H =horizontal force
h
=(subscript) horizontal
k =stiffness
M
=(subscript) centre point of bounding surface
m=aspect ratio of plastic potential surface
N
=(subscript) centre point of bubble surface

=(superscript/subscript) plastic
r =scale ratio of bubble surface to bounding surface
S =wave spectrum
T
peak
=peak spectral period
th =pipe thickness
U=displacement vector =]V, H]
T
u =horizontal displacement
V =vertical force
V
0
=size of bounding surface
W =self-weight
n =vertical displacement
p =shape parameter of bounding surface
A=increment
a =random phase
Received August 30, 2010; revised manuscript received by the edi-
tors April 20, 2011. The original version (prior to the nal revised
manuscript) was presented at the 20th International Offshore and Polar
Engineering Conference (ISOPE-2010), Beijing, June 2025, 2010.
KEY WORDS: Pipelines, on-bottom stability, pipe-soil interaction, Fou-
rier model, calcareous sand, force-resultant model.
q =wave elevation
0 =spreading angle
j=shape parameter of bounding surface
j
t
=aspect ratio of plastic potential surface
w=angular frequency
INTRODUCTION
One of the most fundamental engineering tasks of pipeline
design is to ensure on-bottom stability under the action of hydro-
dynamic loads. In many offshore engineering projects throughout
the world pipeline stabilization is a major cost driver. A typi-
cal example is the North West Shelf of Australia, where shallow
water, complex seabed soil characteristics due to the calcareous
sediments, and severe environmental loading from annual tropical
cyclones, can drive the Capital Expenditure (CAPEX) cost of sta-
bilization up to 30% of the total pipeline CAPEX (Brown et al.,
2002). These scenarios provide strong motivation for new models,
which potentially reduce conservatisms inherent in the traditional
design approaches.
Predicting on-bottom pipeline stability is complex with many
disciplines requiring integration, including soil constitutive
modeling, seabed liquefaction, scour and sediment transport, struc-
tural mechanics and prediction of ocean waves and hydrodynamic
loads. Considering this complexity, most pipelines are designed
using very simplistic methods. For instance, the traditional pipeline
stability design approaches, such as those found in early design
codes (DNV, 1981), use a simplistic Coulomb friction model to
describe pipe-soil behavior and adopt force balance methods to
ensure the pipe does not displace horizontally, i.e. the stability fail-
ure criterion is dened such that the hydrodynamic loading does
not exceed the soil resistance. More recent updates, such as DNV-
RP-E305 (1988) and DNV-RP-F109 (2007), retain simplistic sta-
bility charts, but highlight the advantages of dynamic time domain
and 3D stability analysis. A lateral displacement limit-state is set
as a failure criterion in this latter analysis methodology preferred
by DNV (2007).
Although recommended for some time, and considered the most
comprehensive method, dynamic lateral stability analysis is still
not widely used in practice. Trnes et al. (2009) ascribed the rea-
son to limited software availability and also limitations within
International Journal of Offshore and Polar Engineering, Vol. 21, No. 4, December 2011, pp. 308315 309
those available. However, the industrys desire for using inte-
grated uid-pipe-soil modeling in the time domain is clearly artic-
ulated by the comments of Zeitoun et al. (2008, 2009). This paper
describes a methodology for this type of analysis and presents
example results to highlight the potential advantages.
Pipe-Soil Modelling
For a pipe partially buried in calcareous sand, the 3 pipe-soil
force-resultant models initially proposed by Zhang (2001) and
Zhang et al. (2002a, b), and recently modied and advanced by
Tian and Cassidy (2008a; 2010, 2011), offer an alternative to the
friction-based approach. This has been previously demonstrated
by their implementation as user elements into the nite element
program ABAQUS (Tian and Cassidy, 2010). Named UWAPIPE,
these force-resultant models describe the behaviour of the entire
pipe-soil system within a plasticity framework.
The 4 components of a yield surface, ow rule, hardening law,
and a description of elastic behavior, constitute the formulation
of the UWAPIPE model. The parameters of these components
have been derived through physical geotechnical experiments of a
1-m prototype pipe using calcareous sands collected from offshore
Australia. The incorporation of numerous UWAPIPE nodal ele-
ments, together with the structural pipe elements in ABAQUS, has
made the on-bottom stability analysis of long pipes in the time
domain practical, whilst maintaining realistic behavior of the soil
(Tian and Cassidy, 2008a, b).
Hydrodynamic Loading
