The Safe Design of Hot on Bottom Pipelines

Deep Offshore Technology Conference Vitoria, Espirito Santo, Brazil 2005

Page 1 of 26

The Safe Design of Hot On-Bottom Pipelines with Lateral Buckling using the Design Guideline Developed by the SAFEBUCK Joint Industry Project.

Authors:

David Bruton - Boreas Consultants

Malcolm Carr - Boreas Consultants

Michael Crawford - BP

Edgard Poiate - Petrobras

ABSTRACT

The SAFEBUCK JIP design guideline has already been used by a number of major operators and contractors on recent project developments for both pipelines and pipe-in-pipe systems.

The aim of this JIP is to raise confidence in the lateral buckling design approach and improve understanding of the related phenomenon of pipeline-walking; with experimental work at TWI on low cycle fatigue materials performance, and at Cambridge University on axial and lateral pipe-soil interaction. The JIP partners are Boreas Consultants, TWI, Cambridge University and OTM Consulting.

Phase I of the JIP has attracted leading players from around the world as contributing participants, who all share the common goal of improving safety and reducing installation costs for hot pipelines. BP, ConocoPhillips, ExxonMobil, Petrobras and Shell, as well as the US Government through the MMS participated in Phase I, while installation contractors and suppliers were represented by Allseas, NKK-Mitsubishi, Technip and Tenaris. Additional participants including Statoil and Stolt Offshore have joined Phase II, which will run through 2005.

This paper gives an overview of the design methodologies and key test results, before describing case histories of fields being developed using the SAFEBUCK design guideline.


Page 2 of 26

INTRODUCTION

Subsea pipelines are increasingly being required to operate at higher temperatures and pressures. The natural tendency of a pipeline is to relieve the resulting high axial stress in the pipe-wall by buckling. Such uncontrolled buckling can have serious consequences for the integrity of a pipeline. Consequently, the industry has generally sought to restrain pipelines by trenching, burying and possibly rock-dumping; or relieving the stress with in-line expansion spools.

A far more elegant and cost-effective solution is to work with rather than against the pipeline by controlling the formation of lateral buckles along the pipeline. Controlled lateral buckling is an efficient solution to the relief of axial compression. Indeed, as temperature and pressures increase further, lateral buckling may be the only economic solution.

With the increase in deepwater developments, where pipeline trenching is uncommon, even low temperature pipelines such as water injection flowlines become susceptible to lateral buckling. This challenge has led to a radical advance in pipeline engineering with a greater need for a robust lateral buckling design solutions.

Unfortunately, the industry’s understanding of lateral buckling is not mature, as has been demonstrated by the occurrence of unexpected lateral buckles, leading in some cases to full-bore rupture. As a result, the SAFEBUCK JIP (joint industry project1) was initiated to address the uncertainties and deliver a demonstrably safe and effective lateral buckling design approach, based on targeted experimental research and the development of “Design Guideline”.

The JIP has also been very successful in forming a knowledge base, with relevant data donated by JIP contractors and JIP participants, on operating pipeline behaviour and relevant project-specific test programmes.

With the JIP now entering its second phase, the design guideline and research findings have been applied on a number of projects. This paper is therefore a good opportunity to present the SAFEBUCK design methodologies and early research to a wider audience.

BACKGROUND TO THE JIP

Lateral Buckling, or Euler (bar) buckling, occurs in the horizontal plane. Such lateral buckles have been observed in many operating pipeline systems laid on the seabed without trenching, an example is shown in Figure 1. Such non-trenched pipelines are usually large diameter and operate at low temperature so that fortuitously, although the lateral buckles are often unplanned, they are lowly loaded and often benign.

1 Abbreviations are listed in Appendix B


Page 3 of 26

Figure 1 Side-scan sonar image of a Lateral Buckle

The acceptance of lateral buckling in such systems has helped to establish lateral buckling as an attractive design solution for relief of axial compressive force in hotter pipelines, which is a particular challenge for HPHT (high-pressure, high-temperature) developments, particularly in deep water where trenching and burial is less feasible and considerably more costly.

There are two key concerns about adopting lateral buckling as a design solution for HPHT pipelines:

• High loading in the buckle

• Control of buckle formation

There is no cohesive guidance in the design codes or industry publications on:

• Limit states and what level of loading is acceptable

• Buckle formation (initiation) which must be achieved reliably

• Buckle behaviour in operation

• Methods of analysis

Because of these shortcomings, many projects have undertaken research work and considerable engineering with the aim of developing a robust solution. However, project timescales have necessitated alternative fall-back solutions and this approach has led to some high profile project issues, in some cases risks were not identified early enough, leading to late design changes that incurred considerable project costs. In other cases, pipeline failure has occurred in operation. These include three catastrophic failures in the North Sea, West Africa and Brazil, each leading to full-bore pipeline rupture. In the first two cases, lateral buckling was not addressed correctly in design; in the last case, buckling was caused by the unexpected exposure and global instability of a pipeline buried in very soft clay[5].

It was proposed that these shortcomings should be addressed by a JIP (joint industry project), launched in 2002 as ‘SAFEBUCK’ by Boreas Consultants, TWI and Cambridge University. The JIP was entitled SAFEBUCK or “The Safe Design of Hot On-Bottom Pipelines with Lateral Buckling”.


Page 4 of 26

JIP SCOPE OF WORK

Initial work confirmed that there were considerable gaps in the data required for a robust lateral buckling design. Further gaps have also been identified in the design of more recent deepwater projects. The aim of the JIP was to address these gaps in understanding and initiate appropriate test programmes, which forms the basis for the ongoing JIP.

The starting point of the JIP was to assemble a knowledge base of participant’s and contractor’s experience, including test data and lessons learned from lateral buckling design and operating experience. This process of gathering of data has continued throughout the JIP.

