Pipe-Soil Interaction Analysis with a Three-Dimensional Macroelement Model

Yinghui Tian Mark J. Cassidy

Centre for Offshore Foundation Systems, the University of Western Australia
Perth, Western Australia, Australia







ABSTRACT

Macroelement models have been increasingly utilized in offshore
engineering to simplify complex structure-soil interaction. Sometimes
termed force-resultant models, they attempt to relate resultant forces
directly to the corresponding displacements on the footing, all within a
plasticity framework. Although an increasing number of macroelement
models describing pipe-soil interaction have been developed, through a
combination of centrifuge tests, plasticity limit theory and finite
element analysis, these have in the main been for loading in two-
dimensions (usually vertical-horizontal, or V-H). This paper discusses
the use of such models in three dimensions (vertical, horizontal and
axial force space, or V-H-T). A simple extension to a previous two-
dimensional (2D) model in V-H space is suggested. The axial behavior
of the pipe-soil system, which has traditionally been treated as a purely
friction force, is linked with the plasticity model in the other two
directions (V-H). Details of the force-resultant model in V-H space are
first provided before a discussion of an appropriate method to extend
this to three-dimensions (3D) is given. Further, incorporation of the 3D
model into a structural finite element program is discussed. The 3D
model was integrated into a commercial finite element package as a
user defined element through the user subroutine interface. The
structural pipe is discredited as beam elements and the macroelement
model attached as one nodded element. With this integrated
methodology, pipeline analysis problems where the axial, as well as
vertical and horizontal behaviour, can be taken into account.
Calculation examples are carried out to demonstrate the performance of
the model and the application of the integrated program.

KEY WORDS: Pipe-soil interaction; 3D model; macroelement;
user-defined element

NOMENCLATURE

C
= (subscript) conjugate point
D = pipe diameter
D = stiffness matrix
e
= (superscript) elastic
F = bounding surface
f = bubble (yield) surface
g = plastic potential surface
H = horizontal force
h
= (subscript) horizontal
K = hardening (plastic) modulus
K
max
= maximum hardening modulus
k = stiffness
M
= (subscript) centre point of bounding surface
m = aspect ratio of plastic potential surface
N
= (subscript) centre point of bubble surface
p
= (subscript) plastic
q = the ratio of T to V
r = scale ratio of bubble surface to bounding surface
T = axial force
u = horizontal displacement.
V = vertical force
V
0
= size of the bounding surface
v = axial displacement
w = vertical displacement
= expansion coefficient of steel
= shape parameter of bounding surface
= plastic multiplier
= increment
= distance between two force points
max
= maximum distance between two force points
= displacement vector [ ]
T
u w, =
= parameter of the axial stiffness
= gradient of with increasing vertical embedment
= the axial stiffness of q to v
= shape parameter of the bounding surface
' = friction coefficient of the axial force
0
= with zero embedment
t
= shape parameter of plastic potential surface
= parameter of plastic modulus
= force vector [ ]
T
H V, =
v
= vertical stress
= axial stress
Proceedings of the Nineteenth (2009) I nternational Offshore and Polar Engineering Conference
Osaka, J apan, J une 21-26, 2009
Copyright 2009 by The I nternational Society of Offshore and Polar Engineers (I SOPE)
I SBN 978-1-880653-53-1 (Set); I SSN 1098-618
461

INTRODUCTION

Securing long-term energy supplies is vital to the worlds economy and
security. Our ability to realise the full potential of oil and gas reserves
relies on the development of viable platform and subsea pipeline
solutions to safely access and then transport the hydrocarbons. In the
latter, the analysis of interaction between a pipeline and the seabed it is
laid directly on is imperative. Simple empirical models are still widely
used to describe pipe-soil interaction, though it is increasingly
acknowledged that although they can predict the ultimate resistance
they are theoretically simplistic and inadequate in describe the entire
loading history (Cathie et al., 2005). For more thorough analysis of
discrete problems the finite element (FE) method is often utilised. With
complicated contact or interface elements describing the pipe-soil
interaction, long three-dimensional pipeline analyses are not
practicably feasible due to the extreme computation requirement. A
knowledgeable selection of constitutive relationships is also required if
results are to be consistent.