The Morison equation has been proven inadequate for calculat-
ing hydrodynamic loads acting on a pipe lying on the seabed, as it
is simply based on ambient ow velocity and time invariant coef-
cients. This is especially the case when a current velocity is super-
imposed on an irregular wave velocity. This is because the force
coefcients are highly dependent on the current-to-wave ratio and
the Keulegan-Carpenter number (Verley et al., 1989; Verley and
Reed, 1989; DNV, 2007). More sophisticated models, such as the
Wake model (Lambrakos et al. 1987; Soedigdo et al., 1999; Sabag
et al., 2000) and the Fourier models (Fyfe et al., 1987; Bryndum
et al., 1988; Jacobsen et al., 1988; Verley et al., 1989; Verley and
Reed, 1989), are preferred.
The Fourier methods provide a reliable and accurate prediction
of the hydrodynamic loads on an on-bottom pipe, as the Fourier
coefcients have been back-tted from an extensive experimen-
tal database (Fyfe et al., 1987; Bryndum et al., 1988; Jacobsen
et al., 1988; Verley and Reed, 1989). Youssef et al. (2010)
reviewed the Fourier methods and discussed their implementa-
tion into ABAQUS using the user subroutine DLOAD. Named
UWAHYDRO, this subroutine generates wave kinematics from
specied sea state spectra and evaluates the hydrodynamics based
on the Fourier equations and coefcients of Sorenson et al. (1986).
Adjustment to the loads due to relative movement of the pipe
can also be accounted for, as discussed in Youssef et al. (2010).
Reducing the load due to pipe embedment is particularly bene-
cial when integrated with a pipe-soil model that allows accurate
prediction of vertical movements.
Aims of This Paper
As the detailed technical content of implementing force-
resultant pipe-soil interaction and hydrodynamic models can be
found in the publications referenced within, this paper only
presents a brief review of the implementation and incorporation
of the computer programs UWAPIPE and UWAHYDRO into the
FE commercial package ABAQUS. This paper concentrates on
providing examples of how integrated dynamic analysis of on-
bottom pipes is of value to practitioners. These examples provide
additional insight into on-bottom stability design issues, such as
optimizing pipe weight, the effect of a free span on pipe move-
ments, and the inuence of additional vertical load during a pipe
installation. This represents the rst comprehensive investigation
of these issues using these pipe-soil and hydrodynamic loading
models.
The modeling technique of this paper does not account for
all of the interrelated effects. For instance, pore pressure gen-
eration around the pipe is not considered in the drained pipe-
soil model used, and the effect of scour and sediment transport
is not accounted for. Other signicant research in this regard
has recently been published. Notably Jeng (2003), Liu and Jeng
(2007), Gao et al. (2002, 2003, 2006) and Luan et al. (2008) have
conducted experimental and numerical analysis to investigate the
coupling effect of uid-pipeline-seabed interaction. However, no
method for integrating the effects of pore-pressure generation with
the degradation of soil stiffness due to hydrodynamic cyclic forces
on the pipe itself has been developed, at least within a framework
that allows 3D dynamic structural analysis.
OVERVIEW OF UWAPIPE FORMULATION
The 3 plasticity models introduced by Zhang (2001) and Zhang
et al. (2002a, b) have many common elements. Additional sophis-
tication to account for cyclic kinematic hardening is introduced
in the more advanced 2-surface bubble model (BM), which is
used in this papers study. A brief review of the numerical for-
mulation is provided here, with full details available in Tian and
Cassidy (2008a). Fig. 1 illustrates the sign convention of the load
and displacement of the pipe-soil interactive system.
The BM model is based within the framework of conven-
tional 2-surface isotropic and kinematic hardening plasticity the-
ory (Dafalias and Popov, 1975; Krieg, 1975; Dafalias, 1986; Muir
Wood, 2004). The constitutive formulation comprises an outer
bounding surface (written directly in load space), an inner yield
or bubble surface, isotropic and kinematic hardening laws, a
nonassociated ow rule, and an elastic behaviour denition. It is
mathematically described by the following equations.
Bounding Surface and Bubble Surface
The equations for the bounding and bubble surfaces are written
directly in terms of the load on the model as illustrated in Fig. 2
W
F
v
F
h
u
w
H
V
Iydrodynami% and
se!f2/eight !oads
$esu!tant for%es in
UW"0"1 mode!
Fig. 1 Forces on pipe and sign convention
310 Consideration for On-Bottom Stability of Unburied Pipelines Using a Dynamic Fluid-Structure-Soil Simulation Program
and described by:
E
l
=H j