The scope of work for Phase I of the SAFEBUCK JIP is summarised in Figure 2 and discussed in this paper, which highlights key findings, design methodologies and test results.

Axial testsAxial tests

Review Limit StatesReview Limit StatesMaterial tests

& ECAMaterial tests

& ECALow-cycle

fatigue testsLow-cycle

fatigue tests

Small-scale lateral testsSmall-scale lateral tests

Large-scale lateral testsLarge-scale lateral tests

Pipeline response modelling


Pipe-soil interactionPipe-soil

interaction

Limit StatesLimit StatesData

DonationsData

Donations

Project experience

Project experience

Pipeline walking

Pipeline walking

Analytical models

Analytical models

Buckle formationBuckle

formation

Modelling issues & lessons


Design GuidelineDesign Guideline

Axial testsAxial tests


& ECAMaterial tests

& ECALow-cycle


fatigue tests


& ECAMaterial tests

& ECALow-cycle


fatigue tests

Small-scale lateral testsSmall-scale lateral tests

Large-scale lateral testsLarge-scale lateral tests



Pipe-soil interactionPipe-soil

interaction

Limit StatesLimit StatesData

DonationsData

Donations

Project experience

Project experience

Pipeline walking

Pipeline walking

Analytical models

Analytical models


formation



Pipeline walking

Pipeline walking

Analytical models

Analytical models


formation



Design GuidelineDesign Guideline

Figure 2 Scope of work for the SAFEBUCK JIP

PIPELINE RESPONSE MODELLING

This work addressed the behaviour of the pipeline within a lateral buckle to ensure that the response of the buckled pipeline can be investigated in a robust manner, both in conceptual design and in detailed design, to manage the technical risk associated with lateral buckling.

A key component to the design approach is the adoption of a technique that simplifies design by analysing an isolated lateral buckle between virtual anchors, in what is now commonly called “VAS Analysis”, where VAS refers to the virtual anchor spacing. This approach, which can be applied to conceptual and detailed design, is explained in Appendix A of this paper.


Page 5 of 26

Concept Design Using Analytical Models

To ensure that the risk associated with lateral buckling is managed within a project, the technical challenges must be thoroughly explored at the conceptual design stage. However, the use of non-linear FEA (finite element analysis) at the conceptual stage of a project is not realistic. Current analytical models used for concept design are based purely on elastic stress limits. In reality lateral buckling design is based on moderate plastic deformation on the first load cycle, followed by a fatigue assessment of subsequent elastic cyclic loading for the life of the line. Consequently, the existing elastic models are excessively conservative and do not address the true design limit states.

To address this problem, analytic models were developed, which incorporate first load plasticity and cyclic fatigue in operation. These models were implemented in Mathcad and successfully validated against FE analysis and are already being used by participants to evaluate upcoming HPHT developments. Currently these models are only available to JIP Participants.

Detailed Design Using FEA

In the detailed design phase, analysis will be based upon FE analysis techniques. With the advent of modern FE packages, the analysis technique has become more and more like a black box. However, FE analysis, like any other analysis tool, is simply a mathematical model of the physical world. If the model omits important physical parameters then it is possible to produce results that are misleading at best and catastrophic at worst. The work performed for the JIP investigated the effect of modelling assumptions in four broad categories:

• Mathematical modelling issues;

• Strain localisation phenomena;

• Material modelling;

• Pipe-soil interaction modelling.

The work assessed the importance of the various physical effects to produce guidance on the parameters that must be incorporated within the design analysis.

To ensure a robust assessment of the loads within a buckle, the JIP made a number of recommendations for FE modelling, which can be summarised as follows:- • The model must incorporate the stress associated with the as-laid shape. • The element length at the crown of the buckle should not exceed two diameters. • The model should incorporate a weak joint at the buckle crown. The degree of strength

mismatch requires careful evaluation and does depend upon the dominant limit state. In the absence of a specific assessment, the fully plastic moment of the weak joint should be at least 10% below the fully plastic moment of the nominal pipe.

• If the field joint introduces a significant stiffness discontinuity this should be included within the FE model, the same is true of internal spacers within a pipe-in-pipe system. Discontinuities can greatly amplify loading in the pipe adjacent to the discontinuity.

• For carbon steel pipelines, the material should be modelled as an elastic perfectly plastic material (at least up to 2% total strain), unless it can be shown that the pipe has already experienced sufficient plasticity to improve strain-hardening behaviour prior to commissioning.

• For CRA pipelines, the material can be modelled with smooth strain hardening. However, the material characterisation should be carefully chosen to ensure that the degree of hardening is not overestimated.


Page 6 of 26

• Temperature dependant material properties should be employed in design. At elevated temperature, the thermal expansion coefficient increases while the yield strength and UTS decrease, which exacerbates loading in the buckle. There is also a reduction in Young’s modulus with temperature.

• The full non-linear pipe-soil resistance should be modelled, as discussed later in this paper.

Buckle Formation

Lateral buckling of pipelines can be ‘controlled’ by inducing lateral buckles at predetermined intervals along the pipeline. By initiating buckles at regular intervals along the pipeline, the loads are effectively shared between buckle sites, thus reducing and controlling the load in the pipe. However, this requirement often conflicts with the need to space buckles far apart to ensure reliable buckle formation. The greater the spacing between buckle initiators the higher is the probability of buckles forming at each site. Therefore selecting a suitable spacing is often a difficult design compromise.

One thing is certain, having no strategy to form buckles at regular intervals means that they will form randomly and less frequently than if a strategy is employed. This may be acceptable if a single buckle in a long pipeline will not experience an unacceptably high load. In many cases this is not acceptable, so that an initiation strategy is required.

This work evaluated how buckles form and outlined the key parameters that contribute to the buckling process. There are two strands to this work:-

• Buckle initiation techniques;

• Probability of buckle formation.