The macroelement approach, as highlighted by Cathie et al. (2005) and
Muir Wood (2004), provides an attractive alternative. Force-resultant
models attempts to relate the displacement of the footing in soil (or
segment of pipe in this case) directly to the resultant forces on it. Using
a plasticity framework the pipeline and the surrounding soil are treated
as a system, or a macroelement, with the entire force-displacement
behaviour encapsulated. Therefore, when incorporated within finite
element program of the pipelines structure analysis, special interface or
contact elements between the pipe and seabed are no longer required.
This significantly reduces computation expense.

Several pipe-soil interaction macroelement models have been
developed (Schotman and Stork 1987; Zhang 2001; Hodder et al. 2008;
Merifield et al. 2008; Randolph and White 2008). All describe coupled
vertical and horizontal force-displacement behaviour (2D behaviour in
V-H space), with plasticity properties derived from centrifuge
experiments, dedicated FE analysis and plasticity limit theory. 3D
models, that also incorporate axial (T) force, in the literature seem to be
limited to uncoupled spring (either elastic or elastoplastic) in V, H and
T.

However, Hededal and Strandgaard (2008) recently presented a fully
coupled 3D model based on existing numerical formulations. It used a
simplistic Coulomb friction model as the yield surface. By setting the
vertical component of the plastic displacement gradient as 1 or 0, the
model is able to use associated flow rule or decouple the vertical and
horizontal strength. The adoption of associated flow rule may be
questionable with the Coulomb friction model because the phenomenon
of vertical embedment due to lateral loading can not be simulated.
Decoupling the vertical and horizontal strength is also problematic
although it can model some circumstances.

This paper also presents a simple 3D pipe-soil model. It couples a 2D
V-H force-resultant model (originally developed by Zhang (2001) and
Zhang et al. (2002b) and recently modified and named the UWAPIPE
model by Tian and Cassidy (2008b)) with axial behaviour simulated as
a non-linear elastic-perfectly plastic spring. The latter couples stiffness
to the vertical resistance by providing a relationship between vertical
and axial load. Although the proposed model does not directly couple
the lateral and axial response, it does allow for combined V-H-T
analysis. In the large displacement analysis, the vertical, horizontal and
axial loads can cause not only displacement in its corresponding
direction but also displacements in the other two directions. Therefore,
this model is capable of predicting the 3D behaviour under
hydrodynamic and thermal loading. It is considered a first step in a
fully couple 3D model, though this would require experimental (or
numerical) investigation of the combined behaviour. Importantly,
however, the model allows more realistic analysis of long pipelines
when subjected to 3D loading, which could occur during the pipeline
and through the pipeline service.

3D PIPE-SOIL MODEL FORMULATION

The sign convention of the load and displacement of the pipe-soil
interactive system is illustrated in Fig. 1.

Fig. 1 Model illustration

Three 2D pipe-soil interaction models with increasing sophistication
were introduced by Zhang (2001) to describe the inter-coupled vertical
and horizontal behaviour (V-H) of pipe-soil interaction. Now named
the UWAPIPE models, in increasing complexity they are a single
surface Elastic-Plastic work-hardening model (Tian and Cassidy
2008a), a Bounding Surface model (Tian and Cassidy 2008c) and a
two-surface Bubble model that incorporates kinematic hardening
(Zhang et al. 2002a). The models are all based on centrifuge
experiments on calcareous sand. This latter bubble model is adopted in
this paper to describe the V-H behaviour. The constitutive model of the
bubble model comprises a bounding surface (written directly in V-H
space), an inner yield or bubble surface, a non-associated flow rule,
isotropic and kinematic hardening laws and an elastic behaviour
definition. The summary of the V-H behaviour illustrated in Fig. 2 and
described mathematically by the following equations (see Zhang 2001;
Tian and Cassidy 2008b for details).

