NUMERICAL MODELLING OF PIPE-SOIL …...Numerical modelling of pipe-soil interactions i ABSTRACT This...
Transcript of NUMERICAL MODELLING OF PIPE-SOIL …...Numerical modelling of pipe-soil interactions i ABSTRACT This...
NUMERICAL MODELLING OF PIPE-SOIL
INTERACTIONS
By
SANTIRAM CHATTERJEE
B.Eng., M.Tech.
This thesis is presented for the degree of Doctor of Philosophy
of The University of Western Australia
Centre for Offshore Foundation Systems
School of Civil and Resource Engineering
September 2012
Numerical modelling of pipe-soil interactions
i
ABSTRACT
This thesis described research into the pipe-soil interaction forces during large
movements of deep-water pipelines, using numerical methods. Vertical penetration,
lateral break-out and steady-state lateral resistances were investigated with the help of
numerical models using a large deformation approach implemented within the
ABAQUS finite element software and sophisticated soil constitutive models.
The large deformation finite element methodology is based on a periodic
remeshing and interpolation technique which was developed for this research to
incorporate the effects of changes in the strength and geometry of the seabed during
large movements of the pipelines. Soil constitutive model that accounts for strain rate
effects and remoulding were implemented to simulate realistic behaviour. Coupled
pore-fluid stress analyses were also carried out using the modified Cam Clay plasticity
model to investigate the effects of drainage and consolidation on interaction forces.
The initial vertical penetration of a seabed pipeline is an important parameter for
design of these pipes against lateral buckling and other design conditions. The
penetration rate and strain softening have marked effects on the resistance experienced
during vertical penetration. A simple elastic perfectly plastic soil constitutive model was
modified to incorporate these effects to identify the equivalent shear strength of the soil.
A parametric study considering wide range of parameters was conducted and the results
were unified when the vertical penetration resistance was normalised using this
equivalent shear strength. Simplified equations are presented for ease of application.
Lateral pipe-soil interactions were also studied to observe the effects of the
initial embedment and different pipe weights. Two stages of lateral interaction are dealt
in this research. Firstly, the initial breakout resistance was investigated through large
deformation finite element analyses and also limit analysis using the software OxLim.
Numerical modelling of pipe-soil interactions
ii
Results were presented in terms of plastic failure envelopes in the V-H load space for
different initial embedments. The steady state lateral residual resistance was then
studied using large deformation analyses and an appropriate soil constitutive model. It
was found that steady state resistance is achieved after a lateral displacement of
typically three times the pipe diameter or less, even if the soil berm continues to grow in
size. The increase in berm size is counteracted by a reduction in the soil strength due to
accumulation of plastic strain. The steady state residual friction factor was linked to a
new ‘history’ parameter – termed the effective embedment – in a simple manner,
regardless of the other soil and pipeline parameters.
Finally, coupled consolidation analyses using the modified Cam Clay plasticity
model was carried out to explore the effects of consolidation on penetration and
breakout resistances. Elastoplastic modelling of consolidation beneath partially
embedded pipes was first done to study the pore pressure dissipation time history,
allowing the rate of build-up of pipe-soil resistance to be assessed. The effects of
penetration rate on the vertical resistance were examined and backbone-type curves
showing drained, undrained and partially drained behaviour were presented. It was also
shown that in contractile soil consolidation beneath a shallowly embedded pipe changes
the soil strength significantly and has a marked effect on the lateral breakout resistance.
Again, the results were presented in a normalised manner suite to application in design.
Table of contents
iii
TABLE OF CONTENTS
ABSTRACT……………………………………………………………………………..i TABLE OF CONTENTS……………………………………………………………...iii LIST OF FIGURES.......................................................................................................vii ACKNOWLEDGEMENTS………………………………………………….………xiii CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW
1.1 RESEARCH MOTIVATION................................................................................1-1
1.2 LITERATURE REVIEW......................................................................................1-6
1.2.1 Theoretical solutions ...................................................................................1-7
1.2.2 Numerical analyses .....................................................................................1-9
1.2.3 Design practice..........................................................................................1-13
1.2.4 Model testing observations .......................................................................1-14
1.3 RESEARCH GOALS..........................................................................................1-15
1.4 METHODOLOGY ..............................................................................................1-17
1.5 OUTLINE............................................................................................................1-17
1.6 REFERENCES ....................................................................................................1-20
CHAPTER 2 LARGE DEFORMATION FINITE ELEMENT METHODOLOGY
2.1 NON-LINEAR FINITE ELEMENT ANALYSES................................................2-1
2.2 ANALYSIS PROCEDURES.................................................................................2-2
2.2.1 Lagrangian approach...................................................................................2-2
2.2.2 Eulerian approach........................................................................................2-2
2.2.3 Arbitrary Lagrangian Eulerian approach ....................................................2-2
2.3 RITSS APPROACH ..............................................................................................2-3
2.4 IMPLEMENTATION IN ABAQUS .....................................................................2-4
2.4.1 Superconvergent Patch Recovery – SPR ....................................................2-5
2.4.2 Steps of LDFE analysis ...............................................................................2-6
2.4.3 Effects of strain rate and softening............................................................2-10
2.4.4 Modified Cam Clay model........................................................................2-11
2.5 REFERENCES.....................................................................................................2-12
Table of contents
iv
CHAPTER 3 THE EFFECTS OF PENETRATION RATE AND STRAIN SOFTENING ON THE VERTICAL PENETRATION RESISTANCE OF SEABED PIPELINES
3.1 INTRODUCTION .................................................................................................3-1
3.2 FINITE ELEMENT MODEL ................................................................................3-2
3.2.1 Mesh, boundary conditions and material model .........................................3-2
3.2.2 Strain rate and strain softening....................................................................3-4
3.2.3 Validation of finite element model .............................................................3-5
3.3 PARAMETRIC STUDY .......................................................................................3-7
3.3.1 Effect of unit weight of soil ........................................................................3-8
3.3.2 Geotechnical resistance.............................................................................3-12
3.3.3 Effect of normalised velocity, refγ/Dpv & .................................................3-13
3.3.4 Effect of rate parameter μ .........................................................................3-15
3.3.5 Effect of sensitivity ...................................................................................3-17
3.3.6 Effect of ductility parameter ξ95................................................................3-19
3.3.7 Combining effects of strain rate and softening parameters.......................3-21
3.3.8 Effect of variation of soil shear strength profile .......................................3-23
3.4 SOIL FLOW PATTERN .....................................................................................3-23
3.5 CONCLUDING REMARKS...............................................................................3-24
3.6 REFERENCES ....................................................................................................3-26
CHAPTER 4 LARGE LATERAL MOVEMENT OF PIPELINES ON A SOFT CLAY SEABED: LARGE DEFORMATION FINITE ELEMENT ANALYSIS
4.1 INTRODUCTION .................................................................................................4-1
4.2 SOIL CONSTITUTIVE MODELLING ................................................................4-4
4.3 TYPICAL FINITE ELEMENT MESHES.............................................................4-5
4.4 IDEAL SOIL CASE ..............................................................................................4-8
4.5 REALISTIC SOIL CASE ....................................................................................4-12
4.6 EFFECTIVE EMBEDMENT APPROACH........................................................4-20
4.7 CONCLUDING REMARKS...............................................................................4-23
4.8 REFERENCES ....................................................................................................4-24
CHAPTER 5 MODELLING LATERAL PIPE-SOIL INTERACTIONS
Table of contents
v
5.1 INTRODUCTION .................................................................................................5-1
5.2 PARAMETRIC STUDY........................................................................................5-2
5.2.1 Input parameters..........................................................................................5-2
5.2.2 Typical results – w/D = 0.3 .........................................................................5-3
5.2.3 Initial yield envelopes and breakout resistance...........................................5-6
5.2.4 Residual friction factor..............................................................................5-11
5.3 EFFECTIVE EMBEDMENT APPROACH........................................................5-17
5.4 ASSESSMENT OF THE FULL H/V RESPONSE .............................................5-24
5.5 CONCLUDING REMARKS...............................................................................5-27
5.6 REFERENCES.....................................................................................................5-27
CHAPTER 6 BREAKOUT BEHAVIOUR OF PARTIALLY EMBEDDED PIPES IN UNIFORM CLAY USING LIMIT ANALYSIS
6.1 INTRODUCTION .................................................................................................6-1
6.2 METHODOLOGY.................................................................................................6-2
6.3 YIELD ENVELOPES............................................................................................6-3
6.4 INTERFACE MODIFICATION ...........................................................................6-8
6.5 EFFECT OF SOIL WEIGHT ..............................................................................6-17
6.6 EFFECT OF SOIL HEAVE.................................................................................6-20
6.7 CONCLUDING REMARKS...............................................................................6-24
6.8 REFERENCES.....................................................................................................6-25
CHAPTER 7 ELASTOPLASTIC CONSOLIDATION BENEATH SHALLOWLY EMBEDDED OFFSHORE PIPELINES
7.1 INTRODUCTION .................................................................................................7-1
7.2 NUMERICAL METHODOLOGY........................................................................7-2
7.3 MATERIAL MODEL............................................................................................7-3
7.4 UNDRAINED PENETRATION RESPONSE ......................................................7-6
7.5 CONSOLIDATION RESPONSE..........................................................................7-7
7.5.1 Pore pressure dissipation.............................................................................7-7
7.5.2 Consolidation settlement ...........................................................................7-13
7.6 CONCLUDING REMARKS...............................................................................7-15
7.7 REFERENCES.....................................................................................................7-15
Table of contents
vi
CHAPTER 8 EFFECTS OF CONSOLIDATION ON PENETRATION AND LATERAL BREAKOUT RESISTANCES 8.1 INTRODUCTION .................................................................................................8-1
8.2 MODEL DESCRIPTION ......................................................................................8-2
8.3 SOIL PARAMETERS ...........................................................................................8-4
8.4 EFFECTS OF LOADING RATE ..........................................................................8-4
8.4.1 Penetration resistance..................................................................................8-4
8.4.2 Consolidation settlement...........................................................................8-11
8.4.3 Pore pressure dissipation after undrained penetration ..............................8-13
8.5 LATERAL BREAKOUT RESISTANCE ...........................................................8-15
8.5.1 Background ...............................................................................................8-15
8.5.2 Unconsolidated undrained yield envelopes...............................................8-16
8.5.3 Consolidated undrained yield envelopes...................................................8-17
8.5.4 Simple equation fit ....................................................................................8-19
8.6 CONCLUDING REMARKS...............................................................................8-21
8.7 REFERENCES ....................................................................................................8-22
CHAPTER 9 CONCLUDING REMARKS
9.1 ORIGINAL CONTRIBUTIONS ...........................................................................9-1
9.1.1 Vertical penetration.....................................................................................9-1
9.1.2 Lateral pipe-soil interactions.......................................................................9-2
9.1.3 Coupled consolidation analyses ..................................................................9-4
9.2 LDFE ANALYSIS IN DESIGN............................................................................9-5
9.2.1 LDFE in support of simplified design method............................................9-5
9.2.2 LDFE directly applied in design .................................................................9-5
9.3 LIMITATIONS AND FUTURE RESEARCH......................................................9-6
9.3.1 Dynamic effects ..........................................................................................9-6
9.3.2 Full 3D model .............................................................................................9-7
9.3.3 Whole life behaviour...................................................................................9-7
List of figures
vii
LIST OF FIGURES CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW
Figure 1.1 Existing and proposed pipelines at the North West Shelf of Australia ........1-2
Figure 1.2 Controlled lateral buckling of an on-bottom pipeline (Jayson et al., 2008) .1-3
Figure 1.3 Soil deformation mechanism during lateral pipe movement from centrifuge
modelling (Dingle et al., 2008) ......................................................................................1-4
CHAPTER 2 LARGE DEFORMATION FINITE ELEMENT METHODOLOGY
Figure 2.1 Overview of RITSS approach (Hu and Randolph, 1998a) ...........................2-4
Figure 2.2 Superconvergent Patch Recovery (Zienkiewicz and Zhu, 1992) .................2-5
Figure 2.3 Six noded triangulare elements in ABAQUS ...............................................2-7
Figure 2.4 Determination of position of new Gauss point;............................................2-8
Figure 2.5 Implementation of RITSS in ABAQUS .....................................................2-10
CHAPTER 3 THE EFFECTS OF PENETRATION RATE AND STRAIN SOFTENING ON THE VERTICAL PENETRATION RESISTANCE OF SEABED PIPELINES
Figure 3.1 Mesh and boundary conditions.....................................................................3-3
Figure 3.2 Comparison of penetration resistances with centrifuge result ......................3-6
Figure 3.3 Variation of vertical resistance for different submerged unit weights: (a) κ=0;
(b) κ=20........................................................................................................................3-10
Figure 3.4 Buoyancy factor fb and other non-dimensional parameters with depth: (a) κ =
0; (b) κ = 20 .................................................................................................................3-11
Figure 3.5 Variation of buoyancy factor fb with non-dimensional parameter kD/su,avg ...3-
12
Figure 3.6 Effect of normalised penetration rate on vertical resistance: (a) V normalised
by original shear strength; (b) V normalised by equivalent shear strength..................3-14
Figure 3.7 Effect of rate parameter μ on vertical resistance: (a) V normalised by original
shear strength; (b) V normalised by equivalent shear strength ....................................3-16
Figure 3.8 Effect of sensitivity on vertical resistance: (a) α = 1/St ; (b) constant α (=
0.33) .............................................................................................................................3-18
List of figures
viii
Figure 3.9 Vertical resistances normalised by equivalent shear strength for different
sensitivity values (fs = 0.8)...........................................................................................3-19
Figure 3.10 Effect of softening parameter ξ95 on vertical resistance: (a) V normalised by
original shear strength; (b) V normalised by equivalent shear strength.......................3-20
Figure 3.11 Best fit power law curve for vertical resistance with depth......................3-22
Figure 3.12 Best fit power law curves for vertical resistances for different κ.............3-22
Figure 3.13 Deformation pattern and instantaneous velocity field at w/D = 0.5 for
different κ.....................................................................................................................3-24
CHAPTER 4 LARGE LATERAL MOVEMENT OF PIPELINES ON A SOFT CLAY SEABED: LARGE DEFORMATION FINITE ELEMENT ANALYSIS
Figure 4.1 Mesh and boundary conditions.....................................................................4-6
Figure 4.2 Soil mesh at different stages of movement (Case G) ...................................4-7
Figure 4.3 Trajectory of pipe invert during lateral motion (Ideal soil model, Cases A-D)
........................................................................................................................................4-9
Figure 4.4 Normalised horizontal resistance during lateral motion (Ideal soil model,
Cases A-D) ...................................................................................................................4-10
Figure 4.5 Equivalent friction factor during lateral motion .........................................4-10
Figure 4.6 Trajectory of pipe invert during lateral motion ..........................................4-13
Figure 4.7 Normalised horizontal resistance during lateral motion (Realistic soil model,
Cases E-G) ...................................................................................................................4-14
Figure 4.8 Equivalent friction factor during lateral motion (Realistic soil model, Cases
E-G)..............................................................................................................................4-14
Figure 4.9 Failure mechanisms during Case G ............................................................4-16
Figure 4.10 Soil deformation mechanism from a centrifuge model test (u/D = 3) (Dingle
et al. 2008)....................................................................................................................4-16
Figure 4.11 Soil softening during Case G....................................................................4-17
Figure 4.12 Pipe invert trajectory during lateral motion (Varying vertical loads, Cases
H-J )..............................................................................................................................4-18
Figure 4.13 Normalised horizontal resistance during lateral motion (Varying vertical
loads, Cases H-J)..........................................................................................................4-19
Figure 4.14 Equivalent friction factor during lateral motion (Varying vertical loads,
Cases H-J) ....................................................................................................................4-19
Figure 4.15 Effect of vertical load on steady state embedment ...................................4-20
List of figures
ix
Figure 4.16 Schematic diagram explaining the effective embedment concept (as per
White and Dingle, 2011) ..............................................................................................4-21
Figure 4.17 Variation of normalised lateral resistance with effective embedment......4-22
CHAPTER 5 MODELLING LATERAL PIPE-SOIL INTERACTIONS
Figure 5.1 Typical trajectories of pipes during lateral movement for different pipe
weights ...........................................................................................................................5-3
Figure 5.2 Typical lateral responses of pipe for different pipe weights.........................5-4
Figure 5.3 Friction ratios for pipes with different operating vertical loads ...................5-4
Figure 5.4 Idealisation of pipe response during lateral motion for ‘light’ and ‘heavy’
pipes ...............................................................................................................................5-6
Figure 5.5 Yield envelopes in V-H space (LDFE and parabola fit)...............................5-8
Figure 5.6 Yield envelopes from present study and Merifield et al. (2008) ..................5-8
Figure 5.7 Variation of residual friction factor with initial embedment ......................5-12
Figure 5.8 Variation of residual friction factor with normalised vertical load ............5-13
Figure 5.9 Variation of residual resistance with vertical load normalised by (a) Dsuo,init;
(b) Dsum ........................................................................................................................5-14
Figure 5.10 Variation of residual friction factor with normalised final embedment ...5-16
Figure 5.11 Variation of Hres/su0,finalD with (w/D)final ..................................................5-16
Figure 5.12 Lateral and vertical response using effective embedment approach ........5-18
Figure 5.13 Variation of residual effective embedment with initial embedment ........5-19
Figure 5.14 Ratios of (Hres/V)calculated to (Hres/V)LDFE varying with (a) winit/D; (b)
V/Dsu0,init ......................................................................................................................5-23
Figure 5.15 Typical friction ratio responses with lateral displacement fitted with
exponential equation: (a) Lighter pipes; (b) Heavier Pipes .........................................5-26
CHAPTER 6 BREAKOUT BEHAVIOUR OF PARTIALLY EMBEDDED PIPES IN UNIFORM CLAY USING LIMIT ANALYSIS
Figure 6.1 Schematic of the problem and notation ........................................................6-3
Figure 6.2 Yield envelopes for smooth pipes (Case A) .................................................6-4
Figure 6.3 Yield envelopes for rough pipes (Case B)....................................................6-5
Figure 6.4 Comparison of results for a typical pipe embedment (embedment = 0.4D).6-5
Figure 6.5 Stresses and corresponding displacement vectors of a horizontal interface.6-6
Figure 6.6 Flow vectors for smooth interface ................................................................6-6
List of figures
x
Figure 6.7 Flow directions as per conventional plasticity analysis for rough interface6-7
Figure 6.8 Direction of flow during loss of contact in no-tension surface (after Houlsby
and Puzrin, 1999) ...........................................................................................................6-8
Figure 6.9 Schematic of pipe movement and combination of smooth and rough interface
for OxLim analysis.........................................................................................................6-9
Figure 6.10 Comparison of result with Randolph and White (2008) for w/D = 0.5 after
interface modification (Case C) ...................................................................................6-10
Figure 6.11 Optimizing OxLim Result for modified interface; (a) Before optimisation,
(b) After optimisation...................................................................................................6-11
Figure 6.12 Simple wedge failure mechanism with corresponding hodograph...........6-12
Figure 6.13 Difference between optimised OxLim result and analytical solution for low
vertical loads ................................................................................................................6-13
Figure 6.14 Failure mechanism for pipe movement direction of 1 degree to the vertical
......................................................................................................................................6-14
Figure 6.15 Failure mechanisms for different directions of pipe movement...............6-15
Figure 6.16 Optimised yield envelopes for rough pipe soil interface ..........................6-16
Figure 6.17 Adaptive mesh refinement for pure vertical and horizontal pipe movements
in flat seabed (w/D = 0.5).............................................................................................6-17
Figure 6.18 Growth of yield envelopes due to introduction of soil self-weight (Case D
and E) ...........................................................................................................................6-18
Figure 6.19 Schematic for calculating fb for any direction of pipe movement ............6-19
Figure 6.20 fb values for different directions of pipe movement .................................6-20
Figure 6.21 Heave geometries for different embedments............................................6-21
Figure 6.22 Comparison of yield envelopes from OxLim and LDFE analyses for heaved
soil (Case G).................................................................................................................6-22
Figure 6.23 Adaptive mesh refinement for pure vertical and horizontal pipe movements
in heaved soil (w/D = 0.5)............................................................................................6-23
CHAPTER 7 ELASTOPLASTIC CONSOLIDATION BENEATH SHALLOWLY EMBEDDED OFFSHORE PIPELINES
Figure 7.1 Schematic diagram of the problem solved....................................................7-2
Figure 7.2 Yield envelope and critical state line for MCC model .................................7-5
Figure 7.3 Comparison of penetration responses for smooth and rough pipes..............7-6
Figure 7.4 Contours of excess pore water pressure after penetration (w/D = 0.5) ........7-8
List of figures
xi
Figure 7.5 Excess pore pressure distribution around pipe periphery after penetration
(w/D = 0.5) .....................................................................................................................7-9
Figure 7.6 Excess pore pressure dissipation time history at pipe invert for smooth pipe7-
10
Figure 7.7 Excess pore pressure dissipation time history at pipe invert for rough pipe ..7-
10
Figure 7.8 Average pore pressure dissipation and rise in effective stress along the pipe
periphery ......................................................................................................................7-13
Figure 7.9 Time settlement response for different initial embedments (smooth pipe) 7-14
Figure 7.10 Time settlement response for different initial embedments (rough pipe).7-14
CHAPTER 8 EFFECTS OF CONSOLIDATION ON PENETRATION AND LATERAL BREAKOUT RESISTANCES
Figure 8.1 Schematic of the problem studied.................................................................8-2
Figure 8.2 Finite element meshes before and after pipe penetration .............................8-3
Figure 8.3 Normalised penetration resistances with embedment for different pipe
velocities (smooth pipe) .................................................................................................8-5
Figure 8.4 Normalised penetration resistances with embedment for different pipe
velocities (rough pipe) ...................................................................................................8-6
Figure 8.5 Contours of excess pore water pressure normalised by penetration resistance
........................................................................................................................................8-8
Figure 8.6 Backbone curves for different initial embedment levels (smooth pipe).....8-10
Figure 8.7 Backbone curves for different initial embedment levels (Rough pipe)......8-10
Figure 8.8 Backbone curves for different initial embedment levels (smooth pipe, 1 kPa
surcharge).....................................................................................................................8-11
Figure 8.9 Pipe settlements with time for different initial embedments and pipe
velocities ......................................................................................................................8-12
Figure 8.10 Consolidation settlements following penetration at different speeds under
the same consolidation load (smooth pipe)..................................................................8-13
Figure 8.11 Dissipation of excess pore water pressure with non-dimensional time T
(smooth interface) ........................................................................................................8-14
Figure 8.12 Yield envelopes for different initial embedments for unconsolidated
undrained case (smooth pipe).......................................................................................8-16
List of figures
xii
Figure 8.13 Yield envelopes for different initial embedments for consolidated undrained
case (smooth pipe) .......................................................................................................8-18
Figure 8.14 Comparison of yield envelopes for unconsolidated undrained and
consolidated undrained conditions for w/D = 0.1 and 0.5 (smooth pipe) ....................8-18
Figure 8.15 Contours of ratios of consolidated shear strength to the original shear
strength (smooth pipe)..................................................................................................8-19
Figure 8.16 Maximum vertical penetration resistances for unconsolidated undrained and
consolidated undrained conditions (smooth pipe)........................................................8-21
Acknowledgements
xiii
ACKNOWLEDGEMENTS I would like to thank my thesis supervisors, Professor David White and Professor Mark
Randolph for their insightful guidance and continuous encouragement throughout my
candidature. I could not have expected a better combination of mentors for my research
and this dissertation would not have been possible without their support at key stages of
my candidature.
I also want to express my sincere gratitude to Dr. Dong Wang for his help with
programming the large deformation methodology during the first year of my
candidature.
I wish to thank Engineering Science department of Oxford University, especially
Dr. Byron Byrne and Dr. Chris Martin, for hosting me as a visiting student to work on a
collaborative project. A chapter of this thesis was constructed with the work done at
Oxford. Special thanks are extended to Dr. Chris Martin for allowing me use his limit
analysis software OxLim and for many fruitful discussions.
I would also like to thank Professor Deepankar Choudhury of Indian Institute of
Technology Bombay and Professor Ramendu Sahu of Jadavpur University, India for
recommending me to the University of Western Australia.
Centre for Offshore Foundation Systems at UWA has been a stimulating
working environment and I am indebted to all its faculty members, staffs and post
graduate students for making it such a fun place to work.
Throughout my candidature, I was financially supported by an International
Postgraduate Research Scholarship, a University Postgraduate Award and an ad-hoc
scholarship from COFS, which are gratefully acknowledged.
Finally, I would like to thank my parents, other family members and wife, Divya
for their continuous support, encouragement and love.
Numerical modelling of pipe-soil interactions
Centre for Offshore Foundation Systems 1-1
CHAPTER 1
INTRODUCTION AND LITERATURE REVIEW
1.1 RESEARCH MOTIVATION
Much of Australia’s remote sub-sea oil and gas resources remain undeveloped. Over the
last two decades, offshore oil and gas developments have gradually extended to deeper
water further from shore. This has led to a shifting of focus from fixed platforms to
floating production systems, which in turn has resulted in increasing importance of
pipelines and risers (Randolph & White, 2008a). Australia’s gas industry relies on ultra-
long sea-bed pipelines to bring the oil and gas from remote offshore hydrocarbon fields
to shore.
Figure 1.1 shows the existing (red lines) and proposed (white lines) pipelines at
the North-West Shelf of Western Australia. As seen in this figure, many hundreds of
kilometres of new pipelines are being proposed to transport hydrocarbon fluids from far
field locations. The further from shore, the deeper the water (in general) with seabed
comprised of finer-grained sediments than typically encountered in shallow water.
Within each field development, considerable lengths of flowlines link individual wells
to subsea manifolds and processing plants, prior to export to shore.
In deep water, where hydrodynamic loading is much reduced, pipelines are
generally laid directly on the seabed without trenching or other form of secondary
stabilisation. Assessment of the as-laid pipe embedment is an important step for proper
design of deep-water offshore pipelines, since other aspects such as lateral and axial
resistance, and thermal transfer rates are strongly influenced by embedment. Adopting
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-2
conservative design values is not a safe approach because high and low embedment, and
consequently the probable axial and lateral resistance, may work for or against a
particular design consideration. Design values of pipe-soil interaction forces, which
depend on the embedment, are key input in determining the practicability of a design
solution. Considerable amounts of capital expenditure can be saved by slight fine-tuning
of these values, through reduced requirements for stabilisation and anchoring measures
and a reduced need to tolerate end expansions (Randolph & White, 2008b, Hill & Jacob
2008).
Figure 1.1 Existing and proposed pipelines at the North West Shelf of Australia
In the case of deep-water pipelines, forces from hydrodynamic loading are
generally small and the dominant forces are from high internal temperature and pressure,
which tend to cause expansion (Bruton et al., 2008). Axial resistance between the pipe
and the seabed opposes this expansion. Excessive compressive forces lead to buckling,
but the buckling response depends critically on the lateral soil resistance. When
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-3
buckling occurs, it significantly reduces the net axial load in the pipe. On the other hand,
excessive buckling may lead to high bending strains in the pipe section. So, controlled
buckling (Figure 1.2) may be a feasible solution for relief of thermal loading.
Accumulated axial movement due to repeated thermal cycles may lead to global
displacement of pipes. This phenomenon is termed ‘walking’ (Carr et al., 2006). For
design purposes, it is very important to assess pipeline buckling and walking accurately.
Recent design approaches to control buckling and walking have necessitated predicting
the available soil resistance on pipelines undergoing movement, accounting for the
associated changes in seabed geometry and strength. The existing models are mainly
derived for stability analyses. The challenge is to extend existing models to account for
geometry changes, remoulding and reconsolidation effects that influence large
amplitude cyclic displacements.
Figure 1.2 Controlled lateral buckling of an on-bottom pipeline (Jayson et al., 2008)
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-4
During pipe-lay, deep-water pipelines typically embed by between 10 and 50 % of
their diameter, due to their own weight and the dynamic motions during laying
(Westgate et al., 2009, 2010). Once embedded, pipelines undergo cycles of large
horizontal movement in zones where buckling occurs, due to thermal expansion and
contraction. Key design parameters are the initial peak lateral resistance during the first
‘breakout’ cycle, and then the steady state resistance during subsequent cycles.
Load-displacement responses during lateral breakout have been reported from
centrifuge model studies (Dingle et al., 2008). Figure 1.3 shows soil deformation
mechanisms at several stages during lateral movement.
Figure 1.3 Soil deformation mechanism during lateral pipe movement from centrifuge modelling (Dingle et al., 2008)
At peak breakout resistance, evidence of two-sided mechanisms can be seen. Although
clear slip surfaces can be observed in front of the pipe, no clear slip surface is developed
in the soil behind the pipe, which means full soil strength was not mobilised in this
region prior to separation of the pipe from the soil. After breakout, distinct slip planes
can be seen in front of the pipe, matching mechanisms calculated from plasticity limit
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-5
analysis. At this stage, tensile resistance at the rear is lost and there is full pipe-soil
separation. During large amplitude lateral movement, soil is swept ahead of the pipe and
a berm is generated. Numerical results (Aubeny et al., 2005; Merifield et al., 2008) and
plasticity solutions (Randolph & White, 2008c, Cheuk et al., 2008) have been used to
assess breakout resistance. Although these solutions are theoretically rigorous, they do
not capture the changing seabed geometry or strength and are strictly applicable only for
a horizontal soil surface.
Vertical penetration and lateral movement of pipelines on a soft seabed are typical
examples of large deformation problems. Small strain finite element analyses cannot
replicate the full response because soil elements near the pipeline become extremely
distorted. In recent studies, large deformation finite element (LDFE) analyses of vertical
as well as lateral movements of pipes have been performed (Wang et al., 2010). The
“remeshing and interpolation technique with small strain” (RITSS, Hu & Randolph,
1998) has been used in these methods to divide the total displacement into a series of
small incremental steps, performing small strain analysis for each step. After a given
number of steps, the deformed geometry is remeshed prior to the next series of small
strain analyses. Field variables, e.g. stress and material properties, are updated from the
old mesh to the new mesh. The advantage of this method is that it can be combined with
commercial finite element packages such as AFENA (Carter & Balaam, 1995) or
ABAQUS (Dassault Systèmes, 2007, 2011). This methodology has been implemented
in ABAQUS by the author of this thesis for the current research and various problem-
specific refinements of the approach have been performed wherever necessary.
Most existing finite element studies of (deep-water) pipe-soil interaction assume
simple undrained soil models. However, pipelines undergoing large cyclic deformations
on fine-grained soil cause intermittent episodes of remoulding of the soil (during
pipeline motion) followed by reconsolidation (following movement). For correct
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-6
assessment of the resulting response, it is important to predict the operative soil strength.
Soil strength is decreased by remoulding, but it will increase again during
reconsolidation, as excess pore pressures dissipate. However, there is a scarcity of
models in the literature that can simulate cycles of softening due to remoulding,
followed by strength recovery due to consolidation.
Coupled consolidation finite element analyses to study the consolidation time
history beneath partially embedded seabed pipelines have typically been limited to
small strain elastic solutions in the past. Consolidation analyses using elasto-plastic soil
models, combined with large amplitude displacement of pipelines, have not been
reported in the literature prior to the present work.
1.2 LITERATURE REVIEW
Before proceeding to the original contribution of this research, a detailed review of the
literature on pipe-soil interactions is presented first. Chapters in the book edited by
McCarron (2011) summarised the design considerations for subsea flowlines against
lateral and upheaval buckling. Simple modelling techniques of pipe-soil interactions
based on numerical techniques are also discussed in this book. The chapter called
‘pipeline and riser geotechnics’ in the book by Randolph & Gourvenec (2011) also
discussed key aspects of offshore pipeline design and current practices.
The literature available to date on pipe-soil interaction can be divided into three
main categories. First are solutions based on classical plasticity theory, second are finite
element analysis based solutions and finally there are empirical approaches based on
centrifuge modelling or other physical modelling studies.
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-7
1.2.1 Theoretical solutions
Randolph & Houlsby (1984) first gave a theoretical solution for the limiting pressure on
a circular pile moving laterally through soil under undrained conditions. Although the
context was for laterally loaded piles, in a broader sense the solution applies to any
cylindrical object (such as a pipeline) moving laterally through soil. In that paper,
classical plasticity theory was used to establish lower and upper bound solutions. The
soil was assumed to be a perfectly plastic cohesive material and the pile as a long
cylinder moving horizontally in the infinite medium, reducing the calculations to a
plane strain problem of plasticity theory. In the lower-bound approach, a stress
distribution in equilibrium with a given applied load is assumed. Provided the stress
field does not conflict with the failure criterion, the calculated load is less than or equal
to the true collapse load. On the other hand, in the upper-bound approach, a failure
mechanism is assumed. The collapse load is found by equating the rate of plastic work
within the deforming soil to the work done by the external load. In this case, the
estimated load is greater than or equal to the true collapse load. If both these solutions
match, the solution is said to be exact if an additional conditions is satisfied – that the
stress field is extensible without violating the yield criterion. For the laterally loaded
pile problem the solution is expressed in terms of a resistance factor expressed as N =
P/suD, where P is the lateral capacity per unit length, su is the undrained shear strength
of the soil and D is the diameter of the pile.
At the time it was published, the Randolph & Houlsby solution was considered
exact (having equal upper and lower bounds). However, subsequent work by Murff et al.
(1989), who applied a modified form of the solution to evaluate the ultimate load
capacity of a pipe penetrating into perfectly plastic cohesive material, revealed an error
in the upper-bound solution, apart from the case of a fully rough pile. The authors
showed that the reason for this was the presence of a region of localised conflict
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-8
between the strain rate field and the stress field for any value of α < 1, where α is the
ratio of the limiting pile-soil friction to the shear strength of the soil. The solution was
corrected by taking the absolute value of the maximum shear strain rate, integrating the
particular component of plastic work numerically. This led to divergence of the lower
and upper-bound solutions, with a maximum discrepancy of 9.1% for the case of a
smooth pile.
Martin & Randolph (2006) made an attempt to minimize the gap between lower-
bound and upper-bound solutions for laterally loaded piles in undrained clay. Three
upper-bound solutions were provided for the ultimate load capacity of a circular pile
undergoing lateral translation in undrained clay. The first solution, referred to as the
Randolph mechanism, based on Randolph & Houlsby (1984), was shown to work well
for large α. A second solution was based on what is referred to as the Martin mechanism,
which consists of a crescent-shape body of soil undergoing rigid body rotation about a
point located on the axis of the pile perpendicular to the direction of the motion of the
pile. This mechanism works well for small α. In an effort to obtain an upper-bound
solution that will give close bracketing of the exact collapse load for all values of α,
these two mechanisms were combined in the third mechanism. This mechanism gave
upper-bound solutions that are very close to the lower-bound solution for all values of α.