V
V
0
+p

(V
0
V) =0 (1)
]
l
=H H
N
j

V V
N
rV
0
+
1 +p
2

1 +p
2
rV
0
(V V
N
)

=0 (2)
where E
l
= 0 is the outer bounding surface, and ]
l
= 0 is the
inner bubble surface; j, p are aspect ratios dening the surface
shape, and V
0
is the size of the bounding surface representing
the bearing capacity of the pipe under purely vertical load at the
current embedment; the subscript N denotes the bubble centre,
and r is the size ratio of the bubble to the bounding surface.
Hardening Law
Hardening of the surfaces occurs by (i) isotropic hardening of
the outer bounding surface and (ii) kinematic hardening of the
inner bubble surface. The former is directly correlated to the ver-
tical plastic displacement increment An

as a change in surface
size as:
AV
0
=
k
ve
k
v
k
ve
k
v
An

(3)
Aj=

D
An

(4)
where the superscript denotes a plastic component, is the
slope of j to n

, D is the pipe diameter, and k

ve
and k
v
are
the elastic and plastic vertical stiffness, respectively. Kinematic
hardening of the bubble surface is determined by:
AF
N
=AF
M
+

AV
0
V
0
+
Aj
j

1 r
r
(FF
N
) +AA(F
C
F) (5)
where the subscript M, N, C represent the bounding surface cen-
tre, bubble center and the conjugate point of the current force,
respectively (Fig. 2). F =]V, H]
T
is the force vector. The scalar
AA can either be explicitly evaluated according to the consis-
tency condition of the bubble surface or be implicitly iterated to
integrate the constitutive equations (Tian and Cassidy, 2010). The
bubble surface translates smoothly inside, but never intersects or
lies outside, the bounding surface.
Fig. 2 Model illustration
Flow Rule
Based on experimental evidence, a nonassociated ow rule is
required. It is formulated so that the plastic potential surface main-
tains a similar shape and position with the bubble surface (Fig. 2):
g =H H
N
j
t

V
V
0
+p

m
(V
0
V) =0 (6)
where j
t
and m are aspect ratios controlling the shape of the
plastic potential surface.
Elasticity
For increments inside the bubble surface, the elastic relationship
of the model is:
AF =

AV
AH

=D
e
AU
e
=

k
ve
0
0 k
he

An
e
Au
e

(7)
where U=]n, u]
T
is the displacement vector; k
he
is the horizon-
tal elastic stiffness; A represents an increment, and the superscript
e denotes an elastic component.
OVERVIEW OF UWAHYDRO FORMULATION
The operation of UWAHYDRO can be summarized by the fol-
lowing steps:
Simulate 3D ocean waves:
UWAHYDRO generates random sea state using both the direc-
tional wave spectrum S(w) and spreading function D(w, 0). Time
series of wave elevation q (as well as velocity and acceleration)
is linearly superimposed on a nite number of linear harmonic
waves:
A
i, ]
=

2A0
]
AwS(w
i
)D(w
i
, 0
]
) (8)
q(x, ,, t)
=
N

i=1
M

]=1
A
i, ]
cos(k
i
x cos 0
]
+k
i
, sin 0
]
w
i
t +a
i, ]
) (9)
where A
i, ]
is the wave amplitude for every individual harmonic
wave, and a
i, ]
is a random phase.
Generate wave kinematics at the pipeline level:
The wave velocity (U
n
) and acceleration (

U
n
) normal to the
pipeline are generated using linear wave theory and are described
by Lambrakos (1982) as:
U
n
(x, ,, t) =
N

i=1
M

]=1
o
i, ]
w
i

cosh(k
i
(J +z))
sinh(k
i
J)
cos(
i, ]
) cos(0
]
) (10)

U
n
(x, ,, t) =
N

i=1
M

]=1
o
i, ]
w
2
i

cosh(k
i
(J +z))
sinh(k
i
J)
sin(
i, ]
) cos(0
]
) (11)
where J is the water depth, z is the distance between the pipe and
the still water level, and
i, ]
=k
i
x cos 0
]
+k
i
, sin 0
]
w
i
t +a
i, ]
.
International Journal of Offshore and Polar Engineering, Vol. 21, No. 4, December 2011, pp. 308315 311
Hydrodynamic forces generation:
The hydrodynamic forces acting on the pipeline are then estimated
using the Fourier model, which assumes that a periodic variation
with a certain period T can be reproduced by superposition of a
number of sine waves with periods less than or equal to T . The
general expression of the periodic quantity E (t) is given:
E (t) =o
o
+
N