Buckle Initiation Techniques

Buckle formation is governed by three parameters; compressive force in the pipeline, out-of-straightness (OOS) features and lateral restraint. Although only three parameters are involved, a number of factors feed into each parameter and there is significant uncertainty over the true magnitude of each. The design strategy will usually involve a buckle initiation technique. A number of methods are available to initiate buckling at a controlled spacing; these include: • Snake-lay (Figure 3);

• Vertical upset (Figure 4). • Distributed buoyancy (Figure 5);


Page 7 of 26

Lateral Buckling

Pitch (Typically 2 to 5 km)

Offset (typically 100m)

Pipeline

Lay bend radius (typically 1500m)

Lay Centre-line

Lateral Buckling

Pitch (Typically 2 to 5 km)

Offset (typically 100m)

Pipeline

Lay bend radius (typically 1500m)

Lay Centre-line

Figure 3 Typical Snake Lay Configuration (exaggerated vertical scale)

Sleeper spacing

(typically 2 to 3km)

PipelineSleepers (typically 2 joints of

large diameter pipe

Sleeper spacing

(typically 2 to 3km)

PipelineSleepers (typically 2 joints of

large diameter pipe

Figure 4 Buckle Initiation Using Sleepers (typically 2 to 3 joints of large diameter pipe)

Buckle spacing (typically 2 to 3km)

Distributed buoyancy added to reduce weight(Typically < 100m)

Pipeline

Buckle spacing (typically 2 to 3km)

Distributed buoyancy added to reduce weight(Typically < 100m)

Pipeline

Figure 5 Buckle Initiation Using Distributed Buoyancy


Page 8 of 26

The snake-lay method is becoming much better understood and a wealth of performance data has been donated to the SAFEBUCK JIP. The distributed buoyancy method is being implemented on current projects but performance data is not yet available. The vertical upset (sleepers/rock-dump) method has been applied to pipelines that are now operating but there is limited performance data available.

The distributed buoyancy and sleeper solutions also benefit the design by mitigating the loads in the buckle[8]. Both methods reduce lateral friction across the buckle, by reducing the pipe weight or reducing the zone of contact with the seabed. This reduction in loading means that less lateral buckles are needed to share the load. Therefore, buckles can be spaced further apart, which increases the reliability of buckle formation. The benefit that these techniques bring has resulted in their use on a number of challenging deepwater projects.

All of the buckle initiation techniques attempt to impose relatively large OOS features at known locations. The design methodology then relies on these “engineered” features being more severe than the inherent OOS. However, as project-specific OOS information can never be known prior to pipe lay, there will always be an inherent uncertainty over the buckling response of the system. This uncertainty includes the lack of knowledge about the level and variability of lateral restraint along the pipeline.

Probability of Buckle Formation

The design approach must understand this uncertainty and reduce it to levels whereby the project can proceed with confidence. A structural reliability model of the initiation issue is the most suitable tool to address this problem. This approach allows the designer to quantify the probability of buckle formation, and hence the probability of a robust design.

Basic probabilistic models of the buckle formation process were developed to allow the probability of buckle formation to be quantified. This method has been explained previously by Carr et al[7]

The approach to design must involve some demonstration of the reliability of buckle formation; this implies a target reliability. The tails of the assumed distributions governs the predictions of the probabilistic model. The actual shape of these distributions will affect the calculated reliability very significantly. In reality, we have no information on the shape of these tails. Therefore, it is clear that the absolute value of reliability produced by the model has little real physical justification. The only way to improve confidence in the prediction of the models is to improve the data on which they are based.

This appears to be a significant drawback to the approach. However, for snake-lay and sleepers, real data is available and the comparison of the probabilistic model and the observed response is extremely encouraging. The predicted probability of buckling and maximum spacing between buckles or VAS is consistent with the observed response. It could be argued that the choice of “reasonable” model assumptions is producing a reasonable estimate of actual behaviour.

Pipeline Walking

Start-up/shut-down cycles can lead to axial ratcheting of a pipeline. Over a number of cycles, this movement can lead to very large global axial displacement, leading to overstressed expansion spools or jumpers, or loss of tension in a SCR (steel catenary riser). This cumulative axial displacement is described as ‘pipe-walking’.


Page 9 of 26

Pipe-walking[6] is normally associated with short, high temperature pipelines but it can occur on longer lines, especially where lateral buckling has occurred (the long line is effectively divided into a series of shorter lines).

The conditions that lead to pipe-walking include:

• Tension at the end of the flowline, associated with a SCR;

• Global seabed slope along the pipeline length;

• Steep gradient thermal transients along the pipeline during start-up/shutdown.

This work evaluated the key parameters that influence walking, principally the axial friction response. Analytical equations to predict the rate of walking were developed from first principals for all three conditions and successfully validated against FE (finite element) models. These FE models were also used to assess the significant influence on walking of variations in the axial force-displacement friction response, which influences walking behaviour.

This work has shown that the analytical equations are ideally suited to conceptual design, to assess the likelihood of walking occurring and to calculate the upper bound walk per cycle.

The following examples illustrate how serious an issue pipeline walking can be for a typical 8-inch surface laid pipeline:

Driving Mechanism Pipeline

Length Approximate Rate of

Walking

Thermal transients with gradient of 10°C/km 2km 15mm/cycle



Seabed slope of 5° 2km 170mm/cycle

SCR with 100kN mean tension 2km 350mm/cycle

Table 1 Rate of Pipeline Walking for Different Driving Mechanisms

The rate of walking can appear to be small but when considered over the entire field life, total axial displacement can become excessive. For example, the axial displacement over 500 cycles at 15mm/cycle is 7.5m, while at 350 mm/cycle the displacement would be 175 m. High magnitude pipeline walking could lead to serious pipeline integrity issues.

