See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/284271504

Geotechnics for subsea pipelines

Chapter · October 2010

DOI: 10.1201/b10132-6

CITATIONS READS

38 6,025

2 authors:

David J White David Cathie

University of Southampton Cathie Associates
293 PUBLICATIONS   8,843 CITATIONS    34 PUBLICATIONS   387 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Upheaval buckling and flotation of buried offshore pipelines - DPhil thesis at Oxford University, Department of Engineering Science View
project

ROBOCONE: intelligent robotics for next generation ground investigation and design View project

All content following this page was uploaded by David J White on 13 December 2015.

The user has requested enhancement of the downloaded file.

Frontiers in Offshore Geotechnics II – Gourvenec & White (eds)
© 2011 Taylor & Francis Group, London, ISBN 978-0-415-58480-7

Geotechnics for subsea pipelines

D.J. White
Centre for Offshore Foundation Systems, University of Western Australia, Perth

D.N. Cathie
Cathie Associates, Brussels, Belgium

ABSTRACT: The geotechnical analysis performed for subsea pipeline design involves challenges that are not
common in conventional foundation engineering. This paper reviews recent research in pipeline geotechnics and
shows examples of how this research is being applied in practice. A general theme running through this paper is
the twin challenges of the changes in seabed topography and the changes in soil properties that occur through
the installation and operating life of a pipeline. Results from in situ and element testing of soils that replicate the
loading and disturbance imposed by pipelines are used to show that significant changes in strength are induced.
Soil generally weakens during the episodes of remoulding that accompany pipeline laying, buckling, walking
and storm loading, and during ploughing and trenching. The soil strength recovers during subsequent episodes
of reconsolidation between storms, and between startup and shutdown events. Solutions for incorporating this
behaviour into the estimation of axial and lateral pipe-soil resistance, and the assessment of trenching and
ploughing operations, are discussed. A unifying theme is the relative magnitude of drained and undrained soil
strengths, the evolution of these strengths through cyclic episodes, and the importance of recognising the widely-
varying rates of shearing involved in pipe-soil processes. Pipeline geotechnics can involve drained behaviour in
fine-grained clayey soils – for example, during slow axial expansion of pipelines – and undrained behaviour in
coarse-grained soils – for example during ploughing. Concepts from critical state soil mechanics often provide
a simple framework for clarifying this behaviour.

1 INTRODUCTION soft fine-grained soils, where the management of ther-

mal and pressure-induced expansion is a critical design
1.1 Scope of paper issue. The second is the stability of light large-diameter
pipelines in shallow water, where primary and sec-
The purpose of this keynote paper is to set out the chal-
ondary stabilization measures represent a significant
lenges of pipeline geotechnics and to highlight some
capital expenditure, and where geotechnical analysis
recent developments in this area that the authors have
techniques are not well established. The first scenario
been involved with. This is not an exhaustive treatment
is relevant to pipeline design in almost all deepwater
of the subject, but is intended to
frontiers globally.The second scenario is particular rel-
– provide an overview of areas in which intensive evant off the coast of Australia, where many hundreds
research has recently been published. of kilometers of gas trunkline are currently planned,
– highlight certain novel analysis techniques for to bring gas from deepwater fields to onshore LNG
design that this research is beginning to permit. facilities.
Section 1.2 provides a brief overview of some
More complete introductions to pipeline geotech- key pipeline design considerations that have a strong
nics are provided by the relevant chapters of books geotechnical influence. This overview is intended to
on offshore geotechnical engineering by Randolph provide a brief introduction to some of the most novel
and Gourvenec (2010) and Dean (2010). Cathie et al. aspects of pipeline geotechnics, for those who are
(2005) presented a more exhaustive review of research unfamiliar with this area of geotechnics. In Section 1.3
across the whole of pipeline geotechnics. The present comparisons are made between pipeline geotechnics
paper is intended to be of value to pipeline engi- and conventional foundation engineering.
neers as well as geotechnical specialists, and includes Some of the most relevant aspects of soil behaviour
some basic geotechnical content where we consider are highlighted in Section 2, using recent experimen-
this useful. tal observations, including some from unusual forms
The topics in this paper are generally relevant to of penetration testing. These observations provide a
one of two design scenarios. The first is high pres- backdrop to the mechanisms and analysis techniques
sure high temperature pipelines laid in deepwater on that are outlined in Sections 3, 4 and 5 for assessing

87
Figure 1. Some geotechnical aspects of pipeline design.

pipeline embedment, lateral pipe-soil resistance and seabed. A pipeline is a forgiving structure, being
axial pipe-soil resistance respectively. Sections 6 and 7 able to tolerate significant deformation and gross
are focussed on the geotechnics of pipeline trench con- movements across the seabed, except at points of fix-
struction by ploughing and jetting respectively. Space ity such as end terminations. Such instabilities are
limitations preclude discussion of in-trench pipeline unacceptable for the platforms and foundations that
stability, and the associated upheaval, backfill lique- conventional geotechnical engineering is equipped to
faction and flotation issues. The paper finishes with design. Indeed, if a pipeline was not permitted to move
brief conclusions. across the seabed under thermal loading, this would
often induce unacceptable thermally-induced stresses.
The forgiving flexibility of a pipeline does not,
1.2 Pipe-soil interaction processes
therefore, alleviate the need to quantify the pipe-
Many of the areas of pipeline design that have geotech- soil resistance forces to a sufficient accuracy that the
nical aspects are illustrated in Figure 1. Offshore robustness of the design is demonstrated.
pipelines are often left on the seabed, unburied, if this One of the most difficult aspects of pipeline design,
does not lead to unacceptable instability under hydro- which is an increasing challenge as operating tempera-
dynamic loading. The interaction between the pipeline tures and pressures rise, is the management of thermal
and the seabed feeds into many aspects of the pipeline and pressure-induced loading. Controlled on-bottom
design. If the pipeline must be buried, for stability or to lateral buckling is an attractive design solution but one
avoid fishing gear, the shielding of the pipeline via the which requires the pipe-soil responses to be bracketed:
construction of a trench (possibly backfilled) requires both high and low geotechnical resistance can hamper
geotechnical design. a design (Bruton et al. 2007; AtkinsBoreas 2008).
On-bottom pipelines are increasingly being designed A second and related behaviour that arises from the
to allow movement during their operation, either under thermal and pressure-induced loading is the tendency
hydrodynamic loading or under thermal and pressure- for pipes to ‘walk’ axially over cycles of startup and
induced expansion. Steel catenary risers, which are shutdown (Tornes et al. 2000; Carr et al. 2006). This
extensions of pipelines that connect to surface facil- phenomenon can be driven by the asymmetry of the
ities, inevitably move where they touchdown on the heat-up and cool-down processes or by the presence
seabed, in response to oscillation of the floating of a seabed slope or end-of-line tension (which creates
facility. an asymmetry in the mobilized axial pipe-soil resis-
Throughout the lay process and during subse- tance).Accurate assessment of the axial pipe-soil resis-
quent operating cycles, the pipeline is subjected to tance forces is required for robust modelling of this
geotechnical forces where it is in contact with the process.

88
Table 1. Comparison of pipeline geotechnics and conventional foundation engineering (after White and Gaudin 2008).

Foundation On-bottom pipeline

Problem geometry Known, controlled. Uncertain. Embedment affected by lay process

and metocean conditions. Subsequent pipeline
movements disturb seabed topography.
Design criteria for To remain fixed, movement uD. May be required to displace significantly, uD,
in-service behaviour through hundreds of cycles of operation or
hydrodynamic loading.
Surrounding soil conditions Similar to in situ state. Relatively Soft soil is significantly affected by installation.
unaffected by installation. Remoulding, heave and reconsolidation
affect the local strength.
Soil-structure interaction Usually minimal. Imposed loads Often significant. Local pipe-soil load-displacement
are not strongly affected by relationship affects overall pipeline response.
foundation displacements.
Soil-ocean interaction Scour and wave-induced liquefaction Scour and wave-induced liquefaction can dominate
may require mitigation behaviour
Single conservative design Usually available. Can assume Often unavailable. Both upper and lower bound
approach lowest credible geotechnical geotechnical capacity may adversely effect
capacity structural response

Another significant design issue that is particularly foundations and piles. This is partly because it is
relevant in the shallow waters offshore Australia, is only recently that design codes have permitted gross
pipeline stability under hydrodynamic loading from pipeline movement, and so designers have not needed
storm-induced currents and waves. In this situation a to explicitly assess the interaction forces as pipelines
conservative approach is to adopt a low value of soil sweep across the seabed. Also, it is only recently
resistance. However, the cost of stabilization measures that some of the complexities of the underlying soil
such as concrete coating is huge, and there is a strong behaviour have been recognized. A generally accepted
incentive to refine the geotechnical analysis to remove framework for routine analysis has not emerged.
any unnecessary conservatism in the design seabed The contrasts between pipeline geotechnics and
resistance. conventional foundation engineering are summarised
In shallow water a pipeline may require additional – in Table 1 and illustrated in Figure 2. The designer’s
‘secondary’ – stabilisation for hydrodynamic stabil- task in the geotechnical design of a pipeline is aided by
ity. Secondary stabilisation solutions revolve around the structure’s tolerance of movements and mild defor-
reducing the hydrodynamic loading and increasing the mation, but is hampered by the difficulty of assessing
available lateral resistance. An open trench provides the geometry of the scenario and the operative soil
partial shielding from hydrodynamic load. Burial properties. The laying of a pipeline and any subsequent
of the pipe eliminates direct hydrodynamic load- lateral or axial movements disturb the topography of
ing (although soil liquefaction under hydrodynamic the seabed. The changed geometry and the altered soil
loading can destabilise a buried pipe). Geotechnical properties need to be captured in calculations of the
assessments must be made of the trenching process – available pipe-soil resistance.
which may be by ploughing, cutting, jetting, dredging Even the intact soil properties are difficult to
or a combination. establish at the shallow embedments relevant to
Other secondary stabilisation techniques include pipeline geotechnics. Undisturbed sampling of soft
continuous rockdumping, or engineered solutions to near-surface soils is difficult and penetrometer tests
provide local anchoring at intervals along the pipe. at shallow embedment require particular interpretation
These solutions include flexible concrete mattresses, techniques (Puech and Foray 2002, White et al. 2010a).
anchor blocks or saddles placed over the pipeline, or Some soil properties and parameters such as friction
small piles on either side of the pipeline. The stability angles and undrained strength ratios tend to be differ-
of these objects must also be assessed in design, taking ent at the very low stress levels relevant to pipeline
account of the additional cyclic loading transferred to geotechnics.
them by the unstable pipeline. A further complication is that interaction between
the ocean and the seabed – leading to scour and
liquefaction – can be significant in shallow water. The
1.3 Comparison with foundation engineering
result is a tripartite interaction between the ocean,
Geotechnical design procedures for pipelines and the pipeline and the seabed, which is illustrated in
risers are relatively undeveloped compared to Figure 3. Cross-disciplinary design approaches for

89
Figure 3. Tripartite interaction between the seabed, the
ocean and a pipeline in shallow water during storms.

buckles to sweep laterally across the seabed – and

storms – which create high hydrodynamic loading.
Over the operating life of a pipeline, the surround-
ing soil may therefore be subjected to a large number
of episodes of disturbance followed by recovery and
reconsolidation. The soil strength will generally fall
and rise with each episode, and the net effect may be an
overall increase or decrease in the strength of the soil.
Coupled with the associated changes in seabed topog-
raphy, this may lead to a rise or a fall in the geotechnical
restraint on the pipeline.
Figure 4a illustrates the case of a buckling pipeline
on a fine-grained soil (in deepwater, where hydrody-
namic loading is negligible). Each startup or shutdown
of the buckle causes gross monotonic remoulding of
the surrounding soil, comparable to the high distur-
bance created during passage of a penetrometer. The
startup and shutdown episodes are separated by a
Figure 2. Comparison of pipeline geotechnics and conven- period of time which may or may not be sufficient for
tional foundation engineering (images from Jayson et al. 2008 full dissipation of the excess pore pressures generated
and Fisher and Cathie 2003). during the previous disturbance – depending, obvi-
ously, on the consolidation characteristics of the soil
this interaction are in their infancy. There is not yet relative to the frequency of the startups and shutdowns.
a routine basis for assessing pipeline stability under Typical pipeline designs require several hundred or
hydrodynamic action that incorporates all three inter- even one thousand shutdown and startup events to be
actions concurrently (Damgaard and Palmer 2001, considered.
Cheng et al. 2010). Figure 4b illustrates the case of a pipeline on a
sandy or silty soil in shallow water, where storms cre-
ate high hydrodynamic loading, which in turn causes
the pipeline to exert cyclic loads on the surrounding
2 RELEVANT SOIL MECHANICS
soil. A storm loading event is perhaps best consid-
ered as a pre-failure cyclic disturbance as distinct
2.1 Illustrations of soil behaviour near pipelines
from the gross (undrained) monotonic remoulding of
The soil close to a pipeline is grossly disturbed as the previous example. In this illustration the cyclic
the pipe is laid. If that disturbance happens rapidly loading leads to a weakening, associated with pore
enough for excess pore pressure to be generated then pressure buildup, which is subsequently compensated
the subsequent reconsolidation process generally leads for by reconsolidation and an associated densification
to an increase in the strength and density of the soil. of the soil.
Subsequent events may disturb the pipeline and the Storms occur at a frequency such that full dissi-
surrounding soil further. These events include the pation occurs between events, for the soil considered
startup and shutdown of the pipeline – which cause here. In design, it is necessary to consider the effect

90
Figure 5. Changes in the undrained penetration resistance
of fine-grained soils during cyclic penetrometer tests (after
Gaudin and White 2009).

remoulded state, with this ratio being termed the sen-

sitivity. Much higher sensitivities can sometimes be
found, particularly in carbonate soils. In the analysis of
problems involving significant disturbance, it is neces-
sary to identify the relevant soil strength, which may lie
somewhere between the intact and remoulded values.
Figure 4. Illustrative histories of soil element behaviour Cyclic T-bar or ball penetrometer tests allow this
near unstable pipelines: episodes of disturbance and recovery. behaviour to be quantified. The progressive reduction
in net bearing resistance through cycles of disturbance
of these disturbances on the available lateral pipe-soil is shown for various different soils in Figure 5. The
resistance – which controls the stability. This stability strength degrades exponentially with the number of
is affected by changes in the pipeline embedment as cycles of disturbance, which allows the response to be
well as the changes in the strength of the soil around characterized by two parameters – the sensitivity, St (or
the pipe during disturbance. its inverse, δrem ) and a parameter related to the ducti-
The same history of disturbance and reconsolida- lity, N95 (which is the number of cycles of disturbance
tion shown in Figure 4b is applicable to the backfill after which the resistance has decayed by 95% of the
above a trenched pipeline. These illustrations of the difference between the intact and remoulded values).
history of soil behaviour are clearly very idealized, It is evident that both of these parameters vary sig-
and it is rare that the related changes in soil strength nificantly between soil types (particularly St ), but the
are tracked explicitly within a design analysis. How- form of the decay is similar.
ever, it is important to recognize these effects, since Analysis techniques for design can capture the
they have a significant influence on the geotechnical decaying soil strength by adopting a value that rep-
restraint on a pipeline. resents the relevant level of disturbance. The general
These illustrations provide a convenient back- approach is to firstly convert the penetrometer ductil-
ground to the following three examples of soil element ity parameter, N95 , to an equivalent strain level (e.g.
behaviour. These examples show the changes in soil Zhou and Randolph 2009). The relevant strain level for
strength that can accompany episodes of disturbance the problem being considered is then used to deduce
and recovery, and also showcase novel testing tech- the operative undrained strength.
niques that may in the future be utilized to quantify Such techniques have been proposed for spud-
this behaviour for design. The examples are: can penetration (Erbrich 2005, Hossain and Randolph
– episodes of undrained disturbance and reconsoli- 2009). These methods utilise the type of strength
dation seen by a T-bar penetrometer; degradation curves shown in Figure 5 to link the
– episodes of undrained disturbance and reconsoli- strains and operative strength around a spudcan to
dation seen by a vertical rod penetrometer; those around a T-bar. T-bar and spudcan penetration
– drained and undrained failure of different soils; involves a comparable level of disturbance to mono-
– low stress friction response of fine-grained soils. tonic pipe embedment. However, the dynamic motions
that accompany pipe laying mean that a greater level of
disturbance and hence a lower operative soil strength
is applicable, compared to that mobilised during initial
2.2 Disturbance and recovery: T-bar tests
T-bar penetration.
The strength of soft fine-grained seabed soils typically The dramatic reductions in strength evident in
reduces by a factor of 2–5 from the intact to the fully Figure 5 are partly due to the generation of positive

