SAFEBUCK III Design Guideline Rev A E1 PDF

Safe Design of Pipelines with Lateral Buckling
Design Guideline
SAFEBUCK III
5087471/ 01/ A
Issue Date: 30 August 2011

SAFEBUCK III / Safe Design of Pipelines with Lateral Buckling Page i
Design Guideline
Design Guideline
Prepared for
SAFEBUCK III
COMMERCIAL IN CONFIDENCE
Revision Date Purpose Prepared Approved
A 20/07/2011 For Use M.Carr, D.Bruton, D. Bruton

D.Baxter
Report No: 5087471/ 01 / A

SAFEBUCK III / Safe Design of Pipelines with Lateral Buckling Page ii
Design Guideline
TABLE of CONTENTS
1. GENERAL INFORMATION .......................................................................... 1
1.1. BACKGROUND ............................................................................................ 1
1.2. OBJECTIVE OF GUIDELINE ........................................................................ 1
1.3. SCOPE OF DOCUMENT .............................................................................. 3
1.4. STRUCTURE OF GUIDELINE ...................................................................... 3
1.5. DESIGN CODES........................................................................................... 3
1.6. APPLICABILITY ............................................................................................ 4
1.7. DESIGN EXAMPLE ...................................................................................... 7
1.8. NOTATION, ABREVIATIONS AND DEFINITIONS........................................ 7
1.9. UNITS ......................................................................................................... 13
1.10. DISCLAIMER .............................................................................................. 13
2. BACKGROUND TO LATERAL BUCKLING ............................................... 14
2.1. LATERAL BUCKLING ................................................................................. 14
2.2. BUCKLE DEVELOPMENT .......................................................................... 16
2.3. PIPELINE EXPANSION RESPONSE.......................................................... 18
2.4. VIRTUAL ANCHOR SPACING.................................................................... 22
2.5. PIPE WALKING .......................................................................................... 23
3. DESIGN PROCESS .................................................................................... 24
3.1. METHODOLOGY OVERVIEW .................................................................... 24
3.2. SUSCEPTIBILITY TO BUCKLING .............................................................. 25
3.3. ASSESSMENT OF UNCONTROLLED BUCKLING .................................... 26
3.4. ASSESSMENT OF CONTROLLED BUCKLING ......................................... 27
3.5. BUCKLE INTERACTION AND PIPELINE WALKING .................................. 29
4. CALCULATION OF CHARACTERISTIC VAS ............................................ 32
4.1. OVERVIEW OF METHODOLOGY .............................................................. 32
4.2. CHARACTERISTIC VAS: DETERMINISTIC DEFINITION .......................... 32
4.3. CHARACTERISTIC VAS: PROBABILISTIC DEFINITION ........................... 34
4.4. STRUCTURAL RELIABILITY APPROACH ................................................. 35
4.5. BUCKLE FORMATION MODEL .................................................................. 36
4.6. INPUT PROBABILITY DISTRIBUTIONS ..................................................... 42
4.7. PROBABILISTIC SIMULATION .................................................................. 47
4.8. PROBABILISTIC RESULTS........................................................................ 47
4.9. FE ASSESSMENT ...................................................................................... 49
5. DESIGN ANALYSIS ................................................................................... 51
5.1. SYSTEM MODELLING: VAS ANALYSIS .................................................... 51
5.2. SYSTEM MODELLING: INTERACTION AND WALKING ............................ 53
5.3. DESIGN LOADS ......................................................................................... 55
5.4. CLAD AND LINED PIPES ........................................................................... 57
5.5. MECHANICAL PROPERTIES ..................................................................... 58
5.6. PLASTICITY MODELLING .......................................................................... 61
5.7. PIPE SOIL INTERACTION.......................................................................... 66
6. LIMIT STATES ........................................................................................... 74
6.1. APPROACH TO PIPELINE SAFETY .......................................................... 74
6.2. LOCAL BUCKLING ..................................................................................... 75
6.3. FATIGUE AND FRACTURE ........................................................................ 78
6.4. FATIGUE: S-N APPROACH ....................................................................... 80
6.5. ECA ............................................................................................................ 85
Report No: 5087471/ 01 / A

SAFEBUCK III / Safe Design of Pipelines with Lateral Buckling Page iii
Design Guideline
6.6. PLASTICITY LIMITATIONS ........................................................................ 92

6.7. PROJECT SPECIFIC TESTING.................................................................. 93
7. PROCUREMENT, INSTALLATION AND OPERATION.............................. 98
7.1. PROCUREMENT ........................................................................................ 98
7.2. INSTALLATION ........................................................................................ 100
7.3. OPERATIONAL INTEGRITY..................................................................... 100
8. REFERENCES ......................................................................................... 106
Appendix A Design Examples

Appendix B Pipe Soil Interaction
Appendix C Application of DNV OS-F101
Appendix D Analytic Lateral Buckling and Walking Models
Appendix E Buckle Initiation Techniques and Mitigation Measures
Appendix F SAFEBUCK Deliverables
Report No: 5087471/ 01 / A

SAFEBUCK III / Safe Design of Pipelines with Lateral Buckling Page 1
Design Guideline
1. GENERAL INFORMATION
1.1. BACKGROUND
This document considers the design of on-bottom submarine pipeline systems that are
susceptible to Euler buckling in the horizontal plane (lateral buckling), as well as the
associated pipeline walking phenomena.
Lateral buckling can be a safe and effective way to accommodate the thermal expansion of a
hot pipeline. However there is considerable uncertainty in the initial buckle formation
process and the loads developed in the buckles. High stresses and strains can develop in
the buckles and a conventional stress based design approach is generally not suited to the
design of a laterally buckling pipeline. Consequently, within this guideline the conventional
stress (or moment) limit is relaxed and replaced by a strain limit. However, in doing so the
design must address the following issues:-
1. The amount of feed-in accommodated by each lateral buckle must not be excessive, so
that the bending strains and deformations at the lateral buckle are tolerable. The
following failure modes or limit states must be avoided:
a) Local buckling or wrinkling;
b) Fracture at girth welds, or;
c) Fatigue failure due to low cycle loading associated with operating cycles involving
bending and partial straightening of the line.
2. If a buckle fails to form at an intended location, this not only results in much higher axial
force, but it can also lead to:
a) Excessive feed-in displacement into another buckle, and any of the failure modes
associated with this (under item 1 above), or;
b) Excessive expansion displacements at the ends of the line, which could
overstress the end spools.
3. Unplanned (Rogue) buckles could:
a) Lead to damage of the pipe (item 1 above) if the conditions for lateral buckling
are less favourable, for example due to weight coating or other restraint to free
lateral movements of the pipe, or;
b) Induce a buckle formation failure (item 2 above) by reducing the axial load in the
line.
In theory a single analysis of the entire pipeline (e.g. a finite-element simulation) could
demonstrate that all the above failure modes will be avoided, but in practice there is
considerable uncertainty (and spatial variation) regarding matters such as pipe-soil
interaction, and the as-laid out-of-straightness of the pipe. Furthermore choosing an input
parameter to be conservative for one failure mode can make it non-conservative for another.
Consequently, a more robust design strategy is required. This is the subject of this design
guideline.
1.2. OBJECTIVE OF GUIDELINE

The objective of this guideline is to outline a design approach that addresses the inherent
uncertainty of the lateral buckling problem in a safe and quantifiable way.
Report No: 5087471/ 01 / A

Design Guideline
1.2.1. JIP Partners and Sponsors

The guideline is based upon technical developments investigated within the SAFEBUCK JIP.
The SAFEBUCK JIP has been led by Atkins (formerly Boreas), supported by OTM, TWI,
Cambridge University and the University of Western Australia (UWA). SAFEBUCK GEO is
addressing geotechnical issues, in parallel with Phase III, supported by Oxford University,
Cathie Associates, UWA and Votadini Consultants.
The SAFEBUCK JIP has been supported by a number of sponsors, Figure 1.1.
BOEMRE
Figure 1.1 SAFEBUCK Sponsors (Past and Present)
1.2.2. Who Should Use the Guideline?

The guideline is intended for use by experienced engineers and analysts. A good level of
understanding of basic pipeline engineering is assumed. A brief introduction to lateral
buckling and the relevant design concepts is given in Section 2. The guideline does not
provide a detailed description of all the phenomena of interest, however these are discussed
fully with the SAFEBUCK deliverables (see Appendix F for a full list of deliverables).
1.2.3. Guidance Notes

The guideline is written to be clear and easy to follow. However, there are areas where the
information available is uncertain or only tentative guidance can be provided. Where this is
the case notes are provided to assist the user; these are presented in the following format:-
This is a guidance note.
In some areas the guideline recommendations are based on partial information or involve
extrapolation from a limited research base. In these cases a note is included to highlight the
uncertainty in the guidance:-
This is warning note.
Report No: 5087471/ 01 / A

Design Guideline
1.3. SCOPE OF DOCUMENT

The guideline is focused on the design of a subsea pipeline resting on a flat or modestly
undulating seabed, which is susceptible to lateral buckling. The guideline addresses the
behaviour of conventional pipelines (single walled pipelines) and pipe-in-pipe (PIP) systems.
1.4. STRUCTURE OF GUIDELINE

The aim of the guideline is to facilitate the safe, economic design of pipelines that are
susceptible to lateral bucking. In order to achieve this aim the integrity issues associated
with lateral buckling must be considered at all phases of the pipeline design and operation.
Consequently, the guideline covers the following phases of the pipeline life cycle:-
 Conceptual design;
 Detailed design and procurement;
 Installation and operational integrity.
The basic design approach is applicable to either conceptual or detailed design. The
difference between the phases of design is simply the analysis tools employed, the
availability of design data and the relevant acceptance criteria.
At the conceptual stage the aim of the design process is twofold. Firstly, it is to establish
that the design is feasible, i.e. that the pipeline can be safely designed for the intended
service. Secondly, it is to identify the optimum pipeline design strategy, for example to
evaluate the most economic option between a thin walled CRA and a thick walled carbon
steel. In order to achieve these aims it is necessary to undertake rapid screening of
competing options. Consequently, this stage of the design process is based around the use
of simple analytic tools.
At the detailed design stage the focus is on confirming the integrity of the chosen pipeline
system and optimising the details (for example identifying the optimum buckle initiation
strategy). As a result of the design complexity associated with lateral buckling, a full and
formal verification of the pipeline design should be undertaken during detailed design.
Depending upon the severity of the loading it may be necessary to define enhanced project
specific requirements in the pipeline specification.
There will always be an inherent uncertainty over the buckling response of any pipeline
system. As a result any project adopting a lateral buckling design strategy must undertake
significant post-installation work. This post-installation work must involve measurement of
the actual system performance (i.e. survey) and an update of the design analysis to re-
evaluate the integrity. Guidance on the level of inspection and analysis is provided within
the guideline.
1.5. DESIGN CODES

The guideline addresses the specific issues associated with structural integrity within a
lateral buckle. It does not address the full range of pipeline design activities. Consequently,
the guideline is designed to complement a recognised pipeline design code.
For the extreme loading conditions within a lateral buckle, the two most suitable design
codes appear to be DNV-OS-F101[1] and API 1111[2]. It is likely that both codes will be
employed by future projects; the choice may simply be defined by geography.
Consequently, the aim is that the guideline is consistent with either code. The main body of
the guideline is written around API 1111, modifications and interpretations required for
application to DNV-OS-F101 are given in Appendix C. These guidelines may be applied to
pipelines designed to other codes, provided that the implications of any differences between
the code and DNV-OS-F101 and API 1111 are addressed in a consistent manner.
Report No: 5087471/ 01 / A

Design Guideline
1.6. APPLICABILITY
In developing the guideline a range of pipeline parameters and design scenarios have been
considered. Strictly the guideline is therefore developed from, and is valid within, the set of
parameters outlined here.
The basic approach to the design problem will be applicable to many pipelines that
do not strictly comply with these limitations. However, in this case careful
interpretation of the guideline is required to ensure the inherent assumptions
employed in its development are not compromised and detailed verification of its
applicability is essential.
1.6.1. Pipeline Specification

The guideline has been developed for the following range of pipeline parameters:-
 D/t ratio between 10 and 50;
 Design temperatures up to 180ºC;
 Pipe initial ovality less than 1.5%.
The pipe ovality is defined in accordance with API 1111 (see Section 6.2.2).
1.6.2. Pipeline Material

The guideline has been developed for application to the following materials:-
 API 5L carbon-manganese steels up to grade X65 (carbon steel);
The methodology should be applicable to higher grade steels, such as X70 and X80.
However, for these materials the user must demonstrate that the imposed strain is
consistent with the material post yield characteristics.
 Corrosion resistant alloys; Duplex (22Cr), Super Duplex (25Cr) and 13%Cr stainless
steel (13Cr). These materials are collectively referred to as CRA within this guideline.
The SAFEBUCK JIP has not undertaken any work into the behaviour of clad or lined
pipe. However, the outline methodology can be applied to these pipes. There are
additional design considerations associated with these pipes; some guidance on this
is provided in the Guideline.
 Equivalent steel grades as defined in DNV OS-F101.
1.6.3. PIP Systems

The behaviour of PIP systems is addressed within the guideline. However, the work
performed in producing the guideline only considered a fully bonded PIP system. In the
bonded system, the insulation provides continuous shear transfer between the inner flowline
and the jacket pipe and enforces concentric bending. This leads to a system in which the
membrane axial strain in the two pipes is equal everywhere and the system bends as a
composite with equal curvature in both pipes. However, the outline methodology can be
applied to other types of PIP system; some guidance is given in the Guideline.
The other types of PIP connectivity that have been employed on PIP projects are:
 Regular bulkheads;
 Unconnected.
Report No: 5087471/ 01 / A

Design Guideline
In a regular bulkhead system there is a frequent structural connection (bulkhead)

between the two pipes, typically every two or so pipe joints. Between bulkheads
there may be intermediate spacers, but the main load transfer occurs at the
bulkheads. There is some freedom for the pipes to bend independently (although
there may be little gap inherent in the system). Here the axial strain in the two pipes
is not necessarily equal, however so long as the bulkheads are frequent the strains
are similar. However, at the crown of the buckle significantly different strains can
develop (subject to spacer and gap restraints) and enhanced localisation may occur.
In the unconnected system, the only structural connection is at the ends of the
pipeline (or spaced at a significant distance); typically structural bulkheads are
employed. Interaction between the two pipes is mainly governed by internal friction
of the system and the inner flowline moves axially relative to the jacket pipe. The
axial strains in the two pipes are not equal.
The general design philosophy and acceptance criteria outlined in this guideline can
be employed with these alternative PIP systems. However, the pipeline modelling
must address the degree of strain localisation and stress concentration that may
occur with these systems.
1.6.4. Fluid
The guideline is intended for pipelines transmitting hydrocarbon or water based products.
Project-specific testing may be required if the guidelines are to be applied to hydrocarbon
pipelines under sour service conditions.
This limitation is driven by the fatigue and fracture limit states, for which the presence
of H2S has a detrimental effect.
The SAFEBUCK JIP has performed extensive testing to investigate the effect of sour
service on fatigue and fracture. This information is included in the guideline as
information. However, the range of tests is insufficient to provide general guidance
and projects should investigate the effect based on their own environment and
loading conditions.
1.6.5. Installation Plasticity

The SAFEBUCK JIP has not undertaken any specific work to consider pipelines that are
installed by reeling. However, the outline methodology can be applied to reeled pipelines;
guidance is given in the Guideline where appropriate.
If the pipelines are installed by reeling then the installed pipeline will have
experienced significant plastic straining; will contain reeling residual stresses and
perhaps residual curvatures. There is no evidence to suggest that this is detrimental
(and may be beneficial with respect to buckle formation). In addition, reeling will
have an impact on the fracture response of the pipeline; it may modify the fracture
toughness of the pipe welds and may cause some ductile crack extension of pre-
existing weld defects.
For reeled pipelines, the user must ensure that all relevant influences of reeling have
been considered in the design process.
Report No: 5087471/ 01 / A

Design Guideline
1.6.6. Inspection
The guidance given assumes that the pipeline girth welds are subject to 100% non-
destructive testing. The inspection method (e.g. radiography, manual ultrasonics or
automatic ultrasonics) must be appropriate to the size and type of flaws that need to be
detected, as determined in the ECA.
When defect acceptance criteria based on ECA are employed ultrasonic inspection is
normally required. However, radiography may be sufficient if the acceptable defect
size complies with the workmanship levels defined in the relevant welding
specification.
If there is certainty about where the pipeline will buckle, then it is possible to designate such
areas as „fatigue sensitive zones‟ and apply enhanced inspection methods in those regions
only.
1.6.7. Over-matching
The guidance given assumes that the girth welds are over-matched.
In this context the girth weld strength should be greater than the pipe strength over
the whole strain range of interest and over the whole thermal range of interest. This
is a more demanding requirement than simply over matching the SMYS.
1.6.8. Seabed Flatness

The guideline is focused on the design of subsea pipelines resting on a relatively flat
seabed.
In this context relatively flat implies that the pipeline will not be susceptible to the
development of spans that are suspended significantly above the seabed.
The design approach outlined in the guideline does not address the loads developed
in significant spans, i.e. a basic assumption within the work performed is that
downward vertical displacement of the pipeline is prevented by the seabed. If a load
controlled approach to spanning is adopted (as recommended by DNV-RP-F105[3] for
example) then the local buckling design is likely to be inconsistent with the strain
based approach outlined here. As a rule of thumb, any pipeline in which the
maximum gap beneath the pipeline within a span exceeds the pipeline diameter is
out with the scope of this document.
Modest vertical OOS will increase the frequency of buckling, and is generally
beneficial to the pipeline design. The beneficial effect can be incorporated into the
SAFEBUCK design methodology (Section 4).
1.6.9. Fishing Interaction

Fishing interaction has not been considered during the development of this guideline.
Fishing interaction is usually driven by local buckling considerations during the
pullover event. There is no reason why this limit state cannot be evaluated using the
equations outlined in this guideline. However the fatigue and fracture limit states
associated with any denting that may result from the impact are not addressed in this
guideline.
Report No: 5087471/ 01 / A

Design Guideline
1.7. DESIGN EXAMPLE

In order to illustrate the use of the guideline design examples have been undertaken. These
are presented in Appendix A.
1.8. NOTATION, ABREVIATIONS AND DEFINITIONS
1.8.1. Notation
The following notation is employed in this design guideline.
Roman Symbols
A Constant in SN relationship -
2
Ae External cross sectional area of pipe m
2
Ai Internal cross sectional area of pipe m
2
As Steel cross sectional area of pipe m
-1
Aα Constant in definition of coefficient of thermal expansion °C
C Constant in stress-strain curve -
D Pipeline steel mean diameter m
Deff Effective PIP diameter m
Dfat Fatigue damage ratio -
Di Pipeline steel inside diameter m
Do Pipeline steel outside diameter m
DOC Pipeline overall diameter including external coatings m
e Weld centre-line misalignment m
e Exponential constant -
2
E Young‟s modulus N/ m
EA Pipeline axial stiffness N
2
EI Pipeline bending stiffness Nm
FL Cumulative probability distribution for total length of pipe
F100 Cumulative probability distribution for 100m length of pipe
f Axial pipe soil resistance N/m
fB Bauschinger factor -
fUB Maximum axial pipe soil resistance N/m
fLB Minimum axial pipe soil resistance N/m
fp Strain enhancement factor due to internal overpressure -
f2 API 1111, operational bending safety factor -
g API 1111, collapse reduction factor -
H Height of vertical trigger m
H Lateral pipe soil resistance N/ m
HBE Best estimate lateral breakout resistance N/ m
HLB Minimum lateral breakout resistance N/ m
Report No: 5087471/ 01 / A

Design Guideline
HUB Maximum lateral breakout resistance N/ m

HUBRes Maximum lateral residual resistance N/ m
4
I Pipeline second moment of area m
-1.5
K Stress intensity factor Nmm
Km Girth weld misalignment SCF -
-1.5
Kmax Maximum stress intensity factor Nmm
-1.5
KISCC Threshold stress intensity factor for susceptibility to stress corrosion cracking Nmm
-1.5
KIH Fracture toughness (stress intensity factor) with hydrogen embrittlement Nmm
Kt Thickness effect SCF -
k Thickness exponent -
L Total pipeline length or pipeline length between buckles m
LB The overall buckle length m
LBH The overall buckle length based on Hobbs analysis m
LBT The overall buckle length of an engineered buckle m
Lrmax Load cut-off Ratio in fracture assessment -
m Exponent in SN relationship -
-2
mα Constant in definition of coefficient of thermal expansion °C
n Number of fatigue cycles -
n Exponent in stress strain curve (hardening coefficient) -
N Effective axial force (compression positive) N
N Number of cycles -
NB The post buckle force (compression positive) N
NBH The post buckle force based on Hobbs analysis (compression positive) N
NBT The post buckle force of an engineered buckle (compression positive) N
Nchar Characteristic buckling force N
NcharB Characteristic buckling force when distributed buoyancy is employed N
Ncr Critical buckling axial force (compression positive) N
NcrB Critical buckling axial force at a route bend (compression positive) N
N̂ cr Maximum critical buckling axial force (compression positive) N
N̂ crT Maximum critical buckling axial force at a trigger (compression positive) N
Ni Allowable number of fatigue cycles at a given stress range -

N∞ Hobbs safe force for infinite mode buckling (compression positive) N
Nend Pipeline end reaction (compression positive) N
Nfmax Maximum compressive effective axial force in short pipeline (compression N
positive)
NL Residual lay tension (compression positive) N
Nmax Maximum compressive effective axial force in pipeline (compression positive) N
Nmin Minimum compressive effective axial force that supports buckling N
(compression positive)
Report No: 5087471/ 01 / A

Design Guideline
No Full constrained effective axial force (compression positive) N

NV Critical buckling force for a vertical imperfection N
2
Pc Collapse pressure N/ m
2
PE Elastic collapse pressure N/ m
2
PYC Plastic collapse pressure N/ m
2
pe External pressure N/ m
2
pi Internal pressure N/ m
2
piL Internal pressure during installation N/ m
2
QY True yield stress N/ m
R Target bend radius m
s Bend arc length km
S Effective axial force (tension positive) N
2
S Stress range in SN calculation N/ m
SL Residual lay tension N
So Full constrained effective axial force N
SRC Effective axial force at which route bend is unstable N
Su Shear strength of soil - at the pipe invert (bottom of pipe) kPa
Sw Axial force in pipe wall N
t Nominal pipe wall thickness m
ts Nominal pipe wall thickness of carbon steel in a clad or lined pipe m
tc Nominal pipe wall thickness of cladding in a clad or lined pipe m
tref Reference thickness in Kt calculation m
v Non-dimensional vertical pipe download -
V Vertical pipe download N/m
V‟ Effective vertical pipe download (submerged) N/m
Ws Pipeline submerged weight N/m
W sB Pipeline submerged weight when distributed buoyancy is employed N/m
X Virtual anchor spacing m
Xchar Characteristic Virtual anchor spacing m
XDF Characteristic VAS limited by driving force m
XL Characteristic VAS limited by pipeline length m
XNH Normalised critical force parameter – straight pipe -
XNB Normalised critical force parameter – route bends -
XNBo Normalised critical force parameter – buoyancy -
XNV Normalised critical force parameter – vertical imperfections
XS Characteristic VAS limited by sharing m
2
Y Pipe material SMYS N/ m
YT Yield strength to tensile strength ratio -
Report No: 5087471/ 01 / A

Design Guideline
Greek Symbols
 Coefficient of thermal expansion -
αfat Allowable fatigue damage ratio -
δ Pipe ovality, defined in accordance with API 1111. -
Δθ Operating temperature difference (from installed temperature) °C
εb Bending strain -
εc Critical buckling strain -
ε2 Allowable bending strain -
εeq Equivalent strain -
p
εeq Equivalent plastic strain -
εeqN Neuber equivalent strain -
εL Engineering Lüder strain -
p
εH Hoop direction plastic strain -
p
εL Longitudinal direction plastic strain -
p
εR Radial direction plastic strain -
εy Engineering yield strain
-1
κ Maximum bending curvature at buckle crown m
λ Logarithmic strain -
λL Logarithmic Lüder strain -
λp Plastic logarithmic Lüder strain -
λY Logarithmic yield strain -
μA Axial friction factor (or equivalent coefficient) -
μL Lateral friction factor (or equivalent coefficient) -
μmax Maximum axial friction factor -
μXnb Mean value of XNB -
ν Poisson‟s ratio -
2
ζeq Equivalent stress N/ m
2
ζeqN Neuber equivalent stress N/ m
2
ζf Flow stress N/ m
2
ζh Hoop stress N/ m
2
ζLf Longitudinal Flow stress N/ m
2
ζR Nominal axial stress range N/ m
2
ζRL Local axial stress range N/ m
2
ζy Yield stress N/ m
2
ζyc Compressive Yield strength in the hoop direction N/ m
θ Pipeline Temperature °C
θL Pipeline Temperature during installation (when constraint occurs) °C
Report No: 5087471/ 01 / A

Design Guideline
1.8.2. Abbreviation
The following abbreviations are employed in this design guideline.
API American Petroleum Institute

AUT Automated Ultrasonic Testing
CP Cathodic Protection
CT Compact Tension
CRA Corrosion Resistant Alloy
CTOD Crack Tip Opening Displacement
DTM Digital Terrain Mapping
DVL Doppler Velocity Log
ECA Engineering Critical Assessment
FAD Failure Assessment Diagram
FE Finite Element
FTA Flowline Termination Assembly (also see PLET)
HFI High Frequency Induction
HIPPS High Integrity Pressure Protection System
HPHT High Pressure / High Temperature
H2S Hydrogen Sulphide
INS Inertial Navigation System
JIP Joint Industry Project
KP Kilometre point
LRF Life Reduction Factor
MBES Multi-Beam Echo Sounder
NDT Non-Destructive Testing
OOS Out Of Straightness
PDF Probability Density Function
PIP Pipe in Pipe
PLET Pipeline End Termination (also see FTA)
ppm parts per million
PSI Pipe Soil Interaction
ROV Remote Operated Vehicle
SCF Stress Concentration Factor
SCR Steel Catenary Riser
SENB Singe Edge Notch Bend
SENT Singe Edge Notch Tension
SMYS Specified Minimum Yield Stress
SNCF Strain Concentration Factor
SN Stress-Life
Report No: 5087471/ 01 / A

Design Guideline
SRA Structural Reliability Analysis

SRB Sulphate Reducing Bacteria
SSS Side Scan Sonar
UC Unity Check
UOE „U‟, „O‟ then cold expand (E)
USBL Ultra Short Base Line
UT Ultrasonic Testing
UTS Ultimate Tensile Strength
VAS Virtual Anchor Spacing
YT Yield to Tensile ratio
ZRB Zero Radius Bend
1.8.3. Definitions
The following terms are used in this design guideline.
Buckle Trigger An engineered feature installed to initiate buckling.

Connector A mechanical device used to attach sections of pipeline, typically
employed in PLETs.
Characteristic VAS The Characteristic VAS is the VAS that the design strategy
guarantees will not be exceeded to an acceptable level of reliability.
Clad Pipe A pipe with an internal (corrosion resistant) liner, where the bond
between the backing steel and the liner material is metallurgical.
Engineered Buckle A buckle that occurs at a buckle trigger.
Fully mobile A pipeline is fully mobile if there are no locations of full constraint
during load and unload.
Flowline The inner pipe in a PIP system.
Jacket The outer pipe in a PIP system.
Jumper A type of spoolpiece, generally utilising connectors.
Lateral buckle Bar buckling of the pipeline (predominantly) in the horizontal plane.
Lined Pipe A pipe with an internal (corrosion resistant) liner, where the bond
between the backing steel and the liner material is mechanical.
Lüder Plateau A section on the stress-strain curve in which the strain increases with
no increase in stress following first yield.
Lüder Strain The strain at which a material which exhibits a Lüder plateau starts to
harden.
Out of straightness The geometric imperfections associated with a pipeline.
Rogue Buckle An unplanned buckle that forms remote from buckle trigger locations.
Tolerable VAS The tolerable VAS is the maximum value of VAS at which all design
unity checks are less than one.
Uncontrolled Buckling Uncontrolled buckling describes the buckling behaviour of a pipeline
due to inherent OOS. It results when the pipeline is susceptible to
buckling, but no engineered buckle initiators are employed.
Report No: 5087471/ 01 / A

Design Guideline
Uniform Strain Capacity The engineering strain at UTS.

Unity Check A design check expressed as the ratio of the imposed load divided by
the allowable load. The check incorporates all design factors. A
value less than or equal to one implies the design check is met, a
value greater than one implies that the design check is not met.
Upheaval buckle Bar buckling of the pipeline (predominantly) in the vertical plane.
Virtual Anchor Spacing The spacing between the virtual anchor points adjacent to a buckle.
13Cr Class of CRAs with approximately 13% Chromium content.
22Cr 22%Cr Ferritic-austenitic (duplex ) steel.
25Cr 25%Cr Ferritic-austenitic (super duplex ) steel.
1.9. UNITS
All information is presented in metric units.
Unless stated otherwise, all parameters (stress, strain, force) are tension positive.
The main area where this convention is violated is in the definition of effective axial
force, which is sometimes tension positive and sometimes compression positive. To
clarify the position the symbol S is used when the force is tension positive and N is
used when the force is compression positive.
1.10. DISCLAIMER
All reasonable efforts were made to ensure that the work contained within this guideline
conforms to accepted scientific and engineering practices, but The SAFEBUCK JIP makes
no other representation and gives no warranty with respect to the reliability, accuracy,
validity or fitness of the information presented. Any use or interpretation of the information
contained in the guideline is undertaken at the user‟s own risk.
Report No: 5087471/ 01 / A

Design Guideline
2. BACKGROUND TO LATERAL BUCKLING

2.1. LATERAL BUCKLING
A pipeline laid on the seabed and operated at pressure and temperature above ambient will
tend to expand. If the expansion is restrained in some way, for example by the frictional
restraint of the seabed, then an axial compressive force will develop in the pipeline. If the
compressive effective axial force is large enough then the pipeline will undergo Euler
buckling (sometimes called global buckling).
If the pipeline is laid on a relatively flat seabed, then the Euler mode tends to be in the
horizontal plane (lateral buckling). If the seabed is more uneven, the initial movement may
be in the vertical plane (upheaval buckling), but this is likely to subsequently develop into a
lateral displacement. Only if a pipeline is trenched or buried will a significant buckle remain
in the vertical plane; upheaval buckling is not considered within this guideline.
Lateral buckles have been observed in many operating pipeline systems laid on the seabed
without trenching, an example is shown in Figure 2.1.
Figure 2.1 Side-scan sonar image of a Lateral Buckle

The lateral displacements involved are very significant; displacements of 10 or 20 diameters
are typical (in absolute terms displacement in the range 2 m to 10 m has been observed in
operating pipelines). The length of pipe over which the lateral displacement occurs is
typically between 100 m and 300 m (the lateral scale is exaggerated in Figure 2.1).
Such non-trenched pipelines are often large diameter and operate at low temperature so that
fortuitously, although the lateral buckles are often unplanned, they are lowly loaded and
generally benign. However, the acceptance of lateral buckling in such systems has helped
to establish lateral buckling as an attractive design solution for the relief of axial compressive
force in hotter pipelines. As the operating conditions increase the severity of the loads within
buckles increases. Controlling these loads is a significant challenge for HPHT (high-
pressure, high-temperature) developments, particularly in deep water where trenching and
burial is less feasible and considerably more costly than in conventional water depths.
The problem of lateral buckling in pipelines was first considered by Hobbs [4,5]. Experiments
performed as part of his work observed that the pipeline can deform into a number of
different lateral mode shapes; the most common buckle shape configurations are illustrated
in Figure 2.2.
Report No: 5087471/ 01 / A

Design Guideline
L
L
Mode 2
Mode 1
Mode 4
Mode 3
L L
Figure 2.2 Lateral Buckling Modes

The pipeline can buckle into either symmetric modes (modes 1 and 3 in Figure 2.2) or
asymmetric modes (modes 2 and 4 in Figure 2.2). The actual mode adopted at a lateral
buckle site depends upon a number of factors including the pipeline out-of-straightness
(OOS) and local seabed lateral restraint levels. The modes consist of either one or two
central half-waves, surrounded by a decaying sequence of half waves moving away from the
centre of the buckle. Mode 1 requires a significant concentrated reaction at the end of the
lobes, which the seabed cannot provide and is not normally observedi. The amplitude of
each half-wave decreases rapidly and, almost exclusively, modes shapes 2, 3 and 4
(defined in Figure 2.2) are observed in operating pipelinesii.
The attraction of the lateral buckling design solution is in the reduction in compressive axial
force that occurs in the pipeline. However, the buckling displacement results in significant
bending moment at the crown of the buckle. This moment can be very high and stresses in
excess of yield are normal (although these are relatively limited in axial extent). Thus, under
the lateral buckling philosophy, although the axial force is reduced, the bending stresses are
significantly increased. This can ultimately lead to a local buckling failure with associated
deformation of the pipe cross-section.
On unload the pipeline attempts to return to the as-laid position, but is prevented from doing
so by seabed resistance (both axial and lateral). As a consequence, the pipeline lateral
deformation is reduced, but not to zero; a typical pipeline response is illustrated in Figure
2.3.
i
However Mode 1 buckling is the normal mode of upheaval buckling. A mode 1 shape could only
arise when the lateral deformation is developed from an initial vertical (upheaval buckling)
displacement, which facilitates more concentrated reactions at the touchdown points (perhaps
through local pipe embedment). This has been observed in an operating pipeline, but is unusual.
Some lateral buckles can appear to have a mode 1 shape, particularly when the friction over the
central lobe is very low because it is resting on a sleeper off the seabed, while the outer lobes
become deeply embedded.
ii
In practice, local variations mean that the mode shapes are slightly different to the theoretical
shapes. Nevertheless it is generally possible to categorise actual buckles using these idealised
modes.
Report No: 5087471/ 01 / A

Design Guideline
3.5 Lateral
Displacement (m)
3
2.5
Full Load
2
Shutdown
1.5
1
0.5
0
-100 -80 -60 -40 -20 -0.5 0 20 40 60 80 100
-1 Distance along pipe (m)
-1.5
Figure 2.3 Cyclic Response within a Buckle

The unload deformation is still significant; in this example the crown displacement on unload
is in excess of 1.7 m. As the pipeline experiences fluctuations in pressure and temperature
it cycles back and forth across the same patch of seabed. Surface soil, swept ahead of the
pipe on each cycle, then builds-up into berms at the extremes of pipe displacement (these
are highlighted in Figure 2.1). Subsequent consolidation increases the strength of the soil
berms after disturbance. The soil berms restrict the growth of the buckle so that cyclic
displacements and buckle curvature remain almost constant over a number of cycles, rather
than reducing and relaxing with repeated cycles.
The start-up and shut-down cycle imposes significant stress ranges on the pipeline. Cyclic
loading is generally in the elastic range, although the magnitude of the stress range can be
greater that the uniaxial yield stressiii. This is a high-stress, low-cycle fatigue regime, which
can lead to a fatigue failure of the pipeline.
2.2. BUCKLE DEVELOPMENT

Buckle formation is an imperfection sensitive process and it is difficult to predict the location
and number of buckle sites. The key unknowns are seabed response (pipe-soil friction) and
the size and distribution of initial out of straightness imperfections. The response of a pipe to
various imperfection sizes is illustrated in Figure 2.4.
iii
The cyclic stress range can be as high as twice the uniaxial yield without involving cyclic plastic
strain. In reality the maximum elastic stress range is further limited by the Bauschinger effect to a
maximum of 1.4 to 1.6 times the uniaxial yield stress.
Report No: 5087471/ 01 / A

Design Guideline
Hobbs Solution Hobbs Solution

(perfect pipe) (perfect pipe)
Force Remote from Buckle
Snap
Force
Nmin Remote from
Buckle
In Buckle
A B C B
Amplitude Amplitude
(a) Effect of Imperfection Size (b) Force Remote from and in Buckle
Figure 2.4 Response of Imperfect Pipe
The bold curve is the analytic equilibrium solution for a perfect pipe outlined by Hobbs iv. If a
small imperfection is present in the pipeline, the response is illustrated by Curve A. The
force in the pipeline increases as the system heats up. Very little movement occurs and the
force builds up to a value in excess of the minimum equilibrium force. At some point the
force approaches (but doesn‟t quite reach) the equilibrium curve. At this side of the
minimum force curve, the pipeline is in unstable equilibrium. Any increase in the force in the
pipeline causes a bifurcation and the pipeline snaps through to a stable equilibrium on the
other side of the U shaped curve, as illustrated by the arrow. Further increase in imposed
force causes further (stable) increase in the amplitude of the lateral buckle.
A different response occurs if large imperfections are present in the system; this is illustrated
by curve B or C. Initially the force increases as for curve A, however, initial movement now
occurs at a force below the minimum force. In addition, the displacement is not
accompanied by a snap type bifurcation. Instead the amplitude increases gradually towards
the stable theoretical equilibrium solution.
The force presented in Figure 2.4a is the force outside the buckle at the anchor point. The
variation in force within the buckle is illustrated in Figure 2.4b. The figure shows that the
force within the buckle continues to drop as the amplitude increases. A force difference
develops between the anchor point and the buckle as soon as movement commences, i.e.
the reduction in driving force occurs as soon as the buckle starts to move, irrespective of
snap behaviour. This force difference implies feed-in to the buckle along a slip zone. The
difference between the force within the buckle and at the anchor point continues to increase
– this means that the slip length into the buckle becomes ever longer as the driving force
increases.
iv
This curve describes the post buckle equilibrium configuration for the pipeline. The curve is U-
shaped, and the minimum value on the curve is sometimes called the “safe” force. For forces below
the safe force it is not possible to develop a stable post buckled configuration (that complies with the
inherent assumptions of the analysis).
Report No: 5087471/ 01 / A

Design Guideline
The formation of the buckle is the key uncertainty in the lateral buckling design. The
buckling process is very sensitive to initial pipe imperfections and interaction with the
seabed; neither of these parameters is known with any certainty. Consequently, it is
impossible to define exactly where, or how many, buckles will form. Within the SAFEBUCK
methodology, a structured approach to buckle formation is adopted to ensure that the design
is sufficiently reliable.
2.3. PIPELINE EXPANSION RESPONSE
2.3.1. Long Straight Pipeline, No buckling

The first load expansion response of a straight pipeline is illustrated in Figure 2.1.
Distance along pipeline

Effective Axial Force _
μA
Expansion
Compressive
Fully Constrained Force

Driving Force
(a) Effective axial force (b) Expansion

Figure 2.5 Effective Axial Force and Expansion in a Straight Pipeline
The force presented in Figure 2.5 is the effective axial force in the pipelinev. This is the force
that drives the structural response and is made up of the (true) force in the pipe wall and the
pressure induced axial force. The effective axial force in a pipeline is defined as:-
S  Sw  pe  A e  pi  A i .... 2.1
For a straight pipeline with fixed ends the axial strain is zero everywhere and the force
developed in the system is known as the fully constrained effective force, which is usually
defined as:-
S 0  SL  pi  piL   A i   1  2     E  A s     .... 2.2
v
Route changes in the pipeline may modify this force development.
Report No: 5087471/ 01 / A

Design Guideline
The fully constrained effective force does not depend upon the external pressurevi. Since
pressure and temperaturevii vary along the pipeline length, the fully constrained force also
varies with length (as the pipe cools and the temperature falls the effective axial force
becomes smaller, as shown in Figure 2.5).
For the pipeline with free ends the effective axial force is zero at the ends viii and gradually
increases due to the frictional restraint of the seabed. The slope of the force profile in the
slip zones is defined by the axial frictionix, AW s. At some point the frictional restraint is
sufficient to suppress any expansion and the axial strain in the pipe is zero. Clearly, the
force at this point is the fully constrained effective force defined above.
A similar response occurs at the cold end of the pipeline and the figure shows that the
system develops a fully constrained section in the middle of the pipeline. This is bounded by
points known as virtual anchor points. In the absence of buckling, there is no displacement
over the central section (the pipe is fully constrained) and the pipe expands between the
virtual anchor and the pipe end, reaching a maximum displacement at the pipe end (Figure
2.5b).
2.3.2. Long Straight Pipeline, with Buckling

If the flowline is allowed to buckle the development of effective force is modified. The force
in the buckle drops as the buckle develops. The change in expansion response is presented
in Figure 2.6.

straight buckled
X1 X2 X3
Expansion
Virtual Distance
Lateral Anchors
Buckle

Post Buckling Force Buckle #1 Buckle #2 Buckle #3

Figure 2.6 Effective Axial Force and Expansion in a Buckled Pipeline
vi
However, the external pressure still acts to develop hoop stress, and the wall force is still calculated
from the effective force through equation 2.1.
vii
Strictly, this is the pressure and temperature difference (the difference between the values during
operation and at installation).
viii
If the pipeline is connected into expansion spoolpieces (or PLETS), there will be a reaction to
expansion that means that the effective force at the end has a small compressive value. If the
pipeline is directly connected to a steel catenary riser, the effective force at the end will be tensile.
ix
This assumes that the friction is fully mobilised. For very small axial slip (or large mobilisation
distances) the slope can be lower than the limiting value, i.e. aW s is not fully attained.
Report No: 5087471/ 01 / A

Design Guideline
In the example shown the pipeline has formed three buckles (the number of buckles is
entirely case dependant) and the force profile is somewhat more complex. At each buckle
site the axial force drops and the pipe feeds-in from each side to provide the additional pipe
length required to form the buckle. The force in each buckle varies from site to site;
depending upon the initial out-of-straightness, seabed frictional response and buckle
spacing.
The expansion behaviour along the pipeline is illustrated in Figure 2.6b. In all slip zones the
slope of the force profile is the samex – governed by the axial friction. Clearly, between
adjacent buckles there must be a point at which the direction of pipe expansion changes –
this is termed a virtual anchor point. The system is effectively divided into three short
pipelines anchored at each end. These virtual anchors are positions of zero displacementxi
but not zero strain. The VAS associated with each buckle varies, and is labelled X 1 to X3 in
Figure 2.6b.
2.3.3. Short Pipelines

If the pipeline is short, the overall length may be insufficient to fully restrain the pipeline; this
response is illustrated in Figure 2.7.
Distance along pipeline straight buckled
X1
Expansion
Compressive
Fully Constrained Force Distance

Straight
Buckled
Buckle

Figure 2.7 Effective Axial Force in a Short Straight Pipeline
x
This is true if the axial resistance does not vary with location. In practice the axial resistance will
vary from point to point, with local soil properties and embedment.
xi
These points are not actually positions of zero displacement. Movement can occur during buckle
development.
Report No: 5087471/ 01 / A

Design Guideline
A free-ended short pipeline never reaches a position of full constraint; instead the pipeline
forms a virtual anchor at the centrexii and expands from this point. The maximum axial force
in the pipeline can be significantly below the fully constrained force, but can still be sufficient
to cause buckling. In the event of buckling, the pipe will expand into the buckle and out
towards the ends. There must be a point at which the direction of pipe expansion changes –
i.e. a virtual anchor point. The system effectively contains a short pipeline anchored at each
end – as before the virtual anchors are points of zero displacement but not zero strain. The
VAS associated with the buckle is labelled X1 in Figure 2.7.
In the absence of very high end reactions (from the flowline spoolpieces), the maximum
possible buckle spacing for the short pipe is half of the overall length of the flowline (in
practice the force within the buckle normally reduces this maximum spacing to less than half
of the overall length). The worst place for a buckle to form is normally at the centre of the
flowline.
2.3.4. Lowly Loaded Pipelines

If the pipeline operating conditions are relatively low, it is possible that the feed-in to the
buckle is limited by the fully constrained force, Figure 2.8.


Driving Force
μA
Compressive
X1 X2
Figure 2.8 Effective Force Profile for Buckles in a Lowly Loaded Pipeline
In this case, although the force in the buckle drops it is a relatively high proportion of the fully
constrained force. The VAS is defined by the intersection of the feed-in zones and the fully
constrained force. Two buckles are illustrated in Figure 2.8, although any number is
possible, and these are isolated by a section of fully constrained, straight pipe. This is the
problem of an isolated buckle in an infinitely long pipeline, originally considered by Hobbs [4].
In this case the VAS cannot exceed the value defined by the fully constrained force profile,
although it still varies with position as illustrated.
xii
The virtual anchor does not necessarily form exactly at the centre; variations in axial friction or spool
piece restraint act to move it away from the pipeline centre.
Report No: 5087471/ 01 / A

Design Guideline
2.4. VIRTUAL ANCHOR SPACING
2.4.1. VAS Concept

The behaviour described in the previous sections illustrates the concept of VAS. The
presence of the virtual anchor points, effectively divides the overall line into a series of short
flowlines, which are fully constrained at their ends. In Figure 2.6 to Figure 2.8 these section
lengths are labelled X1 to X3. The response of the flowline between virtual anchors is
analogous to the response between real anchors. For all pipelines, long and short, the
loading caused by pipe feeding into the buckle is characterised by the VAS; the VAS defines
the length of pipe feeding-in to the buckle.
This concept of VAS is fundamental to the SAFEBUCK design methodology and the VAS is
a key parameter in the lateral buckling design process. The closer the buckles are to each
other (the smaller the VAS) the less the axial feed-in which must be absorbed by the lateral
deflection at the buckle. The aim of the design method is for a number of buckles to form at
regular intervals along the flowline. This produces a solution in which the expansion strain is
shared between each site, leading to manageable strain within each buckle.
The design process involves the calculation of two parameters; the Tolerable VAS and the
Characteristic VAS. The Tolerable VAS is the maximum VAS at which all of the relevant
design criteria are met. The Characteristic VAS is the maximum VAS that can be
guaranteed by the design strategy to an acceptable level of reliability. The design process
then becomes a matter of identifying the maximum tolerable VAS and ensuring that buckles
will form regularly enough to ensure this spacing limit is not violated.
In general these tasks lead to competing requirements; the Tolerable VAS requirement leads
to closely spaced buckles whereas the Characteristic VAS (buckle formation reliability
requirement) leads to widely spaced buckles. The SAFEBUCK design guideline outlines a
structured approach to resolving this design problem.
2.4.2. Influence of Bathymetry

The pipeline expansion will be modified by the route bathymetry. If the seabed is relatively
flat then the bathymetry will have only a modest effect on the response. Minor spans will
absorb some initial expansion, but the strain capacity is small before the spans “lock up”.
Once this occurs the expansion will manifest itself in the manner outlined in Section 2.3.
For more severe bathymetric variation, the pipeline expansion may be more complex than
outlined in Section 2.3. If span heights are large enough to prevent touch down then
significant feed-in expansion will occur towards these spans. In this case the span will affect
the force profile in much the same way as a lateral buckle. Virtual anchor points will form
either side of the major spans, in the same way as described for lateral buckles. In extreme
cases the load in the span can exceed the design load. In other cases, where there is
sufficient feed-in to the span, the high points to each side of the span may buckle upwards
and then displace outwards to form a lateral buckle to one or both sides of the span.
The influence of bathymetry does not compromise the use of the VAS concept. The design
process will still involve the identification of the maximum tolerable VAS and ensuring that
buckles will form regularly enough to ensure this spacing limit is not violated.
The tolerable VAS is largely unchanged; in fact it is likely to be increased as the effect of
modest undulations is similar to the effect of residual lay tension (Section 5.3.3).
Report No: 5087471/ 01 / A

Design Guideline
The Characteristic VAS will be reduced in two ways. Firstly, any significant vertical OOS
may promote buckling. Secondly, significant spans will absorb expansion in a similar way to
engineered triggers. If the reliability of these effects can be demonstrated, then this can be
incorporated into the calculation of Characteristic VAS.
Seabeds with severe bathymetric variation have not been considered in detail in the
SAFEBUCK JIP. However, there is no reason why the design methodology cannot
be applied to these cases.
2.5. PIPE WALKING

Pipeline walking can occur for short free-ended pipelines subject to cyclic loading. Pipeline
walking is a phenomena in which start-up/shut-down cycles cause a ratcheting response in
the pipeline axial displacement[6]. Over a number of cycles this ratcheting can lead to very
large global axial displacement with associated overload of the jumper or spoolpiece. The
mechanism mainly affects short, hot pipelines. However, lateral buckles essentially divide
the pipeline into a series of short pipelines; therefore the mechanism is relevant to any
pipeline that adopts the lateral buckling solution[7].
The propensity of the flowline to walking is governed by the pipeline length (or length
between buckles); the axial friction; the maximum inlet temperature; the steepness of
thermal start-up transients; and the steepness of seabed slopes along the flowline. In multi-
phase systems the change in contents density between operation and shutdown can
accelerate the walking mechanisms[8]. For pipelines connected to an SCR, walking is also
driven by the tension in the touch-down zone exerted by the SCR; in this case SCR tension
is likely to dominate the rate of walking.
Pipeline walking should be evaluated as part of the lateral buckling design.
Report No: 5087471/ 01 / A

Design Guideline
3. DESIGN PROCESS
3.1. METHODOLOGY OVERVIEW
The design methodology is summarised in Figure 3.1.
Select basic design
parameters
Modify design parameters Modify design parameters
No
No Is pipeline
Is walking OK? susceptible to
buckling?
Yes
Yes
Is
Yes uncontrolled
buckling
acceptable?
No
Define initiation strategy Modify initiation strategy
Yes
Is Can
controlled No initiation No
buckling strategy be
acceptable? improved?
Yes
buckle
No
interaction and
walking OK?
Yes
Design Complete
Figure 3.1 Basic Design Methodology Philosophy

The initial step is to outline the basic design parameters. This covers the pipe geometry and
material properties, the operating conditions and the pipe-soil interaction model. Guidance
on these parameters is given in Section 5.4 to 5.7.
The design process first evaluates whether the pipeline is susceptible to lateral buckling
(Section 3.2). If it is not there is still a requirement to ensure that pipeline walking is not a
design concern. This can be evaluated in accordance with Section 3.5.
This situation is most likely to occur for a short hot pipeline. The force developed
may be governed by the axial friction and may be insufficient to cause buckling.
However, the shutdown/restart cycle may still be sufficient to induce walking.
If the pipeline is susceptible to lateral buckling a staged design process is followed. Initially,
the design process checks to see if uncontrolled buckling is acceptable (Section 3.3). If
uncontrolled buckling is not acceptable then an engineered initiation strategy is required
(Section 3.4).
Uncontrolled buckling simply means lateral buckling that occurs where no buckle
mitigation methods are implemented. In this case all buckles that form in the pipeline
are classified as rogue buckles. Although it is not possible to guarantee the exact
location of rogue buckles, it is often possible to show that rogue buckling does not
threaten the pipeline integrity.
Report No: 5087471/ 01 / A

Design Guideline
Once an acceptable lateral bucking strategy has been developed, then buckle interaction
and pipeline walking are evaluated (Section 3.5).
The basic design methodology is applicable to either conceptual or detailed design; the
difference between the phases of design is simply the analysis tools employed and the
relevant acceptance criteria.
3.2. SUSCEPTIBILITY TO BUCKLING

At this stage the susceptibility of the system to lateral buckling should be established.
3.2.1. Go/No Go Criteria

The pipeline is susceptible to buckling if the following inequality is true:-
Nmax  Ncr .... 3.1
The maximum compressive effective axial force in the system is defined as:-
Nmax  minN0 , Nf max .... 3.2
Where:-
N0  pi  piL   Ai  1  2    E  A s      L   NL .... 3.3
L
Nf max  Nend  fUB  .... 3.4
2
The fully constrained effective force (N0) governs for long pipelines and the maximum
available frictional force (Nfmax) governs for short pipelines. If the pipeline is anchored
at both ends, the maximum force in the system is the fully constrained effective force
irrespective of length.
The forces outlined in Equations 3.1 to 3.6 are compression positive. Consequently,
in equation 3.3 the lay tension, NL, would normally be negative.
In equation 3.3 the internal pressure term is the difference in internal pressure
between operation and installation. Under normal circumstances the internal
pressure at installation is surface ambient pressure and is usually neglected.
However, if the pipe is laid flooded then the internal pressure at installation is the
local seabed ambient pressure; this is beneficial in reducing the fully constrained
force, buckle susceptibility and associated end expansion.
The critical buckling force is defined as:-
Ncr  min 0.65  N , NcrB 
Where:-
NCrB  HLB  R .... 3.5
EI  HLB
N  3.86  .... 3.6
D
Where HLB is the minimum lateral break-out resistance and N∞ is the minimum force under
which a buckle in a straight pipeline can develop; this is derived from the equations of Hobbs
for an infinite mode buckle.
Report No: 5087471/ 01 / A

Design Guideline
Equation 3.5 defines the critical buckling force associated with any large radius bend
in the pipeline route.
In evaluating Equation 3.5 and 3.6 the minimum pipeline submerged weight should be
employed. This should include an allowance for hydrodynamic lift during storm events.
Storm events may lead to breakout and lateral buckling, particularly if the pipe is not
vertically stable under maximum wave forces. This is a complex dynamic interaction
that requires specific study.
Tentatively it is suggested that the maximum lift force associated with the appropriate
return period significant wave is employed in the design check.
3.2.2. Probabilistic Criteria

In some cases the inequality 3.1 may not be met, but the probability of buckling is low. In
this case it may be possible to demonstrate that although buckling is possible, the probability
of buckling is acceptably low and no further analysis is required.
The pipe is not susceptible to buckling if the probability that at least one buckle forms in the
pipeline is less than 5%. The probability of at least one buckle forming in the pipeline can be
calculated in accordance with Section 4.
An example where this approach might yield a different result to the simple inequality
would be a short pipeline in which the peak driving force under maximum axial
friction only slightly exceeds the critical buckling force over a short length of pipe.
The definition of susceptibility is intended to be consistent with an overall pipeline
failure probability of 10-3 to 10-4. However, a comprehensive calibration analysis has
not been undertaken.
3.2.3. Susceptibility with Length

The susceptibility to buckling will normally vary along the length of a pipeline. The pipe is
not susceptible to buckling in any region where the probability of buckle formation is less
than 1%/km (i.e. less than 1% in any given km of the pipe). The probability of formation can
be calculated in accordance with Section 4.
The most likely application of this is for pipelines in which the operating conditions fall
significantly with length and can be used to define an end to buckle susceptibility.
3.3. ASSESSMENT OF UNCONTROLLED BUCKLING

The uncontrolled lateral buckling behaviour of a pipeline is acceptable if the design unity
check for all relevant failure modes is acceptable (i.e. less than one) at the Characteristic
VAS. The relevant failure modes are defined in Section 6.
To evaluate the design unity checks, a lateral buckling analysis must be undertaken at the
Characteristic VAS. The analysis should be undertaken in accordance with the modelling
requirements outlined in Section 5. The design unity checks should be undertaken at all
relevant locations along the pipe.
For short pipelines or highly insulated pipelines, the Characteristic VAS may not vary
significantly. In this case a single calculation, at inlet conditions would be adequate.
Report No: 5087471/ 01 / A

Design Guideline
3.4. ASSESSMENT OF CONTROLLED BUCKLING

If uncontrolled buckling is not acceptable, then a buckle initiation strategy must be
developed. The aim of the initiation strategy is to ensure that the loads within the
engineered buckles and any rogue buckles that could form are acceptable.
3.4.1. Select Initiation Strategy

There are a number of methods which have been employed or are currently proposed to
initiate buckling at a controlled spacing. These “triggers” include:-
 Snake-lay;
 Vertical upset;
 Zero radius bends;
 Local weight reduction.
These initiation strategies have different implications for the pipeline design. The selection
process depends on many factors such as severity of operating conditions, environmental
conditions, water depth, installation vessel, cost and schedule. The key considerations for
each technique are discussed in Appendix E.
3.4.2. Define Trigger Spacing

In developing the initiation strategy it is necessary to understand how frequently buckles are
required. The trigger spacing must ensure that the loads in the engineered buckles are
acceptable. However, the trigger spacing must also ensure that rogue buckling is tolerable.
To guide the development of the design strategy it is helpful to calculate the Tolerable VAS.
The Tolerable VAS is the maximum value of VAS at which all design unity checks are less
than one.
Calculation of the Tolerable VAS is an iterative process; this is outlined in Figure 3.2.
Initial estimate of Tolerable

VAS
Perform analysis of buckle

for selected VAS
Perform all relevant limit

Increase VAS
state checks
Yes
Max Max
UC <1 Calculate UC >1
Can the VAS UC for each Decrease VAS
be increased? limit state
Max
No UC =1
Buckling tolerable Tolerable VAS
Figure 3.2 Flow Chart for Identification of Tolerable VAS
Report No: 5087471/ 01 / A

Design Guideline
Within Figure 3.2 there is a check on whether the VAS can be increased. There is an upper
limit on VAS; in long lines this is normally the unconstrained VAS (Section 2.3.4); in short
lines it is related to the pipeline length (normally half the pipe length). If all unity checks are
acceptable at this upper limit then buckling is acceptable and no initiation strategy is
required. This situation may apply over the full length in a lowly loaded pipeline, or over part
of the length for a pipeline whose operating conditions fall with distance.
If the design process is followed this situation should not arise for the rogue buckles,
as this implies that uncontrolled buckling is acceptable (assessed in Section 3.3 ).
However, this situation may apply to the response at the triggers.
The lateral buckling analysis should be undertaken in accordance with the modelling
requirements outlined in Section 5.
The assessment will usually involve an analysis of buckling for a range of VAS. This is
desirable, and allows the sensitivity of the pipeline integrity to buckle spacing to be clearly
demonstrated.
The calculations should be performed at sufficient locations along the pipeline to develop the
variation of Tolerable VAS with KP.
For short pipelines or highly insulated pipelines, the tolerable VAS may not vary
significantly. In this case a single calculation, at inlet conditions would be adequate
The Tolerable VAS must be calculated for a planned buckle and a rogue buckle.
In general these will differ. A rogue buckle is an on-bottom buckle whose capacity is
influenced by the lateral pipe-soil resistance. Most initiation strategies modify the
influence of lateral restraint. As a result the feed-in capacity of the engineered and
the rogue buckle is different, and the Tolerable VAS will be different.
In general the trigger spacing should be no more than the minimum of the engineered buckle
Tolerable VAS or twice the rogue buckle Tolerable VAS.
This is based on the assumption that all the triggers fire. In this case the maximum
feasible VAS for a rogue buckle is half the distance between triggers. This strategy
will be adequate so long as the probability of a trigger failing is low. This is
addressed in the calculation of Characteristic VAS (Section 4).
It is possible that the tolerable VAS at the trigger may be much greater than the
tolerable VAS for rogue buckling. This can occur for initiation techniques that reduce
the loading within the buckle, for example distributed buoyancy. In this situation the
definition of trigger location will be driven entirely by the capacity of rogue buckles.
The location of the triggers may also be influenced by sensitive areas of the pipeline.
For example, the triggers may be employed to prevent buckling at a pipeline crossing
or a mid-line tie-in location.
The selection of trigger locations must ensure that the Tolerable VAS is greater than the
Characteristic VAS at all points along the pipeline.
3.4.3. Assessment of Controlled Buckling

The assessment must demonstrate the integrity of the engineered buckles and any rogue
buckles that could form. The relevant failure modes are defined in Section 6.
Report No: 5087471/ 01 / A

Design Guideline
3.4.3.1. Engineered Buckles

The lateral buckling behaviour at a trigger is acceptable if the design unity check for all
relevant failure modes is acceptable at the Characteristic VAS.
To evaluate the design unity checks, a lateral buckling analysis must be undertaken for a
buckle forming at the engineered site at the Characteristic VAS. The analysis should be
undertaken in accordance with the modelling requirements outlined in Section 5.
3.4.3.2. Rogue Buckles

Rogue buckling is acceptable if the design unity check for all relevant failure modes is
acceptable at the Characteristic VAS.
To evaluate the design unity checks, a lateral buckling analysis must be undertaken at the
Characteristic VAS. The analysis should be undertaken in accordance with the modelling
requirements outlined in Section 5. The design unity checks should be undertaken at all
relevant locations along the pipe.
3.4.4. Assessment of Strategy

If both the engineered and rogue buckles are acceptable, then the lateral buckling VAS
analysis is complete.
If either engineered or rogue buckling is unacceptable then the design is unacceptable. If
the initiation strategy can be improved then the optimum strategy should be identified and
the assessment repeated.
Improving the strategy may involve decreasing the spacing between triggers to
reduce the severity of rogues, optimising the chosen technique (for example
decreasing the radius of the as-laid snakes), or it may involve changing the initiation
technique (for example moving from snake lay to distributed buoyancy).
If the initiation strategy cannot be improved then the basic design parameters must be
changed and the design process repeated.
In this case a different pipeline configuration must be adopted, the severity of the
operating conditions must be reduced or mitigation measures must be adopted
(Appendix E).
3.5. BUCKLE INTERACTION AND PIPELINE WALKING

Once the lateral buckling design philosophy has been established, the cyclic expansion
behaviour of pipeline should be verified. This evaluation must incorporate the chosen lateral
buckling design strategy. The design should evaluate:-
 Buckle interaction and stability;
 Route curve pull-out;
 Influence of bathymetry;
 Pipeline walking.
The pipeline behaviour under cyclic loading can be extremely complex. It is possible
for buckles to disappear – especially if they are located near the pipeline end in the
end expansion zone. In addition, preferential growth may occur if the buckles are too
close together (this is only likely if the buckles are very close together – operational
experience is that buckles are quite stable once formed). A key aim of these
analyses is to ensure that undesirable cyclic behaviour does not occur.
Report No: 5087471/ 01 / A