The solution reduced the maximum discrepancy between upper and lower bound
solutions to 0.65% in the case of a smooth pile compared with the previous 9.1% (Murff
et al., 1989).
Upper-bound yield envelopes for shallowly embedded pipes in clay under
combined vertical and horizontal undrained loading were developed in a study by
Randolph & White (2008c). A plane strain plasticity model was assumed and the
solution was based on a generalisation of the mechanism as devised by Martin &
Randolph (2006). The generalisation is that the centre of rotation of the main block of
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-9
failing soil is not constrained to lie on the diameter normal to the direction of the pipe.
‘Break away’ at the trailing edge of the pipe was also assumed. Solutions were
presented for both homogeneous soil and for soil with shear strength varying
proportionally with depth below the soil surface.
Cheuk et al. (2008) also presented results of upper-bound analyses based on
simple slip circle failure mechanism. The results from this study are less optimal than
Randolph & White (2008), but consider non-breakaway cases and soil self weight as
well.
In an effort to allow for the strain-rate dependence of shear strength, and also the
gradual loss of strength due to remoulding, Einav & Randolph (2005) introduced a new
theoretical method involving a modified version of the Tresca soil model to evaluate the
penetration resistance for rigid ‘full-flow’ T-bar (cylindrical) and ball (spherical)
penetrometers. The method, which combines conventional strain path method and
classical upper-bound solutions, was referred to as the upper-bound-based strain path
method (UBSPM). Unlike the conventional strain path method, where the kinematic
mechanisms are based on flow solutions for irrotational inviscid fluid, the analysis was
based on the flow pattern derived from an upper-bound mechanism. The upper-bound
mechanism was optimized for ideal rigid plastic soil and was integrated with the strain
path method. Using this approach, the effects of rigidity index, strain rate and strain
softening on penetration resistance were investigated. The modification to the Tresca
constitutive model proposed by Einav & Randolph (2005) has been applied in the
present research.
1.2.2 Numerical analyses
Aubeny et al. (2005) reported results of small-strain finite element analyses for vertical
loading of pipes ‘wished-in’ place in a vertical trench in clay. Plane strain analyses were
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-10
carried out for different embedment depths. Most of the studies in this area considered
pipes with shallow embedment up to half the diameter of the pipe, whereas this study
was performed for embedments varying from 0.1 times the diameter to 5 times the
diameter of the pipe. Solutions for soils with constant shear strength and for soils with
linearly varying shear strength were presented. The results were compared with
approximate lower-bound and upper-bound plasticity solutions. Curve fitting of the FE
results allowed simplified equations to be developed relating the collapse load to
embedment depth and the shear strength of the soil at the pipe invert.
Merifield et al. (2008) reported results of finite element analyses of shallowly
embedded pipes under combinations of horizontal and vertical load. The results were
compared with the yield envelopes drawn from the upper-bound analyses of Randolph
& White (2008c). The limiting loads obtained from finite element analyses were in
very good agreement with the upper-bound curves. The authors also showed a good
match between the internal displacements calculated from FE analysis and those from
centrifuge tests. For simple assessment of the ultimate resistance of shallowly
embedded pipes, the yield curves were fitted with simple equations.
The above mentioned finite element solutions mainly dealt with small strain
analyses. Hesar (2004) made a first attempt to capture the large movements of pipelines
in soft clay using finite element software ABAQUS. ABAQUS explicit was used along
with adaptive meshing to avoid severe mesh distortion. The effects of initial pipe
embedment and submerged pipe weight on pipe soil lateral interaction forces were
explored. The paper also stressed the importance of obtaining high quality soil data for
very shallow depths which has marked influence on the design of pipelines. Only results
for a few specific cases were presented, and no attempt was made to provide general
solutions.
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-11
Konuk & Yu (2007) and Yu & Konuk (2007) studied large displacement pipe-
soil interaction problem using an Arbitrary Lagrangian Eulerian (ALE) approach
implemented in FE software LS-DYNA. Two-dimensional and three-dimensional
models were prepared and a cap plasticity soil constitutive model was used. These
papers emphasised the inadequacy of traditional design methods of pipelines against
lateral buckling, which are based on Winkler or Coulomb type models. The importance
of large deformation finite element techniques to design high temperature and high
pressure pipelines undergoing cyclic lateral motions was shown through a number of
2D and 3D models.
Merifield et al. (2009) studied the vertical penetration response of pipes and
subsequent horizontal resistance for pushed-in-place (PIP) pipes. A large deformation
finite element methodology, following an ALE approach, was adopted in ABAQUS to
limit mesh distortion problems. The effects of soil weight and local heave generated
during pipe penetration were explored. Simple expressions for vertical and horizontal
bearing capacities were presented incorporating these effects. Archimedes’ principle
was revisited and a modification was suggested to account for the effect of heave.
Zhou & Randolph (2007) performed numerical simulations of deep penetration
of full flow penetrometers (cylindrical T-bar and spherical ball). The RITSS approach
proposed by Hu & Randolph (1998) was adopted to perform the large deformation finite
element analysis, using AFENA as the finite element programme. The simple elastic-
perfectly plastic Tresca soil model was modified to incorporate the effects of strain-
softening and strain-rate. At any stage of analysis, the shear strength was modified to
account for reduction due to strain softening as well as enhancement due to high strain
rate, with the strength given by (Einav & Randolph 2005)
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-12
[ ] 0u/3
remrem
ref
.
ref
.
max
.
u se)1(.max
log1s 95ξξ−δ−+δ
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
⎟⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜⎜
⎝
⎛
γ
⎟⎠
⎞⎜⎝
⎛ γγμ+=
1.1
The first part of this relationship captures the effect of strain rate, with the reference
shear strain, ref
.γ , taken as 3x10-6 s-1 and the rate parameter, μ, giving the rate of strength
increase per decade, taken in the range of 0.05-0.2. The maximum shear strain rate,
max
.γ , was deduced as
tn31
max
.
ΔεΔ−εΔ
=γ 1.2
where 1εΔ and 3εΔ are cumulative major and minor principal strains, respectively, over
n increments between remeshing steps. The time period tΔ is given by
fieldp d/vd/t δ
=Δ 1.3
where δ is the specified incremental displacement for each increment, d and dfield are the
penetrometer diameters in the finite element calculations and field tests respectively,
and vp is the field penetrometer velocity (generally standard at 20 mm/s). The second
part of the equation represents the effect of strain softening. Here, suo denotes the
original shear strength at the reference shear strain rate prior to any softening, δrem is the
ratio of fully remoulded and initial shear strength, ξ is the accumulated absolute plastic
shear strain at the Gauss point and ξ95 is the value of ξ to cause 95% remoulding.
Parametric studies were carried out to investigate the effects of soil ductility,
rigidity index and rate parameter on penetration resistance for both T-bar and ball
probes. The phenomena of periodic shear bands and oscillations in the resistance were
observed in this study, with the peaks in the resistance-penetration response
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-13
corresponding to development of new shear bands and the troughs reflecting the
subsequent softening within each shear band.
Wang et al. (2010) studied large amplitude lateral motion of the pipe following
the RITSS approach, but implemented for the first time in finite element software
ABAQUS. Trajectories and horizontal resistance during lateral motion of the pipe for
different pipe weights were studied. A simple interpretation of the steady-state lateral
resistance was presented for pipes in rate dependent and softening clay. It was shown
that the effects of soil softening and berm build-up ahead of the pipe could be
encapsulated by defining an ‘effective embedment’ of the pipe. The results matched
well with available plasticity solutions and centrifuge test data.
Krost et al. (2011) studied consolidation around partially embedded pipelines
with the help of small strain finite element analysis but assuming the soil response to be
elastic. The generation and dissipation of pore water pressure at different levels of
embedment were studied. The results were compared to those for a strip footing. A
good match of the result was observed with available field data. The mobilisation of
effective contact force due to consolidation was presented. It was shown that up to 35%
increase in the average normal effective stress and hence axial resistance is possible due
to pore pressure dissipation under partially embedded pipes.
1.2.3 Design practice
Cathie et al. (2005) provided a state-of-the-art review of many aspects of pipeline
geotechnics. The paper summarised current models (Wagner et al. 1987, Lieng et al.
1988, Verley & Sotberg 1992, Verley & Lund 1995) used for assessing lateral
resistance of partially embedded pipelines. These two component models, consisting of
a sliding resistance component and a lateral passive pressure component, are totally
empirical and are based on a few model tests. Although these models are practically
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-14
useful they do not capture the underlying mechanics and are not applicable outside the
soil types forming the database.
Bruton et al. (2006), with the help of large and small-scale tests, provided
recommendations for using key parameters that affect the lateral pipe-soil interaction
response in soft clay soils. Different stages of pipe-soil response, including embedment
of pipe during installation, break-out during buckling, large amplitude displacement and
repeated cyclic behaviour, were defined. Suitable empirical equations were developed
for each of these steps to provide design guidelines for lateral buckling of pipelines.
Dendani & Jaeck (2008) presented simplified methods and practical calculations
for assessing pipe-soil interactions. This paper discussed soil-pipe interaction behaviour
properties based on site-specific data. Several phases of lateral resistance including peak
resistance, post-peak resistance, increase of resistance due to build-up of soil berms and
large displacement residual resistance were defined.
1.2.4 Model testing observations
Cheuk et al. (2007) reported results of full scale model tests on kaolin clay and West
African soft offshore clay to study pipe-soil lateral interactions due to repetitive cycles.
Four stages of the force-displacement response were identified: breakout, suction
release, steady berm growth and dormant berm collection. Breakout resistance was
shown to be the peak resistance followed by a drop due to loss of suction at the rear of
the pipe. Increase in resistance due to activating the dormant berm was also shown in
this study. An upper-bound model was proposed which proved to be reasonably
accurate in predicting the experimental results.
Dingle et al. (2008) reported centrifuge model test data for assessing vertical
penetration and lateral break-out resistance of pipelines laid on soft sea-bed sediments.
An advanced digital image analysis technique, using particle image velocimetry (PIV)
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-15
(White et al., 2003), was implemented to observe the soil deformation. The results
closely matched plasticity solutions, with minor differences attributed to the difficulties
in accurately assessing the shear strength of near mud-line soft sediments.
Cheuk & White (2009) reported results of centrifuge model tests using different
clays to investigate dynamic lay effects on pipeline embedment. Results showed that
only a few cycles of small amplitude oscillation were sufficient to double or triple the
static embedment. The combined effects of lateral ploughing and softening were
considered responsible for this additional embedment. The authors also proposed a new
model incorporating the concepts of plasticity theory to estimate dynamic pipe
embedment.
Cardoso & Silveira (2010) reported results of full scale model tests to study the
large deformation lateral resistance of pipe in soft clay. A wide range of parameters
including ‘heavy’ and ‘light’ pipes were chosen and a number of model tests were
performed. Results were expressed in non-dimensional form and empirical equations
were fit. Expressions for breakout resistance, residual resistance and berm resistance,
i.e. the pipeline resistance at different stages of lateral motion, were presented in the
form of simple expressions.
1.3 RESEARCH GOALS
As seen in the review of the literature, most of the available theoretical studies dealt
with small strain finite element analyses. These studies do not capture the change of
geometry during large amplitude movements of the pipe. The limited number of large
deformation studies did not consider all aspects that affect the pipe-soil interaction
forces on moving pipelines. Also, most of these studies assumed non-softening
constitutive soil models, which do not account for remoulding of the soil as it undergoes
large strains, and therefore are not suitable for predicting realistic pipe-soil interaction
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-16
forces. Coupled consolidation stress analyses for drained behaviour of soil under a
partially embedded pipe are scarce in the literature. Experimental studies have been
performed in the industry to study large amplitude movements of pipelines, but these
are mostly case specific, and the results have rarely been generalised other than using
empirical expressions without theoretical basis. A significant quantity of experimental
data related to the topic is becoming available through industry studies and also from
recent academic research. The goal of the present research has been to provide a
theoretical basis in which these experimental results can be framed and also to help to
predict results for the cases where experimental results are not available or experiments
are difficult to conduct. The specific goals of this research are listed below.
1) To develop a methodology based on large deformation finite element (LDFE)
analysis to simulate the movement of offshore pipelines in soft seabed sediments
in the horizontal and vertical directions. The effects of strain rate and strain
softening will be incorporated in the model.
2) To develop a calculation approach to predict pipeline embedment during the
laying process. This will be based on parametric studies, using the methodology
developed in (1), varying parameters such as the relative weight of the pipe
compared with the soil shear strength, the sensitivity of the soil, and the pipeline
motions.
3) To quantify the lateral resistance of pipes in soft soil, with particular focus on
the breakout resistance and the steady state lateral resistance relevant for
buckling analysis. The effects of initial embedments and operating pipe weight
will be explored in detail.
4) To improve the available plasticity-based solutions to predict pipe-soil
interaction forces for pipelines under combined vertical and horizontal loads.
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-17
The specific improvement will be to incorporate uneven berm geometry
available from LDFE analyses.
5) To introduce the effect of soil consolidation into this topic. Coupled
consolidation analysis using the modified Cam Clay soil model will be
incorporated in the LDFE methodology to study partially drained behaviour of
soil around partially embedded pipes.
1.4 METHODOLOGY
A two-dimensional large deformation finite element model was adopted with the help of
“Remeshing and interpolation technique with small strain” (RITSS, Hu & Randolph,
1998). RITSS is a variation of the Arbitrary Lagrangian Eulerian (ALE) method (Ghosh
& Kikuchi, 1991), involving periodic remeshing followed by interpolation of the field
variables from the old mesh to the new mesh, in order to model ‘convection’ of the soil
through the mesh. The approach was implemented in the commercial finite element
software ABAQUS (Dassault Systèmes, 2007, 2011).
Two different soil constitutive models were used. The first was based on the
simple Tresca soil model for undrained response of clays. However, the effects of strain
rate and strain softening were implemented in the analysis according to the model
suggested by previous researchers (Einav & Randolph, 2005; Zhou & Randolph, 2007).
The second model was the more sophisticated constitutive model, modified Cam Clay
(MCC). This was implemented in the LDFE methodology to perform coupled
consolidation analyses.
1.5 OUTLINE
The thesis consists of nine chapters. A brief outline of each chapter is given below.
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-18
1. The first chapter is an introductory chapter outlining the motivation for the
research and providing a review of current literature on pipe-soil interactions.
2. The second chapter explains the large deformation finite element methodology
developed for this study.
3. In the third chapter, pipe-soil interaction during vertical embedment of pipelines
on the seabed is studied using a simple Tresca soil model, modified to
incorporate the combined effects of strain rate and softening. The large
deformation finite element method is validated by comparing the results with
data from centrifuge model tests. Results of a parametric study are then
presented, varying the strain rate and softening parameters to explore their
effects on penetration resistance. Simple expressions for penetration resistance,
incorporating the effects of strain rate and softening, are provided. The effects of
soil strength vertical heterogeneity and buoyancy are also explored.
4. In the fourth chapter, the lateral response of pipelines on a soft seabed is studied
for very large amplitude lateral movement. Initially, pipe soil interaction
simulations are presented for the case of ideal soil, with non-softening strength.
Lateral resistance profiles and trajectories of the pipe during lateral motion are
investigated for different initial embedment of the pipe. A more realistic soil
model incorporating the effects of strain rate and strain softening is then
explored. Lateral resistance profiles and trajectories of the pipes from this
realistic model are compared with the ideal soil case. Finally, the concept of
effective pipe embedment – which accounts for the soil softening and the
geometric changes caused by the soil berm ahead of the pipe – is applied to both
the ideal and realistic soil model responses. The normalized horizontal resistance
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-19
response is shown to be linked to the effective embedment in a simple manner,
regardless of the other soil and pipeline parameters.
5. The fifth chapter also focuses on lateral pipe-soil interactions. Results of a
parametric study varying the pipe weight and initial embedment are presented.
The results show that a steady state is generally reached at large displacements,
reflecting a balance between the growth of a soil berm ahead of the pipe and the
softening of the disturbed soil. The initial breakout response is shown to match
well with previously established failure envelopes and a new interpretation is
proposed to capture the large-displacement response. The ‘effective embedment’
concept is used to rationalise the influence of the soil berm ahead of the pipe.
This leads to simple new relationships for predicting the steady state residual
lateral resistance, which provide more accurate predictions of the LDFE
response than previously published solutions. The complete load-displacement
response over large movements is also shown to be well-fitted by an exponential
relationship, albeit for the specific case of lateral movements under constant
vertical load.
6. In the sixth chapter, the breakout resistance and trajectory of partially embedded
pipelines in seabed is investigated using finite element limit analysis software
OxLim. Although OxLim analyses conform to classical plasticity theory, a slight
modification of the interface condition which violates normality has been
adopted. The pipe-soil interface can sustain neither tension nor shear stress when
separation occurs, providing a more natural solution. Results are compared with
those available in the literature and marginal improvements are demonstrated.
The effect of considering self-weight on the resulting yield envelopes is
explored. Also, the effect of soil heave around the pipe, the geometry of which
has been obtained from large deformation finite element analyses, is investigated.
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-20
7. In the seventh chapter, the consolidation behaviour under partially embedded
pipelines is investigated using large deformation finite element analyses based
on the modified Cam Clay plasticity model. Initially results from undrained
analyses are presented and the effects of initial embedment and pipe-soil
interface friction are explored in a systematic manner. The dissipation responses
are fitted by simple equations to facilitate practical design.
8. In the eighth chapter, results of partially drained analyses are presented for
different penetration rates of the pipe. Backbone-type curves are provided for
both smooth and rough pipes. The lateral breakout resistance and the direction of
pipe movement on breakout depend on the consolidated strength of the soil
around the pipe, as well as the applied loading. The effect of consolidation on
the lateral breakout resistance has also been explored in this chapter. It has been
shown that the envelopes of vertical-lateral combined loading bearing capacity
differ markedly from those predicted assuming undrained behaviour throughout.
9. In the last chapter, concluding remarks based on the research in this thesis are
presented. Future research and challenges in this area are also discussed at the
end of this chapter.
1.6 REFERENCES
Aubeny, C. P., Shi, H. & Murff, J. D. (2005). Collapse load for cylinder embedded in
trench in cohesive soil. Int. J. Geomech. 5, No. 4, 320-325.
Bruton, D. A. S., White, D. J., Cheuk, C. Y., Bolton, M. D. & Carr, M. C. (2006). Pipe-
soil interaction behaviour during lateral buckling, including large amplitude cyclic
displacement tests by the Safebuck JIP. Proc. Offshore Technology Conf., Houston,
Paper OTC 17944.
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-21
Bruton, D. A. S, White, D. J., Carr, M. & Cheuk, C. Y. (2008). Pipe-soil interaction
during lateral buckling and pipeline walking – the Safebuck JIP. Proc. of
Offshore Technology Conference, OTC 19589.
Cardoso, C. O., & Silveira, R. M. S. (2010). Pipe-soil interaction behavior for pipelines
under large displacements on clay soils – a model for lateral residual friction
factor. Proc. Offshore Technology Conf., Houston, OTC 20767.
Carr, M., Sinclair, F. & Bruton, D.(2006). Pipeline walking — understanding the field
layout challenges, and analytical solutions developed for the SAFEBUCK JIP.
Proc. Offshore Technology Conf., Houston, OTC 17945.
Carter, J. P. & Balaam, N. P. (1995). AFENA User Manual 5.0. Geotechnical Research
Centre, The University of Sydney, Sydney, Australia.
Cathie, D. N., Jaeck, C, Ballard, J. C. & Wintgens J-F. (2005). Pipeline geotechnics –
state-of-the-art. Proc. Int. Symp. On Frontiers in Offshore Geotechnics, Perth,
95-114
Cheuk, C. Y., White, D. J. & Bolton, M. D. (2007). Large scale modelling of soil-pipe
interaction during large amplitude movements of partially-embedded pipelines.
Can. Geotech. J. 44, No. 8, 977-996.
Cheuk, C. Y., White, D. J. & Dingle, H. R. C. (2008). Upper bound plasticity analysis
of a partially-embedded pipe under combined vertical and horizontal loading.
Soils and Foundations 48, No. 1, 133-140.
Cheuk, C. Y. & White, D. J. (2011). Modelling the dynamic embedment of seabed
pipelines. Géotechnique 61, No. 1, 39-57.
Dassault Systèmes (2007, 2011) Abaqus analysis users’ manual, Simula Corp,
Providence, RI, USA.
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-22
Dendani, H. & Jaeck, C. (2008). Flowline and Riser: Soil Interaction in Plastic Clays.
Proc. of Offshore Technology Conference, OTC 19261.
Dingle, H. R. C., White, D. J. & Gaudin, C. (2008). Mechanisms of pipe embedment
and lateral breakout on soft clay. Can. Geotech. J. 45, No. 5, 636-652.
Einav, I. & Randolph, M. F. (2005). Combining upper bound and strain path methods
for evaluating penetration resistance. Int. J. Numer. Methods Eng. 63, No. 14,
1991-2016.
Ghosh, S. & Kikuchi, N. (1991). An arbitrary Lagrangian-Eulerian finite element
method for large deformation analysis of elastic-viscoplastic solids. Comput.
Methods Appl. Mech. Eng. 86, No. 2, 127-188.
Hesar, M. (2004). Pipeline-seabed interaction in soft clay. Proc. 23rd Int. Conf. on
Offshore Mechanics and arctic eng., Vancouver, 225-232.
Hu, Y. & Randolph, M. F. (1998a). A practical numerical approach for large
defomation problems in soil. Int. J. Numer. Analyt. Meth. Geomech. 22, No. 5, 327-
350.
Hu, Y. & Randolph, M. F. (1998b). H-adaptive FE analysis of elastoplastic non-
homogeneous soil with large deformation. Comput. Geotech. 23, No. 1-2, 61-83.
Jayson, D, Delaporte, P, Albert, J-P, Prevost, M. E., Bruton, D & Sinclair, F. (2008).
Greater Plutonio Project – Subsea Flowline Design and Performance. Offshore
Pipeline Technology Conf.
Konuk, I. & Yu, S. (2007). Continuum FE modelling of lateral buckling: study of soil
effects. Proc. of 26th Int. Conf. on Offshore Mechanics and Arctic Eng., San Diego,
OMAE2007-29376.
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-23
Krost, K., Gourvenec, S. M. & White, D. J. (2011). Consolidation around partially
embedded seabed pipelines. Géotechnique 61, No. 2, 167-173.
Lieng, J. T., Sotberg, T. & Brennodden, H. (1988). Energybased pipe-soil interaction
models. SINTEF Report to the American Gas Association.
Martin, C. M. & Randolph, M. F. (2006). Upper bound analysis of lateral pile capacity
in cohesive soil. Géotechnique 56, No. 2, 141-145.
McCarron, W. O. (2011) Deepwater foundations and pipeline geomechanics. J. Ross
publishing, Fort Lauderdale, USA.
Merifield, R. S., White, D. J. & Randolph, M. F. (2008). The ultimate undrained
resistance of partially embedded pipelines. Géotechnique 58, No. 6, 461-470.
Merifield, R. S., White, D. J. & Randolph, M. F. (2009). Effect of surface heave on
response of partially embedded pipelines on clay. J. Geotech. Geoenviron. Engng,
ASCE 135, No. 6, 819-829.
Murff, J. D., Wagner, D. A. & Randolph, M. F. (1989). Pipe penetration in cohesive soil.
Géotechnique 39, No. 2, 213-229.
Randolph, M. & Gourvenec, S. (2011). Offshore geotechnical engineering. Spon Press,
Taylor & Francis Group, New York.
Randolph, M. F. (2004). Characterisation of soft sediments for offshore applications,
Keynote Lecture. Proc. 2nd Int. Conf. on Site Characterisation, Porto, Portugal,
1, 209-231.
Randolph, M. F. & Houlsby, G. T. (1984). The limiting pressure on a circular pile
loaded laterally in cohesive soil. Géotechnique 34, No. 4, 613-623.
Randolph, M. F., Wang, D., Hossain, M. S., Zhou, H. & Hu, Y. (2008). Large
deformation finite element analysis for offshore applications. Proc. of 12th
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-24
International Conference of International Association for Computer Methods
and Advances in Geomechanics, Goa, India.
Randolph, M. F. & White, D. J. (2008a). Offshore Foundation Design – A Moving
Target. Proc. BGA International Conference on Foundations, Dundee, IHS BRE
Press, London, 27-59.
Randolph, M. F. & White, D. J. (2008b). Pipeline embedment in deep water: process
and quantitative assessment. Proc. Offshore Technology Conference, Houston, Paper
OTC 19128.
Randolph, M. F. & White, D. J. (2008c). Upper-bound yield envelopes for pipelines at
shallow embedment in clay. Géotechnique 58, No. 4, 297-301.
Verley, R. L. P. & Lund, K. M. (1995). A soil resistance model for pipelines placed on
clay soils. Proc. Offshore Mechanics and Arctic Engineering Conf, Copenhagen,
18–22, Vol V: 225–232.
Verley, R. L. P. & Sotberg, T. (1992). A soil resistance model for pipelines placed on
sandy soils. Proc. Offshore Mechanics and Arctic Engineering Conf, Vol V-A
pipeline technology: 123–131.
Wagner, D. A. Murff J. D. & Brennodden, H. (1987). Pipe-soil interaction model. Proc.
Offshore Technology Conf., Houston, OTC 5504: 181–190.
Wang, D, White, D. J. & Randolph, M. F. (2010). Large deformation finite element
analysis of pipe penetration and large-amplitude lateral displacement. Can. Geotech.
J. 47, No. 8, 842-856.
Westgate, Z., White, D. J. & Randolph, M. F. (2009). Video observations of dynamic
embedment during pipelaying on soft clay. Proc. Conf. on Offshore Mechanics
and Arctic Engineering, Honolulu. Paper OMAE2009-79814
CHAPTER 1: Introduction and literature review
Centre for Offshore Foundation Systems 1-25
Westgate, Z. W., White, D. J., Randolph, M. F. & Brunning, P. (2010). Pipeline laying
and embedment in soft fine-grained soils: Field observations and numerical
simulations. Proc. Offshore Technology Conference, Houston. Paper 20407
White, D. J., Take, W. A. & Bolton, M. D. (2003). Soil deformation measurement using
Particle Image Velocimetry (PIV) and photogrammetry. Géotechnique, 53, No.
7, 619-631.
Yu, S. & Konuk, I. (2007). Continuum FE modelling of lateral buckling. Proc. Offshore
Technology Conf., Houston, OTC 18934.
Zhou, H. & Randolph, M. F. (2007) Computational techniques and shear band
development for cylindrical and spherical penetrometers in strain-softening clay.
Int. J. Geomech., 7, No. 4, 287-295.
Numerical modelling of pipe-soil interactions
Centre for Offshore Foundation Systems 2-1
CHAPTER 2
LARGE DEFORMATION FINITE ELEMENT
METHODOLOGY
2.1 NON-LINEAR FINITE ELEMENT ANALYSES
Over the past several decades, numerical solutions of non-linear problems have received
attention of researchers because of many practical applications. Unlike linear analyses,
the stiffness matrix does not remain constant in a non-linear analysis. Non-linear
behaviours associated with finite element analyses can be classified in to three main
categories – material non-linearity, geometric non-linearity and boundary non-linearity
(Bathe, 1996). In material non-linear analyses, displacements and strains are
infinitesimal, but stress-strain relations are non-linear. Geometric non-linearity is
mainly associated with large deformation of the domain. As per Bathe (1996),
geometric non-linear analyses can be of two types – (i) large displacements, large
rotations, but strains are small and (ii) large displacements, large rotations and strains
are also large. In boundary non-linearity problems, the boundary conditions change
during the motion of the body. As far as geotechnical non-linear analyses are concerned,
material non-linearity, which is an inherent characteristic of soil behaviour, has been the
most studied aspect to date. Thorough and dedicated research to address geometrical
non-linear analyses of geomechanics problems involving large deformations has not
been commenced until recent years. The complexity of large deformation geotechnical
problems coupled with complex constitutive model has encouraged researchers to
invent novel techniques to solve them.
CHAPTER 2: Large deformation finite element methodology
Centre for Offshore Foundation Systems 2-2
2.2 ANALYSIS PROCEDURES
There are three widely used finite element techniques to solve large deformation
problems - Lagrangian approach, Eulerian approach and Arbitrary Lagrangian Eulerian
approach.
2.2.1 Lagrangian approach
In the Lagrangian approach, the nodes of the finite element mesh move with the
associated material point during analysis. The advantages of this approach are that it
only has to satisfy relatively simple governing equations and it allows easy tracking of
free surfaces. In the Total Lagrangian (TL) approach all variables are referred to the
undeformed geometry. In the Updated Lagrangian (UL) approach, all variables are
referred to the current and updated reference geometry. The major limitation that both
these techniques suffer from is gross mesh distortion and entanglement when large
deformations occur within the body.
2.2.2 Eulerian approach
In the Eulerian approach, the computational mesh remains fixed and the material moves
through it as time progresses. This approach is particularly suitable for fluid flow
problems and when there is no moving boundary. This technique was successfully
applied to deep penetration problem in geomechanics by van den Berg (1991).
2.2.3 Arbitrary Lagrangian Eulerian approach
The Arbitrary Lagrangian Eulerian approach was developed to combine the advantages
of the above two techniques and minimise the limitations to address large deformation
problems. This methodology was proposed in a finite difference context by Noh (1964)
and Franck & Lazarus (1964), and later adapted by Hirt et al. (1974). The methodology
CHAPTER 2: Large deformation finite element methodology
Centre for Offshore Foundation Systems 2-3
was implemented in finite element analysis of fluid problems by researchers such as
Donea et al. (1977), Belytschko et al. (1978) and Hughes et al. (1991). Ghosh &
Kikuchi (1988) and Ghosh & Kikuchi (1990) introduced the technique to non-linear
solid mechanics problems. In the Arbitrary Lagrangian Eulerian (ALE) technique, mesh
and material displacements are uncoupled to avoid mesh distortion and entanglement.
As the name suggests, the extent to which material moves through the finite element
mesh or the mesh moves with material is arbitrary. In this description, the motions of
both grid and material are defined, but strictly ALE involves a finite element grid of
constant topology (i.e. without periodic remeshing).
2.3 RITSS APPROACH
All the techniques stated above implement large strain formulations. However, from a
geotechnical perspective, complex constitutive models make large strain formulations
more difficult to be implemented. Hu & Randolph (1998a, b) developed a simple
technique called Remeshing and Interpolation Technique with Small Strain (RITSS). A
series of small strain Lagrangian calculations are performed, followed by remeshing of
the deformed regime and interpolation of stress, strain and material properties from the
old mesh to the new mesh. This is repeated until the required displacement is achieved.
An overview of the technique, as per Hu & Randolph (1998a), is shown in Figure 2.1.
The main advantage of the method is that, due to the Lagrangian small strain
incremental steps, it can be coupled with any commercially available finite element
software. Initially, RITSS was successfully implemented in finite element software
AFENA (Carter & Balaam, 1995) to study several geotechnical applications – shallow
foundations (Hu & Randolph, 1998a, 1998b, 1998c; Hu et al., 1999; Zhou & Randolph,
2006), spudcan foundations (Hu & Randolph 1998a; Hossain et al., 2005; Hossain &
CHAPTER 2: Large deformation finite element methodology
Centre for Offshore Foundation Systems 2-4
Randolph, 2009, 2010), subsea pipelines (Barbosa-Cruz & Randolph 2005; Zhou et al.,
2008) and penetrometers (Lu et al., 2004; Zhou & Randolph, 2007, 2009a, 2009b, 2011).
Initial mesh generation (and optimisation)
Small strain incremental analysis step
Updating of external and internal boundary node positions and remeshing
Interpolation of stresses and material properties from old to new mesh
Loop
as n
eces
sary
Initial mesh generation (and optimisation)
Small strain incremental analysis step
Updating of external and internal boundary node positions and remeshing
Interpolation of stresses and material properties from old to new mesh
Loop
as n
eces
sary
Figure 2.1 Overview of RITSS approach (Hu & Randolph, 1998a)
2.4 IMPLEMENTATION IN ABAQUS
Recently Wang et al. (2006, 2010a, 2010b) implemented RITSS in the commercial
software ABAQUS (Dassault Systèmes, 2007) due to its powerful mesh generation
tools and computational efficiency. The numerical procedure developed for this
research has been based on the same methodology and various problem specific
modifications have been incorporated where necessary.
The finite element software ABAQUS was used for making the model, mesh
generation, analysis and post-processing. The problem was solved using a
displacement-controlled approach. Following the RITSS method, the whole
displacement is divided into a series of small displacement steps. Python, which is the
in-built scripting language of ABAQUS software, is used to execute different ABAQUS
CHAPTER 2: Large deformation finite element methodology
Centre for Offshore Foundation Systems 2-5
functions. The superconvergent patch recovery (SPR) technique (Zienkiewicz & Zhu,
1992) was used for recovery of stresses from the Gauss points to the nodes. Stress-SPR
interpolation scheme was chosen to transfer stress from the old mesh to the new mesh.
2.4.1 Superconvergent Patch Recovery – SPR
There are certain points in each element of a mesh that exhibit superconvergent
properties. At these sampling points, the convergence order of the finite element
functions is at least one order higher than at others. It is seen that the Gauss points are
the superconvergent sampling points for second order triangular elements (Figure 2.2).
Patch assembly point
Superconvergent sampling pointsNodal values recovered
Patch assembly point
Superconvergent sampling pointsNodal values recovered
Patch assembly point
Superconvergent sampling pointsNodal values recovered
Figure 2.2 Superconvergent Patch Recovery (Zienkiewicz & Zhu, 1992)
Let ∧
σ be the stress at superconvergent points. If the values of ∧
σ at these points are
accurate to order p+1, it is possible to find superconvergent stresses *σ at all points in
the element defined by a polynomial of degree p. The polynomial is of the same order
as that in the shape function for displacement and can be fitted to the superconvergent
points using a least squares approach.
The recovered stress component *iσ can be written as
CHAPTER 2: Large deformation finite element methodology
Centre for Offshore Foundation Systems 2-6
a].y,...,y,x,1[pa p*i ==σ 2.1
where, a is a vector of unknown parameters.
An error estimate is provided by
∑=
∧
−σ=∏n
1k
2kkki ]ap)y,x([
2.2
where pk corresponds to the vector associated with the coordinates of the
superconvergent sampling point, (xk, yk).
The value of Π is minimised for a patch with n sampling points. Then, the parameter
vector, a, is deduced as
BAa 1−= 2.3
where
∑=
=n
1kk
Tk ppA 2.4
and
)y,x(pB kkiTk
∧
σ=
2.5
The solution of Equation 2.3 is done component by component .Thus, by means of the
superconvergent sampling points, stress values can be smoothed at each node of a patch.
