Insightsonthedesignof...

rsta.royalsocietypublishing.org

ReviewCite this article: Drago M, Mattioli M,Bruschi R, Vitali L. 2015 Insights on the designof free-spanning pipelines. Phil. Trans. R. Soc. A373: 20140111.http://dx.doi.org/10.1098/rsta.2014.0111

One contribution of 12 to a Theme Issue‘Advances in fluid mechanics for offshoreengineering: a modelling perspective’.

Subject Areas:ocean engineering, structural engineering

Keywords:pipeline, free span, bottom roughness,scouring, load regimes, vortexshedding-induced vibration

Author for correspondence:M. Dragoe-mail: [email protected]

Electronic supplementary material is availableat http://dx.doi.org/10.1098/rsta.2014.0111 orvia http://rsta.royalsocietypublishing.org.

Insights on the design offree-spanning pipelinesM. Drago, M. Mattioli, R. Bruschi and L. Vitali

Saipem S.p.A., Engineering and Construction Business Unit,Via Toniolo 1, 61032 Fano, Italy

The design of free-spanning pipelines is performedwith the aim of ensuring their integrity againstpermanent loads generated by seabed roughness,functional loads induced by internal pressure andtemperature, and dynamic loads induced by marinecurrents and direct wave action. In particular, aload and resistance factored design is applied thatfocuses on extreme environmental loads, and a fatiguelimit state approach is applied as a consequence offree-span dynamics due to vortex shedding-inducedvibration and direct wave action. The pipeline free-span scenario can be permanent, when generated byseabed roughness, or characterized by short- to long-term evolution, when generated by seabed mobilityand scouring in shallow waters. Free-span analysisis generally a task involving a number of disciplinesand should be carried out using a multidisciplinaryapproach. The paper illustrates various themesrelated to free-span analysis: (i) free-span scenarios,(ii) characterization of the environment from deep toshallow water related to proper seabed properties,(iii) hydrodynamic load regimes, (iv) pipeline free-span design assessment aiming to reduce overstressand fatigue damage, (v) erodible seabed mobility andlocal scour, and (vi) some experiences of inspectionsurveys chosen as representative of a free-spanningpipeline in sandy soils.

1. IntroductionIn many deep-water pipeline projects, and also inintermediate water in harsh environments, free spans areone of the major challenges. The complex combinationof uneven/erodible seabed and near-seabed hydro-dynamics potentially leads to very expensive spanintervention works. In order to minimize the relatedcosts, a large focus towards free-span interpretation andvortex-induced vibration (VIV) with associated fatiguedamage has been developed during the past decades.

2014 The Author(s) Published by the Royal Society. All rights reserved.

on May 13, 2018http://rsta.royalsocietypublishing.org/Downloaded from

http://crossmark.crossref.org/dialog/?doi=10.1098/rsta.2014.0111&domain=pdf&date_stamp=2014-12-15

mailto:[email protected]

http://rsta.royalsocietypublishing.org/

2

rsta.royalsocietypublishing.orgPhil.Trans.R.Soc.A373:20140111

.........................................................

(b)(a) (c)

Figure 1. Various scenarios related to suspended spans of pipeline: (a) seabed unevenness, (b) erodible seabed and (c) pipelinecrossing.

(b)(a)

(c) (d )KP 29

KP 5

KP 25

KP 30

KP 35

KP 40

Figure 2. Examples of different seabed roughness generating free-spanning pipelines: (a) iceberg scars induced roughness,(b) rock outcrops and soft clayey sediment, (c) sharp shelf break and (d) continental slope seabed roughness. (Online versionin colour.)

The pipeline free-span scenario can be permanent, when generated by seabed roughness(figure 1a) or by artificial supports for pipeline crossing (figure 1c), or characterized by short-to long-term evolution, when generated by seabed mobility and scouring in shallow waters(figure 1b).

Cases of severe seabed roughness can be encountered, for instance in the Norwegian fjords [1],the Arctic and subarctic areas such as the northern part of the North Sea and the Baltic Sea.A feature of these regions, as illustrated in figure 2a, is a seabed characterized by quite severeunevenness generated in previous geological eras by the migration of icebergs, which, draggingalong the seabed, generated deep and large ‘scars’. This is the scenario encountered for instancealong the Northern Leg of the Langeled project and a potential challenge for the future Shtokmandevelopment.