Design Guideline
The aim of the analysis is to confirm the overall expansion behaviour and evaluate
pipeline walking. It is not to assess the acceptability of the design checks for other
failure modes; that is the role of the VAS analysis, Section 3.3 and 3.4. However, if
the analysis highlights undesirable behaviour (for example pull out of a route curve)
then it may be necessary to reassess the results of the VAS analysis.
If the pipeline is restrained to prevent unacceptable walking, for example by
anchoring the flowline, then it is essential to carry out cyclic loading analysis. In
particular the presence of a pipeline anchor can cause route curve pull-out, or buckle
pullout, after the anchor becomes fully loaded over a number of operating cycles;
while end expansion will usually take some further cycles to stabilise.
A suitable design methodology for evaluating cyclic behaviour is outlined in Section 5.2.
In this stage the pipeline walking response should be evaluated. Pipeline walking is not a
limit state; however, the very significant axial displacements involved can lead to
overstressing and subsequent failure of pipeline risers, jumpers or spoolpieces. The
incremental axial displacement associated with each start-up cycle must be evaluated. The
total axial displacement over the life of the pipeline must be calculated based upon this
incremental displacement and the best estimate of the number of start-up/shut-down cycles
anticipated over the design life.
In the assessment of walking it is important that the number of major shut-down
cycles is not overestimated. The number of cycles should be a realistic best
estimate, not an upper bound. Historical information from similar projects can be
used to assist in quantifying the appropriate number of cycles.
The number of cycles need not be the same as that assumed for low cycle fatigue
design.
The behaviour of the pipeline is acceptable if:-
 The total axial displacement over the life of the pipeline is within the design capacity of
any pipeline connections;
Pipeline connections include off-line tie-ins via jumpers or spools, or in-line
connections such as to a SCR.
The assessment must also confirm the integrity of any mechanical connectors under
the extreme displacement and cyclic loading.
If the number of cycles is large, then very low walking rates per cycle could lead to
very significant movement over the life of field. However, pipeline walking is not well
understood and there is limited operational verification of the design predictions.
When the predicted walking rate is very low, the model uncertainty means there is
little confidence in the prediction. Tentatively, if the walking rate is below 5 mm per
cycle no further assessment is required. However, this may be revised as operational
experience grows.
 The loads within any lateral buckle remain sustainable;
 Route curve pull-out does not occur.
Some movement of the route curves may be acceptable, so long as the walking
response is tolerable and the route curve has stabilised within a reasonable number
of cycles.
If the pipeline connections or lateral buckles are not able to tolerate the total axial
displacement developed over the life of the pipeline, remedial measures must be developed
to control the walking response, for example anchoring.
Report No: 5087471/ 01 / A

Design Guideline
There is significant uncertainty over how accurate the pipeline walking predictions are.
Consequently, an acceptable approach is to recognise the potential for walking, but to delay
significant remedial measures until the actual behaviour of the pipeline has been
established. This approach must be integrated into the pipeline integrity monitoring system
(Section 7.3.3).
Given the uncertainty at low rates of walking, the current recommendation is to
implement monitoring procedures rather than pre invest in expensive mitigation
techniques. However, in this case the monitoring procedures must be capable of
quantifying the pipeline behaviour in a way that allows mitigation measures to be
implemented as necessary.
In addition, if this approach is adopted, the design should identify how any
operational mitigation could be implemented (for example, it may be sensible to
incorporated suitable anchor attachment points into the design).
Report No: 5087471/ 01 / A

Design Guideline
4. CALCULATION OF CHARACTERISTIC VAS

4.1. OVERVIEW OF METHODOLOGY
Buckle formation is an imperfection sensitive process and it is intimately linked to the initial
condition of the pipeline – the OOS. As project specific OOS information is not known prior
to pipe lay, there will always be an inherent uncertainty over the buckling response of the
system. The designer must quantify this uncertainty and reduce it to levels whereby the
project can proceed with confidence.
As the distance between intended buckle sites is decreased, the guarantee of formation
reduces. This means that buckling may not occur at some of the intended sites – i.e. the
robustness of the solution decreases. If the buckle initiation sites are too closely spaced,
some will not develop buckles and the distance between buckles will be greater than
intended. In other words, there is a limit to how closely the buckles can be guaranteed and
this practical limit must be incorporated in the overall design.
Within the SAFEBUCK design methodology these issues are addressed through the
Characteristic VAS. The Characteristic VAS is the design value of VAS; it is the minimum
VAS that will not be exceeded to an acceptable level of reliability. The design process
involves demonstrating that the design unity check for all relevant failure modes is
acceptable at the Characteristic VAS.
A deterministic definition of the Characteristic VAS is given in Section 4.2 and a probabilistic
definition of the Characteristic VAS is given in Section 4.3. The user can choose to use
either definition.
Calculation of the probabilistic Characteristic VAS requires a probabilistic buckle
formation assessment to be performed; the deterministic Characteristic VAS does not
and is simpler to calculate. However, the deterministic Characteristic VAS is more
conservative than the probabilistic definition.
The definition of Characteristic VAS is different for triggers, rogues and rogues between
triggers.
4.2. CHARACTERISTIC VAS: DETERMINISTIC DEFINITION
4.2.1. No Buckle Initiation Strategy

The Characteristic VAS is given by:-
X Char  min X S , X DF , X L  .... 4.1
Where
XL is the Characteristic VAS limited by pipeline length
XDF is the Characteristic VAS limited by the driving force envelope
XS is the Characteristic VAS limited by sharing
4.2.1.1. Characteristic VAS Limited by Pipeline Length

The Characteristic VAS limited by pipeline length depends upon the end conditions. For a
free ended pipeline it is equal to half the pipeline length and for a fixed ended pipeline it is
equal to the pipeline length.
Report No: 5087471/ 01 / A

Design Guideline
4.2.1.2. Characteristic VAS Limited by the Driving Force Envelope

The Characteristic VAS limited by the driving force envelope can be calculated from:-
 N  NBH 
X DF  2.588  L BH  2   0  .... 4.2
 fLB 
Where the Hobbs buckle length, LBH, is calculated by solving the following equation:-
  2 5  
N0  34.06 
EI 
 1.294  fLB  L BH   1  1.668x10  4  EA  HUB Re s  L BH   1
   .... 4.3
2
L BH   fLB  EI 2  
 
The force in the buckle follows from:-
EI
NBH  34.06  .... 4.4
L2BH
This solution is the Hobbs isolated mode III solution [4] for an elastic pipeline. LBH is
the length of the central buckle lobe and 2.588LBH is the overall length of the lateral
buckle.
4.2.1.3. Characteristic VAS Limited by Sharing

The Characteristic VAS limited by sharing is calculated from:-
 N̂  N 
X S   CR BH   2.588  L
BH .... 4.5
 fLB

 
The upper bound critical buckling force for a straight laid pipeline can be calculated from:-
 EI  HUB EI  HBE 
N̂cr  Max  4.4  ,7.1   .... 4.6
 D D 
 
This equation is based upon the critical buckling force model outlined in Section
4.5.2.1 and 4.6.4.1. The upper bound lateral break-out condition is combined with
the best estimate out of straightness parameter and the best estimate lateral break-
out condition is combined with the upper bound out of straightness parameter.
4.2.2. Engineer Buckle Initiation Strategy

For an engineered buckle, the Characteristic VAS is equal to the trigger spacing. For a
rogue buckle between triggers, the Characteristic VAS is equal to 60% of the trigger spacing.
For a short free-ended pipeline with a single trigger near the centre of the pipeline,
the trigger spacing can be taken to be half the pipeline length.
These definitions are applicable so long as the trigger spacing is greater than the sharing
limit associated with the trigger:-
 N̂  NBT  L
X S   CRT   BT .... 4.7
 fLB
 2
 
Report No: 5087471/ 01 / A

Design Guideline
If the trigger spacing is less than the sharing limit associated with the trigger, then the
Characteristic VAS must be evaluated using the Probabilistic Definition (or the trigger
spacing increased).
4.3. CHARACTERISTIC VAS: PROBABILISTIC DEFINITION

This definition of the Characteristic VAS requires that a probabilistic assessment of buckle
formation be performed; guidance on this assessment is provided in Section 4.4 to 0.
The Characteristic VAS is the VAS whose exceedance probability meets the requirements
defined in Table 4.1.
Rogue Buckles Engineered Buckles

1
Full Pipe Length Per km of Pipe
No Engineered Triggers 10% 1% -
Engineered Triggers 10% 1% 10%
Note 1 If there are no engineered triggers, then the full length of pipe is the total pipeline length. If there are engineered
triggers, then the full pipe length is the total length of pipe between initiators.
Table 4.1 Allowable Exceedence Probability: P(X>Xchar)

In general the Characteristic VAS will vary along the length of the pipeline.
It is convenient to present the Characteristic VAS graphically, as in Figure 4.1.
6000
5000
Characteristic VAS (m)
4000
3000
2000 Tout =Ambient
1000 Tout=80% Tin
0
0 5 10 15 KP 20
Figure 4.1 Typical Variation in Characteristic VAS along a Pipeline

The figure shows the Characteristic VAS along the length of the pipeline. The
Characteristic VAS describes the severity of a buckle in any km of the pipe, which is
why the figure is stepped. At each end of the pipeline the Characteristic VAS falls;
this is a result of the interaction of the buckle feed-in zones with the end expansion
zones.
The Characteristic VAS will normally fall along the length of the pipeline; due to
changes in the operating conditions, routing and possibly bathymetry. The blue
curve shows the Characteristic VAS for a pipeline in which the temperature does not
fall significantly along its length.
Report No: 5087471/ 01 / A

Design Guideline
The red curve shows the Characteristic VAS for a pipeline in which the temperature
does fall significantly. The Characteristic VAS reduces significantly with length and in
this example buckling is not a concern beyond KP 14 (the probability of buckling in
any km is less than 1%).
When a buckle initiation strategy is adopted, the calculation of Characteristic VAS must
incorporate the influence of the triggers on buckle formation.
If the triggers are well designed (fire reliably), then the Characteristic VAS associated
with any rogue buckle will reduce, this is illustrated in Figure 4.2.
4 Sleepers (rogues)
6000
No Triggers
5000 Triggers
4000
3000
2000
1000
0
0 5 10 15 KP 20
Figure 4.2 Influence of Triggers on Characteristic VAS of Rogue Buckles

In this example, four triggers are employed. The figure plots the Characteristic VAS
for each trigger (red squares) and the Characteristic VAS for rogue buckling. The
initiation strategy significantly reduces the Characteristic VAS associated with rogue
buckles; this is due to the high probability of each trigger firing
4.4. STRUCTURAL RELIABILITY APPROACH

When the Characteristic VAS is defined in terms of an exceedance probability, it must be
calculated using a probabilistic approach.
A structural reliability model of the pipeline expansion process is required to calculate the
Characteristic VAS. Structural reliability methods are probabilistic methods, which rationally
treat the various sources of uncertainty involved in the buckle formation analysis. The
SAFEBUCK JIP has developed a probabilistic methodology which can be used to calculate
the Characteristic VAS[11]. This assessment methodology is described here and is suitable
for use in the design approach.
The approach outlined here is suitable for use in design. However, any documented
method which is capable of evaluating the Characteristic VAS can be used.
In a reliability analysis, the uncertain parameters are modelled using appropriate probability
distribution functions. The general steps in the process are:-
Report No: 5087471/ 01 / A

Design Guideline
 Define a quantitative model of the underlying process - in this case the buckle formation
process (Section 4.5);
 Establish relevant probability distributions for the key input parameters (Section 4.6);
 Undertake the probabilistic simulation (Section4.7);
 Extract probabilistic results (Section 4.8).
Since frequent buckle formation is desirable, it is important that the simulation does not
overestimate the likelihood of buckling. Specific issues which can compromise buckle
formation are:-
 Higher than anticipated residual lay tension
A high residual lay tension will reduce the driving force for buckling. The lay tension
can be higher than anticipated due to residual thermal loads; installation of in-line
structures or flooded pipeline installation
 Longer than anticipated time delay between installation and start-up
If there is a significant delay between installation and start-up, the on-bottom
condition of the pipe may change. Buckle formation could be compromised if the
pipeline embedment increases significantly or significant self-burial occurs. A similar
concern arises if the pipeline initially operates in a low temperature regime, which
means that the lateral restraint may be higher than anticipated once full operating
conditions are imposed.
Each of these areas of concern can be evaluated using the buckle formation model.
4.5. BUCKLE FORMATION MODEL
4.5.1. Buckle Formation Limit State Equation

The limit state can be formulated in terms of effective axial force. Buckling will occur at any
point in the pipeline at which the driving force equals or exceeds the critical buckling force,
i.e.
N  Ncr .... 4.8
The limit state equation is then:-
gN, Ncr   N  Ncr .... 4.9
In the implementation the limit state can be recast as a load resistance ratio:-
N
R .... 4.10
Ncr
The buckling failure probability is then given by:-
Pf  Pr obR  1 .... 4.11
4.5.2. Calculation of Critical Buckling Force

The appropriate model of critical buckling force varies depending upon the feature
considered. The models outlined here are suitable for conceptual design. For detailed
design they should be verified by project specific analysis.
Report No: 5087471/ 01 / A

Design Guideline
4.5.2.1. Inherent Lateral OOS

The critical buckling force for a straight laid pipeline is given by:-
Ncr  X NH  Nchar  L .... 4.12
Where
EI  Ws
Nchar  3.86  .... 4.13
D
XNH is a normalised critical force parameter. Guidance on the behaviour of XNH is given in
Section 4.6.4.1.
If XNH is equal to unity, then the critical force is the Hobbs minimum force for an
infinite mode buckle. For XNH below 1 buckling would not be feasible based on the
Hobbs analysis. In practice buckling could occur if significant OOS is present. As
XNH increases above 1, lower and lower level of OOS could initiate a buckle.
Consequently, the parameter XNH is a measure of pipeline OOS.
For a PIP system, an effective diameter can be used in equation 4.13:-
8  EI
D eff 
EA
Where EI is the total bending stiffness of the PIP and EA is the total axial stiffness of
the PIP.
4.5.2.2. Route Bends

This model of critical buckling force is applicable to large radius route bends and large radius
bends associated with a snake lay design strategy. The critical buckling force is given by:-
NCr  X NB  L  Ws  R .... 4.14
For a given target bend radius, the actual as-laid shape will contain features that are more
severe than the target. The parameter XNB is essentially a measure of this variation from
nominal radius. Guidance on the behaviour of XNB is given in Section 4.6.4.2.
4.5.2.3. Vertical Trigger

Vertical imperfections (e.g. sleepers) are often employed to initiate buckling. The critical
buckling force from a prop type vertical imperfection is given by:-
Ncr  XNV  N V .... 4.15
EI  Ws
NV  4  .... 4.16
H
The parameter XNV describes the lateral OOS in the vicinity of the vertical imperfection.
Guidance on the behaviour of XNV is given in Section 4.6.4.3.
In practice the local lateral OOS will be very important and the buckling force will be
related to the lateral resistance provided at the vertical upset. As the model does
not consider the impact of lateral OOS explicitly, the lateral resistance is not a
parameter. Care is required in the use of this relationship.
Report No: 5087471/ 01 / A

Design Guideline
4.5.2.4. Buoyancy Trigger

The critical buckling force at a buoyancy trigger is given by:-
Ncr  X NBo  NcharB  L .... 4. 17
Where
EI  WsB
NcharB  3.86  .... 4. 18
D
XNB is a normalised critical force parameter. Guidance on the behaviour of X NB is given in

Section 4.6.4.4.
The approach outlined here is based on the reduction in lateral friction associated
with reduced submerged pipe weight. However, buoyancy also introduces a vertical
OOS due to the different level of embedment of the normal pipe when compared with
the larger diameter, lighter pipe with buoyancy attached. The effect of vertical OOS
can be incorporated into the XNB distribution.
The critical buckling load can be reduced by making the pipe positively buoyant at the
trigger. The critical buckling force in this case can be calculated in accordance with
reference [12].
In this strategy will generally involve the use of temporary buoyancy which produces
a positively buoyant pipe during start-up. For general operation the excess buoyancy
will be removed and the pipe will be negatively buoyant.
4.5.3. Calculation of Driving Force

The critical buckling force varies from point to point along the pipeline. It depends upon the
local out of straightness and the local lateral pipe-soil resistance. The driving force also
varies from point to point along the pipeline. It depends upon the axial pipe-soil resistance,
any pipeline end reactions, pipeline routing and the changing operating parameters. It also
depends on the location and form of any prior buckles.
In order to identify the buckle locations it is necessary to consider the pipeline behaviour as
the effective force increases during hydrotest or during start-up, which is illustrated in Figure
4.3.
Report No: 5087471/ 01 / A

Design Guideline
1
Pre-buckle
0.9
1st buckle stage 1
0.8 Buckle #1 Pre-buckle 2
0.7
N/No max
0.6 Fully Constrained Force

0.5 As system heats-up
0.4
0.3
0.2
0.1
0
0 0.2 0.4 0.6 0.8 x/L 1
Figure 4.3 Buckle Development during Start-up; Stage 1

The dashed black lines in the figure show the development of the fully constrained force as
the start-up develops. The form of the force profile during start-up is different to the steady
state force profile during operation.
The figure shows three stages equivalent to inlet conditions equal to 60%, 75% and
80% of the design conditions – these are arbitrary values for the purposes of
illustrating the behaviour.
The curves presented are simplified. The real curves would be non-linear and if the
system is pressurised would not intersect the x-axis as shown, but extend over the
whole pipeline at a low level equivalent to the pressure effect.
In addition, buckling could occur during hydrotest. Buckling during hydrotest is
usually desirable in that it allows the success of the design strategy to be evaluated
prior to operation. The buckle formation process during hydrotest is essentially the
same as outlined here for first load and so is not explicitly discussed. However, in
general, the buckle formation assessment should evaluate the hydrotest load
followed by the start-up load.
The first force profile (red) illustrates the situation just before a buckle initiates. Initiation will
occur at the first location along the pipeline to achieve the critical combination of axial force
and out-of-straightness.
Within Figure 4.3 this is shown as occurring at the (temporary) hot anchor point; this
need not occur here – the imperfection distribution within the real pipeline will cause
buckling at any point. In the absence of engineered triggers or significant
bathymetric OOS, the shape of the force profile means buckle formation is most likely
in the vicinity of the hot anchor point. The figure also shows a short section of the
pipeline at the fully constrained conditions prior to buckling; this again is not
necessary (or important) and will vary from case to case.
Report No: 5087471/ 01 / A

Design Guideline
As the start-up continues and the operating conditions become more onerous, the buckle
begins to develop. The force within the buckle drops as the pipe feeds-in, and the slip zones
become more developed. The figure shows two stages in the heat-up post buckling. As the
buckle develops the locations of the anchor points change; essentially the term anchor point
is somewhat of a misnomer. Prior to buckling every point between the hot anchor point and
the end moves towards the end of the pipeline as it expands (i.e. to the left in Figure 4.3).
Consequently, when a buckle appears, the first virtual anchor forms at a location that has
already translated away from the buckle site; this effects the feed-in to the buckle, but is
conservatively neglected within the VAS concept.
The green profile in the figure is taken to be the point just before a second buckle initiates.
Here the force developed is higher than that attained in the original pre-buckled profile; again
this may not be the case – the details of whether a second buckle forms when the force is
higher or lower than the initial critical force is entirely determined by the distribution of out-of-
straightness (and seabed frictional response). Further progress of the start-up process is
illustrated in Figure 4.4.
1 Pre-buckle 2
Buckle #1 Buckle #2
0.9 Buckle#2 stage 1
0.8 Pre-buckle 3
0.7
N/No max
0.6
Fully Constrained
0.5 Force
0.4 As system heats-up
0.3
0.2
0.1
0
0 0.2 0.4 x/L 0.6 0.8 1
Figure 4.4 Buckle Development During Start-up; Stage 2

As the second buckle forms, the force begins to fall. This gives rise to feed-in to the second
buckle and the development of an anchor point between buckle #1 and buckle #2. The force
within the first buckle continues to drop as its development continues, but the rate of fall
slows in-line with the response illustrated in Figure 2.4. As for stage 1, the position of the
anchor point varies as the start-up process progresses. Formation of further buckles is
illustrated in Figure 4.5.
Report No: 5087471/ 01 / A

Design Guideline
1
0.9
Pre-buckle 3
0.8
Fully developed
0.7
Steady State No
N/No max
0.6 Buckle #1 Buckle #2 Buckle #3 Buckle #4

0.5
0.4
0.3
0.2
0.1
0
0 0.2 0.4 x/L 0.6 0.8 1
Figure 4.5 Buckle Development During Start-up; Stage 3

Now formation of the third buckle occurs. The force in the other buckles continues to drop
eventually leading to final force profile shown in green (in which a fourth buckle is also
illustrated). This fully developed force profile can occur before the system start-up is
complete. As the frictional slopes now govern the force in the system, any further increase
in operating conditions will not modify the force in the system – but it will lead to further
expansion displacement and growth of the buckles as the potentially high fully constrained
force is dissipated into end expansion and feed-in towards each buckle until the final steady
state distribution is achieved.
Once the buckle formation process is complete the VAS can be identified, this is illustrated in
Figure 4.6 for the example outlined above.
1
0.9
Fully developed
0.8
Steady State No
0.7
N/No max
0.6
D1 D2 D3 D4
0.5
0.4
0.3
0.2
0.1
0
0 0.2 0.4 x/L 0.6 0.8 1
Figure 4.6 VAS for Final Buckle Configuration

Each buckle in the pipeline, which can be either a rogue buckle or an engineered buckle,
has an associated VAS.
Report No: 5087471/ 01 / A

Design Guideline
4.6. INPUT PROBABILITY DISTRIBUTIONS
4.6.1. Input Probability Distributions

To predict the formation of lateral buckles along a pipeline it is necessary to define suitable
probability distributions, which describe the uncertainty in the various key parameters. The
stochastic variables will depend upon the exact buckle formation model adopted, however,
any model will require distributions for the following parameters:-
 Axial pipe-soil resistance;
 Lateral pipe-soil resistance;
 Pipe OOS data.
Guidance on suitable distributions for these parameters is given in the following sections.
4.6.2. Axial Pipe-Soil resistance

Axial pipe soil resistance can be represented by a log-Normal distribution with a mean equal
to the best estimate value. The standard deviation to apply will depend on the confidence in
the axial friction predictions and the level and suitability of test data available to support
design. Guidance is given in Appendix B.
There is little data available on which to base the form of the distribution. The log-
Normal is recommended since it contains only positive values and represents many
engineering phenomena. Given the uncertainty in the distribution, the user must
ensure that the confidence limits employed in defining the distribution are not unduly
optimistic.
4.6.3. Lateral Pipe-Soil resistance

Buckle formation is governed by the lateral breakout friction. This can be represented by a
log-Normal distribution. The standard deviation will depend on the confidence in the
breakout predictions and the level and suitability of test data available to support design.
Guidance is also given in Appendix B.
There is little data available on which to base the form of the distribution. The log-
Normal is recommended since it contains only positive values and represents many
engineering phenomena. Given the uncertainty in the distribution, the user must
ensure that the confidence limits employed in defining the distribution are not unduly
optimistic.
4.6.4. Out of Straightness Data

Pipelines buckle wherever the critical combination of driving force, lateral restraint and out-
of-straightness (OOS) occurs. Of these parameters, the initial OOS in the pipeline is the
most uncertain parameter. It can never be known prior to installation. OOS survey data has
been collated within the JIP and the information obtained from this data is summarised here.
Currently the majority of analysis has been conducted on lateral OOS alone; vertical OOS
associated with uneven seabed bathymetry will also be present, and this can influence the
buckling response of the pipeline. Initial work, on limited projects on relatively flat seabeds,
indicates that the critical buckling force distributions for full 3D OOS (lateral and vertical
OOS) are broadly in line with those obtained using lateral OOS only. However, this will be
investigated in more detail in the Data Review task within SAFEBUCK Phase III.
Report No: 5087471/ 01 / A

Design Guideline
The information provided here is based on data gathered for a small number of
projects. In order to gain confidence in the methodology and develop a more robust
definition of OOS it would be necessary to repeat the process with a large number of
different data sets. Although the analysis undertaken provides useful support to the
proposed methodology, care should be taken in its application. In general project
specific work should be performed to support the chosen distributions.
4.6.4.1. Nominally Straight Pipeline

For straight laid pipe (and sections in between bends), imperfections occur as a result of the
changes in route directions and the lateral movement of the pipe inherent in the lay process.
In the absence of site specific data, the XNH parameter can be described by the distributions
defined in Table 4.2.
Bathymetry Distribution Mean Standard Deviation

Very flat log-Normal 1.18 0.35
Modest vertical OOS log-Normal 1.10 0.20
Table 4.2 Suggested Values for XNH Distribution

The distributions in Table 4.2 are applicable to a 1 km length of straight pipe, i.e. they
describe the most severe out of straightness feature that can be expected in 1 km of straight
pipe.
Identifying a suitable distribution for the parameter XNH is difficult. It must represent
the sort of OOS found in actual pipelines and the effect of this OOS on the buckling
force. The OOS will depend on many factors and consequently XNH will vary from
project to project and sometimes from pipeline to pipeline within a project.
The recommended distribution data has been derived from as-laid OOS data from six
pipelines. To develop the distributions the data has been divided into 1 km sections
and analysed using FE analysis. In each analysis, the pipe was loaded until a buckle
formed. The critical buckling force was recorded and divided by Nchar to give the
OOS parameter XNH; this defines a single point in the distribution. The process was
repeated for the whole data set to develop the distribution. As a consequence the
resulting distributions, Figure 4.7, are associated with 1 km of pipe.
3
Recommended
2.5 1 km pipe Length OOS1
OOS7
2 OOS8
OOS9
Probability
OOS25A
1.5 OOS25B
0.5
0
0 0.5 1 XNH 1.5 2 2.5
Figure 4.7 Distributions of XNH from Example Survey Data
Report No: 5087471/ 01 / A

Design Guideline
The results are consistent with the anticipated behaviour. The distribution implies
that for a 1 km length of pipe the most likely buckling force is the Hobbs safe force
(XNH~1). It would be expected that 1 km of pipe would contain at least one area with
sufficient OOS to trigger buckling at this level. However, it is possible for the force to
be below the Hobbs safe force – in areas of high OOS – but not too far below it.
Conversely, very straight sections of pipe will not buckle until the force is significantly
above the Hobbs safe force.
A total of 158 km of data has been analysed to produce the distributions. The data
covers a large range of pipe bending stiffness; from 44 MNm2 to 822 MNm2. The
variation in the mean value of XNH for the individual pipelines is illustrated in Figure
4.8.
1.5
1.5
1.4
1.4
1.3
1.3
1.2
1.2
Mean XNH
1.1
Mean XNH
1.1
1
1
0.9
0.9
0.8
0.8
0.7 0.7
0.6 0.6
Nchar (kN)
EI (MNm2)
0.5 0.5
0 500 1000 0 5000 10000
Figure 4.8 Mean of XNH for Individual Pipe Data Sets

The recommended distribution is based upon the whole data set.
4.6.4.2. Large Radius Route Bends

Buckling can initiate at a pipeline bend (either route bends or engineered bends for a snake
laid pipeline). For a given target bend radius, the actual as-laid shape will contain features
that are more severe than the target. The parameter XNB is a measure of this variation from
nominal radius.
The XNB parameter can be represented by a log-Normal distribution with a mean given by
equation 4.19 and a CoV of 25%.
 XNB  0.75  0.35  s .... 4.19
In this equation s, the bend arc length, must be defined in km. The distribution is suitable
for conceptual design, but should be verified for use during detailed design.
The statistics are based on a review of high quality as laid data from three pipelines.
The data is summarised in Figure 4.9.
Report No: 5087471/ 01 / A

Design Guideline
1.4
1.2
0.8 5% Prob ex
X NB
0.6
0.4
0.2 95% Prob ex
0
0 0.2 0.4 0.6 0.8 1 1.2
Arc Length (km)
Figure 4.9 Distributions of XNB : Effect of Bend Arc Length

There are 129 data points. Within the data set the arc length varies from about 70 m
to 1200 m. The range of arc length varies between data sets, but there is no
indication that the different data sets belong to separate distributions. Since XNB
measures the most severe OOS within a bend, it is reasonable to expect this to
depend upon arc length (since it is a weakest link property). The data supports this;
the best fit line to the data is equation 4.19.
The figure also shows the confidence limits associated with the distributions. The
data is consistent with these values.
The equation is suitable for bend radii between 1000 m and 3500 m and arc length up to
1 km. If the bend radius is too large, then the critical buckling force may exceed that
anticipated in a nominally straight pipe; in this case the straight pipe response should be
adopted.
For arc lengths greater than 1 km the XNB parameter can be taken to be log-normally
distributed with a mean value of 0.4 and a CoV of 25%. The associated length scale should
be 1 km.
In this case the distribution describes the most severe out of straightness feature that
can be expected in 1 km of pipe bend. This approach is similar to that adopted for
straight pipes
4.6.4.3. Vertical Triggers

Vertical imperfections (e.g. sleepers) are often employed to initiate a buckle. The parameter
XNV describes the OOS in the vicinity of the vertical imperfection. The parameter can be
modelled as a log-Normal distribution with mean 0.725 and standard deviation 0.14.
The distribution is based on a review of high quality as laid data from a single
deepwater project. FE analysis has been employed to calculate the critical force
associated with sleeper triggers using actual pipe OOS. The resulting distribution of
XNV is illustrated in Figure 4.10.
Report No: 5087471/ 01 / A

Design Guideline
1.2
3.5
1 3
0.8 2.5
Probability
2
NV
0.6
X
1.5
0.4
1
0.2
0.5
0 0
0.4 0.6 0.8 1 1.2 0 0.5 1 1.5
Height (m) X NV
Figure 4.10 Distribution of XNV for Vertical Triggers

Care should be taken in the use of this distribution.
4.6.4.4. Buoyancy Triggers

Buoyancy triggers can be employed to initiate buckling. These are likely to involve local
OOS due to the change in submerged weight, and possibly change in outside diameter. The
parameter XNBo describes the OOS in the vicinity of the buoyancy triggers. No useful data
has been obtained by the JIP to provide guidance on this parameter.
Projects adopting this approach will have to develop suitable distributions of critical
buckling force. These can be developed using FEA models with predicted levels of
OOS.
4.6.4.5. Bathymetry
Lateral buckling can initiate from vertical OOS. This is an important driver for buckle
formation, especially if significant spanning can occur. It is not possible to provide generic
data as this is entirely site specific.
If good quality bathymetric data is available to the project then this can be analysed using FE
analysis to provide an estimate of the severity of vertical imperfections. These analyses can
be used to:-
 Update the inherent OOS distribution (XNH) to incorporate the vertical OOS;
 Identify specific significant features which can be incorporated into the buckle formation
analysis in the same way as engineered triggers.
One method to do this is to divide the bathymetric data into discrete sections
(typically a 1 km long) and deform an initially straight pipe to the seabed data. The
operating load in the pipe is then increased until bucking occurs. This will identify the
most severe OOS features. If these are to be incorporated into the buckle formation
assessment, then a number of sensitivity cases should be undertaken to assess the
variation in buckling force with key parameters (for example vertical seabed stiffness,
survey data error, lateral resistance). The results of these analyses can then be
used to define a suitable distribution of critical buckling force for any bathymetric
feature.
 Identify the influence of free spanning on the force profile in the pipeline.
Report No: 5087471/ 01 / A

Design Guideline
If the bathymetry leads to pipeline free spans, then the expansion behaviour of the
pipeline will be modified. The pipeline expansion will feed in to the spanning
sections. This will absorb some of the feed-in that would otherwise drive lateral
buckling, and is generally beneficial. Very significant spans will modify the force
profile in a similar way to lateral buckling, i.e. the effective axial force will drop at the
span as the pipe feeds-in to it and virtual anchor points can be set up between spans
(or between spans and lateral buckles). The effect of this is to reduce the available
driving force for lateral buckling and reduce the Characteristic VAS should lateral
buckling occur.
4.7. PROBABILISTIC SIMULATION

A probabilistic simulation must be undertaken to evaluate the Characteristic VAS.
Since the buckle formation depends upon the development of the force profile, the
simulation must calculate the force profile as the start-up progresses. A Monte Carlo
simulation of the pipeline buckle formation process is the most straightforward approach to
the problem.
There are many different methods available for performing a SRA. The user is free
to adopt any method which is capable of delivering the Characteristic VAS (Section
4.3).
In the Monte Carlo approach the pipeline expansion process is simulated many times. Each
simulation (or trial) involves randomly selected values of the stochastic variables from the
defined distributions. For each trial the locations of lateral buckle formation are calculated by
calculating the force profile development during start-up; this can be undertaken as outlined
in 4.5. The outcomes of these random trials are tallied in an appropriate manner to produce
the desired result.
4.8. PROBABILISTIC RESULTS

The results of the analysis should be processed to:-
 Identify the overall probability of lateral buckling for the pipeline;
 Identify the VAS distribution for each buckle location;
 Identify the Characteristic VAS along the length.
4.8.1. Overall Probability of Lateral Buckling from Monte Carlo Analysis

The overall probability of lateral buckling can be used to evaluate the requirement to
undertake a full assessment of lateral buckling in accordance with this guideline.
If this probability exceeds 5% then the pipeline is susceptible to lateral buckling (See
Section 3.2).
If the number of lateral buckles that form in each expansion simulation is recorded, these
can be expressed as a probability distribution for the number of lateral buckles.
A typical distribution of buckles in a long pipeline is illustrated in Figure 4.11
Report No: 5087471/ 01 / A

Design Guideline
Number of Buckles in Pipeline

0.25
0 5 10 15 20
1.E+00
0.2
Probability of exceeding
0.15 1.E-01
pdf
0.1
1.E-02
0.05
1.E-03
0
0 5 10 15 20
1.E-04
Number of Buckles in Pipeline
Figure 4.11 No Initiation Strategy: Buckle Distribution

In this case several buckles are likely. The probability that there are no buckles in
the pipeline is extremely low. The median number of buckles is between 8 and 9;
this is the best estimate of the number of buckles in the pipeline, from the simulation.
4.8.2. Distribution of VAS from Monte Carlo Analysis

The overall aim of the analysis is to identify the Characteristic VAS (Section 4.3). In order to
do this the distribution of VAS at each point in the pipeline must be recorded.
For a buckle trigger element the distribution of VAS can be obtained by recording the VAS
identified at that trigger in each simulation. The Characteristic VAS is extracted from the
resulting VAS distribution.
In each simulation within the Monte-Carlo simulation the VAS associated with the
buckle at the trigger should be recorded. If no buckle forms at a trigger then the VAS
should be recorded as zero (this allows the probability of trigger failure to be
identified). The resulting VAS distribution can then be plotted, Figure 4.12.
VAS
0 1000 2000 3000 4000 5000
1.E+00
1.E-01
P (X>VAS)
1.E-02
1.E-03
1.E-04
Figure 4.12 Typical VAS Distribution at a Trigger
Report No: 5087471/ 01 / A

Design Guideline
The VAS cumulative probability distribution is shown as the blue line within Figure
4.12. In this example the trigger is quite reliable - there is a 7% chance that the
trigger fails to initiate a buckle - so the distribution starts with a probability of
exceedance very close to 1. For a trigger, the Characteristic VAS has a 10% chance
of being exceeded; this level of confidence is shown within Figure 4.12 as the solid
black line. The Characteristic VAS can be extracted from the distribution, which in
this example is 2800 m.
For a rogue buckle the distribution of VAS can be obtained by recording the maximum VAS
in each 1 km of pipe in each simulation. The Characteristic VAS is extracted from the
resulting VAS distribution.
In each simulation within the Monte-Carlo simulation, if a buckle forms anywhere
along the 1 km section of pipe, the VAS associated with the buckle should be
recorded. If no buckle forms within that km of pipe then the VAS should be recorded
as zero (this allows the probability of no buckling to be identified). The resulting VAS
distribution can then be plotted, Figure 4.13.
VAS (m)
0 2000 4000 6000 8000
1.E+00
1.E-01
P (X>VAS)
1.E-02
1.E-03
1.E-04
Figure 4.13 Typical VAS Distribution for a Rogue Buckle in 1 km of pipe

The VAS cumulative probability distribution is shown as the blue line within Figure 4.13. In
this example buckling is reasonably likely - there is a 32% chance that the pipe buckles
within this kilometre of pipe. For a rogue, the Characteristic VAS has a 1% chance of being
exceeded in any kilometre of pipe; this level of confidence is shown within Figure 4.13 as the
solid black line. The Characteristic VAS can be extracted from the distribution, which in this
example is 5000 m. FE Assessment
It is possible to support the formation reliability analysis using a FE model of the pipeline.
However, it is important that the results of the FE analysis are robust to the uncertainties
within the problem. A FE model of the system will buckle from the OOS inherent in the FE
model; however, the existence of this, or any other OOS, within the pipeline system cannot
be known prior to installation.
For example initial curvature may be introduced into the FE model to initiate buckling
in an otherwise straight model. Although buckling will clearly occur at the
imperfection as intended, the assumption that the rest of the pipeline is straight is
unrealistic. The absence of the real OOS information means that the result of a
single FE analysis gives limited information on the likelihood of buckle formation.
Report No: 5087471/ 01 / A

Design Guideline
This analysis approach may be performed using the anticipated bathymetric profile.
This can provide useful insight into the key vertical OOS features (see Section
4.6.4.5) but it is important that the uncertainty in the bathymetry and other data is
fully explored.
To demonstrate robust formation under all conditions, it is necessary to perform extensive
analysis to address a significant number of additional issues, for example:-
 The effect of global variation in friction; a conservative choice of friction from a pipeline
integrity point of view is not necessarily conservative from the buckle formation point of
view;
 The effect of local variations in friction, including increased lateral restraint in
undesirable locations;
 The effect of undesirable variations in the engineered OOS;
 The effect of large OOS in undesirable locations;
 The effect of survey errors in vertical OOS;
 The effect of vertical seabed stiffness.
The aim of the analysis is to demonstrate that under adverse combinations of the various
parameters, buckles will form as frequently as required.
Report No: 5087471/ 01 / A

Design Guideline
5. DESIGN ANALYSIS
5.1. SYSTEM MODELLING: VAS ANALYSIS
The pipeline should be modelled with an isolated buckle between virtual (or real) anchor
points. The model length should be equal to the appropriate VAS.
For a PIP system both the jacket pipe and flowline must be incorporated within the model,
i.e. an equivalent section model is not acceptable.
5.1.1. Conceptual Design

At the conceptual level the aim of the design process is to evaluate a large range of
competing design options. To facilitate this, an analytic model of the lateral buckling
phenomena can be employed. However, this must include:-
 A model of first load plasticity;
 A model of the cyclic response.
A suitable analytic model has been developed for SAFEBUCK and this is presented in
Appendix D.
The SAFEBUCK analytic model provides a good model of on-bottom buckling from a
straight pipe. It can be used to evaluate on-bottom buckling at a route bend.
However it will overestimate the severity of strains in this case; the error will increase
as the bend radius reduces.
It can be used to assess buckling with buoyancy triggers by modifying the
submerged weight employed. However this is an approximation and the accuracy is
lower than for normal pipe. The models are not suitable for the assessment of
vertical triggers (sleepers).
If a suitable analytic model is not available, FE analysis must be employed. The VAS
concept means that the FE models are relatively simple and involve short analysis
times. However, in many cases FE analysis may be incompatible with the
constraints and requirements of conceptual design.
An alternative approach to the SAFEBUCK model is the scaling solution developed
by Peek and Yun[13]. This only considers the first load response, but can be used to
assess materials which exhibit good strain hardening (which the SAFEBUCK model
does not).
A strain concentration factor (SNCF) should be employed to account for localisation effects.
The SNCF should account for the influence of any field joint at the buckle crown. The
SNCF should also address any other source of strain localisation (for example buckle
arrestors) that may coincide with a lateral buckle.
5.1.2. Detailed Design

The lateral buckling phenomena must be modelled using a non-linear finite element (FE)
program with full quality assurance and verification. The program must be capable of
modelling snap buckling and the appropriate non-linearities including:
 Non-linear geometry response (large displacements and large rotations);
 Material non-linearity (plasticity modelling capability);
 Non-linear elastic-plastic pipe-soil interaction forces.
Report No: 5087471/ 01 / A

Design Guideline
For a PIP system both the jacket pipe and flowline must be incorporated within the model.
The interaction between the two pipes must be carefully modelled to incorporate all relevant
effects (e.g. insulation stiffness and strength, gap modelling, centralisers, structural
connectivity between pipes and internal friction).
The design equations outlined in Section 6 assume that the loads are calculated in
accordance with certain modelling assumptions. If they are not, the design process may be
non-conservative. For example the use of a strain limit for local buckling requires that the
potential for strain localisation be fully quantified; this is achieved using the assumptions
outlined in this section.
The following modelling features must be incorporated into the FE model:-
 The stresses associated with the as-laid condition;
This requirement applies to major as-laid features, for example route or snake bends
or vertical stress over sleepers and bathymetry if there is significant undulation. It is
not necessary to model the stress associated with minor bathymetric variation in the
VAS model. If it is significant, the influence of bathymetry can be evaluated in the
buckle interaction model (Section 5.2 ).
 The element length at the crown of the buckle must be small enough to identify the
curvatures developed within the buckle. The user should perform a sensitivity study to
demonstrate that the chosen mesh is adequate. In the absence of such a study, the
maximum element length in the vicinity of the buckle must be limited to one pipe
diameter.
 A fully non-linear pipe-soil interaction model must be employed, in accordance with
Section 5.7.
The potential for strain localisation at the crown of the buckle due to pipeline field joint
stiffness discontinuity must be fully quantified.
For single pipe systems significant stiffness discontinuity could occur in a concrete
coated pipe[14-16] or a thick insulation coated pipe. The effect of the field joint on strain
localisation should be evaluated through detailed FEA or a test programme.
For a PIP system, the field joint stiffness discontinuities should be modelled.
Any other source of significant strain concentration should also be incorporated into
the design, for example buckle arrestors or change in cross-section.
It is not necessary to model the influence of normal joint to joint strength mismatch on
strain localisation.
The influence of these discontinuities can be incorporated into the assessment either by use
of appropriate SNCF or the inclusion of a length of elements with reduced bending capacity.
If weak elements are incorporated into the FE model, it is prudent to undertake some
analysis without the weak joint. This allows the importance of the weak joint to be
isolated and may assist in defining suitable tolerances within the pipeline
specification.
If a SNCF is employed, it should be evaluated using detailed local FE models of the
buckle crown. The SNCF should be appropriate to the level of pressure, axial force
and bending anticipated at the buckle crown; this may change as a function of
imposed load. With the SNCF approach care should be taken to account for the
change in global curvature caused by the presence of the local discontinuity.
Report No: 5087471/ 01 / A

Design Guideline
For fatigue loading the analysis should incorporate the potential for stress concentration due
to the field joint discontinuity.
5.1.3. Design Considerations for Reeled Pipelines

The reeling process will modify the material characteristics of the pipeline steel. The key
parameters which will be affected are:-
 The stress-strain behaviour of the pipeline;
 The fracture toughness of the pipe and welds.
The subsequent structural response will also be modified:-
 The pipe will be installed with significant residual stress and possibly residual strain;
 The post-reel OOS will modify the pipeline structural behaviour;
The user must ensure that all relevant influences of reeling are incorporated into the VAS
analysis.
5.2. SYSTEM MODELLING: INTERACTION AND WALKING

Once the lateral buckling design philosophy has been established, the cyclic expansion
behaviour of the pipeline should be verified. This evaluation must incorporate the chosen
lateral buckling design strategy.
The aim of the analysis is to investigate the buckle interaction and cyclic expansion
behaviour of pipeline. The approach should evaluate:-
 Buckle interaction and stability;
 Route curve pull-out;
 Influence of bathymetry;
 Pipeline walking.
The number of operating cycles should be carefully evaluated and the best estimate should
be used in the analysis.
5.2.1. Conceptual Design

On unload an axial tension will develop in the pipeline. This may pose a problem for any
horizontal curves in the pipeline route; the tension acting over a route curve may be high
enough to cause the line to slide laterally. The radius of curvature at which the line will be
stable is highly dependent on the break-out lateral soil resistance. Ignoring the small
stiffness component, the tensile force at which a route curve becomes unstable is given by:-
S RC  HLB  R .... 5.1
If the tension on unload in the vicinity of the route bend exceeds SRC, then consideration
should be given to increasing the route bend radius.
Simple models to address walking have been developed within the SAFEBUCK JIP[8,17].
These models are recommended for conceptual design and are outlined in Appendix D.
The selection of appropriate pipe-soil parameters for walking interaction analysis is
discussed in Section 5.7.4
Report No: 5087471/ 01 / A

Design Guideline
5.2.2. Detailed Design

Once the lateral buckling design philosophy has been established, an FE model should be
employed to verify the global response of the pipeline.
The model should incorporate the following features:-
 The whole length of the pipeline (jumper to jumper);
This is necessary if pipeline walking is a major concern. If walking is not a major
concern, it is possible to model the pipeline in a number of shorter sections (to
investigate the effect of bathymetry for example). However, in this case, the end
conditions must be carefully considered and the designer must demonstrate that
these are not influencing the results of the analysis.
 Pipeline route and seabed terrain (bathymetry);
It may not be necessary to include bathymetry if the seabed is very flat.
Significant bathymetric variation will have two main influences on the pipeline
behaviour. It can lead to initiation of buckles, and it can produce pipeline free spans.
Both of these phenomena will be influenced by the accuracy of the survey data and
the model of the seabed vertical stiffness. It is important that a realistic model of
vertical stiffness is employed.
Free spans will absorb expansion in a manner analogous to lateral bucking. This is
beneficial and should be incorporated into the assessment. If the bathymetry has a
beneficial influence on response, sensitivity analysis should confirm that the result of
the assessment is robust to practical variations in the design assumptions.
 Appropriate end conditions (e.g. PLET reactions or SCR tension);
 The lateral buckling response required by the design strategy;
 The transient temperature profiles must capture the steepest transients that occur
(usually early in the start-up sequence);
A transient thermal analysis of the start-up process must be undertaken in order to
develop the thermal profiles required in the walking analysis.
 The number of time steps employed to represent the start-up process must be sufficient
to capture the walking response;
For hot pipelines it is common for the walking response to occur very early in the
heat-up (i.e. significantly before full operating temperature is reached).
Consequently, an adequate number of time steps should be employed over the most
significant portion of the heat-up, rather than a constant step over the whole heat-up.
Sensitivity analysis should be undertaken to ensure the modelling assumptions are
accurate.
 The analysis must continue until the magnitude of the axial displacement with each
start-up/shut-down cycle has reached a steady state.
This will typically involve at least ten start-up/shut-down cycles, and often many
more. For complex operating procedures, it may be necessary to incorporate a
number of different cycle definitions.
As the length of the models can be large, the modelling requirements employed in the VAS
analysis may be relaxed in the following areas:-
 the element discretisation can be coarser than that employed in the VAS models;
 the pipe-soil modelling may be simplified.
Report No: 5087471/ 01 / A

Design Guideline
In all cases the user must ensure that the simplifications do not compromise the applicability
of the analysis.
For example, a pipe-soil model may be employed that does not incorporate the effect
of berms. However, in this case, the model may overestimate the ratcheting growth
of the buckles, which in turn will influence the walking behaviour of the pipeline.
The selection of pipe-soil parameters for walking analysis is discussed in Section 5.7.5
Sensitivity analyses should be undertaken to evaluate the influence of relevant parameters.
Full length models are complex, and for all but the shortest pipelines will involve very
long analysis run time. Generally it is not practical to undertake a large number of
sensitivity analyses using a full length FE model. The use of the VAS concept within
the SAFEBUCK design methodology is intended to minimise the number of analyses
required using full length FE models. Consequently, if the design guideline is
followed, it is anticipated that only one or two sensitivity configurations will require
investigation. A suitable number of sensitivity analyses can be defined by the project
team.
The relevant parameters will be different for each project. Typical parameters
include; axial and lateral pipe-soil resistance; vertical seabed stiffness; bathymetry;
route bend radius; trigger design and location; operating cycles.
5.3. DESIGN LOADS

In assessing the lateral buckling phenomena both the response under initial loading and
subsequent cyclic loading must be evaluated.
5.3.1. Initial Load

The response under initial load should be performed at full local design pressure.
If it can be demonstrated that full pressurisation cannot coincide with the design
temperature, then the associated maximum internal pressure can be employed in the
analysis. However, the integrity of the buckle during increase to full design pressure
(at an appropriate temperature) must also be assessed.
It is acceptable to base the design on the maximum operating temperature, so long as
detailed process simulations have been undertaken. In this case the analysis can take
advantage of the thermal gradient along the pipeline length. However, this maximum
operating temperature must not be exceeded during the life of the pipeline.
The conceptual models outlined in Appendix B assume a constant temperature and
pressure over the VAS. This should be taken as the temperature and pressure at the
buckle location (this approximation is exact if the temperature profile over the VAS is
linear).
For PIP systems the response should be evaluated at both the design maximum and
minimum value of the jacket temperature.
5.3.2. Cyclic Loads

The most onerous stress range developed depends upon the phasing between
pressurisation and thermal load. This is illustrated in Figure 5.1.
Report No: 5087471/ 01 / A

Design Guideline
0.5% API Yield Ellipse Normalised Hoop

1.5
Stress
Elastic Ellipse
1
C
B F
0.5
D A E
0
-1.5 -1 -0.5 0 0.5 1 1.5
-0.5
Normalised Axial Stress
-1
-1.5
Figure 5.1 Load Step Definition

The cycle drawn is the traditional definition of the start-up shut-down cycle:
 Load A No pressure or temperature (but initial lay curvature);
 Load B Pressure up;
 Load C Warm-up;
 Load D Pressure down;
 Load E Cool down.
Subsequent cycles follow this sequence, for example, load F is the pressurised condition,
without temperature, on the second load cycle.
For any given cycle of pressure and temperature, the maximum axial stress range occurs
between the cold pressurised condition (load F) and the hot de-pressurised condition (load
D). For example if the pressure and temperature are applied simultaneously (load E to C)
the stress range calculated is significantly below the maximum value. The analysis must be
based upon the maximum feasible stress range associated with the pressure and
temperature change.
It is possible for the peak stress to occur at an intermediate point in the loading.
Care should be taken to ensure that the maximum stress range that occurs in the
cycle is identified.
If transient thermal analysis has been undertaken by the project, then the actual
phasing of the pressure and temperature cycle can be used in the analysis.
The figure presents the stress response in the flowline on the intrados of the buckle.
This is normally the location of maximum stress range because the axial force cycle
and the bending moment cycle produce a stress change of the same sign. At the
extrados of the buckle the axial force cycle and the bending moment cycle produce a
stress change of opposite sign and the stress range is reduced. The difference is
relatively small in single pipes, but can be significant in PIP systems. Account can
be taken of this beneficial effect in the ECA.
For calculation of the fatigue loads, the normal pipeline operating pressure and temperature
can be employed.
Report No: 5087471/ 01 / A

Design Guideline
5.3.3. Residual Lay Tension

A pipeline laid on the seabed will retain some residual lay tension. The residual lay tension
reduces the fully constrained force in the system in operation (equation 3.3) and therefore
reduces the potential expansion; this is beneficial to the pipeline response. The design
analysis can incorporate the beneficial effect of the residual lay tension.
Definition of a suitable design value of residual lay tension is not straightforward.
The magnitude of the residual lay tension will vary depending upon pipe geometry,
installation technique, water depth and environmental conditions during installation.
If problems are encountered during pipe lay then it may fall to zero in certain
locations.
If the pipeline is highly insulated, then the residual lay tension is likely to increase as
the thermal load imposed during installation may not dissipate until some time after
the pipeline has been laid on the seabed[18].
The residual lay tension may also be higher than normal if in-line structures are to be
installed or if the pipeline is laid flooded.
For lowly loaded systems, the residual lay tension may be sufficient to prevent buckling.
In deep water the lay tension is often increased considerably during the installation of
in-line structures. Consequently, the residual lay tension in the vicinity of these
structures may be high.
5.3.4. Late Life Conditions

In some projects the maximum operation conditions do not occur at the start of life.
Additional considerations are required if a significant increase in operating condition is
anticipated after initial operation has occurred and buckling has been established. Buckles
will quickly develop soil berms at the extent of their displacement. Any subsequent increase
in operating conditions will tend to increase the displacement at the buckle crown and may
cause the buckle to push through the established berm. In this case the buckle will meet an
enhanced lateral restraint at the crown. This may lead to higher strain than would occur if
the same operating conditions were imposed on first start-up.
The project should evaluate the impact of this behaviour.
The key parameter in the assessment is strength of the established berm, which
increases with time and cycles. There has been little research into the change in
strength of berms with time and load. The available information is summarised in
Appendix B. If this is a key issue for the project, then specific pipe-soil interaction
testing may be required.
5.4. CLAD AND LINED PIPES

Clad and lined pipes can be modelled using an equivalent pipe section and equivalent
properties.
5.4.1. Equivalent Section: Clad Pipe

The strength of the pipe cladding may be accounted for in the lateral buckling calculations.
To do this equivalent properties can be employed. The effective wall thickness can be taken
as the nominal wall thickness of the pipe plus cladding. Equivalent elastic properties should
then be calculated as a weighted average of the carbon steel and cladding properties:-
Report No: 5087471/ 01 / A

Design Guideline
 s  ts   c  tc
 eq  5.1
ts  tc
Where ζ is the property of interest (Young‟s Modulus, Poisson‟s ratio and coefficient of
thermal expansion), t is the nominal wall thickness. The subscript s refers to the base pipe
and c refers to the cladding.
The submerged weight of the system must incorporate the weight of the cladding.
5.4.2. Equivalent Section: Lined Pipe

The pipe should be modelled with the base carbon steel dimensions. The submerged
weight of the pipe should be increased to incorporate the weight of the cladding.
No beneficial effect of the liner should be incorporated into the analysis unless sufficient
testing and verification has been performed to calibrate the contribution to bending and axial
stiffness.
An equivalent coefficient of thermal expansion should be employed; this can be calculated
using Equation 5.1.
5.4.3. Plastic Lined Pipe

Only the steel pipe should be modelled. No beneficial effect of the liner should be
incorporated into the analysis.
This is particularly important for the calculation of the effective axial force (Equation
3.3) in which the pipe internal area should be based on the steel inside diameter not
the liner inside diameter.
5.5. MECHANICAL PROPERTIES

To evaluate the structural response of the pipeline system it is necessary to have good
information on the key mechanical properties of the pipeline steel and welds. The data must
be valid over the temperature range of interest for the design. The information provided here
is suitable for use at the conceptual design stage[19]. For detailed design a project specific
testing programme should be undertaken.
Depending upon the severity of the design problem the information presented here
may be suitable for detailed design, supported by modest project specific testing.
However, the user must demonstrate that more detailed data are not required.
The key properties of the steel are:-
 Elastic properties; Young‟s modulus, Poisson‟s ratio and coefficient of thermal
expansion;
 Yield stress;
 Ultimate tensile strength.
The yield stress, UTS and Young‟s Modulus all reduce with temperature. The coefficient of
thermal expansion increases with temperature.
For carbons steel pipelines operating at temperatures up to 50°C it is not necessary
to consider temperature dependant properties; standard structural values are
adequate.
Report No: 5087471/ 01 / A

Design Guideline
5.5.1. Young’s Modulus

The Young‟s modulus of steel reduces with increasing temperature; suitable data are
presented in Table 5.1. Linear interpolation should be applied for intermediate
temperatures.
Temperature (°C) Carbon Steel CRA

0 210 202.8
20 209 201.5
100 205 196.1
150 202 192.3
200 199 188.6
Table 5.1 Recommended Design Values of Young’s Modulus (GPa)
The carbon steel design values are taken from BS5500[20]. This was supported by a
detailed review of project tests on linepipe steel.
The CRA values are taken from ASME B31.3[21] supported by a review of (scarce)
available data. The CRA values can be used for 13Cr, 22Cr and 25Cr.
5.5.2. Poisson’s Ratio

No data were obtained by the JIP to provide guidance on the variation of Poisson‟s ratio with
temperature. A typical value at room temperature is 0.3 for carbon steel and 0.31 to 0.32 for
CRA.
5.5.3. Coefficient of Thermal Expansion

The coefficient of thermal expansion can be defined through the equation:-
  A   m   .... 5.2
Suitable constants are given in Table 5.2.
Material Aα (°C-1) mα (°C-2)

Carbon Steel 11.0 x10-6 8 x10-9
25Cr 13.0 x10-6 5 x10-9
22Cr 12.5 x10-6 5 x10-9
13Cr 11.0 x10-6 8 x10-9
Table 5.2 Definition of Alpha as a Function of Temperature for Carbon Steel
The coefficient of thermal expansion is defined as a total quantity capturing the effect
of temperature change over the whole temperature range from a reference
temperature of 20°C to the chosen design temperature, i.e. these values define the
secant modulus of the thermal strain- temperature curve.
The values are taken from a review of various code sources and project specific test
data obtained from X-grade linepipe steel[19].
Report No: 5087471/ 01 / A

Design Guideline
The CRA data in Table 5.2 are tentative and taken from supplier data sheets. These
should only be used in conceptual design, and in the absence of any more accurate
information.
5.5.4. Yield Strength

The basic yield stress should be taken as the pipe SMYS.
The actual yield stress will exceed the SMYS; measurements of actual linepipe
generally show a normal distribution with a mean between 4% and 10% above
SMYS. However, this distribution has been incorporated into the calibration of the
design equations[9,10]; the mean value should not be used in the design calculations.
The temperature de-rating response outlined in DNV OS-F101[1] is suitable for design, and is
summarised in Table 5.3. Linear interpolation should be applied for intermediate
temperatures.
Temperature (°C) Carbon Steel 25Cr 22Cr

20 0 0 0
50 0 40 40
100 30 90 90
150 50 120 120
200 70 140 140
Table 5.3 Yield Stress Decrement Values (MPa)
The data in Table 5.3 are relative to the yield stress at room temperature (20ºC). Insufficient
data are available to make general recommendations for 13Cr.
In the absence of any other information the yield stress de-rating of 13Cr can be
assumed to be similar to that of carbon steel.
5.5.5. Material Anisotropy

The yield stress information is based upon tensile tests. Some limit states are governed by
the hoop compressive yield strength of the pipe. If the hoop compressive yield stress is not
known it should be equal to the SMYS for seamless pipe and 85% of the SMYS for UOE
pipe.
Enhanced anisotropy has been observed in duplex[22] and may occur even in
seamless pipes. Insufficient data are available to give general guidance; this should
be evaluated during detailed design.
5.5.6. Ultimate Tensile Strength

For carbon steel the UTS temperature de-rating behaviour follows the yield de-rating
response outlined in Table 5.3. Insufficient data are available to make general
recommendations for the CRAs.
In the absence of any other information the UTS de-rating of 25Cr, 22Cr and 13Cr
can be assumed to be similar to the yield de-rating of 25Cr.
Report No: 5087471/ 01 / A

Design Guideline
5.6. PLASTICITY MODELLING

Lateral buckling will involve plastic deformation in all but the most lowly loaded systems. In
this case the analysis must be capable of modelling the elastic-plastic response of the
pipeline.
5.6.1. Monotonic Stress-Strain Curve

The shape of the stress-strain curve is extremely important for the local buckling response
and strain localisation effects. The two commonly observed shapes are illustrated in Figure
5.2.
550 Engineering
Stress (MPa)
500
450
Plateau
Roundhouse
400
350
300
0 1 2 3 4 5
Engineering Strain (%)
Figure 5.2 Typical Stress-Strain Response – X65

Seamless pipes constructed from X-grade steel tend to exhibit Lüder plateaus in the stress-
strain response up to total strains in the order 2-3% on first load. This response can also
occur in HFI pipe, but is less common. The general response of UOE pipe is to exhibit a
roundhouse type of behaviour.
It is not necessary to model the upper yield point associated with the Lüder plateau
response.
If thermal ageing is employed to enhance the collapse behaviour of UOE pipe this
may re-establish the plateau stress-strain response.
CRA materials do not normally exhibit plateau behaviour.
5.6.2. Cyclic Stress-Strain Curve

In standard testing the plateau behaviour is only observed on first load. The cyclic response
is illustrated in Figure 5.3.
Report No: 5087471/ 01 / A

Design Guideline
Stress (MPa)
450
250
50
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

Elastic Range
-150
Twice Yield
Strain (%)
-350
compressive cycle
-550 tensile cycle
Figure 5.3 Cyclic Response with Bauschinger Effect