It should be noted that there are many nodes that belong to more than one patch. In
those cases, the recovered stresses from different patches are averaged to obtain the
final value. In the case of external boundary nodes, the stresses are discontinuous. So,
the nodal values are calculated from the nearest patches for those cases.
2.4.2 Steps of LDFE analysis
The steps of the LDFE analysis are as follows.
CHAPTER 2: Large deformation finite element methodology
Centre for Offshore Foundation Systems 2-7
1. First, a Python script provides details of the model and generates the mesh for
the first step. The model is used as input to ABAQUS for standard analysis with
very small displacement increment.
2. After completion of this standard analysis step, the output database is post-
processed with the help of another Python script. The coordinates of the nodes
of the displaced model, and also stresses at integration points and reaction forces,
are recorded. The python script also generates the list of elements forming the
top free surface. The order of the nodes and faces of a six noded triangular
element in ABAQUS are numbered in the following manner as shown in Figure
2.3.
1 4 2
5
3
6Face 3 Face 2
Face 1
Y
X 1 4 2
5
3
6Face 3 Face 2
Face 1
1 4 2
5
3
6Face 3 Face 2
Face 1
Y
X
Y
X
Figure 2.3 Six noded triangulare elements in ABAQUS
The list generated by the python script groups three types of elements – S1
which have face 1 at the top surface, S2 which have face 2 at the top surface and
S3 which have face 3 at the top surface. From this list of elements, the nodes
that are at the top surface are identified, although it should be remembered that
they are not arranged in order to define the new surface. With the help of a
Fortran subroutine, the nodes at the top boundary are arranged in order and, after
calculating the displaced coordinates, the new top boundary is defined. The old
CHAPTER 2: Large deformation finite element methodology
Centre for Offshore Foundation Systems 2-8
deformed mesh is deleted and with the help of another python script, a new
mesh is generated for the deformed regime.
3. Following the recovery procedure mentioned in the previous section, stresses are
recovered from the old Gauss points to the old nodes. Now, the stress
components are interpolated from the old nodes to the new Gauss points
following the scheme described below.
i. First, one old triangle is chosen at a time. Two horizontal boundaries
passing through the extreme top and bottom points of the triangle are
drawn as shown in Figure 2.4. Two vertical boundaries passing through
the extreme left and right points of the triangle are also drawn.
ii. Then, the new integration points that fall between these boundary lines
are identified. Each integration point (G) falling within the rectangular
region is connected to the vertices of the old triangle. This forms three
sub-triangles for each integration point. These three sub-triangles are
ABG, BCG and CAG.
A
G
C
B
G
A
B
C
(a) (b)
A
G
C
B
A
G
C
B
G
A
B
C
G
A
B
C
(a) (b)
Figure 2.4 Determination of position of new Gauss point;
(a) Gauss point inside old triangle; (b) Gauss point outside old triangle
CHAPTER 2: Large deformation finite element methodology
Centre for Offshore Foundation Systems 2-9
iii. If a new Gauss point is inside the old triangle ABC, the sum of the area
of the sub-triangles is equal to the area of the old triangle. The case is
shown in Figure 2.4a. If not, the sum is greater than the area of the
original triangle - this case is shown in Figure 2.4b. In this way, it is
determined which old element contains the new Gauss point.
iv. The above steps are repeated for all old elements to determine the
position of all the new Gauss points. The stresses at the new integration
points are then interpolated from recovered stresses at the neighbouring
old nodes.
4. For interpolation of the shear strength of the soil from the old mesh to the new
mesh, there is no need to perform the recovery step. The shear strengths at the
new nodes are interpolated directly by quadratic interpolation from the old nodes.
5. The in-built ABAQUS user subroutine SIGINI is used to define the initial
stresses at the new Gauss points. Then, the new remeshed domain with these
initial conditions is submitted to ABAQUS for another incremental small strain
step.
6. All these steps are repeated until the desired displacement is achieved.
The whole process is controlled by a master Fortran program, which repeatedly calls
Fortran subroutines and Python scripts to accomplish the analysis automatically without
the intervention of the user. The default equilibrium and convergence criteria available
in ABAQUS are implemented and convergence is checked at the end of each
incremental step. A flow diagram is shown in Figure 2.5 to describe the entire process.
Essentially this follows the procedure outlined by Wang et al. (2010b), although it has
been implemented in a slightly different way for the present work.
CHAPTER 2: Large deformation finite element methodology
Centre for Offshore Foundation Systems 2-10
Python Script for initial mesh generation
Small strain ABAQUS analysis
Python Script for post-processing to obtain nodal displacements and stresses
and parameters on integration points
Recovery of stresses and other parameters from integration
points to nodes Coordinates of displaced
boundary nodes
Python script for remeshing
Interpolation of stresses and other parameters from old
nodes to new integration points
Small strain ABAQUS analysis
Loop
as n
eces
sary
Figure 2.5 Implementation of RITSS in ABAQUS
2.4.3 Effects of strain rate and softening
To incorporate the effects of strain rate and strain softening, the original shear strength
of the soil, defined within the Tresca soil model, is modified according to the model
proposed by Einav & Randolph (2005). As far as the LDFE methodology is concerned,
two new variables are recorded at each Gauss point – shear strain rate and cumulative
plastic strain. Shear strain rate and cumulative plastic strain are then interpolated to the
new mesh, in addition to stresses and other parameters as described previously. The
original shear strength at each new Gauss point is then modified and updated as a
function of these two variables. The details of this procedure are given in Chapters 3, 4
and 5.
CHAPTER 2: Large deformation finite element methodology
Centre for Offshore Foundation Systems 2-11
2.4.4 Modified Cam Clay model
In the later stages of this research, coupled consolidation analyses were performed to
study partially drained behaviour and consolidation effects. The simple Tresca model is
suitable for total stress analyses, but is not able to replicate soil behaviour under
partially drained conditions. The modified Cam Clay (MCC) critical state soil model
was implemented in the LDFE methodology to perform the study and all details are
provided in Chapter 7. There are several new parameters that are required to be defined
within the MCC model. The effective stresses are recovered at old nodes and
interpolated to the new Gauss points as described earlier. The pore water pressure is a
new parameter in this model and is a nodal variable. Pore water pressure was
interpolated from the old nodes directly to the new nodes as initial conditions and
therefore there was no need for a special recovery technique.
Another parameter that needs to be updated during incremental steps is the pre-
consolidation pressure, a0, which defines the current size (half) of the yield surface. The
current yield surface size is recovered from the Gauss points to the old nodes at the end
of each step and then interpolated from the old nodes to the new Gauss points by
interpolation. At this stage, the yield status is checked at the new Gauss points to
identify whether the initial stress components are inside or on the initial yield surface. If
the stress components are outside the yield surface, the following procedure is adopted
to bring it back to the surface. A line is drawn from the centre of the initial yield surface
to the current stress state. This line cuts the yields surface at a point and the initial stress
components are defined based on that point. This correction is important to ensure an
accurate and smooth response.
CHAPTER 2: Large deformation finite element methodology
Centre for Offshore Foundation Systems 2-12
2.5 REFERENCES
Barbosa-Cruz, E. R. & Randolph, M. F. (2005). Bearing capacity and large penetration
of a cylindrical object at shallow embedment. Proc. Int. Symp. on Frontiers in
Offshore Geotechnics - ISFOG, Perth, Australia, 19-21 September 2005, 615-621.
Bathe, K-J. (1996). Finite element procedures. Prentice Hall, Upper Saddle River, New
Jersey.
Belytschko, T., Kennedy, J. M. & Schoeberle, D. F. (1978). Quasi-Eulerian finite
element formulation for fluid-structure interaction. Proceedings of Joint
ASME/CSME Pressure Vessels and Piping Conference. ASME: New York, 13,
ASME paper 78-PVP-60.
Carter, J. P. & Balaam, N. P. (1995). AFENA User Manual 5.0. Geotechnical Research
Centre, The University of Sydney, Sydney, Australia.
Dassault Systèmes (2007) Abaqus analysis users’ manual, Simula Corp, Providence,
RI, USA.
Donea, J, Fasoli-Stella, P. & Giuliani, S. (1977). Lagrangian and Eulerian finite element
techniques for transient fluid-structure interaction problems. In Trans. 4th Int. Conf.
on Structural Mechanics in Reactor Technology, San Francisco.
Einav, I., & Randolph, M. F. (2005). Combining upper bound and strain path methods
for evaluating penetration resistance. Int. J. Numer. Methods Eng. 63, No. 14, 1991-
2016.
Franck, R. M. & Lazarus, R. B. (1964) Mixed Eulerian-Lagrangian method. In Methods
in Computational Physics, Vol. 3: Fundamental methods in Hydrodynamics,
Academic Press: New York, 47–67.
CHAPTER 2: Large deformation finite element methodology
Centre for Offshore Foundation Systems 2-13
Ghosh, S. & Kikuchi, N. (1988). Finite element formulation for the simulation of hot
sheet metal forming process. Int. J. Eng. Sci. 26, 143-161.
Ghosh, S. & Kikuchi, N. (1991). An arbitrary Lagrangian-Eulerian finite element
method for large deformation analysis of elastic-viscoplastic solids. Comput.
Methods Appl. Mech. Eng. 86, No. 2, 127-188.
Hirt, C. W., Amsden, A. A. & Cook, J. L. (1974) An arbitrary Lagrangian-Eulerian
computing method for all flow speeds. J. Comput. Phys. 14, 227–253.
Hossain, M. S., Hu, Y., Randolph, M. F. & White, D. J. (2005). Limiting cavity depth
for spudcan foundations penetrating clay. Géotechnique 55, No. 9, 679-690.
Hossain, M. S. & Randolph, M. F. (2009). Effect of strain rate and strain softening on
the penetration resistance of spudcan foundations on clay. Int. J. Geomech. 9, No. 3,
122-132.
Hossain, M. S. & Randolph, M. F. (2010). Deep-penetrating spudcan foundations on
layered clays: numerical analysis. Géotechnique 60, No. 3, 171–184.
Hu, Y. & Randolph, M. F. (1998a). A practical numerical approach for large
deformation problems in soil. Int. J. Numer. Analyt. Meth. Geomech. 22, No. 5, 327-
350.
Hu, Y. & Randolph, M. F. (1998b). H-adaptive FE analysis of elastoplastic non-
homogeneous soil with large deformation. Comput. Geotech. 23, No. 1-2, 61-83.
Hu, Y. & Randolph, M. F. (1998c). Deep penetration of shallow foundations on
nonhomogeneous soil. Soils and Foundations 38, No. 1, 241-246.
Hu, Y., Randolph, M. F. & Watson, P. G. (1999). Bearing response of skirted
foundation on nonhomogeneous soil. J. Geotech. Geoenviron. Eng. 125, No. 11, 924-
935.
CHAPTER 2: Large deformation finite element methodology
Centre for Offshore Foundation Systems 2-14
Hughes, T. J. R., Liu, W. K. & Zimmerman, T. K. (1981). Lagrangian–Eulerian finite
element formulation for viscous flows. Comput. Methods Appl. Mech. Eng. 29, 329-
349.
Lu, Q., Randolph, M. F., Hu, Y. & Bugarski, I. C. (2004). A numerical study of cone
penetration in clay. Géotechnique 54, No. 4, 257-267.
Noh, W. F. (1964). A time-dependent two-space-dimensional coupled Eulerian-
Lagrangian code. Methods in Computational Physics, Academic Press, New York,
3,117-179.
van den Berg, P., Teunissen, J. A. M. & Huetink, J. (1991). Cone penetration in layered
media, an ALE element formulation, Proc. Int. Conf. on Computer Methods and
Advances in Geomechanics, 3, 1957-1962.
Wang, D., Hu, Y. & Jin, X. (2006). Two-dimensional large deformation finite element
analysis for the pulling-up of plate anchor. China Ocean Engineering 20, No. 2, 269-
279.
Wang, D., Hu, Y. & Randolph, M. F. (2010a). Three-dimensional large deformation
finite-element analysis of plate anchors in uniform clay. J. Geotech. Geoenviron.
Engng, ASCE 136, No. 2, 355-365.
Wang, D., White, D. J. & Randolph, M. F. (2010b). Large deformation finite element
analysis of pipe penetration and large-amplitude lateral displacement. Can. Geotech.
J. 47, No. 8, 842-856.
Zhou, H. & Randolph, M. F. (2006). Large deformation analysis of suction caisson
installation in clay. Can. Geotech. J. 43, No.12, 1344-1357.
CHAPTER 2: Large deformation finite element methodology
Centre for Offshore Foundation Systems 2-15
Zhou, H. & Randolph, M. F. (2007) Computational techniques and shear band
development for cylindrical and spherical penetrometers in strain-softening clay. Int.
J. Geomech. 7, No. 4, 287-295.
Zhou, H. & Randolph, M. F. (2009a). Resistance of full-flow penetrometers in rate-
dependent and strain-softening clay. Géotechnique 59, No. 2, 79-86.
Zhou, H. & Randolph, M. F. (2009b). Numerical investigations into cycling of full-flow
penetrometer in soft clay. Géotechnique 59, No. 10, 801–812.
Zhou, H. & Randolph, M. F. (2011). Effect of shaft on resistance of a ball penetrometer.
Géotechnique 61, No. 11, 973–981.
Zienkiewicz, O. C. & Zhu, J. Z. (1992). The superconvergent patch recovery and a
posterior error estimates. Part 1: The recovery technique. Int. J. Numer. Meth.
Engng. 33, No.7, 1331-1364.
Numerical modelling of pipe-soil interactions
Centre for Offshore Foundation Systems 3-1
CHAPTER 3
THE EFFECTS OF PENETRATION RATE AND STRAIN
SOFTENING ON THE VERTICAL PENETRATION
RESISTANCE OF SEABED PIPELINES
3.1 INTRODUCTION The problem of pipeline embedment in fine-grained sediments has been an active topic
of research for many years. Failure mechanisms and ultimate loads have been identified
based on classical plasticity theory (Randolph & Houlsby, 1984; Murff et al., 1989;
Martin & Randolph, 2006; Randolph & White, 2008c), small-strain finite element
analyses (Aubeny et al., 2005; Merifield et al., 2008; Merifield et al., 2009) and on
model tests (Verley & Lund, 1995; Dingle et al., 2008). Most of the theoretical studies
assume the pipe to be ‘wished-in-place’ and do not capture the change in geometry
during large amplitude deformation. Model tests performed in the centrifuge (Dingle et
al., 2008) are available, but they are too limited in number and confined to specific
cases to provide general guidance. A recent study has demonstrated the potential of
large deformation finite element (LDFE) analysis for estimating the vertical and lateral
response of pipes over significant displacements (Wang et al., 2010).
Dynamic motions during pipe lay, and potential entrainment of water, result in a
decrease in the shear strength of the seabed soil in the vicinity of the pipe. The amount
of softening depends on the sensitivity and ductility (meaning rate of softening) of the
soil. The effect of strain rate on the shear strength of soil has also been explored widely
(Casagrande & Wilson, 1951; Graham et al., 1983; Lefebvre & LeBoeuf, 1987;
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-2
Biscontin & Pestana, 2001; Lunne et al., 2006; Lunne & Andersen, 2007; Yafrate &
DeJong, 2007, Low et al., 2008). The combined effects of strain rate and softening on
the shear strength of the soil have been modelled recently in theoretical studies of deep
penetration problems (Einav & Randolph, 2005; Zhou & Randolph, 2007, 2009) but has
yet to be applied to the problem of pipe penetration resistance.
This chapter presents the results of a detailed and systematic parametric study of
the vertical embedment of pipelines in clay, incorporating the effects of strain rate and
softening on soil strength. Large deformation finite element analysis has been
performed using the commercial finite element software ABAQUS (Dassault Systèmes,
2007). The simple Tresca soil model was modified to account for the effects of strain
rate and softening. Rate parameters, sensitivity and ductility of the soil were varied to
investigate their effects on penetration resistance. The effects of soil strength non-
homogeneity and buoyancy on the vertical resistance of pipelines were also evaluated.
Simple relationships were then developed, expressing the non-dimensionalised vertical
penetration resistance as a function of the normalised penetration.
3.2 FINITE ELEMENT MODEL The large deformation finite element model developed for this study is based on the
“Remeshing and interpolation technique with small strain” (RITSS, Hu & Randolph,
1998a, 1998b), as described in Chapter 2.
3.2.1 Mesh, boundary conditions and material model A two-dimensional plane strain model was used (as shown in Figure 3.1). The two side-
edges of the model were restrained horizontally but free to move vertically, whereas the
bottom edge was restrained both vertically and horizontally. The pipe was considered as
a rigid body. The 2D plane strain element CPE6 of the ABAQUS element library was
used, which has three vertex nodes and three mid-side nodes. The optimal size of the
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-3
model and mesh density were established after trying several options. An incremental
displacement, taken as 1% of the diameter of the pipe, was imposed at the centre of the
pipe.
Figure 3.1 Mesh and boundary conditions This incremental displacement was verified to be small enough by running one set of
analyses with an incremental displacement of 0.1% of the diameter and confirming
negligible variation in the overall response. Contact between the pipe and the soil was
simulated by defining the pipe surface as the master surface and the soil surface as the
slave surface in ABAQUS. For considering friction between pipe and soil, the penalty
method in ABAQUS was used, with the maximum shear stress at the interface, τmax set
as αsum , where α is the interface roughness factor and sum is the mudline shear strength
of soil. In most cases, as detailed below, a value of α = 1/St was used where St is the
sensitivity of the soil, corresponding to the ratio of the intact and remoulded values. As
a result, the interface resistance was generally equal to the remoulded strength at the
mudline.
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-4
All the analyses were performed using an undrained total stress approach, based
on the Mohr-Coulomb soil model with zero friction angle (equivalent to the simple
Tresca model). For the elastic part of the linear-elastic-perfectly-plastic soil model, a
Poisson’s ratio of 0.499 (~0.5) was adopted to impose negligible volume change. The
Young’s modulus, E, of the soil was taken as 500 times the shear strength, su, at the
given depth.
3.2.2 Strain rate and strain softening The effects of strain rate and strain softening on shear strength were incorporated in the
analysis according to the model suggested by previous researchers (Einav & Randolph,
2005; Zhou & Randolph, 2007). The simple elastic-perfectly plastic Tresca soil model
was modified accordingly. At each remeshing and interpolation stage of the analysis,
the shear strength was modified to account for a reduction due to strain softening as
well as an enhancement due to high strain rate, with the strength due to the combined
effects given by
[ ] 0u/3
remremref
refmaxu s.e)1(),max(log1s 95ξξ−δ−+δ⎥
⎦
⎤⎢⎣
⎡⎟⎟⎠
⎞⎜⎜⎝
⎛γ
γγμ+=
&
&&
3.1
The first part of this relationship captures the effect of strain rate, γ& , with the reference
shear strain, refγ& , taken as 1%/hr or 3×10-6 s-1 and the rate parameter, μ, giving the rate
of strength increase per decade, taken in the range of 0.05-0.2 (Biscontin & Pestana,
2001; Lunne & Andersen, 2007). The maximum shear strain rate at a given location,
maxγ& is defined by
Dv
D/)( p31
max
.
δεΔ−εΔ
=γ
3.2
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-5
Where, 1εΔ and 3εΔ are major and minor principal strain, respectively, resulting from a
displacement increment, δ, D is the pipe diameter and vp is the pipe velocity.
The second part of Equation 3.1 represents the effect of strain softening. Here, su0
denotes the original shear strength at the reference shear strain rate prior to any
softening, δrem is the ratio of fully remoulded and initial shear strength and hence is the
inverse of sensitivity of soil. The sensitivity of the soil (St) ranges from 2 to 6 in the
case of marine clays (Randolph, 2004). ξ is the accumulated absolute plastic shear strain
( )pp 31 εΔ−εΔ at the Gauss point, where 1pεΔ and 3pεΔ are major and minor principal
plastic strain, respectively. ξ95 is the cumulative plastic shear strain for 95% shear
strength degradation, with typical value ranging from 10 to 50 (Randolph, 2004).
3.2.3 Validation of finite element model To validate the finite element model, initially a set of parameters were chosen that
matched those from a centrifuge study (Dingle et al., 2008). The prototype diameter of
the pipe was D = 0.8 m. Shear strength, su0 at any depth z for this simulation was taken
as 2.3+3.6z kPa (with z the equivalent prototype depth in m). Submerged (i.e. effective)
unit weight of the kaolin clay was 6.5 kN/m3. The sensitivity of the clay was considered
as 3.2, reflecting results from cyclic T-bar tests. The friction ratio at the pipe-soil
interface, α, was taken equal to the inverse of sensitivity (0.31).
The pipe was penetrated down to a depth of 0.45 times the diameter of the pipe.
The vertical reaction force during embedment, V was non-dimensionalised by Dsu0,
with su0 the nominal intact (reference shear strain rate) shear strength of the soil at a
level corresponding to the invert of the pipe. The measured and computed variations of
normalised vertical resistance on the pipe, V/Dsu0, with non-dimensionalised
embedment, w/D, are plotted in Figure 3.2.
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-6
0.0
0.1
0.2
0.3
0.4
0 1 2 3 4 5 6 7
V/Dsu0w
/D α=0.31, no strain effects
α=0.31, μ=0.1, St=3.2, ξ95=10,
reference strain rate = 0.000003 s-1, vp/D=0.015 s-1
centrifuge data
Figure 3.2 Comparison of penetration resistances with centrifuge result
When the effects of strain-rate and softening were not considered, the numerical results
are quite different from the centrifuge data. If both strain rate and softening effects are
taken into account, the computed response gives a good match with the centrifuge data
(Figure 3.2) and hence provides some validation of the large deformation finite element
approach. It should be noted that there are some jaggedness in the LDFE results for
normalised penetration resistance which are generally not observed in case of small
strain analyses. This is due to spatial variation of softened and rate dependent shear
strength in the soil domain. Also, this could happen due to interpolation of the shear
strength values from the previous step to the next step.
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-7
3.3 PARAMETRIC STUDY Details of the parameters chosen for the parametric study are shown in Table 3.1. A
base case for rate and softening parameters were chosen with 1000refD/pv =γ& ,
μ = 0.1, St = 3 and ξ95 = 20. The submerged unit weight of the soil, γ' = 5 kN/m3 was
considered for this case, since this is typical of deep water sediments. Keeping other
parameters equal to this base case, one parameter was varied at a time (as in Table 3.1).
Table 3.1 Parameters chosen for LDFE analyses
kD/sum γ'D/sum refp D/v γ& μ St ξ95
0 0.25 0, 100, 1000, 10000
0.1 3 20
0 0.25 1000 0, 0.05, 0.1, 0.15
3 20
0 0.25 1000 0.1 1, 2, 3, 6
20
0 0.25 1000 0.1 3 10, 20, 30
0 0.15, 0.25, 0.35
1000 0.1 3 20
1 1.25 0, 100, 1000, 10000
0.1 3 20
1 1.25 1000 0, 0.05, 0.1, 0.15
3 20
1 1.25 1000 0.1 1, 2, 3, 6
20
1 1.25 1000 0.1 3 10, 20, 30
20 10 0, 100, 1000, 10000
0.1 3 20
20 10 1000 0, 0.05, 0.1, 0.15
3 20
20 10 1000 0.1 1, 2, 3, 6
20
20 10 1000 0.1 3 10, 20, 30
20 6, 10, 14 1000 0.1 3 20
The pipe diameter, D was taken as 0.5 m for all the cases, although all results are
presented in normalised form and may be generalised to other diameters. The shear
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-8
strength of the soil, su0, at any depth z is assumed to vary according to the following
linear variation.
kzss um0u += 3.3
Where, sum is the (intact) shear strength at the mudline and k is the shear strength
gradient. Three values of κ (= kD/sum) corresponding to 0 (sum = 10 kPa and k = 0), 1
(sum = 2 kPa and k = 4 kPa/m) and 20 (sum = 0.25 kPa and k = 10 kPa/m) were selected.
The parametric study was repeated for these three values of κ. The maximum shear
stress at the pipe soil interface, τmax, was taken as αsum, with α equal to the inverse of
the soil sensitivity. So, the maximum shear stress at the interface is actually the mudline
remoulded shear strength. For each case, the pipe was penetrated down to a depth of 1D
and vertical resistance forces were recorded at every increment of displacement (1% of
the diameter of the pipe).
3.3.1 Effect of unit weight of soil The vertical penetration resistance may be considered as the sum of the geotechnical
resistance and a component due to buoyancy as the pipe becomes embedded within the
soil. The geotechnical resistance is generally expressed as a power law (Aubeny et al.,
2005), so the total vertical resistance can be written as:
0u2s
b
b
0u sD
DA
fDwa
DsV γ′
+⎟⎠⎞
⎜⎝⎛=
3.4
Where the first part of the right hand side of the equation denotes geotechnical
resistance (with power law parameters, a and b) and the second part represents the
resistance due to buoyancy. As is the submerged cross-sectional area of the pipe, so that
Asγ' is the (nominal) weight of soil displaced by the pipe. This is adjusted by a factor, fb,
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-9
which accounts for the enhanced buoyancy effect due to heave of the soil adjacent to the
pipe. Merifield et al. (2009) found that fb should be taken around 1.5.
Keeping all other parameters as in the base case, the effect of unit weight was
explored for soils with κ = 0 and κ = 20, taking submerged unit weights of 3, 5 and
7 kN/m3. It is evident from Figure 3.3 that the effect of unit weight is more pronounced
for soil with a high value of κ; this is because the value of the non-dimensional
parameter γ'D/sum lies in the range 0.15 to 0.35 for soil with κ = 0, whereas it is far
higher, varying from 6 to 14, for soil with κ = 20.
To evaluate appropriate values of fb for these cases, parallel sets of analyses
were run with and without considering the self-weight of soil. This allowed the
geotechnical resistance term to be isolated, and then the buoyancy term quantified by
subtracting the geotechnical resistance from the total resistance for analyses that
included self-weight of the soil. The value of fb was thus deduced, albeit on the
assumption that the geotechnical resistance is unchanged between the two cases,
implying no significant change to the penetration mechanism.
As shown in Figure 3.4(a), the average value of fb was ~1.4 for κ = 0. For soil
with a high shear strength gradient (κ = 20), the average value of fb was ~1.75 (Figure
3.4b). A value of fb = 1.5 was obtained for κ = 1. The value of fb is therefore a function
of the shear strength gradient, k. It was found to vary linearly with the non-dimensional
term kD/su,avg, where, su,avg is the average of shear strengths at the mudline and at a
depth of one pipe diameter. Hence kD/su,avg is bounded by 0 (for k = 0) and 2 (for
sum = 0). Figure 3.5 shows the variation of fb with kD/su,avg for these three cases, with a
simple linear trend line given by
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-10
( ) 38.1s/kD2.0f avg,ub += 3.5
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6 7 8
V/Dsu0w
/D
κ = 0
γ'D/sum = 0.15, 0.25, 0.35
(a)
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6 7 8
V/Dsu0
w/D
κ = 20
γ'D/sum = 6, 10, 14
(b)
Figure 3.3 Variation of vertical resistance for different submerged unit weights: (a) κ = 0; (b) κ = 20
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-11
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6 7 8
Non-dimensional Parameters
w/D
κ = 0γ'D/su0 = 0.25
(V/Dsu0)weighty - (V/Dsu0)weightless
(V/Dsu0)weightless
(V/Dsu0)weighty
fb
(a)
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6 7 8
Non-dimensional Parameters
w/D
κ = 20
(V/Dsu0)weighty - (V/Dsu0)weightless
(V/Dsu0)weightless
(V/Dsu0)weighty
fb
γ'D/su0
(b)
Figure 3.4 Buoyancy factor fb and other non-dimensional parameters with depth: (a) κ = 0; (b) κ = 20
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-12
0.6
0.8
1
1.2
1.4
1.6
1.8
2
0 0.4 0.8 1.2 1.6 2
kD/su,avg
f b
Figure 3.5 Variation of buoyancy factor fb with non-dimensional parameter kD/su,avg
3.3.2 Geotechnical resistance It is convenient to quantify the geotechnical resistance using closed-form expressions
suitable for routine design. Previously, power law relationships for the coefficients ‘a’
and ‘b’ in Equation 3.4 have been obtained as a function of embedment (Aubeny et al.,
2005; Merifield et al., 2009). Aubeny et al. (2005) gave power-law fit coefficients for a
‘wished-in-place’ pipe with an open trench directly above it, and hence did not consider
the change in geometry and formation of heave during continuous penetration. They
also did not consider the effects of strain rate and softening. Merifield et al. (2009) gave
results for ‘pushed-in-place’ pipes in uniform strength soil, and also did not consider the
effects of strain rate and softening. Aubeny et al. (2005) gave results for two extreme
values of κ, (0 and ∞) and plotted a best-fit curve between them to predict the resistance
for intermediate values of κ. The present study extends the previous work by
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-13
establishing simple relationships for a range of κ values, but also accounting for rate
effects and strain softening.
To compare results with Aubeny et al. (2005), the geotechnical resistances for different
κ values in this study were averaged for displacements down to 0.5D and fitted to a
best-fit power law curve.
Table 3.2 Comparison of coefficients ‘a’ and ‘b’ from literature and this study
Base case values for strain rate and softening parameters were adopted, as detailed
earlier. Coefficients ‘a’ and ‘b’ of Equation 3.4 for this base case and those from other
researchers are shown in Table 3.2 for rough pipes only. The coefficients are applicable
for the particular strain rate and softening parameters adopted in the base case. The
effect of these parameters on the response is explored below and a more refined set of
coefficients are derived.
3.3.3 Effect of normalised velocity refγ/Dpv &
Three values of refγ/Dpv & , 100, 1000 and 10000, were investigated for the parametric
study with all other parameters similar to the base case. Soil with zero rate-dependency
was also included for comparison. For κ = 1, the variation of vertical resistance for
different values of refγ/Dpv & is shown in Figure 3.6a.
a b Comments Aubeny et al.
(2005) 6.73 0.29 Wished-in-place, no strain effects
Merifield et al. (2008) 7.4 0.4 Wished-in-place,
no strain effects Merifield et al.
(2009) 7.1 0.33 Pushed-in-place, no strain effects
Base Case of Present Study 6.81 0.25
Pushed-in-place, Strain rate and strain
softening effects
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-14
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6 7
V/Dsu0w
/D
κ = 1μ = 0.1ξ95 = 20St = 3
= 100, 1000, 10000
No rate effect
(a)
ref
.
p D/v γ
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6 7
V/Dsu0,eq
w/D
κ = 1μ = 0.1ξ95 = 20St = 3
(b)
Figure 3.6 Effect of normalised penetration rate on vertical resistance: (a) V normalised by original shear strength; (b) V normalised by equivalent shear strength
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-15
As the normalised velocity increases from 100 to 10000, the vertical resistance
increases by more than 20%. It is helpful to identify what equivalent soil strength, su0,eq,
would reflect the effect of the increasing strain rate in the soil. Thus, instead of
normalising V by Dsu0, V was normalised by Dsu0,eq, where
( )[ ]refpr0ueq,0u D/vflog1ss γμ+= & 3.6
The factor fr reflects the average operative shear strain during each increment of pipe
penetration relative to the penetration rate, refp D/v γ& . This factor was varied in order
to find a suitable value to bring the various curves together, and a value of 0.7 was
found to be appropriate. It can be seen from Figure 3.6b that now points from all the
above cases fall in a narrow band, implying that a representative operative strength
during shallow pipe penetration is that at a strain rate corresponding to D/v7.0 p .
3.3.4 Effect of rate parameter μ The rate parameter, μ, was varied across values of 0.05, 0.1 and 0.15 to evaluate the
effect on the vertical resistance. Figure 3.7a shows the variation of vertical resistance
with normalised depth for different values of μ for κ = 1. The response for μ = 0 is also
included for comparison purpose. In this case also, vertical resistance increased by more
than 20% as μ was increased from 0.05 to 0.15. Normalising the vertical resistance by
su0,eq, using Equation 3.6 with fr = 0.7, all these curves fall in a narrow band (Figure
3.7b).
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-16
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6 7 8
V/Dsu0w
/D
κ = 1ξ95 = 20St = 3
μ = 0, 0.05, 0.1, 0.15
(a)
1000D/v ref
.
p =γ
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6 7 8
V/Dsu0,eq
w/D
κ = 1ξ95 = 20
St = 3
(b)
1000D/v ref
.
p =γ
Figure 3.7 Effect of rate parameter μ on vertical resistance: (a) V normalised by original shear strength; (b) V normalised by equivalent shear strength
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-17
3.3.5 Effect of sensitivity The sensitivity of the soil causes the operative strength during pipe penetration to be
lower than the original intact strength. Three values of sensitivity, St = 2, 3 and 6, were
considered and other parameters for all these cases were kept equal to the base case.
Higher sensitivity results in a higher rate of softening and lower penetration resistance.
Figure 3.8a shows variations of vertical resistance for different values of sensitivity for
soil with κ = 1. Results for soil with no softening, i.e. a sensitivity of unity, are also
shown in this figure. The vertical resistance reduces by more than 20% at shallow
depths as the sensitivity increases from 1 to 6. In all cases, the interface resistance was
taken equal to the inverse of the sensitivity times the mudline shear strength (i.e an
interface resistance = 0.167sum, 0.33sum, 0.5sum and sum for sensitivity values of 6, 3, 2
and 1 respectively). If a single value of interface friction is used for all cases,
corresponding to the base case value of 0.33sum, the variation in vertical resistance is
much less (as shown in Figure 3.8b). This is because now there is only the effect of the
change of sensitivity within the soil mass itself, while previously there was an
additional effect of the interface initially being at the different values of fully remoulded
strength, sum/St.
To characterise the effect of the change in soil sensitivity, an equivalent shear strength
was back-calculated as
( )[ ]95s /)D/w(f3remrem0ueq,u e1ss ξ−δ−+δ= 3.7
The term fs(w/D) in Equation 3.7 reflects the equivalent plastic strain (proportional to
w/D) undergone by the soil as it is deformed by the pipe. A value of 0.8 for the constant
fs was required to bring together the curves for the cases with different sensitivities and
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-18
the same α (Figure 3.9), implying that a representative operative strength during shallow
pipe penetration is that at a plastic strain corresponding to 0.8w/D.