i=1
(o
i
cos(iwt) +l
i
sin(iwt)) (12)
where w=2r,T and o
i
and l
i
are the Fourier coefcients deter-
mined from back-tting of the ume experiments (Sorenson et al.,
1986; Jacobsen et al., 1988). The Fourier coefcients are func-
tions of the Keulegan-Carpenter number and the current-to-wave
velocity ratio.
Updating the hydrodynamic forces for pipe movement:
For each pipeline node, both horizontal and vertical movements
are used to calculate hydrodynamic correction factors. This step is
followed at each time increment during the pipeline analysis, and
the corrected hydrodynamics forces used in the following time
increment.
It should be highlighted that in deriving the coefcients of the
Fourier model, as detailed in Sorenson et al. (1986), the velocity
and acceleration for the free eld condition were used to nor-
malise the experimentally measured hydrodynamic loading with
the pipe. Therefore, the effect of a pipe on the ow is accounted
for within the coefcients.
IMPLEMENTATION OF UWAPIPE AND UWAHYDRO
With the UWAPIPE model simulating pipe-soil behaviour and
UWAHYDRO generating the hydrodynamic loading along the
pipeline, the integrated dynamic pipeline stability analysis pro-
gram works in the following way:
i. UWAHYDRO generates the sea state and prepares the
hydrodynamic loading database.
ii. At each new solution increment (or equilibrium iteration),
say at time t
n
, the UWAPIPE is provided with the incremental
displacements of the corresponding node it is attached to.
iii. This incremental displacement is used by each UWAPIPE
model to determine the corresponding internal forces, state vari-
ables and the tangent stiffness matrix at the end of the increment.
These are in turn delivered back to the main program (ABAQUS).
iv. ABAQUS assembles the global stiffness and calls DLOAD
to obtain the incremental load.
v. The overall solution of the system nonlinear equations is
solved, and the increment is ltered and passed to UWAPIPE for
the next iteration.
Steps ii to v are iterated in a Newton-Raphson scheme until
convergence is reached.
COMPUTATIONAL EXAMPLES
Three computational examples demonstrate the application of
the integrated hydrodynamic-pipe-soil program. In all of the
examples, a 1250-m-long pipeline is modeled with 250 beam ele-
ments and 251 UWAPIPE models attached to all the nodes, as
illustrated in Fig. 3. The pipe end conditions are considered to be
free of constraints. The structural and UWAPIPE parameters, and
their values assumed in this paper, are listed in Table 1. Also pro-
vided are details of the sea-state conditions. For all 3 analyses,
1 h (3600 s) of storm loads was generated. It is important to note
that these forces change along the pipe in both space and time.
For illustrative purposes, Fig. 4 shows the horizontal E
h
and the
Fig. 3 Computational model illustration
Parameter Description Value Unit
Pipe
D Pipe diameter 1.0 m
th Pipe thickness 0.03 m
E Youngs modulus 2.1 10
11
Pa
UWAPIPE
k
he
Horizontal elastic stiffness 10000 kN/m/m
k
ve
Vertical elastic stiffness 10000 kN/m/m
k
v
Vertical plastic stiffness 500 kN/m/m
m Aspect ratio 0.18
r Bubble size ratio 0.2
p Shape parameter 0.06
UWAHYDRO
H
s
Signicant wave height 12 m
T