Although the JIP has significantly improved the understanding and prediction of pipeline walking, there are certain areas where further work is ongoing in Phase II: • The presence of lateral buckles was shown to change dramatically the walking response of a

pipeline and the presence of high thermal transients could lead to buckle growth over a number of load cycles. Lateral buckles are a significant influence on the walking response, which highlights the need to understand this interaction. Currently the interaction between walking and buckling is assessed on a case-by-case basis by FEA in detailed design.

• Pipe-Soil interaction is usually based on a simple Coulomb friction model. Sensitivity analyses performed in this study have shown that the rate of walking is strongly influenced by the mobilisation displacement and occurrence of peak breakout loads. These results highlight the lack of current knowledge or published work in this area, leading to further work in Phase II of the JIP.


Page 10 of 26

This work has significantly improved the understanding and prediction of pipe-walking and has highlighted certain areas where further testing and analysis is required. The significance of this work is such that a separate paper is planned to address the JIP findings specific to pipe-walking.

Control of pipe-walking

Hold-back anchors are commonly employed to control pipe walking. However, when a pipeline is shutdown additional tension is induced in the line by the restraining anchors. The concern is that tension in a pipeline could be sufficient to cause instability (ratcheting lateral displacement) of the pipeline at the route curves. Except for the shallowest of curves, this instability can pull-out the curve, allowing further pipe to walk, until the curvature is small enough to be stable. The minimum stable radius of curvature may be so large as to compromise field architecture. The susceptibility of pipelines to curve pullout must be understood prior to fixing a field layout. This issue can lead to major field layout changes; it should be addressed as early as possible in front end engineering design.

FAILURE MODES AND LIMIT STATES

Extreme conditions can be developed within a lateral buckle; on first load the stresses can exceed yield and may involve significant plasticity; in addition, shutdowns will lead to very high stress cyclic fatigue. Lateral buckling if left uncontrolled can seriously compromise pipeline integrity. If the buckling response of the pipeline is not understood and controlled, buckling could occur adjacent to key fittings (for example an in-line tee), leading to integrity issues with the flowline and its connections. The end expansion of the pipeline and loads on the pipeline end terminations will also be influenced strongly by the buckling behaviour of the pipeline.

End expansion limit-states are highly project-specific and dependant upon the design of pipeline end terminations and the capability of pipe-lay vessels to install them. Pipe-walking, even when controlled by an anchor, can lead to significant cumulative end expansion. This limit state has become recognised as a significant issue for current projects.

For lateral buckling, the JIP defined appropriate limit states that must be considered in the design process. The key limit states within a lateral buckle are:

• Local buckling;

• Fatigue;

• Weld fracture.

These are explained further below. In addition to these limit states; the pipeline material specification must be consistent with the high levels of imposed load. A limit is placed on the first load strain, to ensure that the uniform strain capacity of the material is not approached. A limit is also placed on the maximum axial stress range, to prevent cyclic plasticity; this limit considers the biaxial nature of the stresses and incorporates an allowance for the Bauschinger effect.

Local Buckling

Local buckling failure is of most concern when a buckle is formed, and generally defines the maximum plastic deformation on first load.


Page 11 of 26

Within the lateral buckle, the local buckling failure mode is driven by the high imposed bending. This is imposed in combination with the internal pressure, external pressure and axial force. The importance of the different load types varies depending upon the pipeline configuration. Design against local buckling can be based upon a moment (stress) limit or a curvature (strain) limit. These are often termed load or displacement controlled approaches and there is much discussion on the appropriate approach within a lateral buckle. The JIP concluded that a strain limit is appropriate, so long as all relevant sources of strain localisation are accounted for in the calculation of the imposed strain.

There has been very significant amount of research effort focused on the identification of suitable limits to prevent local buckling. The JIP reviewed this work and confirmed the applicability of the existing formulations in OS-F101[1] and API 1111[2] for bending dominated loading, subject to appropriate safety margins. However, little public domain work was identified which considers the effect of internal over pressure and bending, especially in the presence of non-zero effective axial force. Work is ongoing to improve the understanding of these combined load cases.

Low Cycle Fatigue

Cyclic loading is generally in the elastic range and defines the allowable fatigue life and fracture requirements.

The fatigue performance of girth butt welds was assessed for the buckle crown, which usually experiences high levels of plastic strain on first load, followed by a relatively small number of high stress operational cycles with each shut-down and start-up. Fatigue testing concentrated on the 'low-cycle' region of the S-N fatigue design curve. ‘Dog-bone’ specimens were taken from full-sized girth welds and some were subjected to initial plastic pre-strain before applying cyclic fatigue at stress ranges up to 1.8 times SMYS. This involved cyclic compressive and tensile loading (Figure 6).

Figure 6 Low Cycle Fatigue Testing at TWI


Page 12 of 26

Testing captured the behaviour of carbon steel, manufactured by the seamless UOE and HFI methods, with single sided welds and with backing-strip welds appropriate to pipe-in-pipe systems. Low-cycle fatigue tests were also carried with CRA material and in a seawater (3%NaCl) environment with cathodic protection. The influence of pre-strain on fatigue performance was also confirmed across normal to low-cycle behaviour.

The fatigue of girth welds within a lateral buckle can be assessed using the standard SN curve of BS7608[3] or DnV RPC203[3]. The applicability of these curves to the lateral buckling problem has been confirmed through the fatigue test programme. In general, the fatigue lives obtained from the endurance tests were consistent with published data for girth welds and the corresponding design classification. The following conclusions are also made:-

• The results of the fatigue endurance tests indicate that similar fatigue lives can be expected from unstrained and pre-strained girth welds;

• The results of tests carried out at 180°C lie within the scatter-band of all other results obtained at room temperature;

• The apparent reductions in fatigue life observed from tests in 3%NaCl environment are consistent with the down-graded fatigue design curves recommended in RP-C203[3] for service in seawater with cathodic protection;

• The nominal stress-range should be limited to ensure that complex plasticity phenomena such as cyclic softening are avoided. To achieve this, a limit on the nominal stress range is proposed. The limit is based upon the magnitude of the Bauschinger effect and is chosen to prevent cyclic plasticity.