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6 7 8V/Dsu0
w/D
κ = 1ξ95 = 20μ = 0.1
St = 6, 3, 2, 1
(a)
1000D/v ref
.
p =γ
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6 7 8
V/Dsu0
w/D
κ = 1ξ95 = 20μ = 0.1
St = 6, 3, 2, 1
(b)
1000D/v ref
.
p =γ
Figure 3.8 Effect of sensitivity on vertical resistance: (a) α = 1/St ; (b) constant α (= 0.33)
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-19
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6 7 8
V/Dsu0,eq
w/D
κ = 1ξ95 = 20μ = 0.1fs = 0.8
1000D/v ref
.
p =γ
Figure 3.9 Vertical resistances normalised by equivalent shear strength for different sensitivity values (fs = 0.8)
3.3.6 Effect of ductility parameter ξ95
The value of the ductility parameter, ξ95, was varied between 10, 20 and 30 to examine
its effect on the vertical resistance, as shown in Figure 3.10a for κ = 1. The penetration
resistance increases by more than 10% as the soil becomes more ductile, with ξ95
increasing from 10 to 30. Vertical resistances were normalised using an equivalent shear
strength based on Equation 3.7 and a value of fs = 0.8, which brought all the curves of
Figure 3.10a together, as shown in Figure 3.10b.
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-20
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6 7 8
V/Dsu0w
/D
κ = 1μ = 0.1St = 3
ξ95 = 10, 20, 30
(a)
1000D/v ref
.
p =γ
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6 7 8
V/Dsu0,eq
w/D
κ = 1μ = 0.1St = 3fs = 0.8
(b)
1000D/v ref
.
p =γ
Figure 3.10 Effect of softening parameter ξ95 on vertical resistance: (a) V normalised by original shear strength; (b) V normalised by equivalent shear strength
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-21
3.3.7 Combining effects of strain rate and softening parameters It may be seen from the above results that the strain rate and softening parameters have
marked effects on the overall penetration response of pipelines. An attempt is made here
to derive a single curve for the vertical penetration resistance with depth that accounts
for the effects of all these parameters. The current practice is to unify the penetration
response in soils of different strength profiles by normalizing the vertical force by the
shear strength of the soil at the invert of the pipe. In strain rate dependent, softening soil,
this approach is extended by normalizing the vertical resistance force by an equivalent
shear strength of the soil at the pipe invert, incorporating the effects of strain rate and
softening parameters. The equivalent shear strength, su0,eq is expressed as
( )[ ] ( )[ ] 0u)/)D/w(f3(
remremrefpreq,0u se1D/vflog1s 95s ξ−δ−+δγμ+= & 3.8
where fr and fs are the two constant factors introduced earlier and all other symbols are
as in Equation 3.1. The vertical resistance force, V, is normalised by Dsu0,eq. For a
particular κ value, the vertical resistance normalised by the equivalent shear strength at
the pipe invert was plotted in a single graph for all the cases of Table 3.1. It is seen from
Figure 3.11 that all the points fall into a narrow band when fr = 0.7 and fs = 0.8 are used.
These sets of values can be fitted by power law curves of the form V/Dsu0,eq = a.(w/D)b,
although two separate portions are required to provide a good fit, matching the curves at
w/D = 0.1. The best fits are expressed as:
0.1for w/D Dw2.5
DsV
0.1for w/D Dw103.3
Dw10
DsV
DsV
19.0
eq,0u
5.0
1.0D/weq,0ueq,0u
>⎟⎠⎞
⎜⎝⎛=
≤≈⎟⎠⎞
⎜⎝⎛
⎟⎟⎠
⎞⎜⎜⎝
⎛=
= 3.9
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-22
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6
V/Dsu0,eq
w/D
LDFE pointsFitted Curve
κ = 1
V/Dsu0,eq = 5.2.(w/D)0.19
V/Dsu0,eq =
V/Dsu0,eq,w/D=0.1x(10.w/D)0.5
Figure 3.11 Best fit power law curve for vertical resistance with depth
0
0.2
0.4
0.6
0.8
1
0 1 2 3 4 5 6
V/Dsu0,eq
w/D
κ = 20, 1, 0
Figure 3.12 Best fit power law curves for vertical resistances for different κ
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-23
3.3.8 Effect of variation of soil shear strength profile The power law equation fit stated above is valid for a particular shear strength profile
(or value of κ). Two other shear strength profiles were considered: uniform soil with
κ = 0 and soil with high shear strength gradient, κ = 20. The previous procedure for
κ = 1, was repeated for these two cases. For normalised displacement w/D ≥ 0.1,
V/Dsu0,eq = a.(w/D)b equations were fitted and values of coefficients ‘a’ and ‘b’ were
obtained. These coefficients, together with the value of V/Dsu0,eq at w/D = 0.1, are
tabulated in Table 3.3 for different values of κ. For w/D ≤ 0.1, the same curve as in
Equation 3.9 is used. The resulting curves are shown on Figure 3.12. For other values of
κ, the fitted curves can be interpolated.
Table 3.3 Power law fit coefficient ‘a’ and ‘b’ for different shear strength profile
3.4 SOIL FLOW PATTERN The main advantage of the present large deformation finite element analysis is its ability
to capture the changes in geometry and subsequent formation of heave when the pipe is
pushed into the soil. It is of interest to consider the shape and size of the soil heave for
different shear strength profiles, and also the soil flow mechanism during pipe
penetration for different cases. These are illustrated for a normalised displacement
w/D = 0.5 in Figure 3.13. It may be seen that for soil with low κ values, the mechanism
is relatively deep and wide with the heave decaying gradually with distance from the
κ a b (V/Dsu0,eq)w/D=0.1
0 5.4 0.23 3.18
1 5.2 0.19 3.35 20 4.7 0.17 3.18
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-24
pipe. On the other hand, for soil with high κ the mechanism is shallower – tending to
favour the weaker shallower soil – and with the heave concentrated close to the pipe.
κ = 0 κ = 0
κ = 1 κ = 1
κ = 20 κ = 20
Figure 3.13 Deformation pattern and instantaneous velocity field at w/D = 0.5 for different κ
3.5 CONCLUDING REMARKS The large deformation finite element analysis approach described in this chapter
provides a more rigorous analysis for vertical penetration of seabed pipelines than other
theoretical methods currently available in the literature. The method accounts for
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-25
geometry changes in the surface of the seabed in the form of heave, and also the effects
of strain rate and softening by modifying the simple linearly-elastic perfectly-plastic
Tresca soil model. Profiles of vertical penetration resistance with normalised
displacement obtained from this study were compared with results from a centrifuge
model test. Good agreement between these two results was obtained, supporting the
validity of the finite element approach.
The shear strength was modified in each step of the LDFE analysis by
multiplying it with two factors, accounting for (a) enhancement of strength at increasing
strain rates (relative to a reference value) and (b) softening as the soil is remoulded. It
was found that the strain rate and softening parameters have significant effect on the
vertical resistance. However, it was found that normalising the vertical resistance by an
equivalent shear strength, accounting for the strain rate and softening parameters, led to
a narrow band of values when plotted against normalised penetration of the pipe. These
sets of values were fitted by power law expressions, using two different equations for
w/D ≤ 0.1 and w/D > 0.1.
The effect of buoyancy was also explored in this chapter by varying the
submerged unit weight of the soil. The total vertical resistance was divided into a
geotechnical resistance and resistance due to buoyancy. Values for fb (a multiplication
factor on Archimedes’ buoyancy) were provided for different strength profiles, refining
the previous published value of 1.5.
The effect of soil non-homogeneity was also investigated in this chapter, leading
to different power law fit coefficients for soils with different shear strength profile,
expressed as κ = kD/sum. The heave patterns and associated soil flow kinematics were
illustrated for different values of κ.
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-26
It should be noted that the present chapter does not deal with excess embedment of
pipes due to dynamic lay effects. Another limitation of the present work is that due to
numerical difficulty, constant interface shear strength as a fraction of the original
mudline shear strength is assumed. In reality, the interface shear strength should be a
fraction of the remoulded shear strength at the penetration depth.
3.6 REFERENCES Aubeny, C. P., Shi, H. & Murff, J. D. (2005). Collapse load for a cylinder embedded in
trench in cohesive soil. Int. J. Geomech. 5, No. 4, 320-325.
Biscontin, G., & Pestana, J. M. (2001). Influence of peripheral velocity on vane shear
strength of an artificial clay. Geotech. Test. J. 24, No. 4, 423-429.
Bruton, D. A. S., White, D. J., Cheuk, C. Y., Bolton M. D. & Carr M. C. (2006). Pipe-
soil interaction behaviour during lateral buckling, including large amplitude cyclic
displacement tests by the Safebuck JIP. Proc. Offshore Technology Conf., Houston,
Paper OTC 17944.
Casagrande, A. & Wilson, S. D. (1951). Effect of rate of loading on the strength of
clays and shales at constant water content. Géotechnique 2, No. 3, 251-263.
Dassault Systèmes (2007) Abaqus analysis users’ manual, Simula Corp, Providence,
RI, USA.
Dingle, H. R. C., White, D. J. & Gaudin, C. (2008). Mechanisms of pipe embedment
and lateral breakout on soft clay. Can. Geotech. J. 45, No. 5, 636-652.
Einav, I., & Randolph, M. F. (2005). Combining upper bound and strain path methods
for evaluating penetration resistance. Int. J. Numer. Methods Eng. 63, No. 14, 1991-
2016.
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-27
Ghosh, S. & Kikuchi, N. (1991). An arbitrary Lagrangian-Eulerian finite element
method for large deformation analysis of elastic-viscoplastic solids. Comput.
Methods Appl. Mech. Eng. 86, No. 2, 127-188.
Graham, J., Crooks, J. H. A. & Bell, A. L. (1983). Time effects on the stress-strain
behaviour of natural soft clays. Géotechnique 33, No. 3, 327-340.
Hu, Y. & Randolph, M. F. (1998a). A practical numerical approach for large
deformation problems in soil. Int. J. Numer. Analyt. Meth. Geomech. 22, No. 5, 327-
350.
Hu, Y. & Randolph, M. F. (1998b). H-adaptive FE analysis of elastoplastic non-
homogeneous soil with large deformation. Comput. Geotech. 23, No. 1-2, 61-83.
Lefebvre, G., & LeBoeuf, D. (1987). Rate effects and cyclic loading of sensitive clays.
J. Geotech. Engrg. 113, No. 5, 476-489.
Low, H. E., Randolph, M. F., DeJong, J. T. & Yafrate, N. J. (2008). Variable rate full-
flow penetration tests in intact and remoulded soil. Proc. 3rd Int. Conf. on
Geotechnical and Geophysical Site Characterization, Taipei, Taiwan, Taylor &
Francis Group, London, 1087-1092.
Lunne, T. & Andersen, K. H. (2007). Soft clay shear strength parameters for deepwater
geotechnical design. Proc. 6th Int. Offshore Site Investigation and Geotechnics
Conf.: Confronting New Challenges and Sharing Knowledge Vol. 1, Society for
Underwater Technology, London, 151-176.
Lunne, T., Berre, T., Andersen, K. H., Strandvik, S. & Sjursen, M. (2006). Effects of
sample disturbance and consolidation procedures on measured shear strength of soft
marine Norwegian clays. Can. Geotech. J. 43, No. 7, 726-750.
Martin, C. M. & Randolph, M. F. (2006). Upper bound analysis of lateral pile capacity
in cohesive soil. Géotechnique 56, No. 2, 141-145.
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-28
Merifield, R. S., White, D. J. & Randolph, M. F. (2008). The ultimate undrained
resistance of partially embedded pipelines. Géotechnique 58, No. 6, 461-470.
Merifield, R. S., White, D. J. & Randolph, M. F. (2009). Effect of surface heave on
response of partially embedded pipelines on clay. J. Geotech. Geoenviron. Engng,
ASCE 135, No. 6, 819-829.
Murff, J. D., Wagner, D. A. & Randolph, M. F. (1989). Pipe penetration in cohesive
soil. Géotechnique 39, No. 2, 213-229.
Randolph, M. F., & Houlsby, G. T. (1984). The limiting pressure on a circular pile
loaded laterally in cohesive soil. Géotechnique 34, No. 4, 613-623.
Randolph, M. F. (2004). Characterization of soft sediments for offshore applications.
Keynote Lecture, Proc. 2nd Int. Conf. on Site Characterization, Porto, Portugal, Vol.
1, Millpress Science Publishers, Rotterdam, 209-231.
Randolph, M. F. & White, D. J. (2008a). Offshore Foundation Design – A Moving
Target. Proc. BGA International Conference on Foundations, Dundee, IHS BRE
Press, London, 27-59.
Randolph, M. F. & White, D. J. (2008b). Pipeline embedment in deep water: process
and quantitative assessment. Proc. Offshore Technology Conference, Houston, Paper
OTC 19128.
Randolph, M. F. & White, D. J. (2008c). Upper-bound yield envelopes for pipelines at
shallow embedment in clay. Géotechnique 58, No. 4, 297-301.
Verley, R. & Lund, K. M. (1995). A soil resistance model for pipelines placed on clay
soils. Proc. Int. Conference on Offshore Mechanics and Arctic Engineering,
Copenhagen, ASME, Vol. 5, 225-232.
Wang, D, White, D. J. & Randolph, M. F. (2010). Large deformation finite element
analysis of pipe penetration and large-amplitude lateral displacement. Can. Geotech.
J. 47, No. 8, 842-856.
CHAPTER 3: The effects of penetration rate and strain softening…
Centre for Offshore Foundation Systems 3-29
Yafrate, N. J. & DeJong, J. T. (2007). Influence of penetration rate on measured
resistance with full-flow penetrometers in soft clay. Proc. GeoDenver 2007-
Advances in Measurement and Modelling of Soil Behaviour, ASCE, GSP No. 173.
Zhou, H. & Randolph, M. F. (2007) Computational techniques and shear band
development for cylindrical and spherical penetrometers in strain-softening clay. Int.
J. Geomech. 7, No. 4, 287-295.
Zhou, H. & Randolph, M. F. (2009). Resistance of full-flow penetrometers in rate-
dependent and strain-softening clay. Géotechnique 59, No. 2, 79-86.
Numerical modelling of pipe-soil interactions
Centre for Offshore Foundation Systems 4-1
CHAPTER 4
LARGE LATERAL MOVEMENT OF PIPELINES ON A
SOFT CLAY SEABED: LARGE DEFORMATION FINITE
ELEMENT ANALYSIS
4.1 INTRODUCTION Lateral buckling due to internal temperature and pressure can move deepwater pipes a
large distance across the seabed, and can provide an effective mitigation against the
high loads that are generated if the pipe is fully constrained. During on-bottom lateral
buckling a pipeline might sweep laterally by 10 or 20 diameters across the seabed. To
model this behaviour in design, it is necessary to estimate the pipe-soil force-
displacement response. This behaviour cannot be assessed via a single conservative
response – it must be bracketed because both high and low geotechnical resistance can
hamper a design (Bruton et al. 2007).
The existing models for lateral pipe-soil resistance are mainly derived for the
analyses of pipeline stability under hydrodynamic loading, and are not suited to the
large-amplitude movements associated with buckling. The challenge is to extend
existing models to account for changes in the seabed geometry and the soil remoulding
effects that influence pipe-soil resistance during large amplitude cyclic displacements.
An important design parameter is the large-amplitude (‘residual’) lateral resistance,
which is generally assumed to be a steady value, once the pipe has ‘broken out’.
In recent years, lateral pipe soil interaction has been studied extensively by
researchers, with a particular focus on the undrained conditions that generally prevail
during pipe movements on the fine-grained soils found in deep water. It is now well
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-2
known that the resistance from soil during lateral movement is not governed by a
Coulomb friction force at the pipe soil interface. When a pipe is laid on the seabed, it
partially penetrates into the soil due to its self weight and due to other factors such as
the dynamic movement created by the laying process (Randolph & White, 2008a). This
initial embedment has a significant influence on the lateral pipe resistance. When the
pipe is forced to move laterally, it rises or dives depending on its weight relative to the
current bearing capacity (Zhang et al., 2002; Randolph & White, 2008b). The passive
resistance from the soil berm ahead of the pipe governs the lateral pipe-soil interaction
force (White & Cheuk, 2008; White & Dingle, 2010). Also, the large amplitude
movement of the pipe across the seabed during thermal expansion causes remoulding of
the soil. This remoulding has a marked influence on the operative soil strength and the
resulting lateral pipe resistance.
Relatively few previous studies have investigated pipe-soil resistance during
large displacements. Experimental investigations into this behaviour have been
performed at large scale and at reduced scale in a centrifuge, as reported by Bruton et al.
(2006), Cheuk et al. (2007), Bruton et al. (2008) and Cardoso & Silveira (2010). These
studies have led to empirical expressions for the lateral breakout resistance and the
subsequent steady residual resistance. These methods are commonly used in design, but
are subject to significant scatter and uncertainty (AtkinsBoreas 2008). More recently,
studies have examined the detailed mechanisms of soil movement during large
amplitude pipe displacements. Physical modelling of pipe-soil interaction in a
geotechnical centrifuge, using particle image velocimetry (PIV) to quantify the
observed movements, has been performed (Dingle et al. 2008). This study measured
lateral pipe-soil interaction forces and linked them to the observed failure mechanisms.
A more limited number of numerical studies have been performed. Hesar (2004)
reported one of the first finite element studies of large-amplitude lateral movement of
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-3
pipelines using ABAQUS/Explicit together with an ALE (Arbitrary Lagrangian-
Eulerian) adaptive meshing algorithm. The large-amplitude lateral behaviour of pipes
with different self weights was presented. Konuk & Yu (2007) also developed ALE
finite element models for the lateral movement of pipelines, using the software package
LS-DYNA. Both these studies showed the influence of initial embedment and pipe
weight on lateral pipe soil interactions, but did not include any attempt to generalise the
observed behaviour.
Physical model tests have shown the controlling influence of two large deformation
effects: (i) the changing topography of the seabed and (ii) the changing strength of the
sediment, as it is disturbed and remoulded. This chapter shows how finite element
analyses can capture both of these important effects. It is shown that with the help of the
large deformation finite element technique and a particular soil model that can capture
remoulding effects, the key aspects of lateral pipe-soil interaction that are observed in
physical model tests can be reproduced.
In this chapter, the large deformation finite element methodology implemented in
ABAQUS (Dassault Systèmes, 2007) is used to study the lateral movement of a pipe
embedded in soft clay. The pipe is moved a significant distance (seven times its
diameter) to ensure steady state soil resistance is reached. Initially, studies are presented
for ideal soil with shear strength not depending on the strain rate or softening. After that,
a rate dependent and softening soil model is adopted and the results are compared with
the ideal case. The concept of effective embedment – which is a summation of the
original embedment and an increase of embedment due to the berm ahead of the pipe –
is presented. It is shown that the normalised horizontal resistance may be related to the
effective embedment through a simple relationship, regardless of the other parameters.
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-4
4.2 SOIL CONSTITUTIVE MODELLING
Within each small strain analysis, the soil is modelled as an elastic perfectly-plastic
material, with failure according to the Tresca yield criterion. The particular value of the
Tresca shear strength for each element within the mesh is calculated when each small
strain analysis is initialised. This calculation accounts for the influences of cumulative
strain (i.e. the level of disturbance or remoulding) on the soil strength, as well as the
current strain rate.
The same methodology as described in previous chapters has been adopted to
modify the simple Tresca soil model to incorporate the effects of strain rate and
softening. After each analysis step, the original shear strength of the soil at every node,
su0, is modified to an updated value, su, according to the following formula.
( )[ ] 0u/3
remremref
p31u se1
D/vD/
)(,1maxlog1s 95ξξ−δ−+δ×
⎥⎥⎦
⎤
⎢⎢⎣
⎡⎟⎟⎠
⎞⎜⎜⎝
⎛⎟⎟⎠
⎞⎜⎜⎝
⎛γδ
εΔ−εΔμ+=
& 4.1
Where the first part of the equation adjusts the strength according to the strain rate
effect and the second part allows for strain softening. 1εΔ and 3εΔ are the major and
minor principal strains, respectively, resulting from a displacement increment, δ/D,
where D is the pipe diameter. vP is the pipe velocity. μ is the rate of strength increase
per decade of strain rate, refγ& is the reference strain rate. Softening is taken as an
exponential function of the cumulative absolute plastic shear strain ξ. Here δrem denotes
the ratio of fully remoulded strength to the initial strength, hence is the inverse of
sensitivity, St. The parameter ξ95 reflects the relative ductility of soil and is the value of
ξ at which the soil has undergone 95% remoulding.
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-5
Incompressible – i.e. undrained – conditions were imposed and a rigidity index,
E/su = 500 and Poisson’s ratio, ν = 0.49 were prescribed. The penalty method within
ABAQUS was used to model the interaction between the pipe and the soil surfaces. A
constant limiting shear stress, τmax = αsum, with α = 0.5, was imposed at the pipe- soil
interface, along with a zero tension condition. Here, sum denotes the undrained shear
strength of the soil at the mudline and α is termed the interface friction ratio.
4.3 TYPICAL FINITE ELEMENT MESHES A two-dimensional plane strain model with the pipe as rigid and the soil as deformable
was established. Six-noded triangular elements (CPE6 in the ABAQUS standard library)
were used for discretisation. The details of the mesh and boundary conditions are
illustrated in Figure 4.1. Very fine meshing with a minimum size equal to 0.05 times the
pipe diameter is used near the pipe. In each small displacement step, the pipe was
moved by a displacement of 1% of its diameter, before the remeshing and interpolation
procedure was performed.
The extent of the finest meshing from the centre of the pipe at the start of the
analysis was up to 1.5 times the pipe diameter on both sides and 1 time the diameter
below from the mudline. The sufficiency of these criteria is confirmed by the results
shown later in Figure 4.11, where it can be seen that soil in this zone undergoes
extensive strain softening by the pipe movement.
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-6
Figure 4.1 Mesh and boundary conditions
The advantage of this methodology is that the frequent remeshing does not allow
the mesh to become distorted. However, the pipe can be displaced by significant
cumulative distances. Snapshots of the finite element mesh of the soil domain at
different stages of pipe movement are shown in Figure 4.2 for the analysis described
later as Case G (Table 4.4). The initially low mesh density in the far field is replaced by
progressively finer elements as the pipe approaches. This updating is performed
automatically according to the refinement criteria described above. There is no manual
intervention during the analysis. The mesh is not ‘unrefined’ in regions that are no
longer undergoing intensive deformation, with the small elements being combined into
larger elements. However, this could potentially be an additional strategy within the
refinement process if it was desirable to reduce the total number of elements to facilitate
more rapid analysis. However, the distribution of strength and other material parameters
(e.g. the cumulative plastic strain) is highly variable within the ‘trail’ of deformation
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-7
behind the pipe. Any coarsening of the previously-refined mesh would smear these relic
features that are left within the seabed behind the pipe.
(a) Prior to horizontal movement
(b) u = 1D
(c) u = 4D
(d) u = 7D
(a) Prior to horizontal movement
(b) u = 1D
(c) u = 4D
(d) u = 7D
Figure 4.2 Soil mesh at different stages of movement (Case G)
Figure 4.2 shows clearly the growth of a soil berm ahead of the pipe as it sweeps
laterally.
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-8
4.4 IDEAL SOIL CASE Firstly, four cases with soil strength that is independent of strain rate or softening were
simulated. The soil shear strength increased with depth with a linear gradient k and a
mudline strength, sum. The soil and pipe parameters adopted in these cases are tabulated
in Table 4.1 and the details of the four cases are shown in Table 4.2. The parameters
tabulated in Table 4.1 were chosen because they allow the results to be compared with
the centrifuge modelling results presented by Dingle et al. (2008); the model test data
provide an excellent means to validate the present finite element analysis. Vmax in Table
4.2 denotes the penetration resistance experienced by the pipe at the end of initial
vertical embedment and V is the vertical load applied to the pipe during lateral motion.
Table 4.1 Pipe and soil parameters adopted in the study Parameters Values
Pipe Diameter, D 0.8 m
Shear strength of soil at mudline, sum 2.3 kPa
Shear strength gradient, k 3.6 kPa/m
Submerged unit weight of soil, γ' 6.5 kN/m3
Interface friction ratio, α 0.5
Initially the pipe was pushed vertically to different depths, unloaded to a specified
vertical load, and then moved laterally up to a distance of 7 times its diameter (D) under
a constant vertical load. Different initial embedments of w = 0.15D, 0.3D and 0.45D
(centrifuge test) were considered (cases B, C and D respectively). This range represents
typical values of as-laid pipe embedment. The vertical load applied during lateral
movement for all the cases was 0.19 times the maximum vertical load at the 0.45D
vertical penetration (as in the model test). Case A involved no over-penetration - i.e. the
initial embedment was achieved with the same vertical load as during the lateral
sweeping stage.
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-9
Table 4.2 Simulation details for cases of ideal soil
Case V (kN/m) V/Vmax (w/D)ini (w/D)fin (H/V)fin
A 2.96 1.00 0.02 0.026 0.63
B 2.96 0.33 0.15 0.021 0.63
C 2.96 0.24 0.30 0.011 0.78
D 2.96 0.19 0.45 0.002 0.86
The vertical movement of the pipe during the lateral motion – i.e. trajectory of the
pipe – is shown in Figure 4.3. In all cases except for Case A (no over-penetration) the
pipe moves upwards back to the original surface or just below the surface. For the no
over-penetration case, the pipe first moves downward for some distance and then moves
upward.
0
0.1
0.2
0.3
0.4
0.5
0 1 2 3 4 5 6 7Horizontal displacement, u/D
Pipe
inve
rt em
bedm
ent,
w/D
Centrifuge dataNo over-penetrationInitial embedment = 0.15DInitial embedment = 0.30DInitial embedment = 0.45D
Centrifuge data, Cases A, B, C and D
Figure 4.3 Trajectory of pipe invert during lateral motion (Ideal soil model, Cases A-D)
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-10
0
0.5
1
1.5
2
2.5
0 1 2 3 4 5 6 7
Horizontal displacement, u/D
Nor
mal
ised
horiz
onta
l res
istan
ce, H
/Ds u
0 Centrifuge dataNo over-penetrationInitial embedment = 0.15DInitial embedment = 0.30DInitial embedment = 0.45D
Centrifuge data, Cases A, B, C and D
Figure 4.4 Normalised horizontal resistance during lateral motion (Ideal soil model, Cases A-D)
0
0.5
1
1.5
2
2.5
0 1 2 3 4 5 6 7
Horizontal displacement, u/D
Equi
vale
nt fr
ictio
n fa
ctor
, H/V
Centrifuge dataNo over-penetrationInitial embedment = 0.15DInitial embedment = 0.30DInitial embedment = 0.45D
Centrifuge data, Cases A, B, C and D
Figure 4.5 Equivalent friction factor during lateral motion (Ideal soil model, Cases A-D)
The horizontal resistance during lateral motion, H, can be normalised by Dsu0,
where su0 is the shear strength of the soil at the invert of the pipe. The variation in
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-11
normalised lateral resistance, H/Dsu0 during lateral displacement, u/D, is shown in
Figure 4.4. As the shear strength of the soil changes with depth, the change in su0 is also
reflected in the normalised lateral response if H is normalised by Dsu0. For this reason, it
is useful also to normalize H by the operating vertical load, V, which remains constant
throughout the lateral movement. H/V is referred to as the equivalent lateral friction
factor and is shown in Figure 4.5.
Comparing Cases A-D, it is evident that over-penetration of the pipe leads to a
lateral response that softens after the pipe breaks out, as the pipe moves upwards
towards the ground surface. In terms of normalised resistance, H/Dsu0, this effect is
masked by the changing soil strength at the invert elevation. In all cases, a steady
residual resistance is reached after ∼3-4 diameters of lateral movement, when the pipe
invert is located at an embedment of < 0.03D below the original soil surface (and in
cases A and B the pipe was continuing to rise even closer to the surface at the end of the
simulation). This residual resistance is higher for a higher initial embedment (noting
that all cases involved the same weight of pipe). This is because a deeper initial
embedment leads to a larger soil berm being pushed ahead of the pipe.
The same general trend is apparent in the model test – the over-loaded pipes rise
upwards to the surface, and an approximately steady resistance is reached. However,
some significant differences are evident when the large displacement behaviour is
considered:
1. The LDFE trajectories are curved throughout the movement, whereas the model
test trajectory is approximately straight over the range 0.1 < u/D < 0.75.
2. The LDFE trajectories are tending towards an embedment of zero, rather than
the steady value of approximately w/D ∼ 0.05 seen in the model test.
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-12
3. The lateral resistance does not show a significant fall as the pipes move towards
the soil surface. The steady LDFE resistance is H/suD ~ 1.4 but the model test
plateau is H/suD ~ 0.8.
These three effects can all be linked to the influence of soil softening. Although the
LDFE reproduces the initial resistance accurately, once the soil is disturbed and
remoulded (which causes a drop in strength, in reality – i.e. the model tests), the
resistance is over-predicted. Softening is also an explanation for the differences in
trajectory. If a shear plane softens, then it forms a preferential failure mechanism which
the pipe will tend to favour during subsequent motion – hence the straight trajectory
during the initial movement in the model tests and the softening LDFE. Once a failure
plane forms in a particular direction, this is favoured and the pipe does not deviate. In
contrast, the LDFE without soil softening shows a curved trajectory, with a
continuously-changing direction.
Finally, if the soil becomes weaker with accumulating strain then it is possible for
the soil berm to grow in size, but continue to provide the same lateral resistance. The
increase in berm size is counteracted by a reduction in the soil strength. As a result, the
pipe can continue to scrape soil away from the seabed, growing the soil berm, whilst the
resistance remains constant.
4.5 REALISTIC SOIL CASE The influence of soil softening on the pipe trajectory and lateral resistance is confirmed
by the results from a further three LDFE analyses. These use the more realistic modified
Tresca model, which incorporated soil softening and strain rate-enhanced strength, as
defined through Equation 4.1. The adopted parameters for these aspects of the soil
constitutive model are shown in Table 4.3. The effects of rate and softening parameters
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-13
on the lateral response were not explored independently. It should be noted that practical
problems may have different pipe movement velocities and softening parameters.
Three initial embedments of w = 0.45D (the centrifuge test), 0.3D and zero over-
penetration (0.02D) were chosen for the realistic soil cases (Table 4.4). In each case, the
pipe was then moved laterally by 7 times its diameter. For all cases, the operating
vertical load was 0.19 times the maximum vertical load during penetration for the 0.45D
initial embedment case (see Dingle et al. 2008 for further details of the centrifuge
modelling procedures).
Table 4.3 Rate and softening parameters chosen Parameters Values Reference shear strain rate, refγ& 3 x 10-6 s-1
Vertical pipeline penetration rate, vp 0.015 D/s
Horizontal pipeline penetration rate, vp 0.05 D/s Rate of strength increase per decade, μ 0.1 Sensitivity of clay, St 3.2 Accumulated plastic strain at which 95% soil strength reduction occurs by remoulding, ξ95
10
0
0.1
0.2
0.3
0.4
0.5
0 1 2 3 4 5 6 7
Horizontal displacement, u/D
Pipe
inve
rt em
bedm
ent,
w/D
No over-penetration
Initial embedment = 0.3D
Initial embedment = 0.45D
Centrifuge data
Cases E, F, centrifuge data and case G
Figure 4.6 Trajectory of pipe invert during lateral motion (Realistic soil model, Cases E-G)
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-14
Table 4.4 Parameters for realistic soil cases with same load, V
Case V (kN/m) V/Vmax (w/D)ini (w/D)fin (H/V)fin
E 3.4 1.00 0.02 0.051 0.46 F 3.4 0.23 0.30 0.049 0.49 G 3.4 0.19 0.45 0.045 0.54
0
1
2
3
0 1 2 3 4 5 6 7Horizontal displacement, u/D
Nor
mal
ised
horiz
onta
l res
istan
ce, H
/Ds u
0
No over-penetrationInitial embedment = 0.3DInitial embedment = 0.45DCentrifuge data
Cases E, F and G
Figure 4.7 Normalised horizontal resistance during lateral motion (Realistic soil model, Cases E-G)
0
1
2
3
0 1 2 3 4 5 6 7
Horizontal displacement, u/D
Equi
vale
nt fr
ictio
n fa
ctor
, H/V
No over-penetration
Initial embedment = 0.3DInitial embedment = 0.45DCentrifuge data
Cases E, F and G
Figure 4.8 Equivalent friction factor during lateral motion (Realistic soil model, Cases E-G)
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-15
Figure 4.6 shows the pipe trajectory; Figure 4.7 and Figure 4.8 show the normalised
lateral resistance and equivalent friction factor responses respectively. The centrifuge
data from Dingle et al. (2008), which extends up to 3D lateral displacement, is also
superimposed on all three figures for comparison. It is clear that these cases – using a
soil model with softening – capture better the behaviour observed in the centrifuge
model tests. The initial steep increase in friction factor within 0.05D of lateral
movement observed in the centrifuge test is from the mobilisation of tensile resistance
at the rear of the pipe. This phenomenon is not captured in the present finite element
methodology and is hence not seen in the numerical results.
Figure 4.8 shows that the pipe experiences a steady state resistance after travelling
around 2 times its diameter, which is a shorter distance than the ideal soil cases (Figure
4.5). This is consistent with the steeper initial trajectory, which causes the pipe to
approach the soil surface more rapidly (compare Figure 4.6 and Figure 4.3).
It is also evident from Figure 4.6 that, unlike the ideal case, the pipe invert does not
reach as close to the original soil surface. For all cases it stabilizes at an embedment of
w ∼ 0.05D. As the invert of the pipe remains below the original surface, the pipe
continues to displace a significant volume of soil ahead of it, causing the size of the
berm to grow continuously – despite the lateral resistance remaining constant. It is
concluded that the increase in resistance from the growing berm area during the steady
resistance state is counterbalanced by the decrease in the operative soil strength due to
softening. It should be noted that the value of V for the realistic case is slightly larger
than the ideal case – this also has a minor influence on the elevation that the pipe
reaches.
The failure mechanisms associated with Case G are shown in Figure 4.9. Figure
4.10 shows the equivalent failure mechanism, identified through particle image
velocimetry (PIV) analysis of the model test, which was conducted behind a transparent
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-16
window by Dingle et al. (2008). This allowed images to be captured and analysed,
revealing the soil failure mechanism. Again, good agreement is evident between the
centrifuge model test observations (Figure 4.10) and the LDFE results (Figure 4.9d).