H
bubble surface
plastic potential
V
bounding surface
M
N

max

Fig. 2 Model illustration

(i) Bounding surface
462

( ) 0
0
0
=

+ = V V
V
V
H F (1)

(ii) Yield surface (bubble surface)

( ) 0
2
1
2
1
0
0
=

+
+

=
N
N
N
V V rV
rV
V V
H H f

(2)

(iii) Flow rule

( ) 0
0
0
=

+ = V V
V
V
H H g
m
t N
(3)

(iv) Hardening law

Isotropic hardening of bounding surface

p
vp ve
vp ve
w
k k
k k
V &
&

=
0
(4)
D
w
p
+ =
0
(5)

Kinematic hardening of the bubble surface

( ) ( ) +

+ + =
C M N M N
V
V
&
&
&
& &

0
0
(6)

(v) Elastic behavior

= =

=
e
e
he
ve e e
u
w
k
k
H
V
&
&
&
&
&
&
0
0
D (7)

(vi) Kuhn-Tucker complementary condition

( ) ( ) 0 , , , , 0 , , , , 0
0 0
=
N N
V f V f & & (8)

(vii) Consistency condition

( ) 0 , , ,
0
=
N
V f
&
(9)

The plastic modulus K is interpolated according to relative distance of
current load point to the conjugate load point (see Fig. 2).

( )

+ =
max
max c c
K K K K (10)

As shown in Eq.6, the kinematic hardening law contains three parts: (1)
movement of the bounding surface centre; (2) expansion or contraction
of the bounding surface and (3) drag of the load point. The smooth
translation of the bubble surface and a guarantee that it will never
intersect the bounding surface is provided by the third term of the
kinematic hardening law in Eq.6. This requires the motion of the
bubble to travel toward the bounding surface parallel to the line
connecting the current load and the conjugate point, as depicted in Fig.
3 (Mrz 1967; Mrz et al. 1979; Mrz et al. 1981; Hashiguchi 1988;
Rouainia and Muir Wood 2000). Full details of implementing the V-H
UWAPIPE model are provided in Tian and Cassidy (2008b).


Fig. 3 Bubble motion
Though a simplistic assumption, the axial resistance (T) is reasonably
evaluated as proportional to the vertical force through a friction
coefficient ( ' ) (Cathie et al. 2005).

V T ' = (11)

For use in the proposed model for pipe-soil, it is reasonable to take '
as 0.88 as this is a recommended value for calcareous sands (BS8010
1993; Finch 1999) (remembering that the original UWAPIPE model
was based on centrifuge tests on calcareous sand (Zhang 2001)). The
yield surface of the axial and vertical force is illustrated in Fig. 4.
Combining this with the UWAPIPE surface in V-H space (Eq. 1), the
yield envelope of the 3D model can be developed. A schematic of its
shape is illustrated in Fig.5. This surface implicitly implies that there is
no direct coupling of lateral and axial load. This is not necessarily the
case. However, further testing is required to determine an appropriate
relationship.

V
T
1
'
(V
0
,0)
(0, )
0
'V

Fig. 4 Surface in T-V space


Fig.5 3D yield surface in V-H-T space

For behaviour inside of the bounding surface a relationship between the
axial force (T) and displacement (v) is required (similar to Eq. 7 for V-
463

H). In this simple model it is treated as elastic, perfectly plastic with the
initial elastic response following the non-linear behaviour observed by
Finch (1999). As illustrated in Fig.6, Finch (1999) reported an
experimental curve of
v
/ against axial displacement, where and
v
are the (axial) shear and vertical stress respectively. The
dimensionless ratio
v
/ , denoted here as q, can be directly relate to
the load ratio, as

V
T
q
v
= =

(12)

0.0
0.2
0.4
0.6
0.8
1.0
-2 0 2 4 6 8 10 12 14 16
1

v (mm)

0
'
(Fitted curve)
(Finch 1999)
0.0 0.2 0.4 0.6 0.8 1.0
0
400
800
1200
1600

/
v

Fig.6
v
/ versus axial displacement v

Because of the variation in contact area of the pipeline with the seabed
due to the embedment and subsequent heave (Barbosa-Cruz and
Randolph 2005), it is difficult to directly relate T with v. Instead, this
paper uses a nonlinear elastic equation to describe the relationship
between q and v (when q < ' ). For ease of numerical implementation,
the stiffness of q to v (named here) is assumed to decrease from an
initial value
0
to 0.