eok
Peak spectral period 10 s
U
c
Current velocity 1 m/s
J Water depth 55 m

Note: T
peak
= peak spectral period used in denition of wave
spectrum.
Table 1 Computational parameters
Fig. 4 Hydrodynamic loads on Node 60
vertical E
v
loads history acting on an example node (Node 60 in
this case). On the other hand, Fig. 5 is a snapshot of the hydro-
dynamic loads along the pipeline at an example time (in this case
at 700 s).
The parameters adjusted in the 3 examples are (i) the self-
weight of the pipe, (ii) introduction of a free span of different
lengths in the middle of the pipe, and (iii) the vertical load the
pipe was subjected to during the pipe laying.
Example 1: Optimal Weight
Concrete coating is widely accepted as the primary measure
to improve on-bottom lateral stability for unburied pipelines (as
well as to protect against environmental damage). Too little con-
crete coating (i.e. pipe weight) results in an inadequate design,
whilst too much rapidly becomes uneconomic. Although more
pipe weight added during a design will prevent the pipeline mov-
ing laterally, it is not necessarily the optimal value for pipeline
312 Consideration for On-Bottom Stability of Unburied Pipelines Using a Dynamic Fluid-Structure-Soil Simulation Program
Fig. 5 Hydrodynamic loads along pipeline
self-burial under hydrodynamic loading (remembering that the
plasticity model provides for both isotopic and kinematic harden-
ing and can therefore penetrate further into the soil under hydro-
dynamic cyclic loads). Intuitively an optimal pipe weight may
exist, that provides for a maximum level of self-penetration under
the loading conditions. This concept is explored in this example.
Before the hydrodynamic loads are applied in the numerical
modeling, vertical loads are applied to the model and reduced to
the pipe self-weight. This process imitates the pipe laying pro-
cess followed offshore. This is conducted as soil underneath the
pipeline is subjected to larger vertical loads than the pipeline
submerged self-weight during pipe laying. In this example, this
load concentration factor was taken as 2. That is, the pipe was
loaded to twice the submerged self-weight (2W), then unloaded
to the self-weight (W), before hydrodynamic loading generated
from UWAHYDRO was applied.
To investigate the potential existence of an optimal self-weight,
18 analysis cases were conducted, with the self-weight W varying
from 5.06 kN/m to 8.46 kN/m. The pipeline conguration, after
the 1-h storm was applied, is illustrated in Fig. 6 for the case of
W =5.86 kN/m (noting that n represents the vertical penetration
and u the horizontal). It can be observed that, due to the spatial
variation in loads, the pipe displaces laterally and vertically by
different amounts.
The results are summarised in Fig. 7 by averaging the ver-
tical penetration and lateral displacement over all nodes along
the pipeline, as well as by plotting the minimum and max-
imum displacements at any position on the pipe as a range.
Fig. 7 shows that the horizontal displacement u monotonically
reduces as the self-weight increases. However, the vertical pene-
tration n increases with higher self-weight only to a self-weight
W = 7.06 kN/m. After that a heavier pipe actually has a lower
penetration after the storm loading is applied. This maximum ver-
tical penetration case (W =7.06 kN/m) can be considered to be
the optimal weight in this specic computation example.
Fig. 6 Pipeline conguration after 1-h storm (W =5.86 kN/m)
Fig. 7 Displacement range versus self-weight
This example shows that, by using the 2-surface bubble model,
pipelines with an optimal weight benet most from additional ver-
tical penetration while too much weight reduces the self-burial
effect. A traditional Coulomb friction approach (even with addi-
tional passive resistance with depth) is incapable of investigat-
ing this concept because it can only predict the critical resistance
capacity rather than providing extensive pipe-soil behaviour in the
time domain. By also setting a displacement limit criteria, this
example shows the potential for the integrated modeling to inves-
tigate issues such as the optimal weight.
Example 2: Free Span
This computational example investigates the inuence of a free
span on pipe stability. The free span is numerically simulated
by simply removing the UWAPIPE force-resultant models from
the attached structural nodes. It is noted that this unrealistically