The figure illustrates the case when the system is initially loaded in compression, for
comparison the response in initial tension is also shown. Following the initial compressive
load, subsequent yielding under a tensile load commences at a lower stress. This is known
as the Bauschinger effect, a phenomena in which plastic deformation in one direction
decreases the yield strength in other. The example shown has a strong Bauschinger effect;
the cyclic elastic range is only 1.4 times the uniaxial yield stress.
There is no distinct yield point in the cyclic response. Consequently a proof stress
concept must be used to define yield. This means that small levels of plasticity would
occur in the nominally elastic cycle.
It is convenient to define the cyclic yield stress as half the elastic range.
5.6.3. Stress Strain Curve Modelling: Conceptual Design

An elastic-perfectly plastic model can be used.
This will provide a good model for seamless pipes. However, if the steel does exhibit
roundhouse behaviour then this model may be unnecessarily conservative.
The SAFEBUCK analytic model employs an elastic-perfectly plastic model. An
alternative to the SAFEBUCK model, for first load analysis, is the scaling solution
developed by Peek and Yun[13]. This requires a single FE analysis, but can be used
to assess materials which exhibit good strain hardening.
5.6.4. Stress Strain Curve Modelling: Detailed Design
5.6.4.1. Seamless Carbon Steel Linepipe

For seamless carbon steel pipelines, the plateau in the stress-strain response must be
modelled. The recommended model of seamless carbon steel linepipe stress-strain
response is:-
Report No: 5087471/ 01 / A

Design Guideline
E   Y
Q  QY  Y    L ..... 5.3
C  np   L
In this model the true stress, Q is given in terms of the logarithmic strain λ. The real
material will exhibit a plateau in the engineering stress-strain response in both
tension and compression. This leads to conflicting requirements in true stress-strain
space (unless a non-monotonic stress-strain curve is adopted[23,24]). Equation 5.3
defines a plateau in the true-stress true-strain response; this is a compromise
between the observed conflicting responses and should be adequate for most design
purposes.
This model provides a good representation of the stress-strain response from first load to
ultimate tensile strength.
The model employs four parameters; the engineering yield strength, Young‟s Modulus, yield
to tensile strength ratio and the engineering Lüder strain.
From the engineering yield stress and the engineering Lüder strain the true stress and strain
parameters are:-
 y 
 
Q y   y  1   y   y  1 
E

 ..... 5.4
 
 y 
 Y  ln1   ..... 5.5
 E 
L  ln1  L  ..... 5.6
The constant C and the hardening co-efficient n can be calculated from the other parameters
using the following equations:-
QY
C n
..... 5.7
 Q 
 L  Y 
 E 
n n
QY e  Q 
     L  Y  ..... 5.8
UTS  n   E 
Suitable parameters for X60 and X65 linepipe steel at room temperature are given in Table
5.4
Report No: 5087471/ 01 / A

Design Guideline
Parameter X60 X65

Young‟s modulus, E 207 GPa 207 GPa
Engineering Yield stress, ζy 414 MPa 450 MPa
parameters Yield to tensile ratio, YT 0.836 0.850
Lϋder strain, εL 1.6% 1.6%
C 746.02 MPa 784.59 MPa
Model Hardening parameter n 0.137 0.129
parameters True yield stress, Qy 414.8 MPa 451 MPa
Logarithmic Lϋder strain, λL 1.587% 1.587%
Table 5.4 Suitable Parameters for X60 and X65 Steel at Room Temparature
This model of stress strain response is based upon a review of extensive project data
undertaken for the lateral buckling SRA[9]; the review covered API X60 and X65
linepipe steel. The recommended curves are based on the SMYS, combined with
the mean plus 1SD YT ratio and Lüder strain observed in the data set.
The YT ratio was found to be a strong function of the material yield stress. The data
is well represented by a normal distribution with mean given by equation 5.9 and a
CoV of 2.5%.
 y 
YT  0.65  0.16   
 ..... 5.9
 ref 
Where σref is 400 MPa.
The Lüder strain was well represented by a normal distribution with mean 1.6% and
CoV of 15%.
Wherever possible the stress-strain curve parameters should be confirmed by project
specific testing.
5.6.4.2. Other Linepipe

For carbon steel HFI pipe, or pipe in which the fabrication process involves cold expansion,
roundhouse behaviour can be assumed. For CRA pipelines roundhouse behaviour can be
assumed.
The design stress-strain curve should be based upon project specific testing.
The most accurate method of modelling the stress-strain curve is to adopt a piece-
wise linear representation of the measured response; this is the recommended
approach.
Often the stress-strain response of pipes that exhibit a roundhouse response are modelled
using a Ramberg-Osgood power-law relationship. Care should be taken when adopting this
approach, since the power-law curve may not provide a good model of the stress-strain
curve over the whole strain range of interest.
Report No: 5087471/ 01 / A

Design Guideline
The Ramberg Osgood model cannot give an accurate model of the stress-strain
response over the whole strain range of the material. The degree of strain hardening
must be carefully chosen and be appropriate to the strain range of interest. Basing
the hardening on SMYS and SMTS can produce unrealistic curves. The strain
hardening should be fully supported by testing to ensure that an excessively
optimistic model is not employed.
In the absence of project specific data, the parameters defined in Table 5.5 can be
used to represent X65 linepipe.
Yield strength α n
Low hardening 450 MPa 1.4 27
Median hardening 450 MPa 1.35 21
High hardening 450 MPa 1.3 16
Table 5.5 Ramberg Osgood Parameters for X65 Pipe

The parameters defined in Table 5.5 apply to the true stress- logarithmic strain
relationship:-
n
Q Qy  Q 
     ..... 5.10
E E  Qy 
 
These parameters are developed from a brief review of mechanical test data
undertaken as part of the lateral buckling SRA[9].
5.6.4.3. Clad and Lined Pipe

Corrosion resistant alloys usually exhibit good smooth strain hardening properties.
For Clad pipe, the stress-strain response of the equivalent pipe can be obtained through a
process of weighted averaging. In this case the stress values for a given strain shall be
averaged using the following procedure:-
 Calculate the true stress in the carbon steel for defined strain;
 Calculate the true stress in the cladding for defined strain;
 Calculate the true stress in the equivalent pipe using Equation 5.1.
The superior strain hardening behaviour of the cladding may be beneficial in reducing the
effect of any plateau behaviour in the X65 base pipe.
Clad pipe is normally fabricated from plate. The cladding is bonded to the plate, and
the pipe is formed and welded. These pipes occasionally exhibit a plateau in the
stress-strain curve of the carbon steel. Therefore the possibility of a plateau cannot
be discarded. In the absence of better information a plateau stress strain curve
should be assumed for the carbon steel.
A less conservative position would be to use a roundhouse stress-strain curve for
design. This approach could be adopted if project specific test data (for example
from Manufacturing Procedure Qualification Testing) are available which consistently
demonstrate smooth strain hardening.
For a lined pipe, only the base carbon steel response should be modelled.
Report No: 5087471/ 01 / A

Design Guideline
5.6.4.4. Hardening Model

For cyclic loading, an isotropic or kinematic hardening model can be used.
Kinematic plasticity models are normally used for situations in which cyclic plasticity
can occur. However, they are often restrictive in the way in which the shape of the
stress-strain curve can be defined. An isotropic hardening model is not, and is
acceptable since cyclic plasticity is prevented by the outlined design equations
(Section 6.6).
For reeled pipelines it may be necessary to incorporate the strain history associated
with reeling into the operational assessment. In this case a kinematic hardening
model is required.
5.7. PIPE SOIL INTERACTION

Lateral buckling, pipeline walking, route-curve pull-out and flowline anchoring are all
extremely sensitive to pipe-soil interaction forces and there is significant uncertainty
associated with the characterization of these forces in design. Pipe-soil interaction is often
modelled using a simple friction coefficient (Coulomb friction) or friction factor, linking the
pipe weight to the maximum available resistance to axial or lateral movement. A simple
friction factor can be used with care in conceptual evaluation but is generally not appropriate
to detailed design, particularly for lateral displacement.
Axial and lateral resistance should be treated as independent parameters in design for any
given set of soil conditions. Since the axial resistance response is global (affecting long
sections of pipeline), the total axial resistance is negligibly influenced by the localised lateral
friction response.
For pipeline design, it is important to bound behaviour. Upper and lower bound values of
soil resistance are both important. Uncertainty in soil behaviour will lead to a large range
between upper and lower bound behaviour, which is to be avoided. It is also important that
the designer does not assume an unrealistically narrow band of response.
Like all physical properties, pipe-soil interaction is subject to uncertainty. To reflect
the uncertainty in pipe-soil interaction, it can be expressed as a probability density
function. The upper bound, lower bound and best estimate values are characteristic
values, selected to represent the distribution in certain calculations. The term best
estimate is used here to describe the most probable value. For a non-symmetric
distribution, it is useful to define the upper and lower bound values with reference to
the probability that they are exceeded. Within this document the upper and lower
bound values should have an exceedance probability in the order of 2%-5%. The
distribution of the pipe-soil capacity is normally not well known and the estimates
have to rely on engineering judgement.
Good quality geotechnical data for the near surface soils, to about 2m depth, is therefore
essential to design. In weak soils this will require the use of T-bar (or equivalent device) to
measure undisturbed and remoulded shear strength, using box-core samples or in-situ
measurements referenced to the measured seabed surface. Specialist project-specific pipe-
soil interaction testing is also necessary for detailed design, unless the design is relatively
insensitive to variation in soil friction behaviour, or appropriate test data are already
established.
Reference should be made to Appendix B for guidance on geotechnical field data and
laboratory testing methods, as well as the proposed methodologies for predicting
embedment, axial friction and lateral friction responses, including detailed references.
Report No: 5087471/ 01 / A

Design Guideline
Much of the data available to SAFEBUCK JIP is related to very soft high plasticity
deepwater clays. Limited test data is available for high strength soils, such as stiff-
clays and sandy soils. This is an area of ongoing research.
5.7.1. Pipe Embedment

Pipe embedment (the depth of penetration into the undisturbed seabed) affects both the
axial and lateral resistance by changing the pipe-soil contact area, the operative soil
properties (at the depth of embedment) and the volume of soil that must be displaced for the
pipe to move laterally.
Over-embedment, due to the uncertain and variable influence of the dynamic laying process,
is key source of uncertainty in the prediction of pipe embedment. Predictions based on the
static weight generally underestimate the dynamic movement (arising from vertical and
lateral catenary oscillations) as the pipe touches down.
The range of embedment should be based on the contribution of each load case and the
range of measured soil properties, such as shear strength. Load cases include installation,
with increased contact stress along the touchdown region combined with remoulding of the
soil due to dynamic loading, as well as the subsequent flooded and operational submerged
weight cases, that generally occur after the soil has been substantially reconsolidated.
Figure 5.4 Terminology for Pipeline Embedment
5.7.2. Axial friction

Typical axial friction model behaviour, incorporating „breakout‟ and „residual‟ axial friction, is
illustrated schematically in Figure 5.5.
5.7.2.1. Residual Axial Resistance’

Residual axial resistance dominates the shape of the steady state force profiles and the
resulting thermal expansion, pipe-walking and post-buckling behaviour. However, there is
much uncertainty over the assessment of axial resistance in clays because:
1. The contact stress with the seabed (2 to 10 kPa) is below that utilised in traditional
geotechnical testing, or that considered in conventional geotechnical analysis;
2. The generation of excess pore pressure in the soil due to pipe movement is not well
understood.
Report No: 5087471/ 01 / A

Design Guideline
2.5 Elastic-slip Breakout - drained and undrained
2
Axial pipe-soil resistance
1.5
0.5
-0.5
-1
Undrained Brittle
-1.5 2nd Cycle
-2 Undrained Ductile
Drained
-2.5
Axial Displacement
Figure 5.5 Idealised Axial Pipe-Soil Resistance Behaviour

Excess pore pressure, generated by pipe displacement, leads to undrained conditions that
can either: (i) increase the level of resistance due to negative excess pore pressure; or (ii)
reduce the level of resistance due to positive pore pressure. In general, greater excess pore
pressure is generated with increasing pipe velocity; while the fully drained condition (with no
generation of excess pore pressure) usually provides the highest axial resistance. These
undrained (fast) and drained (slow) responses generally define the highest and lowest
values of axial resistance, as illustrated by the „Undrained‟ and „Drained‟ response curves in
Figure 5.5.
The generation of excess pore pressure is not only linked to pipe velocity but also to
pipe weight and pipe roughness. High pipe velocity can also introduce strain-rate
effects that increase the axial resistance. This interaction between pipe velocity,
weight and coating roughness must be considered carefully; the contribution of each
parameter to the axial resistance is the subject of ongoing research.
In sandy soils, axial displacement is not likely to generate excess pore pressure, so the axial
resistance is frictional and defined by the fully-drained response.
5.7.2.2. Axial Breakout Resistance

An initial high breakout resistance can occur under certain undrained conditions, usually
associated with higher velocity displacements, following a long set-up (duration with no
movement). However, a high breakout is often entirely absent and may not be relied upon to
occur. Therefore residual axial resistance should be used (without breakout) to model
thermal expansion, pipe-walking and post-buckling behaviour.
Not incorporating breakout is thought to be conservative in most design cases. Even
if a high breakout occurs, it has little influence on thermal expansion behaviour and
the generation of steady state force profiles.
Report No: 5087471/ 01 / A

Design Guideline
High breakout resistance will inhibit pipe walking under thermal transient loading - if
the breakout occurs on all cycles. However, in most cases, thermal start-up
transients follow a long period of pipe cooling, during which the pipe is continuously
contracting; so a breakout response is unlikely to occur on all walking cycles.
A high breakout resistance can produce a higher driving force that will tend to increase the
propensity for buckling. This is particularly true at hydrotest, where a long set-up time is
likely. If it can be demonstrated that long sections of the pipeline remain fully constrained
(with no axial displacement) prior to buckle formation then incorporating a breakout
resistance into the force profiles may be warranted, provided there is evidence to support a
high breakout resistance.
In most cases accounting for axial breakout is potentially non-conservative for buckle
formation and walking predictions. For this reason modelling the axial breakout
resistance should only be undertaken with great care.
Sensitivity checks on route curve pull-out should include axial breakout resistance in
cases where a long set-up, following an extended shutdown, and a subsequent
reduction in effective force (due to cooling for example) could increase the peak axial
tension along the route curve.
5.7.2.3. Axial Slip and Mobilisation Displacement

It is important to define the „axial slip‟ and „mobilisation displacement‟ carefully, as they are
very different concepts. Elastic slip is a concept within an FE element that usually defines
the elastic displacement to reach a residual resistance; while the mobilisation displacement
describes the true non-linear, elastic-plastic displacement to reach the peak or residual
resistance, as illustrated in Figure 5.5. This parameter influences buckle formation and pipe
walking response under thermal transient loads. Full mobilisation of „elastic slip‟ usually
occurs at much lower displacements than the much larger displacement to „breakout‟,
associated with full plastic deformation of the soil.
The larger mobilisation displacement should not be used to define elastic slip, as the
rate of pipe walking will reduce in proportion to this elastic displacement.
Simple bi-linear models for residual axial friction, incorporating the mobilisation
displacement, are generally sufficient for detailed design of lateral buckling.
Reference should be made to Appendix B for guidance on key parameters, established for
specific soils, and methodologies for extending these data to a range of pipe sizes and
weights.
5.7.3. Lateral Resistance

There are two stages of lateral pipe-soil response behaviour:-
 Monotonic behaviour at first load as the buckle is formed, characterised by a breakout
resistance and a residual or plateau resistance;
 Cyclic displacement in operation, characterised by the accretion of soil-berms and the
development of a trench with increasing cyclic embedment.
The methodology for assessment of each phase of lateral resistance is given in Appendix B.
Report No: 5087471/ 01 / A

Design Guideline
5.7.3.1. Lateral Breakout Resistance

Lateral breakout is associated with buckle formation (initiation) when the pipe breaks out
from its as-installed and embedded condition and starts to move laterally across the seabed.
In clay soils, breakout is normally accompanied by an elevated breakout resistance
associated with suction release. Following breakout there is a phase of elevation correction.
If the pipe is over-embedded the resistance falls away as the pipe is displaced and rises; if
the pipe is lightly-embedded or normally-penetrated this resistance will gradually increase
beyond breakout as the pipe embeds further into the soil. Therefore, the breakout force-
displacement response can take two forms as illustrated schematically in Figure 5.6 and
Figure 5.7.
Breakout resistance Berm resistance

Horizontal resistance
Residual resistance
Horizontal
,
displacement
Figure 5.6 Lateral Pipe-Soil Friction Behaviour – ‘Light’ Pipe
Breakout resistance Cyclic accumulation

of embedment &
resistance
Accumulating passive
Horizontal
resistance
displacement
Figure 5.7 Lateral Pipe-Soil Friction Behaviour– ‘Heavy’ Pipe
Report No: 5087471/ 01 / A

Design Guideline
5.7.3.2. Lateral Residual Resistance

After breakout, two characteristic types of large-amplitude lateral response are found in clay
soils, depending on the ratio of the flowline weight to the seabed strength (v=V/su.DOC). In
general, values of v below 1.5 give a „light-pipe‟ response, characterised by the pipe rising
with displacement (Figure 5.6), while values of v greater than 2.5 give a „heavy-pipe‟
response, characterised by the pipe diving with displacement (Figure 5.7).
For values of v < 1.5 („light‟ pipes), the pipeline usually rises after breakout, while lateral
resistance reduces from the break-out value to a steady residual resistance during the first
sweep. The flowline sweeps horizontally, mobilising an approximately constant (or slightly
rising) horizontal „residual‟ resistance. The residual resistance during the first load cycle
controls the lateral displacement at which the first buckle stabilises, defining the initial shape
of the lateral buckle and the peak bending stress in the pipe.
During the first shutdown cycle, the flowline reverses direction but does not return to the
initial position due to the reversal of the axial resistance along the feed-in section (and the
reversal of lateral resistance along the buckle lobes). Surface soil, swept ahead of the pipe
on each cycle, builds up into berms at the extremes of the pipe displacement. These berms
offer significant resistance to pipe movement and define the shape of the buckle in
operation. Subsequent cycles of lateral movement lead to a steady increase in the restraint
provided by the soil berms.
These soil berms can prevent the growth of buckle amplitude (ratcheting) on
subsequent cycles, so that cyclic displacements remain almost constant. This locks
in the stress range and prevents a reduction in the stress range with growth in the
buckle wavelength.
If feed-in increases in late life, due to an increase in operating temperature or
localised walking, buckle growth will be resisted by the soil berms. If the pipe breaks
through the berm at the point of highest curvature, this may cause excessive
localisation of the bending loads above that experienced in early life. Such an
assessment requires a very good understanding of the rate berm growth and
resistance with cycles, probably based on project specific tests.
For values of v > 2.5 („heavy‟ pipes), the pipeline usually moves downwards after breaking
out. This downward movement, coupled with the growth of a soil berm ahead of the pipe,
leads to a steady increase in the lateral resistance (Figure 5.7). This hardening form of
response (rising residual friction) gradually increases the load in the buckle while
simultaneously inhibiting further lateral displacement.
During the first shutdown cycle, the flowline reverses direction and continues to move
downwards into stronger soil, increasing the resistance to pipe movement with each
subsequent cycle. As the pipe penetrates deeper, soil can flow over the top of the
pipe, further increasing the download and resistance to lateral movement.
Under conditions of constant vertical load, a laterally sweeping „heavy‟ pipeline will
rapidly embed. However, in the areas where there is little cyclic movement, such as
at the stationary pivot points in a lateral buckle, there is little embedment. This
causes the contact pressure to vary, so that the rate of embedment and the lateral
soil resistance varies along the length of the buckle. Therefore, in regions of lateral
sweeping, the contact force and lateral soil resistance gradually reduce as the trench
gets deeper and soil berms are established at the extremes of displacement.
Consequently, the friction curves formulated using the nominal pipe weight can
significantly overestimate the lateral resistance at the crown of the buckle.
Report No: 5087471/ 01 / A

Design Guideline
Lateral buckles with uneven support conditions require careful consideration of the
effect of increasing contact pressure at support points and reducing contact pressure
at low points along the lateral buckle. This will occur with „heavy‟ pipelines but also
where a pipeline passes over a feature such as a sleeper or a vertical out-of-
straightness in the seabed.
In the absence of project specific soils tests, reference should be made to Appendix B for a
representative approach to establishing the values of key parameters such as cyclic frictional
behaviour and berm resistance.
5.7.4. Application in Conceptual Design

Lateral and axial pipe-soil interaction can be modelled using friction coefficients, where the
lateral breakout friction is used to assess buckle formation and the residual friction is used to
assess lateral buckling, walking and other loads.
For probabilistic buckle formation assessment the range of axial residual and lateral
breakout resistance is represented by a distribution with a mean and standard deviation, as
described in Section 4.6.
The assessment of the buckle response should be performed assuming the best estimate
value of axial resistance in combination with the maximum residual lateral resistance.
The assessment of walking may be performed assuming the best estimate axial residual
resistance. However, a sensitivity assessment should also be performed for the lower
bound axial resistance and the weighting of confidence in the walking predictions should be
based on the level of confidence in the axial friction response, which is discussed further in
the next section.
Sensitivity cases for best estimate and upper/lower bound values should also be performed
for the assessment of global response.
5.7.5. Application in Detailed Design

Probabilistic buckle formation assessments are again based on a distribution of axial and
lateral breakout friction with a mean and standard deviation.
There are two stages to the assessment of loading in a lateral buckle. The best estimate
value of axial resistance should be used for both stages[9]. The first stage is associated with
first load when the buckle is formed, and should employ the maximum value of lateral
monotonic resistance, which is a function of pipe weight and the as-installed embedment.
This should incorporate:-
 An estimate of the maximum pipeline embedment (including the effect of dynamics
during installation);
 The non-linear lateral resistance generated at breakout from the embedded condition;
this response is a function of pipe weight and may include an initial peak resistance at
break-out, or increasing resistance following breakout as the soil berm is established.
The second stage is associated with cyclic loading and soil-berm accretion. The full non-
linear pipe-soil resistance should be modelled; this will be focused on the upper bound
lateral resistance but it may also be necessary to evaluate lower bound and best estimate
resistance in some cases. This should incorporate:-
 The steady state cyclic resistance, which may be increasing with each cycle;
 The effect of soil berm resistance at the extremes of cyclic displacement.
Report No: 5087471/ 01 / A

Design Guideline
The assessment of pipe-walking and buckling interaction response may be performed

assuming the best estimate value of axial resistance and lateral resistance. However, it may
also be necessary to evaluate lower bound lateral resistances in some cases where the
walking response is dominated by pipe feeding out of the lateral buckles, causing a
reduction in buckle amplitude and an increase in pipe walking. In this case the lower bound
cyclic lateral resistance can result in the highest rates of walking.
Pipeline walking is not well understood. It has been observed in several operating
pipelines, and has been responsible for at least one failure and the need to intervene
to control unacceptable walking on at least two other pipelines. Although there is no
evidence of walking being a wide spread cause of failure in pipelines; there are many
reasons why this is so (for example there may only be a small number of operating
pipelines which have experienced significant number of cycles in the appropriate
regime). The recommendation here is therefore to assess the phenomenon under
“best estimate” conditions; however, caution is required in adopting this approach.
Pipeline walking is very sensitive to the level of axial restraint and the assessment
could be non-conservative, unless a lower bound resistance is employed. However,
the lower bound may only act over a short section of the pipeline and will probably be
lower than a global lower bound friction, which is the lowest credible value that
defines the true response over a long section of the pipeline. The choice between
using the global lower bound or the absolute lower bound should be based on the
level of confidence in the axial pipe-soil response being truly representative of
expected behaviour. Establishing the true expected behaviour is likely to require a
detailed assessment of axial response that captures the local variation of axial friction
with axial velocity along the pipeline, as well as the effect of the proposed coating
roughness and local pipe weight. In some two-phase flow systems it may also be
necessary to assess the effect of the variation of pipe weight in operation on axial
friction.
The current level of knowledge is such that the best estimate from a data set of tests
may miss the underlying mechanism that results in levels of axial resistance in
operation that are well below or well above the best estimate value. This has
occurred in practice, where the actual response has been more representative of
lower bound axial resistance.
While this status persists, it would be prudent to undertake a sensitivity analysis at
the lower bound condition, which generally would be the worst case for design. If the
results indicate that mitigation is required, then the design may be based on the best
estimate values with provision for mitigation to arrest walking later in the operating
life of the pipeline, if the actual response is closer to the lower bound.
Report No: 5087471/ 01 / A

Design Guideline
6. LIMIT STATES
The relevant failure modes within a lateral buckle are:-
 Local buckling (Section 6.2);
 Fatigue (Section 6.4);
 Fracture (Section 6.5).
In addition to these limit states, the pipeline material specification must be consistent with
the high levels of imposed load; this is addressed in Section 6.6.
The limits outlined in Section 6.1 to 6.6 are intended for general application. It is possible
that project specific assessments of the limit states will lead to enhanced system capacity.
Guidance on the potential for this approach is presented in Section 6.7.
All other limit states, for example pressure containment or hydrostatic collapse,
should be evaluated in accordance with the governing design code.
The design equations presented in this section are based on the requirements of API 1111[2].
For pipelines designed in accordance with DNV OS-F101, relevant modifications and
interpretations required for application of DNV-OS-F101 are given in Appendix C.
The approach outlined here is only intended for use in the assessment of the conditions in a
laterally buckled pipeline.
6.1. APPROACH TO PIPELINE SAFETY

The guideline is intended for pipelines transmitting hydrocarbon or water-based fluids.
The level of reliability required in these instances is not necessarily the same; this can be
expressed in terms of a failure probability. The target failure probability employed in this
guideline is outlined in Table 6.1.
Product DNV-OS-F101 Safety class Target failure probability
Hydrocarbon Medium 10-3 to10-4

Water-based Low 10-2 to10-3
Table 6.1 SAFEBUCK Target Failure Probability (per Pipeline)
The failure probabilities outlined in Table 6.1 are per pipeline due to lateral buckling
(all buckles).
The design process ensures that these target levels of reliability are met through the safety
margins inherent in the underlying design code and the definition of Characteristic VAS.
The definition of Characteristic VAS outlined in Section 4 is consistent with the
pipeline failure probabilities defined in Table 6.1. If the pipeline meets all the
applicable design unity checks at the Characteristic VAS, then the pipeline failure
probability will be equal to or less than the target values. The choice of exceedance
probability employed in the definitions is based upon the results of the SAFEBUCK
SRA [9, 10].
Report No: 5087471/ 01 / A

Design Guideline
To date at least four pipelines have ruptured as a result of lateral buckling; another
has been abandoned early in life with little chance of meeting its original design life.
Three failures involved pipelines in which lateral buckling was not seriously
considered as a design criteria, and two in which it was - albeit with a seriously
flawed design process.
Basing actual failure statistics on these failures alone indicates a historical failure rate
in excess of the target values defined in Table 6.1. Clearly, this is not a like for like
comparison and there are generally a number of causes to any failure. Nevertheless,
this does indicate that previous design practice was not consistent with the desired
levels of reliability. The guidance in this recommended practice is intended to ensure
that the pipeline design does meet these targets.
6.2. LOCAL BUCKLING
6.2.1. Pure Bending

The critical buckling strain in pure bending is:-
t
 c  0.5  .... 6.1
D0
The superior strain hardening behaviour of CRA materials means that Equation 6.1
may be unduly pessimistic. A project specific investigation may produce a significant
enhancement in the buckling strain limit.
The imposed strain should comply with the following equation:-
c
2  .... 6.2
f2
Where the bending safety factor, f2 should be calculated from:-
 D 
f2  1.4  1  0.02  0  .... 6.3
 t 
This recommendation applies to pipelines transporting hydrocarbon. For pipelines

transporting non-flammable water-based fluids, Equation 6.2 can be used with a bending
safety factor of:-
 D 
f2  1.2  1  0.02  0  .... 6.4
 t 
The strain to be employed, ε2, is the absolute value of the maximum compressive axial
mechanical strain, i.e. any thermal strain should be removed from the total compressive axial
strain prior to use in Equation 6.2.
Alternatively the designer can employ the bending strain in the design check:-
D0
b    .... 6.5
2
This approach is consistent with the test data base used to develop the strain limits.
Report No: 5087471/ 01 / A

Design Guideline
6.2.2. Bending and External Overpressure

For bending imposed in the presence of an external overpressure, the imposed strain should
comply with the following equation:-
f2   2

p e  p i   g .... 6.6
c fc  p C
Where g is the collapse reduction factor:-
1
g .... 6.7
1  20  
And the collapse factor for combined loading is defined by:-
1.2  fo
fc  .... 6.8
g
The collapse factor, fo, is equal to 0.6 for cold expanded pipe and 0.7 for seamless pipe.
The collapse pressure is defined by:-
p YC  pE
pC  .... 6.9
p 2YC  pE2
Where:-
 t 
p YC  2   yc    .... 6.10
 Do 
3
2E  t 
pE    
1   2
 Do 
.... 6.11
The pipe ovality is defined in accordance with API 1111:-
Dmax  Dmin
 .... 6.12
Dmax  Dmin
This definition differs from the DNV-OS-F101. For the same cross-section the
API 1111 ovality is half the DNV-OS-F101 ovality, i.e. the pipe ovality must be less
than 3% based on the DNV-OS-F101 definition.
For reeled pipelines, the post installation ovality and any potential for a reduction in
yield stress due to reeling should be employed in the design check.
In applying Equation 6.6 the minimum level of internal pressure that can occur during
operation should be employed. The minimum internal pressure may result from a blowdown
condition, although such a condition may be associated with low operating temperature.
6.2.3. Bending and Internal Overpressure

If the pipeline is subject to an internal overpressure, the critical buckling strain given by
Equation 6.1 may be increased:-
c
2   fp .... 6.13
f2
Report No: 5087471/ 01 / A

Design Guideline
Where the factor[25], fp is:-

2
 
fp  1  4   h  .... 6.14
 y 
 
Where the hoop stress is given by:-

pi  Di  p e  D o
h  .... 6.15
2t
In applying this equation the following interpretations are required:-

 The hoop stress should be based on the minimum level of internal over pressure which
can co-exist with the imposed temperature during loading;
 The SMYS at ambient temperature must be employed.
6.2.4. Axial Force

For modest levels of effective axial force the strain limits for pure bending apply. When high
effective axial forces are present a specific assessment of the pipeline strain capacity must
be undertaken.
The relevant axial force is the post-buckle effective force in the system. For a single
pipe it is likely that the post buckle effective force will be modest. However, for a PIP
system the level of effective axial force in the individual pipes may be significant.
6.2.5. Seamless Linepipe

For a seamless linepipe the local buckling performance of the pipeline is compromised by
the plateau in the stress-strain response. For internal overpressure, f2 should be modified to
incorporate the influence of the plateau[25]:-
 D 
1.6  1  0.02  0 
f2   t 
..... 6.16
  25  fp  D0 / t 
1  0.45  tanh 
  2 .5  f 
 p 
The suitability of the local buckling design equations for use at a lateral buckle has
been evaluated in the SAFEBUCK JIP[9]. This found that the basic equations are not
acceptable when failure is at strains below the Lüder strain. For internal over
pressure this is only likely to apply to for pipes with D/t>30 and σh/σy<0.5.
The SAFEBUCK SRA performed a comprehensive assessment for the internal
overpressure condition. For external overpressure a very limited evaluation was
performed[10]. This showed that the buckling capacity is lower and is significantly
compromised by the Lüder strain. Insufficient work was undertaken to develop
general design guidance. If this limit state is critical the project should undertake
specific work to ensure that the design is conservative.
6.2.6. Treatment of Corrosion

All resistance calculations may be based on the nominal wall thickness, but it must be
demonstrated that this is reasonable. In the absence of such a demonstration, the corroded
wall thickness must be adopted in all resistance calculations.
Report No: 5087471/ 01 / A

Design Guideline
A corrosion allowance is routinely added to carbon steel pipelines, but the corrosion
mechanism is rarely considered in structural design. For example, pitting is a
common form of corrosion for which the corrosion allowance might be intended. It is
probable that small isolated pits will have little or no effect on the buckling resistance
of the pipeline. Corrosion involving more significant areas of metal loss, for example
six o‟clock grooving, may have a greater effect on the collapse resistance. However,
even this might have a limited effect on the bending buckling resistance of a pipeline
when the bending is in the horizontal plane (as it is for lateral buckling). Only if the
corrosion anticipated involved 360º metal loss would removal of the allowance in the
resistance calculation be wholly required.
In addition, corrosion is a time dependant phenomena. If the operating conditions
reduce with time; combining the early life loads with late life corrosion is
unnecessary.
In all cases, the imposed load should be calculated based upon the nominal pipeline
thickness.
Undertaking the load calculations with a corroded pipe cross section will normally
reduce the loads. Using these reduced loads in the design is generally non-
conservative and not representative of the pipe response.
6.2.7. Shutdown Condition

The local buckling capacity of the pipeline does not need to be evaluated in the shutdown
condition.
The local buckling capacity of the pipeline is improved by internal over-pressure. On
unload the strain in the buckle may not reduce significantly. The combination of
unload pressure and strain may develop a higher unity check than the load condition.
However, unloading is typically an elastic process, and local buckling cannot occur
as part of an elastic unloading process. Thus the local buckling check need only be
performed for combinations of strain and internal pressure, during which plastic
deformations increasing the compressive strain at the local-buckling-critical location
could develop.
6.2.8. Clad Pipe and Lined Pipe
6.2.8.1. Clad Pipe

The wall thickness of the pipeline employed in the local buckling design equations may
include the thickness of the liner material.
6.2.8.2. Lined Pipe

The wall thickness of the pipeline employed in the local buckling design equations should be
the thickness of the base carbon steel.
The liner will normally separate from the base steel in pipes with low or no internal pressure.
For pipes subject to these pressure conditions, the liner is likely to wrinkle before the
buckling capacity of the base pipe is reached. For lined pipes the local buckling capacity of
the liner should be subject to specific study.
6.3. FATIGUE AND FRACTURE

Fatigue is a damage process whereby a crack can form and grow under the action of
fluctuating (cyclic) loads. Fracture refers to the failure of a flaw subject to an applied load.
The fatigue and fracture limit states are related, as illustrated in Figure 6.1.
Report No: 5087471/ 01 / A

Design Guideline
LATERAL
BUCKLING
DESIGN
S-N ECA
S-N pipeline geometry
REVISE weld geometry
DESIGN*
IS FATIGUE tensile stress
DAMAGE N tensile properties
ACCEPTABLE?
Y stress range
number of cycles
ECA stress concentration factor
REVISE
fracture toughness
DESIGN*
IS FLAW SIZE S-N curve fatigue crack
ACCEPTABLE growth law
N AND
PRACTICAL? correction to S-N curve / fatigue crack
Y growth law / toughness for environment,
temperature and frequency
FATIGUE & FRACTURE factor of safety*
LIMIT STATES (*this may be different for the S-N and the ECA)
ACCEPTABLE
*REVISIONS MAY INVOLVE MODIFYING SOME

OF THE ASSUMPTIONS IN THE S-N/ECA
CALCULATIONS, CONDUCTING MORE
DETAILED CALCULATIONS, OR CHANGING
THE LATERAL BUCKLING DESIGN
Figure 6.1 The Fatigue and Fracture Limit States and the S-N and ECA Calculations
The two methods used to assess fatigue are:
 S-N curves; and
 Fracture mechanics.
The fatigue limit state should first be addressed through the S-N approach to fatigue. If the
results of the S-N calculations are acceptable, then an ECA should be conducted. The ECA
addresses both the fatigue limit state, through the fracture mechanics approach to fatigue,
and the fracture limit state. Although the ECA calculations require more information than the
S-N calculations, most of the required information is common to both calculations, see
Figure 6.1.
The two sets of calculations are complimentary. S-N curves are only appropriate if the weld
is free from significant defects. The presence of flaws may reduce the fatigue life below that
predicted using S-N curves. One of the objectives of the ECA is to determine this significant
flaw size. The guidance in Section 6.4 is mainly concerned with the S-N calculations and
that in Section 6.5 with the ECA calculations.
Fracture mechanics calculations should not be used to justify a fatigue life in excess
of that predicted using S-N calculations.
In the fatigue assessment, it is necessary to consider all sources of fatigue loading, from
installation through to the end of the design life of the pipeline in order to identify the limiting
case. Typically, in a pipeline designed to buckle laterally, the limiting condition will be
defined by the fatigue loading arising from the movement of the buckle due to start-up and
shut-down cycles. In some cases, other forms of fatigue loading may be significant (e.g.
fatigue loading of spans associated with lateral buckles due to environmental loads or
slugging flow).
When considering the fatigue loading due to start-up and shut-down cycles it is
typically convenient to consider a number of different load cases, based on the
duration of a shut-down (e.g. less than 3, 6 or 12 hours, greater than 24 hours)
Report No: 5087471/ 01 / A

Design Guideline
and/or the operating regime (e.g. normal operations, hot-oiling, cold displacement
etc.).
The guidance here assumes that the fatigue life will be governed by the fatigue
behaviour of the pipeline girth welds. This may not be the case if there are other
sources of significant stress concentrations in the pipeline. For example, the
presence of J-lay collars, buckle arrestors or anode attachment pads will lead to local
increases in stress. If possible, these design features should be avoided in pipelines
that are designed to laterally buckle. If they cannot be eliminated, then they should
be made as fatigue friendly as practical and the design must quantify the stress
concentration and address the influence of these features.
S-N curves and fatigue crack growth laws are specific to the environment, temperature and
loading frequency. Project specific testing may be required to establish suitable values,
particularly if the internal environment is sour or is expected to sour, see Section 6.7.
Souring of the reservoir can occur due to either bacterial or thermo-chemical
reduction of aqueous sulphates, or thermal decomposition of sulphides in sulphur
rich source rocks. Bacterial souring of the reservoir can occur because of bacteria
introduced into the reservoir with the injection water. Typically, bacterial souring of
the reservoir occurs a number of years after start-up. Following souring, typical H2S
production levels through field life are in the region 50 to 200 ppm (dependant upon
reservoir characteristics and injection water treatment).
At conceptual design, some of the information required to address the fatigue and fracture
limit states (e.g. fracture toughness, effect of the environment) may not be available.
Sensitivity calculations should be conducted to determine the significance of the various
assumptions, and hence whether project-specific testing and/or more detailed calculations
are likely to be required in detailed design.
6.4. FATIGUE: S-N APPROACH
6.4.1. S-N Curves for Girth Welds

S-N curves can be used to estimate the fatigue life of a welded joint. An S-N curve presents
the fatigue life, N, as a function of the applied stress range, S. S-N curves of welded joints
are based on endurance tests of workmanship quality welds (containing no significant flaws).
Welded joints are classified according to joint geometry, and for each class there is a
different S-N curve. The class of an S-N curve refers to a particular mode of fatigue failure,
e.g. initiation at the weld toe or the weld root, so more than one class may apply to a
particular welded joint. The effect of stress concentrations due to the weld shape and type
are included in the S-N curves of welded joints.
In addition to classifying the welded joint, to select the appropriate S-N curve, the effect of
plate thickness, the environment, and gross structural discontinuities and deviations from the
intended design shape, need to be taken into account.
The basic system of S-N curves recommended here is that in BS 7608[26] and
PD 5500[20]. S-N curves of welded joints are based on endurance tests of
workmanship quality welds. Design S-N curves, as given here, are mean minus two
standard deviation curves to the experimental data, to account for the scatter in this
data. The S-N curves are referenced to a material with a modulus of elasticity of
209 GPa, corresponding to a ferritic steel at ambient temperature. PD 5500 gives
guidance on correcting the S-N curves for other materials and at other temperature in
terms of the change in the modulus of elasticity. DNV-RP-C203[27] states that the S-
N curves are applicable to duplex and super duplex steel and austenitic steels. It
Report No: 5087471/ 01 / A

Design Guideline
also states that the S-N curves are applicable up to 100°C. The guidance in
PD 5500 is more conservative.
The recommended classifications for single sided pipeline girth welds are outlined in Table
6.2.
Initiation Location Weld Classification Thickness Exponent (k)

Cap D 0.15
Root with no backing or temporary backing E 0
Root permanent backing F 0
Table 6.2 S-N Curves for Single Sided Girth Welds

The classifications are applicable to welds in which the axial misalignment (offset of mid-
thickness of wall) is less than 15% of the wall thickness (up to a maximum of 4 mm). If the
misalignment can exceed these limits, the basic S-N categorisation may not be applicable.
The condition of the root bead has a significant effect on the fatigue performance of a girth
weld. A small hi-lo at the weld root is required to achieve a root with a favourable profile.
Typically, a hi-lo of not exceeding 0.5 mm is required for girth welds to achieve a high fatigue
performance.
Some welding processes result in very poor root profiles, which can compromise
fatigue performance. It is essential to define what weld classification the installation
contractor will achieve for their proposed welding process, prior to detailed design.
This includes welding carried out at the coating yard, spool-base or offshore, and
should consider all possible types of girth weld joint. The welding specification
should include limits on both the maximum acceptable misalignment and the
maximum acceptable hi-lo.
The fatigue life, N, for a given stress range, S, is calculated using the following formula,
where A and m are defined in Table 6.3.
N  A  S m ..... 6.17
Classification A m
D 1.52x1012 3
12
E 1.04x10 3
11
F 6.33x10 3
Table 6.3 Constants in the in-air S-N Curves
The constant A in Table 6.3 is not dimensionless. The values defined are applicable when
the stress range is defined in N/mm2.
Miner‟s Rule can be used to combine the effect of different cycles of different stress ranges
(e.g. full and partial shut-downs) to give the total fatigue damage ratio.
n
D fat   Ni   fat ..... 6.18
i i
Suitable values for the allowable fatigue damage ratio are given in Table 6.4.
Report No: 5087471/ 01 / A

Design Guideline
Service Allowable Fatigue Damage Ratio

Hydrocarbon 0.20
Water-based 0.33
Table 6.4 Allowable Fatigue Damage Ratio
The allowable fatigue damage ratio is the inverse of the factor of safety on the
predicted fatigue life. A limit of 0.2 corresponds to a factor of safety of 5. In some
cases, higher factors of safety may be required.
6.4.2. Stress Range

The stress range, S, used in the S-N curve is the local stress range at the weld, including the
effects of axial misalignment and plate thickness. The local stress range is related to the
nominal stress range, R, by the following expression:
S  R  K t  Km ..... 6.19
The nominal stress range must not exceed the elastic limit defined by the Bauschinger
effect; the appropriate limit is outlined in Section 6.6.
The applicability of S-N curves to the high stress-low cycle regime has been
established in SAFEBUCK. There is no limit on the local stress range, so long as the
nominal stress range meets the requirements of Section 6.6.
The calculation of nominal stress range should account for the stress concentration
associated with field joint geometry.
6.4.2.1. Effect of Thickness

The fatigue performance of welded joints in plates reduces with increasing plate thickness.
This effect is incorporated as follows:
k
 t 
K t    t  t ref ..... 6.20
 t ref 
Where tref is 16 mm and the exponent k is defined in Table 6.2. For thicknesses less than
16 mm, Kt is unity.
6.4.2.2. Effect of Misalignment

The stress concentration factor (SCF) due to axial misalignment should be determined using
the equation:
e
Km  1  3 ..... 6.21
t
Where e is the eccentricity, defined as the axial misalignment between the wall thickness
centrelines of the two pipes.
There are a number of different expressions for calculating SCFs due to axial
misalignment. Several of these expressions, such as those proposed by Connelly
and Zettlemoyer[28], are known to under-predict SCFs for typical flowline dimensions
and misalignments. Equation 6.21 is known to be conservative, and should be used
Report No: 5087471/ 01 / A

Design Guideline
at the early stage of design. As the design progresses, a flowline specific analyses
may be performed to develop a more accurate expression.
In determining an appropriate axial misalignment and hence SCF, due account should be
taken of the line pipe manufacturing tolerances and the welding tolerances. The simplest,
and most conservative, approach is to calculate it from the maximum permitted hi-lo, and the
wall thickness and diameter tolerances from the relevant line pipe specification. However,
this is likely to be over-conservative (and may be detrimental to the design). If line pipe and
welding dimensional data is available then statistical methods can be used to determine an
upper bound misalignment with a given probability of exceedance. In the absence of
project-specific data, data from previous projects can be used (if available) and its
appropriateness confirmed prior to fabrication using project-specific data.
6.4.3. Installation
The extent to which the fatigue loads during installation have implications for operation, and
vice versa, will vary from case to case. The largest cyclic loads during installation will occur
in a plane transverse to those during operation if the pipeline does not rotate and the lateral
buckle is confined to the horizontal plane. In this case, the interaction between the two will
be small (there is likely to be a small component of fatigue loading in the horizontal plane
during installation, and similarly in the vertical plane during operation). If the pipeline does
rotate, or if the lateral buckle is not confined to the horizontal plane, then the interaction
between the two will be more significant.
In conceptual design, it is recommended that 20% of the total allowable fatigue damage is
assumed to be consumed during installation (e.g. if the allowable fatigue damage is 0.20,
then the limit for fatigue loading during operation is 0.16).
In detailed design, the fatigue loading during installation should be determined by the
installation contractor.
6.4.4. Effect of the Environment

S-N curves and fatigue crack growth laws should be appropriate to the environment,
temperature and loading frequency.
In a corrosive environment, the fatigue life is lower and the rate of fatigue crack growth is
higher. The endurance limit and threshold for the initiation of fatigue crack growth are lower
or non-existent.
Corrosive fluids in a pipeline are characterised as sweet (i.e. no H2S) or sour. In these
environments, the fatigue performance of the weld root will be worse than that in air.
The fatigue performance of welds exposed to seawater, even if protected by cathodic
protection, will also be worse than in air. Although this is not recognised in the current
revision of BS 7608, separate design curves are recommended in DNV-RP-C203. At high
stress ranges these are typically a factor of 2-3 lower than the corresponding curve in air. In
any case, neither of these codes take account of the influence of cyclic loading frequency.
Therefore a higher factor is recommended, as detailed in Table 6.5, derived from fatigue
tests undertaken by SAFEBUCK.
The presence of sulphate reducing bacteria in soils may also result in similar effects to those
seen in sour environment. However, there is insufficient published information to offer
quantitative guidance in this case.
If a high-integrity field joint coating is used then the weld cap is protected from the external
environment and in-air behaviour can be assumed.
Report No: 5087471/ 01 / A

Design Guideline
A high integrity field-joint coating should demonstrably exclude sea-water from the
weld under all loads that the pipeline will experience from installation to the end of
the design life.
Lateral buckling imposes a very low frequency loading (less than 10 -5 Hz). Corrosion-fatigue
is sensitive to the loading frequency. In this context, per cycle, low frequency loading is
more severe than high frequency loading.
Knock-down factors (also referred to as fatigue life reduction factors) to be applied to the
in-air S-N curves are given in Table 6.5. In the absence of better data, these are suitable for
use in conceptual design. However, project-specific testing is recommended, see Section
6.7.2.
The guidance assumes low carbon-manganese line pipe steel. Refer to Section 6.4.5 for
other materials.
The tentative recommendations given here are based on fatigue tests undertaken by
SAFEBUCK and the limited number of tests in the published literature[29-49]. They
may be non-conservative. Environmental parameters such as pH, temperature, and
the partial pressures of H2S and CO2 are likely to have an effect on the observed
knockdown factor although quantitative guidance is not currently available. Project
specific testing is therefore recommended.
The range of values provided for sour environment reflect the range of environmental
conditions that may occur in different projects. For low pH conditions and high partial
pressures of H2S, factors towards the upper end of this range may be expected. For
higher pH conditions and/or lower partial pressures of H2S, factors nearer the lower
end of this range may be more likely.
During conceptual design, it is useful to compare the anticipated environmental
conditions within the pipeline with those used in previous testing programmes. It is
noted however that comparison of different test data is not straightforward, as
differences in cyclic loading frequency, stress range, and in-air fatigue performance,
can themselves have an influence on the apparent knockdown, and mask the true
environmental effect. In the latter instance it is recommended that knockdown factors
are determined with respect to an absolute baseline (e.g. a design S-N curve) rather
than the actual in-air performance.
Environment S-N curve knock-down factor

Sweet corrosion 3
Sour corrosion 10-40
Sea-water and cathodic protection1 9
1
Sulphate reducing bacteria No information
Note 1 In-air behaviour should be assumed if a demonstrable high-integrity field joint coating is used.
Table 6.5 S-N Curve Knock-down Factors for Corrosive Environments

There is little test data to support the tentative recommendations for sweet corrosion.
6.4.5. Materials other than Carbon Steel

Corrosion resistant alloys (CRA) may be used in pipelines subject to high levels of corrosion.
The corrosion-fatigue behaviour of CRA materials is better than carbon-manganese steels,
but there remains significant uncertainty.
Report No: 5087471/ 01 / A

Design Guideline
In non-corrosive environments, in-air behaviour can be assumed.

A suitable CRA material should exhibit little degradation of fatigue performance in a
corrosive environment. In this case in-air behaviour can be assumed. However, material
selection to avoid general or localised corrosion does not necessarily imply immunity to
corrosion fatigue.
In particular, there is some evidence to suggest that super duplex stainless steels suffer a
knockdown in sour environment [46]. Therefore it is recommended that some reduction is
applied in this instance, where either the base metal or weld metal are of this alloy type. A
tentative factor of 10 is recommended, although it is acknowledged that this is based on very
few actual test data. Performance is also likely to depend on specific environmental
conditions. Project specific testing is therefore recommended.
Performance of super duplex material in seawater (with cathodic protection) is generally
considered to be comparable to that in air, although there are again very few published test
data, in particular at very low cyclic loading frequencies. However, a high integrity coating
may again be considered to exclude seawater and ensure in-air performance.
The fatigue performance of clad or lined pipe is also potentially inferior to that of solid CRA
or carbon steel materials. In particular, it is noted that conventional fatigue design curves are
derived from tests measuring the number of samples to specimen failure, where failure is
represented by crack growth through a significant proportion of the wall thickness. In the
case of a clad or lined pipe, the growth of a surface root flaw to this extent is not tolerable, as
corrosive fluids would then come into contact with the underlying carbon steel material.
Growth of surface root flaws needs to be confined to the CRA layer, which is typically only 3
mm. Some reduction in the allowable fatigue loading is therefore appropriate. Attention is
drawn to the DNV JIP “Lined and Clad Pipeline Materials” which provides some guidance in
this respect [47]. Project specific testing is recommended.
6.4.6. Alternative Weld Details and Stress Concentrations

Reference should be made to BS 7608 or PD 5500 for the classification of welded details
other than single sided girth welds. If a suitable classification is not clear, then endurance
testing should be undertaken to support the classification.
The effect of other geometric stress concentrations, if present, should be considered. This
may require a detailed stress analysis.
Examples of alternative weld details include anode attachment pads, fillet welds at
PIP field joints or load carrying attachment welds.
6.5. ECA
A design ECA is conducted to determine the tolerable flaw size in all of the structural welds
subject to the loads imposed by the lateral buckling design. It is recommended that the ECA
is conducted in accordance with the guidance in BS 7910[50]. The use of other codes and
standards (e.g. API 579[51] and R6[52]) should give broadly similar results.
In this design ECA, it is necessary to consider all load cases, from installation through to the
end of the design life of the pipeline in order to identify the limiting case, and hence the
smallest tolerable flaw size. Typically, in a pipeline designed to buckle laterally, the limiting
condition will be defined by the maximum load in the buckle at the end of the design life, and
the fatigue loading due to the movement of the buckle.
Examples when there may be other limiting cases include: pipelines installed by
reeling and pipelines operating in sour service.
Report No: 5087471/ 01 / A

Design Guideline
The ECA should consider both surface and embedded flaws. It should consider surface
flaws in both the weld cap and the weld root. In a PIP system it should consider both the
flowline and jacket welds.
The welds should be over-matched.
The tolerable flaw size determined from the ECA should be compared with flaw sizes that
would be acceptable to the relevant welding code, or could realistically escape detection
during production. If the tolerable flaw size is smaller then typical workmanship acceptance
levels, then the constructability of the pipeline may be compromised. The accuracy and
reliability of the inspection technique should be taken into account when determining the
acceptable flaw size.
Typical workmanship acceptance levels are: 3 mm deep by 25 mm long for a surface
flaw, and 3 mm deep by 50 mm long for an embedded flaw.
Project-specific testing may be required to determine the tensile properties, the fracture
toughness, and the fatigue crack growth law to be used in the ECA. Recommendations for
project specific testing are summarised in Section 6.7.
6.5.1. ECAs in Conceptual Design and Detailed Design

An ECA requires a large amount of detailed information, e.g. geometry, material properties
(including fracture toughness), installation and operational loads, welding procedures,
environmental effects, etc. Not all of this information will be available in the early stages of
the design cycle. The following phased approach is recommended:
 A preliminary ECA during conceptual design; and
 A detailed ECA during detailed design.
At the conceptual design stage, the objective of the ECA is to determine whether or not
typical workmanship acceptance levels (e.g. surface planar flaws limited to 3 mm deep by
25 mm long) are fit-for-purpose. Sensitivity calculations should be conducted to determine
the importance and implications of the various assumptions. It may be possible to use
information from previous projects to supplement the limited information available in
conceptual design. The level of complexity of the ECA should be appropriate to the quality
of the available data. The ECA may indicate that the tolerable flaw size is smaller than
typical workmanship acceptance levels. It is important to be aware of such possible
constraints on the fabrication of welds early in the design cycle. The implications should be
discussed with the installation contractor. The results of the preliminary ECA will inform
decisions on the need to:
 Conduct more sophisticated analyses to reduce the conservatism in the calculations;
 Undertake project-specific testing; and
 Modify the design.
At the detailed design stage, the objective of the ECA is to determine the tolerable flaw
sizes. The assessment may simply be to confirm that typical workmanship acceptance
levels are fit-for-purpose, or it may be to derive detailed acceptance limits. Any assumptions
or simplifications in the preliminary ECA due to limited data should be resolved in the
detailed ECA. The level of complexity of the ECA should be appropriate to the quality of the
available data and the relative severity of the loading.
6.5.2. Assessment Procedure

During conceptual design, an ECA should be conducted to Level 2A.
Report No: 5087471/ 01 / A

Design Guideline
If the material is expected to exhibit a Lüder plateau and an actual or estimate stress-
strain curve is unavailable, then the approximate method given in BS 7910 should be
applied.
During detailed design, an ECA should be conducted to Level 2B, 3B or 3C, depending upon
the severity of the loading. A Level 3 assessment would be required if high plastic strains
are introduced during reeling or operation.
An assessment to Level 2B requires a material specific stress-strain curve. An
assessment to Level 3B or 3C requires a material specific stress-strain curve, and a
tearing resistance curve.
6.5.3. Material Tensile Properties

The tensile properties used in the ECA should be those of the line pipe, not the weld. This
approach is conservative provided that the girth weld over-matches the line pipe.
For material in the elastic range, the virgin stress-strain curve and a flow stress based on the
virgin yield strength and tensile strength should be used. For material that has plastically
deformed, a pre-strained stress-strain curve and a flow stress based on the pre-strained
yield strength and tensile strength should be used.
The material tensile properties should be appropriate to the assessment temperature. In the
absence of project specific test data at the temperature of interest, the temperature de-rating
factors given in Section 5.5 should be applied to the yield and tensile strengths.
The tensile properties used in the ECA should be identical to those used in the stress
analysis of the lateral buckles.
The sensitivity of the results to the assumed tensile properties should be investigated. It
does not follow that the use of specified minimum tensile properties is always conservative
(the loads in a buckle may increase as the yield and/or tensile strength increase).
6.5.4. Fracture Toughness

For material in the elastic range, the virgin fracture toughness should be used. For material
that has plastically deformed, the pre-strained fracture toughness should be used.
The fracture toughness should be appropriate to the assessment temperature.
The effect of the local environment and the temperature on the fracture toughness should be
considered (e.g. sour conditions). It may be necessary to use more than one measure of
fracture toughness, e.g. KISCC for surface flaws at the weld root, and KIH for embedded flaws
and surface flaws at the weld cap.
The fracture toughness should be determined using a test specimen appropriate to the flaw
in the structure, considering issues such as microstructure, constraint and applied loads.
The two standard fracture toughness test specimens are the single edge notch bend
(SENB) specimen and the compact tension (CT) specimen. Both of these test
specimens introduce a high level of constraint at the crack tip. The level of constraint
at the tip of a crack in a girth weld in a pipeline subject to bending loads is lower than
that in the SENB or CT test specimens. Therefore, these test specimens will give a
conservative measure of the toughness. There are alternative test specimens that
introduce lower levels of constraint, similar to that of a crack in a girth weld in a
pipeline subject to bending loads. The single edge notch tension (SENT) specimen
is such a specimen. There is currently no standard for conducting SENT tests,
although DNV-RP-F108 gives some guidance[53].
Report No: 5087471/ 01 / A

Design Guideline
It has been demonstrated that the results of SENT tests can be used when the
pipeline is subject to longitudinal loading only (e.g. installation). It has not been
demonstrated that the results of SENT tests can be used when the pipeline is subject
to internal pressure and longitudinal loading (e.g. lateral buckling). The implications
of constraint and bi-axial loading at high levels of plastic strain on the crack driving
force and resistance are unclear. If it is proposed to use the results of SENT tests, or
if a constraint correction factor is to be applied to the results of SENB tests, then a
detailed, project-specific, justification is required.
6.5.5. Stress Concentrations due to Weld Geometry

The weld geometry gives rise to local stress concentrations.
In this context, the weld geometry is defined by the level of misalignment (e) and the
attachment length. The attachment length is the width of the weld cap for surface flaws in the
weld cap, and the width of the weld root for surface flaws in the weld root.
The guidance in section 7.1.1 on SCFs due to misalignment is applicable to both S-N and
fracture mechanics calculations.
An alternative to the stress-based approach to estimating the SCF due to misalignment is a
strain-based approach. A Neuber analysis can be used to estimate the elastic-plastic SCF
that corresponds to an elastic SCF[53], as follows
eqN  eqN  eq  eq  Km

2
.... 6.22
Where Km is defined in Equation 6.21. The intersection of the Neuber curve (Equation 6.22)
with the engineering stress-strain curve for the parent pipe material defines the strain and
stress corresponding to the elastic SCF due to misalignment.
The Neuber approach is only applicable to static loads. The elastic SCF should be used in
fatigue calculations (both S-N and fracture mechanics).
6.5.6. Stresses
The stresses used in the assessment should be principal stresses.
For a circumferential flaw in a girth weld in a lateral buckle, it is normally appropriate
to assume that the axial stress is a principal stress.
The load cases that need to be considered will vary from case to case. The load history
from an FE analysis of several operation cycles is informative. The maximum tensile axial
stress and the maximum axial stress range should be identified. It would be appropriate to
take into consideration any reduction in the maximum stress and stress range over a number
of cycles, if the response of the lateral buckle is shown to stabilise. The maximum stress
and stress range may not be co-incident, and may not occur at the extrados of the crown of
the buckle. It may be necessary to consider flaws at several positions around the
circumference of the pipeline, in order to identify the limiting case. The variation of the
maximum stress and stress range around the circumference is informative. If the toughness
varies significantly over the design temperature range it will be necessary to consider the
variation of the axial stress with temperature; the maximum design temperature may not be
the limiting case.
The use of outer-fibre stresses will typically be conservative. If mid-wall stresses are used
then the variation of the stresses through the thickness should be considered.
Report No: 5087471/ 01 / A

Design Guideline
Secondary stresses such as residual welding stresses should be considered. The residual
stresses should be assumed to be uniform and modified in accordance with the guidance in
BS 7910 on the relief of residual stresses under an applied load.
6.5.7. Limit on Ratio of Applied Load to Yield Load

The cut-off to the ratio of the applied load to the yield load (Lrmax) should be taken to be equal
to the ratio of the tensile strength to the yield strength (rather than the ratio of the flow stress
to the yield strength).
This approach is recommended in DNV-RP-F108[53]. If high plastic strains are
introduced during installation or operation, it may be necessary to confirm the
suitability of this assumption by project-specific testing, or, if possible, by reference to
previous test results.
6.5.8. Bi-axial Loading

A lateral buckle in a pressurised pipeline is in a state of bi-axial loading (ignoring the small
radial stress), and there may be plasticity in the buckle. Depending upon whether the hoop
stress is tensile or compressive, the axial stress at yield will be greater or less than the uni-
axial yield strength. A finite element analysis explicitly takes into account the effect of the
stress state on the yield surface. The ECA is typically based on uni-axial tensile properties.
This can result in inconsistencies between the yield strength in the FE analysis and that in
the ECA, and an overly conservative assessment. The implications of bi-axial loading at
high levels of plastic strain on the failure assessment diagram are unclear. There is limited
guidance for how to address this potential over conservatism, short of conducting an
assessment to Level 3C.
Assuming that the yield surface is described by the von Mises yield criterion, then for
a given hoop stress, H , and uni-axial yield strength, Y, the axial stress at yield is
given by the following equation
H 3 2
L    Y2   H .... 6.23
2 4
Therefore, a simple modification would be to calculate the load ratio as the ratio of
the applied load to the axial stress at yield (rather than the ratio of the applied load to
the uni-axial yield strength).
An alternative modification would be to calculate the plastic collapse moment for the
given flaw geometry, pipeline geometry, tensile properties and applied loads. The
load ratio would then be calculated as the ratio of the maximum moment from the
lateral buckling analysis to the plastic collapse moment. This approach represents a
step towards a Level 3C analysis.
Neither of the above two modifications considers the effect of the bi-axial stress state
on the crack driving force or resistance. This is considered in a Level 3C analysis. In
the absence of a demonstration of the general applicability of either of the above
modifications, a detailed, project-specific, justification would be required if either were
to be applied.
6.5.9. Reeled Pipelines

For reeled pipelines the ECA should be performed with the following modifications:-
Report No: 5087471/ 01 / A

Design Guideline
 The initial defect sizes shall account for any ductile tearing during the reeling process;
 The analysis shall account for the impact of the reeling process on the residual stresses
at the weld;
 Post-reeling tensile stress-strain properties shall be employed;
 Post-reeling fracture toughness properties shall be employed.
The assessment shall evaluate both the case when the plane of reeling and plane of
operational loading are aligned and when they are rotated by 90°. The condition when the
original defect was on the neutral axis during reeling shall be evaluated using the virgin
material stress-strain response.
6.5.10. Fatigue Crack Growth Laws

The fatigue crack growth law should be appropriate to the environment, temperature and
loading frequency.
BS 7910 gives fatigue crack growth laws for in-air and marine environments. In order to
allow for potentially high welding residual stresses, the laws corresponding to a stress ratio
greater than or equal to 0.5 should be used.
The in-air growth laws are appropriate when operating in air or other non-aggressive
environments at temperatures up to 100°C.
The „freely corroding in a marine environment‟ and „in a marine environment with cathodic
protection‟ growth laws are appropriate when operating in marine environments at
temperatures up to 20°C. However, they are based on tests conducted at frequencies of
0.17 to 0.5 Hz, and so are not appropriate when assessing fatigue loads associated with
lateral buckling.
Guidance on fatigue crack growth laws for use in conceptual design is given in Table 6.6 and
Figure 6.2 for low carbon manganese line pipe steel. These laws are defined in terms of an
„acceleration factor to be applied to the two-stage curve for steels in air from BS7910 (stress
ratio greater than or equal to 0.5 as above). The factors provide consistency with the fatigue
life reduction factors or „knockdown factors‟ used in S-N fatigue design. It is recommended
that the same factor be used in each case, at least during conceptual design. Although there
are reasons why the two factors need not necessarily be equal, in cases where the fatigue
behaviour of a welded joint is governed by fatigue crack growth (rather than initiation) this
would be expected to be the case.
Some published crack growth rate data suggest that the influence of environment is
lower at low values of ΔK than it is at higher values of ΔK. This is not consistent with
the use of a constant acceleration factor applied to the baseline in-air curve, as
described in Table 6.6. However, recent tests suggest that measured crack growth
rates at low ΔK are dependent on details of the test technique, and some methods
may under-predict the crack growth rate, possibly as the result of crack closure
effects, rates of hydrogen uptake or variations in crack tip chemistry. Recent data in
sour environment, at high stress ratio, where crack closure effects would be
minimised, suggest a curve that is parallel to the baseline in-air curve[49]. Overall, it is
considered that the use of the curves defined by Table 6.6 is sufficient for conceptual
design. Project specific testing is recommended for detailed design.
The tentative recommendations given here are based on fatigue tests undertaken by
SAFEBUCK and the limited number of tests in the published literature[29-49]. They
may be non-conservative. The influence of environmental parameters (e.g. H2S and
CO2 concentrations, pH, salinity, temperature, inhibitor) on corrosion-fatigue
behaviour is not well understood.
Report No: 5087471/ 01 / A

Design Guideline
Sweet corrosion 3
Sour corrosion 10-40
1
Sea-water and cathodic protection 9
Table 6.6 Fatigue Crack Growth Law Acceleration Factors for Carbon Steels
Exposed to Corrosive Environments
1E-02
1E-03
da/dN, mm.cycle-1
1E-04
1E-05
1E-06
in-air
sweet
1E-07 sea-water and cathodic protection
sour
1E-08
10 100 1000 10000
ΔK, N.mm-3/2
Figure 6.2 The Fatigue Crack Growth Laws for Sweet and Sour Environments
There is little test data to support the tentative recommendations for sweet corrosion.
The fatigue crack growth law should be specific to the type and location of the flaw being
assessed. For an embedded flaw, not exposed to internal or external environment, in-air
behaviour can be assumed. For an internal surface breaking flaw, the crack growth law for
either sweet or sour environment (defined in Table 6.6) should be used, depending on the
nature of the product being carried within the pipeline. For an external surface breaking flaw
the crack growth law for seawater and cathodic protection (defined in Table 6.6) should be
used. However, if a demonstrable high integrity field joint coating is used, in-air behaviour
can be assumed for external surface breaking flaws.
Refer to Section 6.4.5 for other materials. In general, the fatigue crack growth rate
acceleration factor should again be consistent with the S-N life reduction factor. Project-
specific testing is recommended.
For clad or lined pipe, no flaws should be tolerated at the weld root or in the zone bounded
by the thickness of the CRA. The ECA should nevertheless consider hypothetical flaws for
comparison with the detection limit of the NDT techniques used. For an internal surface
breaking flaw, the end-of-life condition should be growth of the flaw through the CRA layer.
An embedded defect located within the carbon steel material, at the interface with the CRA
layer, should also be assessed.
Report No: 5087471/ 01 / A

Design Guideline
6.5.11. Installation
An installation ECA may be conducted by the installation contractor. The installation ECA is
a sub-set of the design ECA, and the results will need to be considered in the design ECA.
The extent to which the loads during installation have implications for operation, and vice
versa, will vary from case to case. The largest static and cyclic loads during installation will
occur in a plane transverse to those during operation if the pipeline does not rotate and the
lateral buckle is confined to the horizontal plane. In this case, the interaction between the
two will be small (there is likely to be a small component of fatigue loading in the horizontal
plane during installation, and similarly in the vertical plane during operation). If the pipeline
does rotate or the lateral buckle is not confined to the horizontal plane then the interaction
between the two will be more significant.
The variation of the maximum stress and stress range around the circumference during
installation and operation is informative.
The allowance for fatigue loading during installation in the ECA calculations should be
consistent with that in the S-N calculations.
6.6. PLASTICITY LIMITATIONS

The imposed strain should not approach the uniform strain capacity of the material. The
maximum equivalent strain developed in the buckle should not exceed the following limitation:-
 eq  0.25  (0.99  YT ) .... 6.24
Where YT is the maximum specified yield to tensile ratio of the pipeline steel.
The limit ensures that excessive strains are not admitted in the design process. The
limit is based upon the test data base gathered as part of the SAFEBUCK JIP and
public sources[54, 55,56], Figure 6.3.
18 SAFEBUCK Denys
X80 Lillig et al
16 Stew art et al Design
14
Uniform Strain (%)
12
10
8
6
4
2
0
0.75 0.8 0.85 0.9 0.95 1
Y/T ratio
Figure 6.3 Uniform Strain Capacity Data

The recommendation is based upon ambient temperature data. The applicability of
the limit at the maximum operating temperature should be confirmed during detailed
design through mechanical testing.
Report No: 5087471/ 01 / A

Design Guideline
Equation 6.24 may be unduly conservative for CRA materials. If sufficient test data is
available, the limit can be defined as 1/3 of the uniform strain capacity of the material.
The equivalent strain can be calculated from the von Mises equivalent stress and
equivalent plastic strain as:-
 eq
 eq  peq  .... 6.25
E
The equivalent plastic strain is defined by:-
p
 eq 
2
3
     
 2 2 2
   Lp   Hp   Rp 
 
.... 6.26
The equivalent plastic strain is usually a standard output from a FE analysis.