(b) u/D = 0.5(a) u/D = 0.1
(c) u/D = 1 (d) u/D = 7
(b) u/D = 0.5(a) u/D = 0.1
(c) u/D = 1 (d) u/D = 7
(b) u/D = 0.5(a) u/D = 0.1
(c) u/D = 1 (d) u/D = 7
Figure 4.9 Failure mechanisms during Case G
Figure 4.10 Soil deformation mechanism from a centrifuge model test (u/D = 3) (Dingle et al. 2008)
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-17
(a) u/D = 0.1
(b) u/D = 0.5
(c) u/D = 1
(d) u/D = 7
(a) u/D = 0.1
(b) u/D = 0.5
(c) u/D = 1
(d) u/D = 7
(a) u/D = 0.1
(b) u/D = 0.5
(c) u/D = 1
(d) u/D = 7
Figure 4.11 Soil softening during Case G
The calculated distributions of soil softening, expressed as the current strength
relative to the initial strength of that soil element, su/su0, are shown in Figure 4.11. These
plots illustrate clearly the softened failure plane that develops initially, and continues to
be mobilised for the first ∼ 0.75D of lateral movement. Once the pipe reaches the stable
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-18
embedment of w/D ∼ 0.05, the failure mechanism is predominantly a sliding failure
beneath the soil berm, but with further shear deformation occurring within the soil berm
(which causes continued softening).
The simulated pipe weight during the lateral sweep is the same for all Cases E - G,
and the steady state embedment is virtually identical for all analyses in each soil type.
To explore the influence of pipe weight on the steady condition, variations on Case G
(0.45D initial embedment) were performed using three other load ratios, V/Vmax = 0.1,
0.3 and 0.4 (Cases H – J, Table 4.5).
Table 4.5 Parameters for realistic soil cases with varying load, V
Case V (kN/m) V/Vmax (w/D)ini (w/D)fin (H/V)fin
H 1.78 0.1 0.45 0.000 0.59 I 5.35 0.3 0.45 0.098 0.54 J 7.12 0.4 0.45 0.170 0.64
-0.1
0
0.1
0.2
0.3
0.4
0.5
0 1 2 3 4 5 6 7
Horizontal displacement, u/D
Pipe
inve
rt em
bedm
ent,
w/D
V/Vmax = 0.40V/Vmax = 0.30V/Vmax = 0.19V/Vmax = 0.10
Cases H, G, I and J
Figure 4.12 Pipe invert trajectory during lateral motion (Varying vertical loads, Cases H-J )
For these cases, the pipe invert trajectory, normalised lateral response and equivalent
friction factors are shown in Figure 4.12, Figure 4.13 and Figure 4.14 respectively,
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-19
together with Case G. These four cases cover a fourfold range in simulated pipe weight,
with all other parameters being the same.
0
0.5
1
1.5
2
2.5
3
3.5
0 1 2 3 4 5 6 7
Horizontal displacement, u/D
Nor
mal
ised
horiz
onta
l res
istan
ce, H
/Ds u
0 V/Vmax = 0.40
V/Vmax = 0.30V/Vmax = 0.19V/Vmax = 0.10Cases H, G, I and J
Figure 4.13 Normalised horizontal resistance during lateral motion (Varying vertical loads, Cases H-J)
0
0.5
1
1.5
2
2.5
3
0 1 2 3 4 5 6 7Horizontal displacement, u/D
Equi
vale
nt fr
ictio
n fa
ctor
, H/V
V/Vmax = 0.40
V/Vmax = 0.30V/Vmax = 0.19V/Vmax = 0.10
Cases H, G, I and J
Figure 4.14 Equivalent friction factor during lateral motion (Varying vertical loads, Cases H-J)
It may be seen from Figure 4.12 that, for a higher vertical load, a greater steady state
pipe embedment is observed. The normalised steady state embedment of the pipe,
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-20
denoted by wres/D, can be related to the normalised vertical load, V/Dsu0. Figure 4.15
shows the LDFE results (taken at u/D = 7) and a simple power law fit with the
following equation:
47.20ures )Ds/V(01.0D/w = 4.2
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0 0.5 1 1.5 2 2.5 3 3.5
V/Dsu0
wre
s/D
LDFE pointsFitted curve
Case H
Case G
Case I
Case J
47.20ures )Ds/V.(01.0D/w =
Figure 4.15 Effect of vertical load on steady state embedment
4.6 EFFECTIVE EMBEDMENT APPROACH These analyses have shown that the lateral resistance is influenced by the size of the soil
berm ahead of the pipe, as well as the strength properties of the soil and the weight of
the pipe.
An elegant way of normalizing the lateral resistances, which incorporates all of
these influences, is to use a concept termed the effective embedment, as proposed by
White & Dingle (2011). When the pipe moves laterally, the berm ahead of it can be
considered as an additional contribution to the embedment, in addition to the vertical
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-21
position of the pipe invert below the original soil surface. A schematic diagram
explaining this concept of effective embedment is shown in Figure 4.16.
Figure 4.16 Schematic diagram explaining the effective embedment concept (as per White & Dingle, 2011)
The effective embedment is the summation of the two embedment components and
can be expressed as:
D/hD/wD/w 'berm
' += 4.3
Where,
η= berm
berm,t
'berm
AS
1h
4.4
Here, the area of the berm, Aberm is idealised as a rectangle with aspect ratio η. As
the pipe moves laterally, Aberm can be calculated as the cross-sectional area swept by the
pipe (i.e. the integral w⋅du) plus the volume displaced during the initial vertical
penetration.
When the realistic (softening) soil model is used, the effective height of the berm is
reduced by a factor St,berm. St,berm is the mobilised sensitivity of the soil in the berm. As
the soil in the berm may on average be only partially softened, St,berm is lower than St by
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-22
some fraction f (Wang et al. 2010). For the ideal soil case, St,berm is taken equal to unity,
since the soil strength is unchanged within the berm.
Using values of η = 1.5 and f = 0.5, all of the normalised lateral resistances for
displacements greater than u = 1D from the ideal (A – D) and realistic cases (E – J) fall
in a narrow band when plotted against effective embedment (Figure 4.17). The data are
fitted by a simple power law expression:
95.0'
0u Dw45.2
DsH
⎟⎟⎠
⎞⎜⎜⎝
⎛=
4.5
0
0.2
0.4
0.6
0.8
1
0 0.5 1 1.5 2 2.5
H/Dsu0
w'/D
Points from ideal soil case
Points from realistic soil case
Fitted power law expression
95.0'
0u Dw45.2
DsH
⎟⎟⎠
⎞⎜⎜⎝
⎛=
Figure 4.17 Variation of normalised lateral resistance with effective embedment
It can be concluded that the normalised horizontal resistance response is very well
correlated with the effective embedment as per the above relationship, for a wide range
of soil and pipeline parameters. This simple observation provides a useful contribution
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-23
to the development of a general model for large-amplitude lateral pipe-soil interaction,
accounting for changes in geometry and soil strength.
4.7 CONCLUDING REMARKS The large deformation finite element analysis method is a robust technique for studying
lateral pipe movements in clay. It is capable of capturing the changing seabed geometry
during large amplitude motion of pipelines on soft soils. But this method should be
coupled with a soil strength model that incorporates the effects of strain rate and
softening to predict the real behaviour.
Initially, the present study adopted an idealised soil model with shear strength not
depending on the strain rate or softening. After that, a more realistic soil model was
adopted and differences in response with the previous case were demonstrated. The
methodology and input parameters were validated by comparing the results with
available centrifuge data. Specific conclusions from the present study are as follows.
The trajectory or the vertical position of the pipe during lateral motion depends on
the weight of the pipe. The heavier the pipe the deeper it remains during lateral
movement. In the case of realistic soil, the pipe traverses with a much steeper upward
trajectory than for the ideal soil case. This phenomenon results in the pipe reaching a
steady state lateral resistance after travelling a shorter lateral distance.
For both ideal and realistic soil models, the deeper the initial embedment of the
pipe, the greater is the steady state lateral resistance. This is because deeper initial
embedment results in a larger area of the berm ahead of the pipe and hence greater
resistance.
In the case of the realistic soil model, the pipeline stays below the original soil
surface and still experiences a steady resistance from the soil. It is concluded that the
growth of the size of the berm is counterbalanced by the softening of the soil during
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-24
lateral motion. The steady state vertical embedment of the pipe is related to the
operating vertical load of the pipe. A simple power law relationship was presented to
express the steady state embedment of the pipe.
Finally, the concept of effective embedment was presented. The results of the study
showed that the large displacement lateral resistance depended on the initial embedment
of the pipe, the overloading ratio and also on the particular soil model. It is obvious that
the lateral resistance for the ideal soil cases is greater than that for the realistic soil
because there is no reduction of the soil strength in the former case. Also, the
differences in initial embedment and overloading ratio result in differences in the berm
size, which in turn result in differences in the lateral resistance.
It is therefore useful to model the lateral pipe-soil resistance with a method that
includes the actual embedment, the berm height and the sensitivity of the soil. The
effective embedment of the pipe, defined by the actual embedment of the pipe below the
mudline plus an increase of embedment due to the berm ahead of it, achieves this. The
lateral resistance responses for both the ideal soil and the realistic soil were related to
the effective embedment of the soil according to the same simple power law expression.
This relationship was valid for all the cases regardless of the other soil and pipe
parameters.
4.8 REFERENCES AtkinsBoreas (2008). SAFEBUCK JIP – Safe design of pipelines with lateral buckling.
Design Guideline. Report No. BR02050/SAFEBUCK/C, AtkinsBoreas.
Bruton, D., Carr, M. & White, D. J. (2007). The influence of pipe-soil interaction on
lateral buckling and walking of pipelines: the SAFEBUCK JIP. Proc. 6th Int. Conf.
on Offshore Site Investigation and Geotechnics, London. 133-150.
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-25
Bruton, D. A. S., White, D. J., Carr, M. & Cheuk, C. Y. (2008). Pipe-soil interaction
during lateral buckling and pipeline walk - the SAFEBUCK JIP. Proc. Offshore
Technology Conf., Houston, Paper OTC 19589.
Bruton, D. A. S., White, D. J., Cheuk, C. Y., Bolton, M. D. & Carr, M. C. (2006). Pipe-
soil interaction behaviour during lateral buckling, including large amplitude cyclic
displacement tests by the Safebuck JIP. Proc. Offshore Technology Conf., Houston,
Paper OTC 17944.
Cardoso, C. O., & Silveira, R. M. S. (2010). Pipe-soil interaction behavior for pipelines
under large displacements on clay soils – a model for lateral residual friction factor.
Proc. Offshore Technology Conf., Houston, OTC 20767.
Cheuk, C.Y., White, D. J. & Bolton, M. D. (2007). Large scale modelling of soil-pipe
interaction during large amplitude movements of partially-embedded pipelines. Can.
Geotech. J.44, No. 8, 977-996.
Dassault Systèmes (2007) Abaqus analysis users’ manual, Simulia Corp, Providence,
RI, USA.
Dingle, H. R. C., White, D. J. & Gaudin, C. (2008). Mechanisms of pipe embedment
and lateral breakout on soft clay. Can. Geotech. J. 45, No. 5, 636-652.
Hesar, M. (2004). Pipeline-seabed interaction in soft clay. Proc. 23rd Int. Conf. on
Offshore Mechanics and arctic eng., Vancouver, 225-232.
Konuk, I. & Yu, S. (2007). Continuum FE modelling of lateral buckling: study of soil
effects. Proc. of 26th Int. Conf. on Offshore Mechanics and Arctic Eng., San Diego,
OMAE2007-29376.
Randolph, M. F. & White, D. J. (2008a). Pipeline embedment in deep water: process
and quantitative assessment. Proc. Offshore Technology Conference, Houston, Paper
OTC 19128.
CHAPTER 4: Large lateral movement of pipelines on a soft clay seabed…
Centre for Offshore Foundation Systems 4-26
Randolph, M. F. & White, D. J. (2008b). Upper-bound yield envelopes for pipelines at
shallow embedment in clay. Géotechnique 58, No. 4, 297-301.
Wang, D, White, D. J. & Randolph, M. F. (2010). Large deformation finite element
analysis of pipe penetration and large-amplitude lateral displacement. Can. Geotech.
J. 47, No. 8, 842-856.
White, D. J. & Dingle, H. R. C. (2011). The mechanism of steady ‘friction’ between
seabed pipelines and clay soils. Géotechnique 61, No. 12, 1035–1041
White, D. J. & Cheuk, C. Y. (2008). Modelling the soil resistance on seabed pipelines
during large cycles of lateral movement. Marine Structures 21, 1, 59-79.
Zhang, J., Stewart, D. P. & M.F. Randolph, M. F. (2002). Modelling of shallowly
embedded offshore pipelines in calcareous sand, Journal of Geotechnical and
Geoenvironmental Engineering ASCE 128, N0. 5, 363–371.
Numerical modelling of pipe-soil interactions
Centre for Offshore Foundation Systems 5-1
CHAPTER 5
MODELLING LATERAL PIPE-SOIL INTERACTIONS
5.1 INTRODUCTION
In the previous chapter, it has been shown that real lateral pipe-soil interaction
behaviours during large movement can be modelled using large deformation finite
element analyses if a softening and rate dependent soil constitutive model is adopted.
The ‘effective embedment’ approach was shown to be successfully capturing the
‘residual resistance’ behaviour. What remains to be established is how to predict the
pipe trajectory, thus allowing the evolution of effective embedment with lateral
displacement to be predicted. In this chapter, a detailed parametric study was performed
varying the initial vertical embedment and pipe weight (a wider range) and the effective
embedment was related to the initial embedment through simple relationships. The
results of the parametric study are compared with existing empirical correlations and
new recommendations are provided.
Also, the initial ‘break-out’ resistance, an important phase of lateral pipe-soil
interaction, was not addressed in the previous chapter. The trajectory and the resistance
during pipe ‘break-out’ is best described within a plasticity framework by means of
failure envelopes. Equations of yield envelopes are available in the literature. In this
chapter, new equations derived from large deformation analyses are proposed. Also, the
entire pipe-soil lateral response from initial break-out to steady residual phases has been
modelled using a simple relationship.
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-2
5.2 PARAMETRIC STUDY
5.2.1 Input parameters A parametric study was undertaken, varying the initial vertical embedments and level of
overloading. Depending on whether the pipe is heavy or light, relative to the vertical
bearing capacity (Vmax) at the current embedment, lateral breakout will be accompanied
by either downward or upward movement. For this study, five values of initial
embedment (w/D = 0.1, 0.2, 0.3, 0.4 and 0.5) and five levels of overloading (V =
0.1Vmax,, 0.2Vmax, 0.4Vmax, 0.6Vmax and 0.8Vmax) were considered. In the previous
chapter, it has been shown that the pipe reaches a steady state after a lateral movement
of two times its diameter. For all the analyses in this chapter, the pipe was moved
laterally up to a distance of three times its diameter. The adopted soil parameters for this
study as given in Table 5.1, represents a typical soft seabed.
Table 5.1 Parameters used for parametric study
Parameters Values Pipe Diameter, D 0.8 m Shear strength of soil at mudline, sum 2 kPa Shear strength gradient, k 4 kPa/m Submerged unit weight of soil, γ' 5 kN/m3
Young’s modulus, E 500 su0 Poisson’s ratio, ν 0.499 Reference shear strain rate, refγ& 3 x 10-6 s-1
Vertical pipeline penetration rate, vp 0.015 D/s
Horizontal pipeline penetration rate, vp 0.05 D/s Rate of strength increase per decade, μ 0.1 Sensitivity of clay, St 3 Accumulated plastic strain at which 95% soil strength reduction occurs by remoulding, ξ95
10
Interface friction ratio, α 0.5
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-3
Interface friction between pipe and soil is a difficult parameter to choose because of
variation of soil shear strength along the pipe perimeter. A mid value of α = 0.5, in
between fully smooth and fully rough case was chosen for all analyses in this chapter.
5.2.2 Typical results – w/D = 0.3 Figure 5.1 to Figure 5.3 show the results from the parametric study for the initial
embedment of w/D = 0.3 (and the 5 values of normalised pipe weight). Figure 5.1
shows that the lighter pipes rise whereas the heavier pipes move downwards. In all
cases the pipe reaches a steady embedment after undergoing a lateral displacement of
around two diameters. The non-dimensionalised horizontal resistance H/Dsu0 for each
case is shown in Figure 5.2. The heavier the pipe, the greater is the lateral resistance,
when normalised by the local initial strength, because the heavier pipes create a higher
berm ahead.
-0.1
0.1
0.3
0.5
0.7
0.9
0 0.5 1 1.5 2 2.5 3u/D
w/D
Initial Embedment = 0.3DV/Vmax = 0.1, 0.2, 0.4, 0.6, 0.8
Figure 5.1 Typical trajectories of pipes during lateral movement for different pipe weights
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-4
0
1
2
3
4
0 0.5 1 1.5 2 2.5 3
u/D
H/D
s u0
Initial Embedment = 0.3D
V/Vmax = 0.1, 0.2, 0.4, 0.6, 0.8
Figure 5.2 Typical lateral responses of pipe for different pipe weights
0
1
2
3
0 0.5 1 1.5 2 2.5 3
u/D
H/V
Initial Embedment = 0.3D
V/Vmax = 0.1, 0.2, 0.4, 0.6, 0.8
Figure 5.3 Friction ratios for pipes with different operating vertical loads
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-5
By non-dimensionalising by the shear strength at the pipe invert, a change in su0 is
reflected in H/Dsu0. To see only the variation in resistance force during lateral motion,
instead of normalizing H by Dsu0, H can be normalised by the vertical force (V) applied
during horizontal motion, as shown in the Figure 5.3. This normalisation shows that the
lightest pipe undergoes a reduction in lateral resistance by a factor of nearly 2.5 in the
first diameter of movement, which is obscured in Figure 5.2 by the simultaneous fall in
su0 as the pipe rises.
Broad trends evident in Figure 5.2 and Figure 5.3 are that:
1. The initial breakout resistance, H/Dsu0, varies by a factor of only 1.6 across all 5
cases, even though the pipe weights different by a factor of 8, indicating the
strong influence of embedment.
2. As a corollary, when the breakout resistance is expressed as a friction factor,
H/V, a high variation is evident, spanning the range 0.3 – 1.7.
3. The residual friction factor spans a range of 2 across all 5 cases, with the highest
value corresponding to the heaviest pipe. The range is far higher in terms of
absolute resistance, H, with the highest value being 5.5 times greater than the
lowest (note that this is not directly evident in either Figure 5.2 or Figure 5.3
since both V and su0 vary among the different analyses at the residual state).
The responses of light and heavy pipes can be idealised as in Figure 5.4. Initially there
is a breakout resistance, Hbrk. After that, light pipes undergo upward movement and the
resistance approaches a steady value, referred to as the residual resistance, Hres. Heavy
pipes move downward resulting in increasing passive berm resistance. After large
displacements the pipe approaches a horizontal trajectory and a steady resistance,
although not as rapidly as in the case of light pipes.
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-6
Hbrk
Hres
Hres
Heavy Pipe
Hor
izon
tal R
esis
tanc
e
Horizontal displacement
Light Pipe
Figure 5.4 Idealisation of pipe response during lateral motion for ‘light’ and ‘heavy’ pipes
5.2.3 Initial yield envelopes and breakout resistance The maximum resistance within the initial movement of u/D = 0.1 was used to define
the breakout resistance, Hbrk. A clear way to present the lateral breakout resistance for
partially embedded pipelines with varying embedment and weight is via yield envelopes
defined in V-H space. Yield envelopes for shallowly embedded pipelines undergoing
small lateral displacements have been derived in previous studies either through
plasticity analysis (Randolph & White, 2008) or small strain finite element analysis
(Merifield et al., 2008). For a given initial embedment, the envelope defines the
different limiting combinations of vertical and horizontal load. The origin of the
envelope is at the point V = 0, H = 0 (assuming a non-bonded pipe-soil interface that
can sustain no tension). The apex point is at V = Vmax, H = 0 where Vmax is the vertical
bearing capacity at the pipe embedment. These envelopes are approximately parabolic
in shape and can be fitted to using (Merifield et al., 2008):
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-7
21
maxmaxmax
max
max VV1
VV
VH
VH
ββ
⎟⎟⎠
⎞⎜⎜⎝
⎛−⎟⎟
⎠
⎞⎜⎜⎝
⎛β= 5.1
Where
( )( )
21
21
21
21ββ
β+β
βββ+β
=β 5.2
The parameters β1 and β2 skew the ellipse, while the ratio Hmax/Vmax gives the relative
proportions of the ellipse.
For the present study, the parameters β1 and β2 and the ratio Hmax/Vmax were found to
vary with initial embedment according to the following linear best-fits:
)Dw66.0(89.01 −=β 5.3
)Dw64.0(87.02 −=β 5.4
)Dw55.0(31.0
VH
max
max += 5.5
The yield envelopes are shown in Figure 5.5 for each initial embedment. The maximum
horizontal load Hmax is reached at around V/Vmax = 0.5. Since the Tresca soil model
obeys normality, the yield envelopes for a pipe embedded in the soil will also obey
normality, so these can also be used to describe the pipe movement at yield by
redefining the axes as the conjugate plastic displacements. The point where pure
horizontal pipe movement occurs is often called the parallel point and lies close to
V/Vmax = 0.5 in the present case. For load levels lower than this point, pipe will undergo
upward movement on breakout, whereas heavier pipes will move downward.
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-8
0
0.1
0.2
0.3
0.4
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
V/Vmax
H/V
max
w = 0.1D
w = 0.5D
LDFE Parabola fit
Figure 5.5 Yield envelopes in V-H space (LDFE and parabola fit)
0
0.1
0.2
0.3
0.4
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
V/Vmax
H/V
max
w = 0.1D
w = 0.5D
present studyMerifield et al. (2008)
Figure 5.6 Yield envelopes from present study and Merifield et al. (2008)
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-9
Figure 5.6 shows comparison of the normalised failure envelopes from the
present study and those from Merifield et al. (2008). The parallel points for Merifield et
al. (2008) are at about V/Vmax = 0.4 compared to ~ 0.5 in the present study and the
envelopes are a slightly different shape.
The maximum values of normalised horizontal resistance, Hmax/Dsu0, from this
study are compared with the Merifield et al. (2008) values in Table 5.2. The latter
values are considerably lower than those from the present study, principally because
they are derived for a ‘wished-in-place’ pipe (without any heave from the penetration
process). In the present study the pipe is ‘pushed-in-place’, leading to local soil heave
and additional passive lateral resistance on breakout. Differences may also be attributed
to the softening and rate-dependency of the soil in this study, and the variation in soil
strength with depth (Merifield et al. 2008 considered uniform/constant strength soil).
In a different study, Merifield et al. (2009) modelled a pushed-in-place pipe
(albeit in non-softening soil) and provided an alternative expression for Hmax (Table 5.2).
These values are slightly lower but closer to those from the present study. The
maximum values of Vmax/Dsu0 from the present study and these two studies are also
compared in Table 5.3. Again, the Merifield et al. (2008) values are considerably lower
than those from Merifield et al. (2009) and from the present study. This reflects the
inclusion of heave during penetration, and also the strength profile increasing with
depth conditions (with kD/sum = 1.6), in the present study.
Merifield et al. (2009) gave expressions for vertical and lateral resistances for
fully smooth and fully rough pipes. The present LDFE analyses are based on an
interface friction ratio of 0.5. So, the results shown in Table 5.2 and Table 5.3 are
average values for fully smooth and fully rough cases presented in that paper.
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-10
Table 5.2 Comparison of Hmax/Dsu0 from present study and literature
Present study
Merifield et al. (2008)
Merifield et al. (2009)
w/D = 0.1 0.90 0.49 0.72 w/D = 0.2 1.20 0.88 1.10 w/D = 0.3 1.58 1.24 1.41 w/D = 0.4 1.78 1.58 1.68 w/D = 0.5 2.02 1.90 1.92
Table 5.3 Comparison of Vmax/Dsu0 from present study and literature
Present study
Merifield et al. (2008)
Merifield et al. (2009)
w/D = 0.1 4.48 2.95 3.27 w/D = 0.2 5.28 3.89 4.00 w/D = 0.3 5.74 4.57 4.49 w/D = 0.4 6.00 5.13 4.87 w/D = 0.5 6.27 5.61 5.19
In conclusion, the normalised shape of the V-H yield envelopes proposed by
Merifield et al. (2008) for uniform soil with constant strength are also broadly
appropriate for the present study, which considers soil softening and strain rate effects,
and slightly modified ellipse parameters to describe the failure envelope shape are
provided in Equations 5.1 to 5.5. The horizontal extent of the envelopes, defined by
Hmax/Vmax, is slightly greater due to the influence of heave during the initial embedment
process.
Although yield envelopes are rigorous methods to express pipe soil interaction
forces in the combined load spaces, they are only valid for the small lateral
displacements required to define the breakout resistance. After the pipe breaks out, the
shape of the soil berm ahead of the pipe compared to the void behind creates an
anisotropic geometry, and a change in the allowable V-H load combinations – including
an anisotropy within the failure envelopes themselves.
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-11
5.2.4 Residual friction factor As shown in Figure 5.3, the equivalent lateral friction factor, H/V, reaches a steady
residual value after the pipe is displaced by about two diameters. The residual friction
factor depends on the weight of the pipe and the initial embedment. These factors
control the pipe trajectory and the final steady embedment.
Table 5.4 Steadiness of the large displacement lateral resistance
winit/D V/Vmax
Horizontal resistance at
u/D = 2.5 (kN)
Horizontal resistance at
u/D = 3 (kN)
Variation over Δu/D = 0.5
(%)
0.1 0.12 0.15 24.5 0.2 0.81 0.82 +1.4 0.4 1.44 1.53 +5.5 0.6 2.75 2.80 +2.0
0.1
0.8 4.21 4.39 +4.1 0.1 0.73 0.66 -9.9 0.2 1.04 1.1 +6.3 0.4 2.36 2.31 -2.2 0.6 4.39 4.33 -4.1
0.2
0.8 6.48 6.75 +3.9 0.1 0.80 0.84 +4.8 0.2 1.37 1.38 +0.3 0.4 3.12 3.15 +0.9 0.6 5.97 6.27 +4.9
0.3
0.8 9.64 9.89 +2.5 0.1 1.03 1.05 +2.6 0.2 1.58 1.53 -2.8 0.4 4.20 4.18 -0.7 0.6 7.90 - -
0.4
0.8 11.50 - - 0.1 1.23 1.18 -4.8 0.2 1.90 1.87 -1.6 0.4 5.40 5.65 +4.3 0.6 9.74 - -
0.5
0.8 14.05 - -
The degree to which a steady state is reached within each analysis can be
defined by the change in lateral resistance during the last half diameter of lateral
movement, denoted by ΔH2.5<u/D<3, divided by the final value, Hu/D=3. These changes are
tabulated in Table 5.4. In most of the cases, the variation in lateral resistances is below
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-12
10%, which indicates that in these cases the pipe has reached a reasonably steady lateral
resistance. As seen from Table 5.4, in some cases the values of resistances at u/D = 3
are left blank. These are the cases where the pipe did not reach a steady state due to the
development of a very large berm; convergence in the numerical calculations was not
achieved beyond u = 2.5D in these cases due to overtopping of the soil berms.
Figure 5.7 shows the variation of residual friction factor with initial embedment
of the pipe. If the pipe starts at a deeper position, it ends up experiencing a higher
residual friction factor. Figure 5.8 shows the variation of residual friction factor with the
initial normalised vertical load. A higher vertical load leads to a higher residual friction
factor, although Hres/V for V/Vmax = 0.2 is lower than for V/Vmax = 0.1 in most cases.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 0.1 0.2 0.3 0.4 0.5 0.6
winit/D
Hre
s/V
V/Vmax = 0.1V/Vmax = 0.2V/Vmax = 0.4V/Vmax = 0.6V/Vmax = 0.8
Figure 5.7 Variation of residual friction factor with initial embedment A complication of this plot is that vertical load, V, is present on both axes. To see the
variation of absolute horizontal resistance with vertical load, both H and V were
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-13
normalised by Dsuo,init (Figure 5.9(a)). It is then clear that for a given initial embedment,
any increase in pipe weight leads to an increase in residual resistance.
0
0.5
1
1.5
0 1 2 3 4 5
V/Dsu0,init
Hre
s/V
initial w/D = 0.1initial w/D = 0.2initial w/D = 0.3initial w/D = 0.4initial w/D = 0.5
Figure 5.8 Variation of residual friction factor with normalised vertical load
Comparisons between different embedments remain complicated by the
variation in su0,init with depth, so Figure 5.9(b) uses Dsum for normalisation of all cases.
This shows that there is a consistent trend through the results for all embedments and
overloading ratios, except at low values of V/Dsum. It should be remembered that Figure
5.9(b) corresponds to a particular kD/sum(= 1.6) ratio and may change for different
strength parameter values.
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-14
0
1
2
3
4
5
0 1 2 3 4 5V/Dsu0,init
Hre
s/Ds u
0,in
it initial w/D = 0.1initial w/D = 0.2initial w/D = 0.3initial w/D = 0.4initial w/D = 0.5
V/Vmax = 0.1
V/Vmax = 0.2
V/Vmax = 0.4
V/Vmax = 0.6
V/Vmax = 0.8
(a)
0
2
4
6
8
10
0 2 4 6 8 10V/Dsum
Hre
s/Ds u
m
initial w/D = 0.1initial w/D = 0.2initial w/D = 0.3initial w/D = 0.4initial w/D = 0.5
V/Vmax = 0.1
V/Vmax = 0.2
V/Vmax = 0.4
V/Vmax = 0.6
V/Vmax = 0.8
(b)
Figure 5.9 Variation of residual resistance with vertical load normalised by (a) Dsuo,init; (b) Dsum
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-15
An alternative form of normalisation is to use the final embedment of the pipe
invert, relative to the initial undisturbed soil surface, (w/D)final (Figure 5.10). A greater
final embedment generally results in greater friction factor, except for the highly
unloaded pipes (low V/Vmax) which rise to the original soil surface. A spread of residual
friction factor is evident for the same final embedment and this is due to differences in
the size of the soil berm ahead of the pipe. Figure 5.11 shows the variation of residual
horizontal resistance (normalised by su0,final) with the final embedment of the pipe. The
differences in the size of the berm in this case also are responsible for different values of
lateral resistance for the same final embedment. In the same figure, the limiting
resistance, Hmax/suD, from Merifield et al. (2009) is also shown, which represents the
resistance to horizontal movement. Merifield et al. (2009) results are valid only for the
range 0 ≥ w/D § 0.5. At the residual state, the pipe is moving horizontally in all of the
LDFE analyses so the large discrepancy between this curve and the LDFE results is due
to the presence of the soil berm ahead of the pipe – which creates additional passive
resistance beyond that captured in the Merifield et al. (2009) resistance using an
embedment relative to the undisturbed soil surface. A method of accounting for this
additional passive resistance is explained in the following section.
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-16
0.4
0.6
0.8
1
-0.2 0 0.2 0.4 0.6 0.8 1 1.2
(w/D)final
Hre
s/V
V/Vmax = 0.1V/Vmax = 0.2V/Vmax = 0.4V/Vmax = 0.6V/Vmax = 0.8winit/D = 0.1
winit/D = 0.2
winit/D = 0.3
winit/D = 0.4
winit/D = 0.5
Figure 5.10 Variation of residual friction factor with normalised final embedment
0
0.5
1
1.5
2
2.5
3
3.5
-0.2 0 0.2 0.4 0.6 0.8 1 1.2
(w/D)final
Hre
s/su0
,fina
lD
V/Vmax = 0.1V/Vmax = 0.2V/Vmax = 0.4V/Vmax = 0.6V/Vmax = 0.8Merifield et al. (2009)
Figure 5.11 Variation of Hres/su0,finalD with (w/D)final
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-17
5.3 EFFECTIVE EMBEDMENT APPROACH For a lateral displacement beyond the initial breakout stage, an elegant way of
expressing the lateral resistance is by its variation with effective embedment. The
concept of effective embedment was proposed by White & Dingle (2011) and was
shown in the previous chapter. The effective embedment of the pipe invert is the
summation of the actual embedment (below the undisturbed soil surface) and an
additional term arising from the height of the soil berm ahead of the pipe. The
normalised effective embedment, w'/D, is expressed as:
η
+=+= berm
berm,t
berm ADS
1Dw
D'h
Dw
D'w 5.6
Aberm is the area of the berm and it is idealised as a rectangular block with aspect ratio η.
The height of the berm, hberm, is given byηbermA
. The value of η is generally in the
range 1.5-2.5. As the soil in the berm is remoulded, the effective berm height is
discounted by a factor of St,berm, to account for the softened soil not offering the same
level of passive resistance as if it were intact: berm,tbermberm S/h'h = . The lateral
resistance from all the LDFE analyses (5 initial embedments and 5 load ratios in each
case) are plotted against effective embedment in Figure 5.12 (the initial mobilisation
and breakout are ignored – only data from u/D > 0.5 are shown). All the responses fall
in a narrow band and are well-fitted by a power-law equation:
y'
0u Dwx
DsH
⎟⎟⎠
⎞⎜⎜⎝
⎛= 5.7
Where x = 2.82 and y = 0.72.
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-18
The results for Hmax/Dsu0 from Merifield et al. (2009), considering the actual
embedment from their study as effective embedment, are shown in the same figure
(Figure 5.12) for comparison. For a particular embedment, the horizontal resistance
from Merifield et al. is slightly higher because they did not consider softening of the
soil.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
0 0.5 1 1.5 2 2.5 3 3.5 4
H/Dsu0
w'/D
0 0.5 1 1.5 2 2.5 3 3.5 4
V/Dsu0
H/Dsu0
power law fit
H/Dsu0 by Merifield et al. (2009)
V/Dsu0
best fit curve
Figure 5.12 Lateral and vertical response using effective embedment approach
The normalised vertical load, V/Dsu0, at the end of lateral displacement, when
the lateral resistance had reached a steady value, are also shown and follow a particular
trend (Figure 5.12), which may be fitted by:
( ) ⎟⎠⎞
⎜⎝⎛ −= − qD/'wp
0ue1A
DsV 5.8
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-19
where A is a constant that corresponds to the limiting value of steady-state vertical
resistance during horizontal movement, in cases of very heavy pipes. The value of A
was estimated by analysing a case with a very heavy pipe (operating vertical load equal
to the 0.9 times the maximum vertical reaction force). The values assigned for A, p and
q to obtain the best fit curve were 3.7, 2.4 and 0.84 respectively.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
0 0.1 0.2 0.3 0.4 0.5 0.6
Initial Embedment, winit/D
Res
. Eff.
Em
b. (w
'/D) re
sidu
al
V/Vmax= 0.1
V/Vmax= 0.2
V/Vmax= 0.4
V/Vmax= 0.6
V/Vmax= 0.8
Figure 5.13 Variation of residual effective embedment with initial embedment
The steady state lateral resistance can be estimated using Equations 5.7 and 5.8.