= =
'
1
0
q
dv
dq
(13)

Using the least square method with the above equation the best-fit to
the data of Finch (1999) are 1513.28 and 0.9 for the parameters of
0

and respectively. This is also shown in Fig. 6.

In summary, the relationship between T and v is simulated as a
nonlinear elastic, perfectly plastic spring. The yield capacity of T is
V ' and the axial stiffness calculated through , which is determined
according to the current load status.

INCORPORATION OF THE 3D MODEL INTO A
STRUCTURAL ANALYSIS FE PROGRAM

The main motivation for incorporating this V-T response with the V-H
model is to allow simulation of long sections of pipeline. With the
model described above, the seabed support to the pipeline can be
simplified as numerous discrete macroelements. This is illustrated in
Fig. 7. It follows the implementation of the UWAPIPE V-H model into
structural finite element program as described in the simple analysis of
pipelines under different spans by Tian and Cassidy (2008b) and in the
analysis of pipelines under 10-Year 3-hr storms by Tian and Cassidy
(2008c). However, in these other analyses no T-V response was
assumed, and this formulation is partly a response to allow for more
accurate modelling of the three-dimensional behaviour.

v
u
w
T
H
V
ep
&
&
&
&
&
&
D

Fig. 7 Supporting of the seabed discretized as footings

In the displacement-driven finite element programs, iteration schemes
are inevitably adopted in nonlinear problems to solve the basic
variables nodal displacement from the governing equations. The
nodal displacements are solved, in incremental form, are then used to
update the stress (force for the macroelement case) and state variables
through the constitutive integration. These two consecutive procedures
continue until the convergence of the equilibrium equations.

For explanation purposes, a Newton-Raphson iteration procedure for a
one-degree-of-freedom nonlinear system is taken as a simple
illustration of the iteration scheme (Fig. 8). The global tangent stiffness
matrix is assembled according to the structural stiffness and
macroelement stiffness provided by the footing module program at time
step
n
t . The displacement increment is accordingly solved, which is
filtered and input into the respective structural and footing modules to
integrate the constitutive laws and update the state variables. This
process is iterated until a proper convergence criterion is met (Chen and
Han 1988; Zienkiewicz and Taylor 2000; Fellin and Ostermann 2002).
The detailed description of the integration process of the macroelement
model can be found in Tian and Cassidy (2008a) and is therefore not
duplicated in this paper. In this paper the macroelements are simulated
as user-defined elements and the constitutive behaviour is written in
FORTRAN90 user subroutine as nodal elements to the structural beam
elements representing the pipe.

) 1 (
1 + n
U
n

) (
1
i
n+
U
1 + n
U

Fig. 8 Iteration scheme

COMPUTATIONAL EXAMPLES

Example 1

464

A numerical computation example has been conducted to explore
pipeline behaviour under thermal loads. A 1245m pipeline with 250
macroelements, all with 5m intervals, is modelled. Pipelines installed
from laying vessels are subjected to a larger vertical loads at the touch-
down area to the seabed during installation than its submerged self-
weight (Cathie et al. 2005; Tian and Cassidy 2008b). The implication
of this to the model is an initial W V >
0
, where
0
V defies the size of the
bounding surface (Eq. 1, Fig 4 and 5) and W is the submerged weight
of the pipeline. In this example, the initial
0
V is taken as twice the self
weight W, (i.e. the load concentration factor during the touch-down of
pipeline laying equals 2). The initial condition of all the 250 macro-
footings is illustrated in Fig. 9 with W=5kN/m and V
0
=2W=10kN/m.


Fig. 9 Illustration of initial condition

During the pipeline service hot hydrocarbons tend to heat up the pipe
and lead to an expansion. As hot fluids enter the cold pipeline, heat is
lost to the surrounding seawater and the fluids cools (Carr et al. 2003).
It takes about 2 hours to fully develop the steady temperature until hot
fluid is discharged at the far end. The temperature profile used in this
computational example is plotted in Fig. 10. The initial ambient
temperature is taken as 4C and the hot fluids heats the pipeline to
130C at the near end. After some time (step 37 in Fig. 10), the whole
pipeline has been heated up. Due to the heat loss, the far end of the
pipeline reaches 120C.