Quite severe seabed roughness can also be encountered in tropical areas (figure 2b). Seabedroughness in the continental shelf is generated by subsequent levels of carbonate outcrops of oldcoralline barriers. In the same area, the coralline barrier might be present at the edge of the shelf,representing a sharp transition (break) between the shelf and the upper continental slope. Manyongoing or near-future field developments in various offshore areas, such as Indonesia and EastAfrica, require pipelines to export the product from the field located in the lowest part of thecontinental slope to the shore. These routes need to cross the sharp shelf break, thus requiringchallenging intervention works and/or managing long and severe free spans (figure 2c).



3


.........................................................

The continental slope itself is typically an area characterized by seabed roughness generatedby various evolution dynamics due to the sediments carried by rivers that gradually accumulatedon the slope and then were further modelled by gravitational movements (landslides, mud flows,etc.). A milestone project particularly challenging in this respect is the Blue Stream in the BlackSea (figure 2d) [2].

In addition, there are also conditions in which the free-span configuration can modify duringthe operating life, thus generating an additional issue into free-span integrity management. Onetypical case is that of sand waves, like those encountered for instance along the Zeepipe, Norfraand Southern Leg of the Langeled project [3]. Other free-span scenarios are those related topipe–soil interaction and scouring in shallow waters [4,5].

Pipelines resting on a sandy seabed in shallow to intermediate water, despite having nofree spans in the as-laid configuration, can develop scouring-induced free spans, which cannotbe deterministically assessed due to the stochastic micro-variability in space of the seabed and intime of the environment (wave and current). This prevents a standard approach to pipeline designand installation, requires one to look at the actual conditions and imposes a statistical assessmentof the possible future pipe–seabed configuration scenarios. Depending on the outcomes of thestudy, the pipeline may need to be trenched soon after laying, when a very short fatigue lifeis assessed, or an adequate pipeline inspection should be planned, when the probability ofoccurrence of unacceptable fatigue loads is marginal.

Understanding the flow that waves and currents induce around these pipelines is critical inassessing the loads that they experience and in identifying necessary measures to ensure stabilityand safety.

The hydrodynamics around a cylinder placed far from a (rigid) wall is completely differentfrom the hydrodynamics around a cylinder touching a wall. The characteristics of the vortexshedding from the cylinder and the ensemble-averaged flow fields were found to be stronglydependent on the gap ratio between the cylinder and the seabed. The wall first influences the flowas a solid boundary inhibiting normal velocities, and then as a surface along which a boundarylayer grows [6]. Moreover, it has been found that regular vortex shedding from a cylinder near aplane wall is suppressed when the gap is less than a certain value. Some of the first investigations[7,8] examined the forces on a cylinder near a plane boundary in a sinusoidally oscillatingfluid. Williamson [9] provided a good description of the flow around a wall-free cylinder inoscillatory flows. Tatsuno & Bearman [10], Sumer et al. [11] and Sumer & Fredsøe [12] provideda fundamental contribution to understanding complex pipe–flow interactions in various flowregimes with sinusoidal forcing.

The hydrodynamics around a cylinder placed on an erodible seabed increases the complexityof the fluid–pipe interaction. However, research on pipe stability [13,14] showed that free spanscan be beneficial for pipeline stability. The formation of a free span can lead to natural self-burialor sinking of the pipeline into the scour hole.

Flow around pipelines in a steady current and under waves has been investigated extensivelyin the past few decades. Most of the studies on local scour below pipelines were concernedwith the so-called two-dimensional scour. Scour induced by waves and combined waves/currenthas been widely studied and is relatively well understood. This includes the onset of scour[15,16], equilibrium scour depth [17–19], time scale of local scour [20] and three-dimensionalscour [21–23]. A comprehensive review on the mechanics of scour is available in [24], whichsummarizes both research results and practical findings to provide engineers and researcherswith the fundamentals of scour.