For that, estimation of the residual effective embedment is necessary. Residual effective
embedment depends on the initial embedment and also on the loading ratios. Figure
5.13 shows the variation of residual effective embedment with initial embedment for
different values of V/Vmax. For each value of V/Vmax, the residual effective embedment
data can be fitted to linear curve as the following equation.
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-20
( ) ( ) nD/wmD/'w initialresidual += 5.9
The values of m and n for different values of V/Vmax are tabulated in Table 5.5, and may
be approximated as
( )maxV/V56.2m = 5.10
( )maxV/V38.0n = 5.11
Table 5.5 Best fit values of coefficients m and n for estimating residual effective embedment (Equation 5.9)
The value of effective residual embedment can now be predicted using
Equations 5.9 to 5.11. Using that value and Equations 5.7 and 5.8, the values of
Hres/Dsu0 and V/Dsu0 can be estimated, and hence the value of Hres/V.
The values of Hres/V predicted in this way have been compared with the four
empirical methods suggested previously. These are the relationships derived from
specific physical modelling studies by White & Dingle (2011) and Cardoso & Silveira
(2010) and the methods devised within the SAFEBUCK Joint Industry Project. White &
Dingle (2011) devised an expression for Hres/V, based on a set of 6 centrifuge model
tests, given by:
V/Vmax m n 0.1 0.31 0.02 0.2 0.45 0.06 0.4 1.03 0.10 0.6 1.57 0.21 0.8 2.03 0.35
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-21
res
initial max
H w V0.3 2V D V
⎛ ⎞= + ⎜ ⎟⎝ ⎠
5.12
Based on the results from a set of large-scale model tests, Cardoso & Silveira (2010)
proposed that Hres/V can be estimated as:
0.586 0.479
u,1Dres
u,1D
sH V0.2 0.929V 'Ds D
−⎛ ⎞ ⎛ ⎞
= + ⎜ ⎟ ⎜ ⎟⎜ ⎟ ⎜ ⎟γ⎝ ⎠ ⎝ ⎠ 5.13
where u,1Ds is the mean undrained shear strength between the soil surface and a depth of
one pipe diameter (i.e. for a linear strength profile, u,1D um sus s k D / 2= + ).
The SAFEBUCK studies generated simple correlations using wide databases of
model test results collated from many studies. The Phase I method was derived in White
& Cheuk (2005) and was published by Bruton et al. (2006), with Hres/V expressed as
res uH s11 0.65 1 expV 2 'D
⎡ ⎤⎛ ⎞= − − −⎢ ⎥⎜ ⎟γ⎝ ⎠⎣ ⎦
5.14
The SAFEBUCK Phase II method was derived in White & Cheuk (2009) and is
currently unpublished. Comparisons between the four methods and the LDFE results
are given in Table 5.6. In the italicised cases the pipe did not reach a steady state in the
numerical calculations (as shown in Table 5.4).
The calculated values of Hres/V using the four methods, normalised by the actual
value obtained through the finite element analyses, have been compared with the initial
embedments and initial vertical resistances, to identify whether any method exhibits
skew with respect to these input parameters (Figure 5.14). The Cardoso & Silveira
(2010) method shows a consistent bias except at very low pipe weights, over-predicting
the resistance by a factor of 2.4 on average. The SAFEBUCK Phase I method also over-
predicts the resistance consistently, with a greater discrepancy evident for lighter pipes.
The White & Dingle (2011) and SAFEBUCK Phase II methods are accurate on average,
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-22
but show more scatter than the method set out in this paper. The proposed equations in
the present study are closest to the parity line on average – which is not surprising since
the method has been calibrated against the data set being considered.
Table 5.6 Comparison of Hres/V value from present and previous studies
Calculated Hres/V
winit/D V/Vmax LDFE result Present
Method
White &
Dingle (2011)
SAFEBUCKPHASE I
SAFEBUCK PHASE II
Cardoso&
Silveira (2010)
0.1 0.20 0.50 0.36 0.83 0.45 0.67 0.2 0.46 0.50 0.39 0.83 0.45 0.91 0.4 0.43 0.53 0.43 0.83 0.45 1.26 0.6 0.56 0.58 0.45 0.83 0.45 1.55
0.1
0.8 0.65 0.63 0.48 0.83 0.45 1.80 0.1 0.59 0.49 0.43 0.81 0.54 0.76 0.2 0.45 0.51 0.48 0.81 0.54 1.04 0.4 0.52 0.57 0.55 0.81 0.54 1.46 0.6 0.66 0.64 0.61 0.81 0.54 1.80
0.2
0.8 0.75 0.72 0.66 0.81 0.54 2.10 0.1 0.61 0.50 0.49 0.79 0.63 0.83 0.2 0.49 0.53 0.57 0.79 0.63 1.14 0.4 0.58 0.61 0.68 0.79 0.63 1.62 0.6 0.76 0.70 0.76 0.79 0.63 2.00
0.3
0.8 0.89 0.80 0.84 0.79 0.63 2.33 0.1 0.63 0.50 0.55 0.76 0.70 0.89 0.2 0.51 0.54 0.66 0.76 0.70 1.23 0.4 0.67 0.65 0.81 0.76 0.70 1.75 0.6 0.82 0.77 0.92 0.76 0.70 2.16
0.4
0.8 0.93 0.89 1.02 0.76 0.70 2.52 0.1 0.66 0.51 0.62 0.74 0.78 0.94 0.2 0.55 0.56 0.75 0.74 0.78 1.32 0.4 0.76 0.69 0.93 0.74 0.78 1.87 0.6 0.90 0.84 1.07 0.74 0.78 2.32
0.5
0.8 0.98 0.98 1.19 0.74 0.78 2.71
Table 5.7 Comparison of mean and standard deviation values from different studies
Mean Standard Deviation (%) Present Study 1.03 11.2
White & Dingle (2011) 1.06 18.1 SAFEBUCK Phase I (2005) 1.38 32.8 SAFEBUCK Phase II (2009) 1.03 21.0 Cardoso & Silveira (2010) 2.44 51.1
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-23
0.1
1
10
0 0.1 0.2 0.3 0.4 0.5 0.6winit/D
(Hre
s/V) ca
lc/(H
res/V
) LD
FE
Proposed ModelParity Line
0.1
1
10
0 1 2 3 4 5V/Dsu0,init
(Hre
s /V)ca
lc/(H
res /V
)LD
FE
Proposed ModelParity Line
0.1
1
10
0 0.1 0.2 0.3 0.4 0.5 0.6winit/D
(Hre
s/V) ca
lc/(H
res/V
) LD
FE
White and Dingle (2011)Parity Line
0.1
1
10
0 1 2 3 4 5V/Dsu0,init
(Hre
s /V)ca
lc/(H
res /V
)LD
FE
White and Dingle (2011)Parity Line
0.1
1
10
0 0.1 0.2 0.3 0.4 0.5 0.6winit/D
(Hre
s/V) ca
lc/(H
res/V
) LD
FE
SAFEBUCK PHASE IParity Line
0.1
1
10
0 1 2 3 4 5V/Dsu0,init
(Hre
s /V)ca
lc/(H
res /V
)LD
FE
SAFEBUCK PHASE IParity Line
0.1
1
10
0 0.1 0.2 0.3 0.4 0.5 0.6winit/D
(Hre
s/V) ca
lc/(H
res/V
) LD
FE
SAFEBUCK PHASE IIParity Line
0.1
1
10
0 1 2 3 4 5V/Dsu0,init
(Hre
s /V)ca
lc/(H
res /V
)LD
FE
SAFEBUCK PHASE IIParity Line
0.1
1
10
0 0.1 0.2 0.3 0.4 0.5 0.6winit/D
(Hre
s/V) ca
lc/(H
res/V
) LD
FE
Cardoso and Silveira (2010)Parity Line
0.1
1
10
0 1 2 3 4 5V/Dsu0,init
(Hre
s /V)ca
lc/(H
res /V
)LD
FE
Cardoso and Silveira (2010)Parity Line
Figure 5.14 Ratios of (Hres/V)calculated to (Hres/V)LDFE varying with (a) winit/D; (b)
V/Dsu0,init
(a) (b)
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-24
However, it is significant that the standard deviation for the new method, at ~11%,
is significantly lower than the standard deviation values for the other methods (Table
5.7), suggesting that the governing mechanisms are better captured by this new analysis
approach.
It is important to note, however, that the method set out in this paper has not been
tested for other soil strength profiles, or against a range of experimental data. A
comparison with the databases used to generate the other calculation methods would be
an important part of validating (and perhaps refining) the proposed method in order to
establish that it is appropriate for use in practice.
5.4 ASSESSMENT OF THE FULL H/V RESPONSE Once the breakout and residual resistances are obtained (as discussed in section 5.2.3
and 5.2.4 respectively), it may be useful to model the complete resistance response for
the laterally sweeping pipeline, assuming that this takes place under constant vertical
pipe-soil load. The friction ratio responses shown in Figure 5.3 are well fitted by the
following expression:
( ) ( )⎟⎠⎞
⎜⎝⎛ −⎟
⎠⎞
⎜⎝⎛ −+⎟
⎠⎞
⎜⎝⎛ −=
μλ−− D/ubrkresD/uabrk e1V
HV
He1V
HVH b
5.15
The first term only controls the initial mobilisation of the breakout resistance, and is not
the main focus of this study. The second term provides a smooth exponential transition
from the breakout resistance to the residual value. To fit the LDFE results, the values of
coefficients a, b and μ remain essentially constant for all values of initial embedment
and pipe weight, but the value of λ (which determines the distance required to mobilise
the steady resistance) changes with pipe weight and initial embedment.
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-25
For any given initial embedment, the values of λ were fitted by a linear
relationship with V/Vmax, given by
d)V/V(c max +=λ 5.16
The values of c and d for different initial embedments are given in Table 5.8 and may be
approximated as
9.4D
w2.8c init −⎟
⎠⎞
⎜⎝⎛= 5.17
5.4D
w8.5d init +⎟
⎠⎞
⎜⎝⎛−= 5.18
The values of a, b and μ were assigned to be 25, 0.5 and 1.5 respectively for all cases.
The parameters a and b only relate to the initial mobilisation of Hbrk, not the transition to
Hres.
Table 5.8 Values of coefficients c and d for different values of initial embedments (Equation 5.16)
Typical fits for an initial embedment value of 0.3 times the diameter are shown in
Figure 5.15(a) and Figure 5.15(b) for ‘light’ and ‘heavy’ pipes respectively. Here ‘light’
pipes are those that rise during lateral movement. As shown previously, pipes with
V/Vmax lower than 0.5 respond in this way. ‘Heavy’ pipes have V/Vmax > 0.5 and dive
winit/D c d 0.1 -3.96 3.94 0.2 -3.16 3.31 0.3 -2.83 3.06 0.4 -1.22 2.05 0.5 -0.83 1.68
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-26
0
0.4
0.8
1.2
1.6
2
0 0.5 1 1.5 2 2.5 3u/D
H/V
(a)
V/Vmax = 0.1
V/Vmax = 0.2V/Vmax = 0.4
0
0.4
0.8
1.2
1.6
2
0 0.5 1 1.5 2 2.5 3u/D
H/V
(b)
V/Vmax=0.8
V/Vmax=0.6
Figure 5.15 Typical friction ratio responses with lateral displacement fitted with exponential equation: (a) Lighter pipes; (b) Heavier Pipes
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-27
during lateral movement. As seen in Figure 5.15, Equation 5.15 captures the lateral
response very well.
5.5 CONCLUDING REMARKS A wide ranging parametric study was performed involving large lateral movements
under constant vertical load. A range of initial embedments and overloading ratios were
considered. Differences in the characteristic response of ‘light’ and ‘heavy’ pipes were
explored and idealised responses for these cases identified. For assessment of the initial
break-out resistance, yield envelopes in V-H load spaces were derived and fitted by
generalised parabolic equations. For lateral movements beyond 50% of pipe diameter,
the concept of effective embedment was found to provide a useful normalisation that
accounts for any berm of soil ahead of the pipe. It was found that a steady state of
embedment and lateral resistance was reached, except for the heaviest cases (highest
V/Vmax). These showed a steady growth in resistance whilst diving to an ever greater
embedment.
The ‘effective embedment’ was related to the initial embedment and the operating
weight of the pipe. Simple relationships to predict the steady residual lateral resistance
were proposed. These perform well in comparison to other empirical correlations. If this
conclusion is shown to apply generally, then the new analysis will provide a useful tool
for the design of seabed pipelines with controlled lateral buckling.
5.6 REFERENCES Cardoso, C. O., & Silveira, R. M. S. (2010). Pipe-soil interaction behavior for pipelines
under large displacements on clay soils – a model for lateral residual friction factor.
Proc. Offshore Technology Conf., Houston, OTC 20767.
CHAPTER 5: Modelling lateral pipe-soil interactions
Centre for Offshore Foundation Systems 5-28
Dassault Systèmes (2007) Abaqus analysis users’ manual, Simulia Corp, Providence,
RI, USA.
Merifield, R. S., White, D. J. & Randolph, M. F. (2008). The ultimate undrained
resistance of partially embedded pipelines. Géotechnique 58, No. 6, 461-470.
Merifield, R. S., White, D. J. & Randolph, M. F. (2009). Effect of surface heave on
response of partially embedded pipelines on clay. J. Geotech. Geoenviron. Engng,
ASCE 135, No. 6, 819-829.
Randolph, M. F. & White, D. J. (2008). Upper-bound yield envelopes for pipelines at
shallow embedment in clay. Géotechnique 58, No. 4, 297-301.
Dingle, H. R. C., White, D. J. & Gaudin, C. (2008). Mechanisms of pipe embedment
and lateral breakout on soft clay. Can. Geotech. J. 45, No. 5, 636-652.
White, D. J. & Cheuk, C. Y. (2009). SAFEBUCK JIP: Pipe-soil interaction models for
lateral buckling design: Phase IIA data review. Report to Boreas Consultants
(SAFEBUCK JIP), UWA report GEO 09497. 185pp.
White, D. J. & Dingle, H. R. C. (2011). The mechanism of steady ‘friction’ between
seabed pipelines and clay soils. Géotechnique 61, No. 12, 1035–1041
Numerical modelling of pipe-soil interactions
Centre for Offshore Foundation Systems 6-1
CHAPTER 6
BREAKOUT BEHAVIOUR OF PARTIALLY EMBEDDED
PIPES IN UNIFORM CLAY USING LIMIT ANALYSIS
6.1 INTRODUCTION
Predicting the resistance and trajectory of a pipeline during buckling-induced lateral
movements is most elegantly described within a plasticity framework by means of yield
envelopes, which in undrained conditions are also plastic potentials, to define the load
combinations that will cause movement, and the resulting direction of that movement.
In this chapter, Finite Element Limit Analysis (FELA) software OxLim (Martin, 2011)
is used to explore the effect of the response at the pipe-soil boundary. Two conditions
are considered: a no-tension associated flow response – which is the formal condition
required for the plasticity bound theorems to apply – and the more natural condition for
an unbonded interface which is zero shear stress to be present when separation occurs
and the normal stress is equal to zero. OxLim computes strict lower and upper bound
plasticity solutions as described by Makrodimopoulos & Martin (2006, 2007 and 2008).
Adaptive mesh refinement using the approach described by Martin (2011) is used to
achieve tight bracketing of the exact collapse load. The adaptivity also reveals the
regions of highest strain, showing the form of the collapse mechanism.
The results from this study provide definitive yield envelopes for the bearing
capacity of a pipeline shallowly embedded on constant/uniform strength weightless
undrained soil for these two pipe-soil interface responses. For the case of a horizontal
soil surface, the results show that the analytical upper bounds provided by Randolph &
White (2008) are also very close to the exact lower bounds. For practical purposes, the
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-2
Randolph & White (2008) solutions provide adequate accuracy, but the FELA results
highlight more optimal failure mechanisms in certain conditions. The increase in the
size of the yield envelope when soil self-weight is introduced is investigated. It is shown
that a superposition approach is adequate to capture the effect of buoyancy, indicating
that the failure mechanism is not altered by the soil self-weight, for typical values of
submerged unit weight.
The growth of the yield envelopes in the presence of soil heave around the pipe
shoulders is also assessed. The shape of the soil heave is defined based on results from
large deformation finite element analyses, the strength of the heaved soil is assumed to
be the original shear strength of the soil, and limit analyses are performed for the
corresponding geometry. These limit analyses show good agreement with the finite
element results, and confirm that the presence of soil heave leads to increase in the
breakout resistance.
6.2 METHODOLOGY
The schematic of the problem studied is shown in Figure 6.1, which introduces the
notation used throughout this chapter. The pipe was modelled as a rigid body with the
shape of a regular polygon with 200 sides. The pipe was rigid and free to move in the
vertical and horizontal directions, but no rotation was allowed. The soil domain was
extended to five times the pipe diameter in the vertical as well as the horizontal
directions from the centre of the pipe, which was sufficient to eliminate boundary
effects. In each OxLim analysis, lower and upper bounds to the exact solution were
calculated iteratively using successive meshes that were automatically refined
adaptively. This automated process was repeated until the difference between the lower
and upper bounds was less than 1%.
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-3
D
w
θ
φ
10D
5D
Direction of pipe movement at failure
Figure 6.1 Schematic of the problem and notation
6.3 YIELD ENVELOPES
All cases analysed in this study are tabulated in Table 6.1. Initially, simple cases were
considered, with the pipe wished-in-place into weightless soil with a horizontal surface.
The soil was idealised as a rigid plastic material, as required in plasticity limit analysis,
failing according to a Tresca yield criterion at constant volume (which is representative
of undrained conditions) mobilising a shear strength of su. The soil strength was
uniform throughout the domain. Initially the pipe-soil interface was modelled in a
manner that obeys normality, and was either fully smooth (i.e. zero shear stress) or fully
rough (i.e. mobilising a shear stress of su).
Five pipe invert embedment cases equalling 0.1, 0.2, 0.3, 0.4 and 0.5 times the
diameter were studied. The pipe was probed in different directions making angles of 0°
to 180° with the vertical at intervals of 5°. The vertical and horizontal resistance forces
have been non-dimensionalised using the pipe diameter D and the shear strength su. The
resulting yield envelopes for smooth and rough interfaces are shown in Figure 6.2 and
Figure 6.3 respectively. The bearing capacity of the pipe increases with embedment and
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-4
with interface roughness. The aspect ratio of the envelopes decreases as the embedment
increases: for a given increase in embedment there is a greater rise in horizontal
capacity than vertical capacity.
Table 6.1 Different cases of OxLim analysis
Case Embedment Pipe-soil interface Soil weight Seabed
geometry Soil shear Strength
A 0.1D, 0.2D, 0.3D, 0.4D and 0.5D
Smooth Weightless Flat Uniform strength
B ,, Fully Rough ,, ,, ,,
C ,,
Fully rough at the front and smooth when breaks away
,, ,, ,,
D ,,
Roughness factor, α = 0.5
at the front and smooth when breaks
away
,, ,, ,,
E ,, ,, Soil weight considered, γ΄D/su = 0.8
,, ,,
F ,, ,, Weightless With soil heave ,,
G ,, ,, Soil weight considered, γ΄D/su = 0.8
,, ,,
0
0.5
1
1.5
2
2.5
3
-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
H /D
s u
V/Dsu
initial embedment = 0.1D, 0.2D, 0.3D, 0.4D and 0.5D
Figure 6.2 Yield envelopes for smooth pipes (Case A)
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-5
0
0.5
1
1.5
2
2.5
3
-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
H /D
s u
V/Dsu
initial embedment = 0.1D, 0.2D, 0.3D, 0.4D and 0.5D
Figure 6.3 Yield envelopes for rough pipes (Case B)
These results are compared with the analytical upper-bound solution given by Randolph
and White (2008) for an embedment of w/D = 0.4 for smooth and rough pipes in Figure
6.4. The results show excellent agreement for the case of smooth pipes, indicating that
the failure mechanism assumed in the Randolph & White (2008) analytical upper-bound
is near-optimal. In contrast, there is a significant discrepancy between the results of
these two studies for the rough pipe-soil interface.
0
1
2
3
-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
H /D
s u
V/Dsu
Smooth pipe (Oxlim analysis)Smooth pipe (Randolph and White, 2008)Rough pipe (Oxlim analysis)Rough pipe (Randolph and White, 2008)
Figure 6.4 Comparison of results for a typical pipe embedment (embedment = 0.4D)
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-6
This difference for rough pipes is due to the different interface behaviour assumed in
each analysis. The response of an interface is defined according to the notation set out in
Figure 6.5, which shows the normal stress σ and shear stress τ acting on the lower half
of the interface and the corresponding displacements of the upper half relative to the
lower half. If the interface is smooth, τ = 0 the yield surface is as shown in Figure 6.6.
From normality the yield surface is also the plastic potential, so the flow at failure is as
shown by the arrows. In the case of rough interface, a shear stress of su is sustained for
all σ > 0, leading to the yield envelope given by the bold line in Figure 6.7. The plastic
displacement vectors at failure are normal to this yield surface.
στ v
u
(a) Stresses (b) Displacement vectors
Figure 6.5 Stresses and corresponding displacement vectors of a horizontal interface
σ
τ
, dv
, du
Figure 6.6 Flow vectors for smooth interface
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-7
σ, dv
τ , du
su
Figure 6.7 Flow directions as per conventional plasticity analysis for rough interface If the two surfaces of the interface are moving apart at an angle (i.e. du ≠ 0) then a shear
stress of su is mobilised. The exception is the special case of zero tangential movement,
in which case the stress state can lie anywhere along the line σ = 0 at ⏐τ⏐ ≤ su. The
implication is that even when the pipe is separating from the soil, and a gap is opening,
the full shear strength of the interface is mobilised. These ‘phantom’ shear stresses are
required by classical plasticity theory, since they emerge from the requirement of
normality. However, they are unrealistic: if the interface is assumed to sustain no
tension and instead separates, then it cannot transmit shear stress through the opening
gap. Instead, a loss of contact means the normal and shear stresses reduce to zero (σ = 0,
τ = 0) so the stress point jumps to origin (O) meaning that whilst the yield envelope
remains unchanged, the plastic flow vectors all radiate from the origin as shown in
Figure 6.8, after Houlsby & Puzrin (1999). In this case the normality condition is
violated. The upper-bound solutions provided by Randolph and White (2008) are
consistent with Figure 6.8 because they do not consider any tangential traction in the
regions where the pipe separates from the soil.
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-8
, du
σ
su
τ
, dv
oA B
Figure 6.8 Direction of flow during loss of contact in no-tension surface (after Houlsby and Puzrin, 1999)
OxLim, following plasticity limit analysis theory, includes tangential traction even
when separation occurs, which leads to the discrepancy compared to the Randolph &
White (2008) solutions for a rough pipe (Figure 6.4). For pure downward vertical
movement of the pipe, when no breakaway occurs, the results agree closely. This
distinction between yield envelopes for no-tension interfaces that either sustain shear
stress at failure obeying normality, or which do not, has been studied in relation to the
bearing capacity of strip foundations by Houlsby & Puzrin (1999).
6.4 INTERFACE MODIFICATION
To replicate more natural condition for an unbonded interface, the pipe yield envelopes
were recalculated in OxLim after modifying the interface condition so that zero shear
stress was mobilised where separation occurs. To do this, the pipe was modelled as fully
rough (A-B in Figure 6.9) except for the portion where separation occurs, which was
modelled as fully smooth (B-C in Figure 6.9). The OxLim analyses were performed by
prescribing a specified direction of pipe movement, consistent with the modelled
interface conditions.
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-9
Smooth interfaceRough interface
Direction of pipe movement
A
B
C
Smooth interfaceRough interface
Direction of pipe movement
A
B
C
Figure 6.9 Schematic of pipe movement and combination of smooth and rough interface for OxLim analysis
The locus of the OxLim results using this interface modification for an embedment of
0.5D is shown in Figure 6.10 and compared with the results from Randolph and White
(2008). The OxLim results for high V/Dsu now match closely with the Randolph and
White (2008) results, compared with the poor agreement in Figure 6.4. However the
OxLim envelope lies slightly outside the Randolph & White (2008) solution for very
low (and negative) vertical loads. Also, the OxLim envelope, if mirrored into the
negative H domain, would not be convex.
These discrepancies at low V/Dsu arise because the OxLim envelope in Figure
6.10 is formed by simply joining the raw (V, H) results from each case. However, these
results all correspond to different boundary value problems because in each case the
separation point is different and so the relative lengths of rough and smooth pipe
interface are different. Following the usual FELA approach, each (V, H) result is
derived by firstly assessing the vector of load combinations that gives the minimum
dissipation rate for the specified direction of pipe movement. From that vector of
resultant loads, the actual (V, H) result is the point on the vector that is tangential to the
yield envelope for that boundary value problem.
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-10
0
0.5
1
1.5
2
2.5
3
-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
H /D
s u
V/Dsu
Randolph and White (2008)
Oxlim with smooth interface at the rear of the pipe
Figure 6.10 Comparison of result with Randolph and White (2008) for w/D = 0.5 after interface modification (Case C)
However, when considering a hybrid yield envelope, composed of segments related to
different separation points and therefore different boundary value problems, the full
load vectors from each solution must be considered, since the relevant (V, H) load point
may no longer be tangential to the yield envelope for the problem analysed.
To obtain the true yield envelope for the hybrid case in which the separation
point and the rough-smooth interface boundary varies with the pipe movement direction,
the load vectors from each solution have been re-combined to define the collapse loads
based on the hybrid case. To illustrate this re-combination process, the case with an
initial embedment of 0.5D is used. The pipeline was displaced upwards at angles from 0
< θ < 50 degrees, at one degree increments, with the separation point and rough-smooth
interface boundary changing for each case. For each raw OxLim (V, H) result – which
is applicable only to the non-varying interface case – the vector of potential load
combinations that would cause this collapse mechanism was reconstructed. These
vectors are perpendicular to the specified direction of movement. Figure 6.11(a) shows
the reconstructed vectors for θ = 10°, 20°, 30° and 40°, along with the full locus of raw
(V, H) results at 1° intervals.
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-11
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
0 0.1 0.2 0.3 0.4
H /D
s u
V/Dsu
Analytical solution (Randolph & White, 2008)OxLim result after optimization
(b)
0
0.2
0.4
0.6
0.8
1
1.2
-0.2 0 0.2 0.4
H /D
s u
V/Dsu
OxLim result
Analytical solution (Randolph & White, 2008)
Vectors perpendicular to the direction of pipe movement
(a)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
0 0.1 0.2 0.3 0.4
H /D
s u
V/Dsu
Analytical solution (Randolph & White, 2008)OxLim result after optimization
(b)
0
0.2
0.4
0.6
0.8
1
1.2
-0.2 0 0.2 0.4
H /D
s u
V/Dsu
OxLim result
Analytical solution (Randolph & White, 2008)
Vectors perpendicular to the direction of pipe movement
(a)
Figure 6.11 Optimizing OxLim Result for modified interface; (a) Before optimisation, (b) After optimisation
The intersections between the reconstructed vectors and the adjacent cases (i.e. at higher
and lower values of θ) then define the new yield envelope that represents the hybrid
case. This process recovers the normality condition (Figure 6.11(b)).
Analytical solution for low vertical loads
For low vertical loads, an analytical solution for the (V, H) collapse can be obtained
using a simple wedge failure mechanism (Randolph & White 2008, Figure 6.12(a)).
When the pipe is displaced at angle θ the only work dissipated is along failure plane L,
which is of length 0.5.D.tanθ. The hodograph for displacement components is shown in
Figure 6.12(b). If the shear strength of the soil is su, the internal work done is equal to
suLδ. The external work done is H.δsinθ - Vδcosθ.
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-12
θθ
θV
H
D/2L
Δu= δ.sinθ
θ
Δw= δ.cosθ
δ
(a) Schematic (b) Hodograph
Figure 6.12 Simple wedge failure mechanism with corresponding hodograph From the principle of virtual work:
H.δsinθ - Vδcosθ = suLδ = 0.5.suDtanθδ 6.1
Hence
H = Vcotθ + 0.5.suD.secθ 6.2
For a particular value of V, the minimum H can be obtained by differentiating with
respect to θ, to give
θθ+θ−=θ
tan.secDs5.0eccosVddH
u2 = 0
6.3
and hence
θθ= 2
u
tan.sin5.0DsV
6.4
Substituting this into Equation 6.2 leads to the maximum value of H, given by
θθ+
=cos2
)sin1(DsH 2
u
6.5
So, for a particular angle of pipe movement θ, Equations 6.4 and 6.5 give optimum
values of V and H. The Randolph and White (2008) analytical upper-bound degenerates
to this simple wedge failure for low vertical loads.
Figure 6.13 compares the wedge solution with the reconstructed upper and lower
bound results from OxLim. For 0º < θ < 32º, the reconstructed OxLim solutions are
better than the simple wedge mechanism. Beyond this angle, the OxLim results bracket
the wedge case.
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-13
0.92
0.93
0.94
0.95
0.96
0.97
0.98
0.99
1
1.01
0 5 10 15 20 25 30 35 40 45 50 55
Hox
lim/ H
wed
ge
Angle of pipe movement with the vertical, θ
Upper bound
Lower bound
θ0ο < θ < 50ο
Figure 6.13 Difference between optimised OxLim result and analytical solution for low vertical loads
The adaptive mesh refinement in OxLim reveals the detail of the failure mechanism,
showing how the breakout resistance is reduced to below the simple wedge case. The
adapted mesh for a near-vertical breakout (θ = 1º) is shown in Figure 6.14. The
adaptivity scheme focuses the elements according to the magnitude of shear strain. The
zoomed portion of this figure shows that the failure mechanism involves distributed
shear strain within a vertical wedge. However, close to the ground surface there is a
hinge point and a rigid block adjacent to a 45º wedge of distributed shear.
This feature is similar to the near-surface detail in the trapdoor failure
mechanism reported by Martin (2009). Although this mechanism is of little practical
relevance, the parallel with the trapdoor solution is interesting, and provides an
explanation for the ∼5% reduction in breakout resistance compared to the simple wedge
case (Figure 6.13). Failure mechanisms for pipe movement at θ = 15˚, 25˚, 35˚ and 45˚
are also shown in Figure 6.15(a-d). The mechanisms show details that differ from the
simple wedge case, explaining the divergence in Figure 6.13.
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-14
Wedge mechanism with an ear near surface
Figure 6.14 Failure mechanism for pipe movement direction of 1 degree to the vertical
To generate the full failure envelopes using this interface modification technique,
further OxLim analyses were performed, spanning all initial embedment cases. These
led to the definitive yield envelopes for the no-tension interface condition, which are
shown in Figure 6.16. For comparison, the results of Randolph & White (2008) are
shown, as well as small strain finite element analyses that were performed for this study,
using the same interface idealisation. In general, the Randolph & White (2008)
analytical upper-bound are equal to or marginally greater than the OxLim results. The
FE results are also generally close to the OxLim results, but are in error for purely
vertical loading by an underestimation of up to 10%.
This error is attributed to some separation occurring in ABAQUS analyses near
the pipe shoulder for pure vertical movement. If the same analysis is repeated with the
pipe and soil surfaces tied together, there is no separation and the FE results fall close to,
but marginally outside the theoretical results – as is typically found in similar plane
strain problems (e.g. Gourvenec & Randolph 2003). This approach cannot be used for
the other directions of loading, since breakaway would not then occur.
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-15
Failure mechanisms revealed by the adaptive mesh refinement in OxLim for
pure vertical and pure horizontal pipe movements are shown in Figure 6.17(a) and
Figure 6.17(b) respectively. These illustrate the similarity between the optimal
mechanism identified by OxLim and the analytical upper-bound reported by Randolph
& White (2008).
(a) 15˚ with vertical (b) 25˚ with vertical
(c) 35˚ with vertical (d) 45˚ with vertical
Figure 6.15 Failure mechanisms for different directions of pipe movement
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-16
0
0.5
1
1.5
2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
H/D
s u
V/Dsu
Randolph and White (2008)
Optimized OxLim result
ABAQUS small strain analysis
w/D = 0.1
0
0.5
1
1.5
2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
H/D
s u
V/Dsu
Randolph and White (2008)
Optimized OxLim result
ABAQUS small strain analysis
w/D = 0.2
0
0.5
1
1.5
2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
H/D
s u
V/Dsu
Randolph and White (2008)
Optimized OxLim result
ABAQUS small strain analysis
w/D = 0.3
0
0.5
1
1.5
2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
H/D
s u
V/Dsu
Randolph and White (2008)
Optimized OxLim result
ABAQUS small stra in analysis
w/D = 0.4
0
0.5
1
1.5
2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
H/D
s u
V/Dsu
Randolph and White (2008)Optimized OxLim resultABAQUS small strain analysis
w/D = 0.5 A
B
Figure 6.16 Optimised yield envelopes for rough pipe soil interface
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-17
(a) Vertical pipe movement (point B in Figure 6.16)
(b) Horizontal pipe movement (point A in Figure 6.16)
Figure 6.17 Adaptive mesh refinement for pure vertical and horizontal pipe movements
in flat seabed (w/D = 0.5)
6.5 EFFECT OF SOIL WEIGHT
To extend the previous analyses, the effect of including soil self weight on the resulting
yield envelopes was explored. The results from analyses denoted as Cases D and E in
Table 6.1 are shown in Figure 6.18. In both cases the interface roughness factor was
chosen to be 0.5. For the weighty soil case, the non-dimensional term γ΄D/su was taken
equal to 0.8, which is typical for a soft seabed soil, where γ΄ is the submerged unit
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-18
weight of the soil. An increase in breakout resistance due to introduction of soil weight
can be observed for all embedment ratios in Figure 6.18.
0
0.5
1
1.5
2
2.5
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
H /D
s u
V/Dsu
w/D = 0.1, 0.2, 0.3, 0.4 and 0.5dashed lines - weightless soilsolid lines - weighty soil (γ'D/su = 0.8)
interface roughness factor, α = 0.5
Figure 6.18 Growth of yield envelopes due to introduction of soil self-weight (Case D and E)
Buoyancy factor
For pure vertical loading, the normalised vertical load V can be expressed as the sum of
components due to soil strength and self-weight as follows:
u
sbc
u DsA
fNDsV γ′
+= 6.6
where, Nc is the bearing capacity factor, As is the area of the soil displaced by the pipe.
fb is a buoyancy factor and is equal to 1 according to Archimedes’ principle, if the soil
surface is horizontal. For pipe motion in a direction other than pure vertical movement,
the resultant load F and the direction of pipe movement are not same. Instead they make
an angle ζ (= π - θ - tan-1(H/V)) as shown in Figure 6.19. To back-calculate fb for any
direction of pipe movement, the component of resultant forces in the direction of pipe
movement should be taken. Equation 6.6 can be rewritten for generalised direction of
movement in the following form:
u
sb0 Ds
Af)cosF(cosF γ′+ζ=ζ =γ
6.7
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-19
Where, ζcosF and ( ) 0cosF =γ′ζ are the components of F in the direction of pipe
movement for weighty (case E) and weightless (case D) soil cases. The value of fb
calculated this way can be confirmed by simple theoretical calculations as described in
the following section.