Fracture

Pipeline girth welds containing defects must be stable, at the start of life and at the end of life following any growth due to fatigue. Because of the extreme loading, the workmanship-based criteria used to set defect acceptance levels may no longer be reliable. To establish weld acceptance criteria, fracture mechanics calculations (within an ECA) are required to define the maximum allowable flaw sizes and to establish requirements for the NDE inspection system.

A review of available assessment methods concluded that FAD based flaw assessment procedures of BS 7910[10], modified in accordance with experience gained from pipe reeling, can be applied to examine the integrity of welds within lateral buckles.

PIPE-SOIL INTERACTION TESTING

Pipe-soil interaction testing evaluated the large displacement, cyclic response that occurs during lateral buckling, with the associated development of soil-berms. Until this JIP, no data were available to validate the use of large displacement response models and the use of Coulomb friction approximations is unrealistic for large lateral movements with developing soil berms.

In these tests, the pipe was allowed to ride-up over, or embed and displace, the soil in its path. The response was a function of the soil properties, pipe diameter, pipe submerged weight and berm building behaviour. These tests evaluated three stages of pipe-soil interaction:

• Break-out, during buckle formation based on different levels of initial pipe embedment;

• Large amplitude displacement as the buckle forms;

• Repeated cyclic behaviour, which is strongly influenced by the building of soil berms.


Full-scale tests were carried out on a flooded soil test bed (Figure 7) with horizontal displacement of about 8D and some assessment of berm consolidation. These tests are backed up by small-scale tests, using a drum centrifuge (Figure 8), to evaluate high numbers of cycles and variations in consolidation for different soils. This has allowed the SAFEBUCK JIP to evaluate typical deepwater soils from the Gulf of Mexico and West Africa to compare with the kaolin clays and West African clay used for the full-scale tests. Centrifuge testing was well correlated with full-scale tests, so that additional small-scale tests are planned for Phase 2 of the JIP.

Figure 7 Pipe-soil Interaction Testing (Large-Scale) at Cambridge University

Fig

La

This sresovthe

d a

Actuated carriage

Soil be

ure 8 Pipe-soil Interaction T

teral pipe-soil modelling

e lateral buckling solution is extrignificant uncertainty over this rearch and testing on low shear er pipe-soil interaction in soft cla world, where on-bottom pipelin

n
Pipe sectionCamer
est

emespstry. e s

Actuated carriage

in

e

eIo

Glass scree

Page 13 of 26

g (Small-Scale) in Mini-drum Centrifuge

ly sensitive to pipe-soil interaction. Unfortunately, there onse. The SAFEBUCK JIP test programme concentrated ngth, cohesive soils because there is much uncertainty n addition, soft clays dominate the deepwater regions of lutions are currently being employed.

Soil bed


Page 14 of 26

Pipe-soil interaction is often modelled as a Coulomb friction, which can be used successfully for some flowline design calculations and can be employed in conceptual evaluation of lateral buckling, if treated with care. However, Coulomb friction is not appropriate in detailed numerical modelling design for lateral buckling.

It is important to appreciate that for lateral buckling design, it is not possible to adopt a ‘conservative’ value for soil friction, which is a common approach in analysis of hydrodynamic instability. For lateral buckling design, it is important to bound behaviour. Upper and lower values of soil resistance are both important. Uncertainty in soil behaviour will lead to a large range between upper and lower bound behaviour, which may preclude a design solution. It is also clear that breakout behaviour has a significant influence upon buckle formation.

Lateral Cyclic Response Observed From Tests

While breakout loads have been the subject of much research and published papers, there is little guidance on modelling lateral friction for the large cyclic displacements of typically 10 or 20 diameters experienced in lateral buckling. A typical large displacement pipe-soil lateral friction response is illustrated in Figure 9.

32

1

1

32

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 2 4 6 8 10

Lateral displacement (Diameters)

Lateral Reaction

Normally penetrated pipeOver-penetrated pipe

Breakout resistance

Figure 9 Schematic of Cyclic Lateral Force-displacement Response

Figure 9 shows the first load monotonic response from steps 0 to 3, described as follows: (0-1) First load monotonic break-out, with elastic response defined by the mobilisation

displacement and a peak that is dependent on the initial penetration; (1-2) Suction release phase and elevation correction, depending on initial pipe embedment; (2-3) Steady accretion phase, characterised by the gradual increase in friction that occurs

following breakout, as an active berm builds up in front of the pipe.

Figure 10 shows the cyclic force-displacement response overlaid on the first load monotonic response.


Page 15 of 26

2 3

1

564

11

1098

7

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

0 5 10 15 20

Lateral displacement (Diameters)Equi

vale

nt L

ater

al F

rict

iona

l Coe

ffiec

ient

First LoadRepeat cool-downRepeat heat-up

First load - monotonic

Cyclic unload

Cyclic load

Figure 10 Schematic of Cyclic Lateral Force-displacement Response

Steps 4 to 11 in Figure 10 are described as follows: (4-5) Cyclic break-out including suction release from the established static soil berm; (5-6) Cyclic phase with a fresh active berm accretion in front of the pipe; (6-7) Berm reaction which increases as the berm is established over initial cycles; (8-9) Cyclic break-out (as 4-5); (9-10) Cyclic accretion (as 5-6); (10-11) Berm reaction (as 6-7).

The cyclic residual friction response (5-6 and 9-10) is generally lower than the earlier monotonic accretion phase response (2-3), despite continued accretion with each sweep, which helps to establish the static berms. The key to cyclic behaviour is the reaction force from the static soil berms. Once the static berms have formed on the first cycle, they quickly provide a significant reaction to further pipe encroachment. This behaviour makes a Coulomb friction approximation increasingly unrealistic.