91
Figure 6. Undrained strength through episodes of remould- Figure 7. Critical state interpretation of episodes of
ing and reconsolidation (test in lightly overconsolidated remoulding and reconsolidation (White and Hodder 2010).
kaolin in the UWA beam centrifuge) (White and Hodder
2010).
the attraction of setting this behaviour within an effec-
tive stress framework, but it does rely on a very crude
excess pore pressure in these contractile materials dur- simplification of the overall behaviour, in which the
ing undrained shearing. As this positive pore pressure response of all the elements of soil around a penetrom-
dissipates and the effective stress rises back to the geo- eter is lumped into a single representative effective
static state the material densifies and the subsequent stress level and specific volume.
undrained shear strength may be higher. More refined models will allow this behaviour to be
Cyclic T-bar penetrometer tests with periods of more accurately quantified and numerical simulations
reconsolidation between episodes of cycling show this will test the validity of this simplification. The key
regain in strength. Figure 6 summarises the results of point, however, is that the rises in soil strength during
a cyclic T-bar test in kaolin clay reported by White and episodes of reconsolidation can eclipse the reduc-
Hodder (2010), expressing the T-bar strength at a par- tions in soil strength during the preceding episodes
ticular depth – 2.25 m – during each cycle. After just of remoulding.
three episodes of full remoulding and reconsolidation,
the current remoulded strength was comparable to the
2.3 Disturbance and recovery: vertical rod tests
original intact strength. These results quantify the con-
trasting effects of disturbance and recovery shown in The example above involves only 3 episodes of recon-
Figure 4a for this soil and the particular disturbance solidation. A significantly larger number of episodes
pattern imposed by a T-bar. of reconsolidation are involved in the second exam-
This behaviour is easily understood within a critical ple. A novel vertical rod penetrometer has been used
state-type framework, since this provides an explicit on a recent centrifuge project at UWA, with the aim
link between moisture content (which reduces as posi- of quantifying the resistance and strength of surficial
tive pore pressures dissipate) and undrained strength. material, as the soil is forced to flow past the penetro-
This interpretation can be extended to a quantitative meter. This device is a cylindrical bar, oriented verti-
treatment, expressed in terms of the operative strength cally, 4 mm in diameter. The device is embedded until
averaged over all of the soil near the penetrometer the tip is typically 5 – 10 diameters below the soil sur-
(rather than of a single soil element). An accurate face. The bar is equipped with multiple levels of strain
back-analysis of the results shown in Figure 6 can gauging located above the soil surface, which allow
be achieved by defining two failure lines in stress- the magnitude and distribution of the pressure on the
volume space, which represent the intact and fully bar to be derived.
remoulded strengths of the soil, as proposed by White In one test the bar was embedded in soft kaolin to a
and Hodder (2010). depth of 45 mm then cycled laterally by a distance of
As shown in Figure 7, this back-analysis of the T-bar 20 mm at a rate of 0.3 mm/s. This rate corresponds to a
resistance at a depth of 2.25 m (which corresponds to dimensionless velocity of vD/cv ∼ 10 which is almost

an in situ effective vertical stress of σvo = 12 kPa) is fully undrained, based on the limits demonstrated by
based on the intact strength line (ISL) being reached Finnie and Randolph (1994) (albeit for a different
during the initial T-bar stroke of a episode, and the geometry of problem). The elapsed time between the
effective stress point migrating towards the remoulded bar passing the mid-point of each lateral stroke was
strength line (RSL) according to an exponential trend 100 seconds, which corresponds to a dimensionless
(i.e. the reduction in effective stress per T-bar stroke consolidation time of T = cv t/D2 = 0.5. This value is
is proportional to the difference between the current indicative of significant (∼50%) pore pressure dissi-
effective stress and the effective stress at the remoulded pation, based on limits provided by Randolph (2003)
state for the current specific volume). This analysis has (again for a slightly different geometry of problem).

92
Figure 8. Lateral resistance on a vertical bar penetrometer Figure 9. Drained and undrained resistance during
through episodes of disturbance and reconsolidation. penetration.

The resulting variation in the average lateral resis-

tance on the bar penetrometer is shown in Figure 8. plough share (Peng and Bransby 2010). In all cases
Initially the resistance reduces, as the remoulding the velocity is normalized by the coefficient of con-
damage exceeds the recovery from reconsolidation. solidation of the seabed and an appropriate drainage
However, a steady rise in resistance is soon evident, distance – generally the size of the foundation.
with the strength after several episodes exceeding the There is some variation in the dimensionless veloc-
initial (intact) value. ities at which fully drained and fully undrained
After this test the level of the soil surface around behaviour occurs, although this may be partly due to
the penetrometer had lowered. This is evidence that the difficulties in establishing appropriate values of
the soil surrounding the device had densified through the coefficient of consolidation.
the episodes of reconsolidation, which is consistent There are more significant differences between the
with the changing strength. relative magnitude of the drained and undrained resis-
This example highlights again the changes in soil tance. These arise from the state of the soil – and
strength that can occur through the episodes of distur- hence its tendency to generate positive or negative pore
bance and recovery that can be imposed on the soil pressure in undrained conditions – and also the partic-
surrounding a seabed pipeline. As shown in Figure 8, ular boundary value problem. For example, for a soil
the strength continued to increase through more than with a particular undrained strength and angles of fric-
25 cycles of heavy remoulding, but ultimately should tion and dilation, the relative magnitude of the drained
reach a limiting value, when the soil has densified suf- and undrained bearing capacity depends on both the
ficient that there is no longer a tendency to generate applicable bearing factors (i.e. Nq and Nγ for drained
positive excess pore pressure when disturbed. conditions and Nc for undrained conditions) as well as
the soil strength properties.
The relative magnitude of the drained and undrained
2.4 Drained and undrained soil responses
strengths of an interface is not affected by the geometry
The shear strength of a given soil in a particular state of the problem, so provides a more simple differentia-
depends on whether drained or undrained conditions tion between the drained and undrained behaviour of
are imposed, as well as the mode of shearing. In condi- a particular soil (albeit in combination with a partic-
tions in which drainage is permitted, the shearing can ular interface). Figure 10 shows the steady residual
be imposed at rates that span from fully drained (i.e. resistance measured during monotonic shearing of
in which no excess pore pressure builds up) to fully normally consolidated kaolin clay over a rough steel
undrained (i.e. in which effectively no pore pressure surface at different velocities. The kaolin was normally
dissipation occurs, despite drainage being permitted). consolidated to a stress of 2.5 kPa prior to shear-
Various authors have recently explored the contin- ing. These tests used a direct shear box at UWA that
uous variation in mobilized soil strength and geotech- has been modified to operate at the low stress levels
nical resistance between drained and undrained relevant to pipeline geotechnics.
conditions. Results summarized in Figure 9 show the The four tests show a trend of increasing resis-
variation in penetration resistance of circular surface tance with reducing velocity. This trend is consistent
foundations (Finnie 1993), and cone penetrometers in with a hyperbolic backbone curve of the same form
soft clay (Randolph and Hope 2004) and dense silt as used in Figure 9, drawn between the fully drained
(Silva 2005). and fully undrained limits. These limits correspond to
Similar relationships have been derived for the an interface friction angle of 30◦ and an undrained

uplift resistance of buried pipelines (Bransby and Ire- strength ratio of su /σvc = 0.25, both of which are con-
land 2009) and the sliding resistance of a pipeline sistent with other published results for this stress level

93
Figure 11. High effective stress friction and a non-linear
Figure 10. Interface shear resistance at varying rates, from failure envelope during low stress interface tests and axial
drained to undrained. pipe-soil movement (data from Bruton et al. 2009, White
et al. 2010b).
(Bolton and Barefoot 1997; Pedersen et al. 2003;
Bolton et al. 2009). The same trend appears in low stress soil-soil and
The movements of an on-bottom pipeline in a given soil-interface shearing of fine-grained soils (Pedersen
soil may span the ranges of velocities and therefore et al. 2003, White and Randolph 2007, Hill and Jacob
drainage conditions shown in Figure 9 and Figure 10. 2008, Bruton et al. 2009, White et al. 2010b). Non-
Thermal expansions and the associated lateral move- linear failure envelopes that express the friction angle
ments span from zero at ‘virtual anchor points’ to or limiting stress ratio as a function of effective stress
millimetres per second of axial movement near ends can capture this variation (e.g. Figure 11).
and buckles, and even metres per second of lateral It is important to recognise that the friction angles
movement during initiation of lateral buckles. measured at conventional geotechnical stresses may
A further range of velocities arises from pipeline not be appropriate for the assessment of drained
movements driven by hydrodynamic action – which pipe-soil resistance. The high friction angles are also
include oscillations in the touchdown zone during lay- reflected in the higher undrained strength ratios found
ing that are created by vessel motion, or oscillations at low stresses – a link highlighted in the approxi-
in response to direct hydrodynamic action on the pipe mate expression for normally consolidated undrained

during storms. strength ratio su /σvc = φ/100 derived from the analy-
A high velocity applies during the ploughing of a sis of Wroth (1984), where φ is the friction angle in

pipeline trench. In combination with the large size degrees and σvc is the consolidation stress. Changes in
of a ploughshare, this leads to undrained conditions friction angle affect both the drained and undrained
even in sands. As a consequence of the varying strengths of soil, since the underlying behaviour is
velocities involved in these processes, it is common principally frictional.
for the geotechnical analysis for pipeline design in
a fine-grained soil to require an assessment of the
2.6 Summary of soil behaviour
drained response. Conversely, in coarse-grained soils
an assessment of the undrained behaviour can be As well as the conventional aspects of soil behaviour
required. Also, there are often occasions when the that are considered in the analysis of foundations,
actual response involves partial drainage, and it is pipeline geotechnics is also often concerned with a
necessary to tie together drained and undrained assess- greater degree of soil disturbance, and intervening
ments in order to predict the most likely behaviour, and periods of recovery and reconsolidation. These pro-
the potential range of responses. cesses also take place at lower stress levels compared to
conventional geotechnics, and can lead to significant
changes in the state and therefore the strength of a soil
2.5 Low effective stress friction through the operating life of a pipeline. Also, due to
the relevant drainage distances and rates of movement,
A final feature of soil behaviour that is particularly rel- it is can be necessary to focus on the drained response
evant to pipeline geotechnics is the variation in friction of fine-grained soils and the undrained response of
angle with stress level. At the low stresses relevant to coarse-grained soils.
pipelines, higher friction angles are found compared
to more usual geotechnical stress levels. The peak fric-
tion angle of sands increases with reducing stress level 3 PIPELINE EMBEDMENT
(Bolton 1986). Results from experiments performed
on Earth (Fannin et al. 2005) and onboard the space 3.1 Pipelaying mechanics
shuttle (Sture et al. 1998) also show that the critical
state or constant volume friction angle is higher at very The as-laid embedment of a pipeline affects the sub-
low stresses. sequent pipe-soil resistance, as well as the thermal

94
Figure 12. Pipeline laying notation (Randolph and White Figure 13. Maximum stress concentration factors for pipe
2008a). laying on an elastic seabed (Randolph and White 2008a).

insulation. The installation process involves soil- Section 3.4, but firstly the significant influence of
structure interaction, since the maximum vertical dynamic pipe movements during the laying process is
pipe-soil force during the lay process will exceed highlighted.
the submerged pipe weight, W , by an amount that During J-lay or S-lay installation, dynamic move-
depends on the seabed stiffness and the geometry of ment of the pipe occurs within the touchdown zone,
the catenary created by the S-lay or J-lay arrangement. driven by the vessel motion and hydrodynamic loading
The configuration of a pipeline during laying is of the hanging pipe. These loads induce a combination
shown in Figure 12. A key parameter is the horizontal of vertical and horizontal motion of the pipeline at the
component of tension, T0 , which is constant through seabed (Lund 2000, Cathie et al. 2005). In addition to
the suspended part of the pipeline and can be assessed vessel motion due to swell and waves at the sea surface,
from the pipe weight, water depth and hang-off angle. cyclic changes in pipeline tension may occur (depend-
The maximum contact force (per unit length) with ing on the accuracy of the tensioning system) if the
the seabed, Vmax , and hence the local force concentra- offloading of the pipe is not smoothly coincident with
tion factor, flay = Vmax /W , is a function of the seabed the vessel advancement.
stiffness, k (defined as the secant ratio of force per This dynamic movement, although often of very
unit length, V, to embedment, w) in addition to the small amplitude, leads to local softening of the seabed
bending rigidity, EI, and T0 . The force concentra- sediments and can push soil away to either side of the
tion factor reduces with increasing water depth and pipe alignment, creating a narrow trench in which the
decreasing seabed stiffness. A characteristic length, pipe becomes embedded.
which relates to the length over which the bending An illustration of the significant additional embed-
stiffness moderates the catenary behaviour, is given ment that can occur simply due to small amplitude
by λ = (EI/T0 )0.5 . cyclic motions is shown in Figure 14. These results
Parametric solutions for the static lay conditions are from a centrifuge model test on lightly over-
have been presented by Randolph and White (2008a), consolidated kaolin clay (Cheuk and White 2010a). A
who showed that for horizontal tension of T0 > 3λW model pipe was penetrated to a normalised embedment
(which holds for most pipelines), results from analyti- of w/D = 0.1 (point A), when the vertical pipe-soil
cal solutions (Lenci and Callegari 2005) and numerical load was fixed constant (the normalised vertical load,
analysis using OrcaFlex (Orcina 2008) all converge to V/su D reduced with pipe embedment, due to the
unique design lines. The value of flay may be expressed increasing su with depth).
approximately as (Figure 13): A series of packets of horizontal oscillations were
then imposed, increasing in amplitude (Figure 14a).
The adopted amplitudes of motion reflect ROV obser-
vations during laying, although these are dependent
on the lay geometry and metocean conditions (West-
3.2 Seabed disturbance during pipelaying
gate et al. 2010a). The aim in this experiment was to
Equation 1 is derived based on a single value of secant represent dynamic lay motions in an idealised manner.
seabed stiffness, V/w, which would be applicable to The pipe initially settled at a rapid rate, by an amount
a purely elastic seabed. The actual seabed response that far exceeds that due to the combined vertical and
is generally non-linear during vertical penetration, horizontal loading alone.
and is stiffer during unloading since the seabed has The lateral soil resistance mobilised during the first
been plastically deformed. As a result, the actual two cycles, when the embedment doubles (to point
operative secant stiffness varies along the touchdown B), corresponds to an equivalent friction factor of
zone. Theoretical solutions for monotonic vertical H/V < 0.25. As the embedment increases, a greater
pipe penetration into undrained soil are described in soil resistance is mobilised for a given amplitude of

95
motion, reflecting the increasing constraint on the
pipeline. The softening of the surrounding soil is evi-
dent in the reductions in lateral resistance at points C
and D, with increasing disturbance.
Complementary assessments of the level of soil
remoulding during pipeline laying can be made using
large deformation finite element analysis. A study
reported by Wang et al. (2009) replicated the first
stage of the model test shown in Figure 14 (with hori-
zontal motions of +/−0.05D) using continuum finite
element analysis. The soil constitutive model included
softening through a reduction of the undrained strength
with accumulated plastic strain, in a manner consistent
with the behaviour shown in Figure 5.
The resulting patterns of lateral resistance and
embedment and the local soil remoulding are shown
in Figure 15. Even small lateral motions of just
+/−0.05D lead to a remoulded zone that extends by
almost one pipe diameter to each side, and the pipe
itself rests on fully remoulded soil.
These results highlight the importance of assessing
pipeline embedment using an appropriately degraded
value of soil strength, as well as accounting for the
catenary overstress via Equation 1 or some comparable
approach.