The maximum axial stress range must comply with the following condition:-
2
R 3  h 
 2  fB  1     .... 6.27
SMYS 4  SMYS 
This equation is intended to prevent cyclic plasticity. Although somewhat different in

form it performs the same function as a hoop strain ratcheting check[57]. No further
assessment of ratcheting is required.
Equation 6.27 can be employed for internal or external overpressure. The hoop
stress used in the check should be the maximum absolute value of the hoop stress
that could occur during operation. The SMYS at ambient temperature should be
employed.
The Bauschinger factor, fB, defines the magnitude of the Bauschinger effect and is the ratio
of the cyclic yield stress to the monotonic yield stress. Recommended values of fB are
presented in Table 6.7.
Pipe Type fB
Seamless carbon steel 0.8
UOE carbon steel 0.7
CRA 0.7
Table 6.7 Default values of the Bauschinger Factor

These values are tentative, based on relatively sparse data. If the limit state is
critical, the applicability of the values of fB should be confirmed through testing in
advance of detailed design.
6.7. PROJECT SPECIFIC TESTING

In the absence of relevant guidance in codes and standards, or information in the published
literature, or from previous projects, it may be necessary to conduct project-specific testing.
Testing requirements depend upon the specific project parameters and will vary significantly
from no additional requirements to detailed investigations into a number of areas. Sensitivity
calculations are informative for determining the significance of a parameter, and hence
whether testing is required. The two areas where testing is most likely are:
Report No: 5087471/ 01 / A

Design Guideline
 Severity of limit states;

 Corrosive environments;
 Pipe-soil Interaction.
6.7.1. Severity of Limit States

The design limits outlined in Section 6.1 to 6.6 are intended to be applicable to any general
design situation. However, under severe loading conditions it may be impossible to comply
with these limits. In this case, it may be possible that a project specific limit state
investigation will demonstrate that an increase in load severity is acceptable.
This approach can also be adopted for situations that are out with the current
guideline.
The investigation will involve a significant full-scale test programme. The test programme
should be supported by a detailed numerical assessment of the structural response. The
results of the test programme should be incorporated into the design process with an
adequate safety factor.
Structural reliability methods should be employed to demonstrate that the pipeline
complies with conventional safety levels.
Since the test programme should be performed using production pipe and production welds,
this approach can only be adopted late in the design cycle and there remains a risk that the
test results may be negative.
This risk could be mitigated by performing exploratory tests early in the design
process (using generic material) and undertaking detailed numerical analysis.
The test programme will depend upon the specific limit state which governs the particular
problem, but may involve:-
 Local buckling tests, with appropriate pressure conditions;
 Low cycle fatigue testing in the appropriate environment;
 Material characterisation testing (cyclic and uni-axial mechanical testing in the
appropriate direction and at the appropriate temperature);
 Fracture toughness testing.
6.7.2. Fatigue Testing in Corrosive Environments

S-N curves and fatigue crack growth laws are specific to the material, environment and
temperature. It is important that any endurance and fatigue crack growth rate tests are
conducted in conditions (e.g. material, geometry, load, frequency, waveform, temperature
and environment) that are representative of the actual conditions.
The internal environment in a pipeline typically varies from project to project. The influence
of environmental parameters (e.g. H2S and CO2 concentrations, pH, salinity, temperature,
inhibitor) on corrosion-fatigue behaviour is not well understood. Consequently, it is difficult
to be specific about testing requirements for any project because much depends on the
corrosive environment to which the pipeline is exposed.
Project-specific testing is likely to be limited to sweet and sour internal environments,
because a high-integrity field joint coating should prevent any interaction between the
external environment and the girth weld.
Lateral buckling imposes a very low frequency loading (less than 10-5 Hz). Corrosion-fatigue
is sensitive to the loading frequency. In this context, per cycle, low frequency loading is
Report No: 5087471/ 01 / A

Design Guideline
more severe than high frequency loading. The S-N curves and fatigue crack growth laws
quoted in codes and standards are based on tests conducted at high frequencies (typically
0.1 Hz and above). The focus of a project-specific test programme is therefore most likely to
be the effect of frequency on the fatigue behaviour.
A significant number of tests are required to produce a new S-N curve or a new fatigue crack
growth law. The extent of the required project-specific testing will depend upon the similarity
of the environment to that in previous projects and the sensitivity of the design to fatigue. It
may be sufficient to conduct a small number of tests to verify that the assumed S-N curve
and fatigue crack growth law is conservative.
A project-specific test programme should consist of tests on project-specific materials and
weld procedures qualified for the project in a simulated project environment.
Fatigue testing at low frequencies is time consuming. The implications of the time taken to
conduct tests on the project schedule should be considered.
SAFEBUCK has conducted low frequency fatigue tests of pipeline welds in the
following environments: sea-water with cathodic protection, sweet environment and
and sour environment[29,30]. A number of recent projects have conducted similar tests
in sour environments, some of which have been published.
6.7.2.1. Fatigue Crack Growth

Fatigue crack growth rate tests are required to determine an appropriate fatigue crack
growth law for the environment. The fatigue crack growth law is used in the ECA.
A typical project-specific test programme would include:
 fatigue crack growth rate tests in air (to establish a baseline);
 frequency scanning crack growth tests in the simulated environment; and
 fatigue crack growth rate tests in the simulated environment and at an appropriate test
frequency.
Frequency scanning tests are fatigue tests conducted at a constant stress intensity factor
range (K), but at various cycling frequencies, with an emphasis on the low frequency
loading experienced in lateral buckling. The aim of frequency scanning crack growth tests is
to establish a plateau frequency below which the incremental fatigue damage per cycle
remains relatively constant. It is not practical to test at the frequency of cycling experienced
by a typical lateral buckle, so this approach helps to define suitable frequencies for
endurance and crack growth rate tests.
It is noted however that the frequency at which crack growth rates saturate, depends
on the applied value of ΔK. This complicates matters and the combined effects of
frequency and stress range are not well understood. Some care is therefore needed
when designing experimental test programmes.
Fatigue crack growth rate tests are typically conducted under conditions of increasing K.
Alternative testing methods (e.g. constant Kmax) may be required if the behaviour at low K is
to be investigated, to avoid problems with crack closure.
It may be more efficient to generate a set of test data at moderate frequency
(typically 0.2 Hz) and then conduct frequency scanning tests to establish the
influence of frequency. This is because the increasing K tests can take a very long
time at the appropriate test frequency, if this is low. If this approach is adopted,
frequency scanning tests should be undertaken at K values corresponding to
beginning and end of life.
Report No: 5087471/ 01 / A

Design Guideline
Due to the uncertainties referred to above, a limited number of endurance tests at

low frequency would provide additional confidence.
6.7.2.2. Endurance Testing

S-N curves are derived from endurance testing.
A typical project-specific test programme would include:
 endurance tests in air (to establish a baseline);
 endurance tests in the simulated environment and at an appropriate test frequency.
SAFEBUCK conducted a limited number of very low frequency endurance tests in
sea-water and sour environments to demonstrate that the effect of frequency
observed in frequency scanning tests conducted at high K was similar to that
observed in endurance tests. Therefore, it may be possible to conduct endurance
tests at frequencies above the frequency plateau (thereby reducing the testing time)
and then adjust the results to account for the effect of frequency.
It is also noted that there are different ways of determining a knockdown factor from a
given set of test data. Tests conducted within SAFEBUCK suggest that this is best
derived from an absolute baseline (e.g. a design S-N curve) rather than the
corresponding in-air data (although the latter are still needed to demonstrate
adequate performance under those conditions). This approach is recommended as it
has been observed that different welds, giving different fatigue lives in air, performed
very similarly in sour environment [30].
6.7.3. Fracture Toughness Testing: Sour Service

Project-specific fracture toughness tests are likely to be required if the internal environment
is sour. The influence of environmental parameters is not well understood (as with fatigue,
see above) so whilst tests from previous projects will be informative they may not be
sufficient. The results of these tests will be used in the ECA.
In sour conditions, active corrosion results in the adsorption and diffusion of atomic hydrogen
into the steel which in turn leads to a reduction in the facture toughness and/or
environmental cracking.
Depending on the environment and the operating conditions (specifically temperature), the
material may be susceptible to hydrogen embrittlement and/or stress corrosion cracking.
Two types of tests are typically required: tests to determine the threshold stress intensity
factor for susceptibility to stress corrosion cracking (KISCC), and tests to determine the degree
of hydrogen embrittlement (KIH). It is preferable to conduct these tests over a range of
temperatures, because it is often not obvious which temperature will represent the limiting
case (in addition to the toughness varying with temperature, the applied loads also vary).
Standard in-air tests should also be conducted to provide a reference toughness.
When conducting tests to measure KIH, it is conservative to test specimens that are fully
charged with hydrogen by immersion in representative sour environment, followed by testing
in air at a displacement rate which is low enough to allow hydrogen diffusion to the crack tip,
but fast enough to avoid significant hydrogen losses from the sample. Measurements of
hydrogen concentration, before and after testing, are recommended to ensure that test
conditions are appropriate. Analysis of the thermal transients in the pipe wall during a shut-
down, and the rates of hydrogen diffusion into and through the steel, may show that the use
of fully charged specimens to measure KIH is over conservative.
Report No: 5087471/ 01 / A

Design Guideline
It is noted that the presence of corrosion inhibitor may have a significant influence on
the rate of hydrogen uptake and therefore the resulting fracture toughness, but
experimental data quantifying the effect is limited.
KISCC is the appropriate measure of toughness when the crack tip is exposed to the sour
environment, e.g. surface flaws at the weld root.
KIH is the appropriate measure of toughness when the crack tip is not directly exposed to
sour environment, e.g. embedded flaws and surface flaws at the weld cap, and when the
sour environment has been displaced, as occurs during cold displacement.
6.7.4. Fracture Toughness and Mechanical Testing: Reeled Pipelines

Tensile and fracture toughness testing should be conducted to characterise the material in
both the unstrained, and strained and aged conditions. The level of pre-strain in testing
should be representative of that introduced during the reeling cycle. Due consideration
should be given to the implications of the different strain histories around the circumference
of the pipe.
Fracture toughness testing to support the reeling installation ECA shall comply with the
requirements of DNV RP F108[53].
6.7.5. Pipe-Soil Interaction Testing

Lateral buckling and pipeline walking behaviour is extremely sensitive to pipe-soil interaction
and there is much uncertainty associated with models for pipe-soil interaction, particularly in
soft clay. In addition, the basic phenomena involved are not fully understood, and detailed
numerical modelling is not yet capable of fully representing the response. This has led to the
need to supplement the conventional geotechnical site investigation with project-specific
model test programmes in which pipe-soil interaction is simulated. Appropriate project
specific pipe-soil testing methods are discussed in Appendix B.
Report No: 5087471/ 01 / A

Design Guideline
7. PROCUREMENT, INSTALLATION AND OPERATION

The design philosophy does place additional burdens on procurement, installation and
operational phases of the pipeline life cycle. The key areas where additional care is required
are discussed briefly here.
7.1. PROCUREMENT
As a result of the severe design conditions it may be necessary to tighten up the basic
linepipe specification. Areas where enhanced project specific requirements should be
considered include:-
 Dimensional tolerances (wall thickness, diameter and out-of-roundness);
 Range of acceptable yield strength;
 Reduced maximum value of YT ratio;
 More frequent mechanical testing (e.g. longitudinal tensile, hardness and toughness);
 Elevated temperature testing;
 Increased NDT coverage (wall thickness, weld flaws);
 More detailed and frequent dimensional measurements of pipe ends.
The feasibility of the requirements should be confirmed with the pipe mill as early in the
project as feasible.
It is important to ensure that the strength of the weld is greater than the strength of the pipe
material (weld overmatching) over the strain range of interest. Conventionally, procedures
aim to achieve this only for the ultimate tensile strength of welds. However, for HPHT
applications, the weld overmatching should be achieved over the strain range expected and
with respect to the pipe upper bound yield strength. The overmatching should be achieved
at ambient temperature and at the operating temperature.
Where reasonably achievable, an upper limit of SMYS + 100 MPa should be placed
on the yield stress of the linepipe.
7.1.1. Fatigue Sensitive Pipelines

SCFs at girth welds result from mismatch at the pipe ends due to manufacturing tolerances
and fabrication tolerances. The SCF is often a critical aspect of design for flowlines with
lateral buckles, where fatigue is the dominating limit state. This section outlines important
aspects for control of SCF during procurement, when fatigue is critical to design.
If there is certainty about where the pipeline will buckle, then it is possible to
designate such areas as „fatigue sensitive zones‟ and apply enhanced inspection
methods in those regions only.
7.1.1.1. Pipe Manufacturing Tolerances

The manufacturing tolerances are on pipe end diameter and wall thickness.
For flowlines designed to DNV-OS-F101 it is recommended that enhanced dimensional
requirements (Supplementary D) for linepipe are specified in accordance with Table 7.26 of
the code.
Report No: 5087471/ 01 / A

Design Guideline
In fatigue-dominated designs, minimising the SCF at a buckle crown can significantly

reduce the fatigue damage. It is possible to manufacture pipe with enhanced
tolerances, or select line-pipe from production with more closely matched
dimensions, for fatigue sensitive areas (planned buckles) and use standard line-pipe
for the remainder. Where individual pipe joints have been measured, pipe sorting
can be adopted to match pipe joints in fatigue sensitive areas (buckle regions) or
match similar joint ends for the whole pipeline. The requirements for pipeline sorting
in fatigue sensitive areas will be borne out of the lateral buckling design. Allowances
for pipe sorting should be made early in the project, prior to contract award.
Alternatively, banding (marking) line-pipe from production with the most closely
matched dimensions, can allow the whole line to be end-matched without significant
logistics issues. Banding pipes into groups with similar end sizes has been used
successfully. End matching of individual pipes for a whole pipeline is difficult
logistically and may be too complex to apply.
API 5L[58] line-pipe specification does not specify enhanced dimensional tolerances,
resulting in slightly less control on dimensions. Pipe mills can typically achieve
tighter tolerances than those quoted in API 5L, with no significant cost penalty. Pipe
tolerances should be considered early in the project, prior to linepipe contract award.
In extreme cases, it is possible to machine pipe ends to improve fit up tolerances.
This raises two key concerns:
1. The wall thickness must be sufficient to meet design requirements following
machining;
2. Machining is likely to increase the localised plastic strain, which could exceed
design limits.
7.1.1.2. Pipe Fabrication Tolerances

Fabrication tolerances are defined by fit-up mismatches between pipe joints. Pipe joint fit-up
is performed using pipe clamps; clamps can be either internal or external. For flowlines
where fatigue at the weld root is dominating then good control of the root hi-lo is required to
minimise the SCF. In fatigue sensitive areas the maximum hi-lo at the root should be as low
as reasonably practical, to achieve the required fatigue performance.
In designs with high fatigue loading the specified maximum root hi-lo is typically
1.0mm but can be as low as 0.5 mm; this is typical of SCR riser design. The
mismatch and hi-lo at individual welds is typically minimised by rotation of pipe joints
to achieve best fit-up.
Excessive hi-lo can lead to deterioration in general fatigue performance (due to poor
weld root profile, for example) that may necessitate a more onerous fatigue curve
than „E‟ to be applied in design.
An allowable or agreed maximum hi-lo should be specified, if the fatigue limit state is likely to
be marginal or critical to the design solution.
7.1.1.3. Data
Project-specific dimensional data should be collected following manufacture, including
diameter and wall thickness around the circumference of each end of every pipe joint, or at a
reduced sampling rate that provides sufficient statistical data. The data shall be used to
assess the range of SCF likely to be achieved, incorporating the proposed line up and pipe
rotation practices of the contractor.
Report No: 5087471/ 01 / A

Design Guideline
Detailed pipe data can be used in statistical analysis to determine a realistic SCF.
Detailed pipe measurements are routinely made at the pipe mill; however, they are
often not made available as a database to the client, hence the need to include this
requirement as part of the procurement contract. Providing there is sufficient
measurement data an SCF can be defined with respect to a probability of
exceedance. This approach has been successfully adopted on some recent projects.
7.2. INSTALLATION
There are a number of areas where the severe design conditions will place additional
burdens on the pipeline installation, these include:-
 Tolerable weld flaws may be smaller than usual and AUT may be required to size flaws;
 Increased testing requirements to confirm weld overmatching;
 Additional restrictions on misalignment at girth welds may be specified (confirmatory
measurements may be required);
 More accurate as-laid survey may be required (Section 7.3.1.1);
 The buckle initiation strategy may complicate the lay (for example a more complex route
or more complex handling because of distributed buoyancy).
The installation contractor should be thoroughly briefed on the importance of the installation
to the design strategy. The feasibility of the requirements should be confirmed with the
installation contractor as early in the project as possible.
7.3. OPERATIONAL INTEGRITY

Prior to start-up there will be an inherent uncertainty over the buckling and walking response
of the system. It is therefore critical to the long-term integrity of the pipeline that operational
monitoring is undertaken. Operational surveys should be used to:-
 Confirm the buckling strategy has been reliable and that all buckles have formed as
planned;
 Identify the number and location of any unplanned (rogue) buckles;
 Determine the buckle shapes and curvatures;
 Monitor axial displacement at end and in-line structures;
 Monitor pipeline/SCR holdback anchors to ensure that pipe-walking has been arrested;
 Monitor route-curves to ensure that no lateral displacement or pull-out has occurred
To comply with the intent of this guideline, the Pipeline Integrity Management System (PIMS)
developed at the project stage, should include requirements for monitoring the formation of
lateral buckles shortly after start-up and subsequent behaviour in operation, together with
monitoring of associated load history, defined by pipeline operating pressures and
temperatures in operation. This operational data should be used to re-qualify the pipeline
integrity throughout its life.
If sleepers, or other fixed mitigation measures are employed then these should include clear,
easy to read, position markings along their length for monitoring buckle amplitude.
Measurement number markings should be on a vertical face to avoid obstruction by soil or
debris deposition.
End terminations or in-line connection structures should be similarly marked to record initial
end expansion and potential pipeline walking. A visual survey of these markings should be
carried out after installation and before flooding (unless the pipeline is laid flooded) to
establish the datum for all future measurements of axial displacement.
Report No: 5087471/ 01 / A

Design Guideline
Consideration should be given to continuous monitoring systems that record bending

strain in a buckle or axial displacement at an end/in-line connections. Such systems
are incredibly valuable in comparison with snap-shot surveys but the technology is
not yet fully proven.
Pipeline surveys generally fall into two broad categories, positional surveys and visual
inspections. Positional surveys, such as high accuracy OOS or Side Scan Sonar (SSS) are
required to assess pipeline buckling and curve stability whilst visual inspections are required
to assess pipeline walking and axial displacements.
7.3.1. Surveys of Pipeline OOS and Buckling Behaviour

Surveys of pipelines susceptible to lateral buckling should be carried out by appropriately
skilled personnel, with sufficient knowledge of lateral buckling design and operational
integrity monitoring to ensure that data obtained by the survey are fit for purpose.
7.3.1.1. As-laid Survey

If buckle formation reliability is a potential concern for system integrity, or if a novel design
solution with inherent uncertainty is adopted, then a high accuracy positional survey of the
as-laid pipeline should be carried out after installation and prior to flooding or hydrotest.
In many cases a high accuracy post lay survey is not required. This need only be
performed if there are specific concerns over the robustness of the buckle initiation
strategy.
Flooding applies pressure to the inside of the pipe altering the as-laid configuration
If required, this survey should establish pipe geometry along the route to check for any
excessive out-of straightness features, as this may cause buckle initiation away from the
intended sites. This survey will also provide a base-line against which unexpected
operational performance can be checked.
The survey should record the position of the pipeline as accurately as reasonably practical;
an ROV survey (see below) or a geometry pig should be able to provide data to the required
level of accuracy.
Traditionally ROV surveys give fairly poor pipeline X-Y positional accuracy but can give good
Z (depth) accuracy. The survey set-up should be designed to give the best possible
absolute and relative accuracy in X, Y and Z. By improving the absolute accuracy, it can be
inferred that the relative accuracy will also be improved. However, the absolute (global)
accuracy is significantly less important than the (local) relative accuracy for this type of
survey. The objective of these surveys is to measure local out-of-straightness over
distances ranging from 10 to 500 m with good accuracy. A range of measures can be
implemented to achieve this, for example:
 ROV positioned directly on top of pipe using an undercarriage system;
 High accuracy GPS surface positioning, giving accuracy better than 20 cm in the X-Y
plane. Vessel heading to be measured to better than 0.1 deg;
 Ultra short base line (USBL) hydro acoustic positioning system to give sub-surface
positioning better than 0.5% water depth;
 ROV mounted Doppler Velocity Log (DVL);
 Inertial Navigation System (INS);
 Gyroscopic direction.
Report No: 5087471/ 01 / A

Design Guideline
In deeper water, measurement of ROV position from surface support vessel is increasingly
inaccurate, so it is essential to use an overarching navigational system that integrates data
from USBL, DVL and INS.
The positional data should be recorded and reported at least one point per metre along the
pipeline. The survey contractor should process the data but no smoothing should be
undertaken.
The vertical position of the pipe should be measured using an ROV mounted digi-quartz
bathymetry unit with measurements reported at 1 m intervals or better.
Extensive calibrations should be undertaken prior to commencing the survey. Calibration of
accuracy and repeatability of local out-of-straightness measurements should be carried out
by surveying the same section of line, a number of times in different directions, preferably
along a section of line with a known OOS feature such as a buckle or a route-curve. This
should be repeated until the system algorithms and position calibration are acceptably
accurate.
Global positioning, while less critical, can be calibrated by holding the ROV at a fixed
location, ideally at known seabed datum point, e.g. a manifold. Over 10-minutes numerous
fixes (e.g. 100) should be taken. The mean position and standard deviation of global
positioning error can then be assessed.
Levels of pipeline embedment should also be assessed after installation or after flooding.
The best quality embedment and seabed data around the pipe comes from Digital Terrain
Mapping (DTM), utilising a dual head Multi-beam echo sounder (MBES) system. Continuous
five-point files (giving depth measurements at top of pipe and two locations – close and far –
to each side of the pipe) should be provided for assessment of embedment and local berm
features along the length of the pipeline. Cross profiles should be provided at 5 m intervals
or better over the region of each lateral buckle.
7.3.1.2. Post Start-up or Hydrotest Survey

A full-length pipeline system position survey should be carried out after steady-state
operation is achieved. Particular attention should be paid to the survey accuracy at the
buckle locations (including planned and rogue buckles). The survey should record as
accurately as reasonably practical the position of the pipeline, the level of embedment and
the size and extent of soil berms to each side of the pipe along the full length of each lateral
buckle.
A full-length survey, to establish the location of each buckle, is vital to understanding
system performance. Measurement of one buckle in isolation does not provide
sufficient information on virtual anchor locations, feed-in to the buckle, or the loads
within the buckle.
The survey should also pay particular attention to the expansion at the pipeline ends.
A high accuracy ROV survey or high definition SSS by AUV should be able to provide data
to the required level of accuracy. The accuracy of the post start-up survey should be similar
to that of the as-laid survey.
The operational data (pressure and temperature) at the time of the survey should be
recorded.
Report No: 5087471/ 01 / A

Design Guideline
7.3.1.3. Operational Surveys

The amplitude and shape of the buckles should be monitored in operation to ensure that the
range of bending in the buckle is not excessive and does not exceed worst-case design
assessments. Ideally the survey should cover the full pipeline length, however, once the
buckles are fully established it is acceptable to limit the extent of the survey to the vicinity of
each buckle and the pipeline ends. The accuracy of the operational surveys should be
similar to that of the as-laid survey.
The operational data (pressure and temperature) at the time of the survey should be
recorded.
7.3.2. Operating Load Condition Monitoring

Temperature and pressure monitoring at the inlet and outlet is required to establish local
operating conditions along the pipeline with sufficient accuracy to evaluate loading in
operation. To ensure significant fluctuations are registered, data should be recorded at
approximately one minute intervals. Lower or higher recording frequencies may be
proposed, depending on the anticipated flow dynamics. Data should be stored in a readily
accessible electronic format for the life of the line.
Surge pressure fluctuations in liquid filled systems have very short durations, typically
a few seconds or less; slug flow incidents have longer durations of tens of seconds or
minutes; while general changes in operating conditions can take place over many
minutes or hours. It is important to record pressure and temperature at sufficient
resolution to capture any load such conditions that will influence system life and long-
term integrity.
Details of measurement frequency, recording methods and the selection of instrument
locations should be discussed with operations personnel to ensure that all data relevant to
pipeline operating conditions are recorded.
For example, topside temperature fluctuations will occur much more frequently than
significant temperature variation in a highly insulated pipeline. Knowledge of the
relationship between instrumentation type and location and pipeline response is vital
in interpreting the load history.
7.3.3. Surveys of Pipeline Walking

If the pipeline walking is identified as an issue (See Section 3.5), and particularly if delayed
remedial measures have been adopted, then the pipeline walking behaviour must be
monitored during operation. Clear easy to read sliding markers should be provided at each
end of the pipeline and at in-line connections for monitoring axial displacement. This is most
practical where a pipeline end termination structure is located at each pipe end.
Measurement markings should be on a vertical face to avoid obstruction by soil or debris
deposition.
7.3.3.1. End Terminations, Spoolpieces or Jumpers

To establish the overall walking behaviour of the pipeline, the following survey activities
should be undertaken:-
 An accurate baseline definition of the markers on termination or in-line structures and
spoolpiece geometry should be established during the as-laid survey, prior to flooding;
 An accurate baseline definition of the markers on termination or in-line structures and
spoolpiece geometry, under steady-state operating conditions, should be established
during the post start-up survey;
Report No: 5087471/ 01 / A

Design Guideline
 The movement of the markers on termination or in-line structures or spoolpieces

geometry should be monitored on a regular basis. The frequency must be based on the
number of shutdowns anticipated and must be chosen to ensure that unacceptable
walking displacements do not occur between surveys;
 Continuous monitoring using a taut wire recorder or similar should be considered.
7.3.3.2. Buckles
The walking behaviour at the buckles should be monitored in the same way as for the
spoolpieces. It may be more difficult to identify axial displacement at the buckles, and
consideration should be given to installing datum marks along the pipe (field-joints are
spaced at least 12 m apart, which is too far for effective measurement), which will facilitate
monitoring. This is more practical at buckle sites employing a fixed structure such as a
sleeper.
7.3.4. Pipeline Integrity Assessment

The operations group should be made aware of the severe loading associated with the
design strategy. Key design parameters should be clearly outlined (for example maximum
allowable temperature and pressure profiles) so that operational monitoring is against a
known limit.
All data from surveys and operating load condition monitoring should be recorded and
readily accessible. The data should be assessed at regular intervals to ensure the pipeline
is operated in line with design intent, to confirm pipeline integrity, and to plan the scope and
timing of future surveys, or mitigation measures in the unlikely event that loading exceeds
design expectations.
7.3.4.1. Post Start-up

The survey data must be reviewed to confirm that buckle initiation has occurred as planned.
In the event of unplanned buckling, an engineering assessment should be carried out to
confirm the acceptability of the pipeline configuration. The post-lay survey should be used to
understand the reason for the unplanned behaviour.
Even if the buckles have occurred as intended it may be prudent to re-assess the pipeline
integrity in the light of the actual shape and amplitude of the buckles, which may differ from
that assumed in design.
7.3.4.2. Operation
The operational survey data must be reviewed to confirm the stability of the buckles. This
must quantify the size and shape of each buckle. In the event of behaviour outside design
predictions, an engineering assessment should be carried out to confirm the acceptability of
the pipeline configuration.
Operational pressure and temperature cycles should be monitored in operation to record
maximum operating load conditions and to ensure that cyclic loading is within the ranges
used in fatigue design. If the number of cycles is considered likely to exceed design
assumptions, a revised fatigue assessment should be considered based on measurement of
buckle shape and amplitude.
Report No: 5087471/ 01 / A

Design Guideline
If walking is detected, the rate of walking should be established based upon the magnitude
of the overall walk and the number of shutdown cycles that have occurred. The long-term
integrity of the spoolpieces and buckles must be assessed using the observed walking rate.
If integrity cannot be guaranteed, then remedial measures, such as anchoring, must be
implemented.
7.3.5. Survey Frequency

Following a post-start-up operational survey, further operational pipeline integrity surveys
should be performed, with the frequency reducing as confidence grows in system
performance. Monitoring allows comparison between early life behaviour and predictions, to
determine whether there is likely to be a problem in the longer term, which may be prevented
with remedial action. The proposed survey frequency should be based on systematic
observation of in-place performance.
Report No: 5087471/ 01 / A

Design Guideline
8. REFERENCES
1. Offshore Standard DNV-OS-F101. Submarine Pipeline Systems 2010. Det Norske

Veritas. October 2010.
2. API Recommended Practice 1111, Recommended Practice for the Design, Construction,
Operation, and Maintenance of Offshore Hydrocarbon Pipelines (Limited State Design),
Fourth Edition, December 2009. Includes Errata (May 2011).
3. Recommended Practice DNV-RP-F105. Free Spanning Pipelines. 2006. Det Norske
Veritas.
4. Hobbs, R.E, In-service Buckling of Heated Pipelines. ASCE Journal of Transportation
Engineering, Vol. 110 No. 2, March 1984.
5. Hobbs, R.E. Liang, F. Thermal Buckling of Pipelines Close to Restraints. OMAE 1989.
6. Tornes, K., Jury, J., Ose, B., Thompson. Axial Creeping of High Temperature Flowlines
Caused By Soil Ratcheting. OMAE 2000.
7. Carr M., Bruton, D. and Leslie, D. Lateral Buckling and Pipeline Walking, a Challenge
for Hot Pipelines. Offshore Pipeline Technology Conference 2003, Amsterdam.
8. Bruton, D.A.S.; Sinclair, F.; Carr, M. Lessons Learned From Observing Walking of
Pipelines with Lateral Buckles, Including New Driving Mechanisms and Updated Analysis
Models. OTC 20750 (2010).
9. Carr M., SAFEBUCK Phase II. Lateral Buckling SRA. Atkins Boreas Report JIP-04/
BR06066/ B. 10 December 2007.
10. Carr M., SAFEBUCK Phase III. Task 1 - Reliability of Design Process. JIP Report, May
2011.
11. SAFEBUCK II. BUCKFAST Manual. TBC
12. Peek, R., and Yun, H. Flotation to Trigger Lateral Buckles in Pipelines on a Flat Seabed.
Journal of Engineering Mechanics. ASCE. April 2007.
13. Peek, R. and Yun, H. Scaling of solutions for the lateral buckling of elastic-plastic
pipelines. OMAE2004-51054.
14. Ness, O.B and Verley, R., (1995), "Strain Concentrations in Pipelines with Concrete
Coating: An Analytical Model," Proc. OMAE'95, Copenhagen, Denmark.
15. Verley, R., and Ness, O.B., (1995), “Strain Concentrations in Pipelines with Concrete
Coating: Full Scale Bending Tests and Analytical Calculations,” Proc. OMAE 1995, Vol.
5, Pipeline Technology, ASME.
16. Savik, S., Storheim, M., Levold, E. Efficient Finite element Evaluation of Strain
Concentrations in Concrete Coated Pipelines. OMAE2008-57373. Proceedings of 27th
Int. Conference on Offshore Mechanics and Arctic Engineering (2008).
17. Carr, M., Sinclair, F., Bruton, D. Pipeline Walking – Understanding the Field Layout
Challenges, and Analytical Solutions developed for the SAFEBUCK JIP. OTC 2006,
paper no. 17945.
18. Anderson, M., Bruton, D., Carr, M. The Influence of Pipeline Insulation on Installation
Temperature, Effective Force and Pipeline Buckling. 26th International Conference on
Offshore Mechanics and Arctic Engineering. 2007.
19. Carr, M. SAFEBUCK JIP. State of the Art Report. Atkins Boreas Report
BR02042/SAFEBUCK/B. Issued July 2003.
20. Published Document PD 5500: 2006. Specification for unfired fusion welded pressure
vessels. British Standards Institution, Third Edition, January 2006.
21. ASME B31.3 –1996 edition. Process Piping. ASME Code for Pressure Piping, B31 An
American National Standard.
Report No: 5087471/ 01 / A

Design Guideline
22. Olsson, A. Plastic Behaviour of Stainless Steel: A Phenomenological Study. Licentiate

Thesis. Luleå, 1998.
23. F. Aguirre, S. Kyriakides and H. D. Yun. Bending of steel tubes with Lüders bands.
International Journal of Plasticity, Volume 20, Issue 7, July 2004, Pages 1199-1225.
24. Corona, E., Shaw, A., and Ladicola, M. Buckling of Steel Bars with Lϋders Bands.
International journal of Solids and Structures 39 (2202) 3313-3336.
25. Carr, M. SAFEBUCK II. Local buckling of Pressurised Pipes under Bending. Atkins
Boreas Report JIP-04/ BR06067/ B. 5 November 2007.
26. BS 7608: 1993. Code of Practice for Fatigue Design and Assessment of Steel
Structures. British Standards Institute.
27. Recommended Practice DNV-RP-C203. Fatigue Design of Offshore Steel Structures.
2008. Det Norske Veritas.
28. Connelly, L.M. and Zettlemoyer, N. Stress Concentration at Girth Welds of Tubulars with
Axial Misalignment. Proceedings of the 5th International Symposium on Tubular
Structures, UK, 1993.
29. Baxter,B., Muhammed,A. SAFEBUCK II Safe Design of Hot On-Bottom Pipelines with
Lateral Buckling Fatigue Testing and Evaluation. TWI Report to Boreas Consultants
Ltd., TWI Report No.: 16050/1/06, June 2006.
30. Baxter,B. SAFEBUCK II Safe Design of Hot On-Bottom Pipelines with Lateral Buckling
Fatigue Testing and Evaluation – Supplementary Corrosion Fatigue Tests. TWI Report
to Boreas Consultants Ltd., TWI Report No.: 17398/1/07, September 2007.
31. Eadie,R.L., Szklarz,K.E. Fatigue Crack Propagation and Fracture in Sour Dilute Brine.
Paper No. 99611, Proceedings of Corrosion 1999, NACE International, Houston, Texas,
USA, 1999.
32. Szklarz,K.E. Aggressive CO2 Corrosion and Fatigue Behavior of Pipeline Girth Welds.
Paper No. 00012, Proceedings of Corrosion 2000, NACE International, Houston, Texas,
USA, 2000.
33. Buitrago,J., Weir,M.S. Experimental Fatigue Evaluation of Deepwater Risers in Mild
Sour Service. Deep Offshore Technology Conference (DOTC 2002), New Orleans,
November 2002.
34. Buitrago,J., Weir,M.S., and Kan,W.C. Fatigue Design and Performance Verification of
Deepwater Risers. Paper No. OMAE2003-37492, Proceedings of OMAE03 22nd
International Conference on Offshore Mechanics and Arctic Engineering, ASME,
Cancun, Mexico, 8–13 June, 2003.
35. Eadie,R.L., Szklarz,K.E., and Sutherby,R.L. Corrosion Fatigue and Near-Neutral pH
Stress Corrosion Cracking of Pipeline Steel in very Dilute Carbonate-Bicarbonate with
and without the presence of Hydrogen Sulfide using the Compliance Technique. Paper
No. 03527, Proceedings of Corrosion 2003, NACE International, Houston, Texas, USA,
2003.
36. Woollin,P., Pargeter,R.J. and Maddox,S.J. Corrosion Fatigue Performance of Welded
Risers for Deepwater Applications. Paper 04144, Proceedings of Corrosion 2004, NACE
International, New Orleans, USA, 2004.
37. Buitrago,J., Weir,M.S., Kan,W.C., Hudak.S.J. and Mcmaster,F.J. Effect of Loading
Frequency on Fatigue Performance of Risers in Sour Environment. Paper No. OMAE
2004-51641, Proceedings of OMAE04, 23rd International Conference on Offshore
Mechanics and Arctic Engineering, Vancouver, British Columbia, Canada, 20-25 June
2004.
38. Woollin,P., Baxter,D. and Maddox,S.J. Corrosion Fatigue of Welded Stainless Steels for
Deepwater Riser Applications. Paper No. OMAE2005-67498, Proceedings of OMAE
2005: 24th International Conference on Offshore Mechanics and Arctic Engineering,
ASME, Halkidiki, Greece, 12-17 June, 2005.
Report No: 5087471/ 01 / A

Design Guideline
39. Maddox,S.J., Pargeter,R.J. and Woollin,P. Corrosion Fatigue of Welded C-Mn Steel
Risers for Deepwater Applications: A State of the Art Review. Paper No. OMAE2005-
67499, Proceedings of OMAE 2005: 24th International Conference on Offshore
Mechanics and Arctic Engineering, ASME, Halkidiki, Greece, 12-17 June, 2005.
40. Baxter,D.P., Maddox,S.J., and Pargeter,R.J. Corrosion Fatigue Behaviour of Welded
Risers and Pipelines. Paper No. OMAE 2007-29360, Proceedings of OMAE2007, 26th
International Conference on Offshore Mechanics and Arctic Engineering, San Diego,
California, USA, 2007.
41. Pargeter, R.J., Baxter, D and Holmes, B. Corrosion Fatigue of Steel Catenary Risers in
Sweet Production. Paper No. OMAE-57075, Proceedings of OMAE2008, 27th
International Conference on Offshore Mechanics and Arctic Engineering, Estoril,
Portugal, June 2008.
42. Baxter, D. SAFEBUCK JIP – Additional Corrosion Fatigue Data for Sweet Operating
Environment. Atkins Boreas Report BR09076/2, October 2009.
43. McMaster, F. et al. Sour Service Corrosion Fatigue Testing of Flowline Welds. Paper
No. OMAE 2007-29060, Proceedings of OMAE2007, 26th International Conference on
Offshore Mechanics and Arctic Engineering, San Diego, California, USA, 2007.
44. McMaster, F. et al. Sour Service Corrosion Fatigue Testing of Flowline and Riser Welds.
Paper No. OMAE-57059, Proceedings of OMAE2008, 27th International Conference on
Offshore Mechanics and Arctic Engineering, Estoril, Portugal, June 2008.
45. Buitrago, J., Hudak, S. and Baxter, D. High Cycle and Low Cycle Fatigue Resistance of
Girth Welds in Sour Service. Paper No. OMAE-57545, Proceedings of OMAE2008, 27th
International Conference on Offshore Mechanics and Arctic Engineering, Estoril,
Portugal, June 2008.
46. Gui, F. et al. Corrosion Fatigue Performance of Duplex 2507 For Riser Applications,
Paper No. OMAE-20609, Proceedings of OMAE2010, 29th International Conference on
Offshore Mechanics and Arctic Engineering, Shanghai, China, June 2010.
47. DNV JIP Lined and Clad Pipeline Materials Phase 2, Guideline For Design and
Construction of Clad and Lined Pipelines, DNV Technical Report 2007-0220.
48. Holtam, C.M., Baxter, D.P., Ashcroft, I.A. and Thomson, R.C. An Investigation into
Fatigue Crack Growth test Methods in a Sour Environment, International Journal of
Offshore and Polar Engineering, Vol. 20, No. 2, pp. 103-109, June 2010.
49. Holtam, C.M. and Baxter, D.P. Fatigue Crack Growth Performance of Sour Deepwater
Riser Welds in the Near Threshold Regime. Paper No. 21279, Proceedings of OTC
2010, Offshore Technology Conference, Houston, Texas, USA, 2–5 May 2011
50. Guide to Methods for Assessing the Acceptability of Flaws in Metallic Structures, BS
7910 : 2005, British Standards Institution, London, UK, July 2005.
51. API Recommended Practice 579. Fitness-For-Service. American Petroleum Institute,
First Edition, January 2000.
52. R/H/R6. Assessment of the Integrity of Structures Containing Defects. British Energy,
Barnwood, Gloucestershire, Revision 3, 2001.
53. Recommended Practice DNV-RP-F108. Fracture Control for Pipeline Installation
Methods Introducing Cyclic Plastic Strain. Det Norske Veritas, January 2006.
54. Denys, R., Lefevre, A and De Baets, P. Weld and pipe material requirements for a
strain-based pipeline design. The Journal of Pipeline Integrity, Q1 2003.
55. Stewart, G., Klever, F.J. and Ritchie, D. An analytical Model to Predict the Burst
Capacity of Pipelines. 13th International Conference on Offshore Mechanics and Arctic
Engineering. 1994.
56. Lillig, D., Hoyt, D., Hukle, M., Dwyer, J., Horn, A. and Manton, K. Materials and Welding
Engineering for ExxonMobil high Strain Pipelines. Sixteenth International Offshore and
Polar Engineering Conference. 2006.
Report No: 5087471/ 01 / A

Design Guideline
57. F.J.Klever, A.C.Palmer & S.Kyriakides. Limit-State Design of High Temperature

Pipelines. Proceedings, Offshore Mechanics and Arctic Engineering Conference.
Houston 1994.
58. Specification for Line pipe. API Specification 5L. Forty-Fourth Edition, 2008.
Report No: 5087471/ 01 / A

SAFEBUCK III / Safe Design of Pipelines with Lateral Buckling Page A1
Design Guideline: Appendix A
SAFEBUCK III
Design Guideline: APPENDIX A
Design Examples
Report No: 5087471/ 01 / A

TABLE of CONTENTS
A1 Introduction ................................................................................................. 3
A2 Design Example #1: Short Hot Pipeline ..................................................... 4
A2.1 BASIC DESIGN PARAMETERS ................................................................... 4
A2.2 IDENTIFY SUSCEPTIBILITY TO BUCKLING ............................................... 6
A2.3 ASSESSMENT OF UNCONTROLLED BUCKLING ...................................... 7
A2.4 ASSESSMENT OF CONTROLLED BUCKLING ......................................... 11
A2.5 PIPELINE WALKING .................................................................................. 11
A2.6 SUMMARY.................................................................................................. 12
A2.7 UNITY CHECK CALCULATIONS................................................................ 13
A3 Design Example #2: Long Hot Pipeline ................................................... 16
A3.1 BASIC DESIGN PARAMETERS ................................................................. 16
A3.2 IDENTIFY SUSCEPTIBILITY TO BUCKLING ............................................. 18
A3.3 ASSESSMENT OF UNCONTROLLED BUCKLING .................................... 19
A3.6 SUMMARY.................................................................................................. 27
A4 Design Example #3: Modest Temperature Pipeline ................................ 29
A4.1 BASIC DESIGN PARAMETERS ................................................................. 29
A4.2 IDENTIFY SUSCEPTIBILITY TO BUCKLING ............................................. 31
A4.3 ASSESSMENT OF UNCONTROLLED BUCKLING .................................... 31
A4.4 INFLUENCE OF BATHYMETRY................................................................. 37
A4.7 SUMMARY.................................................................................................. 39
A5 REFERENCES ........................................................................................... 40
Report No: 5087471/ 01 / A

A1 INTRODUCTION
To illustrate the application of the SAFEBUCK design guideline example calculations are
performed. Notation used in the design example is as shown in Section 1.8 of the
SAFEBUCK Design Guideline.
Three design examples are presented:-
 Design example #1: short hot pipeline;
 Design example #2: long hot pipeline;
 Design example #3: modest temperature pipeline.
These are chosen to illustrate significantly different pipeline behaviour and how the design
approach addresses these very different problems.
The design approach is outlined in Figure 1.
Select basic design
parameters
Modify design parameters Modify design parameters
No
No Is pipeline
Is walking OK? susceptible to
buckling?
Yes
Yes
Is
Yes uncontrolled
buckling
acceptable?
No
Define initiation strategy Modify initiation strategy
Yes
Is Can
controlled No initiation No
buckling strategy be
acceptable? improved?
Yes
buckle
No
interaction and
walking OK?
Yes
Design Complete
Figure 1 – Basic Design Philosophy

Each step in the design process is applied to the design examples in the following sections.
The examples are only undertaken to a conceptual level design. This is sufficient to
illustrate the design process; the detailed design stage would follow the same process, but
with more use of numerical analysis.
Report No: 5087471/ 01 / A

A2 DESIGN EXAMPLE #1: SHORT HOT PIPELINE

The design example is based on a 2 km 10–inch flowline constructed from X65 grade steel.
The maximum operating temperature is 120°C.
A2.1 BASIC DESIGN PARAMETERS
A2.1.1 Pipeline Parameters

The pipeline parameters used in the analysis are shown in Table 1 and Table 2.
Parameter Value
Length 2000 m
Outside Diameter 273.1 mm
Pipeline Material X65
Wall Thickness 25.4 mm
Corrosion Allowance 3 mm
Pipe Submerged Weight (empty) 0.745 kN/m
Pipe Submerged Weight (flooded) 1.137 kN/m
Pipe Submerged Weight (operating) 1.049 kN/m
Residual Lay Tension 50 kN
Water Depth 1000 m
3
Water Density 1025 kg/m
Table 1 Design Example #1: Basic Design Parameters
Parameter Value
Design Temperature 120°C
Seawater Temperature 4°C
Design Pressure 400 bara
Operating Temperature, inlet 120°C
Operating Temperature, outlet 110°C
Operating Pressure 200 bara
Unload Pressure 10 MPa
Slope of start-up transient 20°C/km
Fatigue Environment 200 full shut downs
Table 2 Design Example #1: Basic Operating Parameters
A2.1.2 Friction
The seabed comprises soft to very soft clay. The lateral pipe-soil frictional resistance
developed for the 10-inch pipeline is illustrated in Figure 2.
Report No: 5087471/ 01 / A

Friction Coefficient 1.5
0.5
0
0 0.5 1 1.5 2 2.5 3 3.5
Dis placement/Diameter
Minimum
Median
Maximum
Figure 2 Design Example #1: Lateral Pipe-Soil Resistance

The resistance is modelled using equivalent friction coefficients, calculated from:-
Resistance Force

Submerged Weight
The range of frictional values to be employed in the analysis is outlined in Table 3.