This behaviour is confirmed by the soil berms established on operating pipelines where lateral buckles have occurred. Figure 11 shows a typical cross profile at the crown of a lateral buckle.


Page 16 of 26

Displacement range under start-up/ shutdown cycle

Position on Shutdown

As-layedPosition

MaximumExcursion

Displacement range under start-up/ shutdown cycle

Position on Shutdown

As-layedPosition

MaximumExcursion

Figure 11 Typical Cross Profile at Crown of Lateral Buckle on an Operated Pipeline

Without soil berms, numerical modelling using Coulomb friction shows that buckles will grow in amplitude with each cycle. In reality, soil berms restrict growth of the buckle so that cyclic displacements remain fairly constant over a number of cycles, maintaining increased bending in the buckle and higher stress ranges in operation.

Test performed by the JIP provide a significant set of test data and confirm soil berm resistance established over a number of sweeps. Figure 12 is a typical example of cyclic response.

Figure 12 Cyclic Lateral Force-displacement Response (typical)


Page 17 of 26

Figure 12 clearly shows the reaction at each static berm, which is typical of the JIP tests. Berm response is more significant than expected, meaning that lateral buckling cyclic response is likely to be dominated by the restraining influence of the static berms. This means that the variability in cyclic residual friction is of less importance to cyclic behaviour, except perhaps for very light pipes or very stiff soil.

The lateral pipe-soil interaction test results are currently being reviewed and revised in the light of additional project-specific test data, donated to the JIP. This work will be the subject of a future paper.

Axial response tests on heavy and light pipes in West African clay together with axial test data for kaolin, donated to the JIP, have provided valuable information for axial response behaviour of on-bottom pipelines on a natural seabed. Work on axial response is ongoing in Phase II.

DESIGN GUIDELINE

Summary of the Design Guideline

The key deliverable from the JIP is the “SAFEBUCK Design Guideline”. The aim of the guideline is to facilitate safe, economic design of pipelines susceptible to lateral bucking. To achieve these aims 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 Engineering;

• 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. The behaviour of pipe-in-pipe systems is also covered within the guideline.

The guideline provides:-

• A detailed description of the design process;

• Guidance on modelling and analysis of the lateral bucking process;

• Guidance on buckle initiation and mitigation techniques;

• Data on high temperature material properties, pipe-soil interaction and out of straightness;

• Suitable equations to address all key limit states.

The guideline is consistent with the following current design codes

• DnV OS-F101 [1]

• API RP 1111 [2]

Figure 13 is an overview of the lateral buckling design approach within the design guideline.


Page 18 of 26

Maybe

Select Basic design parameters

Is pipeline susceptible to

buckling?

Yes

No

Step 1

Step 2

Step 3

Step 4

Step 5

Step 6

Identify tolerable VAS

No assessment required

Define initiation strategy and calculate buckle formation reliability

Is initiation reliability acceptable?

Yes No

Modify initiation strategy

Can initiation strategy be improved?

Yes

No

Modify design parameters

Is design tolerable at increased VAS?

Yes

No

Identify VAS for acceptable reliability

Evaluate pipeline walking

Design Complete

Maybe

Select Basic design parameters

Is pipeline susceptible to

buckling?

Yes

No

Step 1

Step 2

Step 3

Step 4

Step 5

Step 6

Identify tolerable VAS

No assessment required

Define initiation strategy and calculate buckle formation reliability

Is initiation reliability acceptable?

Yes No

Modify initiation strategy

Can initiation strategy be improved?

Yes

No

Modify design parameters

Is design tolerable at increased VAS?

Yes

No

Identify VAS for acceptable reliability

Evaluate pipeline walking

Design Complete

Figure 13 Design Guideline Methodology for Lateral Buckling (summary procedure)

APPLICATION OF THE DESIGN GUIDELINE IN PROJECT DESIGN

The SAFEBUCK design guideline has now been applied on a number of projects employing lateral buckling as a design solution. The Guideline has provided a valuable framework for design, often being referenced within the Project Design Basis.

Many lessons have been learned along the way in the application of a robust lateral buckling methodology to live projects:


Page 19 of 26

• A variety of buckle initiation methods have been applied and their performance has generally exceeded expectations, giving better than expected reliability of buckle initiation mainly because the inherent localised levels of OOS were of tighter radius than could be guaranteed at the design stage.

• Demanding fatigue environments have led to wholesale adoption of sleepers and or distributed buoyancy on a number of projects to reduce the fatigue damage sufficiently to meet the fatigue design limit state. Distributed buoyancy that achieves neutral buoyancy at installation has been used on a number of projects. In many cases this has been combined with high-quality welding of pipe at lateral buckle locations, often described as ‘SCR quality welding’.

• Pipe-walking and the thermal transients that drive the process have been shown to modify lateral buckling response. The complexity of this interaction is still not well understood. In some cases, buckle growth occurs under the influence of thermal transients. In other cases, the amplitude of lateral buckles has reduced with operating cycles, which has the unfortunate disadvantage of incremental increases in end expansion.

• End expansion has become a significant issue on a number of deep water projects. End expansion is often limited by pipeline end terminations and the capability of modern pipe-lay vessels to install them. Pipe-walking, even when controlled by an anchor, can lead to significant incremental end expansion. This has led to significant modifications to flowline end terminations to handle the cyclic growth in end expansion.

• Hold-back anchors are now commonly employed to control pipe walking. The size of hold back anchors required to prevent pipe walking is often significant, particularly for longer lines, which can require capacities in the region of 100 to 300 tonnes.