3.3 As-laid pipeline survey observations

Similar conclusions can be drawn from the results of
ROV surveys following pipe laying. The variation in
embedment along a pipeline varies for a variety of rea-
sons including local variations in the seabed strength.
However, consistent variations have also been identi-
fied due to wave height, lay rate, downtime events, and
changes in lay angle.
The first three of these effects influence the level
of dynamic motion that the pipe is subjected to in the
touchdown zone and the latter effect influences the
vertical stress concentration (Westgate et al. 2010a).
The resulting range of embedment can be expressed
as a statistical variation, and this range can be
compared with calculations performed using theo-
retical solutions for the catenary overstress and the Figure 14. Effect of lateral motions on pipeline embedment
(after Cheuk and White 2010a).
seabed penetration resistance, using both intact and
remoulded soil strengths.
Figure 16 shows the distribution of embedment
from survey measurements taken at 1 m intervals along matches well with the most frequent embedment,
a 13 km long pipeline (excluding short lengths of although the agreement was not as good for some other
pipeline that had a far greater embedment due to down- pipelines at the same site.
time events). The pipeline was laid on almost uniform This result and other comparisons from a limited
deepwater soil conditions, in a water depth of 1215– range of post-installation surveys, suggest that the
1450 m (Westgate et al. 2010b). The lay process took fully remoulded soil strength leads to reasonable esti-
several days, during which the sea state varied with a mates of the average pipe embedment for average lay
significant wave height of between 0.6 m and 1.7 m. conditions. However, there remains significant scatter
The calculated pipeline embedment based on the between different pipelines in the same conditions and
static catenary overstress (Equation 1) coupled with variations in embedment along a single pipeline (West-
intact and fully remoulded strengths are highlighted, gate et al. 2010a, Westgate et al. 2010b). An additional
along with estimates based on a dynamic analy- effect appears to be that lighter pipelines are more sus-
sis of the vertical stress concentration performed ceptible to dynamic lay effects – probably due to their
using Orcaflex (Orcina 2008) (Figure 16). The fully reduced inertia – and so a greater degradation in soil
remoulded strength coupled with the static overstress strength applies.

96
Figure 15. LDFE simulation of soil strength after dynamic pipe laying (after Wang et al. 2009).

breakout resistance since it is not immediately obvi-

ous what combination of input parameters will lead to
upper and lower bound outcomes.
It is also attractive to set the geotechnical compo-
nents of pipeline analysis within a full probabilistic
framework. This is consistent with the probabilistic
structural reliability analyses that are increasingly per-
formed during the assessment of pipeline on-bottom
stability or lateral buckling.

3.4 Solutions for vertical pipe penetration

Using an appropriate operative soil strength, su ,
Figure 16. Distribution of as-laid embedment for a single the shallow embedment of a pipeline in undrained
pipeline (Westgate et al. 2010b). conditions can be calculated via a bearing capacity
expression that reflects the appropriate geometry of
failure mechanism. Experimental results presented by
The use of the remoulded soil strength over-predicts Dingle et al. (2008) revealed the internal soil fail-
the embedment in the case of minimal pipeline ure patterns during pipe penetration (Figure 17a).
motions (for example in calm weather or during These detailed observations, coupled with the vertical
lay down of the final catenary section of pipe) and force-displacement response from other tests, have
under-predicts embedment during severe weather or been used to validate numerical simulations (e.g. Fig-
downtime events, again based on limited field data ure 17b, Merifield et al. 2008a) and plasticity limit
(Westgate et al. 2009; Westgate et al. 2010b). analyses (Randolph and White 2008b).
Due to this inevitable scatter, including variability The resulting bearing capacity, Vult , can be cal-
due to effects that cannot be predicted in advance of culated as the superposition of components related
the pipelaying, assessments of pipeline embedment are to the soil strength at the pipe invert (via a bear-
subject to uncertainty. If only an upper or lower bound ing factor, Nc , that varies with embedment, pipe
embedment is required for design then this uncertainty roughness and strength heterogeneity) and the soil
can be circumvented using conservative assumptions. buoyancy – enhanced by heave (Randolph and White
However, both upper and lower bound embedments are 2008a, Merifield et al. 2009):
usually critical for different design considerations.
It is therefore necessary to perform lower and upper
bound assessments using opposite extremes of the
input parameters. These can be extended to a full
probabilistic assessment, using statistical variations in
input parameters, as illustrated in Section 4.4. Such an The nominally embedded cross-sectional area is
approach is desirable when assessing the subsequent denoted A , the soil submerged unit weight is γ and

97
Figure 19. Undrained failure envelopes for lateral breakout
in uniform soil.

4 LATERAL PIPE-SOIL INTERACTION

4.1 Theoretical failure envelopes

The same combination of experimental, numerical and
analytical techniques has been used to derive fail-
ure envelopes in vertical-horizontal load space, which
allow the lateral breakout resistance in undrained con-
ditions to be assessed (Dingle et al. 2008; Merifield
et al. 2008a, Randolph and White 2008b). Differ-
ent forms of failure mechanism apply, depending on
Figure 17. Failure mechanisms during vertical pipe pene- whether tension can be sustained at the rear of the
tration (Dingle et al. 2008, Merifield et al. 2008a). pipeline (Cheuk et al. 2008; Merifield et al. 2008b)
as shown in Figure 19. For the no-tension case (and
w/D < 0.5), the failure envelopes pass through the V-H
origin, and there is a ‘frictional’ cut-off corresponding
to a failure mechanism involving the pipe riding up at
an angle shown as θ in Figure 19, with no soil defor-
mation occurring. In drained conditions, experimental
results show that the failure envelope has a similar
shape to the no-tension undrained case (Zhang et al.
2002a).
Centrifuge modelling results from tests on soft clay
show that the breakout resistance is well predicted by
the unbonded failure envelopes if the pipe is installed
with some level of dynamic movement during laying
Figure 18. Bearing factors during vertical penetration into (such as the simulation in Figure 14) (Cheuk and White
heterogeneous and uniform soil (Chatterjee et al. 2010b). 2010b). In this case, for very soft clay, the appropri-
ate undrained strength appeared to be the intact value.
In this particular case the effects of remoulding and
reconsolidation approximately cancel out.
the factor fb captures the enhancement of soil buoy- Without dynamic movement the breakout resistance
ancy beyond that given by Archimedes’ principle. A is often higher, but is highly brittle, reflecting the open-
value of fb = 1.5 is typical. ing of a crack behind the pipe prior to mobilisation of
Solutions for the bearing factor, Nc , in this expres- a full two-sided mechanism (Dingle et al. 2008). It is
sion have been provided for uniform and linear soil thought that any dynamic movement leaves a skin of
strength profiles (Aubeny et al. 2005, Merifield et al weak remoulded soil at the pipe surface. This can fail
2008a), and with modifications to account for heave in a local mechanism behind the pipe, even if some ten-
(Merifield et al. 2009, Chatterjee et al. 2010a, 2010b). sion is sustained, providing negligible extra resistance
A pair of bearing factor profiles derived from large above the unbonded case.
deformation finite element analyses are shown in Fig- Modified versions of these failure envelopes can be
ure 18, for two different profiles of soil strength created for other boundary conditions, such as a pipe
(Chatterjee et al. 2010b). running parallel to a slope (e.g. Morrow and Bransby

98
Figure 21. Effect of load path on equivalent breakout
friction.

load path reaches the unbonded failure point marked

Figure 20. Equivalent friction factors from failure A, with the equivalent friction factor marked µeq . The
envelopes. load path marked B represents hydrodynamic load-
ing, when the pipe is subjected concurrently to both
2009). For this case the failure envelope is rotated by horizontal drag and upwards lift. In this case the equiv-
the slope angle and changes shape slightly, depending alent friction factor at failure is higher (although the
on the ratio of soil strength to weight. Due to the curved constant V path marked A provides a conservative
shape of the failure envelopes, the equivalent friction assessment).
factor for lateral failure in a downslope direction is not The theoretical failure envelopes of the form shown
simply the value for level ground, adjusted by the slope in Figure 19 can be re-expressed as equivalent fric-
angle according to Coulomb friction (unless failure is tion factors, varying with load path and load level,
via the tangential mechanism). W /Vult . For illustration, results have been derived
from the unbonded envelopes corresponding to a
smooth pipe interface, uniform soil strength and
4.2 Failure envelopes as friction factors
embedments of w/D = 0.25 and 0.45. Load paths vary-
Failure envelopes depict the limiting pipe-soil loads in ing from dV/dH = 0 to dV/dH =−2 (i.e. a lift force of
a different manner to the empirical expressions that are twice the drag) have been used (Figure 20).
commonly used in practice (e.g. DNV 2007a, Verley This form of presentation clarifies how much the
and Sotberg 1994, Verley and Lund 1995, Bruton et al. seabed response differs from the simple Coulomb fric-
2006). These approaches divide the soil resistance into tion that is often assumed. As the pipe weight reduces,
frictional and passive resistance components that are the equivalent friction factor rises, particular at low
superposed. It is then common for the resulting resis- load levels.
tance to be expressed as an equivalent friction factor, The effect of load path is shown in Figure 21 which
by dividing by the vertical pipe-soil load. In this way, compares µeq for two ratios of dV/dH, using the
the geotechnical resistance is linked only to the pipe pure drag case of dV/dH = 0 for normalisation. The
weight (and lift), and any structural analysis software load path has a significant influence on the equiv-
need not include any soil parameters, nor incorporate alent friction factor at breakout. For the load path
explicit calculations of the pipe embedment. with an uplift of dV/dH = −2, the equivalent friction
The theoretical failure envelopes can also be manip- at failure is typically 1.5-2 times higher than pure
ulated to express the pipe breakout resistance as an drag at w/D = 0.25. The difference rises to >2.5 for
equivalent friction factor, µeq . For a given initial state, w/D = 0.45. At low loads, however, the load path does
the equivalent friction at failure, µeq = H/V depends not affect the equivalent breakout friction because the
on the load path in H-V space. The initial state is typi- same tangential mechanism applies.
cally well within the failure envelope (W /Vult << 1) This limiting friction factor at low loads that is
due to the unloading beyond the maximum overstress caused by the tangential mechanism is not captured
of the catenary, dynamic lay effects and also from any in conventional ‘friction + passive’ models for break-
gain in soil strength after laying. A general parame- out resistance. The component of passive resistance in
ter to describe the current level of unloading is the these models is uninfluenced by the vertical load level,
overloading ratio, R = Vult /V, where V is the current leading to a finite breakout resistance under a verti-
vertical load and Vult is the current vertical bearing cal load of zero. This is incorrect and unconservative
capacity (Zhang et al. 2002a; Zhang and Erbrich 2005). for stability design. The tangential failure mechanism
In lateral buckling analysis, the load path is gener- leads to zero breakout resistance under zero vertical
ally assumed to involve horizontal loading, whilst the load for any embedment of w/D < 0.5, if tension can-
vertical load remains constant and equal to the pipe not be sustained at the pipe-soil interface (regardless
submerged weight (i.e. dV/dH = 0). In Figure 19 this of the pipe roughness).

99
4.3 Coupling of geotechnical analysis and Table 2. Deterministic parameters for breakout assessment.
structural and hydrodynamic responses
Parameter Value
To properly capture this strong influence of load path
in an assessment of pipeline stability it is necessary to Pipeline diameter 0.5 m
couple the failure envelopes with the imposed hydro- Submerged pipe weight, W 1.5 kN/m
dynamic loads. As a minimum, the load path of the Specific gravity, SG 1.38
critical wave condition can be combined with the fail- Touchdown lay factor, klay 1.25
ure envelopes to derive a single value of equivalent Soil submerged unit weight, γ 8 kN/m3
friction factor that is then used in an absolute stability Buoyancy factor, fb 1.5
check. The conservative approach of assuming a load
path of dV/dH = 0 will provide a safe under-prediction
of the available breakout resistance (Figure 19). Adop- Table 3. Probabilistic parameters for breakout assessment.
tion of the actual load path when assessing µeq will
Statistical
reduce this conservatism. values
To provide information beyond simply the breakout
resistance, the failure envelopes can be incorporated Parameter P5 P95
within a plasticity macroelement model, which allows
the full load-displacement response to be simulated Operative soil Gradient, ksu (kPa/m) 1.5 3
(e.g. Schotman and Stork 1987, Zhang et al. 2002b, strength profile Dimensionless gradient,
Cocchetti et al. 2009). In essence, these models allow during laying κ = ksu D/sum 20
the pipe embedment and therefore the size and shape Operative soil Mudline value, sum (kPa) 2 4
of the failure envelope to be updated whenever the load strength profile Dimensionless gradient,
reaches the current failure envelope and displaces. If during breakout κ = ksu D/sum 0
the load remains within the failure envelope the pipe
is stable.
The most simple use of this type of model in design parameter being given a statistical description. Such
is to track the response of a single pipe element whilst an approach is consistent with the reliability-based
it is subjected to the most severe hydrodynamic con- approaches being used for the structural aspects of
ditions, unsupported by any neighbouring pipe (which pipeline design (e.g. Carr et al. 2004, DNV 2007a,
is a conservative simplification). AtkinsBoreas 2008, Rathbone et al. 2008).
A more sophisticated approach is to attach multi- The recently-developed solutions for pipeline
ple pipe-soil elements to a full structural model of embedment and combined V-H capacity provide a
the pipeline, allowing longitudinal load-shedding to robust theoretical framework within which a proba-
be incorporated. This approach is described by Tian bilistic assessment of the pipeline response can be
and Cassidy (2008) who incorporated the Zhang et al made.This is illustrated by an example of a 0.5 m diam-
(2002b) model for drained pipe-soil interaction into a eter smooth pipeline with the deterministic parameters
structural analysis of a pipeline subjected to hydro- shown in Table 2. The soil strength profiles relevant
dynamic loading. This allowed full wave-pipe-soil to the embedment and the breakout events are given a
interaction to be simulated using a more sophisti- probabilistic description, as set out in Table 3. For sim-
cated pipe-soil model than had been previously used plicity, only the soil strength profiles are treated prob-
(an inevitable limitation being that the wave-seabed abilistically, although it is straightforward to include
interaction depicted in Figure 3 is not captured). other parameters that have natural variability, such as
the soil unit weight.
The adopted values are typical for a flowline being
4.4 Quantifying uncertainty in lateral breakout
laid on a soft fine-grained soil in deep water. The oper-
resistance: Monte Carlo analysis
ative soil strength profile, representing the disturbed
As discussed in Section 3.3, the as-laid embedment state during laying, increases with depth and has a
of a pipeline is strongly influenced by the dynamic small mudline intercept. The operative value during
effects in the touchdown zone during laying and the breakout is uniform and higher. The upper bound (UB)
consequent reduction in soil strength. These processes and lower bound (LB) parameters for each strength
depend on the laying procedures and the metocean profile differ by a factor of two and are assigned 5%
conditions, which are not known during design. Sim- and 95% exceedence values (i.e. the LB P5 value has a
ilarly, the operative soil strength during breakout is 5% probability of exceeding the actual soil strength).
subject to uncertainty, partly from natural variability This ratio between UB and LB values is broadly typi-
and partly due to potential reconsolidation of the soil cal of practice, but it is strongly affected by the level of
near the pipe after lay-induced disturbance. natural variability and the spatial intensity and quality
Despite these difficulties it is necessary to provide of site investigation data.
some assessment of the pipeline embedment and the The probabilistic soil strength parameters might be
consequent breakout resistance during design. This derived from in situ and ex situ testing on samples dis-
assessment can be performed within a probabilistic tributed along the entire pipeline length, which will
framework, with the uncertainty in each input then capture scatter related to the natural variability

100
Figure 22. Strength distributions in probabilistic analysis
case.
(as well as scatter from the tests themselves). It is
necessary also to include within these soil strengths
some uncertainty to capture the lay process, and any
reconsolidation.
Having chosen P5 (LB) and P95 (UB) values of soil
strength (or any other pair with a specified likelihood,
representing upper and lower bounds) a probabilistic
distribution must be fitted. In this case a gamma dis-
tribution has been chosen because it does not feature
negative values (since negative values of strength are
not possible). A gamma distribution is also free of the
long positive tail that is present in a lognormal distri- Figure 23. Variation in as-laid embedment for example
bution (which is sometimes adopted for non-negative probabilistic analysis.
variables). The resulting cumulative distribution func-
tions for each soil strength parameter are shown in
Figure 22. outputs relative to the inputs due to the particular form
Using these distributions coupled with the embed- of the mechanisms that govern pipe embedment.
ment and failure envelope analyses described earlier, a The next step in the analysis is the calculation
Monte Carlo analysis has been performed using a soft- of breakout resistance, which is expressed here as
ware program called MCWHIPLASH, developed by an equivalent breakout friction factor, Hbrk /V. In this
Advanced Geomechanics, to illustrate the influence of example, the load path to breakout has been assumed
soil strength uncertainty. The analysis involved 10,000 to be purely horizontal loading under a maintained ver-
realisations, which is sufficient to generate smooth tical load corresponding to the submerged pipe weight
distributions of the outputs. (V = W ), which is the usual assumption in a lateral
The first step of the calculation involves estimation buckling analysis.
of the as-laid embedment. The probabilistic operative The breakout resistance is affected by both the as-
strength leads to the variation in embedment shown in laid embedment and also the (probabilistic) strength
Figure 23. The as-laid embedment varies monotoni- at the time of breakout, which are assumed to be
cally with the single variable – the operative strength uncorrelated in this example, for simplicity.
during laying – in a non-linear manner. Although the The variation in breakout friction factor with as-laid
P95 and P5 strength parameters differ by a factor of 2, embedment is shown in Figure 24a and the variation
the ratio between the P95 and P5 embedments is only with soil strength at breakout is shown in Figure 24b.
1.38 (Figure 23b). This reflects the non-linearity of the The resulting distribution of friction factor is shown
bearing capacity expression (Equation 2). A halving of in Figure 24c. The overloading ratio, R = Vult /V, var-
the soil strength, for example, does not lead to a dou- ied between 2.3 and 9.5 at breakout in this example,
bling of the embedment because the bearing capacity with a mean value of 4.9. Consequently, none of the
factor rises with depth. Monte Carlo realisations involved failure via the tan-
Also, the strength of the soil is not the only mech- gential cut-off mechanism (indicated by the cap on
anism supporting the pipeline – the soil buoyancy has Hbrk /V for w/D = 0.25 in Figure 20). A trend line
an influence and this is not given a probabilistic vari- through the centre of the clouds of results in Figure
ation in this example. In practice, the soil unit weight 24a and Figure 24b indicates that the breakout friction
has a narrow range of variability, so the soil buoyancy factor increases approximately linearly with both soil
term generally attenuates the variability arising from strength and embedment (although this is not the case
the soil strength. Overall, there is a narrowing of the for other examples that could have been adopted).