Direction Minimum Median Maximum
Axial 0.45 0.65 1.0
Lateral Break-out 0.53 0.82 1.1
Lateral Residual 0.39 0.59 0.9
Table 3 Design Example #1: Friction Coefficients
A2.1.3 Check Applicability of Pipeline for Guideline

Before using the SAFEBUCK design guideline a number of checks are first made to ensure
the pipeline considered is within the range of validity. Section 1.6 of the design guideline
outlines the range of applicability.
Checks for the design example as shown in the design guideline are outlined in Table 4.
Report No: 5087471/ 01 / A

Parameter Calculated Value SAFEBUCK Rage of Validity

D/t 10.7 ratio between 10 and 50
Design temperature 120°C < 180°C
Pipe initial ovality less than 1.5% < 1.5%
Pipeline Material Check X65 API carbon manganese steel up
to X65
PIP Systems NA
Fluid Oil Hydrocarbon or water based
products
Installation Plasticity No significant plastic deformation Significant plastic deformation
during installation during installation not considered
Inspection Valid 100% girth weld inspection

Seabed Roughness Relatively flat seabed Relatively flat seabed only
Table 4 Design Example #1: Applicability of Pipeline

The pipeline falls within the range of applicability of the guideline.
A2.2 IDENTIFY SUSCEPTIBILITY TO BUCKLING
The susceptibility of a pipeline to lateral buckling is assessed by working through Section 3.2
of the SAFEBUCK guideline. For the design example, only the steel yield stress has been
temperature de-rated.
The SAFEBUCK design guideline states that the pipeline is not susceptible to buckling if the
maximum force in the pipeline is less than the critical buckling force:-
Nmax  Ncr Equation 3.1
where:-
Nmax  minN0 , Nf max Equation 3.2
Where:-
N0  pi  piL   Ai  1  2    E  A s      L   NL Equation 3.3
L
Nf max  Nend  fmax  Equation 3.4
2
N0 is the fully constrained force. This is calculated below:-
 
N0  40MPa  0.039m2  1  2  0.3  2.05  105 MPa  0.020m2  1.17  105 o C1  116o C  50kN
N0  6.14  MN
In this calculation the internal pressure during installation is assumed to be zero and the
temperature at installation is seabed ambient (4°C). The maximum available frictional force
is given by:-
kN 2000m
N f max  100kN  1.0  1.0149   1115  kN
m 2
The maximum force in the system is the minimum of these two values:-
Report No: 5087471/ 01 / A

Nmax  1115  kN
Since there are no route bends in the pipeline, the critical buckling force is based on the
Hobbs infinite mode:-
EI  HLB
Ncr  0.65  N  2.51  Equation 3.6
D
31.7MN / m2  0.53  1.0149kN / m

Ncr  2.51   627  kN
0.273m
The analysis performed does not account for the influence of hydrodynamic lift, or break-out
axial friction, thus the susceptibility to buckling is slightly underestimated. However, the
maximum force in the pipeline (1064 kN) is greater than the 65% of the critical buckling force
(627 kN), and the system is susceptible to buckling.
A2.3 ASSESSMENT OF UNCONTROLLED BUCKLING

The next step is to evaluate whether the uncontrolled lateral buckling behaviour of a pipeline
is acceptable. The uncontrolled lateral buckling behaviour of a pipeline is acceptable if the
design unity checks for all relevant failure modes are acceptable at the Characteristic VAS.
To evaluate this it is necessary to calculate the Characteristic VAS and undertake a lateral
buckling analysis at the Characteristic VAS.
A2.3.1 Calculation of Characteristic VAS: Deterministic

The Characteristic VAS is calculated in accordance with Section 4.2.1.
A2.3.1.1 Characteristic VAS Limited by Pipeline Length

The pipeline is free ended (subject to modest spool reaction), so the Characteristic VAS
limited by pipeline is equal to half the pipeline length, XL=1,000 m.
A2.3.1.2 Characteristic VAS Limited by Driving Force Envelope

The Characteristic VAS limited by driving force envelope is calculated by solving the Hobbs
equations for an isolated buckle in an infinite pipeline (Equation 4.3). The key inputs to this
equation are:-
 Pipe bending stiffness, EI = 31.7 MNm2
 Pipe axial stiffness, EA=4,091 MN
 Lower bound axial frictional force, fLB = 0.45 * 1049.1 = 472.1 N/m
 Upper bound lateral frictional force, HUBres = 0.9 * 1049.1 = 944.2 N/m
Solving Equation 4.3 for these inputs yields the Hobbs buckle length, LBH=95.1 m and the
Hobbs buckle force, NBH=119.4 kN.
The Characteristic VAS limited by driving force envelope is then calculated from:-
 N  NBH 
X H  2.588  L BH  2   0  Equation 4.2
 fLB 
So
 6140  119.4 
X H  2.588  95.1  2     4,185  m
 0.4721 
Report No: 5087471/ 01 / A

A2.3.1.3 Characteristic VAS Limited by Sharing

 N̂  N 
X S   CR BH   2.588  L
BH Equation 4.5
 fLB

 
N̂cr  Max  4.4  ,7.1   Equation 4.6
 D D 
 
The break-out lateral resistance is:-

 Upper bound break-out resistance, HUB=1.1 * 1049.1 = 1.154 kN/m
 Best estimate bound break-our resistance, HBE=0.82 * 1049.1 = 0.86 kN/m
So the upper bound critical buckling force is:-
 31700  1.154 31700  0.86 

N̂ cr  Max  4.4  ,7.1   2243  kN
 0.2731 0.2731 

And the Characteristic VAS limited by sharing is:-
 2243  119.4 
XS     2.588  95.1  4,745  m
 0.472 
A2.3.1.4 Characteristic VAS

The Characteristic Vas is given by:-
X Char  min X S , X DF , X L  Equation 4.1
So,
X Char  min 4745,4185,1000  1,000  m
A2.3.2 Calculation of Characteristic VAS: Probabilistic

The deterministic calculation is sufficient to commence the design process. However, the
designer may wish to undertake a probabilistic calculation; this is illustrated here. The
BUCKFAST computer program[A2] is employed in order to calculate the Characteristic VAS.
A2.3.2.1 Input Distributions

In order to undertaken the Monte Carlo simulation it is necessary to develop suitable
distributions of pipe-soil interaction and critical buckling force.
The deterministic pipe soil friction coefficients presented in Table 3 are used to generate
distributions for use in the reliability modelling. The axial friction and static lateral friction are
assumed to follow log normal distributions. Two points are required to define a log normal
distribution; for axial friction these are taken to be:
 The minimum friction coefficient. The chance of the friction being less than this is
estimated to be a 2.5%;
 The best estimate friction coefficient. This is taken to be the median of the distribution.
Report No: 5087471/ 01 / A

For static lateral friction these are taken to be:

 The maximum friction coefficient. The chance of the friction being greater than this is
estimated to be a 5%;
The parameters of the resulting distributions are:-
 Axial: mean = 0.662, SD=0.125, CoV=18.9%
 Lateral: mean=0.833, SD=0.150, CoV=18.0%
Ncr  XNH  Nchar  L
In accordance with the recommendations of the design guideline, the OOS parameters X NH
is assumed to follow a log normal distribution with mean 1.18 and standard deviation 0.35.
The characteristic force, Nchar is given by:-
EI  Ws 31.7MN / m2  1.049kN / m
Nchar  3.86   3.86   1.347  MN
D 0.273m
The critical force distribution is calculated using the characteristic force, the lateral friction
distributions outlined above and the distribution of the normalised force parameter, XNH. The
resulting distribution of critical buckling force is log normally distributed with a mean of
1.45 MN and standard deviation of 0.45 MN, Figure 3.
1
Pcrit
Mean
0.8
0.6
Pdf
0.4
0.2
0
1 2 3 4
P (MN)
Figure 3 Design Example #1: Distribution of Critical Buckling Force in 1 km of pipe

Since the pipeline route is straight, the seabed is relatively flat and there is little variation in
pipe-soil response, this distribution is applicable to the whole pipeline.
A2.3.2.2 Monte Carlo Simulation

The Monte Carlo analysis employs 106 simulations. The probability of at least one buckle
forming in the pipeline is 0.2%. This is below the threshold value of 5%. In this case it is not
necessary to evaluate the lateral buckling behaviour of the pipeline.
Report No: 5087471/ 01 / A

A2.3.3 Integrity with a Lateral Buckle

Although the probabilistic assessment shows that it is not necessary to evaluate the lateral
buckling behaviour of the pipeline, the assessment is undertaken based on the results of the
deterministic calculation. The lateral buckling design is undertaken at a Characteristic VAS
of 1000 m using the pipeline inlet conditions.
A2.3.3.1 Lateral Buckling Analysis

The analytical lateral buckling models developed for SAFEBUCK (see Appendix D) are
utilised to perform the buckling analysis.
The pipeline inlet operating conditions are employed and the analysis is performed for the
median undrained axial friction coefficient (0.65) and the maximum lateral residual friction
coefficient (0.9).
The calculated response is illustrated in Figure 4 and Table 5.
6 0.03
Elastic Elastic
Plastic Plastic
Unload 0.02
4
Curvature (m-1)
0.01
y (m)
2
0
0
 0.01
2  0.02
0 20 40 60 80 0 20 40 60 80
x (m) x (m)
Figure 4 Design Example #1: Lateral Buckle Results: Characteristic VAS=1000 m
Parameter Value
First Load compressive strain 0.28%
Maximum equivalent strain 0.28%
Maximum stress range 400 MPa
Table 5 Design Example #1: Loads in Lateral Buckle: Characteristic VAS=1000 m

Figure 4 shows that a well-developed buckle is predicted at the Characteristic VAS; the
maximum lateral displacement is approximately 6 m (~22 D). However, the first load strain
is low and the stress range is modest.
A2.3.3.2 Design Unity Checks

The results are subject to the appropriate limit state checks. Limit state checks are
performed for local buckling and low cycle fatigue (failure from the weld root and weld cap)
and the plasticity limitations. The calculations are performed in accordance with API 1111
as outlined in Section 6 of the SAFEBUCK guideline.
Report No: 5087471/ 01 / A

For local buckling the unity check is defined as the imposed strain divided by the allowable
strain. In calculating the imposed strain a strain concentration factor of 1.2 is employed to
account for the effect of the coating stiffness mismatches.
For fatigue the unity check is defined as the fatigue damage due to the 200 full shutdown
cycles divided by the allowable fatigue damage (0.2). The calculation is performed for
fatigue from the weld root (sweet environment assumed) and cap (seawater plus CP
environment).
The unity checks are summarised in Table 6 (detailed calculations are presented in Section
A2.7).
Check Imposed Load Value Unity Check
Local Buckling Maximum compressive strain 0.335% 0.14
Fatigue Local stress range 530 MPa 0.88
Plasticity Maximum equivalent strain 0.28% 0.16
Stress range Nominal stress range 400 MPa 0.61
Table 6 Design Example #1: Unity Check Summary: Characteristic VAS=1000 m

The table shows that all of the unity checks are met at the Characteristic VAS.
A2.4 ASSESSMENT OF CONTROLLED BUCKLING

As all of the unity checks are met at the Characteristic VAS, uncontrolled buckling is
tolerable and no buckle initiation or mitigation methods are required.
A2.5 PIPELINE WALKING
The seabed has no significant slope and is not connected to a SCR. Consequently, the only
active driver for walking is the thermal transient; the maximum transient slope is 20°C/km.
Since the start-up shutdown cycles involve relatively slow axial movement, the drained axial
response is appropriate.
A2.5.1 Pipeline Walking – No Buckle
Pipeline walking is assessed using mathematical models outlined in Appendix D.
The analysis is performed using the median drained axial friction coefficient, 0.65 and
employs 100 increments (k=100). The predicted walk is 26 mm per cycle.
A2.5.2 Pipeline Walking with a Buckle
In the event of a buckle the effective length of the pipeline is reduced. The pipeline is split
up into sections between the buckle locations and the ends of the pipeline. In this case one
buckle near the centre of the pipeline is the most likely buckled configuration. This gives two
virtual pipelines of equal length for the walking analysis. The length is effectively half the
pipeline length.
The analysis is repeated and gives a walk of 6.5 mm per cycle.
Report No: 5087471/ 01 / A

A2.5.3 Assessment
If buckling occurs then the walking rate is 6.5 mm per cycle. If buckling does not occur then
the walking rate is 26 mm per cycle. The design number of shutdown cycles is 200.
However, this is the maximum expected number of cycles. The process group estimate that
the most likely number of shutdown cycles is 100; the walking assessment is based on this
estimate. Over 100 cycles this walking rate is equivalent to an end expansion of 2.5 m. A
displacement of this magnitude could lead to overloading of the end spools.
The phenomenon should be further evaluated in detailed design using FE analysis with
accurate transient profiles.
Given the uncertainty in the phenomenon, the recommendation is to implement monitoring
procedures rather than pre invest in expensive mitigation techniques. It may be also
sensible to incorporated anchor flanges into the design.
A2.6 SUMMARY
The pipeline is susceptible to lateral buckling, although the probability of buckling is low (the
probability of a buckle forming is approximately 1 in 100). However, the assessment
demonstrates that uncontrolled buckling is tolerable and no initiation strategy is required to
control the response.
Pipeline walking may be a problem, particularly if no buckling occurs.
Report No: 5087471/ 01 / A

A2.7 UNITY CHECK CALCULATIONS
SAFEBUCK II Unity Checks
1.0 Introduction
This file performs design code checks in accordance with the SAFEBUCK II Design Guideline.
Units
2.0 Design Data

2.1 Flowline Data
Outside Diameter D fo : 273.1  mm
Select Pipe
If steel is
Wall Thickness tf : 25.4  mm Material:
"user"
complete
Corrosion Allowance tfcorr : 3  mm
material
5 1 definition
Expansion Coefficient f : 1.18  10 C
2.2 Operating Data
Operating Pressure pi : 20  MPa Operating Temperature f : 120  C
Operational design life (years) Life : 20 Number of full shutdowns per year nfull : 10
2.3 Enviromental Data

3
Water Depth d : 1000  m Water Density w : 1025  kg  m
Ambient Temperature a : 4 C
2.4 SAFEBUCK Data
Service
Bauschinger Factor f B : 0.8
Strain Concentration Factor SNCFf : 1.2 Maximum yield to tensile ratio YT : 0.92
Allowable damage D fa : 0.2 Girth weld eccentricity ef : 2  mm
SN knock down factor: SN knock down factor: Ne : 9

internal Ni : 3
external
2.5 Results from Lateral Buckling
Peak Compressive Strain cmax: 0.142  % Peak Equivalent Strain eqmax : 0.280  %
Full Shutdown (MPa) R : 400  MPa
Report No: 5087471/ 01 / A

3.0 Calculations
3.1 Select Basic Properties
Derating
Yield Strength at Temperature

SMYS f  412 MPa
3.2 Local Buckling
External Pressure pe : w g d pe  100.552 bar
Inside Diameters D fi : D fo  2  tf D fi  222.3 mm
corroded WT t : t f  tfcorr
t 22.4 mm
D fo
D over t (corroded) Dot : Dot  12.192
t
t
Critical Strain c : 0.5  c 4.101 %
D fo
Safety factor f2 f 2 : 1.2 

1  0.02  Dot
 if Service = "water"
f2  1.741
1.4 
1  0.02  Dot
 otherwise
pi  D fi  pe  D fo
Hoop stress h : h 33.463 MPa
2  tf
2
 
h
Hoop stress correction f p : 1  4 fp  1.022
SMYSfamb
Peak Compressive
fme : cmax  f f fme 
 
a 0.279 %
Mechanical Strain
Design strain 2 : fme  SNCFf
2 0.335 %
c fp
Allowable strain All :
f2 All  2.407 %
2
Local buckling unity check UCLB :
All
UCLB  0.139
3.3 Fatigue
ef
Misalignment SCF Km : 1  3 Km  1.236
tf
Report No: 5087471/ 01 / A

Thickness exponent Kt : 1 if t f < 16  mm

Kt  1.072
0.15
tf
 
mm
otherwise
16
R
Local Stress range RL :  Kt  Km
MPa
RL  529.984
Number of cycles nf : Life  nfull nf  200
12.015 3 log( )
Root hot spot - 10
xE ( , ND) :
E Curve ND
12.182 3 log( )
Cap hot spot - 10
xD ( , ND) :
D Curve ND
Allowable Cycles, internal

N fi : xE RL , N fi 
 
Ni 2317.9
Allowable Cycles, external

N fe : xD RL , N fe 
 
Ne 1134.9
Imposed Damage nf
D fat : D fat  0.176
min N fi , N fe
 
Fatigue unity check D fat
UCfat : UCfat  0.881
D fa
3.4 Plasticity Checks
Maximum nominal e : 0.25  ( 0.99  YT ) e 1.75  %

equivalent strain
eqmax
UC : UC  0.16
First load plasticity
e
Unity Check
 
3 h
Maximum allowable Rall : SMYSf  2  f B  1  Rall  657.6  MPa
stress range 4 SMYS f
Stress range Unity R

Check
UC R : UC R 0.608
Rall
Report No: 5087471/ 01 / A

A3 DESIGN EXAMPLE #2: LONG HOT PIPELINE


The pipeline parameters used in the analysis are shown in Table 7.
Parameter Value
Length 20 km
Maximum YT Ratio 0.9
Corrosion Allowance 3 mm
Water Depth 1000 m
3

The pipeline operating parameters are shown in Table 8 and Figure 5.
Parameter Value
Unload Pressure 10 MPa
Report No: 5087471/ 01 / A

25 140
120
20
Temperature (°C)
100
Pressure (MPa)
15 80
60
10
40
5
20
0 0
0 5 10 KP 15 20 0 5 10 15 20
KP
Figure 5 Design Example #2: Pressure and Temperature Profiles
A3.1.3 Friction
The seabed comprises soft to very soft clay. The range of frictional values to be employed
in the analysis is outlined in Table 9.
Axial 0.3 0.5 0.8

Report No: 5087471/ 01 / A


to X65
PIP Systems NA
Fluid Gas Hydrocarbon or water based
products


of the SAFEBUCK guideline. For the design example, only the steel yield stress has been
temperature de-rated.
maximum force in the pipeline is less than the critical buckling force. The fully constrained
force is (equation 3.3):-
 
N0  40MPa  0.039m 2  1  2  0.3  2.07  10 5 MPa  0.020m 2  1.18  10 5 o C 1  116 o C  50kN
This gives N0 = 6.17 MN. In this calculation the internal pressure during installation is
assumed to be zero and the temperature at installation is seabed ambient (4°C). The
maximum available frictional force is given by equation 3.4:-
kN 20,000m
Nf max  100kN  0.8  0.826   6.708  MN
m 2
The maximum force in the system is the minimum of these two values:-
Nmax  6.17  MN
The critical buckling force is defined as (equation 3.6):-
EI  HLB
Ncr  0.65  N  2.51  Equation 3.6
D
31.7MN / m2  0.53  0.826kN / m

Ncr  2.51   566  kN
0.273m
maximum force in the pipeline (6.17 MN) is greater than the 65% of the critical buckling force
(566 kN), and the system is susceptible to buckling.
Report No: 5087471/ 01 / A


buckling analysis at a number of locations along the pipeline to determine if uncontrolled
buckling at the Characteristic VAS is acceptable.



equation are:-
 Lower bound axial frictional force, fLB = 0.3 * 826 = 247.8 N/m
 Upper bound lateral frictional force, HUBres = 0.9 * 826 = 743.4 N/m
Solving Equation 4.3 for these inputs yields the Hobbs buckle length, LBH=113 m and the
Hobbs buckle force, NBH=84 kN.
The Characteristic VAS limited by driving force envelope is then calculated from:-
 N  NBH 
X H  2.588  L BH  2   0  Equation 4.2
 fLB 
So
 6170  84 
X H  2.588  113  2     49.4  km
 0.2478 

 N̂  N 
X S   CR BH   2.588  L
BH Equation 4.5
 fLB

 
N̂cr  Max  4.4  ,7.1   Equation 4.6
 D D 
 
Report No: 5087471/ 01 / A

 Upper bound break-out resistance, HUB=1.1 * 826 = 0.909 kN/m

 Best estimate bound break-our resistance, HBE=0.82 * 826 = 0.677 kN/m
 31700  0.909 31700  0.677 

N̂cr  Max  4.4  ,7.1   1990  kN
 0.2731 0.2731 
 
 1990  84 
XS     2.588  113  7,985  m
 0.2478 

So,
 
X Char  min  7985 ; 49,400 ; 10,000   7,985  m
 
 

calculated Characteristic VAS is large, so it is logical to undertake a probabilistic calculation
to optimise the design point. The BUCKFAST computer program[A2] is employed in order to
calculate the Characteristic VAS.

The parameters of the resulting distributions are:-
 Axial: mean = 0.517, SD=0.137, CoV=26.5%
Report No: 5087471/ 01 / A

In accordance with the recommendations of the design guideline, the OOS parameters X NH
is assumed to follow a log normal distribution with mean 1.18 and standard deviation 0.35.
EI  Ws 31.7  MN / m2  0.826  kN / m
Nchar  3.86   3.86   1.195  MN
D 0.2731  m
The critical force distribution is calculated using the characteristic force, the lateral friction
distributions outlined above and the distribution of the normalised force parameter, XNH. The
resulting distribution of critical buckling force is log normally distributed with a mean of
1.29 MN and standard deviation of 0.40 MN, Figure 6.
1.5
Pcrit
Mean
1
Pdf
0.5
0
1 2 3 4
P (MN)
Figure 6 Design Example #2: Distribution of Critical Buckling Force in 1 km of Pipe

Since the pipeline route is straight, the seabed is relatively flat and there is little variation in
pipe-soil response, this distribution is applicable to the whole pipeline.

The Monte Carlo analysis employed 106 simulations. At least one buckle formed in every
simulation. The mean number of buckles expected to form is 5. Clearly the probability of at
least one buckle forming is above the threshold value of 5%, so the buckling behaviour of
the pipeline must be assessed in accordance with the design guideline.
The distribution of Characteristic VAS calculated in the analysis is present in Figure 7.
Report No: 5087471/ 01 / A

8000
7000
6000
5000
4000
3000 Deterministic
2000 Probabilistic: 1%/km
1000 Probabilistic: 10% Total Pipe
0
0 5 10 15 KP 20
Figure 7 Design Example #2: Characteristic VAS

For an on-bottom pipeline the Characteristic VAS is the VAS whose exceedance probability
is less than 1% for any 1 km of pipe. The Characteristic VAS calculated in the analysis for
each KP is presented in the plot. The lowest Characteristic VAS is at each end of the
pipeline. At these locations the Characteristic VAS is reduced by the interaction between the
buckle feed-in zones and the end expansion zones. The maximum Characteristic VAS
calculated in the analysis is 5650 m and this occurs between KP12 and 13.
The Characteristic VAS based on the whole pipeline is also shown in the plot. This is the
VAS whose exceedance probability is 10% over the whole pipe length. Using this definition,
the Characteristic VAS is 5050 m. However, this would have to be employed in conjunction
with inlet operating condition.
The figure also shows the deterministic Characteristic VAS. This is much greater than the
value derived from the probabilistic assessment; this illustrates the benefit of undertaking the
probabilistic assessment.

Lateral buckling analyses are undertaken to determine whether the relevant unity checks are
acceptable at the Characteristic VAS for the pipe lying on the seabed.

utilised to perform the buckling analysis.
The pipeline operating conditions along the length of the pipe are employed (Figure 5) and
the analysis is performed for the median undrained axial friction coefficient (0.5) and the
maximum lateral residual friction coefficient (0.9).

and the plasticity limitations. The calculations are performed in accordance with OS-F101
(2010)[A1].
Report No: 5087471/ 01 / A

account for the effect of the coating strength mismatches.
fatigue from the weld root (sweet environment assumed) and cap (seawater plus CP
environment).
The results are presented in Figure 8.
2
1.8
1.6
1.4
Unity Check
1.2
1
Local Buckling
0.8
SAFEBUCK Strain
0.6
Fatigue
0.4
0.2
0
0 5 10 15 KP 20
Figure 8 Design Example #2: Unity Check for Uncontrolled Buckling

A lateral buckling analysis is undertaken at each KP for the calculated characteristic VAS
shown in Figure 7. From each lateral buckling analysis the relevant unity checks are
calculated and plotted as a function of KP. For example, at KP 2, the lateral buckling
analysis is performed for a VAS of 2700 m (the Characteristic VAS for KP 1 to 2) and
3700 m (Characteristic VAS for KP 2 to 3).
The figure shows that the critical unity check is the SAFEBUCK equivalent strain limit (the
allowable strain is 2.25%, based upon the maximum specified yield to tensile ratio of 0.9).
This limit state is violated between KP 1 and KP 14. The local buckling limit state is also
violated over most of this section of the pipeline. Beyond KP 14 the unity checks for all limit
states are acceptable, as a result of the fall in operating conditions.
Consequently, uncontrolled buckling is not acceptable, and a buckle initiation or mitigation
strategy is required.

A buckle initiation strategy using vertical imperfections in the form of pipeline sleepers is
selected. The design employs four 0.9 m high sleepers, located at KP2.5, 5.6, 8.7 and 12.0.
A3.4.1 Critical Buckling Load at a Sleeper

The critical buckling force from a prop type vertical imperfection is given by:-
Ncr  XNV  N V
where,
Report No: 5087471/ 01 / A

EI  W 31.7  MNm 2  0.826  kN / m

NV  4   4  0.682  MN
H 0.9  m
The parameter XNV describes the OOS in the vicinity of the vertical imperfection and is
represented by a log-normal distribution with a mean of 0.725 and standard deviation of
0.14.
The critical force distribution is calculated using the distribution of XNv and the deterministic
value of Nv. The resulting distribution of critical buckling force is log normally distributed with
a mean of 0.495 MN and standard deviation of 0.096 MN, Figure 9.
5
Vertical Prop
Mean
4
3
Pdf
0
0.2 0.4 0.6 0.8 1
Pcrit (MN)
Figure 9 Design Example #2: Critical Buckling Force at a Pipeline Sleeper

This distribution applies to each sleeper. The on-bottom pipe sections still employ the
distribution presented in Figure 6.
A3.4.2 Buckle formation

The Monte Carlo analysis employs 106 simulations. The minimum number of buckles
observed in any simulation is 2. The mean number of buckles is increased to 6. The
distribution of Characteristic VAS calculated in the analysis is presented in Figure 10.
Report No: 5087471/ 01 / A

4 Sleepers (rogues)
6000
No Triggers
5000 Triggers
4000
3000
2000
1000
0
0 5 10 15 KP 20
Figure 10 Design Example #2: Distribution of Characteristic VAS with Triggers

The red squares present the characteristic VAS for buckling at each sleeper. This is the
VAS whose exceedance probability is 10% for a given sleeper. The solid red line presents
the characteristic VAS for rogue buckling. This is the VAS whose exceedance probability is
less than 1% for any 1 km of pipe. For comparison, the figure also plots the Characteristic
VAS for uncontrolled buckling (in blue).
The severity of rogue buckling can be compared to the uncontrolled buckling (since both
involve on-bottom buckles at unknown locations). For the proposed initiation strategy the
Characteristic VAS for the rogue buckles is much lower than the Characteristic VAS for
uncontrolled buckling. This is due to the high probably that the sleepers will produce a
buckle, which limits the VAS that a rogue can experience.

Lateral buckling analyses are undertaken to determine whether the relevant unity checks are
acceptable at the Characteristic VAS for the engineered buckles at the trigger and for rogue
buckling on the seabed.
A3.4.3.1 Engineered Buckles

The analytic models are not suitable for assessment of buckling on a sleeper.
Consequently, a FE model of the sleeper is constructed to evaluate the design loads. The
FE model is employed to perform a lateral buckling assessment at the Characteristic VAS.
The unity check results are presented in Table 11.
Report No: 5087471/ 01 / A

Trigger Characteristic Unity Check
VAS (m) Local buckling Fatigue SAFE strain Stress range
KP 2.5 2850 0.116 0.521 0.131 0.669
KP 5.6 3500 0.128 0.539 0.142 0.672
KP 8.7 3700 0.125 0.488 0.139 0.645
KP 12.0 5650 0.166 0.512 0.181 0.647
Table 11 Design Example #2: Unity Check for Engineered Buckles

The table show that the unity checks are acceptable at all triggers. The sleepers reduce the
lateral restraint imposed on the buckle and hence develop much lower strain for the same
VAS. As a result, the most critical unity check is fatigue.
A3.4.3.2 Rogue Buckles

utilised to perform the buckling analysis. The assessment is the same as described for
uncontrolled buckling (Section A3.3.3.1). The results are subject to the appropriate limit
state checks, as outlined in Section A3.3.3.2.
The resulting unity checks are presented in Figure 11.
1.2 Local Buckling SAFEBUCK Strain

Fatigue
1
0.8
Unity Check
0.6
0.4
0.2
0
0 5 10 15 KP 20
Figure 11 Design Example #2: Unity Check for Rogue Buckling

The figure shows that the unity checks for rogue buckling are significantly reduced. The
maximum unity check is below 1 over the whole route. Consequently, rogue buckling is
acceptable and the outlined initiation strategy is suitable for implementation.
Report No: 5087471/ 01 / A

response is appropriate.
A3.5.1 Pipeline Walking – No Buckling
Pipeline walking is assessed using mathematical models outlined in Appendix D of the
Design Guideline.
The analysis is performed using the median drained axial friction coefficient, 0.65, rather
than the undrained values defied in Table 9. In the absence of buckling the pipeline
becomes fully constrained. In this case the criteria for walking to stop is f/f θ > 1 (Section
3.1.2 of Appendix D).
f   Ws 0.65  826  N / m
 A   1.12  1
f EA    q 4.091x10  N  1.17 x10  5  C 1  0.01  C / m
9
Consequently, walking is not a problem in the absence of buckling.

In the event of buckling the effective length of the pipeline is reduced. The most likely
response is for the sleepers to each form a buckle. This behaviour will split the pipeline into
a series of shorter sections. The first section of pipe is 2.5 km long, and the walking
assessment is repeated for an effective length of 2.5 km. The analysis gives a walk of
11.7 mm per cycle.
A3.5.3 Assessment
If buckling occurs then the walking rate is 11.7 mm per cycle. The pipeline is highly
susceptible to lateral buckling and this is the most likely prediction. The design number of
shutdown cycles is 200. However, this is the maximum expected number of cycles. The
process group estimate that the most likely number of shutdown cycles is 100; the walking
assessment is based on this estimate. Over 100 cycles this walking rate is equivalent to an
end expansion of 1.17 m. The end spools could be designed to accommodate an
incremental end expansion of this magnitude. Alternatively, a more sophisticated analysis
could be undertaken to evaluate the interaction of the lengths of pipe between buckles; this
is likely to reduce the overall walking rate.
A3.6 SUMMARY
The pipeline is highly susceptible to lateral buckling. If no buckle initiation strategy is
employed, then the most likely number of buckles along the pipeline is five. The assessment
predicts that uncontrolled buckling is not acceptable; the limit state unity checks are violated
between KP 1 and KP 16.
A buckle initiation strategy using vertical imperfections in the form of pipeline sleepers is
selected. The design employs four 0.9 m high sleepers, located at KP 2.5, 5.6, 8.7 and 12.0.
This successfully reduces the unity checks to an acceptable level, Figure 12.
Report No: 5087471/ 01 / A

1.6
Uncontrolled
1.4 Rogues
1.2 Sleepers
Unity Check
0.8
0.6
0.4
0.2
0
0 5 10 15 KP 20
Figure 12 Design Example #2: Unity Checks

The figure shows the maximum unity check (the unity check for the most onerous limit state)
from each analysis. This shows how the initiation strategy reduces the limit state associated
with on-bottom buckling. The unity checks for the engineered buckles (at the triggers) are
quite high; the spacing requirements are governed by the requirements of the engineered
buckles.
A low level of pipeline walking is predicted. Over 100 cycles this is equivalent to an end
expansion of 1.17 m. The end spools should be designed to accommodate an incremental
end expansion of this magnitude. Alternatively, given the uncertainty in the phenomenon
and the low walking rate, it may be adequate to implement monitoring procedures rather
than pre invest in expensive mitigation techniques.
Report No: 5087471/ 01 / A

A4 DESIGN EXAMPLE #3: MODEST TEMPERATURE PIPELINE


The pipeline parameters used in the analysis are shown in Table 12.
Parameter Value
Length 20,000 m
Maximum YT ratio 0.89
Corrosion Allowance 1.0 mm
3
Contents Density 200 kg/m
3
Concrete Coating 40 mm @ 2400 kg/m
Water Depth 250 m
3

The pipe has 40 mm low density (2400 kg/m3) concrete coating.
A4.1.2 Operating Conditions

The pipeline operating parameters are shown in Table 13 and Figure 13.
Parameter Value
Unload Pressure 2.5 MPa
Report No: 5087471/ 01 / A

60 25
50
20
Temperature (°C)
Pressure (MPa)
40
15
30
10
20
5
10
0 0
0 5 10 KP 15 20 0 5 10 KP 15 20
Figure 13 Design Example #3: Pressure and Temperature Profiles

A4.1.3 Friction
The seabed comprises fine sand. The range of frictional values to be employed in the
analysis is outlined in Table 14.

Axial 0.4 0.6 0.8

Report No: 5087471/ 01 / A


to X65
PIP Systems NA
Fluid Gas Hydrocarbon or water based
products


of the SAFEBUCK guideline. For the design example, no yield stress temperature de-rating
has been employed.
maximum force in the pipeline is less than the critical buckling force. The fully constrained
force is (equation 3.3):-
 
N0  25MPa  0.110m2  1  2  0.3  2.07  105 MPa  0.0195m2  1.17  105 o C1  56o C  150kN
This gives N0=3.59 MN. In this calculation the internal pressure during installation is
assumed to be zero and the temperature at installation is seabed ambient (4°C). The
maximum available frictional force is given by equation 3.4:-
kN 20000m
Nf max  100kN  0.8  1.172   9.5  MN
m 2
The maximum force in the system is the minimum of these two values, 3.59 MN.
The critical buckling force is defined as (equation 3.6):-
EI  HLB 77.1  MN / m2  0.5  1.172  kN / m

Ncr  0.65  N  2.51   2.51   0.837  MN
D 0.4064m
maximum force in the pipeline (3.59 MN) is greater than the 65% of the critical buckling force
(0.837 MN), and the system is susceptible to buckling.

Report No: 5087471/ 01 / A

buckling analysis at the Characteristic VAS.



equation are:-
 Lower bound axial frictional force, fLB = 0.4 * 1172 = 468.8 N/m
 Upper bound lateral frictional force, HUBres = 0.9 * 1172 = 1054.8 N/m
Solving Equation 4.3 for these inputs yields the Hobbs buckle length, LBH=100.3 m and the
Hobbs buckle force, NBH=261 kN.
The Characteristic VAS limited by driving force envelope is then calculated from equation
4.2:-
 N  NBH   3377  261 

X H  2.588  L BH  2   0   2.588  100.3  2     13.553  km
 fLB   0.4688 

 N̂  N 
X S   CR BH   2.588  L
BH Equation 4.5
 fLB

 
N̂cr  Max  4.4  ,7.1   Equation 4.6
 D D 
 

 Upper bound break-out resistance, HUB=1.2 * 1172 =1.406 kN/m
 Best estimate bound break-our resistance, HBE=0.8 *1172 = 0.938 kN/m
 77100  1.406 77100  0.938 

N̂ cr  Max  4.4  ,7.1   2995  kN
 0.4064 0.4064 
 
Report No: 5087471/ 01 / A

 2995  261 
XS     2.588  100.3  6,091  m
 0.4688 

So,
 
X Char  min  6,091 ; 13,553 ; 10,000   6,091  m
 
 

calculated Characteristic VAS is large, so it is logical to undertake a probabilistic calculation
to optimise the design point. The BUCKFAST computer program[A2] is employed in order to
calculate the Characteristic VAS.

In order to undertaken the Monte Carlo simulation it is necessary to develop suitable
distributions of pipe-soil interaction and critical buckling force.
The resulting distributions are illustrated in Figure 14.
Report No: 5087471/ 01 / A

4 3
log normal
min
3 median
2 max
Occurence
Occurence
2
1
1
0 0
0.5 1 0.5 1 1.5
Axial Friction Coefficient Lateral Friction Coefficient
Figure 14 Design Example #3: Friction Distributions

The parameters of the distributions are:-
 Axial: mean = 0.613, SD=0.128, CoV=20.9%
The design guideline recommends a log normal distribution with mean 1.18 and standard
deviation 0.35 for the OOS parameters XNH. However, more specific data available to the
project indicates a mean of 1.11 and standard deviation 0.15 is more appropriate; this is
used in the example.
EI  Ws 77 .1MN / m 2  1.172 N / m
Nchar  3.86   3.86   1.820  MN
D 0.4064 m
The critical force distribution is calculated using the lateral friction distributions outlined
above and the distribution of the normalised force parameter, XNH. The resulting distribution
of critical buckling force is log normally distributed with a mean of 1.95 MN and standard
deviation of 0.63 MN, Figure 15.
Report No: 5087471/ 01 / A

1.5
Pcrit
Mean
1
Pdf
0.5
0
0.5 1 1.5 2 2.5 3 3.5
P (M N)
Figure 15 Design Example #3: Distribution of Critical Buckling Force in 1 km of pipe

Since the pipeline is laid in a straight line, the seabed is relatively flat and there is little
variation in pipe-soil response, this distribution is applicable to the whole pipeline.

The Monte Carlo analysis employs 106 simulations. The probability of at least one buckle
forming in the pipeline is 96%. The mean number of buckles is 1.6. The distribution of
Characteristic VAS calculated in the analysis is presented in Figure 16.
7000
6000
5000
4000
3000
Deterministic
2000
Probabilistic
1000
0
0 5 10 15 KP 20
Figure 16 Design Example #3: Design Example #1: Distribution of Characteristic

VAS
For an on-bottom pipeline the Characteristic VAS is the VAS whose exceedance probability
is less than 1% for any 1 km of pipe. The Characteristic VAS calculated in the analysis for
each KP is presented in the plot. At the inlet end of the pipeline the Characteristic VAS falls;
this is a result of the interaction of the buckle feed-in zones with the end expansion zone.
The Characteristic VAS rises to a peak at about KP 5, and then falls as the operating
temperature and pressure reduce. At KP 10 the probability of buckling falls below 1% /km
and beyond this point the pipe is not susceptible to buckle.
Report No: 5087471/ 01 / A

The figure also shows the deterministic Characteristic VAS. This is much greater than the
value derived from the probabilistic assessment; this illustrates the benefit of undertaking the
probabilistic assessment.

The lateral buckling design is undertaken at the Characteristic VAS at each KP along the
pipeline.

utilised to perform the buckling analysis. The pipeline operating conditions along the route
are employed (Figure 13) and the analysis is performed for the median axial friction
coefficient (0.6) and the maximum lateral residual friction coefficient (0.9).

and the plasticity limitations. The calculations are performed in accordance with OS-F101
(2000).
account for the effect of the concrete coating.
fatigue from the weld root (F1 curve in air) and cap (D curve in seawater plus CP
environment).
The unity checks are summarised in Figure 17.
1.6
Local Buckling
1.4
SAFEBUCK Strain
1.2 Fatigue
Unity Check
0.8
0.6
0.4
0.2
0
0 2 4 6 8 KP 10
Figure 17 Design Example #3: Unity Checks for Uncontrolled Buckling

The figure plots the UC over the first 10 km of the pipe. The maximum UC is the due to local
buckling. The UC exceeds 1 between KP 2 and 3; over this region lateral buckling is not
acceptable. The UC drops rapidly, and beyond KP 3 the response is very benign.
Report No: 5087471/ 01 / A

A4.4 INFLUENCE OF BATHYMETRY

The bathymetry along the pipeline route is reasonably uneven. An assessment of the survey
data shows that there are several vertical features which have a critical buckling load below
that expected from inherent lateral OOS. A detailed analysis of the survey data is beyond
the scope of the conceptual design. However, to understand the impact of the bathymetry a
simplified model is adopted. In this the vertical OOS is represented by a 0.4 m prop type
high imperfection, and one feature of this magnitude is assumed to occur in each kilometre
of pipe.
A4.4.1 Critical Buckling Load at 0.4 m high Bathymetry

The critical buckling force from a prop type vertical imperfection is given by:-
Ncr  XNV  N V where,
EI  W 77 .1  MNm 2  1.172  kN / m
NV  4   4  1.901  MN
H 0.4  m
The parameter XNV describes the OOS in the vicinity of the vertical imperfection. For a
sleeper this is represented by a log-normal distribution with a mean of 0.725 and standard
deviation of 0.14. There is no reason for this distribution to apply to the bathymetric feature.
However, no better information is available at this stage in the design, and the critical force
distribution is calculated using the distribution of XNV and the deterministic value of Nv.
The resulting distribution of critical buckling force is log normally distributed with a mean of
1.378 MN and standard deviation of 0.266 MN.
A4.4.2 Buckle formation

The Monte Carlo analysis employs 2x105 simulations. The probability of at least one buckle
forming in the pipeline is increased to almost 100% and the mean number of buckles
increases to 3. The distribution of Characteristic VAS calculated in the analysis is presented
in Figure 18.
5000
4500 with Bathymetry
4000 Flat
3500 Location of vertical OOS

3000
2500
2000
1500
1000
500
0
0 5 10 15 KP 20
Figure 18 Design Example #3: Distribution of Characteristic VAS: Effect of

Bathymetry
Report No: 5087471/ 01 / A

The figure shows the Characteristic VAS for a flat seabed (in blue) and incorporating the
bathymetry (in red). The bathymetry is modelled as discrete locations of vertical OOS (the
axial locations are shown as green squares in Figure 18.
Incorporating the bathymetry in the model reduces the Characteristic VAS over the first
9 km. This is as a result of the increased likelihood of buckles occurring. Between KP 9 and
12 the Characteristic VAS increases, this is also due to buckling from the bathymetric
features in a region where the force is too low to cause buckling from the lateral OOS.

The lateral buckling design is undertaken at the revised Characteristic VAS at each KP along
the pipeline using the methodology outlined in Section A4.3.3. The unity checks are
summarised in Figure 19.
1.6
1.4 With Bathymetry
1.2 Flat
Unity Check
0.8
0.6
0.4
0.2
0
0 2 4 6 8 KP 10
Figure 19 Design Example #3: Unity Checks for Controlled Buckling

The figure shows the maximum unity check (the unity check for the most onerous limit state)
from each analysis (with and without the bathymetric features). This shows how accounting
for the bathymetric OOS reduces the limit state associated with on-bottom buckling.
All unity checks are now acceptable.

As all of the unity checks are met at the Characteristic VAS, uncontrolled buckling is
tolerable and no buckle initiation or mitigation methods are required.
response is appropriate. Pipeline walking is assessed using mathematical models outlined
Appendix D of the Design Guideline.
Report No: 5087471/ 01 / A

A4.6.1 Pipeline Walking – No Buckle

In the absence of buckling the system reaches a position of full constraint. In this case
walking will cease if f/fθ>1. At the median axial friction coefficient, 0.6 the axial restraint, f, is
703 N/m. The thermal resistance is
f  EA    q  4.04x109  1.17x105  0.01  472  N / m
Hence f/fθ =1.49 and no walking will occur.

In the event of a buckle the effective length of the pipeline is reduced. The mean number of
buckles identified in the formation analysis is 3. These are predicted to occur over the first
12 km, with a bias towards the first 10 km. This means that the most likely distance between
buckles is between 2 and 3 km; these lengths are employed in the assessment. The
analysis employs 100 increments (k=100).
The predicted walk is 5.4 mm per cycle for a 2 km section and 12 mm per cycle for a 3 km
sections.
A4.6.3 Assessment
The predicted walking rate is low, between 5 mm and 12 mm per cycle. Over 100 cycles the
total axial displacement would be between 0.5 m and 1.2 m. The end spools should be
designed to accommodate an incremental end expansion of this magnitude.
A4.7 SUMMARY
The pipeline is susceptible to lateral buckling. The assessment illustrates that the seabed
bathymetry is sufficient to initiate regular buckling and uncontrolled buckling is acceptable.
However, the conceptual analysis is based upon a very approximate model of the seabed
bathymetry. Detailed FE analysis should be employed as early in the project schedule as
practical to confirm the reliability of buckle initiation from the inherent bathymetry.
A very low level of pipeline walking is predicted in the post buckle configuration. Over 100
cycles this is less than an end expansion of 1.2 m. The end spools should be designed to
accommodate an incremental end expansion of this magnitude. Alternatively, given the
uncertainty in the phenomenon and the low walking rate, it may be adequate to implement
monitoring procedures rather than pre invest in expensive mitigation techniques.
Report No: 5087471/ 01 / A

A5 REFERENCES
A1. Det Norske Veritas, “Submarine Pipeline Systems”, DNV-OS-F101, January 2000
(reprint with amendments and corrections as of January 2003).
A2. SAFEBUCK Phase II. Probabilistic Buckle Formation: BUCKFAST Program. Atkins
Boreas Report JIP-04/BR08054/A. December 2008.
Report No: 5087471/ 01 / A

SAFEBUCK III / Safe Design of Pipelines with Lateral Buckling Page B1
Design Guideline - Appendix B
SAFEBUCK III
Pipe-Soil Interaction
Report No: 5087471/01/A

TABLE of CONTENTS
B1. Introduction ................................................................................................. 3

B1.1. Background ................................................................................................... 3
B1.2. Scope............................................................................................................ 3
B1.3. Symbols ........................................................................................................ 4
B1.4. Abbreviations ................................................................................................ 5
B2. Modelling pipe-soil interaction................................................................... 6
B3. Soil Properties Required For Design ......................................................... 8
B3.1. Geotechnical field and laboratory testing ....................................................... 8
B3.2. Soil Types ..................................................................................................... 9
B3.3. Specialised Geotechnical Model Testing of Pipe-Soil Interaction .................. 9
B3.4. Specialised Geotechnical in-Situ Testing of Pipe-Soil Interaction ................ 11
B3.5. Pipe Roughness .......................................................................................... 12
B4. Pipe Embedment ....................................................................................... 13
B4.1. Embedment and Soil Response .................................................................. 13
B4.2. Definition of Pipe Embedment ..................................................................... 13
B4.3. Effects of Installation ................................................................................... 14
B4.4. Effective Pipe Weight .................................................................................. 14
B4.5. Model to Predict Initial Embedment in Clay soil ........................................... 16
B4.6. Model to Predict Initial Embedment in Sandy Soil ....................................... 17
B5. Axial Pipe-Soil Interaction ........................................................................ 18
B5.1. Definition of Axial Resistance ...................................................................... 18
B5.2. Axial Breakout Resistance .......................................................................... 21
B5.3. Axial Residual Resistance ........................................................................... 22
B6. Lateral Pipe-Soil Interaction ..................................................................... 28
B6.1. Stages of Lateral Response ........................................................................ 28
B6.2. Lateral Resistance Models – Clay soil ......................................................... 34
B6.3. Lateral Resistance Models – Sandy soil ...................................................... 44
B7. Treatment of Uncertainty .......................................................................... 48
B7.1. Overall Uncertainty Factors ......................................................................... 48
B7.2. Individual Uncertainty Factors ..................................................................... 48
B8. References................................................................................................. 50
Report No: 5087471/01/A

B1. INTRODUCTION
B1.1. BACKGROUND
Lateral buckling, pipeline walking, route-curve pullout and flowline anchoring are all
extremely sensitive to pipe-soil interaction forces and there is significant uncertainty
associated with the characterization of these forces in design. For design, it is important to
bound behaviour. Upper and lower bound values of soil resistance are both important.
Uncertainty in soil behaviour will lead to a large range between upper and lower bound
behaviour, which is to be avoided. It is also important that the designer does not assume an
unrealistically narrow band of response.
B1.2. SCOPE
This APPENDIX B to the SAFEBUCK Design Guideline provides guidance on the pipe-soil
interaction response and the derivation of axial and lateral friction parameters for use in
pipeline design, as summarised in Lateral and Axial Interaction.
Response Description
Embedment Initial embedment, defined by the soil conditions and the loads during and
following installation is a significant influence on the axial and lateral
response that follows:
Axial Friction Axial breakout resistance First load response, a peak generally
only occurs after some set-up time
Axial residual resistance Large displacement response as the
pipe expands and contracts
Cyclic axial resistance Long term cyclic response under
shutdown/restart loading
Lateral Friction Lateral break-out Initial resistance to pipe displacement
from as-installed position, usually as a
buckle is triggered on first load
Lateral residual resistance Large displacement response as a
lateral buckle forms
Cyclic lateral resistance Long term response with soil-berm
growth under shutdown/restart loading,
with potential for increasing resistance
with embedment after repeated cycles
Table 1.1 Characteristic Pipe-soil Responses
SAFEBUCK has invested significant research effort to evaluate, quantify and understand the
complex lateral and axial pipe-soil interaction mechanisms for the low shear strength, very
soft, highly plastic clays that dominate deepwater regions, where on-bottom pipeline
solutions are currently being employed.
An extensive review of lateral response data from several recent projects donated to the
SAFEBUCK JIP, combined with tests carried out by the JIP, has led to a radical over-haul of
established geotechnical practice in pipe-soil interaction. While significant uncertainty
remains over pipe-soil interaction in such soils, a rigorous theoretical understanding of the
lateral response has emerged. Meanwhile work continues on developing a similar level of
understanding for the axial response.
Current understanding is presented, with models based on the most recent research.
Report No: 5087471/01/A

Limited test data is available for high strength soils, such as stiff-clays and sandy soils. This
is an area of ongoing research
Although this guideline is based on a relatively flat seabed, guidance is also provided on the
approach to uneven seabeds and spans that occur with some methods of buckle initiation.
Work continues under Phase GEO of the Safebuck JIP, to develop models for axial
pipe-soil interaction and improve guidance for lateral interaction. This work is not yet
complete but will be captured by an update of this Appendix during the current phase
of the JIP.
B1.3. SYMBOLS
a, b, c, d,... Calibration parameters that vary with pipe soil conditions -

Doc Outside pipe diameter - over coatings m
D‟ Projected bearing area on soil, per unit length of pipe m
Dref Reference pipe diameter m
EI Pipe bending stiffness N m2
FA Axial friction force per unit length N/m
Fp Passive soil resistance N/m
fdyn Dynamic embedment lay factor -
fcoat Coating efficiency factor -
Hbrk Horizontal resistance at breakout kN/m
Hres Residual horizontal resistance kN/m
Hberm Total horizontal resistance at established soil berm kN/m
Hcyc Horizontal resistance mid-sweep (between berms) kN/m
klay Touchdown lay factor -
m Indentation overload parameter -
R Touchdown indentation reaction N/m
Roughness measurement (average absolute deviation from
Ra µm
mean height)
SL Residual (horizontal) lay tension N
St Sensitivity -
su Undrained shear strength, referred to pipe invert level kN/m2
su_R Remoulded shear strength , referred to pipe invert level kN/m2
V Vertical unit pipe load kN/m
v Normalised vertical pipe-load (V/SuD) -
W Submerged pipe weight kN/m
Wi Submerged pipe weight at installation, usually empty N/m
Report No: 5087471/01/A

Wf Submerged pipe weight when flooded (for hydrotest) N/m

Ws Submerged pipe weight in operation N/m
z Depth of embedment at start-up m
zinit Initial pipe embedment m
 Adhesion factor -
‟ Submerged soil unit weight kN/m3
η Uncertainty factor -
Subscripts: L-Lower bound (factor); m-modelling; s-shear
strength; U-upper bound (factor); w-soil submerged weight.
Superscripts: L-Lower bound (value); U-upper bound (value)
 pipe-to -soil interface angle of friction °
Δz Additional embedment with each sweep m
Δu Average excess pore pressure kPa
µBO Lateral friction factor – breakout
µR Lateral friction factor – residual
µA Axial friction factor (or equivalent coefficient) -
µς Normal stress axial friction factor (FA/V. )
2.3% Lower Bound 2.3 percentile value for  -
50% Best Estimate value for  -
97.7% Upper Bound 97.7 percentile value for  -
σ'n Normal effective stress between pipe and soil kPa
 soil friction angle °
θ Chord angle – defines pipe-soil contact surface °
 Embedment enhancement factor (or „wedging factor‟) -
B1.4. ABBREVIATIONS
FEA Finite Element Analysis BE Best Estimate

HKU Hong Kong University LB Lower Bound
JIP Joint Industry Project LE Lower Estimate
NGI Norwegian Geotechnical Institute UB Upper Bound
PSI Pipe-soil Interaction UE Upper Estimate
UWA University of Western Australia
Report No: 5087471/01/A

B2. MODELLING PIPE-SOIL INTERACTION

B2.1.1. The Use of Friction Coefficients and Factors
Pipe-soil interaction is often modelled using a simple equivalent friction coefficient (Coulomb
friction) or friction factor, linking the pipe weight to the maximum available resistance to axial
or lateral movement. A friction factor is an unusual concept to apply to the limiting
resistance of clay, which is usually characterised by an undrained strength. However, this is
a convenient way of presenting the data for input into design, using analytical or finite
element based models, and remains comparable with approaches used previously for pipe-
soil interaction. However, this terminology should not be taken to indicate that lateral pipe
behaviour is purely frictional; the limiting lateral and axial resistance is not solely dependent
on the pipe weight, but is influenced by pipe embedment, the soil response to displacement,
and the previous history of pipe movement.
A simple axial and lateral friction factor can be used successfully for some flowline design
functions (e.g. simple stability calculations or end-expansion) and can be employed in
conceptual evaluation, if treated with care. Although lateral resistance at large
displacements is closely related to pipe weight, the use of a simple friction factor is not
appropriate in detailed numerical modelling design; particularly for lateral movement, where
a frictional model represents a significant simplification of true behaviour.
B2.1.2. Structural Finite Element Analysis

The interaction between the pipe and the seabed is incorporated into the structural analysis
of a pipeline – which is usually finite element-based – by attaching pipe-soil elements at
intervals along the pipe, to represent the axial and lateral forces applied by the soil to the
pipe. This approach is analogous the „t-z‟ and „p-y‟ load transfer methods of analysing pile
response.
The simplest pipe-soil element models are spring-sliders which provide a bi-linear elastic -
perfectly plastic response in the axial and lateral directions, where the limiting value of axial
or lateral pipe-soil resistance is calculated using a simple friction factor.
However, in order to capture the more advanced effects of interaction – particularly the large
displacement lateral behaviour – it is necessary to introduce subroutines in which the
element response is modified to account for (a) brittle breakout behaviour, (b) suction
release, (c) residual resistance at large displacements and (d) cyclic berm growth. This non-
linear force-displacement response is normally specified in a piece-wise linear fashion, with
an appropriate procedure adopted to handle cyclic behaviour.
Nevertheless, it is common to model many tens of kilometres of pipeline within a single FE
analysis. It is therefore necessary to keep the pipe-soil interaction models relatively simple.
It is not currently feasible to model the soil domain around the pipe along the pipeline length,
and so it remains necessary to encapsulate the pipe-soil behaviour into the response of a
single node. Non-linear force-displacement responses are therefore used to represent pipe-
soil behaviour. In most cases this is defined as a non-linear friction factor, which is updated
automatically throughout the analysis to simulate the underlying behaviour.
B2.1.3. Lateral and Axial Interaction

Axial and lateral resistance should be treated as independent parameters in design for any
given set of soil conditions.
Since the axial resistance response is global (affecting long sections of pipeline), the total
axial resistance is negligibly influenced by the localised lateral friction response.
Report No: 5087471/01/A

Caution is necessary with some FEA software that includes built-in soil elements,
which often combine axial and lateral friction using a combined yield surface. This is
non-conservative as it results in reduced lateral resistance when axial resistance is
mobilised and a reduced lateral resistance when axial resistance is mobilised. In
short, any combination of lateral and axial resistance causes a reduction of limiting
resistance on both axes.
Treating axial and lateral friction as independent, can underestimate the true axial
resistance where lateral displacement occurs, typically at specific regions of a lateral
buckle away from the crown. In such regions, the lateral loading may tend to
increase the axial friction locally due to the additional load on the soil. If so, treating
axial and lateral resistance as independent quantities in such regions is likely to
underestimate the local axial resistance in regions of a lateral buckle; this approach
is usually conservative.
Report No: 5087471/01/A

B3. SOIL PROPERTIES REQUIRED FOR DESIGN

The force-displacement responses given for each flowline should cover the expected range
of soil strength, and submerged unit weight. Low soil strength associated with the minimum
shear strength profile results in greater embedment, which leads to high lateral breakout
loads; whereas high soil strength results in shallow embedment and a reduced breakout
load.
It is essential that good soils data are available. Uncertain soil conditions will result in a
large range between upper and lower bound behaviour, which will increase the design
challenge and potential mitigation costs; in some cases, it may preclude the ability to
demonstrate a robust design solution. It is therefore important to define a range that
includes realistic estimates of upper and lower bound values with depth.
B3.1. GEOTECHNICAL FIELD AND LABORATORY TESTING

For pipelines, the near surface soil properties are critical and historically these are not well
measured. In weak deepwater soils, box-core samples are preferred over drop-core or soil
boring samples. In-situ T-bar tests, including cyclic loading, usually provide good data on in-
situ strength and remoulded strength for the near-surface soils[B9, B17]. Tests performed
directly in the box-core (using mini T-bar or equivalent) can produce highest quality data,
although discrepancies have been observed between in-situ tests and box core tests, which
require careful reconciliation. A real example of good remoulded shear strength design data,
based on in-situ T-bar tests, compared with genuine poor quality data for very similar soils,
based on more traditional test methods, is shown in Figure 3.2.
Shear strength (kPa)

0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0
0.0
0.1
Depth below mudline (m)
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5
1.6
Good HE Poor HE
Good BE Poor BE
Good LE Poor LE
Figure 3.1 Comparison of good and poor shear strength profiles for similar soils
In this case the upper and lower bound values for good quality T-bar data are based on
± one standard deviation of measurements taken at over ten sites in one field development.
It is notable that the good data includes a sound datum to the seabed surface and a good
record of the weak near-surface soils.
Report No: 5087471/01/A

In-situ test data of shear strength often does not register an increase in resistance at
the observed soil surface level. This can either be because the measuring device
has already penetrated the surface during placement of the seabed frame, leading to
some depth offset between the visual observations and the in situ test data.
Alternatively, the visible surface layer may be so weak that its significance can be
overlooked. In deep water, a very weak layer of soil about 50 mm to 200 mm thick is
commonly found at the surface.
Scopes for in-situ soils investigation and specialist laboratory pipe-soil testing are being
developed and refined on a project-by-project basis. The geotechnical investigation should
include in situ measurements of soil strength and ex situ measurements of pipe-soil interface
resistance – using an appropriate low-stress shear device to measure the friction angle at
the low levels of effective stress associated with pipeline bearing pressures (typically 1kPa to
10kPa). Further discussion of soil characterisation techniques for pipeline design is
presented by Randolph & White (2008)[B17].
B3.2. SOIL TYPES

In this Appendix soil types are broadly categorised as „clays‟ and „sands‟, where clays are
generally „fine-grained‟ and „plastic‟, while sands are „coarse-grained‟ and non-plastic‟. The
distinction is made principally to differentiate between calculation methods for conditions that
are „undrained‟ (associated with clays) or „drained‟ (associated with sands). It is also
recognised that in some cases, a drained response occurs in clay soils, where displacement
rate is slow enough, which is captured in the design guidance for clays. It is for the designer
to elect which approach is appropriate for the soil and loading condition under consideration.
B3.2.1. Clays
Key data for clays include:
 Shear strength profiles in-situ to about 2.0 m depth, in soft clays this should
preferably be carried out by in-situ T-bar and/or box core T-bar measurements;
 Sensitivity based on cyclic T-bar response or similar;
 Submerged soil unit weight;
 Project specific laboratory test data using surface soil samples may be carried out
using specialist testing methods.
The soil strength is characterised by the undrained strength (Su), the unit weight () and the
sensitivity (St), which is the ratio of undisturbed shear strength to the fully remoulded shear
strength.
As the soil strength usually increases with depth, it is necessary to select a representative
value of soil strength. Models presented in this guideline generally employ the soil strength
at the pipe invert (directly under the embedded pipe).
B3.2.2. Sands
Key data for sands include:
 Particle size distribution
 Relative density
 Submerged soil unit weight;
 Project specific laboratory test data, using surface soil samples may be carried out
using specialist testing methods.
B3.3. SPECIALISED GEOTECHNICAL MODEL TESTING OF PIPE-SOIL INTERACTION

Specialist ex-situ equipment for low-stress shear testing includes:
Report No: 5087471/01/A

a) Tilt table device for the measurement of drained interface friction, at the University of
Texas at Austin [B13];
b) Cam shear device, for measuring drained and undrained friction, at the University of
Cambridge;
c) Low-stress shear box, for measuring drained and undrained interface friction, at the
University of Western Australia.
These devices are used to characterise axial friction, for which it is also essential to measure
the roughness of the pipe coating interface with the soil.
Such tests may be combined with large- or small-scale tests of axial and lateral response, as
appropriate. Pipe-soil interaction in soft clay is not well understood, and detailed numerical
modelling is not yet capable of truly representing or predicting the response. Therefore,
conventional geotechnical site investigations may be supplemented with project-specific
model test programmes in which pipe-soil interaction is simulated. This is an area of
research that is constantly evolving and being rationalised on a project-by-project basis. This
type of testing has greatly improved understanding of the pipe-soil interaction mechanisms.
A large-scale pipe-soil test facility at the Norwegian Geotechnical Institute is described by
Langford et al.[B10]. Small scale modelling can be conducted in a geotechnical centrifuge, as
described by Chuek et al [B5, B6] and White et al[B20].
Large-scale model tests typically use 3m3 to 5m3 of soil collected from the field and re-
consolidated in a large tank. The model pipe is placed into the soil bed and swept axially or
laterally under appropriate force or displacement control. A number of such project-specific
tests have been carried out over recent years (Figure 3.2). The practical limitations of such
tests include the time taken to reconsolidate the soil in the tank and the maximum pipe
diameter of typically 300 mm. Nevertheless, such tests provide detailed measurements of
pipe-soil interaction in near full-scale conditions.
Large-scale lateral tests have been supplemented by small-scale centrifuge tests which
were initiated by the SAFEBUCK JIP at Cambridge University[B5] and have since been
carried out at the University of Western Australia (UWA)[B3, B7, B6] (Figure 3.3). The
advantage of centrifuge testing is that only a small quantity of soil is required and the testing
duration is much shorter.
The most advanced large-scale or centrifuge model testing facilities can replicate the
complete load and displacement patterns imposed in real conditions, including a simulation
of lay effects, changes in pipe weight due to hydrotest and operating cycles, and any pre-
determined sequence of lateral motion (or loading). In the centrifuge, the accelerated rates
of consolidation allow long periods of pipeline operation to be simulated within a single
continuous centrifuge flight. In-flight T-bar testing allows the model seabed to be
characterised in the same way as the field case, or large-scale tests. It is important that
rates of displacement in T-bar tests are correlated against offshore T-bar testing rates and
pipe model displacement rates.
Report No: 5087471/01/A

Axial test set up
Lateral test set up

Figure 3.2 Pipe-soil model testing in the NGI facility at large-scale
Figure 3.3 Pipe-soil model testing in the UWA geotechnical beam centrifuge
Many lessons have been learned in developing tests and interpreting test data for projects
and for the SAFEBUCK JIP. These lessons are vital to the ongoing value of project-specific
pipe-soil tests and the development and improvement of test facilities. Such testing forms
the basis for the generic models for the lateral response proposed in this guideline and
ongoing tests are expected to provide the same level of confidence to the development of
improved axial response models. More recently, data from operational pipelines has also
provided good validation for existing test results. Further operational data will be reviewed in
SAFEBUCK Phase III, to validate theoretical models.
B3.4. SPECIALISED GEOTECHNICAL IN-SITU TESTING OF PIPE-SOIL INTERACTION

Specialist in-situ equipment developed for low-stress shear testing includes a specialist pipe-
soil testing rig called SMARTPIPE® [B9,B11] which is mounted on a seabed frame and can
measure axial and lateral pipe-soil resistance in situ. This equipment has been deployed in
deepwater and compared with model tests in laboratories, although the full details this
comparison has yet to be published in full
Report No: 5087471/01/A

Figure 3.4 Pipe-soil in-situ testing using SMARTPIPE

Alternative methods of in-situ testing are also under development.
A key issue in all these tests is he need for good vertical load control and detailed
measurements of pore-pressure response to allow sound interpretation of the data.
The JIP is endeavouring to adopt simplified methods for evaluating pipe-soil
response, by referencing soil properties back to detailed and comprehensive tests,
this work is progressing and further work is ongoing.
B3.5. PIPE ROUGHNESS

It is clear that pipe coating roughness is an essential parameter in the assessment of pipe-
soil resistance, in particular for axial resistance, and yet there is no standard method for
assessment of coating roughness is employed by projects, nor has coating roughness been
measured in many pipe-soil test datasets.
Current recommendations are to measure roughness using the Ra value (average absolute
deviation from mean height) measured over a given area (frame size) with a laser
interferometer, or over a given distance using profile gauge, where the size of the gauge pin
(tip radius of curvature) is also recorded. The intention is to achieve some agreement
between these approaches for a consistent and repeatable roughness measurement
technique. This is discussed further in Section B5.3.5
Report No: 5087471/01/A

B4. PIPE EMBEDMENT

B4.1. EMBEDMENT AND SOIL RESPONSE
Pipe-soil interaction influences the behaviour of the pipe from the moment that the pipe
touches down on the seabed during installation. The interaction between the pipeline and
the seabed surface soil at touch-down, defines the initial pipeline embedment. Remoulding
of the soil, due to the dynamics of the pipe catenary during installation, and the
reconsolidation of the soil under the weight of the pipe, that occurs between installation and
operation, influences both the axial and lateral resistance.
This reconsolidation leads to a small increase in pipe embedment, and changes the axial
response associated with pipeline start-up. This is particularly relevant for soft clays, where
consolidation effects can significantly increase (a) the effective stress between the pipe and
the soil, and (b) the soil strength close to the pipe. These processes both increase the
resistance to first movement.
B4.2. DEFINITION OF PIPE EMBEDMENT

Nominal pipe embedment is defined as the depth of penetration of the invert (bottom of pipe)
relative to the undisturbed seabed („far‟ embedment). Pipeline embedment influences the
pipe-soil contact area, which affects the axial and lateral resistance. Heave of soil during
penetration increases the „local‟ embedment of the pipe („near‟ embedment), by raising the
soil surface against the shoulders of the pipe. The typical geometry of heave created during
monotonic vertical embedment of a pipe is such that the „local‟ embedment is typically 50%
greater than the nominal embedment relative to the original soil surface[B17]. While this level
of heave appears to be supported from field observations, recent data being collected by
SAFEBUCK indicates that this heave dissipates with time and should not be relied upon to
provide additional axial friction or thermal insulation. The nominal embedment is therefore
the conventional definition used in design to define the key pipe-soil contact arc length
parameter.
Figure 4.1 Terminology for Pipeline Embedment [B2]

It is important to capture the range of embedment that results from the range of measured
soil properties. Embedment is dependent on the following parameters:
 Soil properties;
 Pipe submerged weight;
 Pipe outside diameter;
 Additional vertical installation load at touchdown (stress concentration at touchdown);
 Dynamic installation loads at touchdown (vertical & lateral oscillation of pipe catenary);
 Seabed bathymetry.
Report No: 5087471/01/A
B4.3. EFFECTS OF INSTALLATION
The as-laid pipeline embedment usually exceeds predictions based on the static weight
alone. The vertical load induced by the laying process and particularly the dynamics of the
lay process generally lead to over-embedment (greater than embedment from self weight) of
the pipe; yet these secondary variables are not known prior to installation.
Two effects arise during the laying process that cause over-embedment: the stress
concentration at the touchdown point (arising from the catenary shape), and the uncertain
and variable influence of dynamic movement as the pipe touches down (arising from vertical
and lateral catenary oscillations). The touchdown load is actually cyclic in nature (due to
dynamic effects including vessel heave, surge, pitch and roll), combined with hydrodynamic
loading of the pipe catenary, which both contribute to the vertical and lateral catenary
oscillations.
The initial embedment is therefore calculated for the weight of pipe during installation, where
the pipe weight is multiplied by a „touchdown factor‟ to account for the static overstress
reaction at the touchdown during pipelay (defined in Section B4.4.1). The dynamic motion
at touchdown in clay soils is captured by using the fully remoulded strength of the soil, as
explained in Section B4.4.2. This embedment may also be increased by a „dynamic
embedment factor‟ to account for other influences on the lay catenary dynamics that
influence motion in the touchdown region during installation (discussed in Section B4.4.2).
The final embedment shall include any additional embedment due to the increase in the
weight of the pipe when flooded (for hydrotest) or in operation.
Special consideration is required in zones where the pipe weight changes due to loss of
contact with the seabed, for example at spans, sleepers, or zones of localised weight
reduction (distributed buoyancy). Such features cause variations in the pipe contact force.
In addition, the touchdown regions during installation, when the pipeline is air-filled, may
differ from the touchdown region in operation, when the pipe contents are usually heavier. In
this case, the touchdown load and resulting embedment may be defined by the hydrotest or
operating weight alone.
B4.4. EFFECTIVE PIPE WEIGHT