assumes that the span is innitely deep, and that the sagging pipe
will not touch down on the bottom of the hole (Fig. 8).
The calculation model and the hydrodynamic loading used were
the same as in Example 1, with the self-weight assumed to be
W =7.06 kN/m. (See Fig. 3, Fig. 4 and Table 1.) Only one free
span is assumed, and this is located in the middle of the pipeline.
Sixteen computational cases were conducted, with the free span
length varying from 10 to 160 m. This represents the removal of
1, 3, 5. . .31 UWAPIPE elements attached to structural nodes, cen-
tred in the middle of the pipelines (remembering that UWAPIPE
models were simulating a pipe length of 5 m).
The nal pipeline congurations after the 1-h storm are illus-
trated in Fig. 9. The gure compares all of the calculation cases,
together with the original at sea-bed (the no free span case
from Example 1). We can observe that the longer the free span,
the more uneven the lateral displacement. With a free span, the
reaction to the hydrodynamic loading is redistributed along the
pipeline. Compared with no free span, the lateral displacement of
a pipeline with free span becomes more uneven. Another obser-
vation is that the free span only inuences the vertical penetration
Fig. 8 Pipeline free span illustration
International Journal of Offshore and Polar Engineering, Vol. 21, No. 4, December 2011, pp. 308315 313
Fig. 9 Pipeline conguration after 1-h storm
at the span shoulder to a marginal degree. Taking the extreme
160-m free span as an example, the shoulder length is shown to be
about 120 m (or 0.63 of the free span). It is interesting that close
to the free span (for example, 20 m in 160-m free span case),
the shoulder penetrated deeper whilst further away from the span
(100 m in 160-m free span case) the pipe was upheaved. This is
due to the load redistribution of the pipe beam elements close to
the free span and the beam bending rigidity.
By plotting the displacement history of the middle node (Node
126) in Fig. 10, we can observe that the middle of the free span
oscillates more with a longer free span. That is, the length of
free span increases the vibration of the spanning pipe. This shows
the importance of a vortex induced vibration (VIV) and fatigue
analysis, and the possible role that force-resultant modeling could
plan in accurately modeling a changing pipe location and pipe-
soil stiffness in the soil shoulders. Of course, this simple example
is not conducting such a rigorous free span investigation.
The maximum absolute values of each UWAPIPE vertical and
horizontal reaction force are shown for the length of pipe in
Fig. 11. These illustrate the envelope of maximum vertical V and
horizontal H values, rather than any of the reactions at any dis-
tinct time. Also shown, by the heavy lines, are the envelope of
the absolute maximum of the horizontal drag force E
h
and the
Fig. 10 Middle node (node 126) displacement history
Fig. 11 Maximum UWAPIPE reaction and combined loads
Fig. 12 UWAPIPE reaction and combined loads at 1000 s
self-weight W acting on each node along the pipeline. This gure
shows that the span shoulder, especially close to the free span,
bears signicantly more loads than other parts of the pipe. In
the case of the 160-m free span (as annotated on Fig. 11), the
20 m of span shoulder close to the free span shared more load.
This is consistent with the pipeline conguration as observation
in Fig. 9. Fig. 12 shows the reaction forces V and H of each
UWAPIPE model along the pipeline in comparison with the com-
bined applied loads (W E
v
, E
h
as demonstrated by the thick
lines).
Example 3: Pipe Laying
During pipe laying, the vertical force pushing the pipe into the
soil is larger than the eventual submerged weight. As highlighted
by Cathie et al. (2005), this load concentration factor (LCF) can be
further increased due to vessel motion and installation issues. An
LCF results in deeper initial pipe embedment, which is benecial
for pipe stability. This example explores the inuence of LCF on
the pipeline stability in a storm.
Numerically in the integrated program, the load concentration
factor (LCF) embodies the relationship of an initial bounding sur-
face to self-weight (i.e. V
0 initial
=LCF W and referring to Fig. 2
and Eq. 5 for denition). The UWAPIPE models the pipe-soil
undergoing vertical loads of V
0 initial
, and then unloaded to W
314 Consideration for On-Bottom Stability of Unburied Pipelines Using a Dynamic Fluid-Structure-Soil Simulation Program