101
Figure 25. Normalised variability of input and output
parameters in example probabilistic analysis.

1.54.As for the embedment, this variability is narrower

than the range of input strengths. The cumulative
distributions for the strengths and the resulting embed-
ment and breakout friction factor are compared in
Figure 25, with each variable being normalised by
its P50 value. This form of presentation highlights the
relative uncertainty in each parameter.
As noted previously, the analysis leads to a narrower
output range than input range: a given uncertainty in
soil strength leads to a lower uncertainty in breakout
friction factor. This is a useful outcome, given the dif-
ficulty in assessing accurately the soil strength close
to the seabed. This effect can be contrasted with the
response of a surface foundation (with or without short
skirts). The lateral capacity of the foundation varies
in proportion to the surface soil strength, so the soil
strength and the foundation capacity have the same
level of variability.
The deterministic results based on combinations of
the P5 (LB) and P95 (UB) values are also marked on
Figure 24a and Figure 24b. These results highlight
the value of a Monte Carlo analysis. The friction fac-
tors found using the extreme values of soil strength
in combination (i.e. LB/LB and UB/UB) lie within
the centre of the distributions (at the P34 and P55 per-
centiles respectively). These do not provide a useful
indication of the potential variability.
The friction factors calculated using the more
unlikely assumptions of the UB/LB and LB/UB soil
strength combinations give breakout values that have
percentiles of P1 and P99 . The (un)likelihood of the
results found using these assumptions is also strongly
dependent on the particular combination of pipe and
soil parameters being used, due to the non-linearity of
the governing mechanisms. There is therefore no sim-
ple method for linking the probability of the inputs to
the probability of the outputs, without a Monte Carlo
Figure 24. Variation in breakout friction factor for example analysis or a similar statistical technique.
probabilistic analysis. If it is assumed that the operative strengths during
laying and breakout are correlated, it might be assumed
For this example, the P5 and P95 values of Hbrk /V that the LB/LB and UB/UB cases would then provide
are 0.94 and 1.45 respectively, which have a ratio of a true indication of the potential variability in Hbrk /V.

102
However, this is not the case. In fact, for the example
shown here, with fully correlated strengths, the high-
est value of Hbrk /V is found for an analysis based on
the P31 strength during laying and breakout. The max-
imum breakout resistance does not arise from either
the weakest or strongest soil, but an intermediate case.
Why does this surprising observation arise? Soft
soil gives high embedment, and therefore a high wall
of passive soil on breakout. However, the beneficial
embedment is countered by the weakness of the pas-
sively loaded soil. Due to these counteracting effects
coupled with the non-linear shape of the V-H failure
envelope (and its non-linear growth with embedment),
the extreme values of breakout resistance are often not
associated with the extreme values of soil strength.
This same interaction underlies the narrower range of
breakout friction factors compared to the input range
of soil strengths.
The form of Monte Carlo analysis shown here
provides a solution to this difficulty, capturing the
uncertainties associated with soil strength and other
parameters that influence pipe embedment and break-
out. The simple example shown here includes only the
variability associated with soil strength. In practice the
other geotechnical inputs such as the soil unit weight
can also be treated probabilistically. It is also possible,
using MCWHIPLASH, to incorporate other parame-
ters such as the pipeline bottom tension during laying,
via the solution given in Equation 1.
The uncertainty associated with the geotechnical
parameters arises from the site variability and the
quantity and quality of site characterisation data. To Figure 26. Deformation mechanisms during large ampli-
provide a consistent approach throughout the entire tude lateral sweeping of a pipe on soft clay.
analysis process – from interpretation of the geotechni-
cal SI data through to the pipeline structural analysis –
pipe rests on the seabed. It is therefore necessary
a probabilistic approach can also be used to assess
to assess the lateral pipe-soil resistance over long
the soil strength parameters (e.g. Lumb 1966, DNV
distances of lateral movement across the seabed.
2007b, Lacasse et al. 2007). However, these methods
Flowlines are typically designed for several hun-
do require a sufficient quantity of site characterisa-
dred startup and shutdown events throughout the field
tion data, which is often not the case. The uncertainty
life. During the early years of operation these will be
related to the changes in soil strength during laying
more frequent. In later years, as the operating tempera-
and recovery can be reduced through model tests that
ture and pressure reduce, the feed-in to the buckle may
simulate these processes (White and Gaudin 2008).
reduce. However, if new wells are tied-in, or if the field
recovery is enhanced, then the length of feed-in into
a buckle, and the consequent amplitude of movement,
4.5 Large amplitude monotonic lateral response
may increase in later life.
The engineered buckles that are often used to relieve As discussed previously, the overloading ratio
thermal and pressure-induced expansions typically of a pipeline during operation is typically greater
involve lateral movements of several pipe diameters, than 2, which puts the load point on the ‘dry’
as is evident in the picture within Table 1. This exam- side of the unbonded breakout failure envelopes
ple is a buckle that was initiated over a sleeper, which (V/Vult < ∼0.45). In this case the pipe moves upwards
provides a vertical upset and therefore reduces the at failure, leading to a reduction in resistance after
buckling load locally, providing a reliable buckle ini- breakout as the pipe rises towards the ground sur-
tiation point. Only the central part of the buckle is face. Model tests using PIV image analysis have shown
lifted away from the seabed, and a significant length that in soft clay the steady failure mechanism involves
of grounded pipeline sweeps back and forth across basal sliding of a berm of soil ahead of the pipe, with a
the seabed as the buckle expands and contracts dur- thin layer of additional material being ploughed from
ing operating cycles. Buckles are sometimes initiated the seabed (Figure 26a). The same mechanism has
on-bottom at lay route curves, and designers always been replicated in large deformation finite element
need to consider the likelihood of a ‘rogue’ buckle analysis (Figure 26b). The subsequent steady ‘resid-
being initiated anywhere along the route, where the ual’ resistance depends on the size of the berm and

103
the strength of the soil within it (as well as the seabed
strength). The size of the berm depends on the initial
embedment of the pipe and its trajectory, whilst the
strength within the berm depends on the remoulding
it has undergone, and the sensitivity of the soil.
At a given point in a large amplitude lateral sweep,
these two effects can be quantified by defining an
‘effective embedment’, w /D, which amalgamates the
embedment below the original soil surface, w/D, and
an additional component of embedment arising from
the berm, hberm /D (Figure 26c). The volume of the
berm is calculated from the area swept by the pipe
invert. By assuming that the berm has a particular
aspect ratio, η = length/height, its height can be found
from the area. This height is discounted by a sensitivity,
St,berm , which represents the reduction in berm strength
due to remoulding, to give the effective additional
embedment, hberm /D:

This normalisation approach provided excellent

agreement between a set of centrifuge model tests on
soft clay, simulating pipes of different weights and
initial embedments (White and Dingle 2010). The
tests all showed differing values of steady residual
lateral resistance, ranging from Hres /V = 0.32 to 0.65
(Figure 27a, b). There was no clear influence of simu-
lated pipe weight, V, but a scattered trend with initial
embedment, (w/D)init . However, the profiles of nor-
malised lateral resistance, H/su D, when plotted against
effective embedment, match closely regardless of the
vertical load (Figure 27c). All of the responses lie close
to a power law relationship between H/su D and w /D.
The same behaviour is seen in large deformation finite
element analysis (Wang et al. 2010).
This effective embedment approach alone does not
provide a predictive tool, only a robust normalisa-
tion of the response. A method for calculating the
pipe trajectory is required, in order that the evolving
size of the soil berm and the local embedment can be
assessed. This trajectory is affected by the pipe weight,
even though the resistance at a given local embedment
appears not to be.
This missing link has been filled by project-specific
empirical models, and is currently being tackled within
the SAFEBUCK JIP, via the development of a plastic-
ity macroelement model. This model will operate in
the usual macroelement manner of combining yield
envelopes, a hardening law and a flow rule to describe
the general force-displacement response of an element
of pipe (following Zhang et al. 2002b). However, it will
incorporate the influence of the changing soil strength
and berm geometry on the shape of the failure enve-
lope. This will allow the macroelement approach to be
extended to large amplitude movements.
In the meantime, various empirical expressions
have been proposed to estimate the residual lateral Figure 27. Centrifuge modelling observations of resid-
resistance with reference only to the as-laid state (not ual lateral resistance on soft clay (White and Dingle
the post-breakout trajectory) (e.g. Bruton et al. 2006, 2010).

104
Cardoso and Silviera 2010). Based on the six tests on
soft kaolin clay shown in Figure 27, White and Din-
gle (2010) proposed the a simple relationship between
initial embedment, (w/D)init , overloading ratio, R, and
residual lateral friction factor, Hres /V:

with suggested parameters of µ = 0.3 and k = 2

(Figure 27d).
The form of Equation 4 encapsulates some of the
factors that influence the residual resistance: a high
(w/D)init leads to a large initial berm size and a low
overloading ratio, R, leads to a deeper pipe trajec-
tory; both of which increase the passive horizontal
resistance. An obvious limitation of this tentative Figure 28. Typical full scale modelling observations of
expression is that it does not include any soil parame- cyclic lateral response on soft clay (White and Cheuk 2008).
ters, so does not explicitly capture the varying strength
as soil is remoulded ahead of the pipe, or the variation
in soil strength with depth.
Surprisingly, this empirical expression has a differ-
ent form to another one derived by a recent confidential
study for the SAFEBUCK JIP (White and Cheuk
2009), and also from that published by Cardoso and
Silviera (2010). In all cases these expressions are based
on a particular collection of model tests, rather than
any theoretical mechanism. Such contrasts provide a
warning that these empirical expressions may not be as
accurate when applied beyond the conditions in which
they were calibrated.

4.6 Large amplitude cyclic lateral response

During large amplitude cyclic lateral movements, the
pipe-soil resistance remains dependent on the berm
material being swept back and forth. Each time the
pipe changes direction the berm is left behind, and is
remobilised if the pipe approaches that point during a
subsequent cycle.
These aspects of behaviour are illustrated by a
typical model test involving cycles of large ampli-
tude lateral movement, between fixed displacement
limits (Figure 28, White and Cheuk 2008). This
test involved cycles of fixed lateral amplitude under
constant simulated pipe weight.
The general form of the response involves initial
breakout of the pipe (point A in Figure 28) followed
by a gentle increase in resistance associated with the
growth of a small ‘active’ berm ahead of the pipe (B).
On reversal of the sweeping direction, this response
is repeated (C) and the berm generated during the Figure 29. Soil berms during large amplitude lateral
previous sweep is left behind, becoming ‘dormant’. movement.
When the pipe again approaches this point during a
later sweep, an increase in resistance is experienced
as the dormant berm is collected (D). With repeated pipe. An image of berms created by this type of large
cycles of movement, the berms at the limits of the pipe amplitude pipe motion are shown in Figure 29a.
movement grow, causing a corresponding increase in In the design of lateral buckles, it is important to
resistance. The first sweep encounters slightly higher model the constraint imposed by the soil berms, in
steady resistance than later sweeps due to the larger order to provide an adequate assessment of fatigue.
active berm arising from the initial embedment of the If the pipe-soil resistance is assumed to be constant,

105
with the berms ignored, then a buckle will pro-
gressively lengthen through cycles of expansion and
contraction, which reduces the peak bending stresses
near the crown (Cardoso et al. 2006, Bruton et al.
2007). Soil berms inhibit this lengthening, causing
the high stresses generated during buckle initiation to
be locked-in. Although the restraint provided by the
berms also attenuates the amplitude of the pipe motion
within each cycle – and therefore the cyclic stresses –
the overall effect on fatigue is usually harmful, due
to the higher mean stresses that are locked-in (Bruton
et al. 2007).
Models for the cyclic large-amplitude lateral
behaviour can be based on the accumulation and
deposition of berm material, as shown schematically
in Figure 29b. The current berm size can be used
as a hardening parameter that governs the passive
resistance, rather like in Equation 4. For undrained
conditions, the rate that the berm grows with lateral
pipe movement is equal to the depth of soil scraped
away by the pipe, from conservation of volume (White
and Cheuk 2008). Re-consolidation of the soil that has
been remoulded and transported ahead of the pipe will
increase the berm resistance.
The cyclic resistance is not only affected by the
changing geometry. Depending on the soil type, pore
pressure dissipation may occur during lateral sweep-
ing, and is also likely to occur between startup and
shutdown events. This leads to reconsolidation of the Figure 30. Topography of a soft clay seabed after several
disturbed soil within the berm, and also swelling of the years of operation of a lateral buckle (Cardoso and Silviera
2010).
unloaded seabed that is exposed by the scraping action
of the pipe.
For design, the variation in lateral resistance can be
estimated from a geotechnical analysis that considers
in detail the mechanisms described above. Then, to
incorporate the results into the structural analysis of
the pipeline, they can be converted into more simple
relationships. For example, the results of a geotechni-
cal analysis can be converted into equivalent friction
factors (H/W ) for the residual and berm resistance as
a function of cycle number (Bruton et al. 2009).
Over the design life of a lateral buckle, the depth of
the trench created by the sweeping action of the pipe
can be significant. As the trench deepens, soil debris
may collect in the base, raising the residual resistance.
Soil may even flow over the crown of the pipe. The
seabed topography near the crown of an on-bottom
buckle that has been operating for several years is
shown in Figure 30 (Cardoso and Silviera 2010). The
pyramidal berms on each side of the pipe are similar in Figure 31. Accumulation of pipeline embedment during
shape to the model test observations in Figure 29. The cyclic large-amplitude sweeping: SAFEBUCK model test
database (White and Cheuk 2009).
pipe itself is almost hidden from view, being partly
buried under softer soil that has accumulated within
the trench. in Figure 31. These results are from a database
The situation evident in Figure 30 is analogous of large-amplitude cyclic lateral tests collated by
to the cyclic rod penetrometer results shown in the SAFEBUCK JIP. These tests involved soft fine-
Figure 8. Repeated remoulding and reconsolidation of grained model seabeds.
the seabed will tend to increase the soil strength and Short elements of pipe were modelled, exerting
consequently increase the constraint on the pipe. a constant bearing pressure in the range V/D = 0.5–
The accumulation of pipe embedment during a 8 kPa. Fixed-amplitude lateral cycles of 1–10 pipe
set of full-scale and centrifuge model tests is shown diameters were imposed. The pipes embedded at rates