In the 1970s, pipelines laid at a water depth where surface wave action was still significantwere used in order to be buried; exposure to wave loads or potential hydroplastic phenomenawere definitely avoided [25]. In the case of deep-water pipelines, the analysis procedure waslimited to the identification of the maximum free-span length referred to a 100 years return perioddesign current in order to make negligible the risk of tuning of vortex shedding. The so-called ‘nooscillation’ approach was due to the fact that in most cases a resulting cross-flow oscillation of theorder of one diameter of amplitude was unacceptable [26].



4


.........................................................

The early 1980s saw the issue of more comprehensive guidelines, including the concept ofsynchronization of vortex shedding with oscillation in the direction of the current [27]. Thefirst time this criterion was applied to a submarine pipeline, it proved to be the one steeringthe determination of the limit length, at the same time showing that pre-existing situations ofpipelines laid on an irregular seabed would have to be redesigned in some way as they did notcomply with the acceptability criteria [26].

In the following decades, in situ stability and long-term integrity of offshore pipelines acrossactive seabeds became topical. A number of joint industry projects were addressed to free-spanning pipeline design and extensive experimental campaigns were undertaken by differentEuropean hydraulic laboratories. The issue currently in force for free-spanning pipelines is datedFebruary 2006 [28]. The sections regarding free-span analysis and in-line VIV fatigue analyses arebased on the published results (see [29–34] among others).

The paper illustrates various themes related to free-span analysis: (i) free-span scenarios,(ii) characterization of the environment from deep to shallow water related to proper seabedproperties, (iii) hydrodynamic load regimes, (iv) pipeline free-span design assessment aiming toreduce overstress and fatigue damage, (v) erodible seabed mobility and local scour, and (vi) someexperiences of inspection survey chosen as representative of a free-spanning pipeline in sandysoils.

2. Pipeline free-span scenarios and environmental loadsSea bottom morphology exhibits a large variability: (i) flat, almost featureless situations with littleor no evolution over the short and long terms, (ii) uneven, passive undulations due to semi-permanent bedforms like large sand ridges, pockmarks, relict iceberg scars, palaeo-channels,isolated or random rocky outcrops, and (iii) active, morphodynamics due to either the presenceof active bedforms, i.e. bottom structures that evolve and migrate under the actions of wavesand currents, like sand ripples and sand waves, or the erosion/deposition of sediments causedby the action of sea currents and waves, as commonly occurring in nearshore areas, where boththe shoreline and the bathymetry exhibit modifications, regular or irregular both in time and inspace [35,36].

The variability of the bearing or containment provided by active soil, activated by loadingeffects from near-bottom hydrodynamics or service loads, may cause additional workingconditions for a pipeline, which is moved to interact with varying (in time and in space)boundaries/soils. We will refer to different scenarios:

— Pipelines crossing active seabeds. The undulations of mega-ripples and sand waves are notnecessarily permanent during the operating life. Bedform migration can be anticipated,and the bed-line for the pipeline should be defined in such a way as to minimize the riskof pipe exposure and suspension. This is a risk across sandy seabeds in shallow waters,which can be managed due to a time scale, years, which enables mitigation works. Incoastal regions, the morphodynamics can be worse due to a faster time scale, weeks, forwhich the pipe burial is to be completed just after laying to avoid unexpected pipelineexposure to severe near-bottom hydrodynamics. In terms of pipe–soil interaction, sand isa good bearing soil, geotechnical embedment is minor and only underneath erosion canprovide changes in equilibrium and state of stress.

— Exposed pipelines affected by local, near pipeline, seabed erosion. In sandy seabeds and shallowto medium water depths, scouring in the proximity of and under the pipeline can developeven in the short run. Self-lowering or free spanning is the outcome. Where current isdominant, modifications are not fast except in the presence of considerable tidal currents;should wave-induced conditions be relevant, modifications can be fast and temporaryexposure of long free spans to cross-flows during a single wave storm may cause severepipeline vibration, often unacceptable for overstresses and cumulative fatigue damage.This is a risk across sandy seabeds, in particular where the sand layer is not thick, the



5


.........................................................

underlying soil is stiff and undulations may cause conditions for long permanent spans.Pipe–soil geotechnical interaction may be relevant at free-span shoulders.