Fig. 13 Pipe laying process in UWAPIPE modeling
before the hydrodynamic loads are applied. This process is illus-
trated in Fig. 13.
In this example the pipe-soil properties and hydrodynamic load-
ing are the same as in Example 1, with the self-weight assumed
to be W =5.06 kN/m (Fig. 3 and 4 and Table 1). Ten computa-
tional cases were carried out with the LCF varying from 2 to 20,
in increments of 2.
The pipeline congurations after the 1-h storm are illustrated in
Fig. 14. The trend shows that the pipe became more stable (less
lateral displacement) with increasing LCF.
Physically the pipe obtains deeper penetration when the soil
underneath the pipe undergoes larger pipe laying loads (higher
LCF). This establishes a larger capacity surface, and for the
UWAPIPE model this is numerically a larger initial bounding sur-
face (as dened by Eq. 5).
Fig. 15 depicts the resultant forces (V =WE
v
, H =E
h
) acting
on an arbitrary node as scattered dots. From this gure, the initial
bounding surfaces with larger LCF encompass the majority of the
resultant force dots. In these cases nearly no plastic behaviour
occurs, since most vertical and horizontal reactions are contained
within the bounding surface. The smaller initial surfaces require
expansion of the surface. This causes a reduction in pipe stiffness
and permanent plastic displacements result.
Fig. 14 Pipeline congurations
Fig. 15 Initial bounding surface and resultant forces on Node 200
CONCLUSION
This paper investigated 3 example questions related to pipeline
on-bottom stability using an integrated uid-structure-soil pro-
gram. A concept of optimal weight was proposed. Dened as
the pipe weight that causes maximum penetration in a storm, this
example highlighted the benet of an integrated model and the use
of a pipe-soil model that allows kinematic hardening under cyclic
storm loading. The effect of a free span on pipe movements in a
storm was also discussed. The benets of an integrated analysis
method were again highlighted, including a more accurate predic-
tion of the redistribution of forces in the span shoulders. In the
nal example, the benet of accounting for the increased vertical
load during pipe laying was quantied. Though it is difcult to
accurately predict the load concentration factor, the importance of
this value to the nal pipe movements was shown. It is hoped that
by articulating the potential benets of running 3D and dynamic
nite element analyses, their use will become a more common
practice in industry.
ACKNOWLEDGEMENT
This research is being undertaken within the CSIRO Wealth
from Oceans Flagship Cluster on Subsea Pipelines with funding
from the CSIRO Flagship Collaboration Fund. The second author
is the recipient of an Australian Research Council Future Fel-
lowship and holds the Chair of Offshore Foundations from The
Lloyds Register Educational Trust, an independent charity work-
ing to achieve advances in transportation, science, engineering
and technology education, training and research worldwide for the
benet of all. He gratefully acknowledges this support.
REFERENCES
Brown, NB, Fogliani, AG, and Thurstan, B (2002). Pipeline Lat-
eral Stabilisation Using Strategic Anchors, Presented at Soc
of Petroleum Engineers (SPE), Asia Pacic Oil and Gas Conf,
Melbourne.
Bryndum, MB, Jacobsen, V, and Tsahalis, DT (1988). Hydro-
dynamic Forces on Pipelines: Model Tests, Proc 7th Offshore
Mech and Arctic Eng Conf, pp 123131.
Dafalias, YF (1986). Bounding Surface PlasticityI. Mathe-
matical Foundation and Hypoplasticity, J Eng Mech, ASCE,
Vol 112, No 9, pp 966987.
Dafalias, YF, and Popov, EP (1975). A Model of Nonlinearly
Hardening Materials for Complex Loading, Acta Mechanica,
Vol 21, No 3, pp 173192.
International Journal of Offshore and Polar Engineering, Vol. 21, No. 4, December 2011, pp. 308315 315
DNV (1981). Rules for Submarine Pipeline Systems, http://www
.scribd.com/doc/7300254/DNV-Rules-for-Submarine-Pipeline-
Systems-1981.
DNV (1988). On-Bottom Stability Design of Submarine Pipelines,
DNV-RP-E305.
DNV (2007). On-Bottom Stability Design of Submarine Pipelines,
DNV-RP-F109.
Fyfe, AJ, Myrhaug, D, and Reed, K (1987). Hydrodynamic
Forces on Seabed Pipelines: Large Scale Laboratory Experi-
ments, Proc 19th Offshore Tech Conf, pp 123131.
Gao, FP, Gu, XY, and Jeng, DS (2003). Physical Modeling
of Untrenched Submarine Pipeline Instability, Ocean Eng,