106
of 1–50% of the pipe diameter per cycle and there was walking per cycle can increase rapidly as the available
no tendency for a stable embedment to be reached. It axial resistance decreases (Bruton et al. 2007).
is worth noting that these tests involved up to only 60 Compared to the lateral behaviour, the axial pipe-
cycles of movement, whereas most pipeline systems soil response involves a more simple geometry, with
are designed for an order of magnitude more startup failure being constrained to occur at, or close to, the
and shutdown events. pipe-soil surface. By analogy with pile design, both α
An effect not accounted for in these tests was the (total stress) and β (effective stress) approaches can be
longitudinal flexural stiffness of the pipe. In all tests considered. The β-method expression for the ultimate
the simulated pipe weight was held constant. In prac- axial resistance per unit length, T, is:
tice, any vertical movement of the pipe down into the
trench can lead to a reduction in the vertical pipe-soil
contact force, with load being transferred longitudi-
nally to adjacent sections of pipe. This in turn will where µ is the pipe-soil friction coefficient, which can
attenuate the rate at which the pipe embeds. be alternatively expressed in terms of a pipe-soil fric-
This soil-structure interaction is difficult to capture tion angle, δ, where µ = tan δ. The parameter ζ is a
in design without a pipe-soil interaction model that factor that accounts for the enhancement of the nor-
simulates the trajectory of the pipe. The macroele- mal pipe-soil contact force due to a ‘wedging’ action
ment plasticity models (e.g. Zhang et al. 2002b; Tian (White and Randolph 2007).
and Cassidy 2008) have this capability, but have not The α-method expression is
yet been extended to capture large deformation effects
related to soil berms and remoulding.
In cases where the pipe settlement has been consid-
ered important to capture, some recent projects have
where αsu is the shear stress acting on the pipe sur-
used a cycle-by-cycle approach in which the trench
face at failure and DθD is the pipe-soil contact length
geometry is updated after each startup and shutdown,
around the pipe perimeter, with θD being the angle
based on a geotechnical analysis of the pipe trajec-
subtended by radii to the limits of the pipe-soil con-
tory that is performed independent of the structural
tact. The adhesion factor, α, captures the roughness of
analysis (Bruton et al. 2009). This approach allows the
the surface (i.e. the relative strength of pipe-soil and
changing vertical pipe-soil contact force to be captured
soil-soil shearing). Also, if su is taken as the in situ
through a structural analysis of many operating cycles,
soil strength then α also encompasses any differences
albeit in a manner that requires significant user inter-
between that strength and the strength of the soil at
vention. Given the settlement implied by extrapolation
the pipe surface, due to downdrag, and the processes
of Figure 31 to the typical design number of operating
of remoulding and reconsolidation since the pipe was
cycles, it is likely that some longitudinal load shedding
laid.
will occur during the design life of most on-bottom
The argument in support of β-methods is rooted –
lateral buckles.
perhaps idealistically – in the fundamental concept that
Centrifuge and large-scale model testing is cur-
soil strength is controlled by effective stress friction. In
rently a widely-used technique to aid the assessment
addition, pipeline movements can often be so slow that
of cyclic large-amplitude lateral pipe-soil resistance
drained conditions prevail even in fine-grained soils.
(White and Gaudin 2008; Langford et al. 2007). In the
However, excess pore pressures are generated at even
long term, it is envisaged that analysis techniques for
modest speeds in some clays. For this situation, the
this behaviour will become more well-established, and
β-method can be adjusted by using an excess pore pres-
model testing will not be required as often. However,
sure ratio, ru = u/σn,av , based on the average excess
given the complexity of the mechanisms shown here,
pore pressure (u) and total stress (σn,av ) around the
and the difficulty of predicting the behaviour based on
pipe surface: Equation 5 is then multiplied by (1 − ru ).
conventional soil parameters alone, it is currently com-
If the response is fully undrained it is instead tempting
mon to perform project-specific model tests to assess
to use an analysis based on the relevant undrained shear
appropriate cyclic pipe-soil model parameters.
strength. Estimation of the reconsolidated su beneath
the pipe might be thought easier than assessing ru ,
although the two types of calculation are essentially
5 AXIAL PIPE-SOIL INTERACTION
interchangeable and based on the same principles.
As a pipe penetrates the seabed, the surficial soil
5.1 Effective stress and total stress models
is dragged downwards. As a consequence, the pipe is
The axial resistance between an on-bottom pipeline bearing not on soil from below the surface, but on soil
and the seabed affects the feed-in response towards that was previously at or very close to the surface –
lateral buckles and also controls the end expansions. In as illustrated by the experimental data (Dingle et al.
addition, it influences the ‘walking’ behaviour through 2008) and numerical results (Zhou et al. 2008) shown
cycles of startup and shutdown (Carr et al. 2006). Low in Figure 32. Also, the excess pore pressure generated
axial friction can be particularly problematic, and the during the undrained penetration process will take time
relationship between pipeline walking rate and axial to dissipate. Solutions for the dissipation of excess
pipe-soil resistance is generally non-linear. The rate of pore pressure around a pipe based on an elastic soil

107
Figure 32. Seabed distortion after vertical pipe penetration
(Dingle et al. 2008 (left) vs. Zhou et al. 2008 (right)).

Figure 34. Interface shearing: kaolin clay and Storæbelt

clay till.

related to the sliding capacity of shallow foundations

supporting the Storebælt Crossing in Denmark, but
the results provide useful insights relevant to axial
pipe-soil interaction.
The responses of both the kaolin clay and also the
Storebælt clay till are very similar, in terms of both the
limiting drained and undrained resistances and also
Figure 33. Post-laying pore pressure dissipation around the velocity range over which this transition occurs.
pipe perimeter (Gourvenec and White 2010). A hyperbolic relationship provides a convenient fit to
the response. However, these results are not generally
model are presented by Krost et al. (2010) and Gour- applicable to all soils. Other proprietary tests show the
venec and White (2010). Figure 33 shows the variation drained-undrained transition at a velocity that is two
in normalised excess pore pressure, averaged around orders of magnitude different, reflecting the particular
pipe, with dimensionless time. characteristics of the soil.
These dissipation curves allow the level of consoli- It is not immediately obvious how to normalise the
dation at the pipe-soil interface prior to any in-service drained-undrained transition for shearing on a plane.
movements to be assessed – which is analogous to The soil permeability, k, is a relevant parameter, since
the ‘set-up’ of a driven pile. The pipe weight often it controls the rate of pore water flow and excess
applies a bearing pressure that exceeds the apparent pore pressure dissipation, but it can be argued that
pre-consolidation pressure of the seabed, so the soil the stiffness may not be: the volume of pore water
strength will generally increase during dissipation of to be expelled may depend principally on the shear
the excess pore pressure created by the laying of the zone thickness (and its potential to dilate or contract),
pipe. rather than the bulk stiffness of the soil. This argument
A final complication is that if the response is is pursued by Palmer (1999) in his analysis of plough-
undrained, then the state of the soil beneath the pipe ing and is discussed in Section 6. If the soil stiffness
will inevitably change through the hundreds of oper- is not relevant then the coefficient of consolidation is
ating cycles, and therefore hundreds of episodes of not required within the normalisation, unlike the cases
failure followed by reconsolidation. The tendency for shown in Figure 9.
excess pore pressure to be generated during axial It remains to be established whether the appropri-
movement will be reduced, as described below. ate dimensionless group to normalise the velocities
in Figure 34 is vL/cv (with L being some relevant
drainage distance) or v/k. In practice, it can be more
5.2 Interface shearing and drainage
efficient to perform multiple interface shear tests at
The response of a fine-grained soil during interface varying velocities (and at the low stress relevant to
shear varies between fully-drained and fully undrained pipelines) in order to identify this response directly,
across a range of velocities that typically spans two rather than assessing it indirectly from a normalised
or three orders of magnitude. It is useful to consider theoretical curve via measurements of cv and k.
firstly the response in tests where the apparatus forces The form of response shown in Figure 34 can be
the failure to occur on a particular plane, rather than used to assess the relevant axial resistance for the ini-
the case of a curved pipe surface on the seabed. The tial axial movement of a pipeline. Over a longer period,
interface direct shear box tests from Figure 10 are and after a series of movements, this relationship no
shown again in Figure 34, and supplemented by results longer applies. The apparent friction coefficient dur-
from shearing of a concrete – clay till interface at ing undrained and partially drained sliding is lower
varying rates, reported by Steenfelt (1993). This study than the drained case due to the generation of positive

108
pore pressure when shearing is initiated. Subsequent
dissipation of this pore pressure leads to a rise in effec-
tive stress. This is accompanied by contraction of the
interface zone, which can be observed as settlement of
the platen in a direct shear test.
An example of this behaviour is shown in Fig-
ure 35. These are results from shearing of normally
consolidated kaolin clay on a steel interface.The shear-
ing rate was close to the fully undrained limit and
a sequence of 2 cycles of +/−5 mm is shown. No
consolidation period was permitted between cycles.
The initial apparent friction (based on total stresses) is
τ/σn = 0.28. However, over the following two cycles
the resistance rises (Figure 35a) whilst the sample
settles (Figure 35b).
The rate of settlement is approximately constant
through the two cycles, with a slight tendency for more
rapid settlement following each reversal. Meanwhile,
the absolute shearing resistance, when plotted against
the cumulative horizontal movement, shows a contin-
uous rise, interspersed with reductions to zero at each
reversal point (Figure 35c). After only 35 mm of move-
ment, or an elapsed time of approximately 3 hours, the
apparent friction coefficient has risen from 0.28 to 0.46
(Figure 35d).
This response can be illustrated schematically in
stress – specific volume space (Figure 36a). For the
type of test shown in Figure 35, the stress path ini-
tially heads to the critical state line in an undrained
manner (path OU). If sliding continues, then the state
moves along the critical state line (CSL) following the
increase in effective stress that accompanies the dis-
sipation of excess pore pressure (UP). Ultimately the
test path and the strength (or apparent friction) reaches
the same state as a drained test would (point D) (Figure
36b). The drained test follows the path OD. At point D
the effective stress equals the applied total stress.
The idealised strength response in Figure 36b
ignores the possible influence of a changing mobilised
friction angle, and we are also not differentiating
between a critical state line approached after mono-
tonic shearing, and one at a denser state approached
through cyclic shearing.

5.3 Episodic interface shearing and consolidation

The three test paths shown in Figure 36 cover three
possibilities for an initial episode of sliding. For the
fully undrained case, after sliding halts, the consoli-
dation will be accompanied by contraction following
an unload-reload (κ) line (U-R, Figure 36a) – exactly
analogous to the reconsolidation stages shown in
Figure 7. Following this reconsolidation, a smaller Figure 35. Settlement and apparent friction behaviour dur-
excess pore pressure and hence a higher undrained ing interface shearing of kaolin clay.
strength will be mobilised in a subsequent fast sliding
event – just as shown in Figure 6. followed by reconsolidation will take the soil state to
These idealised responses of a single soil element the point on the critical state line at the applied total
through episodes of undrained sliding and reconsolida- normal stress. In this case there is no longer a ten-
tion can be used to estimate the long-term cyclic axial dency for excess pore pressure to be created during
pipe-soil resistance during movements at undrained shearing, and the apparent friction will be the drained
rates. Eventually, the process of undrained shearing value regardless of the shearing rate.

109
Figure 36. Critical state interpretation of drained, undrained
an partially-drained interface shear tests.

5.4 Application to axial pipeline resistance

These observations from shear box testing are not
directly applicable to pipeline behaviour due to the
potential for failure to occur on multiple planes at dif-
ferent radii from the pipe surface. At first sight, Figure
35 and Figure 36 suggest that once a drained failure Figure 37. Schematic illustration of successive undrained
condition has been reached then the apparent sliding failures on adjacent surfaces during axial pipeline movement.
friction will always be the drained value irrespective
of the rate of movement, since the soil state at the
interface has reached the critical state line, and has stress. If the soil is assumed to be uniform, then an
no tendency to contract or dilate. However, in a shear initial axial movement, fast enough to be undrained,
box test the failure plane is prescribed by the boundary will cause failure within the shear zone immedi-
conditions of the apparatus, so only this single plane ately adjacent to the pipe (zone A in Figure 37a, b).
needs to be hardened. In contrast, failure can occur at The mobilised axial resistance will be controlled by
different radii close to the surface of a pipeline. the effective stress during undrained failure of this
If the pipe surface is rough then both pipe-soil shear- element, which will follow the stress path OA in
ing and soil-soil shearing have comparable resistance, Figure 38.
and failure will occur at whichever circumferential After a subsequent period of pore pressure dissipa-
surface offers the least resistance. We illustrate how tion, shear zone A will have contracted (Figure 37c)
this affects the hardening behaviour by considering a and hardened to point A , under the stress of σn
pipe that undergoes intermittent fully undrained move- imposed by the pipe (Figure 38). However, although
ments, with intervening consolidation periods. For this element is now able to offer increased shearing
simplicity we are setting aside the effect of partial resistance during a subsequent cycle, zone B has not
drainage during shearing. been hardened, and will form the ‘weak link’ during
The shear stresses created within the soil by the axial the next undrained axial movement (Figure 37d).
traction at the pipe-soil interface decay approximately This process can continue with each successive
with the inverse of the radial position. Given that a cycle leading to hardening of another zone of soil – or,
shear plane is typically a fraction of a millimetre in as the shear zones progressively harden, to the remo-
thickness in fine-grained soils, there are many poten- bilisation of a previously-sheared zone. Ultimately all
tial shear planes carrying essentially the same shear zones will reach the steady state (SS) point on the CSL

110
interpret them within the framework described here,
to support assessments of axial pipe-soil resistance.

6 TRENCH CONSTRUCTION BY PLOUGHING

6.1 Mechanics of ploughing

Pipeline ploughs are used to excavate soil from a trench
to permit a pipeline to be lowered below the seabed
(Palmer et al. 1979; Reece and Grinsted, 1986; Cathie
and Wintgens, 2001) and for burying cables (Jordon
and Cathie, 2004). The soil mechanics of plough-
ing involves large soil deformations, mobilization of
Figure 38. Critical state interpretation of failures on adja- peak and residual shear strengths simultaneously in
cent planes during axial pipeline movement (see Figure 37).
different zones around the share, partially drained fail-
ure, strain rate effects on soil strength, and interface
(Figure 38). Hereafter, the mobilised soil strength and shearing.
hence the axial pipe-soil resistance will be the drained In this section, some of these themes are developed
value, regardless of the rate of movement. However, with knowledge gained recently. Plough-soil mecha-
this process of hardening the multiple potential fail- nics is complex and field soil conditions are rarely
ure zones means that a considerably greater number uniform.The operational environment towing a plough
of cycles and episodes of consolidation are required, with a long catenary is analogous to a stick-slip resis-
compared to in a shear box test. tance on the end of a large spring hanging from a
Further research involving model testing, supported heaving vessel. Trying to keep the plough on track
by episodic interface shear tests, will shed light on is difficult, let alone making clear field observations
these mechanisms. What we have presented here is and measurements of performance.
clearly idealised, but it captures the most important These conditions can lead trenching contractors
features of cyclic axial pipe-soil behaviour in fine- to be satisfied with approximate correlations of
grained soils, namely: plough performance in terms of trench depth, tow
force applied and progress rate. However, this is
1. The significant difference between the undrained
unsatisfactory, at least scientifically, and it is certain
and drained sliding resistance of ‘virgin’interfaces;
that improvements can be made to the current pub-
2. The potential for partial drainage within the shear
lic domain methods for pipeline plough assessment,
zone during a long slow sliding event;
which are summarised by Cathie and Wintgens (2001).
3. The consolidation and hardening that follows
A large pipeline plough and the key loads acting on
undrained sliding episodes in soft soils (on the wet
it are shown in Figure 39. The plough advances due
side of the CSL);
to a tow force (T) being applied which overcomes the
4. The effect of the multiple potential failure surfaces
soil resistance acting on the engaged parts. In normal
at, or close to, a pipe-soil interface.
situations these are the base of the skid and the base
The critical state framework provides a conve- and upper surface of the share. Figure 40 is a close-up
nient basis for interpreting and analysing these events. of a typical share of a pipeline plough for producing a
Ultimately, given sufficient cycles of movement and V-shaped trench.
reconsolidation, any shear box or pipe-soil interface Referring to Figure 39, in steady-state ploughing a
should reach a state where there is no further tendency force and moment balance exists between all the forces
for the shear zone to contract and excess pore pressure acting on the plough. In actual operations the line of
to be generated. At this point a sliding resistance equal action of the base share forces (Shv , Shh ) in particular
to the high drained value would be reliably achieved is continuously changing. The line of action moves
for all rates of movement. However, it remains unclear towards the tip as the plough pitches forward – for
how many cycles (or what cumulative distance of example if the operators raise the skids by rotating
shearing) is required to reach this state in a given soil. the skid linkage counter clockwise in Figure 39 – and
The form of interface shear box test used to illustrate towards the rear if the reverse happens. Forward or aft
this behaviour in an individual soil element is rarely pitch also occurs if the downward component of the
conducted on fine-grained soils, precisely because soil resistance on the top of the share changes, or if
of the ill-defined drainage boundary conditions, and the normal component of soil resistance on the base
the possibility of partially-drained behaviour occur- of the share changes (e.g. if harder or softer soils are
ring. In most normal geotechnical applications such encountered).
behaviour would be unwanted. However, here it illus- Ploughs appear to move forward in a series of accel-
trates the influence on interface friction of partial erations and decelerations, or start-stops. This is due to
drainage from a shear band concurrent with sliding. the tow wire catenary behaviour, and the time depen-
Given the insights shown here, it can be worthwhile dent soil shear strength mechanisms (brittle failure,
to conduct project-specific interface shear tests and partial drainage, and thixotropy or setup).