The vertical unit pipe weight V must account for the maximum vertical download defined by:-
V  max( k lay  Wi ,Wf ,Ws ) ... B4.1
Where:
Wi is the pipe submerged weight at installation, usually the empty submerged weight
klay is the a multiplier to account for the touchdown reaction during pipelay
Wf is the flooded pipe submerged weight
Ws is the operating pipe submerged weight
Equation B4.1 is intended to capture the maximum vertical download that could occur during
installation, hydrotest or in operation. On clay soils, the remoulded soil strength should be
employed for assessing embedment in the installation case (as discussed in Section B4.4.2).
The appropriate submerged weight at installation and the touchdown factor will depend on
project specific installation procedures.
Report No: 5087471/01/A

B4.4.1. Touchdown Factor
The touchdown factor (klay) is required to account for the increased touchdown reaction due
to stress concentration effects during pipelay. Typical values of this factor lie between 1.0
and 3.0 depending on the seabed stiffness, lay tension, departure angle and pipeline
bending stiffness. The touchdown factor increases in stiffer soils where the touchdown
reaction is concentrated over a shorter length of pipe. In softer soils, where the reaction is
spread over a long length of pipe, the touchdown factor can equal one (no increased
reaction). The touchdown factor can never be less than one.
The SAFEBUCK report “Touchdown Indentation of the Seabed During Pipelaying” by
Palmer[B14] derives an analytical solution. This can be used to predict the additional
indentation load that occurs in the touchdown zone, based on the pipeline configuration and
lay tension, assuming the pipe is being laid onto rigid-plastic seabed. The touchdown
embedment can be derived directly by finding the depth at which the seabed penetration
resistance equals the required indentation reaction.
The depth for a given indentation reaction is defined by:
EI .Wi  1 
2
z  1 
  m  ln( 1  )   ... B4.2
2.D.SL  m  1  m 
2
D 
Where
R
m  1 ... B4.3
Wi
B4.4.2. Dynamic Embedment

The dynamic motion at touchdown in clay soils is captured by using the fully remoulded
strength of the soil.
This approach is logical, since the adjustment of the soil strength is a more direct
representation of the dynamic lay process than an arbitrary multiplier on the static
embedment into intact soil. However, the level of dynamic pipe motion and the rate at
which the pipe is laid (and therefore the number of cycles of motion that a given
element of pipe is exposed to) can also influence the as-laid embedment.
In many cases, the fully remoulded strength coupled with the static overstress
matches well with the average embedment, although the agreement is not as good
for some pipelines. This result and other comparisons from a limited range of post-
installation surveys, suggest that the measured range of fully remoulded soil strength
leads to reasonable estimates of the average pipe embedment for average lay
conditions along a pipeline.
This embedment from installation may be increased by a „dynamic embedment factor‟ (f dyn)
to account for the level of lay catenary dynamics in the touchdown region during installation.
The default value of dynamic embedment factor is one. However, there remains
significant scatter between different pipelines laid in the same conditions. It is known
that the embedment can be under-predicted in shallow water, where the departure
angle from the lay vessel is low and the lay tension high; a review by Westgate[B19]
estimates a dynamic factor of 1.5 in one example.
Report No: 5087471/01/A

It is also known that the embedment can be under-predicted for light pipes of low
bending stiffness and over predicted for heavy pipes with a high bending stiffness.
Data from existing pipelines in the same soil region is most helpful in providing
information to calibrate this dynamic embedment factor. According to the cases
evaluated by Westgate[B19], it is estimated to range between 1.0 and 2.5, increasing
with reducing pipe weight. This study also showed that incorporating the dynamic
variation in contact stress at the touchdown gives a close match to field conditions;
however, such load conditions are not known during the design phase of a project.
This dynamic embedment factor is under investigation in SAFEBUCK Phase III,
based on a large set of donated as-installed embedment data.
The parameters that influence this „dynamic embedment‟ response are thought to include:
 pipe bending stiffness (EI);
 residual lay tension(SL);
 water depth;
 submerged weight of the pipe (V);
 outside diameter of the pipe (Doc).
If the pipe is lowered without installation dynamics, then the dynamic embedment factor
should be set to unity (fdyn = 1) and the undisturbed soil strength is used. For example an
expansion jumper lowered by crane, or a suspended region of the pipeline that touches
down only when pipeline contents are introduced.
A further complication related to pipe embedment is the definition of the soil surface,
which forms the datum for observed levels of embedment on installed pipelines and
may include a very weak layer near-surface soil that is commonly ignored in lateral
and axial resistance assessments. However, any visual assessment of pipeline
embedment from post-installation surveys will include this weak layer.
Finally, the presence of small berms on each side of the pipe due to soil heave at
penetration can provide a source of error for visual observations of the pipe
embedment, unless ‘near’ and ‘far’ measurements are taken.
Further work to assess field data for initial embedment and the influence of cyclic
loading on long-term embedment is being carried out during Phase III
B4.5. MODEL TO PREDICT INITIAL EMBEDMENT IN CLAY SOIL

White and Cheuk[B24] provide the best current guidance on predicting pipe embedment,
which supersedes earlier models. Embedment calculations should be based on the
theoretical plasticity limit analysis solution, which has now been calibrated from model
tests[B16]:
b
V z 
 a init  ... B4.4
Doc .SuR  Doc 
Where a and b are parameters that vary with soil type and pipe roughness. Rounded values
of a = 6 and b = 0.25 are convenient for design.
The effect of heave and increased buoyancy when the pipe is surrounded by soil (where the
sediment density is greater than the seawater density), will also influence embedment.
Modelling of heave and embedment is also detailed by White and Cheuk[B24], with the
equation rearranged to give the vertical load with embedment:
Report No: 5087471/01/A

Erratum corrected in equation B4.5 – October 2011
 z 
b
 '.Doc   z  z    2.z  z  z 
sin 1  4
2
V  a  .Doc .SuR  1.5. 1     21   1    ...B4.5
 Doc  4   Doc  Doc    Doc  Doc  Doc  
   
 z 
Where 
 D   0.5
 oc 
b
 z   '.Doc
2

V  a  .Doc .SuR  1.5. ... B4.6
 Doc  4 2
 z 
Where 
 D   0.5
 oc 
Modelling Uncertainty
• Uncertainty shall account for variation in each input parameter (Section B7)
• Tests tend to follow the trend of the theoretical solutions for static penetration
• Variation of 10% to 20% from the calculated value may be due to
• rates of pipe penetration
• roughness of model pipe
• uncertainty of soil strength in the near-surface zone
• soil strength ratios, leading to varying contributions from soil buoyancy
• rates of increase of soil strength with depth
• These effects are overshadowed by the dynamic lay effects
• A modelling uncertainty factor of 1.5 proposed
B4.6. MODEL TO PREDICT INITIAL EMBEDMENT IN SANDY SOIL

In sandy soil, initial embedment depth is defined as a function of contact force (pipe weight),
pipe outside diameter and submerged unit weight of the soil. The best fit to empirical data
presented by Verley & Sotberg[B18] is:
 zinit 
2/3
 
   0.037  V  ... B4.7
D    ' D2 
   oc 
Where:
Doc is the outside diameter (over coatings)
γ' is the soil submerged unit weight
V is the effective pipe weight
zinit is the initial prediction of pipe embedment
This equation does not account for static or dynamic lay effects, covered separately above.
No assessment of the applicability of this model or testing of embedment in sandy
soils has been undertaken as part of the SAFEBUCK JIP. The potential to carry out
such testing and assessment is under consideration.
Report No: 5087471/01/A

B5. AXIAL PIPE-SOIL INTERACTION

B5.1. DEFINITION OF AXIAL RESISTANCE
Axial pipe-soil resistance is often modelled using a simple friction factor, linking the pipe
weight to the available resistance to axial movement. However, the response may not be
truly frictional, in that changes in pipe weight may not give corresponding changes in axial
resistance. A finite axial displacement must occur to reach full axial resistance, so that small
axial movements can occur at quite low loads, which influences the initiation of buckling and
the pipe-walking response.
Typical axial friction model behaviour, incorporating „breakout‟ and „residual‟ axial friction, is
illustrated schematically in Figure 5.1.
2.5 Elastic-slip Breakout - drained and undrained
2
Axial pipe-soil resistance
1.5
0.5
-0.5
-1
Undrained Brittle
-1.5 2nd Cycle
-2 Undrained
Drained
-2.5
Axial Displacement Design
Figure 5.1 Axial Resistances with Mobilisation Displacement and Breakout
There are three stages of axial pipe-soil interaction:-

Axial Elastic Slip – describes the displacement associated with recoverable elastic
deformation of the soil used in FEA, which occurs at quite small displacements.
Axial Breakout Mobilisation – describes the displacement associated with the peak or
maximum friction, usually associated with full plastic deformation of the soil. The
displacement at which the breakout resistance is reached is defined as the „mobilisation
displacement‟.
Axial Residual Resistance – once the pipe has started to move, the friction falls or rises to a
„residual‟ i friction value at larger displacements. Residual axial resistance dominates the
shape of the steady state force profiles and the resulting thermal expansion, pipe-walking
and post-buckling behaviour.
i
The term „residual‟ is used by analogy with the residual friction angle that is mobilised within fine-
grained soils after continued shearing along a single plane.
Report No: 5087471/01/A
This response can vary significantly for different types of soil, different weights of pipe and
different coating roughness. In clay soils, the velocity of the pipe also has a significant effect
on the breakout and residual axial friction, because pipe displacement generates excess
pore pressure, leading to undrained conditions that can either: (i) increase the level of
resistance due to negative excess pore pressure; or (ii) reduce the level of resistance due to
positive pore pressure. High velocity can also introduce strain-rate effects that increase the
axial resistance.
In general, excess pore pressure tends to increase with increasing pipe velocity,
giving a low residual axial resistance. If pipe displacement occurs very slowly, such
that excess pore pressure is dissipated as fast as it is generated, this ‘fully drained’
condition exhibits a gradual increase to a significantly higher level of residual friction.
This is illustrated in Figure 5.1 by the ‘drained’ (shown green) force-displacement
curve, which provides a high residual axial resistance.
While dissipation of the excess pore pressure from pipe-lay is expected to occur
relatively quickly, that does not prevent undrained conditions being re-established
under conditions of rapid or large pipe displacements, indeed SAFEBUCK tests have
demonstrated this. The reduction in axial friction due to increasing velocity is not well
understood but has been demonstrated in a number of tests carried out for the
Safebuck JIP, as illustrated in Figure 5.2 and Figure 5.3 reviewed by White et al[B26].
These undrained (fast) and drained (slow) responses generally define the highest
and lowest values of axial resistance, as illustrated by the ‘Undrained’ and ‘Drained’
response curves in Figure 5.1.
In typical field conditions, the response is likely to lie between the fully drained and
undrained conditions.
Currently the important transition from undrained to drained conditions is not well
understood. So that the global average value of axial friction (Best Estimate) might
lie close to the upper bound or close to the lower bound friction value measured in
tests.
Since undrained (fast) and drained (slow) responses generally provide very different
values of axial resistance it is important to consider the interaction between the
available resistance and the resulting pipe velocity. A structural analysis based on
drained parameters that yields high pipe velocities is meaningless, and vice versa.
A significant peak – leading to a brittle breakout response – can occur when the pipe
moves axially for the first time, or after some time at rest, at a rate which is fast
enough to generate negative excess pore pressure. This brittle response appears to
be influenced more by excess pore pressure generation at the pipe-soil interface,
rather than by changes in mobilised friction angle.
The ductile response – with no obvious breakout – can occur when the pipe has not
been at rest for sufficient time to allow excess pore pressure to dissipate.
Report No: 5087471/01/A

0.7
“drained” residual friction
0.6
Residual friction factor (H/W')
0.5
0.4
0.3
0.2 JIP2-AS1
JIP2-AS2
0.1 JIP2-AS3
“undrained”
JIP2-AS4 residual friction
0
0.001 0.01 0.1 1 10
Speed (mm/s)
Figure 5.2 Effect of axial pipe velocity on residual friction factors in short pipe tests
0.8
“drained” residual friction
0.7
0.6
Residual friction factor(H/W')
0.5
0.4
0.3
Test JIP2-AL1
0.2
Test JIP2-AL2
0.1 Test JIP2-AL3

“undrained” residual friction
Test JIP2-AL4
0
0.001 0.01 0.1 1 10
Speed (mm/s)
Figure 5.3 Effect of axial pipe velocity on residual friction factors in long pipe tests
Report No: 5087471/01/A

More work is ongoing under the SAFEBUCK JIP to improve understanding and better
define the most appropriate axial friction models and optimum testing methods.
Finally, there is thought to be a link between the generation of excess pore pressure
and cumulative displacement, so that after a sufficient number of cycles the axial
resistance will tend towards the drained value, regardless of the rate of shearing[B21].
The hypothesis is that the repeated episodes of shearing will cause consolidation
within the soil adjacent to the pipe that ultimately suppresses any tendency for
excess pore pressure to be generated; this corresponds with the soil reaching its
critical state. However, the design life typically involves an order of magnitude more
cycles than those completed in axial testing. It should not be assumed that at a
particular stage in the design life of a pipeline the drained value is reached, unless
tests and analysis have demonstrated this for the site conditions.
This interaction between pipe velocity, pipe weight, coating roughness and cumulative
displacement must be considered carefully; the contribution of each parameter to the axial
resistance is the subject of ongoing research.
In sandy soils the axial friction is defined by a drained response.
B5.2. AXIAL BREAKOUT RESISTANCE
B5.2.1. Axial Elastic Slip and Mobilisation Displacement

It is important to define the „axial slip‟ and „mobilisation displacement‟ carefully, as they are
very different concepts. Elastic slip is a concept within an FE element that usually defines
the elastic displacement to reach a residual resistance; while the mobilisation displacement
describes the true non-linear, elastic-plastic displacement to reach the peak or residual
resistance, as illustrated in Figure 5.1. This elastic slip, used in FEA is a significant influence
on buckle formation and pipe walking response under thermal transient loads. Mobilisation
of „elastic slip‟ usually occurs at much lower displacements than the much larger
„mobilisation displacement‟ to „breakout‟, associated with full plastic deformation of the soil.
An elastic slip displacement less than 5mm or 0.01D for initial loading conditions and about
0.0025 D for unload-reload conditions is recommended.
The displacement required to mobilise the full axial resistance must be treated with
extreme caution in design when using finite element analysis, since the mobilisation
displacement to breakout is usually treated in FEA analysis as elastic (usually called
elastic-slip). This means that if a section of pipe is moved to its mobilisation
displacement and the load is then removed, it will return to its starting position. In
reality, a small part of the initial mobilisation displacement is elastically recovered
when the pipe is unloaded, but the majority is not. As FE analysis usually treats
mobilisation displacement as elastically recoverable, the elastic-slip value should be
kept small.
An unrealistically large elastic slip distance should not be used for pipe-walking
predictions, as this would be non-conservative. It has been shown that the rate of
pipe walking (or walk per cycle) will reduce in proportion to an increase in elastic slip
distance. An unrealistically large elastic-slip displacement (equal to the breakout
mobilisation displacement) should not be used.
The mobilisation displacement to maximum friction usually occurs after full plastic
deformation of the soil and should only be included in FE analysis if a full cyclic
elastic-plastic subroutine is available. Such a subroutine is not currently thought to be
available in the industry.
Report No: 5087471/01/A

B5.2.2. Axial Peak Breakout Resistance
An initial high breakout resistance can occur under certain undrained conditions, usually
associated with higher velocity displacements, following a long set-up (duration with no
movement). However, a high breakout is often entirely absent and should not be relied upon
to occur. Therefore simple bi-linear models for upper and lower bound residual axial friction,
incorporating a small elastic-slip mobilisation with no peak breakout resistance, are generally
sufficient for detailed design of thermal expansion, pipe-walking and post-buckling
behaviour.
Not incorporating breakout is thought to be conservative in most design cases. Even if a
high breakout occurs, it has little influence on thermal expansion behaviour and the
generation of steady state force profiles.
High breakout resistance will inhibit pipe walking under thermal transient loading (if
breakout occurs on all cycles). However, in most cases, thermal start-up transients
follow a long period of pipe cooling, during which the pipe is continuously contracting;
so a breakout response is unlikely to occur on all restart cycles. Breakout is more
likely to occur on shutdown - following a long period in steady operation, with little
axial displacement – however, this does not generally affect walking, as in most
cases cooling occurs uniformly along the length of the pipeline, with no steep
transients that could induce walking.
If it can be demonstrated that long sections of the pipeline remain fully constrained (with no
axial displacement) prior to buckle formation then incorporating an axial breakout resistance
may be warranted, provided there is evidence to support a high breakout resistance.
A high axial breakout resistance will improve the reliability of buckle formation.
However, in most cases including axial breakout is potentially non-conservative for
buckle formation and walking predictions. For this reason modelling the axial
breakout resistance should only be undertaken with great care.
Sensitivity checks on route curve pullout should include axial breakout resistance, in cases
where initial cooling occurs at restart.
A long set-up, following an extended shutdown, and a subsequent reduction in
effective force (due to cooling for example) could increase the peak axial tension
along the route curve. Route curve pullout is known to have occurred during a
system restart. In this case the pressure in the pipeline reduced initially, causing
Joules-Thompson cooling (adiabatic expansion of the gas), resulting in a small
reduction in temperature along the pipeline, causing increased tension in the pipe
that was resisted by a relatively high axial breakout friction along the whole pipeline.
B5.3. AXIAL RESIDUAL RESISTANCE
B5.3.1. The ‘Alpha’ Approach for Axial Resistance in Clay Soils

Traditional axial resistance models for „clay‟ soils define the axial resistance as the product
of the shear strength (Su), the contact area between pipe and soil (which is itself governed
by the pipe embedment) and an „adhesion factor‟ multiplier (). As the pipe is displaced
during breakout, it is commonly assumed that the shear strength reduces to the remoulded
shear strength of the soil, defined by the soil sensitivity, giving peak and residual values of
resistance (Figure 5.1). However, such traditional models are very difficult to apply due to
uncertainties about the applicable contact area and shear strength. This model is also found
to be a poor fit to experimental data[B26]. This approach is no longer recommended by the
SAFEBUCK JIP
Report No: 5087471/01/A

The contact area between the pipe and the soil depends on the pipe embedment and
is enhanced by heave around the pipe shoulders [B28,B12]. Also the strength of the soil
in contact with the pipe is not the strength at that elevation within the in situ strength
profile; instead, the pipe will initially be in contact with soil dragged down from the
surface, which will then consolidate under the weight of the pipe, becoming stronger
by an amount related to the contact stress imposed by the pipe. In addition variations
in pipe weight for the same embedment would have no effect on the axial friction
(when using the alpha approach); which is not reflected in test results.
B5.3.2. The ‘Beta’ Approach for Axial Resistance

Axial friction can be defined using a frictional model – analogous to the „beta‟ effective stress
approach for axial pile shaft capacity. In contrast to pile design, the contact stresses
between a pipe and the seabed are known, being due to the pipe weight. There is also a
small increase in axial friction due to the embedment enhancement or „wedging‟ factor,
which provides a horizontal component of contact force between the pipe and the soil,
raising the total normal force between the pipe and the soil to exceed the weight alone.
The wedging factor should be applied with caution where lateral installation dynamics
might be expected to leave gaps to each side of the pipe, which might reduce the
estimated axial friction in early life. Such gaps are unusual but have been observed
over short sections of a pipeline.
This wedging factor[B20] increases the axial resistance with increasing pipe embedment and
given by the following equation:
2 sin 
  ... B5.1
  sin  cos
Where:   Min( / 2),(cos1(1  2  z / Doc )
The axial resistance per unit length is then expressed as:-

FA  V     A (fully-drained conditions) ... B5.2
In fully-drained conditions the axial friction factor A = tan , where  is the pipe-soil interface
angle of friction, as measured on a flat shear plate or tilt-table device.
In undrained conditions, excess pore pressure, generated around the pipe-soil interface,
reduces the contact stress with the soil matrix to the „effective stress‟, which discounts the
load carried by excess pore pressure, based on the average excess pore pressure ratio:-
 u 
ru    ... B5.3
  n  average
So that equation 5.2 becomes:

FA  V     A 1  ru  (undrained conditions) ... B5.4
However a difficulty remains is in establishing the average excess pore pressure

ratio, which is a function of pipe weight, coating roughness, pipe velocity and
cumulative displacement. There is currently little guidance available to projects on
this generic response, which is why this topic is the subject of ongoing research
within SAFEBUCK GEO and other current projects.
Report No: 5087471/01/A

B5.3.3. Derivation of Friction Factors
Results from axial friction tests are often plotted against the normal effective stress which in
the absence of excess pore pressure (drained response) is equal to the total normal stress
on the test sample (in a tilt-table or shear box). For a test involving a pipe, the normal
effective stress acts around the chord of the circumference of the pipe that is in contact with
the soil. Presenting data in this way extracts the influence of embedment (or wedging) on
the axial resistance, making the flat and curved pipe shaped tests comparable.
The normal effective stress, which is the normal load divided by the area of pipe
circumference in contact with the soil, is defined as:-
V  ,
 'n  ... B5.5
Doc  
Where θ is defined above, for equation B5.1.

This reaches a maximum value at V/Doc = 0.5
The effective stress can also be defined as the vertical effective stress (or bearing load),
based on the projected area in contact with the soil, defined as:-
 'v  V / D'oc ... B5.6
Where D‟oc is the width of the contact area defined by the pipe embedment.
This reaches a maximum value of V/Doc, for ≥ 50% embedment.
The normal (depth adjusted) interface friction is then defined as:-

  FA /(V   ) ... B5.7
Currently it is recommended that projects where axial pipe-soil interaction is critical to

design undertake their own pipe-soil interaction testing, based on the framework
described above. The aim should be to measure the drained and undrained interface
friction across the appropriate correct stress levels for the pipelines being
considered, using an interface that represents that actual pipe coating interface.
B5.3.4. Drained Axial Residual Resistance

A limited number of low effective stress axial friction data are available to the SAFEBUCK
JIP from tilt-table tests, or large-scale pipe tests carried out in particular project soils.
Fully-drained data from tilt table tests are extremely good at defining the drained interface
friction and clearly show that the friction angles are influenced by the low level of effective
stress (Figure 5.4). Similar tests may be carried out by other means provided that the test
equipment is reliable at low stress levels and provides fully-drained conditions.
The friction angles ( and ) are significantly influenced by the level of effective
stress. The use of an interface friction angle measured at a more usual geotechnical
stress level would significantly under-predict the axial resistance. For an on-bottom
pipeline, the effective stress levels of 2 to 10 kPa generated by typical pipeline
weights is below that considered in conventional geotechnics. At these stress levels,
the drained friction angle of soft clays significantly exceeds that measured by
traditional testing apparatus operating at usual effective stress levels. The difference
can be critical to pipeline design. A consistent trend of residual friction is
demonstrated for different soils in Figure 5.4.
Report No: 5087471/01/A

Figure 5.4 presents donated tilt-table data to illustrate the variability of interface
friction in different soils, across a good range of effective stress. While these data
may be helpful in providing general guidance on axial friction factors, the specific
values should not be used for design because frictional strength varies between soil
types as well as being affected by the roughness and interface material. Indeed
some of the scatter within each data set is due to variations in coating roughness or
types of coating.
0.85
Carbonate Clayey SILT
Carbonate Clayey Silty SAND
0.80
Carbonate SAND
Drained Residual Friction Factor
(Tan of Secant Friction Angle)
West African CLAY

0.75 West African CLAY - smooth coat
Carbonate Clayey SILT - smooth coat
0.70
0.65
0.60
0.55
0.50
0 2 4 6 8 10 12
Effective Stress (kPa)
Figure 5.4 Typical Measured Drained Friction Factors for Specific Soils & Coatings
B5.3.5. The Influence of Coating Roughness on Residual Drained Resistance

Axial friction appears to be modified by roughness of the pipeline coating, particularly in
course-grained sandy soils but also in clay. With reasonably smooth pipeline coatings, the
shear failure surface occurs between the pipe and the soil, rather than in the soil. For this
reason many current projects are adopting roughened pipeline coatings to maximise the
available axial friction.
A „coating efficiency factor‟, fcoat has been used previously to identify the difference between
pipe-soil interface strength and soil-soil shear strength in fully drained tests. For a fully
rough pipeline coating, the interface is often assumed to be as strong as the soil. In drained
conditions this implies that  =  (the soil friction angle).
Under drained conditions (no excess pore pressure), the surface roughness of most
typical pipeline coatings clearly modifies this frictional response. Tests by Najjar et
al[B13] measured coating efficiency factors of 52% to 66% on nominally smooth
pipeline coatings (polypropylene and polyurethane) in deepwater soils. The above
tests gave drained frictional resistance values (FA/V) for these coatings of around 0.5
(range of 0.42 to 0.53) at a normal stress of 2.9 to 4.1 KPa (60 to 86psf) for the
specific soils tested. Deliberately roughened coating is known to achieve a coating
efficiency of close to 100%. Other tests donated to the JIP on West African soils give
coating efficiencies under drained conditions of 78% to 93% for smooth coatings and
95% to 96% for roughened polypropylene coating. Most of these tests have not
included detailed measurements of coating roughness.
Report No: 5087471/01/A

However, it is clear that that these test results are specific to both coatings and soils.
Little guidance can currently be given on the expected value of coating efficiency as
there is insufficient data on coating roughness to correlate with measured coating
efficiency. It is now recommended that coating roughness should be measured and
quantified for future tests, as outlined in Section B3.5.
The planned SAFEBUCK axial tests have measured roughness values from 2µm
(smooth) to 200 µm (very rough) over a 4mm frame size, with a few moderately
rough coatings (typical of factory applied sintered PP) with Ra in the region of 10 µm
to 50 µm.
B5.3.6. Undrained Axial Residual Resistance – Clay soils

The SAFEBUCK JIP has carried out undrained axial tests at a range of pipe velocities and
weights and evaluated the generic mechanisms for a range of parameters using typical
offshore soil samples available to the JIP. Generic guidance is not yet available and is the
subject of a review and future testing in SAFEBUCK GEO.
Data from large-scale tests in West African clay, using pipes of 6-inch, 8-inch and
12-inch diameter has been donated to the SAFEBUCK JIP, presented in Figure 5.5.
These tests cover a good range of typical effective stress for pipelines. This is one of
the best current datasets available to provide general guidance on undrained axial
friction factors. However, the specific values should be used with caution because
frictional strength is known to vary between soil types, as well as being influenced by
the interface roughness and material.
1.6
1.4
Measured Residual Friction Factor
Test Data
1.2
UB
1.0
BE
0.8 LB
0.6
0.4
0.2
0.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0
Note: 'Wedging' embedment Total Vertical Stress (kPa)
enhancement f actor eliminated
Figure 5.5 Axial resistances for moderately rough pipe coating in WA soil
These tests employed polypropylene (PP) coating that has been moderately
roughened by the application of sintered PP in the final stages of manufacture.
There is significant scatter in the data for a variety of reasons, including some
uncertainty in the test conditions. To reduce the spread in the data, it is normalised
by the embedment enhancement (or wedging) factor, which increases with
embedment. The data is presented as friction factor against total vertical stress, with
estimated curves of the most appropriate Upper Bound (UB), Best Estimate (BE) and
Lower Bound (LB) values, defined by the equation:-
Report No: 5087471/01/A
μς = p.σv-q ... B5.8
Where the parameters ‘p’ and ‘q’ are given in Table 5.1.
Parameters p q
UB 1.40 0.40
BE 0.88 0.40
LB 0.35 0.20
Table 5.1 Axial friction test curve parameters for fits to data
These curve fits are not applicable below 1kPa and the minimum value of friction at
high pipe weights should not be taken as less than 0.25. These correlations are
known not to encompass some of the data from another site for which SAFEBUCK
has generated undrained axial friction data, shown earlier in Figure 5.2 and Figure
5.3, so specific values should be used with caution.
Report No: 5087471/01/A

B6. LATERAL PIPE-SOIL INTERACTION

The lateral resistance to pipe displacement during lateral buckling is a combination of
(1) passive soil resistance provide by the soil berm pushed in front of the pipe, (2) soil-pipe
friction at the bottom of the pipe, (3) passive self-weight resistance from lifting soil ahead of
the pipe, and (4) potential suction at the trailing side of the pipe, during breakout. For
drained loading conditions this third component is likely to be negligible. The passive
resistance and weight components are the most difficult to determine at large displacements.
The cyclic displacements that occur with each shutdown and restart of the pipeline create
soil berms at the extremes of displacement, which are a significant influence on pipe loading.
Depending on the berm geometry, pipe geometry and pipe weight, the pipe can either dive
down into the soil or move upwards through the berm back towards the seafloor.
B6.1. STAGES OF LATERAL RESPONSE

There are two stages of lateral pipe-soil response behaviour:-
 First load displacement (monotonic) as the buckle is formed, characterised by a
breakout resistance and a residual or plateau resistance;
 Cyclic displacement in operation, characterised by the accretion of soil-berms and the
development of a trench with increasing cyclic embedment.
B6.1.1. General First Load Response

Breakout is associated with buckle formation (initiation) when the pipe breaks out from its as-
installed and embedded condition and starts to move laterally across the seabed. Breakout
is frequently accompanied by an elevated breakout load. Following breakout there is a
phase of elevation correction. If the pipe is over-embedded the resistance falls away as the
pipe rises with displacement; if the pipe is lightly-embedded or normally-penetrated this
resistance will gradually increase beyond breakout as the pipe dives further into the soil.
These two forms of breakout force-displacement response are illustrated in Figure 6.1 or
Figure 6.2.
Breakout friction is used in the assessment of critical buckling force for deterministic or
probabilistic analysis of buckle formation. The larger the breakout mobilisation displacement
the more influence breakout will have on the first load resistance, which is usually dominated
by the lateral residual friction. Lower bound breakout and residual friction may also be used
in the assessment of route-curve pullout.
After breakout, two characteristic types of large-amplitude lateral response can occur,
depending on the ratio of the flowline weight to the seabed strength. In clay soils, „Heavy‟
and „light‟ pipes can be distinguished by the ratio of the pipeline weight to the seabed
strength (V/su.D). In simple terms, values of V/suDOC < 1.5 give a „light-pipe‟ response,
characterised by the pipe rising during the initial lateral breakout, while values of v greater
than 2.5 give a „heavy-pipe‟ response, characterised by the pipe diving with displacement.
The contrasting form of the lateral load-displacement response for these two cases is
illustrated in Figure 6.1 and Figure 6.2, in terms of the equivalent friction, H/V.
In sandy soils behaviour is expected to follow the light-pipe response. However test
data for sandy soils is limited and the parameters that that define the level of
embedment at which the trajectory becomes stable have not been defined for sand.
Figure 6.1, illustrates how a „light‟ pipe (v < 1.5) generally rises after breakout. As the pipe
rises, lateral resistance usually reduces from the break-out value to a steady residual
resistance, during the first lateral sweep. The flowline sweeps horizontally, mobilising an
approximately constant (or slightly rising) horizontal „residual‟ resistance. The residual
resistance during the first load cycle controls the lateral displacement at which the first
Report No: 5087471/01/A

buckle stabilises, defining the initial shape of the lateral buckle and the peak bending stress
in the pipe.
Breakout resistance Berm resistance

Residual resistance
Horizontal
,
displacement
Figure 6.1 Lateral Pipe-Soil Friction Behaviour – ‘Light’ Pipe

Figure 6.2 illustrates how the downward movement of a „heavy‟ pipe (v > 2.5) coupled with
the growth of a soil berm ahead of the pipe, leads to a steady increase in the lateral
resistance. This hardening form of response (often termed „rising residual friction‟) gradually
increases the load in the buckle while simultaneously inhibiting further lateral displacement.
Breakout resistance Cyclic accumulation

of embedment &
resistance
Accumulating passive
Horizontal
resistance
displacement
Figure 6.2 Lateral Pipe-Soil Friction Behaviour– ‘Heavy’ Pipe
Report No: 5087471/01/A

B6.1.2. General Cyclic Response
During the first shutdown cycle, the flowline reverses direction but does not return to the
initial position due to the reversal of the axial resistance along the feed-in section (and the
reversal of lateral resistance along the buckle lobes). This behaviour is confirmed by the
soil berms established on operating pipelines where lateral buckles have occurred. Figure
6.3(a) shows a typical cross profile at the crown of a lateral buckle, where the as-installed
and maximum excursion positions are inferred from survey data. Figure 6.3(b) shows the
pair of soil berms established by a laterally-sweeping pipe section during a centrifuge model
test for SAFEBUCK JIP.
Displacement range under

start-up/ shutdown cycle
Maximum Position on As-laid

Excursion Shutdown Position
(a) Side scan at buckle crown on operating pipeline (b) centrifuge modelling of cyclic displacement
Figure 6.3 Typical cross-sectional profiles and soil berms
Surface soil, swept ahead of the pipe on each cycle, builds up into berms at the extremes of
the pipe displacement. These berms offer significant resistance to pipe movement and
define the shape of the buckle in operation. Subsequent cycles of lateral movement lead to
a steady increase in the restraint provided by the soil berms.
The reaction at the soil berm is a function of:-
 Cumulative pipe displacement, defined by the number and size of previous sweeps, and
hence the volume of material added to the berm;
 The rate of displacement into the berm and the drainage characteristics of the soil,
hence whether the response tends towards undrained or drained behaviour;
 The strength of the berm, defined for clay by remoulding under repeated cyclic loading
and consolidation;
 Encroachment into the berm, defined by the additional displacement into the berm
compared to the previous sweep;
 The pipe weight relative to the strength of the berm.
The amplitude of the cyclic sweeps and the depth to which the pipe descends defines the
volume of soil added to the berms. After the initial cycles, some soil from the berms tends to
collapse back into the trench, creating a bowl shaped trench and additional cyclic constraint.
B6.1.3. Comparison of ‘Light’ and ‘Heavy’ Pipe Response

„Heavy‟ pipe continues to move downwards into stronger soil with cycles, increasing the
resistance to pipe movement with each cycle (Figure 6.2). As the pipe penetrates deeper,
soil from the berms can remain on top of the pipe as it retreats from the berm, or soil can
even flow over the pipe at the berm, further increasing the download and resistance to lateral
movement. The contrasting lateral response of „light‟ and „heavy‟ pipes is illustrated
schematically in Figure 6.4. These tests simulated a pipe of constant vertical weight, with a
diameter of 0.8 m; the soil was soft, lightly-overconsolidated clay with an undrained strength
profile increasing with depth. The values of V/suD (based on su at a depth of one diameter)
were 0.75 and 2.8 for the „light‟ and „heavy‟ pipes respectively.
Report No: 5087471/01/A

Figure 6.4 Comparison of ‘light’ and ‘heavy’ pipe cyclic lateral responses
The „light‟ pipe (Figure 6.4a,b) showed a lateral response that matches closely the schematic
shown in Figure 6.1, with the residual friction factor remained approximately constant (0.3)
throughout each lateral sweep until the static berms were reached. The embedment of the
pipe progressively increased with cycles, reaching approximately 0.5D after 8 cycles. The
trajectory within each sweep was approximately horizontal (noting that the spatial axes of
Figure 6.4a are distorted), although the pipe rose slightly as the static berms grew, with more
soil being added. The pipe therefore reached the berms earlier in each sweep, and a gradual
rise in the mobilised berm resistance was recorded, reaching a maximum of 1.5 at the end
of the test, after 21 cycles.
By contrast the „heavy‟ pipe (Figure 6.4c,d) showed a contrasting lateral response, which is
comparable to Figure 6.2. For comparison, the first sweep from the „heavy‟ pipe test is
shown superimposed on the „light‟ pipe data (Figure 6.4b,c). During the first lateral sweep of
only 2D amplitude the pipe embedded deeper than 1D. The increasing pipe depth caused a
sharp increase in the passive resistance and no steady residual value was reached. During
the return sweep the same behaviour was observed and video footage showed that pipe
was buried beneath remoulded soil which passed over the pipe crown during each sweep.
Within 6 cycles of only 2D amplitude the pipe had reached an embedment of 3D, mobilising
a friction factor of almost 5.
Report No: 5087471/01/A

A further challenge with „Heavy‟ pipe is that sections of the buckle experiencing large
amplitude lateral displacements and increasing penetration will also experience reducing
contact pressure with the seabed. This can be a significant advantage, causing the cyclic
lateral friction to stabilise. This occurs because, with increasing penetration along the buckle
lobes, the sections between the lobes (the pivot points) experience little lateral displacement,
instead contact pressure and embedment increase to take up the vertical load of the pipe in
the buckle lobes. Ultimately this response should approach a steady state, where the
contact pressure reduces sufficiently to limit further penetration and the buckle lobe starts to
behave as a „light‟ pipe. Meanwhile, the pivot points will gradually become restrained and to
some degree resist rotation. Modelling this response is very challenging but it has been
done[B3] by taking account of the reduced contact pressure in the lobes with embedment
depth.
Figure 6.5 Model illustrating cyclic seabed penetration & contact pressure variation
[Note: This complex model was used to validate simplified ‘flat’ seabed
frictional response curves]
B6.1.4. Response Models for Uneven Vertical Loading

Special consideration is required in zones where the local effective pipe weight (vertical
load) changes due to loss of contact with the seabed, for example at sleepers, or at the start
of a distributed buoyancy section. Such features, introduced deliberately to control lateral
buckling, require careful implementation of the pipe-soil response in FEA. A distinction must
be made between the pipe effective weight, W, and the local contact force between the pipe
and the soil, denoted V. The same approach is also applicable at natural seabed features
that can cause spans to occur along the line that may also act as buckle initiators.
Particular attention should be given to the definition of soil resistance at:
 The span touchdown points (TDP) to each side of a sleeper or span;
 The start and end of distributed buoyancy or light-weight sections, where there is a
sudden change in outside diameter and pipe weight.
Report No: 5087471/01/A

The touchdown loads to each side of a sleeper may typically be 60% greater than the
effective pipe weight. The pipe will experience significant additional lateral restraint
in these touch-down regions to each side of the sleeper, which strongly influences
the first load in the lateral buckle. This effect is shown in Figure 6.6, which shows the
plan, elevation and vertical load across a lateral buckle over a sleeper, for a lower
and upper bound value of lateral resistance.
13 Low lateral resistance 13 High lateral resistance

11 11
Buckle amplitude (m)

9 9
7 7
5 5
3 3
1 1
-1 -1
850 900 950 1000 1050 1100 1150 850 900 950 1000 1050 1100 1150
-3 -3
-5 -5
1.0 1.0
Height (m)
Height (m)
0.5 0.5
0.0 0.0
850 900 950 1000 1050 1100 1150 850 900 950 1000 1050 1100 1150
-0.5 -0.5
1st Heat-up 1st Cool-down 2.5 1st Heat-up 1st Cool-down

2.5 2nd Heat-up 2nd Cool-down
2nd Heat-up 2nd Cool-down
Contact Force [kN/m]
2nd Heat-up 2nd Cool-down 2nd Heat-up 2nd Cool-down

2.0 Limits of buoyancy
1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.0
850 900 950 1000 1050 1100 1150 850 900 950 1000 1050 1100 1150
Distance along pipe [m] Distance along pipe [m]
Figure 6.6 Influence of touchdown load and lateral resistance on buckle shape
This example in Figure 6.6 includes a reduction in pipe weight by the addition of
buoyancy between 900m and 1100m, thus combining the vertical upset and local
weight reduction techniques. It is demonstrated here that the highest vertical load
occurs within the lateral buckle and is a significant influence on the shape of the
lateral buckle.
A design approach that provides an increase in lateral resistance in these zones of high
contact stress is necessary, usually by defining the lateral resistance as a function of the
local contact force between the pipe and the soil (lateral friction factor). This approach
allows the FEA to take account of seabed stiffness and bending of the pipe in the vertical
plane.
The use of a non-linear lateral friction factor tends to overestimate the breakout
friction slightly (which is generally conservative) while residual lateral friction factor is
not much affected.
The actual seabed stiffness should be calculated for a specific pipeline using the
embedment calculation in Section B4.
Report No: 5087471/01/A

A realistic value of seabed stiffness is necessary in design to arrive at realistic
contact pressures. A very stiff seabed (hard contact) can over-estimate the contact
pressures and lateral resistance at the TDP to each side of the sleeper. A realistic
seabed stiffness to be adopted in design for soft seabeds is typically in the range of
15 kN/m/m to 50kN/m/m but default seabed stiffness values in FEA are often much
higher than this.
The variation in diameter that occurs at sections of distributed buoyancy should also be
captured in FEA models to ensure correct modelling of the pipe-soil contact regions.
The initially high contact stresses that occur at buoyancy transitions due to adjacent
pipe, without buoyancy, spanning above the seabed may not be sustained under
cyclic loading as the pipe embeds and the contact stresses are redistributed.
The touch-down regions to each side of a sleeper may not experience the increased
embedment during installation if the pipe is laid empty. Instead, the pipe can first touch-
down in these regions during hydrotest or in operation. In this case, pipeline embedment
should be based on the hydrotest or operating weight, without a touch-down or dynamic
embedment factor. This will result in reduced lateral breakout resistance at buckle formation
and reduced residual resistance following breakout.
B6.1.5. Cyclic Load Modelling

Cyclic lateral resistance dominates fatigue loading and can strongly influence walking on
long pipelines.
Soil berms tend to limit the growth of buckle amplitude and constrain the buckle
shape and wavelength, inhibiting any relaxation of bending and the associated
reduction in stress range with increasing cycles.
Unless an appropriate berm resistance is modelled, the fatigue assessment should not
include the potentially unrealistic reduction in stress range with increasing cycles.
Soil berms are a significant constraint to buckle growth. If feed-in increases in late
life, due to an increase in operating temperature or localised walking, buckle growth
will be resisted by the soil berms. If the pipe breaks through the berm at the point of
highest curvature, this may cause excessive localisation of the bending loads above
that experienced in early life. Such an assessment requires a very good
understanding of the rate berm growth and resistance with cycles, possibly based on
project specific tests.
Under cyclic loading, the peak that occurs during each suction release as the pipe
pulls away from the berm, can occur some distance from the berm when soil from the
berm adheres to the pipe. Suction release contributes to soil falling back into the
self-created trench. This mechanism of crumbling soil from the fixed berms falling
back into the trench is what creates the bowl shaped trench observed with real
pipelines. The suction release load that occurs under shutdown-restart is not thought
to significantly influence lateral buckle loading.
B6.2. LATERAL RESISTANCE MODELS – CLAY SOIL

Complex force-displacement resistance models are required to capture the complete lateral
resistance response, including the initial breakout resistance, the residual resistance at large
displacements and the complex cyclic resistance due to cyclic lateral displacement.
A comprehensive review of lateral response data in clay soils was carried out for the
SAFEBUCK JIP by White & Cheuk (2010)[B25], which forms the basis of this design guidance.
This supersedes earlier guidance based on the Phase I data review (2005)[B2,B25], which was
based on limited tests of mixed quality, at a time when no theory existed for lateral pipe-soil
resistance, beyond friction factors inherited from on-bottom stability design.
Report No: 5087471/01/A
The most recent review was a major task that captured donated datasets from four
participant Operators and JIP partners UWA, supported by HKU. The quality of data far
exceeds the Phase I database, as all tests are accompanied by T-bar strength profiles and
all tests featured good or perfect vertical load control. The quantity of data (28 large-scale
tests at NGI and 55 small-scale centrifuge tests at UWA) far exceeds the Phase I database,
including up to 60 cycles of lateral movement and pipe embedment that reaches up to 3
diameters below the original soil surface during lateral cycles. The tests included five
different fine-grained soils (with shear strengths less than 10 kPa) and two coarse-grained
soils.
The review also benefited from recent work by that established empirical correlations based
on relevant dimensionless groups. The review complements these advances by cross
comparison of data using these new models.
The majority of tests were project-specific addressing specified pipe weight, diameter and
(usually) embedment, although the data does address the full range of typical pipe-weights.
However, systematic patterns of behaviour such as the varying of pipe weight or initial
embedment were rarely investigated; also the majority of tests involved fixed-amplitude
lateral cycles. This approach was generally to minimise the number of parametric variables
that were considered to be of lesser importance. Combining the datasets into one review
was expected to be very useful path to generic understanding of lateral pipe-soil response,
for example the effect of pipe weight on lateral residual response.
Consideration was also given to the development of advanced step-wise models that form
the basis of current work under SAFEBUCK GEO.
Some key parameters that affect the response were varied during the tests but are less
easily captured in generic models including:
a) Monotonic penetration - versus - simulated dynamic laying
b) Pipe velocity
c) Consolidation times between events
B6.2.1. Model to Predict Lateral Breakout Resistance – Clay Soil

The breakout resistance is defined by the following equation using a new normalisation
method that reduces the scatter within the data and demonstrates the dominance of
„passive‟ resistance over „friction‟.
d f h
H brk  z   V    ' Doc  z 
 c    e   g    ... B6.1
su .Doc  Doc   su .Doc   su  Doc 
To emphasise that the passive strength term is the most significant, that is given first; the
second term is the frictional component, while the third term captures the passive self weight
resistance, from lifting the soil ahead of the pipe. The calibration parameter values are
presented in Table 6.1.
Report No: 5087471/01/A

Calibration Parameter Value

c 1.7
d 0.61
e 0.23
f 0.83
g 0.6
h 2.0
Model Uncertainty 1.50
Table 6.1 Lateral Breakout Equation – Calibration Parameters
The scatter between this model and the database is consistent with a lognormal distribution
with a mean value of 0.99 and standard deviation of 0.25. The corresponding P5 and P95
values are reached by multiplying or dividing the calculated value by 1.5, which defines the
modelling uncertainty. Other probabilities of exceedence can be assessed directly from the
lognormal distribution. It is essential to rationalise the uncertainty associated with each of
the input parameters using the methodology outlined in section B7.
Sources of Error
Sources of error between the calculated and measured breakout resistance arise partly from
experimental error and partly from difficulties in determining the soil shear strength near to
the sample surface, which is particularly relevant to shallow embedments. The correlation is
weighted towards the pipe roughness, pipe velocity and bonding condition to the rear of the
pipe and other aspects of soils response that are not captured by the in-situ strength at the
pipe invert (for example the strength gradient and effects of remoulding during laying and
subsequent consolidation, which was present in some of the tests). Nevertheless these
effects are linked because test observations show that bonding reduces with remoulding and
increases with higher pipe velocities.
The soil deformation mechanisms during lateral breakout were assessed within the
SAFEBUCK JIP by analysing images captured during centrifuge model tests using
Particle Image Velocimetry (PIV)[B23]. It was found that the peak resistance during
breakout coincided with tensile failure at the rear of the pipe. Figure 6.7 shows the
instantaneous soil velocity fields at peak breakout resistance and immediately
afterwards. At breakout there is evidence that a two-sided mechanism is forming, but
no fully-mobilised slip plane is evident behind the pipe. After breakout, there is a
distinct slip plane in front of the pipe. These observations indicate that the breakout
resistance is limited by tensile failure at the rear of the pipe, rather than the shear
strength within the soil – which confirms why empirical methods for predicting
breakout resistance generally fall between theoretical solutions for the two extreme
cases of tensile bonding.
For very slow loading, the breakout load will tend towards the zero tension case.
However, the initial breakout of the pipeline is usually a fast event, and some level of
tension is sustained at the rear of the pipe, leading to a two-way failure mechanism
involving soil ahead of and behind the pipe.
Report No: 5087471/01/A

(a) At peak breakout resistance (b) Immediately post-peak

Figure 6.7 Soil deformation mechanisms observed using PIV image analysis [B23]
The effect of velocity is clearly illustrated in Figure 6.8, showing breakout resistance for the
same soil, pipe diameter and embedment. These rates are not necessarily scalable to other
soils since the time scale for drainage and crack development is likely to differ with the
coefficient of consolidation as well as the level of embedment.
Velocity is not captured in the existing model; as further testing would be required to
create a generic model for different soil types.
2
1.8 Breakout rate: 4.5 mm/min
1.6
Breakout rate: 1.5 mm/min
1.4
Friction factor, H/V (-)
1.2
1
0.8
0.6 Breakout rate: 0.9 kN/m/day
0.4
0.2
0
0 0.5 1 1.5 2
Horizontal displacement, u/D (-)
Figure 6.8 Example of the effect of velocity on breakout response

Breakout Friction Factors >> 1.0
The above model may predict very high breakout resistances for light pipes that are deeply
embedded; gas pipelines that were flooded after installation are a typical example.
Report No: 5087471/01/A

A breakout friction factor much greater than one is difficult to justify. It is important to
recognise that all pipe tests in the test database were displaced horizontally at
breakout (simulating a lateral buckle). If the load had been applied in a different
direction (vertically upwards for example) breakout may have occurred at a lower
resistance. If no bonding occurs between pipe and soil, the vertical breakout
resistance would equal the pipe weight (Hbrk/V = 1). If breakout occurred at an angle
to the horizontal then some intermediate breakout resistance may apply.
The critical breakout force at which buckling of the pipe is triggered is a function of
breakout resistance and out-of-straightness (OOS). OOS can occur in any plane and
buckles can readily initiate vertically and then become lateral buckles. Therefore in
situations where a breakout resistance factor much greater than one is predicted, the
use of a mean breakout factor equal to one may be justified in the assessment of
buckling probability, provided that bonding resistance with the soil is low.
Care should be taken in applying this approach to post-formation buckle loading, as
once the buckle has broken out and starts to expand laterally across the seabed,
higher levels of horizontally applied lateral breakout resistance may be experienced
away from the crown of the buckle.
It is recognised that non-horizontal breakout friction should be assessed and possibly
linked more closely to the levels of OOS. Further work is needed in this area.
B6.2.2. Model to Predict Lateral Mobilisation Displacement – Clay Soil

The mobilisation to reach breakout typically occurs at a displacement of less than 10% of a
pipe diameter for a statically embedded pipe, reducing with less embedment to around 5% of
diameter at 25% embedment. However, dynamically embedded pipes can experience much
higher mobilisation displacements, although the breakout resistance is often reduced.
0.1
uMU/D=0.025+0.10(z/D)
0.09
Mobilisation displacement [u/D]
0.08
0.07
0.06
0.05
uMB/D=0.016+0.04(z/D)
0.04
0.03
0.02 uML/D=0.004+0.02(z/D)
0.01
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Initial Embedment [z/D]
Figure 6.9 Examples of lateral breakout displacements for large-scale tests with
static penetration (these exclude heavy pipe tests with no peak breakout)
Values in (Figure 6.9) are based on a limited dataset of large-scale West African clay
and Ønsoy clay tests, excluding ‘heavy’ pipe response where there is no discernable
breakout. However, pipes that have been ‘dynamically’ embedded by applying small
amplitude cyclic lateral displacements during the embedment phase, can incur much
Report No: 5087471/01/A

larger mobilisation displacements to breakout because the pipe has to climb up from
the initially embedded condition as the pipe displaces laterally (Figure 6.10).
In such cases the breakout resistance for a dynamically embedded pipe is usually
less than for a statically embedded pipe at the same embedment. However, a
general rule to assess these two types of response has not yet been developed.
Breakout Displacement
This is based on a review of all tests, excluding heavy pipe response
1.20
A
Mobilisation Displacement [u/D]
1.00 B
0.80 C
D
0.60
E
0.40 F
0.20 G
H
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40
Initial Embedment [z/D]
Figure 6.10 Examples of lateral breakout displacements for large- and small-scale
tests that include simulated dynamic loading during penetration
The mobilisation displacements shown above may be also exceeded in soils that
differ from those employed in these tests.
B6.2.3. Model to Predict ‘Residual’ Resistance – Clay soil

The lateral residual resistance is defined by the following equation:
k
H res  z 
i j   ... B6.2
V  Doc 
The calibration parameter values are presented in Table 6.2
Calibration Parameter Value
i 0.32
j 0.8
k 0.8
Modelling uncertainty 1.50
Table 6.2 Lateral Residual Equation – Calibration Parameters
Report No: 5087471/01/A

This equation is normalised by H/V, rather than H/suD used previously (in Phase I with the
reduced dataset). This reduces the scatter within the data and indicates that the pipe
reaches a stable depth at which a balance is achieved between:
• depth of cut into undisturbed soil
• growth of the berm (which raises the passive resistance)
• decrease in the strength of the soil within the berm (due to remoulding)
It is clear that vertical load governs the pipe trajectory (high V/su.D leads to high w/D), and
the passive resistance is governed by the embedded depth (high w/D leads to high H/su.D),
the result is that the residual resistance, Hres, varies with V
The second term in the equation captures the influence of the size of the active berm (being
pushed ahead of the pipe), which is defined by the initial embedment.
This approach also removes the uncertainty in establishing the local shear strength in the
near-surface soils and provides better agreement with the „heavy‟ pipe response.
The scatter between this model and the database is consistent with a lognormal distribution
with a mean value of 1.00 and standard deviation of 0.25 (which is the same as the breakout
model). The corresponding P5 and P95 values are reached by multiplying or dividing the
calculated value by 1.5, which defines the modelling uncertainty. Other probabilities of
exceedence can be assessed directly from the lognormal distribution. The scatter shows no
skew with respect to the initial embedment, soils strength ratio, or normalised vertical load.
Sources of Error
Sources of error between the calculated and measured residual resistance arise partly from
experimental error and partly from difficulties in determining:
 the loss of soil shear strength with the remoulding of the soil berm;
 the trajectory followed as the pipe moves to the steady elevation;
 effects of pipe roughness,
 effects of rate of movement;
However, the correlation is inevitably weighted towards the actual pipe roughness, pipe
velocity and other aspects of soils response from the tests that were intended to be as
realistic as possible for a pipeline in the types of deepwater soil used in the tests.
B6.2.4. Enhanced Model to Predict Lateral Resistance – Clay soil

An enhanced model that captures the non-linear response by defining the pipe displacement
trajectory and berm growth, using a step-wise analysis is likely to reduce some of the
uncertainty in the residual response.
This model is based on the effective embedment concept[B22], outlined in Figure 6.11 and
detailed by White & Cheuk[B25]. Predictions of residual resistance are certainly improved with
this approach and it is well suited to „heavy‟ pipe response. However the model usually
requires calibration to capture the kinematics observed in project specific tests[B3].
Report No: 5087471/01/A

(a) velocity field during centrifuge test (b) modelling idealisation
Figure 6.11 Lateral sweeping mechanisms and the effective embedment concept[B22]
B6.2.5. Models to Predict ‘Cyclic’ Resistance – Clay soil

The cyclic response is strongly influenced by the increasing embedment of the pipe.
Therefore, the „cyclic residual resistance‟ (Hcyc) and „mobilised berm resistance‟ (Hberm) are
both defined in terms of the mid-sweep embedment for each cycle. The cyclic resistance
should be based on the actual vertical load (V) exerted by the pipe, which may be greater or
less than the submerged weight of the pipe.
Cyclic Embedment:
3 2
 V  z  V 
0.01    0.15  ... B6.3
 sU .Doc  Doc  sU .Doc 
Cyclic Mid-sweep Resistance:

Hcyc
For z/Doc<~0.8 0.25   0.9 ... B6.4
V
 z  Hcyc  z 
For z/Doc>~0.8 0.25  0.3  0.8    0.9  1.5  0.7  ... B6.5
 Doc  V  Doc 
Cyclic Berm Resistance:
0.4
 z  H berm  z 
0.9    4.5  ... B6.6
 Doc  sU .Doc  Doc 
The berm resistance is intended to model the overall resistance to lateral movement when
the pipe is in contact with the berm. The full berm resistance will typically start to be
mobilised over a distance of 1D (one diameter) from the face of the berm that was
established during the previous cycle.
Lateral buckles with uneven support conditions will require careful consideration of the effect
of increasing contact pressure at support points and reducing contact pressure at low points
along the lobes of the lateral buckle. This will occur where the pipeline passes over a
feature such as a sleeper, or seabed out-of-straightness; or where significant levels of
increasing embedment occur with lateral displacement; as discussed in Section B6.1.5.
These are bounding equations that capture the large and scattered database of results
including all the uncertainty in the tests. The fit to the data is illustrated in Figure 6.12 to
Figure 6.15. The approach is to establish broad guidance rather than definitive calculation
methods for the assessment of cyclic behaviour. A large range is inevitable, as the database
spans many soil types and test conditions. „Heavy‟ pipe can fall outside these bounds.
Sources of Error
The cyclic response is more complex than the embedment and first sweep response:
 Soil berms at the end of the sweeping range are remoulded and grow with each cycle
 There is no established theoretical basis for predicting resistance during each
successive sweep across fresh seabed
Report No: 5087471/01/A

 It is difficult to assess the changes in soil strength, as the soil surface is exposed by
each sweep it swells and softens, depending on the elapsed time between sweeps
 Over-topping behaviour as soil is transported on top of the pipe or passes over the
pipe crown is challenging to assess but is captured within the test data
If detailed quantification of cyclic response is required then project-specific analysis,
based on appropriate tests may be needed
Figure 6.12 Increasing embedment of the pipe based on normalised vertical load
Report No: 5087471/01/A

Figure 6.13 Mid-sweep cyclic resistance with mid-sweep embedment – shallow depths
Figure 6.14 Mid-sweep cyclic resistance based on mid-sweep embedment – all depths
Report No: 5087471/01/A

Figure 6.15 Berm resistance based on mid-sweep pipe embedment
B6.3. LATERAL RESISTANCE MODELS – SANDY SOIL

Limited testing on sand was undertaken during SAFEBUCK; consequently, the models
recommended here are based on public domain information focussed on breakout
resistance for pipeline stability; and a participant donation of a study by Peek (2006)[B15] on
lateral residual friction. In addition, recent project specific tests on sandy calcareous soils
donated to the JIP, were reviewed in Phase II[B27] and compared with the current
SAFEBUCK models including those proposed in the study by Peek, now included as an
Appendix to the review[B27].
The current SAFEBUCK guidance, which was not developed for calcareous soils,
provides a poor prediction of the response from these tests and the discrepancy is
not consistent; trends encapsulated in the guidance are not evident in parametric
model test studies. This is not unexpected since the recommended equations have
no soil properties as input parameters, apart from soil weight. This suggests that
improved recommendations for both siliceous and calcareous sands should be based
on a re-evaluation of the controlling mechanisms rather than a simple recalibration of
the empirical parameters in the current guidance. Key parameters are likely to
include: friction angle, interface friction (or coating efficiency), soil dilatancy and unit
weight.
Since systematic differences in behaviour were observed between different sands,
quantitative conclusions are not applicable to all sandy soils; however, these results
do provide the most comprehensive advice currently available.
Further work is being considered under SAFEBUCK GEO to address pipe-soil
responses in sandy soils. For now, the guidance differentiates between siliceous and
calcareous sands.
Report No: 5087471/01/A

B6.3.1. Models to Predict Breakout Resistance - Sandy Soil
For sandy soil, Verley & Sotberg[B18] define the lateral breakout resistance from full scale
model tests performed on siliceous sandsii. The breakout resistance is divided into a
frictional and a passive component that provides additional resistance with penetration. For
higher values of V/γ‟D2oc, the passive component also varies with pipe weight:-
Hbrk   V  Fp B6.7
Where the passive resistance is defined as:
0.15 '.Doc
2
V
 0.05 Fp   '.Doc (5  )  ( z / Doc )1.25
2
If B6.8
 '.Doc
2
V
V
 0.05 Fp  2 '.Doc  (z / Doc )1.25
2
If B6.9
 '.Doc
2
A value of 0.6 is adopted for the friction coefficient (μ) by Verley & Sotberg[B18].
Extensive data used in developing this model covers diameters from 0.3 to 1.0m,
embedments from 1% to 35% and V/γ‟.D2oc values from 1 to 0.04.
The level of uncertainty associated with these equations is not known.
A set of centrifuge model tests in two sandy carbonate soils was donated to the JIP
capturing embedment, breakout and large amplitude sweeping. A SAFEBUCK
review[B27] compared these results with the above breakout models and found that
they over-predict the breakout resistance by a factor of 1.5 on average.
B6.3.2. Models to Predict Lateral Residual Resistance – Sandy Soil