Fig. 10 Temperature profile

The computation parameters are listed in Table 1 (with the meaning of
all parameters given in the nomenclature).

The axial force of the pipeline during the heat loading process is shown
in Fig. 11. Negative values imply compressive axial force of the
pipeline. As the temperature advances along the pipeline, the axial
force is building up until a virtual anchor is formed close to the pipeline
centre (Carr et al. 2003).

Table 1 computational parameter

Parameter Dimension Value Unit
D L 1.0 m
E F/L
2
2.110
11
Pa
k
he
F/L
2
/L 10000 kPa/m
k
ve
F/L
2
/L 10000 kPa/m
k
vp
F/L
2
/L 500 kPa/m
m - 0.18 -
r - 0.2 -
t L 0.03 m
W F/L 5 kN/m
L/T 1.410
-5
m/C
- 0.06 -
- 0.9 -
- 0.65 -
0
L
-1
1513.28 1/m
0
- 0.4 -
t
- 0.6 -
' - 0.88 -
- 0.3 -

- 2.0 -

0 250 500 750 1000 1250
0
-500
-1000
-1500
-2000
-2500
-3000
19 -
18
17
16
15
14
13
12
11
10
9
8
6
7
5
3
4
2
1
A
x
i
a
l

F
o
r
c
e

(
k
N
)
Distance (m)
0
Virtual Anchor

Fig. 11 Axial force

As illustrated in Fig. 12, the axial displacement of the pipeline
increases as the temperature transits from the close end to the far end.
The pipeline extends toward the directions of the two ends, except in
the middle of the pipeline where a virtual anchor develops.

465

0 250 500 750 1000 1250
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Heating up
Heating up
A
x
i
a
l

d
i
s
p
l
a
c
e
m
e
n
t

(
m
)
Distance (m)
Virtual Anchor

Fig. 12 Axial displacement

The axial displacement versus temperature of some nodes along the
pipeline is plotted in Fig. 13. The pipeline expands with the increasing
temperature. The two ends extend by about 1 m while the middle of the
pipe almost remains at the original position.

0 20 40 60 80 100 120 140
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
Hot end
Node250
240
230
220
210
200
190
180
170
160
150
140
130
120
110
100
90
80
70
60
50
40
30
20
10
A
x
i
a
l
D
i
s
p
l
a
c
e
m
e
n
t

(
m
)
Temperature (
o
C)
Node1
Cold end

Fig. 13 Axial displacement vs. temperature

Fig. 14 depicts the relationship between axial forces and temperature of
all pipe elements. From this figure, we can see the axial forces
accumulate until the soil underneath has no more frictional capacity.
The axial force in the middle part of the pipeline is much larger than at
both ends, which have released the compression by axial movement.

0 20 40 60 80 100 120 140
-2800
-2400
-2000
-1600
-1200
-800
-400
0
Cold end
Middle
A
x
i
a
l
F
o
r
c
e

(
k
N
)
Temperature (
o
C)
Element249
240
230
220
210
200
190
180
170
160
150
140
130
120
110
100
90
80
70
60
50
40
30
20
10
Element1
Hot end

Fig. 14 Axial force vs. temperature

Example 2
Although there in no direct coupling of H and T in the current 3D
model (only through changes in V), horizontal loads cause axial
displacement and accordingly axial resistance. This example shows the
influence of axial soil resistance on the horizontal and vertical
behaviour. A 500m pipeline with 101 macroelements at 5m intervals
was modelled. The initial condition and calculation parameters are the
same as those in Example 1 (see Fig. 9 and Table 1). However, in this
example a point horizontal load (560 kN) was gradually exerted at
the middle node of the pipeline. Two computational cases were
analysed; one using the 2D model (V-H model) and the other using the
3D model described in this paper.