• When an anchored pipeline is shutdown, additional tension is induced in the line by the restraining anchor. This tension can be sufficient to cause lateral instability at route curves, resulting in ratcheting lateral displacement of the pipeline. The minimum stable radius of curvature may be so large as to compromise field architecture. This issue has caused major field layout changes on a number of recent projects and should be addressed as early as possible in front end engineering design.

• Uncertainties in soil-pipe friction and initial pipe-embedment can have a significant effect on pipeline design. Recent work has significantly improved the understanding of the soil-pipe response, but there are still gaps in the ability to accurately test and predict this behaviour.

• Sour service is known to reduce fatigue life significantly. There is limited data to support this but indications are that the fatigue life can be reduced by an order of magnitude (factors of 10 or 20 have been suggested depending on the level of souring). While the SAFEBUCK JIP in Phase II and a number of current projects are investigating phenomena, there considerable uncertainty remains. The reduction in fatigue life means that for flowlines on this type of development, fatigue becomes the dominant limit state and designing for such a challenging criteria is extremely demanding.

• Given the complexity of lateral buckling and walking design, the survey and monitoring of flowlines requires enhancement from established practices (i.e. side-scan sonar or ROV fly-over). More high accuracy data is required from in-service behaviour, to provide adequate verification of the design for these flowlines and to assess the remaining fatigue life of the flowline throughout the life of the field.


Page 20 of 26

ONGOING WORK – PHASE II

The SAFEBUCK Phase I design guideline provides a solid foundation for the design of pipelines with lateral buckling. However, there are a number of areas of design where the state of knowledge is uncertain. The guideline puts the responsibility on the individual project to address these gaps.

To improve the guidance and assist future projects, Phase II of the JIP is ongoing. The key areas of work for Phase II JIP are:

• Structural reliability modelling of lateral buckling to improve the design guidance. The aims are twofold; firstly to develop a methodology for the application of SRA to the lateral buckling problem; secondly to employ the SRA model to evaluate the design rules outlined in the design guideline. The aim is to specify less conservative requirements for design. Specifically this would provide optimised guidance on the range of pipe-soil models that should be employed and the tolerable buckle spacing that must be achieved.

• Local buckling behaviour of seamless pipes under internal pressure and non-zero effective axial force. There is a large public domain database of tests undertaken to investigate bending buckling. However, within the D/t range of interest to submarine pipelines (between 10 and 45), there are very few tests involving internal pressure or axial load, particularly for pipe which exhibits a plateau in the stress-strain response. To address this, an experimental program is planned. This will be supported by finite element analysis to develop guidance on the incorporation of these loads into existing local buckling correlations.

• Investigation into sour service fatigue performance under high-stress, low-cycle loading. The fatigue loading within a lateral buckle is quite unusual. It is characterised by a relatively low number of high stress cycles. The design guideline limits the imposed loading to elastic cycling, thus this problem can be categorised as near low cycle fatigue (typically stress range up to 1.6 times SMYS). Fatigue testing was undertaken in Phase I to evaluate the applicability of existing SN curves to the low cycle regime. However, the imposed environment can adversely affect the fatigue performance of pipeline girth welds. Work is planned to evaluate the effect of seawater and H2S on the fatigue performance. The work is focused on the high stress ranges and low frequency experienced within a lateral buckle. This work will complement ongoing research efforts into fatigue in sour environments that are focused at the low stress end of the loading spectrum.

• Pipe-soil interaction, axial response testing. The SAFEBUCK JIP has already identified pipe-walking as a major pipeline design issue and detailed aspects of pipe-soil interaction were shown to modify the walking response. It became clear that the basic understanding of axial pipe-soil interaction was inadequate for the complex loading response associated with walking. To address this, pipe-soil interaction testing (full and small scale) will be undertaken to define the nature of cyclic axial behaviour. This will allow a clear assessment of the walking behaviour to be developed. Further, in order to bring clarity to the basic phenomenon a small-scale walking experiment is being considered. This would provide solid experimental evidence for the phenomenon, as no experimental investigation into walking has ever been undertaken.

• Pipe-soil interaction, lateral response testing. Further pipe-soil interaction testing is planned for Phase II. Within Phase I, experimental apparatus was developed to facilitate the testing of lateral pipe-soil interaction within a centrifuge mini-drum. Phase II will take advantage of this unique facility to further investigate the pipe-soil interaction behaviour and to expand the lateral interaction database produced during Phase I. This will investigate the effect of enhanced lateral excursions (up to 25 diameters), a large number of cycles, variable amplitude cycling and additional soil types.


Page 21 of 26

REFERENCES 1. Offshore Standard OS-F101. Submarine Pipeline systems 2000. Det Norske Veritas.

January 2000.

2. Design, Construction, Operation and Maintenance of Offshore Hydrocarbon Pipelines (Limit State Design). API Recommended Practice 1111. Third Edition, July 1999. American Petroleum Institute.

3. BS 7608: Code of Practice for Fatigue Design and Assessment of Steel Structures. British Standards Institute. 2005.

4. Recommended Practice RP-C203. Fatigue Strength Analysis of Offshore Steel Structures. 2000. Det Norske Veritas.

5. Soil-Structure Interaction of Heated Pipeline Buried In Soft Clay. Costa, A.M., Oliveira Cardoso, C. Santos Amaral, C., Andueza, A. International Pipeline Conference 2002. IPC02-27193

6. Lateral Buckling and Pipeline Walking, a Challenge for Hot Pipelines. Carr M., Bruton, D. and Leslie, D. Offshore Pipeline Technology Conference 2003, Amsterdam.

7. Penguins Flowline Lateral Buckle Formation Analysis and Verification. Matheson, I., Carr, M., Peek, R., Sanders, P. and George, N. 23rd International Conference on Offshore Mechanics and Arctic Engineering. Vancouver, OMAE 2004.

8. King Flowlines – Thermal Expansion Design and Implementation. Harrison, G.E., Brunner, M.S, and Bruton, D.A.S. Proceedings of the Annual Offshore Technology Conference. OTC 15310, 2003.