During lateral buckling of pipelines on a sandy soil, the resistance to lateral movement is
controlled not only by frictional between the pipeline coating and the sand, but also by the
passive resistance of the soil berm pushed ahead of the pipeline. The lateral residual friction
factor is defined as the ratio of the lateral force to maintain steady-state lateral movement, to
the submerged weight of the pipe.
Although lateral buckling may at first occur dynamically, the buckles will subsequently grow
under further increases in temperature. It is during this subsequent, quasi-static response
that the largest pipe stresses and strains develop. It is for this part of the response that the
effective friction coefficients are estimated for siliceous sands, based on the methodology
provided by Peek[B15,B27].
The median value of the friction factor may be calculated from:
(Hres/V)50% = 0.71 (V/γ‟.Apipe)0.12 .(Dref/Doc)0.18 ... B6.10
Where:
Apipe is the cross sectional area of the pipe ( D2/4),
Dref is a reference diameter taken as 20inches (508mm)
The following lower and upper bounds for the friction factor were proposed:
(Hres/V)2.3% = 0.05 + 0.7 (Hres/V)50% ... B6.11
(Hres/V)97.7% = 0.18 + 1.15 (Hres/V)50% ... B6.12
ii
There is a typographic error in the second equation from this reference which is corrected here.
Report No: 5087471/01/A
These are not bounds in the strict sense of the word. The estimated probability that the
actual value falls outside the bounds is 2.3%iii on either side.
These results apply for typical sand deposits. Conditions that could result in exceptionally
loose deposits, such as shallow gas, or artesian water are excluded. Also excluded are any
conditions that would result in a value of 50% of less than 0.5. (For this case the upper
bound could still be calculated from Equation 6.8 using 50%=0.5.)
Whereas the relative density and the variation of relative density with depth seems to be
important, the effect of relative density is not well understood and the upper bound estimate
has been raised slightly based on engineering judgement rather than statistical theory to
account for this. Additional effective lateral friction data, coupled with careful measurements
of the relative density could be used in future to reduce the uncertainty, possibly enabling
narrower bounds to be justified.
In addition, coating roughness is an essential parameter to measure in any future tests.
This recommendation assumes that the residual resistance is independent of initial
embedment, which is in contrast to Verley and Sotberg[B18], who assume that the residual
embedment is half the embedment at breakout, and then use equations B6.7 to B6.9 to
assess residual lateral resistance. Based on the dataset review for carbonate sands, this
approach does not provide an improved basis for assessment of lateral residual friction over
the Peek model[B27].
Often conditions are encountered in which a veneer of sand overlies a stiffer substrate. If
the thickness of the sand layer is less than half a pipe diameter, then some reduction in the
friction coefficients may be appropriate. In any case the upper bound may be used as a
conservative value no matter how thin the sand layer may be.
It must be emphasized that these lateral friction coefficients only account for the berms that
form for monotonic steady-state lateral movements of the pipe. Cyclic displacements of the
pipe during shutdown-restart are likely to result in larger berms and increasing resistance to
growth in lateral displacements, this is addressed, to a limited extent for sandy soils in
Section B6.3.3.
Centrifuge model tests in two sandy carbonate soils were donated to the JIP. The results for
large amplitude sweeping were compared with the above model in a SAFEBUCK review[B27],
which showed that the response is fairly well represented, with values of measure/predicted
residual friction factor ranging from 0.6 to 1.1 with an average of 0.9.
A key conclusion from this SAFEBUCK review is that improved recommendations for
both siliceous and calcareous sands should be based on a re-evaluation of the
controlling mechanisms rather than a simple recalibration of the empirical parameters
in the current guidance.
B6.3.3. Prediction of Lateral Cyclic Resistance – Siliceous Sandy Soil

Little data is available on the cyclic response of pipelines in sandy soil.
The following results are indicative and should only be used with caution.
For the „lower bound‟ (2.3%) or „upper bound‟ (97.7%) value, the probability of exceeding this value is
iii
the same as the probability that a normally distributed random variable exceeds the mean value plus
two standard deviations.
Report No: 5087471/01/A
Tests performed by SAFEBUCK[B1, Annex B] in 2004 using a 250mm diameter pipe in
siliceous sand (D60 = 250 µm, D10 = 110 µm) were most informative about the
general response, although the test rig had some issues, common to most early
tests, due to uncertain vertical load control. Nevertheless, cyclic frictional behaviour
was reasonably consistent; the equivalent friction rising from an initial value of about
0.4 to a final value of about 1.0. Fluctuations are also present, as the pipe appears
to climb the active berm and then accrete a new berm in a series of waves. These
tests observed extensive berm building. The resistance experienced by the pipe at
the berms was between 3 and 5 times the vertical load. Berm encroachment (berms
collapsing back into the trench) was observed as the berms become established over
a number of cycles, giving increased resistance as the pipe reaches maximum
amplitude on each cycle. Over a large number of cycles, this is likely to result in a
bowl-shaped lateral trajectory for the pipe; such behaviour has been observed in
practice.
The SAFEBUCK review[B27] of a set of centrifuge model tests donated to the JIP also
captured large amplitude sweeping in calcareous sand and calcareous silty sand.
These tests showed that the berm walls stabilised at 10° to 15° slope, with a flat
trench between them, when the cyclic amplitude was large enough. Over 50 cycles
the mid-sweep friction factor (Hmid/V) appeared to be stabilising at between 0.3 and
0.6 at mid-sweep. However, this is not a steady value across the sweep (as happens
on the first sweep); instead (after a few sweeps) the resistance is almost zero as the
pipe retreats from one berm (moving down the slope), increasing more rapidly with
every sweep as it advances (up the slope) towards the other berm. Indeed the mid-
sweep resistance rises to merge with the increasing berm resistance, reaching
resistance factors (Hberm/V) of 0.8 to 1.5 after 10 cycles, and 1.25 to 2.5 after 50
cycles. Pushing through the berms mobilised broadly similar levels of resistance,
with no sudden rise to a much higher resistance.
The range of responses observed here will be affected by the pipe diameter, pipe
weight and interface conditions as well as the soil properties. So, although these
cyclic results provide an indication of the cyclic behaviour and the governing
mechanisms, values of pipe-soil resistance that lie outside of these ranges are likely
to be observed in other conditions. In addition, the number of cycles in these tests is
likely to be significantly exceeded in practice and the trends may vary at higher
numbers of cycles.
Report No: 5087471/01/A

B7. TREATMENT OF UNCERTAINTY

Clearly, there is a significant degree of uncertainty associated with the models, which arises
from a number of sources. The main sources of uncertainty are modelling uncertainty,
uncertainty in the soil properties such as in-situ shear strength and submerged weight of the
clay. Design calculations must account for all of these sources of uncertainty. Advice on the
statistical treatment of uncertainty in soils data is given in DNV-RP-C207[B8].
The recommended approach for the short-term axial resistance, the breakout lateral
resistance and the residual lateral resistance is to calculate a median resistance and then
apply uncertainty factors to the median value to calculate upper and lower bound values.
The details of this approach[B4] are outlined in this section.
Uncertainty in the long-term axial resistance or berm resistance is addressed through the
upper and lower bound friction factors defined earlier, until models have been developed that
can define these parameters in terms of the soil properties.
B7.1. OVERALL UNCERTAINTY FACTORS

The upper and lower bound responses are calculated using overall uncertainty factors, as
shown in Equation B7.1.
U
 Hbreakout  u Hbreakout
Hbreakout L
Hbreakout  ... B7.1
L
Equation B7.1is written in terms of the lateral breakout resistance; a similar approach should
be adopted for the axial resistance and the residual lateral resistance.
The overall uncertainty factors cover the combined effect of the uncertainty in each individual
parameter (in this case model uncertainty, shear strength uncertainty and soil submerged
weight uncertainty). The individual uncertainties are combined probabilistically, based on
the assumption that the individual uncertainties are log-normally distributed; this results in
the definition of Equations B7.2 and B7.3.
ηU  exp  lnηm 2  lnηSU 2  lnηWU 2  ... B7.2

 
ηL  exp  lnηm 2  lnηSL 2  lnηWL 2  ... B7.3

 
B7.2. INDIVIDUAL UNCERTAINTY FACTORS

The uncertainty factors must be calculated for each parameter in turn and then combined to
produce the overall uncertainty factor using Equation B7.2 and B7.3.
B7.2.1. Model Uncertainty Factor

The model uncertainty factor reflects the inherent uncertainty in the prediction of the
equations outlined in Sections B4 to B6, i.e. the uncertainty that would remain in the
prediction even if all of the input parameters were known with absolute certainty. In principle
the factor should vary depending upon which equation is being evaluated, however in
practice significant judgement is required in selecting an appropriate value and at this stage
a model uncertainty factor of 1.5 is considered suitable for all of the equations, except for
cyclic lateral response where the equations are bounding.
Report No: 5087471/01/A

B7.2.2. Shear Strength Uncertainty Factor
The approach is described here with reference to the break-out lateral resistance. The same
approach should be adopted to address the uncertainty in the embedment. Although, the
best estimate (median) of embedment should be calculated using the median shear strength
profile and the median soil submerged weight in combination with Equation B4.4, B4.5 and
B4.6 and then Equation B4.1 to determine the best estimate embedment.
L
The lower bound breakout resistance, H BreakoutS , should be calculated using the upper
bound shear strength profile and the median soil submerged weight in combination with
Equation B6.1 (to define the lower bound embedment)
U
The upper bound breakout resistance, H BreakoutS , should be calculated using the lower
bound shear strength profile and the median soil submerged weight in combination with
Equation B6.1 (to define the upper bound embedment).
The shear strength uncertainty factors are then defined by Equation B7.4.
HUBreakoutS H Breakout
ηUS  ηLS  ... B7.4
HBreakout HL BreakoutS
B7.2.3. Soil Unit Weight Uncertainty Factor

The approach is described here with reference to the lateral breakout resistance; a similar
approach should be adopted to address the uncertainty in the embedment and lateral
residual resistance.
L
The lower bound breakout resistance, H BreakoutW , should be calculated using the lower
bound soil submerged weight and the median shear strength profile in combination with
Equation B6.1 (to define the lower bound breakout)
U
The upper bound breakout resistance, H BreakoutW , should be calculated using the upper
bound soil submerged weight and the median shear strength profile in combination with
Equation B6.1 (to define the upper bound embedment).
The soil unit weight uncertainty factors are then defined by Equation B7.4.
H U BreakoutW H Breakout
WU  SL  L ... B7.5
HBreakout H BreakoutW
Report No: 5087471/01/A

B8. REFERENCES
B1. Bolton, M.D., Cheuk, C.Y., White, D.J., and Bonab, M.H. Modelling of soil-pipeline
interaction for large amplitude deflections for the SAFEBUCK JIP. Cambridge
University Technical Services. July 2004.
B2. Bruton, D., White, D., Cheuk, C., Bolton, M., Carr, M. “Pipe/Soil Interaction Behaviour
During Lateral Buckling”. SPE Journal of Projects, Facilities and Construction.
1(3):1-9 (2006).
B3. Bruton, D., White, D., Langford, T., Hill, A. “Pipe-Soil Interaction Testing For Design
of a Deepwater Project Offshore Angola”. Society for Underwater Technology, First
Annual Subsea Technical Conference, Perth. February 2009.
B4. Carr M., Matheson I. & Peek R. Design Specification for Clad Pipelines Subject to
Lateral Buckling on a Flat, Soft Clay Seabed. SIEP Report EP 2005-5154 (Restricted
Issue). (2005)
B5. Cheuk C.Y., Bolton M.D. “A technique for modelling the lateral stability of on-bottom
pipelines in a small drum centrifuge.” Proc. International Conference on Physical
Modelling in Geotechnics, Hong Kong (2006).
B6. Cheuk, C.Y. and White, D.J. “Centrifuge modeling of pipe penetration due to dynamic
lay effects.” Paper OMAE2008-57302 Proc. Int. Conf. on Offshore Mechanics and
Arctic Engineering, Estoril, Lisbon (2008)
B7. Dingle, H.R.C., White, D.J. and Gaudin, C. “Mechanisms of pipe embedment and
lateral breakout on soft clay.” Canadian Geotechnical Journal (accepted for
publication 2007).
B8. DNV-RP-C207 Recommended Practice: Statistical Representation Of Soil Data
October 2010.
B9. Hill, A.J. and Jacob, H “In-Situ Measurement of Pipe-Soil Interaction in Deep Water”
Proc. Offshore Technology Conference, Houston, USA. Paper OTC 19528 (2008).
B10. Langford, T.E., Dyvik, R. & Cleave, R., “Offshore pipeline and riser geotechnical
model testing: practice and interpretation.” OMAE2007-29458. Proc. of the
Conference on Offshore, Marine and Arctic Engineering (2007)
B11. Looijens, P. and Jacob, H.: “Development of a Deepwater Tool for In-Situ Pipe-Soil
Interaction Measurement and its Benefits in Pipeline Analysis”. Proc. 31st Offshore
Pipeline Technology Conf., Amsterdam, The Netherlands (2008).
B12. Merifield R.S, White D.J. & Randolph M.F. “The effect of pipe-soil interface conditions
on undrained breakout resistance of partially-embedded pipelines.” Int. Conf. on
Advances in Computer Meth. & Analysis in Geomechanics, Goa, India. (2008).
B13. Najjar, S.N., Gilbert, R.B., Liedtke, E.A., McCarron, W. “Tilt Table Test for Interface
Shear Resistance Between Flowlines and Soils.” Proc. Conf. on Offshore Mechanics
and Arctic Engineering (2003).
B14. Palmer, A. C., “Touchdown indentation of the seabed.” Applied Oceans Research
(Submitted for publication 2008). Based on Safebuck Report “Touchdown Indentation
of the Seabed During Pipelaying” Feb 2007.
B15. Peek, R., “Pipeline on Sand – Resistance to Lateral Movements” SIEP Internal
Report Jan 2006.
Report No: 5087471/01/A
B16. Randolph M.F. & White D.J.. “Upper bound yield envelopes for pipelines at shallow
embedment in clay.” Géotechnique, (in press 2008).
B17. Randolph, M.F. & White, D. J., “Pipeline Embedment in Deep Water: Process and
Quantitative Assessment”. Offshore Technology Conference Houston USA Paper
OTC 19128 (2008).
B18. Verley, R.L.P. and Sotberg, T. A Soil Resistance Model Pipelines Placed in Sandy
Soils. International Conference on Offshore Mechanics and Arctic Engineering.
OMAE, 1992
B19. Westgate, Z.J., White, D.J., Randolph, M.F., & Brunning, P. Pipeline Laying and
Embedment in Soft Fine-grained Soils: Field Observations and Numerical
Simulations. OTC 20407. Offshore Technology Conference, Houston (2010 )
B20. White D.J. & Randolph M.F. “Seabed characterisation and models for pipeline-soil
interaction.” Proc. 17th Int. Offshore & Polar Engng. Conference, Lisbon, Portugal
(2007). [Note: a typographic correction to the published equation was agreed with
the authors]
B21. White D.J., and Cathie D.N. “Geotechnics for subsea pipelines”. Proc. 2nd Int. Symp.
on Frontiers in Offshore Geotechnics. Perth. 87-123 (2010).
B22. White D.J., Dingle H.R.C. & Gaudin C.. SAFEBUCK JIP Phase II: “Centrifuge
modelling of pipe-soil interaction”: Factual Report for Boreas Consultants
(SAFEBUCK JIP), ref. UWA report GEO 07396. 57pp. (2007) (confidential)
B23. White D.J., Take W.A. & Bolton M.D. “Soil deformation measurement using Particle
Image Velocimetry (PIV) and photogrammetry.” Géotechnique 53 7:619-63 (2003).
B24. White, D.J., Cheuk, C.Y. Cambridge University Technical Services. Lateral pipe-
soil interaction: data review – Report prepared for the SAFEBUCK JIP.
October 2005. Report SC-CUTS-0502R02
B25. White, D.J. and Cheuk, C.Y. “SAFEBUCK JIP: Pipe-soil interaction models for lateral
buckling design [- Clay soils]: Phase IIA data review”. Report to Atkins Boreas and
the SAFEBUCK JIP, UWA report GEO 09497v3. 185pp. June 2010.
B26. White, D., Ganesan, S.A., Bolton, M.D., Bruton, D.A.S., Ballard, J-C., Langford, T.
SAFEBUCK JIP: Observations of axial pipe-soil interaction from testing on soft
natural clays. Offshore Technology Conference. OTC 21249 May 2011.
B27. White D.J. “SAFEBUCK JIP: Pipe-soil interaction models for lateral buckling design -
Sandy soils: Phase IIA data review”. Report to Atkins Boreas and the SAFEBUCK
JIP, UWA report GEO 10520v2. 56pp. October 2010
B28. Zhou H., White D.J. & Randolph M.F. “Physical and numerical simulation of shallow
penetration of a cylindrical object into soft clay.” Proc. ASCE GeoCongress, New
Orleans (2008).
Report No: 5087471/01/A

SAFEBUCK III / Safe Design of Pipelines with Lateral Buckling Page C1
Design Guideline: Appendix C
SAFEBUCK III
Design Guideline: APPENDIX C
Application of DNV OS-F101
Report No: 5087471/ 01 / A

Design Guideline: Appendix C
TABLE of CONTENTS
C1 General ........................................................................................................ 3
C1.1 SAFETY CLASS ........................................................................................... 3
C1.2 LOAD AND CONDITION FACTORS ............................................................. 3
C2 Limit States.................................................................................................. 4
C2.1 LOCAL BUCKLING ....................................................................................... 4
C2.2 FATIGUE AND FRACTURE .......................................................................... 4
C2.3 PLASTICITY LIMITATIONS .......................................................................... 4
C3 REFERENCES ............................................................................................. 5
Report No: 5087471/ 01 / A

Design Guideline: Appendix C: Appendix C
C1 GENERAL
An enormous amount of effort has been expended in developing the formulation outlined in
OS-F101[C1] and there is common acceptance within the pipeline industry that this code is
the most widely applicable set of design rules. The review performed for SAFEBUCK
confirmed that the formulations outlined in OS-F101 are suitable for application to the lateral
buckling problem. However, the requirements of OS-F101 should only be applied to a
pipeline that is wholly designed, specified and fabricated to OS-F101.
C1.1 SAFETY CLASS

The safety classification must be decided on a project-by-project basis. However, for
hydrocarbon service the flowline will typically be categorised as a medium safety class
pipeline. If the fluid is a non-flammable water based product the flowline will typically be
categorised as a low safety class pipeline.
This categorisation assumes that any buckles that form are in location class 1 (i.e.
outside the platform safety zone). In some cases this may not be true and a different
categorisation should be applied.
If a PIP system is employed the jacket pipe should be given the same safety classification as
the flowline.
Since the jacket pipeline does not contain hydrocarbon there is an argument for
assigning a lower safety classification. However, while failure of the jacket pipe does
not immediately lead to loss of containment it does open the way for incremental
failure of the flowline and hence loss of containment. Thus to achieve a given safety
class both pipelines require the same categorisation. This requirement may be
relaxed if engineering is undertaken to demonstrate that a catastrophic failure of the
jacket pipe does not compromise the integrity of the flowline.
C1.2 LOAD AND CONDITION FACTORS

The design factors defined in the code should be employed with the following
interpretations:-
 γF = 1.1
 γC = 1.07
Report No: 5087471/ 01 / A

C2 LIMIT STATES
C2.1 LOCAL BUCKLING
The buckling limit state can be checked in accordance with the displacement-controlled
criteria of OS-F101, Section 5 D608 and 609i.
For a seamless linepipe the local buckling performance of the pipeline is
compromised by the plateau in the stress-strain response. For failure on the Lüder
plateau the design equations may result in unacceptably high failure probabilities.
The suitability of the local buckling design equations for use at a lateral buckle has
been evaluated in the SAFEBUCK JIP[C2]. This found that the basic equations are
not acceptable when failure is at strains below the Lüder strain. For internal over
pressure this is only likely to apply to for pipes with D/t>30 and σh/σy<0.5.
C2.2 FATIGUE AND FRACTURE

The fatigue life of the pipeline welds shall be checked in accordance with OS-F101 and
Recommended Practice RP-C203[C3].
Detailed recommendations for ECA are given OS-F101. The assessment should be
undertaken in accordance with the recommendations of OS-F101 Appendix A.
The guidance given in Section 6.3 to 6.5 of this guideline should be consistent with the
recommendations of OS-F101. This can be used where OS-F101 is silent.
C2.3 PLASTICITY LIMITATIONS

The limitations outlined in Section 6.6 of this guideline should be applied.
i Alternatively, the user is free to employ the moment controlled criteria of OS-F101.
Report No: 5087471/ 01 / A

C3 REFERENCES
C1. Offshore Standard DNV-OS-F101. Submarine Pipeline Systems 2010. Det Norske
Veritas. October 2010.
C2. SAFEBUCK Phase II. Lateral Buckling SRA. Atkins Boreas Report JIP-04/ BR06066/ B.
10 December 2007.
C3. Recommended Practice DNV-RP-C203. Fatigue Design of Offshore Steel Structures.
2008. Det Norske Veritas.
Report No: 5087471/ 01 / A

SAFEBUCK III / Safe Design of Pipelines with Lateral Buckling Page D1
Design Guideline: Appendix D
SAFEBUCK III
Design Guideline: APPENDIX D
Analytic Lateral Buckling and Walking Models
Report No: 5087471/ 01 / A

Design Guideline: Appendix D
TABLE of CONTENTS
D1 First Load Elastic Plastic Lateral Buckling Model .................................... 3
D1.1 MODEL FORMULATION .............................................................................. 3
D1.2 MOMENT CURVATURE RESPONSE .......................................................... 7
D1.3 SOLUTION PROCEDURE .......................................................................... 11
D1.4 MODEL VALIDATION ................................................................................. 12
D2 Cyclic Elastic Lateral Buckling Model ..................................................... 27
D2.1 MODEL FORMULATION ............................................................................ 27
D2.2 VALIDATION............................................................................................... 32
D3 Pipeline Walking........................................................................................ 47
D3.1 THERMAL TRANSIENTS ........................................................................... 47
D3.2 SEABED SLOPES ...................................................................................... 48
D3.3 RISER TENSION ........................................................................................ 48
D3.4 LIQUID HOLD-UP ....................................................................................... 49
D3.5 APPLICATION TO PIPELINES WITH LATERAL BUCKLING...................... 49
D3.6 NOMENCLATURE ...................................................................................... 49
D4 REFERENCES ........................................................................................... 51
Report No: 5087471/ 01 / A

Design Guideline: Appendix D: Appendix D
D1 FIRST LOAD ELASTIC PLASTIC LATERAL BUCKLING MODEL

D1.1 MODEL FORMULATION
D1.1.1 Basic Assumptions

The model makes the same key assumptions employed by Hobbs. These are:-
 The lateral friction is fully mobilised over the whole of the buckled pipe;
 The axial friction is fully mobilised on the slip zones;
 The analysis uses linear differential equations which are only valid for small slopes;
 The axial force is constant within a buckle;
 The pipe has no initial geometric imperfection;
D1.1.2 Development of Governing Equations

The basic mode III shape is presented in Figure 1.
LB q
Mode 3
y y
μW μW
z x
P P
q
q
L2 L
Figure 1 Mode III Buckle Deformed Shape

The force in the buckle is P (compression positive) and the lateral restraint is q (=μL·W).
The length of the central lobe is termed L, this is in accordance with the terminology
introduced by Hobbs. The length of the secondary lobe is L2; this gives an overall buckle
half-length of LB, where:-
L
LB   L2 … D1.1
2
Initially the central lobe is considered. If the slope of the pipe is small, then the free-body
diagram developed is shown in Figure 2.
Report No: 5087471/ 01 / A

q
S
P
P M+dM
S+dS
y+dy
M
y
dx
y
Figure 2 Free-Body Diagram for Buckled Pipe

From vertical equilibrium:-
dS
q … D1.2
dx
From moment equilibrium:-
dM dy
P S 0 … D1.3
dx dx
Differentiating equation D1.3 and combining with equation D1.2 yields:-
d2M d2 y
P q0 … D1.4a
dx 2 dx 2
This is the governing differential equation. A similar approach for the secondary lobe yields:-
d2M d2 y
P q0 … D1.5
dx 2 dx 2
Since the derivation is based entirely on statics, there is no assumption on how the moment
varies, i.e. there is no assumption of linearity.
D1.1.3 Boundary Conditions

The boundary conditions for the problem are, at the centre of the buckle, x=0:-
y0  0 (Symmetry) …i
M0  0 (Zero reaction force at centre) … ii
where ‘ indicates differentiation with respect to x. At x=-L/2 we get:-
 L
y    0 (Definition of lobe end) … iii
 2
Report No: 5087471/ 01 / A

 L  L
y    y2    (Continuity of slope) … iv
 2   2
 L  L
     2    (Continuity of curvature) …v
 2  2
 L  L
M    M2    (Continuity of shear force) … vi
 2  2
where subscript 2 refers to the response in the secondary lobe. At x=-LB we get:-
y2  LB   0 (Zero displacement at end of lobe) … vii
y2  LB   0 (Zero slope at end of lobe) … viii
y2  LB   0 (Zero curvature at end of lobe) … ix
D1.1.4 Solution in the Central Lobe

In the central lobe the pipeline is allowed to develop plasticity, i.e. the moment is not a linear
function of the curvature. In order to solve equation D1.4a it is necessary to employ a
numerical method. However, before this can be applied it is necessary to convert the
governing differential equation to a more convenient form. Consider the integral form of
equation D1.4a.
dM dy
P  q x  0 … D1.4b
dx dx
In which, the integration constant is zero as a result of boundary conditions (i) and (ii). Now
the moment is a general function of the curvature, κ, and curvature is defined as:-
 y
 … D1.6a
1  y 
2 3/2
In this equation the sign arises since convex upwards (as drawn in Figure 2) is a negative
curvature. Since the model has been developed for small slopes this equation should be
simplified to:-
  y … D1.6b
However, at this stage this simplification is not employed; this is mathematically inconsistent,
but has been found to provide superior results. Equation D1.4b and D1.6a can be recast in
terms of the pipeline slope, θ (which is simply y’):-
dM
P   q x  0 … D1.4c
dx
 
 … D1.6c
1   
2 3/2
Now since M is a function of κ we can write:-
dM dM d
  … D1.7
dx d dx
Report No: 5087471/ 01 / A

and from equation D1.6c:-
 
d    1  2  3    2
 … D1.8
dx 1  2 
5/2

So substituting equation D1.8 into equation D1.4c and re-arranging yields:-
 

 P    q  x   1  2 
3/2
3    2
dM

1   
2
… D1.9
d
This is a second order differential equation, which can be integrated to obtain a solution.
However, to undertake this procedure a relationship between the moment and curvature is
required; this relationship is considered later. In addition, initial conditions for θ and θ’ are
required. These are provided by the compatibility conditions (iv) and (v). In order to
evaluate these it is necessary to consider the solution in the secondary lobe.
D1.1.5 Solution in the Secondary Lobe

In principle the non-elastic behaviour could be modelled in the secondary lobe. However, it
is simpler if a wholly elastic response is assumed for this lobe. In practice only modest
plasticity could occur in this lobe for a realistic pipeline solution. If the response is elastic,
the moment is given by:-
M  EI  y 2 … D1.10
Which is consistent with equation D1.6b for small slopes. Substituting this into equation
D1.5 yields:-
yiv2  2  y2    0 … D1.11
Which is the normal small slope form of the differential equation, where:-
P
  … D1.12a
EI
q
 … D1.12b
EI
The general solution of equation D1.11 is:-
  z2
y 2  A  cos  z   B  sin  z    Cz D … D1.13a
2  2
Where the axial ordinate z is used rather than x (see Figure 1), since this simplifies the
mathematics. Applying boundary conditions (iii), (vii), (viii) and (ix) yields the solution:-
y2 
 
cos  z     sin  z     z  
  z2  1 … D1.13b
4  2 

where:-
Report No: 5087471/ 01 / A

cos  L 2  
  L 2 2 1
 2 … D1.13c
sin  L 2     L 2
There is still one unknown, the length of the lobe, L2. To evaluate this we consider boundary
condition (vi). In the secondary lobe, equation D1.10 holds, so differentiating this yields:-
dM2
 EI  y2 … D1.14a
dx
In the primary lobe the moment derivative is defined by equation D1.4b. Noting the
compatibility condition (iv), the boundary condition (vi) yields:-
L
y2 L 2   2  y2 L 2   0 … D1.14b
2
Substituting the solution for y2, equation D1.13b, into this yields:-
sin  L 2     L 2      L 2    L   1    L 2 
2
 cos  L 2  … D1.15
 2  2
Equation D1.15 can be solved to yield the unknown length L2.
D1.1.6 Modifications to the Model

An improvement to the results produced by the model outlined above can be achieved by
modifying the assumptions for the secondary lobe. Two modifications are proposed:-
 The curvature is calculated accounting for slope, using equation D1.6a;
 The length of secondary lobe is calculated from the elastic value, L 2 =4.63249, rather
than equation D1.15.
Both of these modifications are less coherent than the assumptions they replace, however
they have been found to improve the predictions of the model when compared to finite
element analysis.
D1.2 MOMENT CURVATURE RESPONSE

In order to develop the elastic-plastic bending response the following assumptions are
made:-
 Thin wall assumption is employed for the pipes;
 No ovalisation of the cross-section is considered;
 The steel is elastic-perfectly plastic with a von Mises yield criteria;
 All stresses are tension positive.
D1.2.1 Elastic Response

The stress distribution under elastic bending is illustrated in Figure 3.
Report No: 5087471/ 01 / A

dθ
y y
z
θ σL
x z0 z0
Axial
force
Figure 3 Elastic Stress Distribution

There is an axial force on the section. As a result the neutral axis is located a distance z0
from the centroidal axis. The axial force in the pipe wall, Fw, is defined as:-
Fw   L  dA … D1.16a
A
Where σL is the axial stress. Since thin wall theory is employed, the equation can be re-cast
as:-
/2
Fw  2  R  t  L  d … D1.16b
 / 2
Where R is the pipe mean radius and t is the wall thickness. Similarly, the moment in the
pipe is given by:-
/2
M   L  y  dA  2  R  t  L  y  d … D1.17a
A
 / 2
The axial strain at any point is given by:-
L    z … D1.18
Where κ is the imposed curvature and z is the distance from the neutral axis. Equation
D1.18 is true for both elastic and elastic-plastic deformation. The stress in the pipe is:-
L  E    z … D1.19a
Equation D1.19a is true only for elastic deformation. The distance z can be written in terms
of the angular coordinate, θ as:-
z  z0  R  sin … D1.20
Substituting equation D1.19a and D1.20 into D1.16b and D1.17a yields:-
Fw
z0  … D1.21
2    R  t   E  
and
Report No: 5087471/ 01 / A

M  EI   … D1.22
D1.2.2 Elastic-Plastic Response

Once yield occurs the stress distribution changes, as illustrated in Figure 4.
Y+
dθ
z2
y y
z
θ σL
x z0 z0
z1
Axial
force
Y-
Figure 4 Elastic-Plastic Stress Distribution
The material yields in accordance with the von Mises yield criteria. Under the thin wall
assumption, the von Mises yield condition is:-
2 2 2
 -    Y
L L H H
… D1.23a
Where σH is the hoop stress and Y is the material yield stress. This can be rearranged to
provide the axial stress at yield as:-
H 3 2
Y   Y 2   H … D1.23b
2 4
in tension and:-
H 3
Y   Y 2   H2 … D1.23c
2 4
in compression.
The curvature to cause first yield can be evaluated by combining equations D1.19a to D1.21
with equations D1.23b and D1.23c to give:-
  Fw Fw 
Y   Y 
 y  min 2   R  t , 2   R  t  … D1.24
 E R E R 
 
 
For a curvature in excess of the yield curvature, the distance from the neutral axis to the first
yield point is simply:-
Y Y
z1  ; z2  … D1.25
E E
Report No: 5087471/ 01 / A

Which can be written in terms of θ using equation D1.20:-
 Y z   Y z 
1  a sin  0 ; 2  a sin  0 … D1.26
E  R R  E  R R 
   
The strain distribution is unchanged (defined by equation D1.18), but the stress distribution
is then:-
Y   1
L  E    z 0  R  sin 1     2 … D1.19b
Y   2
The lateral buckling model often fails to find a solution because the moment curvature
response goes flat when severe bending occurs (resulting in the denominator dM/dκ→0). To
counteract this, the true stress can be considered to provide a more robust solution and the
following approximation is used within the model:-
Y   1  L    1
L  E    z 0  R  sin 1     2 … D1.19c
Y   1  L    2
The axial force in the pipe can then be found by combining equations D1.16b and D1.19c
as:-
       
 Y   1  2   1    z 0   Y   2  2   1    z 0  
     
Fw  2  R  t  


   R  Y   cos2   Y   cos1    

… D1.16c
 E    z 0  2  1   R  cos2   cos1 
 
 
Equation D1.16c and D1.26 can be solved to yield the neutral axis offset, z0. Then
equations D1.17a, D1.19c and D1.26 can be combined to yield:-
   Y    
cos1     Y   1    z 0   E    z0   
   E   
   
 
       
 cos 2    Y  1    z 0 
Y   E    z0  
MR t
2
 … D1.17b
  E  
    
 
        
    R   E   2  1   Y   1    Y     2  
  2 2  
 
Equation D1.17b is the elastic-plastic bending response of the pipeline.

The differential equation D1.9 requires the derivative of M with respect to the curvature. This
can be obtained numerically by assuming a linear variation in moment over a small curvature
interval dκ. This involves evaluating equation D1.17b twice, at κ+dκ/2 and κ-dκ/2.
Report No: 5087471/ 01 / A

D1.3 SOLUTION PROCEDURE

To solve the problem, equation D1.9 must be integrated numerically. There are a number of
techniques for example, the second order Runge-Kutta integration scheme. The integration
starts at x=-L/2, with initial conditions defined by the secondary lobe:-
 L
    y2 L 2
 2
  … D1.27a
 L
 
y1    y2 L 2
 2
… D1.27b
There are still two unknowns in the solution; the axial force in the buckle, P, and the length of
the central lobe, L. These must be evaluated by iteration.
D1.3.1 Iteration on Buckle Length

For a given value of the buckle force, P, the elastic value of L (defined by Hobbs) is initially
assumed. This will result in a non-zero slope at the centre of the buckle, i.e. boundary
condition (i) is violated. The iteration on L is therefore performed until the slope at the centre
of the buckle is zero. Since boundary conditions (i) and (ii) were used in the derivation of the
governing differential equation, satisfaction of (i) implies satisfaction of (ii). The iteration is
performed using the length correction:-
0
L  … D1.28
0
Iteration continues until the slope is less than the acceptable tolerance. The scheme
becomes inefficient as the solution is approached, and could be replaced by a more
sophisticated scheme, for example based on Newton’s method.
D1.3.2 Iteration of Buckle Force

Once the value of L is found the validity of the solution is evaluated by considering the axial
compatibility. The axial feed-in to the buckle is given by:-
P P   W  L2s
 0   Ls  … D1.29
 EA  2  EA
Where P0 is the fully constrained force, EA is the pipe axial stiffness, μ is the coefficient of
axial friction and Ls is the slip length, which is defined as:-
X
Ls   LB … D1.30
2
Where X is the buckle spacing. An alternative form of Equation D1.29 is required for an
isolated buckle in a very long pipeline. This is trivial to implement, but is not considered
here.
The axial displacement within the buckle is:-
1 L / 2 2 
L2
 P0  P 
uLB      LB     y  dx   y22  dz
 … D1.31a
 EA  2  0
 0 
Report No: 5087471/ 01 / A

The integration for the central lobe must be undertaken numerically, however, the integration
over the secondary lobe can be performed analytically:-
L
1 2 2
  y2  dz  9.5535  10  5  2  L72 … D1.31b
2 0
The error in the solution can be written as:-
du    uLB  … D1.32
A Newton iteration scheme is employed:-
Pj 1  Pj 

du j  Pj  Pj 1  … D1.33
du j  du j 1
Iteration continues until the error in the feed-in is less than an acceptable tolerance.
D1.3.3 Iteration Scheme

A nested iteration scheme is employed, which adopts the following sequence.
 Step 1: Calculate the Hobbs elastic solution;
 Step 2: Solve equation D1.9 for the elastic buckle force and elastic buckle length;
 Step 3: Iterate on L until slope at centre is zero;
 Step 4: Check axial compatibility condition, update value of force in buckle;
 Step 5: repeat steps 3 and 4 until solution.
D1.4 MODEL VALIDATION

The model developed is validated against FEA models for four typical design examples; two
single pipelines and two PIP systems as presented in Table 1.
Report No: 5087471/ 01 / A

Case 1 2 3 4
Type Single Flowline Single Trunkline PIP PIP
Material X65 X65 25Cr X65
Pipe Seamless UOE Seamless Seamless
Size 8-inch 24-inch 8-inch 16-inch

Flowline
Dfo (mm) 219.1 609.6 219.1 406.4
WT (mm) 15.9 20.6 12.7 21.4
CA (mm) 6 3 0 5
Insulation 30 mm PP Concrete 50mm PUF PUF
OHTC (W/m2K) 5.8 43 1.1 0.6
Material X65 X65
Pipe HFI HFI

Jacket
Size 12-inch 22-inch
Djo (mm) 323.9 558.8
WT (mm) 22.2 15.9
ODoc (mm) 280.1 710.6 329.9 564.8
Pressure (MPa) 20 10 30 20
Temperature (°C) 200 100 200 110
Water depth (m) 1500 500 2000 200
Seabed Clay Sand Clay Sand

Table 1 Definition of Design Examples
The design examples are chosen to represent the whole range of parameters of interest.
The examples consider shallow and deep-water projects (to 2000 m) and modest to high
design temperatures (100ºC to 200ºC).
Report No: 5087471/ 01 / A

The flowline wall thicknesses are calculated in accordance with OS-F101. The jacket wall
thicknesses are calculated in accordance with the collapse requirements of OS-F101.
However, the minimum jacket wall thickness is then increased to provide some bending
resistancei. For example the jacket wall thickness of design case 3 has been increased from
the collapse requirement of 15.9 mm to 22.2 mm. The PIP systems are assumed to be fully
bonded systems.
The FEA models all employ the following assumptions:-
 The seabed is modelled using a Coloumb friction representation;
 The axial friction coefficient is 0.4 and the lateral friction coefficient is 0.7.
 Pipe elements are employed;
 The models are initially straight. Buckling is initiated employing a small trigger force.
The force is removed during initial heat-up, such that it is zero when 50% of the
temperature has been applied;
 The pipe material is elastic-perfectly plastic, with constant temperature properties.
D1.4.1 Design Case 1: Deep Water Flowline

The validation is performed using the parameters outlined in Section D1.4 for a buckle
spacing of 2 km. Results are presented for three temperatures during the heat-up; θ=100ºC,
141ºC and 200ºC.
D1.4.1.1 Displaced Shape

The displaced shape developed is presented in Figure 5 to Figure 7.
8 P=20 MPa
7 θ=100ºC
6
5
Mathcad
4
Abaqus
Offset (m)
3
2
1
0
-10.88 0.90 0.92 0.94 0.96 0.98 1.00
-2
Distance along pipe (km)
-3
Figure 5 Case 1: Displaced Shape at 100ºC
i
The local buckling of the jacket pipe is a critical design issue within the buckle. This failure mode is
an interaction between the external pressure and the imposed bending. If the jacket is sized for
collapse, then there is no reserve left to accommodate significant bending. Consequently, an
increase in the basic thickness is required.
Report No: 5087471/ 01 / A

10 P=20 MPa
θ=141ºC
8
6
Mathcad
Offset (m)
4 Abaqus
0
0.88 0.90 0.92 0.94 0.96 0.98 1.00
-2
-4
12 P=20 MPa
θ=200ºC
10
8
6 Mathcad
Offset (m)
4 Abaqus
2
0
-20.88 0.90 0.92 0.94 0.96 0.98 1.00
-4
-6

The shape of the buckle is excellent, however the analytic model overestimates the
magnitude of the buckle; this is a maximum of 12% at 200ºC.
D1.4.1.2 Curvature
The curvature developed in the model is presented in Figure 8 and Figure 9.
Report No: 5087471/ 01 / A

0.02
0.01
Curvature (1/m)
0
0.90 0.92 0.94 0.96 0.98 1.00
-0.01
-0.02
200 C MCAD 141 C MCAD 100 C MCAD

-0.03
200 C Abaqus 141 C Abaqus 100 C Abaqus
-0.04
Figure 8 Case 1: Curvature over Buckle Length

0
0.98 0.99 0.99 1.00 1.00
-0.005
-0.01
Curvature (1/m)
-0.015
-0.02
-0.025
-0.03
-0.035
-0.04
Figure 9 Case 1: Curvature over Buckle Crown

The figures show that the curvature prediction of the analytic model is excellent. At the
crown the error in curvature is negligible, below 1%. In addition, the model is capable of
picking up the localisation of curvature that occurs as the plasticity becomes significant, i.e.
as the plastic hinge begins to form (see response at 200ºC).
The difference is greater in the secondary lobe; higher curvatures are predicted at the peak
of the secondary lobe and the start of the buckle is less clear in the Abaqus model (this is
due to the point load required to support a true mode 3 buckle). These differences will not
effect the integrity assessment, since this is based on results at the crown, which are
excellent, but are the source of the errors in the displaced shape observed above.
D1.4.1.3 Moment
The moment developed in the model is presented in Figure 10 and Figure 11.
Report No: 5087471/ 01 / A

200
150
100
50
Moment (kNm)
-500.90 0.92 0.94 0.96 0.98 1.00

-100 Distance along pipe (km)
-150
-200
-250
-300
Figure 10 Case 1: Moment over Buckle Length
0.98 0.99 0.99 1.00 1.00

-50
Moment (kNm)
-100
-150
-200
-250
-300
Figure 11 Case 1: Moment over Buckle Crown

The results follow a similar trend to the curvature response; the maximum error in moment at
the crown is below 0.5%.
D1.4.2 Design Case 2: Deep water Trunk Line

The validation is performed using the parameters outlined in Section D1.4. As the pipeline
exhibits a very severe buckling response, the comparison is performed for a buckle spacing
of 1 km. Results are presented for the maximum temperature, θ=100ºC.
The displaced shape is presented in Figure 12.
Report No: 5087471/ 01 / A

4
Analytic
3 Abaqus
Offset (m)
2 P=10 MPa
θ=100ºC
1
0
0.36 0.38 0.40 0.42 0.44 0.46 0.48 0.50
-1
-2
Figure 12 Case 2: Displaced Shape

The figure shows excellent agreement between the analytic model and the FEA; the
difference in maximum offset is 1%. The curvature developed is presented in Figure 13.
0.006
0.002
Curvature (1/m)
-0.0020.36 0.38 0.40 0.42 0.44 0.46 0.48 0.50
-0.006
Analytic
-0.01 Abaqus
-0.014
-0.018
Figure 13 Case 2: Curvature

The comparison between the models is reasonable except at the crown of the buckle, where
the analytic model overestimates the curvature by 17%. This is due to the very flat moment-
curvature response associated with the pipe, as illustrated by the moment comparison in
Figure 14.
Report No: 5087471/ 01 / A

1500
500
Moment (kNm)
-5000.36 0.38 0.40 0.42 0.44 0.46 0.48 0.50
-1500 Analytic
Abaqus
-2500
-3500
Figure 14 Case 2: Moment

The moment is in excellent agreement; the difference in the calculated moment at the crown
is less than 3%. This small difference in moment results in the very large curvature
difference.
D1.4.3 Design Case 3: Deep Water PIP

The validation is performed using the parameters outlined in Section 2 for a buckle spacing
of 2 km. The yield stress of the flowline is 550 MPa; the yield stress of the sleeve is 448
MPa. Results are presented for three temperatures during the heat-up; θ=118ºC, 141ºC and
200ºC.
D1.4.3.1 Displaced Shape

The displaced shape developed is presented in Figure 15 to Figure 17.
4
Abaqus
3
Analytic
Offset (m)
2
P=30 MPa
1 θ=118ºC
0
0.90 0.92 0.94 0.96 0.98 1.00
-1
Report No: 5087471/ 01 / A

4
Abaqus
3
Analytic
Offset (m)
P=30 MPa
2
θ=141ºC
1
0
0.90 0.92 0.94 0.96 0.98 1.00
-1
-2
5 Abaqus
4
Analytic
Offset (m)
3
P=30 MPa
2 θ=200ºC
0
0.90 0.92 0.94 0.96 0.98 1.00
-1
-2

The shape of the buckle is excellent; the maximum difference in the magnitude of the buckle
is 3%.
D1.4.3.2 Curvature
The curvature developed in the model is presented in Figure 18 and Figure 19.
Report No: 5087471/ 01 / A

0.015
0.01
0.005
Curvature (1/m)
0
0.90 0.92 0.94 0.96 0.98 1.00
-0.005
-0.01
-0.015
200 C Analytic 141 C Analytic 118 C Analytic
-0.02 200 C Abaqus 141 C Abaqus 118 C Abaqus
Figure 18 Case 3: Curvature over Buckle Length

0
-0.0020.98 0.99 0.99 1.00 1.00
-0.004
-0.006
Curvature (1/m)
-0.008
-0.01
-0.012
-0.014
-0.016
-0.018 200 C Analytic 141 C Analytic 118 C Analytic
-0.02 200 C Abaqus 141 C Abaqus 118 C Abaqus
Figure 19 Case 3: Curvature over Buckle Crown

The figures show that the curvature prediction of the mathcad model is good. At the crown
the maximum difference in curvature is 8%, with the analytic model providing consistently
conservative results. In addition, the model is capable of picking up the localisation of
curvature that occurs as the plasticity becomes significant, i.e. as the plastic hinge begins to
form (see response at 200ºC).
There is little difference in the secondary lobe, which is reflected in the excellent displaced
shape correlation observed above.
D1.4.3.3 Moment
The moment developed in the model is presented Figure 20 to Figure 22.
Report No: 5087471/ 01 / A

600
400
200
Total Moment (kNm)
0.90 0.92 0.94 0.96 0.98 1.00

-200
-400
-600
-800 118 C Abaqus 141 C Abaqus 200 C Abaqus
-1000 200 C Analytic 141 C Analytic 118 C Analytic
Figure 20 Case 3: Total Moment over Buckle Length
100
50
Flowline Moment (kNm)
0.90 0.92 0.94 0.96 0.98 1.00

-50
-100
Figure 21 Case 3: Flowline Moment over Buckle Length
600
400
Jacket Moment (kNm)
200
0.90 0.92 0.94 0.96 0.98 1.00

-400
-600 118 C Abaqus 141 C Abaqus 200 C Abaqus
Figure 22 Case 3: Jacket Moment over Buckle Length
Report No: 5087471/ 01 / A

The analytic model is in good agreement with the Abaqus results; the difference in the
maximum moment in either pipe is below 6%.
D1.4.3.4 Axial Stress

The axial stress developed in the buckle at maximum temperature is presented in Figure 23
and Figure 24.
150 Abaqus Extrados Abaqus Intrados

Analytic Extrados Analytic Intrados
Flowline Axial Stress (MPa)
50
-50 900 920 940 960 980 1000

Distance (m)
-150
-250
-350
P=30 MPa
-450 θ=200ºC
Figure 23 Case 3: Axial Stress in Flowline at 200ºC
400
300
Jacket Axial Stress (MPa)
200
100
Distance (m)
-100 900 920 940 960 980 1000

-200
-300
Abaqus Extrados Abaqus Intrados
-400
-500
Figure 24 Case 3: Axial Stress in Jacket at 200ºC

There is good agreement in the axial stress distributions. The analytic model predicts a
slightly higher stress at the crown when the material is still elastic.
D1.4.4 Design Case 4: Shallow Water PIP

spacing of 2 km. The yield stress of the (X65) flowline is de-rated to 420 MPa; the yield
stress of the sleeve is 448 MPa. Results are presented for the maximum design
temperature of 110ºC.
The displaced shape is presented in Figure 25.
Report No: 5087471/ 01 / A

8
7
Analytic P=20 MPa
6 θ=110ºC
5 Abaqus
Offset (m)
4
3
2
1
0
-10.84 0.86 0.88 0.90 0.92 0.94 0.96 0.98 1.00
-3

maximum offset is 2%. The curvature developed in the buckle is presented in Figure 26.
0.006
0.004 Analytic
0.002 Abaqus
Curvature (m-1)
0
0.84 0.86 0.88 0.90 0.92 0.94 0.96 0.98 1.00
-0.002
-0.004
-0.006 P=20 MPa

θ=110ºC
-0.008
-0.01
Figure 26 Case 4: Curvature

maximum curvature is approximately 4%. The moments developed in the buckle are
presented in Figure 27 to Figure 29.
Report No: 5087471/ 01 / A

2000 Analytic
1500
Total Moment (kNm)
1000 Abaqus
500
-5000.84 0.86 0.88 0.90 0.92 0.94 0.96 0.98 1.00

-1000
P=20 MPa
-1500
θ=110ºC
-2000
-2500
Figure 27 Case 4: Total Moment
600
P=20 MPa
400 θ=110ºC
Flowline Moment (kNm)
200
0.84 0.86 0.88 0.90 0.92 0.94 0.96 0.98 1.00

-200
-400
Analytic
-600
Abaqus
-800
Figure 28 Case 4: Flowline Moment
1500
P=20 MPa
1000
θ=110ºC
Jacket Moment (kNm)
500
0.84 0.86 0.88 0.90 0.92 0.94 0.96 0.98 1.00

-500
-1000
Analytic
-1500 Abaqus
-2000
Figure 29 Case 4: Jacket Moment
Report No: 5087471/ 01 / A

maximum moment in either pipe is below 3%. The axial stress developed in the buckle is
presented in Figure 30 and Figure 31.
350 Analytic Extrados Analytic Intrados

Flowline Axial Stress (MPa)
250
150
50 Distance (m)
-50 `
900 920 940 960 980 1000
-150
-250
-350 P=20 MPa

θ=110ºC
Figure 30 Case 4: Axial Stress in Flowline
500
400
Jacket Axial Stress (MPa)
300
200
100
Distance (m)
-100 900 920 940 960 980 1000

-200
-300 Analytic Extrados Analytic Intrados
-400 Abaqus Extrados Abaqus Intrados
Figure 31 Case 4: Axial Stress in Jacket

The model accurately predicts the stress developed in both pipes. The yield condition is
identical in the analytic and FE models, so the stresses are exact when yielding occurs. The
analytic model predicts a slightly higher stress at the crown (difference approximately 5%)
when the material is still elastic.
Report No: 5087471/ 01 / A

D2 CYCLIC ELASTIC LATERAL BUCKLING MODEL

D2.1 MODEL FORMULATION
D2.1.1 Basic Assumptions

The model makes the same key assumptions employed by Hobbs. These are:-
 The lateral friction is fully mobilised over the whole of the buckled pipe;
 The axial friction is fully mobilised on the slip zones;
 The analysis uses linear differential equations which are only valid for small slopes;
 The axial force is constant within a buckle;
In order to evaluate the behaviour of the system under cyclic conditions an incremental
solution is required to follow the deformation from the initially displaced shape.
D2.1.2 Development of Governing Equations

On unload the pipe will attempt to return to its initial configuration. However, it will be
prevented from doing so by lateral friction within the buckled region and axial friction along
the feed-in length. The resulting response is presented in Figure 1.
y
Load
Unload
Stationary Point
Figure 32 Mode III Buckle Deformed Shape

The figure shows how pipe will straighten out. As a result of this straightening process, the
response is split into two areas:-
 Near the crown the pipe moves back towards its original location;
 Near the end of the buckle the pipe moves away from its original location.
A (lateral) stationary point separates these two zones. Based upon this basic response, the
loads imposed on the unload configuration is illustrated in Figure 33.
Report No: 5087471/ 01 / A

Unload
Stationary Point
q
q
y L/2
μW
x
S
Figure 33 Free-Body on unload

As a result of the contraction, the force in the buckle, S is tensile; the convention employed
here is tension positive. The lateral restraint is q (=μL·W) and the length of the buckle on
unload is LBu. The length of the central portion, which contracts towards the original axis is
termed L.
Since the model is based on elastic response, the governing differential equations are well
known, these are:-
y1iv  2  y1    0 for x ≤ L/2 … D2.1
yiv2  2  y2    0 for x > L/2 … D2.2
where:-
S
2  … D2.2a
EI
q
 … D2.2b
EI
D2.1.3 Boundary Conditions

The boundary conditions for the problem are, at the centre of the buckle, x=0:-
y1 0  0 (Symmetry) …i
y10  0 (Zero reaction force at centre) … ii
where ‘ indicates differentiation with respect to x. At x=L/2 we get:-
L L
y1   y 2   (Continuity of displacement) … iii
 
2 2
L L
y1    y2   (Continuity of slope) … iv
2 2
Report No: 5087471/ 01 / A

L L
y1   y2   (Continuity of curvature) …v
 
2 2
L L
y1   y2   (Continuity of shear force) … vi
2 2
At x=LBu we get:-
YLu for LBu  LB

y 2 LBu   (Displacement at end of lobe) … vii
0 for LBu  LB
 for LBu  LB
YLu
y2 LBu   (Slope at end of lobe) … viii
0 for LBu  LB
 for LBu  LB
YLu
y2 LBu   (Curvature at end of lobe) … ix
0 for LBu  LB
L 
y1   YLu (Displacement at stationary point) … vii
2
 is the slope under load at x=LBu

Where YLu is the displacement under load at x=LBu, YLu
 is the curvature under load at x=LBu
and YLu
D2.1.4 Solution of Equations

The general solution to equations D2.1 and D2.2 is:-
  x2
y1  A1  cosh  x   A 2  sinh  x   A 3    x  A 4   YLu … D2.3a
2  2
  x2
y 2  A 5  cosh  x   A 6  sinh  x   A 7    x  A 8   YLu … D2.3b
2  2
The boundary conditions (i) to (viii) can be used to solve for the eight unknown coefficients.
This yields the solution:-
2    L4  x 2 
y1   B  cosh x   B   YLu … D2.4a
L4  4 
1 4
2    L4  x 2 
y2   B  cosh x   B  sinh x   B 
x
 B   YLu … D2.4b
L4  4 
5 6 7 8
Lu
Where the coefficients are:-
  L  3
L  LBu  2  coshLBu   sinh   
YLu
 2  
 … D2.5a
2  sinhLBu 
Report No: 5087471/ 01 / A

 L 
B1    cosh  … D2.5b
 2 
 L  L L2 L2
B 4    coshLBu   sinh   sinhLBu    LBu  Bu  1 … D2.5c
 2  2 4 8
B5   … D2.5d
 L 
B 6  sinh   … D2.5e
 2 
 L
B7  … D2.5f
2
 L  L L 2
B8    coshLBu   sinh   sinhLBu    LBu  Bu … D2.5g
 2  2 4
The two remaining boundary conditions, ix and x, are used to obtain the relationship
between the overall buckle length, LB, the central length, L, and the unload buckle length,
LBu:-
 L  2
  coshLBu   sinh   sinhLBu   
1

 YLu … D2.6a
 2  2 2
 L   L  L
  cosh   cosh 2     L Bu 
LBu 2  3  L 2  1
 2   2  2 4 16 2
… D2.6b
 4  Y  
   YL 2  YLu  Lu 
2   2 
Equations D2.6a and D2.6b can be solved for L and LBu.
D2.1.5 Force in Buckle

There is one unknown remaining, the force in the buckle on unload. This must be found by
considering the axial compatibility. The effective force history under the cyclic is illustrated in
Figure 34.
Load
Effective Force
Buckle Unload
P
Ls LB
Report No: 5087471/ 01 / A

Figure 34 Effective Axial Force Over Load Cycle

On unload, the axial feed-in to the buckle is given by:-
   W  L2s  S  S0  
      L s  … D2.7
 2  EA  EA 
 
Where S0 is the fully constrained force, EA is the pipe axial stiffness, μ is the coefficient of
axial friction and Ls is the slip length, which is defined as:-
X
Ls   LBu … D2.8
2
Where X is the buckle spacing. An alternative form of Equation D2.7 is required for an
isolated buckle in a very long pipeline. In addition, this model is only applicable if the change
in force at the anchor points (extreme left and right of Figure 34) is less than the change in
fully constrained force. This will be true unless the pipeline is lowly loaded or the buckle
spacing is very high. These modifications are trivial to implement within the model, but are
not considered here.
The axial displacement within the buckle is:-
1 L / 2 2 
LB
 S  S0 
uLBu      LBu     y1  dx   y22  dx
 … D2.9a
 EA  2  0
 L/2 
Substituting equations D2.4 to D2.6b yields,
 S  S0  2
uLBu      LBu  7  f L, LB  … D2.9b
 EA  
Where
 2  coshLBu   sinhLBu   LBu 

  L  
 4  sinh   2  L  LBu   coshLBu   sinhLBu 
  2  
  
  2  sinh L   cosh2 LBu   L  cosh L  
 
  2   2  
 L 
f L, LBu    coshLBu   sinhLBu   LB  L   sinh2    … D2.9c
 2 
 L    L  
 sinh    2  LBu  L   sinhLBu   2  coshLBu   5  cosh   
 2    2 
L2  LBu L  L2Bu L3Bu L3 3  L
    
2 2 6 8 2
Using equations D2.7 and D2.9b yields the compatibility of axial displacements
requirement:-
Report No: 5087471/ 01 / A

  W  L2s  S  S0   LBu  LB  2

      7  f L, LB   0 … D2.10
2  EA  EA     
This equation can be solved to yield the force in the buckle, S.
D2.1.6 Solution Procedure

There is no closed form solution to equation D2.10. In order to solve the problem a simple
iteration scheme is adopted. Denoting the error in equation D2.10 by ε we can define the
iteration scheme as:-
S j 1  S j 
 j  S j1   j1  S j  … D2.11
 j 1   j
Two values of S are required to start the iteration; these are taken to be the magnitude of the
compressive force in the buckle on load and twice this value. Iteration continues until the
error in the feed-in is less than an acceptable tolerance.
D2.1.7 Limitations
The model has the following limitations:-
 It is only applicable for a buckle between virtual anchors;
 The force in the buckle on unload must be tensile (this may not be true for small cycles);
 The change in force at the virtual anchors must be less than the change in fully
constrained force.
D2.2 VALIDATION
The model developed is validated against FEA models for the four design cases developed.
The FEA models all employ the following assumptions:-
 The seabed is modelled using a Coulomb friction representation;
 The axial friction coefficient is 0.4 and the lateral friction coefficient is 0.7.
 Pipe elements are employed;
 The models are initially straight. Buckling is initiated employing a small trigger force.
The force is removed during initial heat-up, such that it is zero when 50% of the
temperature has been applied;
Since the analytic model is wholly elastic, the validation is performed against an elastic FEA.
However, to illustrate the errors associated with plasticity, the stress range is also compared
with an elastic-perfectly plastic FEA.
D2.2.1 Design Case 1: Deep Water Flowline – Full Shutdown

spacing of 2 km. The displaced shape over the cycle is presented in Figure 35.
Report No: 5087471/ 01 / A

12
10
Abaqus Load Abaqus Unload
8
Analytic Load Analytic Unload
6
Offset (m)
-20.86 0.88 0.90 0.92 0.94 0.96 0.98 1.00
-4

The initial displacement is overestimated by the Hobbs model and the unload condition is
underestimated. However, the cyclic moment is reasonable, as illustrated in Figure 36.
7
Offset Range (m)
5 Abaqus
Analytic
3
-1
0.86 0.88 0.90 0.92 0.94 0.96 0.98 1.00
-3
-5
Figure 36 Case 1: Offset Range

The resulting moments are in good agreement, as illustrated in Figure 37.
Report No: 5087471/ 01 / A

200
150
100
50
Moment (kNm)
-50 860 880 900 920 940 960 980 1000

-100
Distance along pipe (m)
-150
-200
-250 Analytic Load Analytic Unload
-300 Abaqus load Abaqus Unload

The resulting stress ranges are compared in Figure 38.
600
400
Axial Stress Range (MPa)
200
0
860 880 900 920 940 960 980 1000
-400
Figure 38 Case 1: Stress Range

The stress range is in good agreement; the analytical model overestimates at the stress
range by approximately 8%. The stress range is compared with the values developed in an
elastic-perfectly plastic FEA analysis in Figure 39.
Report No: 5087471/ 01 / A

600
400
200
0
860 880 900 920 940 960 980 1000
-400
Figure 39 Case 1: Stress Range – Plasticity in FEA

The figure shows that the first load plasticity leads to an increase in the stress range at the
crown. However, the stress range is still well predicted by the analytic model, which now
underestimates the FEA by approximately 5%.
D2.2.2 Design Case 1: Deep Water Flowline – Half Shutdown

The displaced shape over the cycle is presented in Figure 40.
14
12
10 Abaqus Load
8 Abaqus Unload
6
Offset (m)
Analytic Load
4 Analytic Unload
2
0
-20.86 0.88 0.90 0.92 0.94 0.96 0.98 1.00
-4
-6

The initial displacement is overestimated by the Hobbs model and the unload condition is
underestimated. However, the cyclic moment is reasonable, as illustrated in Figure 41.
Report No: 5087471/ 01 / A

3 Abaqus
Analytic
2
Offset Range (m)
0
0.86 0.88 0.90 0.92 0.94 0.96 0.98 1.00
-1
-2
-3
Figure 41 Case 1: Offset Range

200
150
100
50
Moment (kNm)
-50 860 880 900 920 940 960 980 1000

-100
-150
-200

Report No: 5087471/ 01 / A

400
300
200
100
0
-100 860 880 900 920 940 960 980 1000
-200
-300
-400
Figure 43 Case 1: Stress Range

The stress range is in good agreement.
D2.2.3 Design Case 2: Deep water Trunk Line – Full Shutdown

The validation is performed using the parameters outlined in Section D1.4. As the pipeline
exhibits a very severe buckling response, the comparison is performed for a buckle spacing
of 1 km. The displaced shape over the cycle is presented in Figure 44.
7
6
5 Abaqus Load
4 Abaqus Unload
Offset (m)
3 Analytic Load
Analytic Unload
2
1
0
-10.38 0.40 0.42 0.44 0.46 0.48 0.50
-2

Report No: 5087471/ 01 / A

2500
1500
Moment (kNm)
500
-500 380 400 420 440 460 480 500
-2500

500
400
300
200
100
0
-100 380 400 420 440 460 480 500
-200
-300
Figure 46 Case 2: Axial Stress Range