Vol 30, No 10, pp 12831304.
Gao, FP, Gu, XY, Jeng, DS, and Teo, HT (2002). An Experimen-
tal Study for Wave-Induced Instability of Pipelines: The Break-
out of Pipelines, Applied Ocean Res, Vol 24, No 2, pp 8390.
Gao, FP, Jeng, DS, and Wu, YX (2006). Improved Analysis
Method for Wave-Induced Pipeline Stability on Sandy Seabed,
J Transp Eng, ASCE, Vol 132, No 7, pp 590596.
Jacobsen, V, Bryndum, MB, and Tsahalis, DT (1988). Prediction
of Irregular Wave Forces on Submarine Pipelines, Proc 7th
Offshore Mech and Arctic Eng Conf, pp 123131.
Jeng, DS (2003). Numerical Modelling for Non-Linear Wave-
Pipe-Seabed Interaction, Proc Computat Fluid and Solid Mech
2003, Vols 1 and 2, pp 13811385.
Krieg, RD (1975). A Practical Two Surface Plasticity The-
ory, Trans ASME, Series E, J Applied Mech, Vol 42, No 3,
pp 641646.
Lambrakos, KF, Chao, JC, Beckmann, H, and Brannon, HR
(1987). Wake Model of Hydrodynamic-Forces on Pipelines,
Ocean Eng, Vol 14, No 2, pp 117136.
Liu, HJ, and Jeng, DS (2007). A Semi-Analytical Solution
for Random Wave-Induced Soil Response and Seabed Lique-
faction in Marine Sediments, Ocean Eng, Vol 34, No 89,
pp 12111224.
Luan, M, Qu, P, Jeng, DS, Guo, Y, and Yang, Q (2008). Dynamic
Response of a Porous Seabed-Pipeline Interaction Under Wave
Loading: Soil-Pipeline Contact Effects and Inertial Effects,
Computat Geotech, Vol 35, No 2, pp 173186.
Muir Wood, D (2004). Geotechnical Modelling, Spon Press, Lon-
don.
Sabag, SR, Edge, BL, and Soedigdo, I (2000). Wake II Model
for Hydrodynamic Forces on Marine Pipelines Including Waves
and Currents, Ocean Eng, Vol 27, No 12, pp 12951319.
Soedigdo, IR, Lambrakos, KF, and Edge, BL (1999). Predic-
tion of Hydrodynamic Forces on Submarine Pipelines Using
an Improved Wake II Model, Ocean Eng, Vol 26, No 5,
pp 431462.
Sorenson, T, Bryndum, M, and Jacobsen, V (1986). Hydrody-
namic Forces on PipelinesModel Tests, Danish Hydraulic Inst
(DHI), Contract PR-170-185, Pipeline Res Council Int Catalog,
No L51522e .
Tian, Y, and Cassidy, MJ (2008a). Modelling of Pipe-Soil Inter-
action and Its Application in Numerical Simulation, Int J
Geom, ASCE, Vol 8, No 4, pp 213229.
Tian, Y, and Cassidy, MJ (2008b). A Practical Approach to
Numerical Modeling of Pipe-Soil Interaction, Proc 18th Int
Offshore and Polar Eng Conf, Vancouver, ISOPE, Vol 2,
pp 533538.
Tian, Y, and Cassidy, MJ (2010). The Challenge of Numerically
Implementing Numerous Force-Resultant Models in the Stabil-
ity Analysis of Long on-Bottom Pipelines, Comp and Geotech,
Vol 37, pp 216312.
Tian, Y, and Cassidy, MJ (2011). "A Pipe-Soil Interaction
Model Incorporating Large Lateral Displacements in Calcare-
ous Sand," J Geotech and Geoenviron Eng, ASCE, Vol 137,
No 3, pp 279287.
Tian, Y, Cassidy, MJ, and Gaudin, C (2010). Advancing Pipe-Soil
Interaction Models Through Geotechnical Centrifuge Testing in
Calcareous Sand, Appl Ocean Res, Vol 32, No 3, pp 294297.
Trnes, K, Zeitoun, H, Cumming, G, and Willcocks, J (2009).
A Stability Design RationaleA Review of Present Design
Approaches, Proc ASME 2009 28th Int Conf on Ocean, Off-
shore and Arctic Eng, Honolulu, CD-ROM.
Verley, RLP, Lambrakos, KF, and Reed, K (1989). Hydro-
dynamic-Forces on Seabed Pipelines, J Water Port C, ASCE,
Vol 115, No 2, pp 190204.
Verley, RLP, and Reed, K (1989). Use of Laboratory Force Data
in Pipeline Response Simulations, Proc Int Offshore Mech and
Arctic Eng Symp, pp 157165.
Youssef, BS, Cassidy, MJ, and Tian, Y (2010). Balanced
3D Modelling of the Fluid-Structure-Soil Interaction of an
Untrenched Pipeline, Proc 20th Int Offshore and Polar Eng
Conf, Beijing, ISOPE, Vol 2, pp 123130.
Zeitoun, H, Trnes, K, Cumming, G, and Brankovi, M (2008).
Pipeline StabilityState of the Art, Proc ASME 27th Int Conf
on Offshore Mech and Arctic Eng, Estoril, Portugal, CD-ROM.
Zeitoun, H, Trnes, K, Li, J, Wong, S, Brevet, R, and Will-
cocks, J (2009). Advanced Dynamic Stability Analysis, Proc
ASME 2009 28th Int Conf on Ocean, Offshore and Arctic Eng,
Honolulu, CD-ROM.
Zhang, J (2001). Geotechnical Stability of Offshore Pipelines in
Calcareous Sand, PhD Thesis, Univ of Western Australia.
Zhang, J, Stewart, DP, and Randolph, MF (2002a). Kinematic
Hardening Model for Pipeline-Soil Interaction Under Various
Loading Conditions, Int J Geomech, ASCE, Vol 2, No 4,
pp 419446.
Zhang, J, Stewart, DP, and Randolph, MF (2002b). Modelling of
Shallowly Embedded Offshore Pipelines in Calcareous Sand,
J Geotech and Geoenviron Eng, ASCE, Vol 128, No 5,
pp 363371.