The final configuration of the pipeline, in all vertical, axial and
horizontal directions, is shown in Fig. 15, which suggests that the 2D
model (effectively setting the axial stiffness as zero) reports larger
displacements than the 3D model. The maximum horizontal
displacement at the middle of the pipeline of 2D model is about 1.4 that
of the 3D model. The horizontal force H of the macroelements is
illustrated in Fig. 16. The peak horizontal resistance of the UWAPIPE
macroemelents (at the pipe centre) was recorded to be higher in the 2D
analysis. This is because the pipe was required to penetrate further
(larger w in Fig. 15), thus expanding the bounding surface. The H of
the macroelement at the middle of the pipeline is about 1.43 times that
of the 3D model. Therefore, considering axial behaviour (i.e. including
the axial stiffness and resistance) reduces the horizontal displacement
and (peak) forces considerably. Neglecting the axial behaviour may
result in over estimate of the displacement and a conservative design.

0 10 20 30 40 50 60 70 80 90 100
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
-0.06
-0.03
0.00
0.03
0.06
0.09
0.06
0.03
0.00
0 50 100 150 200 250 300 350 400 450 500
3D model
2D model
u

(
m
)
Node
3D model
2D model
v

(
m
)
3D model
Distance (m)
w

(
m
)
2D model

Fig. 15 Configuration

466

0 20 40 60 80 100
-15
-10
-5
0
5
10
15
20
25
0 100 200 300 400 500
3D model
Distance (m)
H

(
k
N
)
Node
2D model

Fig. 16 H of all macroelements

CONCLUSION

This paper presents a simple extension to the force-resultant modelling
of pipes to allow for three-dimensional pipe-soil behaviour. The two-
dimensional V-H relationship is based on the force-resultant model
developed from centrifuge tests (Zhang 2001) and the V-T relationship
on Coulomb friction behaviour (Finch 1999). The friction coefficient of
the axial force ' is highly depended on the nature of the soils. In this
paper, ' is taken as 0.88 as the V-H model was developed for the
calcareous sand. However, the value should be chosen with caution. As
a simple extension to the 2D model, there is no direct coupling between
the behaviour of H and T. However, it provides a framework for further
developments, including fully coupled V-H-T behaviour. The model
has also been shown to be easily incorporated into structural finite
elements analyses.

The first computational example was shown to simulate the behaviour
of a pipeline shallowly embedded on the seabed during the heating up
process. The temperature profile is taken only for explanatory purpose
and may vary remarkably between different oil and gas fields. Precise
temperature description should be accordingly adopted for a specific
place. The second was to the influence of the axial behaviour on the
other two directions behaviour. The 3D pipe-soil interaction model has
the potential to investigate three dimensional problems, such as non-
coherent storm loading, walking, buckling and pipe-laying.

ACKNOWLEDGEMENT

The 2D pipe-soil model used in this paper is based on the PhD thesis of
Dr. Jianguo Zhang. The authors thank him for his discussion, help and
encouragement. The constructive suggestion and references provided
by Prof. Dave White are also appreciated. This research is being
undertaken within the CSIRO Wealth from Oceans Flagship Cluster on
Subsea Pipelines with funding from the CSIRO Flagship Collaboration
Fund.