111
Figure 39. General arrangement of a large pipeline plough showing actions and resistances.

where F is the horizontal component of the plough-

ing resistance, W is the plough weight (increased by
the roller loads i.e. the weight of pipe supported), Cw
is a friction coefficient, γ is the soil submerged unit
weight, and Cs is a coefficient, similar to a passive
pressure coefficient. The first term is, of course, the
frictional resistance of the plough and is assumed to
be independent of speed.
The dynamic resistance component of Equation 7
(Fdynamic = Cd vD2 ) principally arises from the poten-
tial of the soil to dilate during shear, and the change in
pore pressure (normally suction) and effective stress
that this induces, offset by any drainage that may
occur. Palmer (1999) has shown that, for a simpli-
fied representation of a triangular plough share, the
velocity-dependent component of force, Fdynamic , is a
Figure 40. Pipeline plough share (mouldboards that dis-
place the excavated soil to the side of the trench are
function of the forward velocity, v, the soil dilation
retracted). potential, S and the soil permeability, k:

The equations of motion and the catenary response

can be coupled with a soil resistance model quite eas- The function f is linear until pore water cavitation
ily in a finite difference algorithm but the unknowns begins to occur in some area around the share. Simi-
are largely associated with the soil resistance models lar relationships have been developed by van Os and
and the simulation is of academic value except in very van Leussen (1987), and van Rhee and Steeghs (1991).
deep water where the catenary is particularly long and The former validated the theory with plane strain tests
flexible. in 2D while the latter performed tests with a plough
geometry (i.e. a triangular trench) and performed mul-
tipasses to simulate several plough shares at different
6.2 Existing ploughing resistance model for sand depth settings.
A ploughing resistance model for sand was described The soil dilation potential, S, or volumetric strain
by Cathie and Wintgens (2001) based on some basic to critical state (Van Leussen and Nieuwenhuis, 1984)
theory and calibrated to a limited data set of full size may be defined as
ploughing:

112
where ec is the critical void ratio and e is the in situ
void ratio.
Considering that the value of S can only vary up
to about 0.4, while the permeability k can vary by at
least an order of magnitude in sands of broadly similar
grain size, it is clear that the dynamic resistance to
ploughing will be dominated by the soil permeabi-
lity and variations in this parameter. It is likely that
much of the scatter that is observed in progress rates
when ploughing in sand is due to small local variations
in grain size distribution and therefore permeability.
Therefore, Cd , the dynamic force coefficient, will be Figure 41. Model plough showing accumulation of soil
a function of (S/k), increasing gently with density and above the share (Lauder, 2010).
increasing strongly with reducing permeability.
The term in D3 in Equation 8 arises from a term in D
for the magnitude of the suction pore pressure and D2 this material is likely to be close to the critical state)
arising from the cross-sectional area of the trench to and the spoil heaps.
be cut. This ignores the actual three-dimensional shape The effect of any build-up of spoil in the body of the
of the share and takes no account of the actual defor- plough was taken account of implicitly in calibrating
mation mechanism experienced by the soil around the the lumped Cs and Cw terms of the model of Cathie
share. and Wintgens (2001) (Equation 7). However, one could
The share can be considered as a horizontally argue that the drained resistance is made up of several
advancing ‘penetrometer’increasing in size up to a cer- separate components:
tain point. When the share tip first influences the soil,
the cavity formed may be considered “deep”, i.e. unaf- 1. Share base friction or adhesion due to the downward
fected by the presence of the soil surface. At this point, component of soil resistance on the share face;
the suction pressures will be related to the dimensions 2. Shearing resistance of the undisturbed soil ahead
of the tip and not the depth below seabed. As the share of the share (passive pressure) augmented by the
advances the cavity increases in size and breaks out weight of the surcharge;
on the seabed so a depth effect (“shallow” failure) is 3. Shearing resistance of the disturbed soil acting on
correct. the mouldboards.
Considerable forward movement (and therefore The passive resistance component of Equation 7
time – several seconds) is associated with transition- (Fpassive = Cs γ D3 ) is the basic component of share
ing from the “deep” to the “shallow” failure at any resistance and the vertical component also creates
cross-section, during which time drainage can occur. additional friction on the base of a share (leading to the
These 3D aspects of the share geometry are one reason first term in Equation 7). This can be assessed approx-
why Cathie and Wintgens considered that the dynamic imately using 3D finite element analysis. Results
resistance actually increases with something nearer to are presented in Figure 43 for a range of peak fric-
D2 (Equation 7) rather than D3 (Equation 8). tion angles representing loose to very dense sand. A
Drucker-Prager soil model with a cap was used in
ABAQUS. It can be seen that the passive resistance
6.3 Components of ploughing resistance in sand does not increase with D3 and is more correlated with
Further insights into the ploughing resistance formu- D2 or the trench cross-sectional area, A. The effect of
lation are provided by more recent research discussed surcharge is not included in this analysis and the soil-
below. steel friction angle was taken as φ – 5◦ which may be
rather high for the denser soils.
6.3.1 Static ploughing resistance Peng and Bransby (2010) performed 2D FEA mod-
Model testing in dry sand at 1g performed by Lauder eling and concluded that the static resistance parameter
(2010) has provided some useful understanding of Cs was linearly related to Kp tanφ or K2p for the 2D
drained ploughing resistance and the soil flow mech- model, varying with depth as D2 . This study did not
anism around the share and mouldboards (Figure 41). include the effect of surcharge.
There can be a large accumulation of spoil above the Bransby et al (2005) and Lauder et al (2008) report
share. This adds weight to the share (thus increasing 1 g model test results that appear to broadly confirm
share friction), adds weight as a surcharge to the soil the static resistances suggested by Cathie and Wint-
being sheared by the share, and adds resistance due to gens (2001) but with a margin of uncertainty due to
the pressure of the spoil on the mouldboards. the lower effective stress level and therefore possibly
The flow mechanisms seen in model testing are higher operational friction angle in the 1g tests.
confirmed by the pattern of scour marks on full-size The formulation of the drained ploughing resis-
plough mouldboards. Figure 42 provides a front view tance therefore still needs to be resolved finally. The
of ploughing showing schematically the area of undis- dimensionless parameter group would be expected to
turbed soil in the trench, the disturbed soil (for sand, be Fpassive /γ D3 but this does not appear to be the case.

113
Figure 42. View of soil moving operation with a pipeline plough.

Figure 43. Drained resistance of share only as a function of Figure 44. Suction pressures during soil cutting (He et al.
depth and angle of friction. 2005) (low and negative to hydrostatic: blue, green, yel-
low, red).

The effect of surcharge in a two-dimensional ideal-

isation has been considered by Hettiaratchi and Reece static resistance and their relative magnitudes before
(1974, 1975) and is a function of surcharge pressure (q) measured data can be properly interpreted to establish
and trench depth (D). For a 3D diamond shaped share, the dynamic resistance components.
the surcharge pressure is (as a first approximation) a
function of the trench cross-sectional area divided by 6.3.2 Dynamic ploughing resistance in sand
the trench width (see Figure 42). Therefore, the sur- As noted in Section 6.2, rate-dependent effects in sub-
charge pressure is a function of depth (q = f(γ’D)) and merged sands are largely due to the suction created by
the effect on soil passive resistance a function of D2 . dilation potential during undrained shearing. This sub-
Surcharge pressure also increases the vertical load on ject has been studied extensively in respect of cutting
the share and therefore increases friction. This com- sands during dredging (Van Leussen and Nieuwenhuis,
ponent would be related to the volume of soil retained 1984; van Os and van Leussen,1987; Miedema, 1987;
above the share and so both D2 and the length of the Miedema, 2005).
share, i.e. D3 for a diamond-shaped share. Experimental and numerical analysis (e.g. van Os
In respect of the component of resistance arising and van Leussen, 1987; He et al, 2005) demonstrates
from mouldboard pressure, this should be related to that the normalized cutting speed v/k is a key variable
the surcharge depth as discussed above, and therefore controlling this resistance (Figure 44).
the resistance would be a function of D2 . All of the research into the cutting of sand for dredg-
Finally, work must be done to lift the soil from the ing, which is centred on groups based in Delft, has
trench to the spoil heap level. For an increment of for- used v/k for normalisation rather than vD/cv , which
ward movement, the volume of soil to be raised is a is more commonly considered for normalizing pen-
function of trench area. Assuming that the average dis- etrometer and foundation speed effects in geotechnical
tance it must be raised is related to D the contribution engineering. The answer may lie in the localisation
to resistance is unsurprisingly related to γ D3 . during shearing of sand, limiting the dilation to a nar-
Clearly, a combination of these terms will con- row plane. The resulting pore pressure response is
trol experimental or field data. It is essential to get essentially a seepage problem, rather than one of con-
an improved understanding of these components of solidation. Interestingly, there are exceptions within

114
the interface properties: sud = f(S,v/k, φ, δ). Dynamic
resistance would then be related to D2 considering
the effect of this operative undrained strength on the
passive resistance (Hettiaratchi and Reece, 1974) as
suggested by Cathie and Wintgens (2001). It is possi-
ble that a large component of the dynamic resistance
arises from the increased normal stresses on the base
of the share due to suction. The use of an operative
undrained strength would facilitate an integrated inter-
pretation for ploughing in sands and clays and enable
the resistance to be evaluated in terms of a normalised
ploughing speed (which would be consistent with the
framework shown in Figure 9, which is used elsewhere
in pipeline geotechnics).

6.4 Ploughing resistance in clay

Cathie and Wintgens (2001) also proposed an empiri-
Figure 45. Pore water pressures around share during cal model for ploughing resistance in clays:
ploughing (low and negative to hydrostatic: black, blue,
green, yellow, red; the upper figure is at the share tip, the
lower figure is further back along share).

where Fw is the adhesion of the underside of the skids

geotechnical engineering: Elsworth and Lee (2005) and share to the soil during ploughing, Cc is a coef-
analyse excess pore pressure generation during cone ficient similar to a bearing capacity factor, su is the
penetration using only v/k, treating the problem as soil undrained shear strength, D is the trench depth,
seepage not consolidation. v is the plough speed and Cd is a coefficient relating
Cathie Associates has performed 3D finite element the strength of the soil at normal shear strain testing
modeling of the plough share problem to investi- rates (su ) to the strength of the soil at ploughing rates
gate the dynamic component of ploughing resistance. of strain.
A Mohr-Coulomb cap soil model with limited dila- Recent numerical studies by Cathie Associates have
tion was used within ABAQUS. Differing from the highlighted the sensitivity of Cc to the working inter-
He et al. (2005) work where cutting rates are faster, face resistance between the share and the soil. Near
our FEA suggests that maximum suction pressures the tip the resistance is probably close to the intact su .
are relatively localized around the share, although of Further along the share the strength almost certainly
course the zone would grow for higher speeds or lower drops to a residual value, possibly corresponding to
permeability. the remoulded shear strength. The working interface
Ahead of the share, the suction pressures are lower resistance has a strong impact on the values of Cc
due to the increased mean stress resulting from the computed.
advancing share. Figure 45 also shows clearly the Probably insufficient attention has been given to
effect of the free surface higher up the share where experience and research on tillage tools in the agri-
suction pressures drain much more quickly. Partial cultural sector. For example, Karmakar (2005) inves-
dissipation of excess pore water pressures is limited tigated the behaviour of a tool shown in Figure 46
near the tip but quite substantial further back along the which has some similarities to a pipeline plough share.
share, particularly near the top. The length of the share He found that brittle failure dominated the response of
is a factor affecting the dynamic ploughing resistance the soil above the share. While his work was in unsat-
just as it is for static resistance. urated materials, ploughing experience confirms that,
A further conclusion arising from the 3D FEA work in stiff clays, the soil is excavated as lumps and our
was that the average plastic volumetric strain around FEA work also demonstrated the importance of brittle
the share (a measure of the dilation potential) was not tensile failure for all but soft clays.
very sensitive to the soil density (or dilation angle) The value of Cc indicated in Cathie and Wintgens
even if very localized high dilations occurred. (2001) did not take account of brittle failure and con-
Finally, the dynamic ploughing resistance also con- sidered that it was largely independent of su . This
tains a contribution from the dynamic shearing of the appears to result in over-predictions of the ploughing
accumulated spoil against the mouldboards. resistance in very stiff to hard clays. We assume this is
In the view of the authors, a promising approach because the clays actually shear in a brittle manner with
to assess dynamic resistance is to consider a opera- the operational strength being somewhere between the
tive undrained strength, sud , that would be a function intact, residual or (submerged) tensile strengths.
of the average pore pressures generated around the While for clays there is still plenty of scope for
share (capped in dilatant conditions by cavitation) and theoretical work on ploughing resistance, it receives

115
Figure 46. Crack propagation in brittle unsaturated clays
(Karmakar, 2005).

Figure 48. Examples of jet trenchers.

Figure 47. Schematic of pipeline lowering using jet trench-
ing (Vanden Berghe et al. 2008). 7.2 Jetting in sands
The main physical processes involved in jetting of
less attention in industry because ploughing most clays coarse-grained soils are:
is much easier than ploughing sands. 1. Sediment entrainment (erosion and fluidization of
coarse-grained soils) or soil cutting (fine-grained
soils);
7 TRENCH CONSTRUCTION BY JETTING 2. Sediment transport;
3. Trench collapse;
7.1 Mechanics of jet trenching 4. Sediment deposition.
Jet trenching involves the use of water jets to erode and The water jets erode coarse-grained soil due to their
fluidise sands, or to cut clays, in order to permit low- high energy and carry the suspended particles away
ering and burial of cables (Jordon and Cathie, 2004) from the forward face, transporting them rearward.
and pipelines as shown in Figure 47 (Vanden Berghe Sediment erosion is a well understood and for the
et al, 2008). typical flow rates used by jet trenchers, erosion and
Jetting systems utilize the power of water jets to fluidization is normally achieved. However, if there
erode or cut the soil, and to transport or fluidise it to is insufficient flow compared to the trench volume
allow the cable or pipeline to be lowered into the trench and progress rate of the trencher, the fluidization will
(Figure 47). Jetting systems range in power from about be incomplete and the jetting swords will experience
250 kW to 2 MW and jet swords generally comprise resistance from the soil still in place.
quite large numbers of nozzles facing forwards, down- A nominal measure of the soil volume to be exca-
wards and rearwards depending on the manufacturer’s vated is defined by the excavation rate ER :
design. Jetting pressures are typically in the range 2–10
bar with flow rates between 100–3000 m3 /hr.