The stress range is in good agreement; the analytical model overestimates the stress range
by approximately 5%. The stress range is compared with the values developed in an
elastic-perfectly plastic FEA analysis in Figure 47.
Report No: 5087471/ 01 / A

500
400
300
200
100
0
-100 380 400 420 440 460 480 500
-200
-300
Figure 47 Case 2: Stress Range – Plasticity in FEA

crown. However, the stress range is still well predicted by the analytic model, which now
underestimates the FEA by approximately 7%.
D2.2.4 Design Case 2: Deep water Trunk Line – Half Shutdown

The displaced shape over the cycle is presented in Figure 48.
7
6
5 Abaqus Load
4 Abaqus Unload
Offset (m)
3 Analytic Load
Analytic Unload
2
1
0
-10.38 0.40 0.42 0.44 0.46 0.48 0.50
-2

Report No: 5087471/ 01 / A

3000
2000
1000
Moment (kNm)
380 400 420 440 460 480 500

-1000
-2000
-3000

400
300
200
100
0
-100 380 400 420 440 460 480 500
-200
-300
-400
Figure 50 Case 2: Axial Stress Range

The stress range is in good agreement
D2.2.5 Design Case 3: Deep Water PIP – Full Shutdown

spacing of 2 km. The yield stress of the flowline is 550 MPa; the yield stress of the sleeve is
448 MPa. The displaced shape over the cycle is presented in Figure 51.
Report No: 5087471/ 01 / A

7
6
5 Abaqus Load
4 Abaqus Unload
Offset (m)
3 Analytic Load
Analytic Unload
2
1
0
-10.88 0.90 0.92 0.94 0.96 0.98 1.00
-2

600
400
200
Moment (kNm)
-200 880 900 920 940 960 980 1000
-400
-600

The resulting stress ranges are compared in Figure 53 and Figure 54.
Report No: 5087471/ 01 / A

50
Flowline Stress Range (MPa)
-50 860 880 900 920 940 960 980 1000
-150
-250
-450

-550
Figure 53 Case 3: Flowline Stress Range
500
400
Sleeve Stress Range (MPa)
300
200
100
0
-100 860 880 900 920 940 960 980 1000
-200
-300
-400
Figure 54 Case 3: Sleeve Stress Range

The stress range is in good agreement; the analytical model overestimates the flowline
stress range by approximately 3% and the sleeve stress range by approximately 6%. The
stress range is compared with the values developed in an elastic-perfectly plastic FEA
analysis in Figure 55 and Figure 56.
Report No: 5087471/ 01 / A

50 Distance Along Pipe

-50 880 900 920 940 960 980 1000
-150
-250
-350
-450
Figure 55 Case 3: Flowline Stress Range – Plasticity in FEA
500
400
300
200
100 distance (m)
0
-100 880 900 920 940 960 980 1000
-200
-300
Figure 56 Case 3: Sleeve Stress Range – Plasticity in FEA

crown. However, the stress range is still well predicted by the analytic model, which is
approximately 1% of the FEA for both pipes.
D2.2.6 Design Case 4: Shallow Water PIP – Full Shutdown

spacing of 2 km. The yield stress of the (X65) flowline is de-rated to 420 MPa; the yield
stress of the sleeve is 448 MPa. The displaced shape over the cycle is presented in Figure
57.
Report No: 5087471/ 01 / A

8
7
6
Abaqus Load
5
Abaqus Unload
4
Offset (m)
Analytic Load
3
Analytic Unload
2
1
0
-10.88 0.90 0.92 0.94 0.96 0.98 1.00
-2
-3

2000
1500
1000
500
Moment (kNm)
-500 880 900 920 940 960 980 1000

-1000
-1500
-2000

The resulting stress ranges are compared in Figure 59 and Figure 60.
Report No: 5087471/ 01 / A

200
100
0
860 880 900 920 940 960 980 1000
-100
-300

-400
Figure 59 Case 4: Flowline Stress Range
500
400
300
200
100
0
-100 860 880 900 920 940 960 980 1000
-200
-300
Figure 60 Case 4: Sleeve Stress Range

The stress range is in good agreement; the analytical model overestimates the flowline
stress range by approximately 3% and the sleeve stress range by approximately 4%. The
stress range is compared with the values developed in an elastic-perfectly plastic FEA
analysis in Figure 61 and Figure 62.
Report No: 5087471/ 01 / A

200
100
0
860 880 900 920 940 960 980 1000
-200
-300

-400
Figure 61 Case 4: Flowline Stress Range – Plasticity in FEA
500
400
300
200
100
-100 860 880 900 920 940 960 980 1000

-200
-300
Figure 62 Case 4: Sleeve Stress Range – Plasticity in FEA

crown. However, the stress range is still well predicted by the analytic model; the maximum
stress range is within 1% of the FEA for both pipes.
Report No: 5087471/ 01 / A

D3 PIPELINE WALKING
Simple models to address the driving mechanisms have been developed within the
SAFEBUCK JIP[D1,D2]; these are outlined here.
These models are applicable to single pipe flowlines. While the tension and seabed
slope models are applicable to PIP, the transient model is not due to the influence of
the outer pipe which may also be heated to above ambient temperatures. FEA is
required to assess of walking of PIP under thermal transient loading.
D3.1 THERMAL TRANSIENTS
D3.1.1 Fully mobilised Pipeline

These equations are valid for a fully mobile pipeline, or pipeline section. A pipeline is fully
mobilised if the following equality is true:-
S0
f  f*  .... D3. 1
L
In this equation L is the relevant length of pipe. In the absence of buckling it is the
overall pipe length of pipe between expansion spools. If buckling occurs it is the
distance between buckles.
The analysis involves a simple incremental calculation. To start the calculation an
appropriate number of increments, k, must be defined; a value of k=50 is typical. The
anchor point increment is then given by:-
L
xA  .... D3. 2
k
The values of xθ are then defined by:-
xA k0
x k  .... D3. 3
2  f  x A  L  x A 
k  xA  k  x A  x k 1 
2

f
k0
Where, f  EA    q .
The incremental walk is given by:-
f  x 2A
 2  k  1
L
x k 
EA 2
2
 L
f   x k  
f  x A  2  k  1
2
 2 L
w k   x k 1   x k .... D3. 4
EA 2  EA 2

  2
 
f  x 2A  2  k  1 f  L  x k 1  x k  x k  x k 1
2
 x k 1 
L
EA 2  EA 2
The total walk per thermal cycle is given by:-
Report No: 5087471/ 01 / A

k max
T   maxw k ,0 .... D3. 5
k 1
The analysis presented here is based upon a constant heat-up gradient of qθ (ºC/m).
If a non-linear profile is known, an average slope (over the mobilised hot end) should
be used in these equations.
The model is valid for f/fθ<1, but will yield reasonable result for higher values of this
ratio.
This mechanism will cause the pipeline to walk towards the cold end of the pipeline.
D3.1.2 Constrained Pipeline

If the pipeline is constrained (or partially constrained) then walking can persist only if the
thermal gradient is high. A reasonable condition for walking to stop is f/fθ > 1.
This is not a condition for walking to cease if the pipe is fully mobile.
For pipes with f/f*>1 and f/fθ < 1 there is no simple model available.
FEA is required to evaluate the problem in this case. However, unless f θ is much
greater than f, walking is unlikely to be a major issue for constrained pipelines.
D3.2 SEABED SLOPES

For seabed slopes along the pipeline, the incremental distance walked per cycle can be
calculated from:
 
S0  Ws  L  sin  f  L  cos Ws  L  tan .... D3. 6
EA  f
The equation is based upon a constant seabed slope. If the slope varies slowly then
the average slope over the pipeline can be employed. If the slope varies significantly
along the length, a more complex FE assessment is required.
The equation is independent of the sign of φ; for this mechanism walking will always occur
down hill.
If the pipeline reaches a position of full constraint on first load or cyclically then no walking
will occur for this mechanism.
D3.3 RISER TENSION

The incremental distance walked per cycle can be calculated from:
( S 0  S R  f  L )  S R
S 0  S R  f  L
EA  f
R  .... D3. 7
0 S 0  S R  f  L
For this mechanism walking will always occur towards the riser.
If the pipeline reaches a position of full constraint on first load or cyclically then no walking
will occur for this mechanism.
Report No: 5087471/ 01 / A

D3.4 LIQUID HOLD-UP

For pipelines transporting multi-phase fluid the walking mechanisms outlined above can be
accelerated due to liquid hold-up. During operation the fluid has an average contents
density, which varies slowly along the pipeline length. However, on shutdown the fluid will
separate into gas and liquid phases, with the liquid phase gathering at the bottom of any
slopes along the pipeline length. This leads to significant variation in contents density and
hence pipe submerged weight. On start-up this leads to further asymmetry in the force
profile and a modified walking response.
Simple models to evaluate this behaviour can be found in reference D2ii. Combinations of
Drivers
Equations D3. 1 to D3. 7 are for individual load components. If a pipeline has more than one
component, for example thermal transients and seabed slope, then the overall walking can
be assumed to be the sum of the individual components.
The components may add or subtract from one another. The summation should take
account of the walking direction.
D3.5 APPLICATION TO PIPELINES WITH LATERAL BUCKLING

A preliminary walking assessment may be undertaken using the analytic models outlined in
Section D3.1 to D3.4. These models consider the impact of different driving causes
individually. However, it is possible to combine these individual models to provide a good
prediction of pipeline walking in complex pipelines using the following approach:-
 Step 1: Divide the pipeline into short sections between buckles;
 Step 2: Calculate the walk on each section due to thermal transients;
 Step 3: Assume the anchor points are half way between each buckle (and between the
first and last buckle and the ends.
 Step 4: Calculate the feed-in to each buckle based on the steady state operating
conditions and the average slope over the anchor length. Add the thermal walk to the
calculated feed-in.
 Step 5: Calculate the force in each buckle due to calculated feed-in using the Hobbs
elastic solution[D3,D4].
 Step 6: Update the position of the anchor points based on the buckle forces.
 Step 7: Repeat Steps 4 to 6 until a stable solution is found.
 Step 8: Repeat Steps 3 to 7 for the unload condition.
 Step 9: Repeat Steps 2 to 8 for as many cycles as required.
This process will provide a good preliminary estimate of the walking behaviour. Sensitivity
cases to variations in key parameters such as thermal transients, number of buckles and
axial friction coefficient should be undertaken.
D3.6 NOMENCLATURE
The following notation is employed in the analytic walking models.
ii
Reference D2 was re-issued to OTC correcting Equation 11, and the corrected version is available
on the SAFEBUCK website. The definition of SC was corrected with the addition of a minus sign at
the start of each expression. This correction brought the definition of SC in line with the theory
presented for other walking mechanisms, which require the change in force between load and unload
to be defined as a negative value.
Report No: 5087471/ 01 / A

Roman Symbols
EA Pipeline axial stiffness N
f Axial pipe soil resistance N/m
fθ Thermal axial resistance, N/m
f* Mobilisation axial resistance N/m
k Integer counter in walking calculation -
kmax Maximum value of k in walking calculation -
L Total pipeline length or pipeline length between buckles m
qθ Gradient of the transient heat-up profile °C/m
So Full constrained effective axial force N
SR Riser tension applied to pipeline N
Ws Pipeline submerged weight N/m
xA Location of virtual anchor point in transient heat up m
xθ Location of thermal anchor point in transient heat up m
Greek Symbols
 Coefficient of thermal expansion -
ΔT Walk per cycle due to thermal transient loading m
Δφ Walk per cycle due to seabed slope m
ΔR Walk per cycle due to SCR loading m
ΔSo Change in fully constrained force due to change in operating conditions N
Δw Incremental walk m
φ Seabed slope radians
Report No: 5087471/ 01 / A

D4 REFERENCES
D1. M. Carr, F. Sinclair, D. Bruton. Pipeline Walking – Understanding the Field Layout
Challenges, and Analytical Solutions developed for the SAFEBUCK JIP. OTC 2006,
paper no. 17945.
D2. Bruton, D.A.S.; Sinclair, F.; Carr, M. Lessons Learned From Observing Walking of
Pipelines with Lateral Buckles, Including New Driving Mechanisms and Updated Analysis
Models. OTC 20750 (2010).
D3. Hobbs, R.E, In-service Buckling of Heated Pipelines. ASCE Journal of Transportation
Engineering, Vol. 110 No. 2, March 1984.
D4. Hobbs, R.E. Liang, F. Thermal Buckling of Pipelines Close to Restraints. OMAE 1989.
Report No: 5087471/ 01 / A

SAFEBUCK III / Safe Design of Pipelines with Lateral Buckling Page E1
Design Guideline: Appendix E
SAFEBUCK III
Design Guideline Appendix E
Buckle Initiation Techniques and Mitigation Measures
Report No: 5087471/01/A

TABLE of CONTENTS
E1 Introduction ................................................................................................. 3
E2 Buckle Initiation Techniques ...................................................................... 4
E2.1 Snake Lay ..................................................................................................... 4
E2.2 Local Weight Reduction: Distributed Bouyancy ............................................. 9
E2.3 Local Weight Reduction: Concrete Removal ............................................... 11
E2.4 Vertical Upset.............................................................................................. 11
E2.5 Zero Radius Bend ....................................................................................... 15
E2.6 Influence of Bathymetry............................................................................... 16
E2.7 Combined approaches ................................................................................ 17
E2.8 Other Issues................................................................................................ 17
E3 Mitigation Techniques for Lateral Buckling ............................................ 20
E3.1 Increase Axial friction .................................................................................. 20
E3.2 Decreased Lateral Friction .......................................................................... 20
E3.3 Hot lay......................................................................................................... 20
E3.4 Reduced design conditions ......................................................................... 20
E3.5 Influence of Pipeline Configuration .............................................................. 21
E4 Alternatives to Lateral Buckling ............................................................... 22
E4.1 In-line expansion spools .............................................................................. 22
E4.2 Trench and Bury ......................................................................................... 22
E5 REFERENCES ........................................................................................... 23
Report No: 5087471/01/A

E1 INTRODUCTION
The lateral buckling design strategy will often involve a buckle initiation technique, to
guarantee the formation of regular buckles. All of the buckle initiation techniques considered
to date attempt to impose relatively large OOS features at known locations. The design
methodology then relies on these “engineered” features (or triggers) being more reliable than
the inherent OOS.
There are a number of methods which have been employed or are currently proposed to
initiate buckling at a controlled spacing; each buckle initiation method is described in Section
E2 of this Appendix.
Although the various initiation techniques are designed to develop buckles consistently,
there is a practical limit as to how close together the buckles can be formed. As the distance
between intended buckle sites is reduced, the guarantee of formation reduces. If the buckle
initiation sites are too closely spaced, some sites will not develop buckles and the
Characteristic VAS will be greater than the distance between triggers. Nevertheless, placing
triggers at a closer spacing than that required to meet the tolerable VAS could reduce the
Characteristic VAS. For example, if triggers are spaced at half the Tolerable VAS spacing
and 50% (every other) of the triggers fail to initiate buckling, the design would still be
acceptable. The advantage of this approach is the potentially improved probability of
meeting the Tolerable VAS. If the Characteristic VAS is insufficient to ensure integrity of the
pipe, mitigation strategies could be adopted to reduce the severity of the design problem.
Some mitigation techniques are described in Section E3 of this Appendix and some design
alternatives are described in Section E4 of this Appendix.
Report No: 5087471/01/A

E2 BUCKLE INITIATION TECHNIQUES

The formation of the buckle is the key uncertainty in the lateral buckling design. The
buckling process is very sensitive to initial pipe imperfections and interaction with the
seabed; neither of these parameters is known with any certainty. To reduce this uncertainty
recent projects have installed engineered triggers, with the aim of increasing the reliability of
buckle formation. There are a number of methods which have been employed to date to
initiate buckling at a controlled spacing. These include:-
 Snake-lay;
 Vertical upset;
 Local weight reduction;
 Zero radius bends.
All of these techniques seek to improve the reliability of buckle formation by lowering the
buckle initiation force. The buckle initiation force is governed by (i) out-of-straightness
(OOS) features and (ii) the lateral breakout resistance. The initiation techniques work by
modifying one or both of these parameters.
There is a further advantage in the use of engineered buckle initiators: the buckle shape
generally becomes more benign with an initiator and the severity of the bending (for a given
design condition) is also reduced. The buckle shape is modified by the influence of the
original trigger shape (OOS feature) and in most cases by the reduction in lateral friction in
key regions of the buckle. Consequently, the integrity of the pipe within the buckle is
improved. This has the benefit of allowing an increase in the Tolerable VAS, which further
increases the reliability of buckle formation.
Over the last seven or eight years a large number of pipelines have been designed to
operate with lateral buckles. Therefore, there is an increasing volume of data which defines
the actual response of pipelines in this complex operating regime. This data is extremely
valuable in evaluating the suitability of the design methodologies available. One of the aims
of the Phase III data review is to collect and collate this information for as many projects as
possible.
E2.1 SNAKE LAY
E2.1.1 Background
In snake-lay, the pipeline is laid in a series of gentle curves, as illustrated in Figure 1.
Report No: 5087471/01/A

Pitch (Typically 2 to 5 km)

Lay bend radius
(typically 1500m)
Pipeline
Offset Plan
(typically view
100m)
Lay Centre-line
180
Offset (m)
175
170
165 Buckled
As-Laid
160
6.1 6.2 6.3 6.4 Lateral Buckling
Figure 1 Typical Snake Lay Configuration (Exaggerated Vertical Scale)

Figure 1 shows the key parameters in the snake lay configuration. The snake pitch is the
half-wavelength, defined as the distance between successive snake crowns. The offset is
the distance from the lay centre-line to the crown of the snake and the bend radius is the
average lay radius around the bend. The offset and bend radius define the arc length, which
also influences the buckle response.
The propensity for buckling is controlled by the bend radius of the snake. The radius is
designed to act as the buckle initiator, with the aim of developing a lateral buckle at some
point on the curve. This is a true localised lateral buckle (see insert in Figure 1), not a
benign expansion of the snake crown. Decreasing the bend radius increases the likelihood
of buckling (although this is ultimately limited by the minimum lay radius capability of the lay
vessel for the given water depth and seabed). However, it is also influenced by the snake
pitch and offset. The offset is usually defined to provide an arc length in the range 100 m to
300 m.
The frequency of buckling can be increased by reducing the snake pitch; although this is
limited by the potential for buckles to interact, localise, feed-through and coalesce. The
approach does give the designer some degree of control and by a careful choice of these
parameters the success of the snake lay can be maximised.
E2.1.2 Installation Issues

The minimum lay radius selected from design must lie within capability of the lay vessel for
the given water depth and seabed. Installation must be monitored by sufficiently accurate
survey to ensure that the minimum radius is achieved. The surveyed route corridor should
make allowance for snaking. The additional length of pipe required is generally not
significant, being in the order of a few pipe joints.
Report No: 5087471/01/A

E2.1.3 Approach Philosophy and Weakness
In the idealised situation the pipeline is laid in a series of gentle curvesi. The propensity for
buckling is controlled by the bend radius of the snake. The radius is designed to act as the
buckle initiator, with the aim of developing a lateral buckle at the some point on the curve.
The intention is that a buckle forms at the crown of each snake. This results in a situation
where the VAS is equal to the snake pitch.
This ideal situation is illustrated for a typical as-laid configuration in Figure 2.
400 P1 P2 P3 P4 P5 P6 P7
350
300
250
200
y (m)
150
100
X1 X2 X3 X4 X5 X6 X7
50
0
-50 0 2000 4000 6000 8000 10000 12000
x (m)
-100
Figure 2 Typical Real Snake Lay Route Profile

The figure shows the first seven route curves from as-laid data. Routing considerations
mean that the lay path is somewhat irregular. Figure 2 shows that the actual pitch varies
significantly (the pitch is labelled P1 to P7) to accommodate the design pitch within given
routing considerations. However, the longest straight runs are chosen to comply with the
pitch requirement (in this case 2 km). If the system forms a buckle at each snake, then the
resulting VAS would be in the range of values X1 to X7 illustrated in Figure 2. Clearly, in this
situation, the maximum VAS is less than or equal to the design pitch.
The arrangement illustrated in Figure 2 is the most common form of snake lay. An
alternative approach is to remove the straight sections and have a continuously curved
route. In this case there are no straight sections in the route; the exit tangent to one curve is
the entry tangent to the next; a section of the as-laid data is illustrated in Figure 3.
i
The as-laid curves have a relatively large radius. The range 1000 m to 2500 m is typical.
Report No: 5087471/01/A
40 P1 P2 P3 P4 P5 P6 P7
30
20
10
x (m)
0
y (m)
-10 0 1000 2000 3000 4000 5000
-20
-30
-40 X1 X2 X3 X4 X5 X6 X7
-50
Figure 3 Continuous Snake Lay Route Profile

If the system forms a buckle at each snake, then the resulting VAS would be X1 to X7
illustrated in Figure 3. Again, in this situation, the maximum VAS is less than or equal to the
design pitch.
However, there are significant challenges with the method and uncertainty over whether the
method is robust enough to work given practical variations in seabed response. The key
concern is that the buckling process may “miss” a snake crown, i.e. it may feed-through a
crown without buckling. The main reasons this can occur are:-
 Variation in lateral friction;
 Variation in axial friction;
 Inherent OOS;
 Bend radius is consistently greater than target;
 Snake pitch is too small.
For lateral friction, the concern is that the friction at the snake may be higher than
anticipated. In this case, the critical buckling force will be higher than anticipated and the
resistance may be large enough to suppress buckling. For axial friction, the concern is that
the friction is lower than anticipated. In this case, the force will not develop so quickly with
distance and there may be insufficient driving force at the snake to buckle the line.
The concern with the inherent OOS is that a buckle may form away from the snake crown.
This may then mean that the adjacent snake is missed, producing a VAS above the pitch. It
is very difficult to guarantee that the OOS imposed by the snake is always above the level of
OOS that may be inherent in the “straight” pipeline. This concern is magnified if the seabed
is relatively uneven, and significant vertical OOS is feasible.
The final two problems often result from of a poor understanding of the buckling process.
Since the aim of the design is to introduce a large number of buckles, the simplistic design
approach is to continually reduce the pitch between buckle initiators until the design criteria
are met. However, as the distance between intended buckle sites is reduced, the guarantee
of formation at each site reduces. Alternatively defining a bend radius that is at the limits of
what the lay vessel can achieve may be counter-productive. In either case buckling may not
occur at some of the intended sites – i.e. the robustness of the solution decreases.
Report No: 5087471/01/A

The effect of these undesirable behaviours is incorporated into the calculation of the
Characteristic VAS (Section 4 of the Design Guideline). In general the Characteristic VAS
will exceed the snake pitch. The design is not compromised so long as the unity checks at
the Characteristic VAS are acceptable.
An additional issue which should be considered with snake lay is the stability of the bends
associated with tension during shutdown. The tension in the line can increase with cycles,
particularly if pipeline walking occurs. If the tension becomes large enough to overcome the
lateral resistance provided by the pipe-soil interaction, the bends may pull out which in turn
can increase the expansion observed at the ends of the pipeline, or overload existing lateral
buckles (if they absorb the additional pipe).
E2.1.4 Experience with Snake Lay

Snake lay is the most common buckle initiation approach employed to date. The method is
becoming much better understood and the data obtained from operating pipelines is that
snake lay is a highly reliable buckle formation technique. The operational behaviour of
pipelines with data donated to the SAFEBUCK JIP is presented in Table 1.
Project Pipe Number of Bend Radius Number of VAS (km)

Length (km) Bends (m) Buckles Typical Max
A 51.5 25 1500 24 + 1 2.0 3.0
B 23 7 1000/1500 7 2.5 4.0
C 30 12 2500 11+2 2.5 4.0
Table 1 Observed Performance of Snake Lay Strategy

Project A is the Penguins project; a comprehensive description of the implementation of the
snake lay strategy for this pipeline is in the public domain[E1-E3]. Since start-up of the
Penguins pipeline in early 2003 two high precision surveys have been undertaken. Of the
twenty five 1,500 m radius bends, twenty four produced a lateral buckle. Thus only one
1500 m bend in the main route has failed to initiate a buckle; this is a 96% success rate.
There were two 2,000 m bends which did not buckle; however bucking was not anticipated
at these bends since these were near to each end of the pipeline. The observed VAS is
generally below 2 km, with a maximum value of 3 km. In addition to the planned buckles,
one rogue buckle formed in a straight section of pipe.
Project B employed eight bends (3 route bends and 5 snake lay initiator bends). As the final
route bend occurred within 800 m of the end of the pipeline and was not expected to cause
buckling, only seven were trigger bends. The bend radius employed was 1500 m (with a
single bend having a reduced radius of 1000 m). Buckling was observed at all seven of the
route bends, i.e. there was a 100% initiation success rate.
Project C employed 12 bends, but with a large bend radius (2500 m). Thirteen buckles did
initiate, but two of these occurred away from the bends, i.e. from inherent OOS. The
initiation success rate associated with the bends is therefore 92%. In this case the selected
bend radius may have been too high; the critical buckling load associated with the bends
was not sufficiently different to that associated with inherent OOS. Nevertheless, an
adequate number of buckles formed to meet the design objectives.
Although data from only three projects is currently available to the JIP, the technique is
known to have been used on many more projects all over the world. There has been no
report of the technique suffering a significant failure rate.
In addition, buckling has been observed at route bends in a number of projects which did not
employ the snake lay strategy. This supports the contention that large radius bends make a
highly reliable buckle initiator.
Report No: 5087471/01/A
E2.2 LOCAL WEIGHT REDUCTION: DISTRIBUTED BOUYANCY
E2.2.1 Background
In the distributed buoyancy method discrete lengths of typically 60m to 200m of the
pipeline are installed with additional buoyancy on them; these lengths are the intended
buckle initiation sites as illustrated in Figure 4.
Pipeline Plan view
Buckle spacing
Buoyancy added (typically 2 to 3km)
to reduce weight
(Typically < 100m)
Figure 4 Buckle Initiation Using Buoyancy

The buoyancy is chosen so that the operational submerged weight is a small fraction
(typically 10%) of the normal pipe submerged weight. A related approach is to use discrete
buoyancy (buoyancy bags) to aid buckle initiation[E4]. In this the buoyancy is simply an
initiation aid, and is removed once buckle formation has occurred.
The distributed buoyancy method works in two ways. Firstly, during lay the section of pipe
with buoyancy will be positively buoyant. This means that it tends to form vertical
imperfections as it is laid down. This is further exaggerated by the increase in outside
diameter and the reduction in vertical load associated with the buoyancy modules (tending to
reduce the embedment). These vertical imperfections, coupled with the effect of
hydrodynamic loading, will tend to produce a natural out-of-straightness at the chosen
location. Secondly, because the operational submerged weight is so low, the lateral
frictional restraint is reduced. Consequently, the buckle initiation force is also reduced.
These combined advantages mean that the pipe is more likely to buckle at the locations
where the buoyancy is applied.
There is a further advantage; since the lateral friction is reduced, the severity of the bending
(for a given design condition) is also reduced. Consequently, the integrity of the pipe within
the buckle is improved. This has the knock-on benefit of allowing an increase in the
Tolerable VAS, which further helps in achieving reliable buckle initiation.
Report No: 5087471/01/A

Buoyancy can be added to the pipe over short sections by increasing the thickness of
insulation coating, provided the installation vessel can handle the diameter change during
lay, or by offshore installation of buoyancy half-shells, which can pass through the lay ramp
or lay vessel stinger. The optimum method of installation is highly dependent upon the type
of lay vessel and pipe assembly method.
The length of buoyancy modules can be adjusted from a practical maximum length of
typically 200 m, equal to the length over which each buckle occurs, down to about 60 m,
equal to the length of the central lobe of the buckle. In some cases two (approximately
60 m) lengths of distributed buoyancy have been installed with a gap (of approximately
12 m) between them for installation reasons. This configuration was expected to form stable
mode-II buckles. However, in most cases Mode-I buckles formed in preference to the
planned Mode-II buckle.
Therefore, it is difficult to predict the mode shape of lateral buckles in practice and there
appears to be a tendency for buckles to transform from higher mode-shapes to lower mode
shapes under cyclic loading. Local effects, such as the presence of lateral OOS or lower (or
greater) than expected embedment at the transition regions, can have a significant influence
on the shape of a buckle. The consequences of such behaviour should be addressed in
design.
The buoyancy should remain on the pipe in operation and should be designed for the life of
the pipeline, including cyclic interaction with the seabed.
E2.2.3 Approach Philosophy and Weakness

The distributed buoyancy approach has the advantage that the submerged weight and the
length of pipe over which it is applied can be defined with some accuracy, to meet design
requirements. The diameter of the buoyancy can also be chosen to provide a known height
of vertical OOS to assist buckle initiation.
This method also reduces the importance (and uncertainty) of the pipe-soil interaction,
although understanding pipe-soil interaction is still important to predicting buckle behaviour,
particularly towards the end of the buoyancy where high vertical loads occur at the transition,
and at regions away from the buoyancy, where unplanned buckling may occur.
The application of buoyancy may cause local hydrodynamic instability, although global
instability is prevented by the normal-weight pipe to each side of the buoyed section.
Clearly, the OOS created by this instability will further assist buckle initiation. However,
there is a concern particularly in shallow water that the buckle would be unstable, flipping
from one side of the pipe centreline to the other. Such behaviour may compromise fatigue
performance by introducing stress cycles that are larger than those experienced in
operation. This possibility must be addressed in design.
Distributed buoyancy could result in pipeline spans that may be susceptible to vortex-
induced vibrations and may be a snagging (fishing) hazard. This is more likely to be an
issue in shallow water.
E2.2.4 Experience with Distributed Buoyancy

The distributed buoyancy method has been employed on at least two projects. The
installation experience of these projects is good. The distributed buoyancy was installed
accurately and without major delays to the lay process.
However, these projects have either not started up or not yet undertaken a sufficiently
detailed operational survey. Consequently, there is no operational performance data
available of distributed buoyancy as a method to control lateral buckling. The Phase III data
review hopes to provide additional information regarding the performance of buoyancy.
Report No: 5087471/01/A
E2.3 LOCAL WEIGHT REDUCTION: CONCRETE REMOVAL
To achieve local weight reduction for concrete coated pipelines, short sections of the
pipeline are installed with no concrete coating, or a reduced weight concrete coat. These
sections of pipe then behave in a similar manner to the sections of distributed buoyancy.
This is a very low cost design solution.
As well as improving the reliability of buckle initiation, the no-concrete approach has the
advantage of removing the strain concentration associated with the concrete coating field
joint. This can significantly improve the lateral buckling performance of the pipeline.
A section of reduced concrete thickness, or reduced concrete density, will still incur strain
localisation at field joints but the reduction in stiffness and lateral load should reduce strain
localisation.
E2.4 VERTICAL UPSET
E2.4.1 Background
Pipeline buckling can be initiated in either the horizontal or the vertical direction. If the
pipeline is laid on an uneven seabed, then an initial vertical movement is highly likely to
develop into a lateral buckle. This approach takes advantage of this fact. The technique
works by deliberately introducing significant vertical OOS at a number of points along the
pipeline. Two techniques have been employed:-
 Sleepers[E6,E7];
 Gravel dump berms[E8,E9].
In the gravel dump option, the sleeper is replaced by a gravel berm; otherwise, the concept
is identical, although the more gradual rise in slope towards a gravel berm can modify the
formation response, by encouraging upward displacement of the pipe before it falls over to
displace laterally. On a sleeper upheaval is immediately accompanied by lateral
displacement so that the pipe should not lift off the sleeper. Sleepers and gravel dump
options have already been adopted by specific projects.
The sleepers are installed prior to (or during) pipe lay at the appropriate spacing, as
illustrated in Figure 5.
Report No: 5087471/01/A

Sleepers (typically 2 to 3 joints

Pipeline of large diameter pipe
Elevation
view
Sleeper spacing
(typically 2 to 3km)
Pipeline
Sleeper
Figure 5 Buckle Initiation Using Sleepers

This approach has the attraction that the friction between the pipe and the sleeper can be
modified by coatings. In addition, the pipe to each side of the sleeper is suspended above
the seabed and therefore experiences no frictional restraint. This has the knock-on benefit
of reducing the cyclic load in the buckle allowing an increase in the design buckle spacing.
E2.4.2 Sleeper Design

There have been three distinct approaches to the design of sleepers. The simplest
approach is to employ a number of large diameter pipe joints welded together to form the
sleeper. The diameter of the sleeper pipe must be sufficient to provide the desired vertical
imperfection height following any anticipated settlement. The number of joints depends on
the length required, which in turn depends upon the anticipated buckle amplitude. Typically
two or three joints have been used. The joints are coated in order to provide wear
resistance and to control the friction between the sleepers and pipe. Thick insulation coating
on the pipe reduces the contact pressure, which improves wear resistance but increases the
friction coefficient at the interface[23]. This approach is illustrated in Figure 6.
Report No: 5087471/01/A

Sleeper
Flowline
Figure 6 Simple Sleeper Arrangement

A modification to this basic design approach is to incorporate a mud mat with the large
diameter pipe. This approach should reduce the uncertainty associated with the sleeper
settlement and allow larger vertical imperfections to be achieved. The final approach
involves a much more elaborate structure, employing concrete or steel. This approach
allows the incorporation of additional features, such as a vertical reaction post.
All structural steel sleepers are fitted with independent cathodic protection. Coating integrity
between pipe and sleepers is important to avoid potential corrosion issues. A few projects
have now carried out submerged seawater tests of coating integrity for this interface, using
the design contact loads and repeated cyclic loading. Some of the test data has been made
available to the SAFEBUCK JIP and will be reported separately.

Sleepers must be preinstalled prior to (or during) pipe lay, at intervals along the pipeline
route. Accuracy of installation is clearly important. The lay vessel is then required to lay the
pipe over the centre of the sleepers, with sufficient accuracy to allow for buckling in
operation. This will require touch-down monitoring and possibly special lay procedures to
ensure that the pipe is laid gently onto the sleepers to minimise sleeper embedment.
E2.4.4 Approach Philosophy

The height of the sleeper upset defines the buckle initiation force. The approach has the
advantage that the height of the vertical OOS can be defined and measured with some
accuracy. The method also reduces the importance (and uncertainty) of the pipe-soil
interaction, although pipe-soil interaction at the touch-down points is a significant influence
on buckle behaviour and buckle loads. Pipe-soil interaction also remains important at
regions away from the sleepers, where unplanned buckling may occur.
The vertical upset can result in significant pipeline spans; these are susceptible to vortex-
induced vibrations and may be a snagging (fishing) hazard. In many cases, this loading
limits the maximum allowable height of the upset. In extreme circumstances this concern
may prevent the application of the technique.
With this solution, the intention is to form a lateral buckle at each vertical upset. This results
in the VAS being equal to the upset spacing. However, initiation cannot be guaranteed and
the Characteristic VAS is sometimes greater than the sleeper spacing.
Report No: 5087471/01/A

E2.4.5 Experience with Sleepers
Operating data has been obtained from nine pipelines which have employed sleepers as a
buckle initiation technique. The performance of the projects is summarised in Table 2.2.
Pipe Number Number Number Buckle modesii Notes

of of of
sleepers buckles rogues
P1 4 4 0 All mode III
P2 14 12/14 4 First two initiators were not
surveyed. 12 out of 12 sleepers
produced buckles
P3 15 14/15 No data 14 symmetric Limited survey. One sleeper
may have failed or may have
formed a mode II buckle
P4 6 6 0 All mode I 5 mode I buckles formed as
mode II on hydrotest.
P5 9 9 1 All mode I 7 mode I buckles formed as
mode II on hydrotest.
P6 11 6 4 4 mode I,2 mode II 1 mode I started as a mode II on
hydrotest
P7 7 5 3 4 mode I,1 mode II
P8 4 4 2 2 mode I,2 mode II
P9 5 5 1 4 mode I,1 mode II 2 mode I started as a mode II on
hydrotest
Table 2.2 Outline Sleeper Information
The operational experience with vertical triggers is good; buckle formation at vertical triggers
appears to be very reliable. Only on two very similar pipelines (P6 and P7) have any triggers
failed to initiate a buckle. In this case the failure was due to a very high residual lay tension
(which was not anticipated in the design) compounded by the relatively low driving force in
the system due to the low operating temperature.
An unexpectedly high post-lay residual tension due to post-installation cool-down[E10] also
prevented buckling during hydrotest in another project[E6] (although buckling did
subsequently occur in operation). Clearly for relatively lowly loaded systems it is important
not to underestimate the potential magnitude of the residual post-installation tension. For
the other projects 57 triggers have been employed, all triggering a buckle.
The sleeper initiation technique develops both symmetric (mode I or III) and asymmetric
(mode II) buckles, Figure 7.
ii
The buckles are described by their mode shape, defined by the number of buckle lobes. When
pipes buckle on sleepers in a symmetric mode there is usually a large central lobe on the sleeper,
while the outer lobe displacements are quite small, as they embed into the soil. Such a shape is very
close to being mode-I and are therefore described as such.
Report No: 5087471/01/A
Figure 7 Typical Buckles (a) Mode I (b) Mode II

The initiation force for all these modes is very similar, and the actual mode is driven by the
local imperfection induced during pipe lay. It is possible that mode II buckles can develop
higher strains than a mode I buckle (because the location of peak bending can be in the
seabed). Thus the design must be robust to mode II buckles, or provision made to intervene
if mode II buckle occurs, so that at a suitable time the buckle can be pulled out to a mode
I/III.
A number of projects have enhanced the reliability of the buckle formation by reducing the
pipe submerged weight at the trigger. This has been achieved either permanently (by
removing concrete coating, incorporating a low density coating, or application of buoyancy
modules) or temporarily (by use of buoyancy bags). The permanent option also has the
benefit of reduced in-service loading in the buckle.
One project has employed a reduced length sleeper. The aim is for the pipe to buckle at the
sleeper, but fall to the seabed (to avoid VIV problems). This approach was successful;
however a mode II buckle would have compromised the strategy.
A number of the projects developed rogue buckles in addition to the engineered buckles.
This was not a problem for these projects as the integrity of rogue buckles was assessed as
part of the design in all cases. The rogue buckles did not prevent the triggers from working.
There has been one significant problem identified with the sleeper technique. One of the
projects reviewed has observed flow induced vibration at sleeper spans as a result of
slugging. The pipelines operate in a severe slugging regime, and the presence of the
sleepers appears to exacerbate slugging. Any project in which slugging is anticipated
should review the implications of cyclic loading (fatigue) due to density changes at the
anticipated slugging frequency before employing sleepers as the buckle initiation technique.
This is expected to be of particular concern for lighter pipes, where the weight of the
contents makes up a significant portion of the submerged weight.
E2.5 ZERO RADIUS BEND

A number of pipelines have employed the zero radius bend technique[E5]. In this technique a
vertical trigger is pre-installed on the seabed and the pipeline is initially laid straight towards
it. The straight lay continues until the pipe is firmly seated on the trigger. At this point lay
vessel moves laterally and changes heading by a small angle. The lay vessel then continues
laying in a straight line after the change in heading has been achieved.
Report No: 5087471/01/A

This lay procedure creates a bend in the pipeline route, concentrated within the span created
by the trigger. This is referred to as a (nominally) zero-radius bend, although in reality the
flexural rigidity of the pipeline means that the bend radius is non-zero. The bend radius
achievable is much smaller than can be implemented in a convention route curve, which
means that the critical buckling load is similarly reduced.
The lateral load imposed on the trigger during installation is small, and can easily be reacted
by incorporating a vertical post into the trigger.
E2.5.1 Experience with Zero Radius Bend

The zero radius bend has been employed on at least three projects to date, although
operating data has only been obtained from two of these. The Phase III data review is
expected to add several more instances to this database of operational performance.
For the first project eight triggers were employed approximately between 2.5 km and 4 km
apart. A buckle formed at each of the triggers. Seven of the buckles were mode I and one
was mode II. No rogue buckles where observed.
For the second project three triggers were employed. A buckle formed at each of the
triggers. All of the buckles were mode I. No rogue buckles where observed.
The zero radius bend technique has been employed to enhance the reliability of buckle
formation. This has been very successful (11 buckles at 11 triggers). Using this technique it
is possible to significantly reduce the buckling force, which facilitates closely spaced
buckling. Further, this technique is much less likely to produce mode II buckles.
E2.6 INFLUENCE OF BATHYMETRY

The work undertaken during the SAFEBUCK JIP is focused on the behaviour of pipelines on
relatively flat seabeds. However, the SAFEBUCK design approach is suitable for application
to projects with modest bathymetric OOS.
Significant bathymetric variation will have two main influences on the pipeline behaviour:-
 Buckle initiation;
 Spanning.
E2.6.1 Buckle Initiation

If the seabed exhibits significant vertical OOS, then this can act as a buckle initiator and can
potentially compromise the buckle formation at planned buckle sites. Buckling from
bathymetric features may involve initial movement in the vertical plane, but the final buckle
will form in the lateral plane unless there is very significant lateral restraint at the lift off
points. Even quite significant embedment may be insufficient to prevent lateral movement.
The pipeline is only likely to remain in the vertical plane if the pipeline were fully buried prior
to buckling.
The design can take advantage of the bathymetry. If the measured bathymetry is
incorporated in FE models of the pipeline then the most severe OOS features can be
identified. These features can then be incorporated into the buckle formation strategy as
planned initiation sites. The reliability of these sites can be evaluated with the buckle
formation analysis in the same way as engineered triggers (Guideline Section 4).
If the bathymetric features are modest, it may still be possible to take benefit from then. For
example, locating engineered triggers at the bathymetric OOS it may be possible to improve
the reliability of the trigger.
Report No: 5087471/01/A

E2.6.2 Spanning
If the bathymetry leads to pipeline free spans, then the expansion behaviour of the pipeline
will be modified. The pipeline expansion will feed in to the spanning sections. This will
absorb some of the feed-in that would otherwise drive lateral buckling, and is generally
beneficial. Very significant spans will modify the force profile in a similar way to lateral
buckling, i.e. the effective axial force will drop at the span as the pipe feeds-in to it and virtual
anchor points can be set up between spans (or between spans and lateral buckles). The
effect of this is to reduce the available driving force for lateral buckling and reduce the
Characteristic VAS should lateral buckling occur.
An FE analysis of the expansion process can be used to evaluate the effect of the free-
spanning on the development of driving force, which can be employed to modify the buckle
formation assessment.
E2.6.3 Areas of Concern

The response predicted in an FE analysis that incorporates bathymetric data will be sensitive
to the vertical pipe-soil interaction modelling. Specifically, the accuracy of the survey data
and the model of seabed vertical stiffness will influence the propensity for buckling and
spanning.
If the bathymetry has a beneficial behaviour on the pipeline behaviour, then the analysis
should ensure that the model of vertical interaction is reasonable and adequately
incorporates the uncertainty in the response.
E2.7 COMBINED APPROACHES

Sections E2.1 to E2.6 discuss the issues with individual buckle initiation techniques. It is
possible to combine different techniques in order to improve buckle formation reliability. For
example, the local weight reduction approach can easily be combined with the vertical upset
solution (and has been adopted on a number of pipelines). However, it cannot easily be
combined with the snake lay solution due to the difficulty of laying a light pipe around a
curve.
E2.8 OTHER ISSUES
E2.8.1 Pipe-Soil Modelling Issues

Engineered buckle initiators tend to reduce the influence of pipe-soil interaction on the
operating loads in the lateral buckle. However, understanding pipe-soil interaction is still
important to predicting buckle behaviour at any initiator, particularly at the touch-down points
and at regions away from the sleepers or reduced weight sections, where unplanned
buckling may occur.
Special consideration is required in zones where the local effective pipe weight (vertical
load) changes due to loss of contact with the seabed, for example at sleepers, or at the start
of a distributed buoyancy section. Such features, introduced deliberately to control lateral
buckling, require careful implementation of the pipe-soil response in FEA. A distinction must
be made between the pipe effective weight, W, and the local contact force between the pipe
and the soil, denoted V. The same approach is also applicable to spans, which can occur
along the line and may also act as buckle initiators.
Particular attention should be given to the definition of soil resistance at:
 The span touchdown points (TDP) to each side of a sleeper or span;
 The start and end of distributed buoyancy sections, where there is a sudden change in
diameter and effective pipe weight.
Report No: 5087471/01/A

The touchdown loads to each side of a sleeper are typically 60% greater than the effective
pipe weight. The pipe will experience significant additional lateral restraint in these
touchdown regions to each side of the sleeper, which strongly influences the load in the
lateral buckle. This effect is shown in Figure 8, which shows the plan, elevation and vertical
load across a lateral buckle over a sleeper, for a lower and upper bound value of lateral
resistance.
13 Low lateral resistance 13 High lateral resistance

11 11

9 9
7 7
5 5
3 3
1 1
-1 -1
850 900 950 1000 1050 1100 1150 850 900 950 1000 1050 1100 1150
-3 -3
-5 -5
1.0 1.0
Height (m)
Height (m)
0.5 0.5
0.0 0.0
850 900 950 1000 1050 1100 1150 850 900 950 1000 1050 1100 1150
-0.5 -0.5
1st Heat-up 1st Cool-down 2.5 1st Heat-up 1st Cool-down

2.5 2nd Heat-up 2nd Cool-down
2nd Heat-up 2nd Cool-down
2nd Heat-up 2nd Cool-down 2nd Heat-up 2nd Cool-down

1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.0
850 900 950 1000 1050 1100 1150 850 900 950 1000 1050 1100 1150
Distance along pipe [m] Distance along pipe [m]
Figure 8 Influence of touchdown load and lateral resistance on buckle shape

This example in Figure 8 includes a reduction in pipe weight by the addition of buoyancy
between 900m and 1100m, thus combining the vertical upset and local weight reduction
techniques. It is shown that the highest vertical load occurs within the lateral buckle and is a
significant influence on the shape of the lateral buckle. A design approach that provides an
increase in lateral resistance in these zones is necessary, usually by defining the lateral
resistance as a function of the local contact force between the pipe and the soil. This
approach leaves the FEA to take account of seabed stiffness and bending of the pipe in the
vertical plane. This approach may tend to slightly over-predict the breakout friction but that
assumption is generally considered to be conservative.
A realistic value of seabed stiffness is necessary in design to arrive at realistic contact
pressures. A very stiff seabed (hard contact) can over-estimate the contact pressures and
lateral resistance at the TDP to each side of the sleeper. A realistic seabed stiffness to be
adopted in design for soft seabeds is typically in the range of 15 kN/m/m to 50kN/m/m but
default seabed stiffness values in FEA are often much higher than this. The variation in
diameter that occurs at sections of distributed buoyancy should also be captured in FEA
models to ensure correct modelling of the pipe-soil contact regions.
E2.8.2 Pipeline Trigger Zones

If engineered triggers are employed then it is possible to modify the design of the pipeline at
the triggers to improve the overall response. For example, the design parameters that could
be modified include:-
Report No: 5087471/01/A

 Increased WT pipe in the vicinity of the triggers;
 Pipe sorting;
 Removal of concrete coating;
 Modified field joint;
 Improved tolerances at triggers;
 Machined ends; stiffness mismatch
These modifications can improve the local buckling or fatigue performance of the engineered
lateral buckles.
However, this approach does not improve the response of rogue buckles, which must still be
assessed in accordance with the Design Guideline.
Report No: 5087471/01/A

E3 MITIGATION TECHNIQUES FOR LATERAL BUCKLING

Techniques which can mitigate the severity of the loads within a lateral buckle include:-
 Increase axial friction;
 Decreased lateral friction;
 Hot lay – installing the pipeline at temperatures above ambient;
 Reduced design conditions;
 Modify the pipe configuration.
E3.1 INCREASE AXIAL FRICTION

Increasing the axial friction reduces the degree of feed-in to a buckle; it essentially has the
same effect as reducing the VAS. One way of achieving this is to rock dump the straight
sections of the pipeline. This could be undertaken before start-up between the straight
sections. Alternatively, rock is perhaps more usefully applied post start-up, once the position
of the buckles is known.
Another method would be to roughen the pipeline coating deliberately, or to introduce
regular anchor flanges, both techniques are highly dependent on soil interaction.
Roughened pipeline coating has been employed on a number of pipelines where tests, on
soil collected from the field, have demonstrated an increase in axial friction. There is no
record of anchor flanges being used to increase friction, although pipelines have been
installed with J-lay collars. In such cases, it appears that the flange will displace soil during
early life pipeline expansion, leaving an open annulus around the pipe, thus only influencing
friction at the extremes of displacement.
E3.2 DECREASED LATERAL FRICTION

Reducing the lateral friction at the buckle location has the advantage of reducing the severity
of the bending (for a given design condition) in the buckle. Consequently, the VAS can be
increased whilst maintaining integrity of the pipe within the buckle. There are various
methods for reducing lateral friction, including seabed preparation by gravel dumping, used
on the Åsgard project. However, the simplest method is to reduce the operational
submerged weight of the pipeline by the reduction of weight coating or the application of
buoyancy (described in Section E2.2). Another method is the use of sleepers (Section
E2.4), which elevates the pipeline off the seabed.
E3.3 HOT LAY

For a long (fully restrained) pipeline there are benefits in laying the line while hot. As it
cools, during flooding, tension is introduced and restrained by seabed friction[E10]. This
reduces the maximum axial compression in operation, thus reducing the feed-in to each
buckle. To lay a pipeline hot is not difficult if the line is well insulated. An installation
temperature of 5° to 10°C above ambient is quite achievable. However pre-tension would
be lost if the lay operation were interrupted.
For a short pipeline, as-laid temperature only affects buckle formation, since once expansion
has occurred over the length of the line, the history of installed tension is lost.
E3.4 REDUCED DESIGN CONDITIONS

The severity of the buckling problem can be mitigated by reducing the design conditions;
both the pressure and temperature can be modified.
Report No: 5087471/01/A

The pressure can be reduced through the use of a HIPPS or more practically an OPPS
system. This approach is mainly applicable to developments in which the shut-in pressure is
significantly in excess of the maximum operational pressure.
Temperature can also be reduced by cooling the product, using a cooling system or subsea
cooling spool. Both of these measures will reduce the maximum compressive force in the
pipeline. Cooling spools have been used on projects such as Erskine (replacement), Cook
and Mallard.
E3.5 INFLUENCE OF PIPELINE CONFIGURATION

The pipe configuration has a large influence on the compressive force in the pipeline and on
the resistance of the pipeline to buckling. For example, adding an outer jacket pipe for a
traditional pipe-in-pipe system will reduce the propensity for buckling. This is because the
pipeline bending capacity is increased and, for a buried pipe-in-pipe, the larger OD and
increased pipe weight greatly increases the restraint offered by the soil.
Even for a single pipe flowline there may advantage in increasing or decreasing the wall
thickness to mitigate buckling loads; the influence of such changes is case-specific.
Report No: 5087471/01/A

E4 ALTERNATIVES TO LATERAL BUCKLING

E4.1 IN-LINE EXPANSION SPOOLS
If the length of the system is not too great, in-line expansion spools could be introduced into
the pipeline, as illustrated in Figure 9.
Pipeline length
Expansion spool spacing
Figure 9 Expansion spool spacing

Introducing expansion spools reduces the maximum compressive axial force in the pipeline.
This is an obvious and technically simple solution but is generally expensive. Intermediate
spools were employed on projects such as Heron and Jade.
E4.2 TRENCH AND BURY

Trenching and burial constrains a pipeline from lateral or vertical movement, which inhibits
buckling. Backfill of sufficient depth and strength must be laid over the pipe to prevent
upheaval buckling. By controlling OOS levels within the trench, the amount of backfill cover
required over the pipe can be reduced. It is usual to measure the OOS along the route, after
installation, to confirm the minimum level of backfill required. Backfill can be either seabed
soil excavated from the trench, or imported rock, or a combination of both. Upheaval
buckling control by burial is used on many pipelines but in some cases is impractical or too
expensive; for example HPHT pipelines where the loads are severe.
Report No: 5087471/01/A

E5 REFERENCES
E1. Carr, M., Matheson, I., Peek, R., Sanders, P. and George, N. Load and Resistance
Modelling of the Penguins Pipe-in-pipe Flowline Under Lateral Buckling. 23rd
International Conference on Offshore Mechanics and Arctic Engineering. Vancouver,
OMAE 2004.
E2. Peek, R., Carr, M., Matheson, I., Sanders, P. and George, N. Thermal Expansion by
Lateral Buckling: Structural Reliability for the Penguins Flowline. 23rd International
Conference on Offshore Mechanics and Arctic Engineering. Vancouver, OMAE 2004.
E3. Matheson, I., Carr, M., Peek, R., Sanders, P. and George, N. Penguins Flowline
Lateral Buckle Formation Analysis and Verification. 23rd International Conference on
Offshore Mechanics and Arctic Engineering. Vancouver, OMAE 2004.Peek, R., and
Yun, H. Flotation to Trigger Lateral Buckles in Pipelines on a Flat Seabed. Journal of
Engineering Mechanics. ASCE. April 2007.
E4. Peek, R., and Yun, H. Flotation to Trigger Lateral Buckles in Pipelines on a Flat
Seabed. Journal of Engineering Mechanics. ASCE. April 2007.
E5. Peek, R., Kristiansen, N. Zero Radius Bend Method to Trigger Lateral Buckles.
OMAE2008-58046. Proceedings 27th International Conference on Offshore Mechanics
and Arctic Engineering. June 2008.
E6. Harrison, G.E., Brunner, M.S, and Bruton, D.A.S. King Flowlines – Thermal Expansion
Design and Implementation. Proceedings of the Annual Offshore Technology
Conference. OTC 15310, 2003.
E7. Jayson, D., Delaporte,P., Albert, J., Prevost, M., Bruton, D, Sinclair, F. Greater
Plutonio Project – Subsea Flowline Design and Performance. Offshore Pipeline
Technology Conference, 2008.
E8. Nystrom, P., Tornes, K., Karlsen, J., Endal, G., Levold, E. Design of the Asgard
Transport Gas Trunkline for Thermal Buckling. ISOPE 2001. Proceedings 11th
International Offshore and Polar Engineering Conference.
E9. Slettbo, H., Seek, D., Jorgensen, GM. And Dertvik, S. Asgard production start up
confirms Dynamic flowline Design Approach. Offshore Pipeline Technology (OPT)
2001, Amsterdam.
E10. Anderson, M., Bruton, D., Carr, M. The Influence of Pipeline Insulation on Installation
Temperature, Effective Force and Pipeline Buckling. 26th International Conference on
Offshore Mechanics and Arctic Engineering. 2007.
Report No: 5087471/01/A

SAFEBUCK III / Safe Design of Pipelines with Lateral Buckling Page F1
Design Guideline: Appendix F
SAFEBUCK III
Design Guideline Appendix F
SAFEBUCK Deliverables
Report No: 5087471/01/ A

Task Author Phase Last Rev
Volume 1 Design Guideline and Overview

SAFEBUCK Design Guideline III July 2011
State of the Art Report Atkins Boreas I Aug 2004
Andrew Palmer
Third Party Comments on SAFEBUCK Phase I Graham Stewart I Aug 2006
Volume 2 Pipeline Behaviour

Response within a Buckled Pipeline Atkins Boreas I Aug 2005
Mathcad (Ver 12) analytical model -

SAFEBUCK-01 V1.1 (Single Pipeline) Atkins Boreas I Aug 2005
Mathcad (Ver 12) analytical model -

SAFEBUCK-02 V1.1 (Pipe-in-pipe) Atkins Boreas I Aug 2005
Dynamic Modelling of Lateral Buckling Atkins Boreas II Feb 2007
Influence of Lay Tension on Embedment Andrew Palmer I Feb 2007
Lateral Buckling SRA Atkins Boreas II Dec 2007
Reliability of Design Methodology Votadini & Atkins III May 2011
Probabilistic Buckle Formation Report Atkins Boreas I Jan 2009
Probabilistic Buckle Formation: BUCKFAST model Atkins Boreas II Sep 2009
Buckle Formation Atkins Boreas I Aug 2003
Pipeline Walking Atkins Boreas I Aug 2004
Volume 3 Pipeline Limit States

Fatigue Testing and Evaluation Report TWI II Jun 2006
Local Buckling of Pressurised Pipes under Bending

Load Atkins Boreas II Nov 2007
4 Point Bend Tests of 6" Pipe, Examining Effects of

Internal Pressure & Axial Loading on the Buckling Point Mitsui-Babcock II Aug 2006
Low-cycle Fatigue/Cyclic Softening Literature Review TWI II Sep 2006
Failure Modes and Limit States Atkins Boreas I Dec 2003
Review of Fatigue Aspects of Pipeline Welds TWI I Jul 2003
Review of Fracture Toughness and Defect Tolerance of

Pipeline Welds TWI I Jul 2003
Material (Fatigue) Testing and Recommendations for

ECAs TWI I Jul 2004
Additional Corrosion Fatigue Data for Sweet Operating

Environment Atkins Boreas II Oct 2009
Report No: 5087471/01/ A

Task Author Phase Last Rev
Volume 4 Pipe-Soil Interaction

Lateral Pipe-Soil Interaction Data Review CUTS II Oct 2006
Review of Pipe-Soil Modelling Atkins Boreas II Feb 2007
Centrifuge Modelling of Pipe-Soil Interaction - Factual

Report UWA II Feb 2007
Force-Resultant Plasticity Model Cathie Assoc II Aug 2007
Force-Resultant Plasticity Model Chris Martin II Aug 2007
Axial Pipe-Soil Interaction Testing using the Cam-Shear

Device CUTS II Sep 2007
Lateral Pipe-Soil Interaction Testing in the Minidrum

Centrifuge CUTS II Aug 2007
Axial Pipe-Soil Interaction Testing - using Short and Long

Pipes CUTS II Apr 2008
Axial pipe-soil resistance - Summary Report CUTS II Dec 2008
Pipe-Soil Interaction Models for Lateral Buckling Design – D.J. White & C.Y.
Clay Soils Cheuk II Jun 2010
Pipe-Soil Interaction Models for Lateral Buckling Design –

Sandy Soils UWA II Oct 2010
Pipe-Soil User Friction for Abaqus

Friction Subroutine SAFEBUCK Atkins Boreas II Oct 2008
UBOR 2008 Atkins Boreas II Oct 2008
Report No: 5087471/01/ A


SAFEBUCK III Design Guideline Rev A E1 PDF

Uploaded by

Copyright:

Available Formats

SAFEBUCK III Design Guideline Rev A E1 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SAFEBUCK III Design Guideline Rev A E1 PDF

Uploaded by

Copyright:

Available Formats

Safe Design of Pipelines with Lateral Buckling

Issue Date: 30 August 2011

Revision Date Purpose Prepared Approved

A 20/07/2011 For Use M.Carr, D.Bruton, D. Bruton

Report No: 5087471/ 01 / A

Report No: 5087471/ 01 / A

6.6. PLASTICITY LIMITATIONS ........................................................................ 92

Appendix A Design Examples

Report No: 5087471/ 01 / A

1.2. OBJECTIVE OF GUIDELINE

Report No: 5087471/ 01 / A

1.2.1. JIP Partners and Sponsors

Figure 1.1 SAFEBUCK Sponsors (Past and Present)

1.2.2. Who Should Use the Guideline?

1.2.3. Guidance Notes

Report No: 5087471/ 01 / A

1.3. SCOPE OF DOCUMENT

1.4. STRUCTURE OF GUIDELINE

1.5. DESIGN CODES

Report No: 5087471/ 01 / A

1.6.1. Pipeline Specification

1.6.2. Pipeline Material

1.6.3. PIP Systems

Report No: 5087471/ 01 / A

In a regular bulkhead system there is a frequent structural connection (bulkhead)

1.6.5. Installation Plasticity

Report No: 5087471/ 01 / A

1.6.8. Seabed Flatness

1.6.9. Fishing Interaction

Report No: 5087471/ 01 / A

1.7. DESIGN EXAMPLE

1.8. NOTATION, ABREVIATIONS AND DEFINITIONS

Report No: 5087471/ 01 / A

HUB Maximum lateral breakout resistance N/ m

N̂ cr Maximum critical buckling axial force (compression positive) N

N̂ crT Maximum critical buckling axial force at a trigger (compression positive) N

Ni Allowable number of fatigue cycles at a given stress range -

Report No: 5087471/ 01 / A

No Full constrained effective axial force (compression positive) N

Report No: 5087471/ 01 / A

Report No: 5087471/ 01 / A

API American Petroleum Institute

Report No: 5087471/ 01 / A

SRA Structural Reliability Analysis

Buckle Trigger An engineered feature installed to initiate buckling.

Report No: 5087471/ 01 / A

Uniform Strain Capacity The engineering strain at UTS.

Report No: 5087471/ 01 / A

2. BACKGROUND TO LATERAL BUCKLING

Figure 2.1 Side-scan sonar image of a Lateral Buckle

Report No: 5087471/ 01 / A

Figure 2.2 Lateral Buckling Modes

Report No: 5087471/ 01 / A

-1 Distance along pipe (m)

Figure 2.3 Cyclic Response within a Buckle

2.2. BUCKLE DEVELOPMENT

Report No: 5087471/ 01 / A

Hobbs Solution Hobbs Solution

Report No: 5087471/ 01 / A