REFERENCES

Barbosa-Cruz, ER, and Randolph, MF (2005). "Bearing capacity and
large penetration of a cylindrical object at shallow embedment."
Proc. Int. Symposium on Frontiers in Offshore Geotechnics - ISFOG
2005, Perth, Australia, pp 615-621.
BS8010 (1993). "Code of Practice for Pipelines. Part3. Pipelines
Subsea: Design, Construction and Installation." British Standard
Institution.
Carr, M, Bruton, D, and Leslie, D (2003). "Lateral buckling and
pipeline walking, a challenge for hot pipelines." Offshore Pipeline
Technology Conference, Amsterdam.
Cathie, DN, Jaeck, C, Ballard, J-C, and Wintgens, J-F (2005). "Pipeline
geotechnics state-of-the-art." Proc. of the Int. Symp. on the
Frontiers in Offshore Geotechnics: ISFOG 2005, Perth, Australia, pp
95-114.
Chen, WF, and Han, DJ (1988). Plasticity for structural engineers,
Springer-Verlag, New York.
Fellin, W, and Ostermann, A (2002). "Consistent tangent operators for
constitutive rate equations." International Journal for Numerical and
Analytical Methods in Geomechanics, Vol 26, No 12, pp 1213-1233.
Finch, M (1999). "Upheaval Buckling and Floatation of Rigid
Pipelines:The Influence of Recent Geotechnical Research on the
Current State of the Art." OTC, Houston, Texas.
Hashiguchi, K (1988). "A mathematical modification of two surface
model formulation in plasticity." International Journal of Solids and
Structures, Vol 24, No 10, pp 987-1001.
Hededal, O, and Strandgaard, T (2008). "A 3D elasto-plastic spring
element for pipe-soil interaction analysis." Offshore Pipeline
Technology Conference, Amsterdam.
Hodder, MS, Cassidy, MJ, and Barrett, D (2008). "Undrained response
of shallow pipelines subjected to combined loading." BGA
International Conference on Foundations ICOF 2008, Dundee,
Scotland.
Merifield, R, White, DJ, and Randolph, MF (2008). "The ultimate
undrained resistance of partially embedded pipelines." Geotechnique,
Vol 58, No 6, pp 461-470.
Mrz, Z (1967). "On the description of anisotropic work-hardening."
Journal of the Mechanics and Physics of Solids, Vol 15, No 1, pp
163-176.
Mrz, Z, Norris, VA, and Zienkiewicz, OC (1979). "Application of an
anisotropic hardening model in the analysis of elasto-plastic
deformation of soils." Geotechnique, Vol 29, No 1, pp 1-34.
Mrz, Z, Norris, VA, and Zienkiewicz, OC (1981). "An anisotropic,
critical state model for soils subject to cyclic loading."
Geotechnique, Vol 31, No 4, pp 451-469.
Muir Wood, D (2004). Geotechnical modelling, Spon Press, London.
Randolph, MF, and White, DJ (2008). "Upper-bound yield envelopes
for pipelines at shallow embedment in clay." Geotechnique, Vol 58,
No 4, pp 297-301.
Rouainia, M, and Muir Wood, D (2000). "A Kinematic hardening
constitutive model for natural clays with loss of structure."
Geotechnique, Vol 50, No 2, pp 153-164.
Schotman, GJM, and Stork, FG (1987). "Pipe-Soil Interaction: A
Model for Laterally Loaded Pipelines in Clay." OTC, Houston,
Texas.
Tian, Y, and Cassidy, MJ (2008a). "Explicit and implicit integration
algorithms for an elastoplastic pipe-soil interaction macroelement
model." the 27th International Conference on Offshore Mechanics
and Arctic Engineering, Estoril, Portugal.
Tian, Y, and Cassidy, MJ (2008b). "Modelling of Pipe-Soil Interaction
and Its Application in Numerical Simulation." Int. J. Geom., Vol 8,
No 4, pp 213-229.
Tian, Y, and Cassidy, MJ (2008c). "A Practical Approach to Numerical
Modeling of Pipe-Soil Interaction." The 18th International Offshore
(Ocean) and Polar Engineering Conference, Vancouver, BC, Canada.
Zhang, J (2001). "Geotechnical stability of offshore pipelines in
calcareous sand," Thesis (Ph.D.), University of Western Australia.
Zhang, J, Stewart, DP, and Randolph, MF (2002a). "Kinematic
Hardening Model for Pipeline-Soil Interaction under Various
Loading Conditions." The International Journal of Geomechanics,
Vol 2, No 4, pp 419-446.
Zhang, J, Stewart, DP, and Randolph, MF (2002b). "Modelling of
467

shallowly embedded offshore pipelines in calcareous sand." Journal
of Geotechnical and Geoenvironmental Engineering, Vol 128, No 5,
pp 363-371.
Zienkiewicz, OC, and Taylor, RL (2000). The finite element method-
Volume 2: Solid Mechanics, Butterworth-Heinemann, Oxford;
Boston.


468