116
than coarse sands and gravels. Therefore, the fluidized
zone is longer and burial often better in fine soils (see
Figure 50 and Figure 51).
A mathematical model of the jet trenching process
has been developed by Vanden Berghe et al (2008)
and Peng and Capart (2008) under contract to CTC
Marine Projects. The model is based on the funda-
mental physical processes that occur and has been
calibrated by a series of 1g model tests. Both the tests
and the mathematical modelling confirm the depen-
dence of the progress rate on jetting power and sand
density, and capture the main phenomena of sidewall
collapse, sedimentation rate and overspill.
Overspill (or loss of sediment from the trench)
occurs due to dispersion of the sediment in the tur-
bulent flow. The effect of cross-currents, which may
be quite severe in practice, has not yet been consid-
ered. The model has also been used to optimize sword
design and to assess the value of rearward facing jets
Figure 49. Relationship between fluidization ratio and to lengthen the fluidized region aft of the trencher.
trencher speed. Note that the burial depth of the cable or pipeline
also depends also on the weight, stiffness and residual
tension in the cable or pipeline. Lower weight, higher
stiffness or higher tension all lead to a longer span
where v is the trencher progress rate (m/hour), DSw is
length and therefore reduced burial depth (Figure 51).
the embedded sword depth and WSw is the sword width.
A quantitative treatment of this interaction is provided
The ratio of the water pumped into the soil and the vol-
by Vanden Berghe et al (2010).
ume to be excavated at a given speed can be called
The unit weight (or specific gravity) of the pipeline
the fluidization ratio (i.e. jetting system flow rate/
or cable during installation must be greater than that
excavation rate). This is also a measure of the result-
of the fluidized soil. Moreover, there must be suffi-
ing water/solids ratio (or the solids concentration, as
cient weight margin to ensure that the lowering of the
is considered in the materials transport industry). It is
product is not affected by the turbulence and upward
useful for any trencher to develop charts of fluidization
flow of water behind the trencher. A minimum spe-
ratio for various speeds, sword width and sword depth
cific gravity is believed to be about 1.8 (but without
to provide a rapid method of assessing likely progress
conclusive experimental support).
rates that could be anticipated. An example is shown
Figure 49.
Fluidisation ratio is one indicator of the likely max-
7.3 Jetting in clay
imum rate of progress of a jet trencher and must be
sufficient to enable the soil to be transported at the Jetting of fine-grained soils is essentially a cutting pro-
flow rates available. Fluidisation ratios above 5 appear cess rather than an erosion process. The jets must cut
to be sufficient but this also depends on the soil grain and break up the material ahead of the sword. After that
size and the configuration of the jets. the essential mechanisms are similar except the clay
Following erosion and fluidization of the sand is not fully disaggregated but remains in lumps which
grains, the sand must be transported behind the must be transported and maintained in suspension if a
trencher before it is deposited in the lower flow regime pipeline or cable is to be lowered. Conventional wis-
behind the swords. The process is depicted in Figure dom is that clay lumps should be removed (educted)
50 based on experimental work at the University of from the trench by a dredging tool located on the
Taiwan on behalf of CTC Marine Projects. trencher if good lowering is to be achieved.
Both the power and orientation of the forward jets, A water jet from a static nozzle impinging on a
and the power of any rear facing jets affects the sedi- bed of clay creates a circular depression or cylindri-
ment transport and to some extent defines the length cal cavity in the clay often up to 3 times the nozzle
of the turbulent region through which the pipeline or diameter (Machin and Allan 2010). The depth of the
cable is lowered. Lateral inflow of soil is also observed cavity will depend on the jet pressure, the undrained
in model tests, which merges with the turbulent flow shear strength and other properties of the soil, the
behind the swords to increase the solids content and offset between the bed and the nozzle (stand-off dis-
reduce the energy in the turbulent region. Inflow is less tance), and the time the jet is allowed to act. Machin
in fine, dense sands which can support a near vertical et al (2001) show that the depth of cut consists of a
wall for longer than in coarse, loose sands. quasi-instantaneous cut depth formed in a time span
Eventually, sediment is deposited in a hindered set- of 0.1–0.5 ms. After this, time-dependent erosion of
tling regime at a rate that is largely dependent on the the already-formed cut takes place. Under some cir-
mean grain size. Fine sands settle much more slowly cumstances, and particularly in fissured clays or clays

117
Figure 50. Mechanisms observed during jet trenching (Vanden Berghe et al. 2008).

Figure 52. Pressure distribution of a submerged water jet

acting on a target (Kondo et al. 1974).

Figure 51. Lowering of a pipeline or cable in fluidized zone with the deformation of the first liquid to strike the
behind a jet trencher.
solid;
with silt or sand seams, hydraulic fracture mechanisms 2. A lower, quasi-steady pressure as the jet begins
also operate. to flow outwards; this is associated with the pres-
A basic model for clay cutting using jets requires an sure required to deflect the liquid sideways over the
understanding of the mechanics of a submerged jet. If surface;
the center-line velocity of a submerged jet at the nozzle 3. Shear stresses caused by liquid moving over the
is v0 , the fluid velocity decays with distance due to surface at high speed from the centre of impact.
spreading of the jet and entrainment of the surrounding Considering the normal impact of a steady jet of
water. The velocity, v, at the centreline of the flow at a inviscid, incompressible fluid against a rigid surface,
stand-off distance x from the nozzle can be estimated and assuming the stand-off distance is small enough,
by: the pressure distribution may be considered as uniform
in the core area of the jet. The pressure applied to the
surface is known as the stagnation pressure, p:

where C is a dimensionless hydrodynamic drag coef-

ficient (C ≈ 6.2), and d is the jet diameter. Within the
where ρw is the mass density of the water jet.
initial region, up to about 6d, there is no loss in velocity.
A simple model for assessing the required jet pres-
When a jet impinges on a surface, the necessary
sure to cut clay soil can be developed by considering
deflection of the jet fluid results in a pressure applied
that the normal pressure of the jet, p, acting on the
to the surface as illustrated in Figure 52 (Kondo
soil (Figure 52) creates a bearing failure. The reality
et al. 1974). A solid target impacted by a liquid jet
is probably much more complex with both normal and
is subjected to the following pressures:
shear stresses applied to the soil, local spalling occur-
1. A strong initial compression pulse with a duration ring, localized drainage of the surface, erosion and
on the order of a microsecond; this is associated variations in water pressure (hammering).

118
Bearing capacity failure will occur if:

where Nc is a bearing capacity factor (typically

Nc = 6), or more simply the condition that p > qc , the
cone tip resistance, can be used. Equations 11-13 can
be combined to relate the pressure at the nozzle (p0 )
required to cut a soil of a given shear strength as a func-
tion of stand-off distance, considering that Equation 12
also applies to the pressure and velocity at the nozzle
(following Bernoulli). As shown also by Machin and
Allan (2010), the resulting relationship between stand-
off distance and the nozzle pressure to cause bearing
failure is: Figure 53. Required jet pressure for cutting.

As a first approximation, x may also be considered

the depth of cut that can be made for a jet located at
the surface of the soil. The relationship is shown on
Figure 53. With a jetting pressure of 20 bar from a
20 mm OD nozzle, in a soil of 150 kPa shear strength,
a cut depth of about 0.2 m can be anticipated from each
jet excluding any stand-off distance.
From a practical perspective, considering a jetting
sword of the type shown in Figure 48 with multiple jets,
it is unlikely that zero stand-off distance can ever be
achieved for all jets simultaneously. A minimum aver-
age stand-off might be 0.1 m if the trencher is being
driven hard into the trench face. Assuming a mini-
mum acceptable cut depth for each jet (of 20 mm OD) Figure 54. Cutting depth as a function of jet trenching
of 0.2 m gives x/d of 15. A typical 5 bar low pres- speed.
sure trenching system could be expected to operate
acceptably in shear strengths of less than about 20 kPa. of 1000 psi (69 bar), being less for coarse-grained soil
This agrees with general experience in the trenching than fine-grained. It was found that the jet penetration
industry. Low pressure (say 5 bar) systems which work depth, x, in a given soil is related exponentially to the
effectively in sand can also create a trench in very soft corresponding time from initial impact. Machin and
to soft clays. Allan (2010) indicate that it takes several seconds to
As the bearing capacity failure occurs and the hole reach the full cavity depth in clays.
is deepened, the actual failure mechanism becomes Atmatsidis and Ferrin (1983) also show that the pen-
more complex. The bearing capacity increases with etration depth, x, could be related exponentially to the
depth, and the flow is no longer as shown on Figure jet’s traversing velocity, v:
52 but is constrained inside the hole. The mechanism
of entrainment of water leading to the reduction in jet
velocity is quite different. However, Equation 14 is
considered to provide a good indication of the ability
of a static jet to cut the soil and experimental work where xmax is the limiting penetration for a stationary
does seem to confirm that this approach provides a jet after infinite time, and ζ is an empirical constant
conservative assessment (Machin and Allan 2010). (with units of velocity) depending on jet and material
Jet trenching involves the jets traversing rather than properties.
remaining static. Atmatzidis and Ferrin (1983) investi- Combining the static and traversing equations pro-
gated the influence of time and traversing speed in the vides an indication of the likely depth of cut that a spe-
laboratory with a 1 mm diameter nozzle in clean sand, cific jetting system could provide, if the value of ζ were
silty sand, silt and clay. The depth of jet penetration known. Machin and Allan (2010) suggest that a quasi-
into the soil target and jet effectiveness was measured instantaneous cavity depth is achieved for translation
as a function of exposure time, the degree of saturation speeds exceeding 0.1–0.5 m/s (360–1800 m/hr) but
of the soil, the dry density of the soil and the traversing slower speeds are required for maximum penetration.
velocity of the jet over the soil target. Figure 54 shows a typical result for cutting
Their research showed that the time required for stiff clay (su = 100 kPa) considering that the quasi-
the jet to approach maximum penetration in the soil instantaneous cavity depth is achieved at a speed of
target was about 15–20 seconds for a driving pressure 500 m/hr (assumed).

119
Successful lowering of a cable or pipeline also soil deformations to be observed. Other insights have
depends on the orientation of the jets so that the soil is emerged through finite element analysis – with notable
actually cut into blocks which can be educted or trans- advances being the use of coupled methods in sand
ported away from the zone where the product must to capture undrained and partially-drained ploughing
sink. As far as the authors are aware, the first dis- behaviour, and the use of large deformation tech-
cussion of this issue in the public domain is given in niques to capture gross remoulding of fine-grained
Machin and Allan (2010). soils during large lateral movements.

8 CONCLUSIONS
ACKNOWLEDGEMENTS
This paper has reviewed various aspects of pipeline
geotechnics, by reference to recent and emerging The work described here forms part of the activi-
research activity from both academia and industry. ties of the Centre for Offshore Foundation Systems
The design challenges in pipeline geotechnics differ (COFS), established at the University of Western Aus-
somewhat from conventional foundation engineering. tralia in 1997 under the Australian Research Council’s
Two particular challenges are the changes in both the Special Research Centres Program. COFS is now sup-
seabed topography and also the soil properties that can ported by Centre of Excellence funding from the State
occur throughout the installation and operating life of Government of Western Australia. The first author is
a pipeline: these are themes that run throughout this supported by an Australian Research Council Future
paper. Fellowship (grant FT0991816)
Results from novel forms of repetitive in situ testing The assistance from a number of colleagues at UWA
have been used to illustrate the response of soil to the and Cathie Associates during the preparation of this
forms of loading and disturbance that are induced by a paper is acknowledged. In particular, Fauzan Sahdi
pipeline. In soft fine-grained soils, these tests illustrate kindly provided the data in Section 2.3 from his PhD
the balance between the reduction in strength from studies. The interface shear box tests in Section 5.2
remoulding and the recovery that accompanies subse- were performed by Nat McNab assisted by Binaya
quent reconsolidation. Concepts from critical state soil Bhattarai. The MCWHIPLASH software (Section 4.4)
mechanics provide a useful framework for capturing was written by the first author with David Bonjean
this behaviour. of Advanced Geomechanics, Perth. Helpful comments
Solutions for incorporating this behaviour into the provided by George Zhang, also of Advanced Geome-
estimation of axial and lateral pipe-soil resistance, chanics, who reviewed a draft of this paper are also
and the assessment of trenching and ploughing oper- acknowledged.
ations, are discussed. A common theme is the relative Some of the research described in this report
magnitude of drained and undrained soil strengths, has been guided by the SAFEBUCK Joint Indus-
the evolution of these strengths, and the importance try Project, which is coordinated by David Bruton
of recognising the widely-varying rates of shearing of AtkinsBoreas. The support of the SAFEBUCK
involved in pipe-soil processes.The slow rates at which participants is gratefully acknowledged.
pipes move under thermal loading and the high rates
at which trenching machines are driven mean that
REFERENCES
pipeline geotechnics often involves a drained response
in fine-grained soils and undrained behaviour in sands. AtkinsBoreas. 2008. SAFEBUCK JIP: Safe design of
The conventional laboratory tests used to characterise pipelines with lateral buckling; design guideline. Report
these soils often require modification in order to BR02050/C, AtkinsBoreas 252pp.
extract the properties that are relevant for pipe-soil Atmatzidis, D.K. and Ferrin, F.R. 1983. Laboratory inves-
interaction. tigation of soil cutting with a water jet, 2nd US Water
Some of the concepts discussed in this paper repre- Jet Conference, Rollo, Missouri, University of Missouri,
sent merely a snapshot of an evolving understanding, 101–110.
Aubeny, C.P., Shi, H., and Murff, J.D. 2005. Collapse loads
which will no doubt advance in the coming few years. for a cylinder embedded in trench in cohesive soil. ASCE
For example, the changes in soil strength through Int. J. Geomechanics 5(4):320–325.
episodes of disturbance and recovery, and through Bolton, M.D. 1986. The strength and dilatancy of sands.
cycles of partially-drained interface shearing, have Géotechnique 36(1):65–78.
only recently been observed in experiments. The Bolton, M.D. and Barefoot, A.J. 1997. The variation of critical
underlying mechanisms are not fully established and pipeline trench back-fill properties. Proc. of Conference
quantitative calculation methods for design use are in on Risk-Based and Limit State Design and Operation of
their infancy. Pipelines, Aberdeen.
New modelling technologies have been recently Bolton, M.D., Ganesan, S.A. and White, D.J. 2009.
SAFEBUCK Phase II: Axial pipe-soil resistance: sum-
applied to pipeline geotechnics. The mechanisms mary report. Cambridge University Technical Services,
of pipe-soil penetration and breakout behaviour Report for Boreas Consultants (SAFEBUCK JIP), ref.
have been quantified through sophisticated centrifuge SC-CUTS-0705-R01. 54pp.
model tests, which allow complex load sequences to Bransby, M.F. Yun, G.J. Morrow, D.R. and Brunning, P.
be imposed on a model pipe and detailed internal 2005. The performance of pipeline ploughs in layered

120
soils, International Symposium on Frontiers in Offshore three-dimensional approach. Part 1: Theoretical formula-
Geotechnics (ISFOG), Perth, September. tion. Canadian Geotechnical Journal 46:1289–1304.
Bransby, M.F. and Ireland, J. 2009. Rate effects during Damgaard, J.S. and Palmer, A.C. 2001. Pipeline stability on
pipeline upheaval buckling in sand. Proc. ICE Geotechni- a mobile and liquefied seabed: a discussion of magni-
cal Engineering 162: 247–256. tudes and engineering implications. Proc., 20th Int. Conf.
Bruton, D.A.S., White, D.J., Cheuk, C.Y., Bolton, M.D. and on Offshore Mechanics and Arctic Engineering, Rio de
Carr, M.C. 2006. Pipe-soil interaction behaviour dur- Janeiro.
ing lateral buckling, including large amplitude cyclic Dean, E.T.R. 2010. Offshore Geotechnical Engineering.
displacement tests by the Safebuck JIP. Proc. Offshore Thomas Telford.
Technology Conference, Houston. OTC17944. Dingle, H.R.C., White, D.J., and Gaudin, C. 2008. Mech-
Bruton, D., Carr, M. and White, D.J. 2007. The influence anisms of pipe embedment and lateral breakout on soft
of pipe-soil interaction on lateral buckling and walking clay. Canadian Geotechnical Journal, 45(5):636–652.
of pipelines: the SAFEBUCK JIP. Proc. 6th Int. Conf. DNV 2007a. Recommended Practice DNV RP-F109, On-
on Offshore Site Investigation and Geotechnics, London. bottom Stability Design of Submarine Pipelines. (with
133–150. amendments, April 2009). Det Norske Veritas.
Bruton, D., White D.J., Langford, T.L. and Hill, A. 2009. Tech- DNV. 2007b. Recommended Practice DNV-RP-C207, Statis-
niques for the assessment of pipe-soil interaction forces tical representation of soil data. Det Norske Veritas.
for future deepwater developments Proc. Offshore Tech- Elsworth, D. and Lee, D.S. 2005. Permeability determination
nology Conference, Houston, USA. Paper OTC20096. from on-the-fly piezocone sounding.ASCE J. Geotech and
Cardoso, C.O., da Costa, A.M. and Solano, R.F. 2006. HP-HT Geoenvironmental Engng., 131(5):643–653.
pipeline cyclic behaviour considering soil berms effect. Erbrich, C.T. 2005. Australian frontiers – spudcans on the
Proc. 25th Int. Conf. on Offshore Mechanics and Arctic edge. Proc. Int. Conf. on Frontiers in Offshore Geotech-
Engineering, Hamburg, Germany. nics, Perth. 49–74.
Cardoso, C.O. and Silviera, R.M.S. 2010. Pipe-soil inter- Fannin, R.J., Eliadorani,A. and Wilkinson, J.M.T. 2005. Shear
action behavior for pipelines under large displacements strength of cohesionless soil at low stresses. Géotechnique
on clay soils – a model for lateral residual friction fac- 55, No. 6, 467–478.
tor. Offshore Technology Conference, Houston. Paper Finnie, I.M.S. 1993. The performance of shallow founda-
OTC20767. tions in calcareous soil, PhD thesis, University of Western
Carr, M., Matheson, I., Peek, R., Saunders, P. and George, N. Australia.
2004. Penguins flowline lateral buckle formation analysis Finnie, I.M.S., and Randolph, M.F. 1994. Punch-through and
and verification. Proc. Int. Conf. on Offshore Mechanics liquefaction induced failure of shallow foundations on
and Arctic Engineering, OMAE2004-51202. calcareous sediments. Proc., Int. Conf. on Behaviour of
Carr, M.C., Sinclair, F., and Bruton, D.A.S. 2006. Pipeline Offshore Structures, BOSS’94, Boston, 217–230.
walking – understanding the field layout challenges, Fisher, R. and Cathie, D. 2003. Optimisation of gravity
and analytical solutions developed for the SAFEBUCK based design for subsea applications. Proc. International
JIP. Proc. Offshore Technology Conf., Houston, Paper Conference on Foundations, ICOF, Dundee. 283–296.
OTC17945. Gaudin, C. and White, D.J. 2009. New centrifuge modelling
Cathie, D.N. and Wintgens, J.F. 2001. Pipeline trenching techniques for investigating seabed pipeline behaviour.
using plows: performance and geotechnical hazards. Proc. Proc. XVIIth Int. Conf. on Soil Mechanics and Geotech-
Offshore Technology Conf., Houston. Paper OTC13145 nical Engineering. Alexandria.
Cathie, D.N. Jaeck, C.. Ballard, J.C and Wintgens J.F. 2005. Gourvenec, S.M. and White, D.J. 2010. Elastic solutions
Pipeline Geotechnics: State of the Art. Proc. Interna- for consolidation around seabed pipelines. Proc. Offshore
tional Symposium on Frontiers in Offshore Geotechnics Technology Conference, Houston. Paper 20554.
(ISFOG) Perth, Australia. 95–114. He, J. Vasblom, W.J. Miedema, S.A. 2005. FEM analyses
Chatterjee, S., Randolph, M.F., White, D.J. and Wang, D. of cutting of anisotropic densely compacted and satu-
2010a. Large deformation finite element analysis of ver- rated sand, Western Dredging Association Conference,
tical penetration of pipelines into the seabed. Proc. 2nd WEDAXXV and TAMU37, New Orleans, USA, June and
Int. Symp. on Frontiers in Offshore Geotechnics. Perth. www.dredgingengineering.com/dredging
Chatterjee, S., Randolph, M.F. and White, D.J. 2010b. The Hettiaratchi, D.R.P. and Reece, A.R. 1974. The calculation of
effects of strain rate and strain softening on the verti- passive soil resistance, Géotechnique, 24(3):289–310.
cal penetration resistance of pipelines on an undrained Hettiaratchi, D.R.P. and Reece, A.R. 1975. Boundary wedges
seabed. Submitted for publication. in two-dimensional passive soil failure, Géotechnique,
Cheng, L., White, D.J., Palmer, A.C., Jas, E., Czajko, A, 25(2):197–220.
Fogliani, N, Fricke, R. and An, H. 2010. A new facility for Hill, A.J. and Jacob, H. (2008) In-Situ Measurement of Pipe-
research into the stability of pipelines on unstable seabeds. Soil Interaction in Deep Water. Proc. Offshore Technology
Offshore Pipeline Technology Conference, Amsterdam. Conference, Houston, USA. Paper OTC 19528.
Cheuk, C.Y., White, D.J. and Dingle, H.R.C. 2008. Upper Hossain, M.S. and Randolph, M.F. 2009. Effect of strain
bound plasticity analysis of a partially-embedded pipe rate and strain softening on the penetration resistance
under combined vertical and horizontal loading. Soils and of spudcan foundations on clay. ASCE Int. J. Geomech.
Foundations 48(1):137–148. 9(3):122–132.
Cheuk, C.Y. and White, D.J. 2010a. Modelling the dynamic Jayson, D., Delaporte, P., Albert, J-P, Prevost, M.E., Bruton,
embedment of seabed pipelines. Géotechnique. Accepted D. and Sinclair, F. 2008. Greater Plutonio project – Sub-
August 2009, in press. sea flowline design and performance. Offshore Pipeline
Cheuk, C.Y. and White, D.J. 2010b. The influence of breakout Technology Conf., Amsterdam.
conditions on the lateral resistance of on-bottom pipelines. Jordon P, and Cathie, D.N. 2004. Developments in cable
Canadian Geotechnical Journal, in review. protection by burial. SubOptic 2004.
Cocchetti, G., di Prisco, C., Galli, A. and Nova, R. 2009. Karmakar , S. 2005. Numerical modelling of soil flow
Soil-pipeline interaction along unstable slopes: a coupled and pressure distribution on a simple tillage tool using