9. SAFEBUCK JIP. Safe Design of Pipelines with Lateral Buckling - Design Guideline BR02050/SAFEBUCK/A December 2003. [Available to JIP Participants only]

10. Guide on methods for assessing the acceptability of flaws in fusion welded structures. BSI 7910: 1999.


Page 22 of 26

APPENDIX A

LATERAL BUCKLING, EFFECTIVE AXIAL FORCE AND ‘VIRTUAL ANCHOR SPACING’

Effective Force and Expansion of Straight Pipelines

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 axial compressive force will develop in the pipeline, as illustrated in Figure 14.

00.10.20.30.40.50.60.70.80.9

1

0 0.2 0.4 0.6 0.8 1x/L

S/S0

max

Free EndsFixed Ends

Virtual Anchor

Figure 14 Effective Axial Force in a Straight Pipeline

The force presented in Figure 14 is the effective axial force in the pipeline, which drives the structural response. The effective force is made up of the (true) force in the pipe wall and the pressure induced axial force.

Since pressure and temperature vary along the pipeline length, the fully constrained force also varies along the length as the pipe cools. This is shown in Figure 14 by the fall in the curve between x=0 and x=L. For a pipeline with free ends, the effective axial force gradually increases from zero at the ends, due to the frictional restraint of the seabed, until the force reaches full restraint. For a short pipeline, the overall length may be insufficient to reach full restraint, as illustrated in Figure 15.


Page 23 of 26

00.10.20.30.40.5

0.60.70.80.9

1

0 0.2 0.4 0.6 0.8 1x/L

S/S 0

max

Free EndsFixed Ends

Virtual Anchor

Figure 15 Effective Axial Force in a Short Straight Pipeline

The compressive effective axial force in a pipeline therefore depends upon the operating condition of the pipeline and the axial friction. If the compressive force is large enough, then the pipeline may be susceptible to lateral buckling. A lateral buckle results from instability due to axial compressive loading, otherwise known as Euler 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. Trenched or buried pipelines will buckle in the vertical plane, but these are not considered here, as the upheaval buckling design approach restrains the pipe against buckling, whereas the lateral buckling relieves the axial force by encouraging and controlled buckling.

Extreme conditions can develop within a lateral buckle:

• On first load, the stresses can exceed yield and may involve significant plasticity;

• In normal operation, regular shutdowns can lead to very high stress cycles that cause fatigue.

These limit states are addressed in design by controlling the regular formation of lateral buckles along the flowline length using buckle initiators. This is an efficient solution to the high axial force problem and allows the load to be shared between the lateral buckles. An example axial force profile for a straight flowline and a laterally buckled flowline is illustrated in Figure 16.


Page 24 of 26

0500

100015002000250030003500400045005000

0 2 4 6 8 10 12KP

Effe

ctiv

e Fo

rce

(kN

)Straight PipePost Buckling

buckle buckle

Virtual Anchor

Figure 16 Effective axial force profile (example) with and without lateral buckling

In the straight flowline example, the effective force (drawn compression positive) is essentially zero at the flowline ends (subject to some tie-in or riser loads) and becomes more compressive as the seabed friction restrains the expansion. At some distance from the ends the flowline becomes fully restrained (KP 4.6 in Figure 16), and the compressive force begins to reduce after KP 4.6 due to the falling temperature of the fluid. A similar expansion response occurs at the cold end of the flowline, with the force reducing to zero at the free-end

If the flowline is encouraged to form a number of discrete lateral buckles along its length, the force in each buckle depends on a number of factors (operating conditions, seabed parameters, buckle shape, feed-in lengths) and is shown in Figure 16 as approximately 900 kN. Approximately half-way between the buckles the pipe forms virtual anchor points, from where feed-in occurs towards each buckle. The level of feed-in (flowline axial expansion) defines the extent of the growth of the buckles. This behaviour raises the concept of virtual anchor spacing (VAS), as illustrated in Figure 17.

0200400600800

100012001400160018002000

0 2 4 6 8 10 12KP

Effective Force (kN)

virtual anchors

bucklebuckle buckle

D1 D2

bucklebuckle

D5 D4 D3

virtual anchors

Figure 17 Buckle VAS (virtual anchor spacing)


Page 25 of 26

The presence of the virtual anchor points, effectively divides the overall line into a series of short flowlines, which are constrained at their ends. In Figure 17 the VAS (virtual anchor spacings) are labelled D1 to D5. The VAS is a key parameter in the lateral buckling design process. The response of the flowline between virtual anchors is the same as it would be between real anchors. If the buckle spacing is close (small VAS) there is less axial feed-in to the buckle, which reduces lateral deflection and load in the buckle. The aim of the design method is for a large number of buckles to form at regular intervals along the flowline. This produces a solution in which the thermal strain is shared between several sites, leading to manageable strains within each buckle.

It should be recognised that this requirement conflicts with the need to space buckles far apart to ensure reliable buckle initiation. Therefore selecting a suitable VAS is often a difficult design compromise.

The design process is much simplified by adopting VAS models, which are only as long as the proposed distance between virtual anchor locations. This approach, commonly known as VAS analysis, simplifies both concept and detailed design calculations by allowing a large number of parametric analyses to be completed in a relatively short time.


Page 26 of 26

APPENDIX B

ABBREVATIONS

CRA Corrosion Resistant Alloy

ECA Engineering Criticality Assessment

FAD Failure Assessment Diagram

FEA Finite Element Analysis

HPHT High-Pressure, High-Temperature

JIP Joint Industry Project

MMS Minerals Management Service

NDE Non-destructive Examination

OOS Out-of-Straightness

SCR Steel Catenary Riser

VAS Virtual Anchor Spacing

Documents

The Safe Design of Hot on Bottom Pipelines