Direction of pipe movement
F
h
θ ζ
ψ
dψ
P Q
R S w
R
Figure 6.19 Schematic for calculating fb for any direction of pipe movement
A very small arc, ‘P-S’ in the periphery of the pipe in contact with the soil is considered
(Figure 6.19). This makes an angle ψ with the pipe radius. The arc itself makes a small
angle dψ in the centre of the pipe. When the pipe moves at an angle θ to the vertical, the
area of soil lifted with ‘P-S’ is denoted by an approximate rectangle ‘PQRS’, the area of
which is equal to R.dψ.δ.sin(θ - ψ). Here, δ is an incremental pipe movement in the
direction of pipe movement. The normalised depth of this portion from the mudline can
be calculated as:
Dw5.0sin5.0
Dh
−−ψ= 6.8
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-20
Therefore, the incremental work done (ΔW) for this small arc can be expressed as:
')Dw5.0sin5.0)(sin(RdW γ−−ψψ−θψδ=Δ
6.9
If this incremental work is integrated around the pipe periphery in contact with the soil
and divided by the work done by the component of F in the direction of pipe movement
against the weight of the soil displaced by the pipe, fb is obtained. The theoretically
calculated fb values are shown in solid lines and the buoyancy factors calculated using
OxLim results are shown by different symbols in Figure 6.20. As seen from this figure,
an excellent match is obtained between the theoretical calculation and the OxLim results.
0
0.2
0.4
0.6
0.8
1
1.2
0 30 60 90 120 150 180
Angle of pipe movement with vertical, θ
Buo
yanc
y fa
ctor
, f b
w/D = 0.1w/D = 0.2w/D = 0.3w/D = 0.4w/D = 0.5
Solid lines - theoretical calculationsMarkers - OxLim results
Figure 6.20 fb values for different directions of pipe movement
6.6 EFFECT OF SOIL HEAVE
The effect of soil heave around the pipe was also investigated. Initially, large
deformation finite element analysis using ABAQUS was performed to study the
breakout resistances of partially embedded pipelines to explore more realistic cases.
This analysis used the approach described in Chapters 2. The pipe was gradually pushed
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-21
into the soil and a berm was generated on each side. The pipe-soil interface was
assigned a roughness value of α = 0.5, in between the fully smooth and rough cases.
OxLim analyses were performed with the same soil conditions and using the berm
geometry extracted from the LDFE analyses after the initial penetration. The heave
geometries for w/D = 0.1 to 0.5 are shown in Figure 6.21.
(a) w/D = 0.1
(b) w/D = 0.2 (c) w/D = 0.3
(d) w/D = 0.4 (e) w/D = 0.5
Figure 6.21 Heave geometries for different embedments Two cases (Case F and Case G) without and with soil self-weight were considered. The
breakout resistance for pure horizontal motion for w/D = 0.5 was increased by 9.1% due
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-22
to the effect of soil self weight. For Case G, with heaved soil and γ'D/su = 0.8, the value
of fb for pure vertical loading was found to be in the range of 1.35 to 1.38 for uniform
soil. Yield envelopes for the berm soil case with soil weight are shown in the Figure
6.22.
0
0.5
1
1.5
2
2.5
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
H /D
s u
V/Dsu
w/D = 0.1, 0.2, 0.3, 0.4 and 0.5
solid lines - OxLim resultscircles - LDFE points
A
B
Figure 6.22 Comparison of yield envelopes from OxLim and LDFE analyses for heaved soil (Case G)
Results of Case D (no soil heave, weightless soil) and Case G (weighty soil with soil
heave) were compared to explore the effects of soil heave and soil weight together on
the breakout resistance. An increase of 10% (w/D = 0.1) – 18% (w/D = 0.5) in breakout
resistance for pure horizontal motion was observed in the case of heaved soil compared
to the weightless flat seabed. Case F (weightless soil with soil heave) and Case D
(weightless soil with no soil heave) were compared to see the effect of heave only. An
increase of 8.1% in the horizontal capacity for w/D = 0.5 for pure horizontal movement
was observed. The sum of the effect of heave (8.1%) and the effect of soil self weight
(9.1%) gives a total increase of 17.2%, which is close to the combined effect of 18%
increase. This indicates that superposition is an acceptable approach for assessing these
two effects.
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-23
Data points from LDFE analyses are also superimposed in the same figure and a
good match is observed. Displacement vectors revealing the failure mechanisms for
vertical penetration and pure horizontal movement for heaved soil case are shown in
Figure 6.23(a) and Figure 6.23(b). There are no fundamental differences in the failure
mechanism compared to the flat seabed case (as shown in Figure 6.17(a) and Figure
6.17(b)). The additional resistance is primarily due to the additional weight of soil to be
lifted, and the slight increase in the length of the failure planes that extend to the soil
surface.
(a) Vertical pipe movement (point B in Figure 6.22)
(b) Horizontal pipe movement (point A in Figure 6.22)
Figure 6.23 Adaptive mesh refinement for pure vertical and horizontal pipe movements
in heaved soil (w/D = 0.5)
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-24
6.7 CONCLUDING REMARKS In this chapter, the breakout resistance of shallowly embedded pipelines has been
addressed using finite element limit analysis methodology and verification has been
carried out using finite element software ABAQUS.
In the initial part of this study, a flat seabed and weightless soil was considered
and fully smooth and fully rough pipe soil interfaces were explored. The results were
compared with previously published solutions. In the case of a no-tension rough pipe-
soil interface, differences were highlighted between cases with an interface obeying an
associated flow rule, and a more natural one where the shear stress was limited to zero
where the pipe separates from the soil. In the latter case, although the associated flow
rule of classical plasticity theory is violated, the solution relates to a more logical
condition and leads to a different solution for a rough pipe soil interface. The OxLim
analysis was repeated using a modified interface behaviour, to mimic the more natural
breakaway condition. This led to results that are equivalent to published upper-bound
results in which breakaway is modelled. The resulting failure envelopes are definitive
solutions for the breakaway case. The OxLim approach revealed marginal
improvements in the collapse loads relative to analytical upper-bounds.
After exploring the weightless case in detail, the effect of soil self-weight was
investigated. It was shown that, for a flat seabed, a superposition approach is adequate
to capture the influence of buoyancy for a typical value of normalised soil self-weight.
This is because the soil failure mechanism is negligibly altered due to introduction of
self-weight.
The effect of soil heave around the pipe was also investigated. Initially, large
deformation finite element analyses were performed to simulate gradual penetration of
pipe and formation of heave around its shoulder. The seabed geometry with soil heave
was then extracted to perform upper-bound analyses in OxLim to study the breakout
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-25
resistance. It was confirmed that the buoyancy factor fb exceeds unity for purely
downward motion in the case of heaved soil surface. It was also concluded that the
presence of a soil berm around the pipe has a marked effect on the breakout resistance,
with an increase of up to 18% in capacity due to cumulative effect of soil weight and
berm observed. The results were compared with LDFE analyses and a good match was
observed.
6.8 REFERENCES Bruton, D. A. S., White, D. J., Carr, M. & Cheuk, C. Y. (2008). Pipe-soil interaction
during lateral buckling and pipeline walk – the SAFEBUCK JIP. Proc. Offshore
Technology Conf., Houston, Paper OTC 19589.
Chatterjee, S., Randolph, M. F. & White, D. J. (2012).The effects of penetration rate
and strain softening on the vertical penetration resistance of seabed pipelines.
Géotechnique 62, No. 7, 573-582.
Dassault Systemes (2007). Abaqus analysis users’ manual, Simulia Corp, Providence,
RI, USA.
Gourvenec, S. and Randolph, M. F. (2003). Effect of strength non-homogeneity on the
shape and failure envelopes for combined loading of strip and circular foundations on
clay. Géotechnique, 53, No. 6, 575-586.
Houlsby, G. T. & Puzrin, A. M. (1999). The bearing capacity of a strip footing on clay
under combined loading. Proc. R. Soc. Lond. A, 455, 893-916.
Martin, C. M. (2009). Undrained collapse of a shallow plane-strain trapdoor.
Géotechnique 59, No. 10, 855-863.
Martin, C. M. (2011). The use of adaptive finite-element limit analysis to reveal slip-
line fields. Géotechnique Letters, pp. 1-7.
CHAPTER 6: Breakout behaviour…
Centre for Offshore Foundation Systems 6-26
Merifield, R. S., White, D. J. & Randolph, M. F. (2009). Effect of surface heave on
response of partially embedded pipelines on clay. J. Geotech. Geoenviron. Engng.,
ASCE 135, No 6, 819-829.
Makrodimopoulos, A. & Martin, C. M. (2006). Lower bound limit analysis of cohesive
frictional materials using second-order cone programming. Int. Journal for Numerical
Methods in Engineering 66, No. 4, 604-634.
Makrodimopoulos, A. & Martin, C. M. (2007). Upper bound limit analysis using
simplex strain elements and second-order cone programming. Int. Journal for
Numerical and Analytical Methods in Geomechanics 31, No. 6, 835-865.
Makrodimopoulos, A. & Martin, C. M. (2008). Upper bound limit analysis using
discontinuous quadratic displacement fields. Communications in Numerical Methods in
Engineering 24, No. 11, 911-927.
Randolph, M. F. & White, D. J. (2008). Upper-bound yield envelopes for pipelines at
shallow embedment in clay. Géotechnique 58, No. 4, 297-301.
Numerical modelling of pipe-soil interactions
Centre for Offshore Foundation Systems 7-1
CHAPTER 7
ELASTOPLASTIC CONSOLIDATION BENEATH
SHALLOWLY EMBEDDED OFFSHORE PIPELINES
7.1 INTRODUCTION
In the fine grained, low permeability soils typically encountered in deep-water,
undrained conditions prevail during laying, resulting in excess pore pressures and
relatively low effective stresses in the vicinity of the pipe-soil interface. Estimation of
the axial friction available between pipeline and soil, which is required for design
calculations such as expansion and contraction of the pipeline due to temperature
changes, must therefore consider the time-scale of consolidation beneath the pipeline.
Pore pressure dissipation under shallowly embedded pipes has been addressed by
Gourvenec & White (2010) and Krost et al. (2011), but those results were limited to
elastic response of the seabed, and with a uniform coefficient of consolidation, cv. This
chapter extends those results to a more realistic elasto-plastic response of the soil.
Coupled consolidation finite element analyses have been undertaken using the
Modified Cam Clay soil model. Results from large deformation finite element (LDFE)
analyses are compared with the previous findings for elastic soil. Extremes of pipe
roughness (fully smooth or fully rough) and cv profiles (either uniform or increasing
proportionally with depth according to the effective stress level) are considered.
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-2
7.2 NUMERICAL METHODOLOGY
The large deformation finite element methodology developed for total stress analysis as
described in previous chapters was extended to coupled effective stress analysis for this
chapter.
The problem was solved as a two-dimensional plane strain problem with elasto-
plastic soil response and a rigid pipe, as shown in Figure 7.1. A sequence of small strain
analyses, with the pipe advanced by 1 % of its diameter at each step, were combined
with remeshing and interpolation of stress and material properties to penetrate the pipe
to the target embedment level.
Impe
rmea
ble
Impe
rmea
ble
Impermeable pipe-soil interface
Rigid pipe
Permeable surface Permeable surface
Impermeable
D
Uniform surcharge = 200 kPa Uniform surcharge = 200 kPa
Impe
rmea
ble
Impe
rmea
ble
Impermeable pipe-soil interface
Rigid pipe
Permeable surface Permeable surface
Impermeable
D
Uniform surcharge = 200 kPa Uniform surcharge = 200 kPa
Impe
rmea
ble
Impermeable pipe-soil interface
Rigid pipe
Permeable surface Permeable surface
Impermeable
D
Uniform surcharge = 200 kPa Uniform surcharge = 200 kPa
Figure 7.1 Schematic diagram of the problem solved
A graded mesh of second order (coupled consolidation-stress) triangular
elements, type CPE6MP in ABAQUS, was used with the smallest element of side equal
to 2 % of the pipe diameter. No drainage was allowed during the penetration process.
After the pipe had reached the target embedment, drainage was allowed at the top
surface to dissipate pore pressures generated during penetration. As for the small strain
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-3
analyses, extremes of a perfectly smooth and fully rough pipe-soil interface were
modelled.
The initial consolidation time increment, Δtinitial, was chosen according to the
criterion (Vermeer & Verruijt, 1981):
k'E6ht w2
initialγ
=Δ 7.1
where h is a typical element dimension, γw is the unit weight of water, E' is the effective
Young’s modulus of the soil (calculated here using the MCC reloading stiffness,
inversely proportional to κ) and k is the permeability of the soil.
7.3 MATERIAL MODEL
The soil response was modelled using Modified Cam Clay (MCC, Roscoe & Burland,
1968), as implemented in ABAQUS. The soil is defined as a porous elastic material
before yielding. All parameters used for the numerical analyses are listed in Table 7.1.
Table 7.1 Input parameters for numerical study Parameter Value Slope of CSL in p'-q space, Μ, (friction angle in triaxial compression, φ'tc)
0.92 (23.5°)
Void ratio at p' = 1 kPa on CSL, ecs 2.14 Slope of NCL in e-ln(p') space, λ 0.205 Slope of swelling and recompression line in e-ln(p´) space, κ 0.044 Poisson’s ratio, ν (LDFE) 0.3 Saturated bulk unit weight, γsat 15.0 kN/m3 Unit weight of water, γw 10.0 kN/m3 Permeability of soil, k 1.0e-9 m/s Pipe diameter, D 0.5 m
A limitation of the MCC model is that the soil stiffness, and hence the coefficient of
consolidation, cv, varies in proportion to the mean effective stress. In order to
investigate how the time-scale for consolidation varies with the distribution of cv,
separate series of analyses were undertaken: (a) with an artificial surcharge of 200 kPa
applied at the soil surface (including beneath the pipe) (Figure 7.1), giving an
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-4
approximately uniform value of cv within the soil domain; and (b) with a very small
surcharge (1 kPa – the smallest to allow numerical stability), giving cv approximately
proportional to depth. Comparison of the two series allowed assessment of an
equivalent cv for the latter case in order to obtain similar consolidation timescale for the
homogeneous case. For the large deformation analyses, heave grows on either side of
the pipe. The curved soil surface is modelled as a series of small straight lines and the
surcharge is maintained throughout the analysis by applying that pressure on each
straight line.
In all analyses the soil was considered to be K0 (normally) consolidated, with K0
given by
tc0 sin1K φ−= ~ 0.6 7.2
where φtc is the friction angle for triaxial compression conditions. In situ effective
stresses and pore pressure varied according to the respective self-weights (see Table
7.1).
The initial size of the yield envelope is determined by p'c, expressed as
tc
tc0
02
20
c sin3sin6
M withppM
qp
φ−φ
=′+′
=′ 7.3
where '0p and 0q are the initial effective mean stress and deviatoric stress. Initial void
ratio, e0, is calculated from
c010 pln)(plnee ′κ−λ−′κ−= 7.4
where
)2ln()(ee cs1 κ−λ+= 7.5
and κ and λ are the usual swelling and compression indices in MCC.
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-5
For these initial conditions, the starting point of the analysis for a given depth is denoted
by ‘O’ in p' - q and e – ln p' spaces, as shown in Figure 7.2.
q
p'
M1
Critical state line
C
O
BA
pa' pb' po' pc'
ln p'
e
λ1
A
BO
C
Critical state line
Normal compression line
κ1
q
p'
M1
Critical state line
C
O
BA
pa' pb' po' pc'
ln p'
e
λ1
A
BO
C
Critical state line
Normal compression line
κ1
Figure 7.2 Yield envelope and critical state line for MCC model
The stress path to reach critical state during undrained penetration is denoted by OB.
For triaxial compression conditions, the undrained shear strength ratio, su/σ'v, for K0
consolidated soil can be calculated from the MCC parameters using (Wroth, 1984):
Λ
⎟⎟⎠
⎞⎜⎜⎝
⎛ +φ=
σ 21a
a2sins 2
tc'v
u 7.6
For plane strain conditions the undrained shear strength ratio can be expressed as,
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-6
Λ
⎟⎟⎠
⎞⎜⎜⎝
⎛ +φ=
σ 21a
a2sin
32s 2
tc'v
u 7.7
where
)sin23(2sin3
atc
tc
φ−φ−
= and λ
κ−λ=Λ 7.8
This leads to (su/σ'v0)nc for plane strain conditions of 0.29 and mudline strengths of
0.29 kPa and 57.2 kPa for the 1 kPa and 200 kPa surcharge cases.
7.4 UNDRAINED PENETRATION RESPONSE
Normalised undrained penetration responses are shown in Figure 7.3.
0
0.1
0.2
0.3
0.4
0.5
0 1 2 3 4 5 6V/Dsu0
w/D
MCC Model -Smooth
Tresca Mode - Smooth
MCC Model - Rough
Tresca Model - Rough
1 kPa surcharge, smooth
1 kPa surcharge, rough
Figure 7.3 Comparison of penetration responses for smooth and rough pipes
The majority of the results, unless otherwise stated, are for uniform strength soil with a
surcharge of 200 kPa at the soil surface, although comparative LDFE results are also
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-7
shown for the nominal surcharge of 1 kPa. The geotechnical penetration resistance, V,
has been calculated by subtracting the buoyancy effect, assessed following Chatterjee et
al. (2012). V is normalised by the pipe diameter, D, and the undrained shear strength,
su0, at the invert level, calculated from the in situ profile based on Equation 7.7.
Comparative results of LDFE analyses using a simple Tresca soil model are also shown,
giving close agreement.
Comparison with the rigid plastic limit analyses of Randolph & White (2008)
shows that the partial mobilisation effect applies to the LDFE analyses at shallow
penetration, for the soil stiffnesses used here. The responses in Figure 7.3 lie slightly
below the equivalent rigid plastic solutions for w/D less than 0.2 (smooth) to 0.3
(rough). However, the relative trends agree, so the 1 kPa surcharge cases (increasing
strength with depth) show a higher normalised penetration resistance at shallow
embedment compared to uniform soil, reversing at greater embedment. The latter effect
is enhanced by dragdown of weaker sediments from the surface – an effect neglected in
the limit analysis solutions.
7.5 CONSOLIDATION RESPONSE
7.5.1 Pore pressure dissipation After penetration, the final resistance was maintained as a constant load while
consolidation was permitted. Contours of initial excess pore pressure normalised by the
value at the pipe invert for w/D = 0.5 are shown in Figure 7.4, and the corresponding
variation around the pipe periphery is shown in Figure 7.5. For the 1 kPa surcharge case,
the excess pore pressure is concentrated at the pipe invert, reflecting the increasing soil
strength with depth, and therefore the concentration of load at the pipe invert. In
uniform soil conditions the excess pore pressure is almost uniform (+/-15%) across
most of the pipe (-0.4 < x/D < 0.4).
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-8
Figure 7.4 Contours of excess pore water pressure after penetration (w/D = 0.5)
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-9
0
0.2
0.4
0.6
0.8
1
1.2
0 0.1 0.2 0.3 0.4 0.5 0.6
x/D
Δu/
Δu i
nv
Pipe embedment, w/D = 0.5Results shown for both sides of pipe (x/D positive and negative)
Rough, 200 kPa
Smooth, 200 kPa
Rough, 1 kPa
Smooth, 1 kPa
Figure 7.5 Excess pore pressure distribution around pipe periphery after penetration (w/D = 0.5)
The differences in the shape of the initial excess pore pressure field contribute to
differences in consolidation rate. Consolidation time is normalised as T = cvt/D2, where
cv is expressed in terms of the permeability, k, and (plastic) isotropic compressibility,
mv, as
( )w
0
wvv
pe1km
kcλγ
′+=
γ= 7.9
For simplicity, the initial value of cv has been used for the normalisation.
Figure 7.6 and Figure 7.7 show variations of excess pore water pressure at the
pipe invert, normalised by the initial value, with non-dimensional time T.
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-10
0
0.2
0.4
0.6
0.8
1
1.2
0.00001 0.0001 0.001 0.01 0.1 1 10 100T = cvt/D
2
Δu/
Δu
iw/D = 0.1w/D = 0.2w/D = 0.3w/D = 0.4w/D = 0.5
w/D = 0.1, 1 kPa surcharge
w/D = 0.5, 1 kPa surcharge
Markers: LDFE resultsLines: Equation 10
Figure 7.6 Excess pore pressure dissipation time history at pipe invert for smooth pipe
0
0.2
0.4
0.6
0.8
1
1.2
1E-05 0.0001 0.001 0.01 0.1 1 10 100T = cvt/D
2
Δu/
Δu i
w/D = 0.1w/D = 0.2w/D = 0.3w/D = 0.4w/D = 0.5
w/D = 0.1, 1 kPa surcharge
w/D = 0.5, 1 kPa surcharge
Markers: LDFE resultsLines: Equation 10
Figure 7.7 Excess pore pressure dissipation time history at pipe invert for rough pipe
Simple hyperbolic equations (solid lines) have been fitted to the FE data (symbols),
according to
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-11
m50i )T/T(1
1uu
+=
ΔΔ 7.10
where T50 is the value of T for 50% dissipation and m is a constant. The values of T50
and m for different embedment levels are tabulated in Table 7.2.
Table 7.2 Values of T50 and constant ‘m’ of hyperbolic fits
Smooth pipe Rough pipe Initial embedment ratio, w/D
T50 m T50
m
0.1 0.022 0.028 0.2 0.040 0.055 0.3 0.060 0.075 0.4 0.081 0.110 0.5 0.096
1.05
0.135
1.05
It can be seen from Figure 7.7 that, for a rough pipe, there is an initial increase in excess
pore pressure for all embedments. This is due to the Mandel-Cryer effect (stress transfer
phenomenon in which the dissipation process creates a local rise in total stress, resulting
in an increase rather than a decrease in excess pore pressure) (Mandel, 1950; Cryer,
1963 and as discussed for this problem by Gourvenec & White, 2010). In contrast to the
elastic consolidation results of Krost et al. (2011), the effect is evident for each
embedment, although in the LDFE analyses is more prominent for shallower
embedments. Due to the Mandel-Cryer effect, the hyperbolic fit does not capture the
initial portion of the dissipation response.
The majority of the results in Figure 7.6 and Figure 7.7 are for the uniform cv
case, but two example responses for the LDFE analyses using 1 kPa surcharge (cv
approximately proportional to depth) are shown. For those cases, the cv value that gives
a good match to the uniform cases at T50 and during the latter part of the consolidation
curve is higher than the invert value. The ratio, χ, of the operative value, cv,operative to the
invert value is given in Table 7.3.
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-12
Table 7.3 Operative cv for different initial embedment values
χ = cv,operative/cv,invert Depth of operative cv
(normalised by pipe diameter) Initial
embedment ratio, w/D Smooth pipe Rough pipe Smooth pipe Rough pipe
0.1 1.19 1.19 0.20 0.20 0.2 1.54 1.54 0.55 0.55 0.3 1.53 1.78 0.70 0.90 0.4 1.69 1.91 1.00 1.20 0.5 1.61 2.00 1.10 1.50
The increase in cv by 1.2 - 2 times indicates more rapid dissipation than if the
entire soil domain had the same cv value as at the pipe invert. This can be linked to (i)
the higher cv within the consolidating soil beneath the pipe invert and (ii) the different
initial pore pressure field (Figure 7.4 and Figure 7.5). An alternative interpretation of
this effect is to consider the depth at which this operative cv is found, which is also
given in Table 7.3.
The decay in the average excess pore pressure around the pipe periphery, Uav
(= Δuav/Δuav,init), and the corresponding rise in average normal effective stress, Σ, are
useful quantities since they indicate the build-up of potential axial resistance between
the pipe and the seabed. Σ is defined as:
( )( )'
init,av,n'
f,av,n
'init,av,n
'av,n
σ−σ
σ−σ=∑ 7.11
where σ'n,av denotes the average normal effective stress around the pipe periphery and
σ'n,av,init and σ'n,av,f are the values before and after dissipation. These trends are shown in
Figure 7.8 for a smooth pipe. The pore pressure and inverted effective stress responses
agree to within 4% throughout the decay process, indicating that changes in total stress
– which can occur due to the wedging effect, even if the applied pipe weight is constant
– are small. The MCC results are similar to the elastic results, but show more rapid
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-13
dissipation as consolidation progresses compared to the elastic solution, reflecting the
increasing stiffness as the effective stress rises.
0
0.2
0.4
0.6
0.8
1
0.00001 0.0001 0.001 0.01 0.1 1 10 100
T = cvt/D2
Uav
or (
1-Σ
)w/D = 0.1w/D = 0.2w/D = 0.3w/D = 0.4w/D = 0.5
MCC analyses:Lines: Effective stress, (1 - Σ)Markers: Pore pressure, Uav
Elastic analysis, w/D = 0.1(Gourvenec & White, 2010)
Elastic analysis, w/D = 0.5(Gourvenec & White, 2010)
Figure 7.8 Average pore pressure dissipation and rise in effective stress along the pipe periphery
7.5.2 Consolidation settlement Consolidation settlement, wc normalised by the diameter of the pipe is plotted against
non-dimensional time T in Figure 7.9 (smooth pipe) and Figure 7.10 (rough pipe). The
settlement increases with increasing initial embedment, consistent with the increase in
load applied during consolidation. Apart from the shallowest embedment, where the
pipe-soil interface condition makes little difference, the rough interface leads to a slight
delay in the onset of consolidation settlement, and T50 values that are up to twice as high.
Since the pipe is subjected to the full vertical bearing capacity throughout consolidation,
these settlements are greater than will typically occur in practice. The duration of the
consolidation settlement is substantially less than the equivalent elastic results – scaled
to give the same total consolidation settlement (Figure 7.9, Krost et al., 2011).
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-14
0
0.05
0.1
0.15
0.2
0.25
1E-05 1E-04 0.001 0.01 0.1 1 10 100 1000T = cvt/D
2w
c/D
w = 0.1Dw = 0.2Dw = 0.3Dw = 0.4Dw = 0.5D
V/Dsu0=2.13
V/Dsu0=3.29
V/Dsu0=3.84
V/Dsu0=4.13V/Dsu0=4.35
Elastic solutionKrost et al. (2011)
Figure 7.9 Time settlement response for different initial embedments (smooth pipe)
0
0.05
0.1
0.15
0.2
0.25
1E-05 1E-04 0.001 0.01 0.1 1 10 100 1000T = cvt/D
2
wc/D
w = 0.1Dw = 0.2Dw = 0.3Dw = 0.4Dw = 0.5D
V/Dsu0 = 2.16
V/Dsu0 = 3.52
V/Dsu0 = 4.48
V/Dsu0 = 5.26
V/Dsu0 = 5.91
Figure 7.10 Time settlement response for different initial embedments (rough pipe)
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-15
7.6 CONCLUDING REMARKS The consolidation process after partial embedment of pipelines is an important
consideration for design, as it governs the rate at which axial friction develops. The
effects on consolidation of embedment, pipe-soil interface condition and the large
deformations associated with the penetration have been investigated. In an advance over
previously-published elastic solutions, the more realistic Modified Cam Clay plasticity
soil model was used.
For both smooth and rough pipe-soil interfaces, consolidation time increased with
increasing initial embedment, and was greater for the rough case. An initial increase in
excess pore water pressure was observed at the invert for rough pipes due to the
Mandel-Cryer effect. Simple hyperbolic equations were fitted to the dissipation curves
at the pipe invert, with parameters tabulated for ease of use in design. Also, the
averaged pore pressure and effective stress quantities around the pipe-soil contact were
assessed, since these control the available friction.
The dissipation responses were compared to those from elastic solutions,
highlighting the effects of different initial excess pore pressure distributions and some
stiffness increase during consolidation arising from the Modified Cam Clay model.
Comparison of results between small strain and large deformation analyses showed the
effect of the soil berms on the consolidation behaviour.
7.7 REFERENCES Chatterjee, S., Randolph, M. F. & White, D. J. (2012). The effects of penetration rate
and strain softening on the vertical penetration resistance of seabed pipelines.
Géotechnique 62, in press, doi: 10.1680/geot.10.P.075.
Cryer, C. W. (1963). A comparison of the three dimensional consolidation theories of
Biot and Terzaghi. Q. J. Mech. Appl. Math. 16, No. 4, 401–412.
CHAPTER 7: Elastoplastic consolidation…
Centre for Offshore Foundation Systems 7-16
Dassault Systèmes. (2010). Abaqus Analysis Users' Manual. Simula Corp, Providence,
RI, USA.
Ghosh, S. & Kikuchi, N. (1991). An arbitrary Lagrangian-Eulerian finite element
method for large deformation analysis of elastic-viscoplastic solids. Comput.
Methods Appl. Mech. Eng. 86, No. 2, 127-188.
Gourvenec, S. M. & White, D. J. (2010). Elastic solutions for consolidation around
seabed pipelines. Proc. Offshore Technology Conf., Houston, Texas, USA, Paper
OTC 20554.
Hu, Y. & Randolph, M. F. (1998). A practical numerical approach for large deformation
problems in soil. Int. J. Numer. Analyt. Meth. Geomech. 22, No. 5, 327-350.
Krost, K., Gourvenec, S. M. & White, D. J. (2011). Consolidation around partially
embedded seabed pipelines. Géotechnique 61, No. 2, 167-173.
Mandel, J. (1950). Étude mathématique de la consolidation des sols. Actes du Colloque
International de Mécanique, Poitier, France, 4, 9–19.
Randolph, M. F. & White, D. J. (2008). Upper bound yield envelopes for pipelines at
shallow embedment in clay. Géotechnique, 58, No. 4, 297-301
Roscoe, K. H. & Burland, J. B. (1968). On the generalised stress-strain behaviour of
'wet clay'. Engineering Plasticity, Cambridge University Press, 535-609.
Vermeer, P. A. & Verruijt, A. (1981). An accuracy condition for consolidation by finite
elements. Int. J. Numer. Analyt. Meth. Geomech. 1, 1–14.
Wang, D, White, D. J. & Randolph, M. F. (2010). Large deformation finite element
analysis of pipe penetration and large-amplitude lateral displacement. Can. Geotech.
J. 47, No. 8, 842-856.
Wroth, C. P. (1984). The interpretation of in situ soil tests. Géotechnique 34, No. 4,
449-489.
Numerical modelling of pipe-soil interactions
Centre for Offshore Foundation Systems 8-1
CHAPTER 8
EFFECTS OF CONSOLIDATION ON PENETRATION
AND LATERAL BREAKOUT RESISTANCES
8.1 INTRODUCTION
To design a pipeline reliably for lateral buckling, it is necessary to predict the lateral and
axial pipe-soil resistance forces, which both depend on the pipe embedment and the
strength of the surrounding soil. In deep water, the seabed typically comprises soft fine-
grained sediments, which can consolidate and change in strength over the time periods
relevant to the operating life of a pipeline. In this chapter, the effects of consolidation on
two key aspects of deep water pipeline design are studied: firstly, the effect of
consolidation on pipeline embedment; and secondly the effect of consolidation on the
lateral breakout behaviour.
The current design practice (White & Cheuk 2005, 2009; AtkinsBoreas 2008)
assumes that undrained conditions apply throughout the lay process and subsequent
consolidation settlements are generally neglected. Also, it is usually assumed that the
soil strength during pipe breakout is unaffected by consolidation under the pipe weight
during the period between laying and breakout. The effect of these assumptions is
investigated in this chapter.
A large deformation finite element methodology combined with the Modified Cam
Clay plasticity soil constitutive model was developed as described in the previous
chapter to study the pore pressure dissipation beneath partially embedded seabed
pipelines. In this chapter, the same methodology has been used to study the effect of
consolidation on penetration behaviour and subsequent lateral breakout resistance.
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-2
8.2 MODEL DESCRIPTION
The same model as described in the previous chapter is used here. A two-dimensional
plane strain model was constructed with the pipe as a rigid body and the soil as
deformable. The extent of the model was 10 times the pipe diameter in the vertical
direction and 8 times the pipe diameter in the horizontal direction on both sides of the
pipe. The side boundaries of the model were free to move in the vertical direction, but
restrained against horizontal movement. The bottom boundary was fixed, preventing
vertical and horizontal movement. Drainage was allowed only at the top soil surface and
the pipe-soil interface was taken as impermeable. A schematic diagram of the problem
studied and main notation used are shown in Figure 8.1.
10D
Impermeable pipe-soil interface (smooth / rough)
Permeable Permeable
D
Uniform surcharge = 200 kPa Uniform surcharge = 200 kPa
w
HV
Rol
ler /
Impe
rmea
ble
Rol
ler /
Impe
rmea
ble
Hinge / Impermeable
16D
Pipe
Soil10D
Impermeable pipe-soil interface (smooth / rough)
Permeable Permeable
D
Uniform surcharge = 200 kPa Uniform surcharge = 200 kPa
w
HV
Rol
ler /
Impe
rmea
ble
Rol
ler /
Impe
rmea
ble
Hinge / Impermeable
16D
Pipe
Soil
Impermeable pipe-soil interface (smooth / rough)
Permeable Permeable
D
Uniform surcharge = 200 kPa Uniform surcharge = 200 kPa
w
HV
Rol
ler /
Impe
rmea
ble
Rol
ler /
Impe
rmea
ble
Hinge / Impermeable
16D
Pipe
Soil
Figure 8.1 Schematic of the problem studied
A very small displacement of 1% of the pipe diameter, D, was applied at the
pipe reference point in each step. Six-noded triangular plane strain elements of type
CPE6MP within the ABAQUS library were used for discretization of the soil domain. A
fine mesh with minimum side length of the triangular elements of 0.02D was adopted
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-3
near the pipe. The extent of the finest meshing from the centre of the pipe at the start of
the analysis was up to 1.25D on both sides and below from the mudline. Figure 8.2
shows the finite element mesh before any pipe displacement and after the pipe had been
penetrated into the soil by half its diameter.
(a) Initial Mesh
(b) Mesh after pipe penetration
Figure 8.2 Finite element meshes before and after pipe penetration
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-4
8.3 SOIL PARAMETERS
The modified Cam Clay (MCC) soil constitutive model (Roscoe & Burland, 1968;
Schofield & Wroth, 1968) was adopted and the numerical parameters used for all the
analyses are same to that of the previous chapter. A uniform pressure of 200 kPa was
applied at the top soil surface. This alleviates numerical problems associated with the
very low shear strength of the soil at the mudline when using the Cam clay soil model
and normally consolidated conditions.