121
computational fluid dynamics, PhD thesis, University of Pedersen, R.C., Olsen, R.E. and Rausch, A.F. 2003. Shear and
Saskatchewan. interface strength of clay at very low effective stresses.
Kondo, M. Fujii, K. and Syoji, H. 1974. On the destruction of ASTM Geotechnical Testing J., 26(1):71–783.
mortar specimens by submerged water jets, 2nd Interna- Peng W, and Bransby M.F. 2010. Numerical modelling of
tional Symposium on Jet Cutting Technology, Cambridge, soil around offshore pipeline plough shares, 2nd Interna-
paper B5, 69–88. tional Symposium on Frontiers in Offshore Geotechnics
Krost K., Gourvenec, S.M. and White, D.J. 2010. Consol- (ISFOG), Perth, November.
idation around partially-embedded submarine pipelines. Peng A.T.H and Capart H. 2008 Underwater sand bed ero-
Géotechnique, Accepted December 2008, in press. sion and internal jump formation by travelling plane jets,
Lacasse, S., Gutteromsen, T., Nadim, F., Rahim, A., Lunne, T. Journal of Fluid Mechanics, 595:1–43.
2007. Use of statistical methods for selecting design soil Puech, A. and Foray, P. 2002. Refined model for interpret-
parameters. Proc. 6th Int. Offshore Site Investigation and ing shallow penetration CPTs in sands. Proc. Offshore
Geotechnics Conference, London, UK, 449–460. Technol. Conf., Houston, Paper OTC14275.
Langford, T.E., Dyvik, R. and Cleave, R. 2007. Offshore Randolph, M.F. 2003. Science and empiricism in pile foun-
pipeline and riser geotechnical model testing: practice and dation design. Géotechnique. 53(10): 847–875.
interpretation. Proc. Conf. on Offshore Mech. and Arctic Randolph, M.F. and Hope, S.N. 2004. Effect of cone veloc-
Eng., San Diego. Paper OMAE2007-29458. ity on cone resistance and excess pore pressure. Proc.
Lauder, K. 2010. Predicting pipeline plough performance Conf. on Engineering Practice and Performance of Soft
by scale model testing, Presentation to CTC, April, Deposits, Osaka, 147–152.
University of Dundee. Randolph, M.F. and White, D.J. 2008a. Pipeline embedment
Lauder, K. Bransby, F. Brown, M. Cathie, D. Morgan, N, in deep water: processes and quantitative assessment.
Pyrah, J. and Steward, J. 2008. Experimental testing of Proc. Offshore Technology Conference, Houston, USA.
the performance of pipeline ploughs, 18th Int. Offshore Paper OTC19128-PP.
and Polar Engineering Conf. ISOPE-2008, Vancouver, Randolph, M.F. and White, D.J. 2008b. Upper bound yield
Canada, 212–217. envelopes for pipelines at shallow embedment in clay.
Lenci, S. and Callegari., M. 2005. Simple analytical models Géotechnique, 58(4):297–301.
for the J-lay problem. Acta Mechanica, 178:23–39. Randolph M.F. and Gourvenec S.M. 2010. Offshore Geotech-
Lumb, P. 1966. The variability of natural soils. Canadian nical Engineering. Taylor and Francis.
Geotechnical Journal. 3:74–97. Rathbone, A. Hakim, M. A. Cumming, G. and Tørnes,
Lund, K.M. 2000. Effect of increase in pipeline soil penetra- K., 2008. Reliability of lateral buckling formation from
tion from installation. Proc. of ETCE/OMAE 2000 Joint planned and unplanned buckle sites. Proc. 27th Inter-
Conference; Energy of the New Millennium OMAE2000/ national Conference on Offshore Mechanics and Arctic
PIPE-5047. Engineering, OMAE2008-57300, Estoril, Portugal.
Machin, J.B. and Allan, P.J.A. 2010. State-of-the-art jet Reece, A.R. and Grinsted, T.W. 1986. Soil mechanics of sub-
trenching analysis in stiff clays, 2nd International Sym- marine ploughs, Offshore Technology Conference, OTC
posium on Frontiers in Offshore Geotechnics (ISFOG), 5341.
Perth, November. Schotman, G.J.M. and Stork, F.G. 1987. Pipe-soil inter-
Machin J.B. Messina, F.D. Mangal, J.K. Girard, J. and Finch, action: a model for laterally loaded pipelines in
M. 2001. Recent research on stiff clay jetting, Offshore clay. Proc. Offshore Technology Conference, Houston,
Technology Conference,Houston, Paper OTC 13139. OTC5588.
Merifield, R., White, D.J. and Randolph, M.F. 2008a.Analysis Silva, M.F. 2005. Numerical and physical models of rate
of the undrained breakout resistance of partially embedded effects in soil penetration. PhD thesis, University of
pipelines. Géotechnique, 58(6)461–470. Cambridge.
Merifield, R.S, White, D.J. and Randolph, M.F. 2008b. Steenfelt, J.S. 1993. Sliding resistance for foundations on
The effect of pipe-soil interface conditions on undrained clay till. Proc. Wroth Memorial Conference, Predictive
breakout resistance of partially-embedded pipelines. Proc. Soil Mechanics. Thomas Telford. 664–684.
Int. Conf. on Advances in Computer Meth. and Analysis Sture, S., Costes, N. C., Batiste, S.N., Lankton, M.R., Al-
in Geomech. Goa, India. Shibli, K.A., Jeremic, B., Swanson, R.A. and Frank, M.
Merifield, R., White, D.J. and Randolph, M.F. 2009. The 1998. Mechanics of granular materials at low effective
effect of surface heave on the response of partially- stresses. ASCE J. Aerospace Engng 11(3): 67–72.
embedded pipelines on clay. ASCE J. Geotechnical and Tian,Y., and Cassidy, M.J. 2008. Explicit and implicit integra-
Geoenvironmental Engineering, 135(6):819–829. tion algorithms for an elastoplastic pipe-soil interaction
Miedema, S.A. 1987. Calculation of the cutting forces when macroelement model. Proc. 27th International Confer-
cutting water saturated sand, PhD dissertation, Delft ence on Offshore Mechanics and Arctic Engineering,
University of Technology. OMAE2008-57237, Estoril, Portugal.
Miedema, S.A. 2005. The cutting of water saturated sand, the Tornes, K., Jury, J. and Ose, B. 2000. Axial creeping of
final solution, WEDAXXV and TAMU37, New Orleans, high temperature flowlines caused by soil racheting. Proc.
USA, June, and www.dredgingengineering.com/dredging. Conf. on Offshore Mechanics and Arctic Engineering,
Morrow, D.R. and Bransby, M.F. 2009. The influence of slope OMAE-PIPE5055.
on the stability of pipelines subjected to horizontal and Verley, R. and Lund, K.M. 1995. A soil resistance model for
vertical loading on clay seabeds. Proc. Conf. on Offshore pipelines placed on clay soils. Proc. International Con-
Mechanics and Arctic Engineering. OMAE2009-79050. ference on Offshore Mechanics and Arctic Engineering
Orcina. 2008. OrcaFlex software, Orcina Ltd, Ulverston, UK (OMAE), Copenhagen, Denmark, 5: 225–232.
Palmer, A.C., Kenny, J.P., Perera, M.R. and Reece, A.R. 1979. Verley, R.L.P. and Sotberg, T. 1994. A soil resistance model
Design and operation of an underwater pipeline trenching for pipelines placed on sandy soils. ASME J. Offshore
plough. Géotechnique 29(3):305–322. Mechanics and Arctic Engineering, 116(3):145–153.
Palmer, A.C. 1999. Speed effects in cutting and ploughing. Van Leussen, W. and Nieuwenhuis, J.D. 1984. Soil mechanics
Géotechnique 49(3):285–294. aspects of dredging, Geotechnique 34(3): 359–381.

122
Van Os, A.G. and van Leussen, W. 1987. Basic research on White, D.J. and Cheuk, C.Y. 2009. SAFEBUCK JIP: Pipe-soil
cutting forces in saturated sand, Journal of Geotechnical interaction models for lateral buckling design: Phase IIA
Engineering, 113 (12):1501–1516. data review. Report to Boreas Consultants (SAFEBUCK
Van Rhee, C. and Steeghs, H.J.M.G. (1991) Multi-blade JIP), UWA report GEO 09497r1. 185pp.
ploughs in saturated sand; model cutting tests. Dredging White, D.J. and Dingle, H.R.C. 2010. The mechanism of
and Port Constructions, June, 37–39. steady ‘friction’ between seabed pipelines and clay soils.
Vanden Berghe, J.-F. Pyrah, J. Gooding, S. and Capart, H. Géotechnique. Accepted April 2009, in press.
2010. Development of a jet trenching model in sand, White, D.J. and Hodder, M. 2010. A simple model for the
2nd International Symposium on Frontiers in Offshore effect on soil strength of remoulding and reconsolidation.
Geotechnics (ISFOG), Perth, November. Canadian Geotechnical Journal. Accepted November
Vanden Berghe, J.-F. Capart, H. and Su, J.C.C. 2008. 2009, in press.
Jet induced trenching operations: mechanisms involved. White, D.J., Gaudin, C., Boylan, N and Zhou, H. 2010a.
Offshore Technology Conference, Houston. Paper OTC Interpretation of T-bar penetrometer tests at shallow
19441. embedment and in very soft soils. Canadian Geotechnical
Wang, D., White, D.J. and Randolph, M.F. 2009. Numerical Journal, 47(2):218–229.
simulations of dynamic embedment during pipelaying on White, D.J., Hill, A.J., Westgate, Z. and Ballard, J-C 2010b.
soft clay. Proc. Conf. on Offshore Mechanics and Arctic Observations of pipe-soil response from the first deepwa-
Engineering, Honolulu. Paper OMAE2009-79199. ter deployment of the SMARTPIPE. Proc. 2nd Int. Symp.
Wang, D., White, D.J. and Randolph, M.F. 2010. Large defor- on Frontiers in Offshore Geotechnics. Perth.
mation finite element analysis of pipe penetration and Wroth, C.P. 1984. The interpretation of in-situ tests. 24th.
large-amplitude lateral displacement. Canadian Geotech- Rankine Lecture, Géotechnique, 34(4):449–489.
nical Journal. Accepted April 2009, in press. Zhang, J., Stewart, D. P., and Randolph, M. F. 2002a.
Westgate, Z., White, D.J. and Randolph, M.F. 2009. Video Modelling of shallowly embedded offshore pipelines in
observations of dynamic embedment during pipelaying on calcareous sand. ASCE Journal of Geotechnical and
soft clay. Proc. Conf. on Offshore Mechanics and Arctic Geoenvironmental Engineering, 128(5):363–371.
Engineering, Honolulu. Paper OMAE2009-79814. Zhang, J., Stewart, D. P., and Randolph, M. F. 2002b. Kine-
Westgate, Z.J., Randolph, M.F., White, D.J. and Li, S. 2010a. matic hardening model for pipeline-soil interaction under
The influence of seastate on as-laid pipeline embedment: various loading conditions. ASCE International Journal
a case study. Applied Ocean Research, Available online, 4 of Geomechanics, 2(4):419–446.
February 2010. doi:10.1016/j.apor.2009.12.004. Zhang, J., and Erbrich, C.T. 2005. Stability design of un-
Westgate, Z.W., White, D.J., Randolph, M.F. and Brun- trenched pipelines – geotechnical aspects. Proc. Interna-
ning, P. 2010b. Pipeline laying and embedment in soft tional Symposium on Frontiers in Offshore Geo-technics
fine-grained soils: Field observations and numerical sim- (ISFOG) Perth, Australia. 623–628.
ulations. Proc. Offshore Technology Conference, Houston. Zhou, H., White, D.J. and Randolph, M.F. 2008. Physical and
Paper 20407. numerical simulation of shallow penetration of a cylin-
White, D.J. and Randolph, M.F. 2007. Seabed characterisation drical object into soft clay. Proc. GeoCongress 2008, New
and models for pipeline-soil interaction Proc. 17th Int. Orleans,ASCE Special Geotechnical Publication No. 179,
Offshore and Polar Engng. Conference, Lisbon, Portugal. 108–117.
CD 11pp. Zhou, H. and Randolph, M. F. 2009. Penetration resis-
White, D.J. and Gaudin, C. 2008. Simulation of seabed tance ofcylindrical and spherical penetrometers in
pipe-soil interaction using geotechnical centrifuge mod- rate-dependent and strain-softening clay. Géotechnique.
elling. Proc. 1st Asia-Pacific Deep Offshore Technology 59(2)79–86.
Conference, Perth, Dec 2008.
White, D.J. and Cheuk, C.Y. 2008. Modelling the soil resis-
tance on seabed pipelines during large cycles of lateral
movement. Marine Structures 21(1):59–79.

123

View publication stats