As well as providing numerical stability, this surcharging technique minimises the
variation in soil properties with depth. This makes normalisation of the results more
straightforward, since properties such as the coefficient of consolidation and the initial
undrained shear strength are essentially invariant with depth. The strength heterogeneity
affects the vertical and horizontal soil resistance. Adoption of an artificial surcharge of
200 kPa restricts the study to strength heterogeneity of close to zero. However, the main
focus here is to evaluate the general trends of response observed as a result of partial
consolidation, compared with corresponding results for purely undrained conditions.
The interface between the pipe and the soil was assumed to be fully smooth
(mobilising zero shear stress during tangential movement) or fully rough (with adjacent
pipe and soil nodes being tied). The pore water pressure distribution was initially
hydrostatic.
8.4 EFFECTS OF LOADING RATE
8.4.1 Penetration resistance
Fully coupled consolidation stress analyses following the RITSS approach were
performed. The pipe was penetrated to an embedment of 50% of its diameter with
different velocities, v, and thus different values of the non-dimensional velocity vD/cv.
The consolidation coefficient, cv, can be determined from
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-5
wvv m
kcγ
= 8.1
Where, mv is the volume compressibility, wγ is the unit weight of water and k is the
permeability.
A wide range of values of vD/cv from a very high velocity (vD/cv = 100) down
to the lowest velocity corresponding to vD/cv = 0.025 were considered. The highest
velocity means that negligible excess pore water pressure can dissipate and undrained
conditions are approached. In contrast, slower velocities of the pipe lead to a partially
drained and ultimately a fully drained response. Figure 8.3 shows the variation of
normalised penetration resistance with depth for a smooth pipe-soil interface, and
Figure 8.4 shows the same results for a rough pipe-soil interface.
0
0.1
0.2
0.3
0.4
0.5
0 2 4 6 8V/Dsu0
w/D
vD/Cv=0.025vD/Cv=0.1vD/Cv=1vD/Cv=10vD/Cv=30Dv/Cv=60vD/Cv=100Tresca Model
Figure 8.3 Normalised penetration resistances with embedment for different pipe velocities (smooth pipe)
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-6
0
0.1
0.2
0.3
0.4
0.5
0 2 4 6 8 10V/Dsu0
w/D
vD/Cv=0.05
vD/Cv=0.1
vD/Cv=1
vD/Cv=10
vD/Cv=30
vD/Cv=100
Tresca Model
Figure 8.4 Normalised penetration resistances with embedment for different pipe velocities (rough pipe)
The resistance force is normalised using the undrained shear strength, su0 at the pipe
invert obtained from the modified Cam Clay parameters for K0-consolidated soil. The
initial undrained shear strength (Wroth, 1984) obtained was 57.2 kPa at the mudline and
64.4 kPa at the bottom of the mesh.
At the two highest penetration rates, the resistance profiles are similar,
suggesting that fully undrained conditions are almost reached. This is confirmed by the
results of an analysis performed using the same numerical technique but with the Tresca
soil model and the equivalent undrained shear strength. These results are also shown in
Figure 8.3 and Figure 8.4. Excellent agreement is evident, indicating that the fastest
cases correspond to practically undrained conditions.
At lower pipe velocities, the penetration resistances are higher (Figure 8.3).
Numerical convergence problems were observed for a very low penetration rate
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-7
simulating fully drained conditions. However, there are negligible differences in the
resistance responses for vD/cv = 0.1 and vD/cv = 0.025 or 0.05. This indicates that pipe
velocities corresponding to vD/cv = 0.1 or lower lead to essentially fully drained
resistance, even though small excess pore pressures are still present. The contours of
excess pore water pressure normalised by the resistance experienced by the pipe at w/D
= 0.5 for the highest and the lowest penetration rates (nominally undrained and drained
cases) for smooth and rough pipes are shown in Figure 8.5. For the lowest pipe
penetration rate, the excess pore pressure generated beneath the pipe is much less
compared to that for the undrained cases.
The ratio between the drained and undrained penetration resistance increases
with pipe embedment. Compared to the smooth case, more resistance is observed at a
particular embedment level for the rough pipe-soil interface. However, the relative
increase in resistance from the fully undrained to the fully drained case is lower in the
case of rough pipe. At the embedment level of w/D = 0.5, for the smooth pipe, a 72%
increase in resistance is observed from the fully undrained condition to the fully drained
condition, whereas a difference of approximately 48% is found for the rough pipe.
To illustrate the transition between drained and undrained conditions, the
resistance for a particular pipe velocity (V) is normalised by the undrained resistance
(Vundrained) at that depth and plotted against the non-dimensional velocity. Figure 8.6 and
Figure 8.7 show the resulting ‘backbone’ curves at different embedment levels for
smooth and rough pipes respectively. For vD/cv ≤ 0.1, the resistance is independent of
velocity and the response is essentially drained. For vD/cv ≥ 100, the response stabilises
and the response is fully undrained. The non-dimensional velocities in between these
limits correspond to partially drained behaviour. The backbone curves for smooth and
rough pipes can be fitted to a simple hyperbolic equation of the form:
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-8
Figure 8.5 Contours of excess pore water pressure normalised by penetration resistance
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-9
[ ]c50vvundrained )c/vD/()c/vD(1ba
VV
++=
8.2
where (vD/cv)50 is the normalised penetration rate that gives a response midway
between the drained and undrained limits. For higher values of vD/cv, i.e. for undrained
cases, V/Vundrained tends to unity, so the value of parameter ‘a’ is always 1. The
parameter ‘b’ controls the drained limit of the backbone curve as vD/cv → 0. The
quantities ‘c’ and (vD/cv)50 were varied to obtain best-fit curves for all embedment
levels for smooth as well as rough pipes. It was found that values of c = 1 and (vD/cv)50
= 2 gave reasonably fitting curves for all initial embedment and smooth and rough pipes.
The parameter ‘b’ depends on the initial embedment level and can be expressed as a
power law function of the embedment depth. For a smooth pipe
b ~ 1.45(w/D) 8.3
And for a rough pipe:
b ~ 0.92(w/D)0.9 8.4
For comparison, a number of analyses were run for a low surcharge of 1 kPa at
the top surface for the smooth pipe. For the case of 1 kPa surcharge, the value of cv
varies considerably with depth. Hence, while calculating non-dimensional velocity
vD/cv, cv was chosen at depth of 1D. Figure 8.8 shows the backbone curves for different
embedments for 1 kPa surcharge. The V/Vundrained ratios are generally higher at the
drained end for this case. However, for embedment levels of w/D = 0.2 or more, the
backbone curves are closer to each other compared to the 200 kPa case. This indicates
that for 1 kPa surcharge, V/Vundrained ratio increases less as the pipe penetrates deeper.
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-10
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
0.01 0.1 1 10 100 1000
vD/cv
V/V
undr
aine
d LDFE result, w/D=0.1LDFE result, w/D=0.2LDFE result, w/D=0.3LDFE result, w/D=0.4LDFE result, w/D=0.5
Figure 8.6 Backbone curves for different initial embedment levels (smooth pipe)
1
1.1
1.2
1.3
1.4
1.5
0.01 0.1 1 10 100 1000
vD/cv
V/V
undr
aine
d
LDFE result, w/D=0.1LDFE result, w/D=0.2LDFE result, w/D=0.3LDFE result, w/D=0.4LDFE result, w/D=0.5
Figure 8.7 Backbone curves for different initial embedment levels (Rough pipe)
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-11
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
0.01 0.1 1 10 100 1000vD/cv
V/V
undr
aine
d
w/D = 0.1w/D = 0.2w/D = 0.3w/D = 0.4w/D = 0.5
Figure 8.8 Backbone curves for different initial embedment levels (smooth pipe, 1 kPa surcharge)
8.4.2 Consolidation settlement
At different levels of vertical embedment, i.e. at w/D = 0.1, 0.2, 0.3, 0.4 and 0.5, the
consolidation settlement behaviour was also studied. The excess pore water pressure
generated during the pipe penetration was allowed to dissipate under the full penetration
resistance load experienced at the respective embedment level. This simple case may
not represent practical conditions. In general, the maximum vertical pipe-seabed load
during pipe laying is higher than the pipe weight alone, so the applied vertical load
during consolidation is less than the full bearing capacity.
The settlement (Δw) variations for different embedment levels are shown in
Figure 8.9, as a function of non-dimensional time factor T (= cvt/D2), for initial
penetrations at speeds of vD/cv = 0.1 (drained), and 100 (undrained). For the (nominally)
drained penetration case, the pore water pressure is partially dissipated during
penetration and hence the subsequent consolidation settlement should be less than for
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-12
0
0.05
0.1
0.15
0.2
0.25
0.3
0.0001 0.001 0.01 0.1 1 10 100 1000T = cvt/D
2Δ
w/D
vD/Cv = 0.1
vD/Cv = 100
Curves in orderw/D = 0.1, 0.2, 0.3, 0.4 & 0.5
Curves in orderw/D = 0.5, 0.4, 0.3, 0.2 & 0.1
(a) Smooth interface
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.0001 0.001 0.01 0.1 1 10 100 1000T = cvt/D
2
Δw
/D
vD/Cv = 0.1vD/Cv = 100
Curves in orderw/D = 0.1, 0.2, 0.3, 0.4 & 0.5
Curves in orderw/D = 0.5, 0.4, 0.3, 0.2 & 0.1
(b) Rough interface
Figure 8.9 Pipe settlements with time for different initial embedments and pipe velocities
the undrained case. However, this is more than compensated for by the greater
resistance experienced during drained penetration, and thus higher loads applied during
the consolidation phase. If the same load were to be applied during consolidation, the
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-13
consolidation settlement would indeed be much less following drained penetration
compared to the undrained case. This phenomenon is illustrated in Figure 8.10 for the
smooth pipe, where for the drained case the loads were reduced (following penetration,
prior to consolidation) to the same as for the corresponding undrained case.
-0.05
0
0.05
0.1
0.15
0.2
0.25
0.0001 0.001 0.01 0.1 1 10 100 1000T = cvt/D
2
Δw
/D
vD/Cv = 0.1
vD/Cv = 100
Curves in orderw/D = 0.1, 0.2, 0.3, 0.4 & 0.5
Curves in orderw/D = 0.1, 0.2, 0.3, 0.4 & 0.5
Figure 8.10 Consolidation settlements following penetration at different speeds under the same consolidation load (smooth pipe)
It may be seen that the overall time-scale of consolidation is the same, regardless
of the degree of consolidation during initial penetration. The time-scale for
consolidation settlement is essentially dictated by the far-field pore pressures, and is
nearly two orders of magnitude greater than for pore pressure dissipation adjacent to the
pipe (Chatterjee et al. 2012).
8.4.3 Pore pressure dissipation after undrained penetration The axial and lateral resistance of the pipeline are affected significantly by the degree of
consolidation following installation. This may be characterised by the average excess
pore pressure around the embedded pipe perimeter (Δuav), normalised by the initial
average value (Δuav,init), as plotted for different initial embedment levels against the
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-14
non-dimensional time T for a smooth pipe in Figure 8.11. The consolidation response
can be fitted by a simple hyperbolic equation of the form
m50init,av
av
)T/T(11
uu
+=
ΔΔ
8.5
where T50 is the non-dimensional time required for 50% dissipation of the average
excess pore pressure. The values of T50 and index ‘m’ for different embedment levels
are tabulated in Table 8.1. The T50 values from the present study are less than published
previously for elastic soil (Gourvenec & White, 2010), indicating faster dissipation
(Table 8.2).
0
0.2
0.4
0.6
0.8
1
1.2
0.00001 0.0001 0.001 0.01 0.1 1 10 100
T = cvt/D2
Δu a
v/ Δu a
v,in
it
w/D = 0.1w/D = 0.2w/D = 0.3w/D = 0.4w/D = 0.5
Symbols - LDFE resultsLines - hyperbolic fits
Figure 8.11 Dissipation of excess pore water pressure with non-dimensional time T (smooth interface)
Table 8.1 Values of T50 and constant ‘m’ of hyperbolic fits
Initial embedment ratio, w/D T50 m
0.1 0.015 0.85 0.2 0.032 0.88 0.3 0.052 0.93 0.4 0.075 1.00 0.5 0.090 1.05
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-15
Table 8.2 Comparison of values of T50 from the present study and elastic solution (Gourvenec & White, 2010)
T50 Initial embedment ratio,
w/D Present study
Gourvenec & White (2010)
0.1 0.015 0.018 0.2 0.032 0.042 0.3 0.052 0.068 0.4 0.075 0.096 0.5 0.090 0.121
8.5 LATERAL BREAKOUT RESISTANCE
8.5.1 Background After the pipe is partially embedded into the seabed, it can be displaced laterally in
response to internal temperature and pressure, or as a result of external hydrodynamic
loading. The breakout resistance, i.e. the peak lateral resistance experienced by the pipe
as it displaces laterally, depends strongly on the strength of the surrounding soil. The
direction of the pipe movement at this stage also depends on the weight of the pipe
relative to the strength of the seabed (although note that this will not be simulated well
in the present study, because of the artificially high shear strength resulting from the
surcharge of 200 kPa).
The available solutions for breakout resistance in the literature are mainly
confined to undrained breakout, with the strength of the surrounding soil being
unaffected by consolidation. In reality, there is always a significant duration between
pipe laying and operation. During this time, consolidation of the soil below and around
the pipe can occur under the weight of the pipe and contents. Dissipation of positive
excess pore water pressure leads to an increase in the shear strength of the soil near the
pipe. So, before breakout occurs, the strength distribution around the pipe is altered, and
the breakout resistance is potentially raised.
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-16
The breakout resistance depends on the load path in V-H space, so the best basis for
describing the potential breakout resistance is to determine the yield envelope in V-H
space. In this study, yield envelopes in V-H load space have been evaluated, for a
smooth pipe-soil interface only, for two conditions: (i) immediately after undrained
penetration (referred to as unconsolidated, undrained); and (ii) after full consolidation
following undrained penetration (referred to as consolidated, undrained). In both cases,
the pipeline movement during penetration and breakout was at rate corresponding to a
normalised velocity of 100, giving nominally undrained conditions. The failure loads in
V-H space were obtained by displacing the pipe by 10 % of its diameter in different
directions.
8.5.2 Unconsolidated undrained yield envelopes Firstly, the unconsolidated undrained case is considered. The limiting values of vertical
and horizontal resistances have been normalised by the initial undrained shear strength
at that depth. The resulting yield envelopes in V-H space for w/D = 0.1, 0.2, 0.3, 0.4
and 0.5 are shown in Figure 8.12.
0
1
2
3
0 1 2 3 4 5 6
V/Dsu0
H/D
s u0
Randolph and White (2008)
Present study
Merifield et al. (2009) horizontal resistance
Merifield et al. (2009) vertical resistance
Curves in orderw/D = 0.1, 0.2, 0.3, 0.4 and 0.5
Figure 8.12 Yield envelopes for different initial embedments for unconsolidated undrained case (smooth pipe)
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-17
These results exceed the equivalent results presented by Randolph & White (2008) by
typically 15% because the latter analyses considered only a flat seabed (with the pipe
wished into place) and ignored the self-weight of the soil. Buoyancy effects due to soil
self-weight are minimal in the present study, because of the high shear strength.
However, the berm of soil displaced during penetration in the present study results in
greater soil resistance. Merifield et al. (2009) reported results of finite element analyses
using a Tresca model and also considered the effects of soil berms on the vertical
penetration and horizontal breakout resistances. Results from that study for pure vertical
and pure horizontal pipe movements are also shown in Figure 8.12. Their results are
close to the present study, with a maximum error below 9 % (except for w/D = 0.1). The
close agreement confirms the correct operation of the Modified Cam Clay soil model
for fully undrained conditions.
8.5.3 Consolidated undrained yield envelopes Figure 8.13 shows the consolidated undrained yield envelopes for different initial
embedment levels after full pore pressure dissipation. To make a comparison between
the unconsolidated and consolidated cases, results for two initial embedment levels of
w/D = 0.1 and 0.5 are plotted in Figure 8.14. The growth in the size of the yield
envelope was 66 % for pure vertical movement for both w/D = 0.1 and 0.5. For pure
horizontal movement, the horizontal resistance is 53 % greater for the consolidated case
for w/D = 0.1, whereas the increase is 83 % for w/D = 0.5.
Contours of the strength increase compared to the original undrained shear
strength at the embedment level of w/D = 0.5 (w/D ~ 0.75 following consolidation) are
shown in Figure 8.15. The consolidated strength reaches 2.1 times the original strength
just beneath the pipe. The shape of the bulb of strength increase is consistent with the
change in shape of the yield envelopes. The increase is strength is greatest beneath the
pipe – within the soil that controls the response to vertical loading – and a lower
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-18
increase in strength is evident to the side of the pipe – within the soil that controls the
response to horizontal loading.
0
1
2
3
4
0 1 2 3 4 5 6 7 8
V/Dsu0
H/D
s u0
Curves in orderw/D = 0.1, 0.2, 0.3, 0.4 and 0.5
Figure 8.13 Yield envelopes for different initial embedments for consolidated undrained case (smooth pipe)
0
1
2
3
4
0 1 2 3 4 5 6 7 8
V/Dsu0
H/D
s u0
Consolidated undrainedUnconsolidated undrainedParabola fit
w/D = 0.1
w/D = 0.5
Figure 8.14 Comparison of yield envelopes for unconsolidated undrained and consolidated undrained conditions for w/D = 0.1 and 0.5 (smooth pipe)
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-19
Figure 8.15 Contours of ratios of consolidated shear strength to the original shear strength (smooth pipe)
8.5.4 Simple equation fit
The finite element results shown in Figure 8.12 and Figure 8.13, which form the
undrained yield envelopes, can be fitted by an equation with the form of a distorted
ellipse:
21
0u0u
max
0u0u DsV
DsV
DsV.f
DsH
ββ
⎟⎟⎠
⎞⎜⎜⎝
⎛−⎟⎟
⎠
⎞⎜⎜⎝
⎛β=
8.6
where
( )( )
21
21
21
21ββ
β+β
βββ+β
=β 8.7
Here, the parameters β1 and β2 skew the ellipse and f is a factor determining the aspect
ratio of the ellipse; Vmax is the undrained resistance under pure vertical loading. The
values of the parameters f, β1 and β2 for different initial embedment levels and
unconsolidated and consolidated cases are tabulated in Table 8.3.
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-20
Table 8.3 Values of f, β1 and β2 for unconsolidated and consolidated conditions
Unconsolidated Undrained
Consolidated Undrained Initial embedment ratio,
w/D f β1 β2 f β1 β2
0.1 0.14 0.83 0.64 0.26 0.81 0.53 0.2 0.16 0.70 0.65 0.32 0.75 0.53 0.3 0.22 0.65 0.61 0.39 0.46 0.52 0.4 0.31 0.53 0.58 0.47 0.41 0.70 0.5 0.37 0.45 0.59 0.55 0.35 0.70
The values of normalised Vmax versus normalised embedment for unconsolidated and
consolidated cases are plotted in Figure 8.16. These responses can be fitted by a simple
power law equation of the form:
b
0u
max
Dwa
DsV
⎟⎠⎞
⎜⎝⎛=
8.8
Coefficients ‘a’ and ‘b’ for unconsolidated and consolidated cases are listed in Table 8.4.
In the same figure (Figure 8.16), results from Merifield et al. (2009) are also plotted for
comparison with the unconsolidated undrained case and show good agreement.
Table 8.4 Power law fit coefficient ‘a’ and ‘b’ for unconsolidated and consolidated
conditions
Results for 1 kPa surcharge are also shown for the unconsolidated and the consolidated
cases in Figure 8.16 to compare with the more natural case. There are considerable
buoyancy effects in the overall resistance for the 1 kPa surcharge case, unlike the 200
kPa case. The maximum penetration resistances for the 1 kPa surcharge case that are
plotted in Figure 8.16 are the geotechnical capacity after correction for buoyancy.
Conditions a b
Unconsolidated Undrained 5.7 0.29
Consolidated Undrained 9.3 0.28
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-21
0
0.1
0.2
0.3
0.4
0.5
0 1 2 3 4 5 6 7 8Vmax/Dsu0
w/D Unconsolidated undrained
Consolidated undrained
Merifield et al. (2009)
Unconsoildated undrained, 1kPa surchargeConsolidated undrained, 1 kPasurcharge
Fitted power law curve
Figure 8.16 Maximum vertical penetration resistances for unconsolidated undrained and consolidated undrained conditions (smooth pipe)
8.6 CONCLUDING REMARKS The consolidation of soil around deep-water pipelines is an important phenomenon to
consider for correct prediction of pipe-soil interactions. A large deformation finite
element (LDFE) methodology combined with the Modified Cam Clay plasticity soil
model was developed for this study to explore the coupled consolidation behaviour
beneath partially embedded seabed pipelines.
The penetration resistance during embedment of the as-laid pipes depends
markedly on the rate of penetration, with the resistance increasing with the degree of
consolidation during penetration. Results have been presented for both smooth and
rough pipe-soil interfaces for normalised embedment, w/D, from 0.1 to 0.5. Up to 72 %
increase in resistance was observed from fully undrained to fully drained conditions.
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-22
Backbone curves, i.e. penetration resistance versus non-dimensional velocities
for smooth and rough pipes, have also been presented. From these curves, fully drained
conditions pertain for vD/cv = 0.1 or lower, while undrained conditions pertain for
vD/cv = 100 or higher. The backbone curves are presented in terms of simple hyperbolic
equations fitted to the LDFE results.
The strength of the soil beneath and around the pipe, and hence the breakout
resistance, depends on the extent to which consolidation occurs following penetration.
The resulting changes in the size and shape of the undrained yield envelopes were
shown to be significant for initially normally consolidation conditions. It was found that
the bearing capacities during vertical and horizontal movements were increased by 66 %
and 83% respectively, for an embedment of w/D = 0.5 if consolidation under the full
vertical bearing capacity was permitted following penetration. These effects of
consolidation – which strongly affect the stability of an on-bottom pipeline – are
important to consider in design in order to provide realistic prediction of pipe-soil
interaction forces.
8.7 REFERENCES AtkinsBoreas (2008). SAFEBUCK JIP – Safe design of pipelines with lateral buckling.
Design Guideline. Report No. BR02050/SAFEBUCK/C, AtkinsBoreas.
Chatterjee, S., Yan, Y., Randolph, M. F. & White, D. J. (2012). “Elastoplastic
consolidation beneath shallowly embedded offshore pipelines.” Géotechnique
Letters , 2, 73-79.
Dassault Systèmes. (2011). Abaqus analysis users' manual, Simula Corp, Providence, RI,
USA.
CHAPTER 8: Effects of consolidation…
Centre for Offshore Foundation Systems 8-23
Gourvenec, S. M. & White, D. J. (2010). “Elastic solutions for consolidation around
seabed pipelines.” Proc. Offshore Technology Conf., Houston, Texas, USA, Paper OTC
20554.
Merifield, R. S., White, D. J. & Randolph, M. F. (2009). “Effect of surface heave on
response of partially embedded pipelines on clay.” J. Geotech. Geoenviron. Engng,
ASCE 135, No. 6, 819-829.
Randolph, M. F. & White, D. J. (2008). “Upper-bound yield envelopes for pipelines at
shallow embedment in clay.” Géotechnique 58, No. 4, 297-301.
Roscoe, K. H., & Burland, J. B. (1968). On the generalised stress-strain behaviour of
'wet clay'. Engineering plasticity, Cambridge University Press.
Schofield, A. & Wroth, C. P. (1968). Critical State Soil Mechanics, McGraw-Hill, New
York.
White, D. J. & Cheuk, C. Y. (2005). SAFEBUCK JIP: Lateral pipe-soil interaction:
Data review. Report to Boreas Consultants (SAFEBUCK JIP), ref. SC-CUTS-0502-
R02. 77pp.
White, D. J. & Cheuk, C. Y. (2009). SAFEBUCK JIP: Pipe-soil interaction models for
lateral buckling design: Phase IIA data review. Report to Boreas Consultants
(SAFEBUCK JIP), UWA report GEO 09497. 185pp.
Wroth, C. P. (1984). “The interpretation of in situ soil tests.” Géotechnique 34, No. 4,
449-489.
Numerical modelling of pipe-soil interactions
Centre for Offshore Foundation Systems 9-1
CHAPTER 9
CONCLUDING REMARKS
9.1 ORIGINAL CONTRIBUTIONS
This thesis shows how large deformation finite element (LDFE) analysis can be used to
shed light on the pipe-soil interaction forces during embedment and lateral buckling of
on-bottom pipelines, and reports parametric studies that have shown the underlying soil
deformation mechanisms and led to improved calculation methods for use in design.
The remeshing and interpolation technique with small strain (RITSS) approach
combined with commercial software ABAQUS has been adopted to develop the LDFE
methodology. The specific outcomes of this research are as follows.
9.1.1 Vertical penetration
The design parameters that define the soil resistance to lateral and axial motion of the
pipeline are a function of the amount of vertical embedment. However, vertical
embedment is difficult to estimate, partly because of the effects of soil heave around the
pipeline as it penetrates, and partly because the soil shear strength depends on the strain
rate and the degree of softening as the soil is sheared and remoulded. The large
deformation finite element approach was implemented in ABAQUS firstly to study
pipe-soil interaction during vertical embedment of pipelines on the seabed. This
implementation utilises a soil constitutive model that captures the effects of penetration
rate and softening on soil shear strength.
The contributions of shearing resistance and buoyancy to penetration resistance
were characterised in simple expressions. A detailed parametric study was performed,
CHAPTER 9: Concluding remarks
Centre for Offshore Foundation Systems 9-2
varying the strain rate and softening parameters to explore their effects on geotechnical
resistance. It was found that strain rate and softening have significant effects on the
penetration resistance of shallowly embedded pipes and vertical resistances can vary
widely with the change of these parameters.
An ‘operative shear strength’ incorporating the effects of strain rate and softening
was proposed. Normalising the vertical resistance with this equivalent shear strength led
to narrow band of values. Simple power law expressions for penetration resistance
varying with normalised embedment were presented suited to application in design.
9.1.2 Lateral pipe-soil interactions
Accurate assessment of the lateral pipe-soil interaction forces during large-amplitude
movements of subsea pipelines is required to support lateral buckling design. In this
thesis, two important stages of pipe-soil lateral interactions – initial break-out behaviour
and steady-state residual behaviour – were studied in detail.
Initial yield envelopes and break-out resistance
The initial break-out behaviour of the pipe during lateral motion was studied
numerically using large deformation finite element analyses as well as limit analysis
software OxLim. The present LDFE methodology suggests improvement over the
available numerical solutions in the literature because it incorporates a softening soil
model and generation of soil heave around the pipe is accounted for. Marginal
improvement of the presently available plasticity solution is also discussed. For the first
time, yield envelopes for heaved soil, with consideration of the soil self weight, were
presented and a good match of results between LDFE and limit analyses were obtained.
It was shown that for a flat seabed the soil failure mechanism is negligibly altered due
to introduction of self-weight and a superposition approach is adequate to capture the
influence of buoyancy. It was also concluded that the presence of a soil berm around the
CHAPTER 9: Concluding remarks
Centre for Offshore Foundation Systems 9-3
pipe has a significant effect on the breakout resistance and an increase of up to 18% in
lateral capacity was observed.
Residual friction factor
It was found that a steady residual resistance and embedment is achieved after the pipe
has moved laterally by 2-3 times its diameter. In the LDFE analyses, a very large lateral
displacement ranging from three to seven times the diameter was imposed to ensure the
steady state was reached. Both ideal (no rate effect and softening) and realistic (effects
of strain rate and softening incorporated) soil models were explored. If soil softening
was not included, the response differed in some critical aspects from the behaviour
observed in model tests available in the literature. The lateral resistance was over-
predicted and the trajectory of the pipe was too shallow. When softening was included
in the analysis, the trajectory deepened but this was compensated by a lower operative
soil strength in the partly-remoulded soil ahead of the pipe. Good agreement with model
test data was achieved – both the resistance and the pipe trajectory were correctly
reproduced.
A detailed parametric study was also performed to explore the effects of the
initial embedment and pipe weight on the residual friction factor. For the range of
parameters chosen, the residual friction factor varied from 0.5 to 0.98, which is
comparable to the SAFEBUCK JIP database. Initial embedment and operating pipe
weight have considerable effects on the steady state lateral resistance. The steady state
lateral friction factor was shown to depend on the ‘effective embedment’ of the pipe
irrespective of other soil and pipeline parameters. The ‘effective embedment’ is defined
as the height of the growing soil berm ahead of the pipe, discounted for remoulding
effects. The evolution of the effective embedment and the resulting lateral resistance
was shown to depend on the initial embedment and the operating pipe weight. This
CHAPTER 9: Concluding remarks
Centre for Offshore Foundation Systems 9-4
conceptual approach provides a useful contribution towards the development of a
general model for describing large-amplitude lateral pipe-soil interaction. At any instant,
the effect of the soil softening and changing seabed topography can be distilled into the
effective embedment parameter. Based on this concept, simple correlations were also
provided to estimate the steady lateral resistance. These were shown to provide more
accurate predictions than other empirical relationships that have been proposed.
The complete load-displacement response over large movements was also
shown to be well-fitted by an exponential relationship interpolating between the break-
out and residual resistances.
9.1.3 Coupled consolidation analyses
Most previous studies using LDFE and the RITSS approach have been based on total
stress analyses. In this research, a further advance was to combine LDFE and RITSS
with an effective stress approach using the modified Cam Clay soil model. Analyses
were performed to study the coupled consolidation behaviour of partially embedded
seabed pipelines. The pore water pressure dissipation time history and consolidation
settlement were studied after the pipe had been partially penetrated to different
embedment levels in undrained conditions. Results were fitted by simple equations to
facilitate application in practice. The effect of the rate of penetration on the degree of
consolidation was also studied. Backbone type curves defining zones of drained,
partially drained and undrained behaviour, depending on the non-dimensional velocity
were also presented.
Changes in lateral breakout behaviour due to drainage and consolidation were
also explored. For normally consolidated soil, the shear strength near the pipe increases
significantly due to consolidation under the weight of the pipe. Yield envelopes defined
in V-H space for unconsolidated undrained and consolidated undrained cases were
CHAPTER 9: Concluding remarks
Centre for Offshore Foundation Systems 9-5
compared and it was shown that the size of the V-H yield envelope can almost double
following consolidation.
9.2 LDFE ANALYSIS IN DESIGN
9.2.1 LDFE in support of simplified design method
The analyses reported in this research have shown that LDFE techniques, coupled with
an appropriate soil model, can capture faithfully some of the behaviour observed during
large-amplitude lateral pipeline movements. The geometric and soil strength changes
can be replicated well, at least for an initial lateral sweep of several diameters.
LDFE and physical model testing are complementary tools. Both allow
parametric studies to be performed. It is easier to replicate tests with controlled
variations of the input parameters via LDFE, but physical modelling has the advantage
of utilising ‘real’ soil. Given these complementary attributes, both LDFE and model
testing currently play a significant role in the development of analysis methods for the
assessment of pipe-soil interactions. Having assembled databases of results it is
convenient to devise approximate relationships that capture the same behaviour. These
are then a convenient tool for use in practice.
9.2.2 LDFE directly applied in design
LDFE techniques could be used directly in design on at least two levels. Firstly, specific
plane strain analyses could be performed to develop site-specific (and pipeline-specific)
lateral force-displacement curves, which would then be converted to non-linear ‘p-y
springs’, making some assumptions regarding generalisation to arbitrary load and
displacement paths.
A second, more detailed approach would be to model the soil and structural
responses concurrently, either by creating a full three-dimensional soil model, or by
CHAPTER 9: Concluding remarks
Centre for Offshore Foundation Systems 9-6
attaching two-dimensional plane strain soil domains at each node along the structural
model of the pipeline. Both of these latter approaches are hugely demanding from a
computational perspective.
Realistically, this type of analysis still lies beyond the time constraints of real
projects. However, as computational power increases, and the obstacles to realistic
modelling of the soil response are overcome, this situation is likely to change.
9.3 LIMITATIONS AND FUTURE RESEARCH
This research has shown the capabilities of LDFE analysis in successfully modelling the
pipe-soil interaction forces during vertical penetration and lateral displacement. Despite
the promising results and findings from this research, a number of limitations still exist
and there is significant scope for future research.
9.3.1 Dynamic effects
Partial embedment of deep water as-laid pipes into the seabed depends on the weight of
the pipeline and the strength of the soil, although the laying process and dynamic
motions of the laying vessel (leading to cyclic motions of the pipe) make it significantly
more complicated. In this research, though static embedment of the pipe using an
advanced soil model incorporating the effects of strain rate and softening has been
studied, it was not possible to include the dynamic and cyclic effects within the scope of
the thesis. However, it has been observed that the as-laid pipe penetration is generally
much greater than that calculated from the effect of the self-weight alone. It is, therefore,
suggested that future research effort to numerically model embedment of deep water
pipes should be directed towards capturing cyclic effects.
CHAPTER 9: Concluding remarks
Centre for Offshore Foundation Systems 9-7
9.3.2 Full 3D model
All the analyses in this thesis were based on a two-dimensional plane strain model
assuming the pipe to be a rigid body. During lateral buckling, the pipe may undergo
large lateral bending strains. Hence the assumption of a rigid pipe may be too simplistic.
As mentioned in the previous section, concurrent modelling of structural and
geotechnical responses using a full three-dimensional (3D) model may be a feasible
solution. However, full 3D modelling of large deformation effects of both structural and
soil domains is a significant challenge from a numerical perspective. Also, the high
computation time required for such analyses poses a considerable challenge for real
projects. With the advent of supercomputers, and also new solver techniques, full 3D
modelling of simultaneous structural and geotechnical responses could be achieved in
the near future.
9.3.3 Whole life behaviour
During lateral bucking, the pipe moves several times its diameter to and fro for a
number of cycles. During this time, the soil around the pipe may undergo periods of
strength reduction due to softening and strength gain due to consolidation. In the present
research, only the first cycle of lateral movement has been studied. Also, consolidation
effects after large lateral motion has not been considered. It is therefore necessary for
future work to consider the whole life behaviour of a pipe element sweeping repeatedly
within a buckle and to extend the soil constitutive modelling to capture consolidation
effects within the large lateral deformation framework.
Despite restrictions of the current work and scope for future research, it is
considered that the outcomes of this research present several useful contributions for the
investigation of pipe-soil interaction. They also provide validation and calibration of
CHAPTER 9: Concluding remarks
Centre for Offshore Foundation Systems 9-8
existing calculation methods for pipe-soil interaction, and have generated new analysis
techniques that can be utilised to investigate project-specific refinements.