— Buried pipelines in soft soils with similar specific gravity, where sinking or flotation dependson weak imbalance between weight and buoyancy, which may vary over time dueto densification or liquefaction under cyclic loads. Soil strength capacity to withstandpressure cycles from waves, as well as uplift due to thermal expansion of the pipeline, arerelevant topics. Whether natural in situ soils or material engineered to meet the needs ofin situ vertical stability, it is relevant in different environments, shallow and deep watersor sandy and clayey seabeds.

— Pipelines across the continental slopes to abyssal planes, part of strategic continental links orof export from deep-water production, resting on soft soil affected by seismic activityand susceptible to impact from geo-hazards. In terms of pipe–soil interaction, a numberof efforts have been and are currently being made to understand the load conditionsimposed by impacting plastic flows, but it is far more important to define the potentialtrajectories of the flows with the aim to define pipeline routes that minimize the potentialinterference of events with impact angle close to the perpendicular to the pipeline. Thisis a new field of research that is going to be energized by a number of field developmentsthat are expected in imminent projects to face the difficulties of routing export pipelinesacross very uneven bottom roughness and unstable slopes potentially triggering massflows and turbidities.

The near-seabed environment is very complex from a fluid-dynamic point of view. The boundarylayer at the seabed–water interface is driven by bottom roughness, which can be smooth oruneven, and kind of flow, which can be almost steady current, unsteady current or oscillatorywaves. Environmental conditions cover different phenomena, which may contribute to structuraldamage or operational disturbances as well as seabed characteristics. Wave conditions to beconsidered for structural design purposes may be described either by deterministic designwave methods or by stochastic methods applying wave spectra. For the quasi-static responseof structures, it is sufficient to use deterministic regular waves characterized by wave heightand corresponding wave period. Structures with significant dynamic response require stochasticmodelling of the sea surface and its kinematics by time series. The interaction between waves andcurrent gives rise to a complex fluid–pipeline interaction. The current can cause (i) VIV, (ii) anincrease in drag and lift forces and (iii) seabed scouring around submarine pipelines.

Theoretical modelling allows investigation in detail of the phenomena occurring close to theinterface and provides explanations for causes and consequences of events. But very often, owingto its complexity, it has no practical application in engineering. Usually, simplified semi-empiricalinterpretative models are applied. Semi-empirical models, having a strong feedback fromtheoretical models, try to quantify and express in formulae the outcome of experimental tests.

3. Near-seabed flow conditions and load regimes around free-spanningpipeline

Impinging flow conditions that a pipeline resting, more or less continuously, on the seabed canexperience are in general due to steady—more or less turbulent—currents of tidal, barotropic,storm surge or density nature, therefore characterized by a time scale of variation of some hours,combined with oscillatory flows due to perturbation to the water column caused by travellingsurface waves, characterized by a time scale of variation of 10–20 s.

One can make the following straightforward assumptions:

— In deep waters, where ‘deep’ here means that the surface waves are not able to produceany significant perturbation at the seabed, steady currents are relevant flow conditionsimpacting on the pipeline.

— In shallow-to-medium water depths, oscillatory and steady components of the impactingvelocity may be contemporary or not. In some circumstances, steady currents are



6


.........................................................

not correlated to waves: the pipeline can be impacted by a steady current without asuperimposed oscillatory component as well as by an oscillatory flow devoid of anysuperimposed steady component.

Correlation between waves and currents can influence either the combination of extreme eventsfor both waves and currents or even the ordinary near-seabed flow conditions. Unfortunately, ameasure of the correlation is hardly quantifiable at the design stage and, therefore, is not used inthe design process. Conservative assumptions are generally applied, as for instance proposed in[37] for extreme conditions.

A series of extensive experimental activities performed in both laboratory and field conditionshave allowed hydrodynamic researchers to classify flow conditions impacting on a near-seabedpipeline as a function of the parameter α, expressing the ratio between steady current (Uc) andthe steady current plus oscillatory flow (Uc + Uw), particularly:

— wave dominant, for α < 4/5; and— current dominant, for α > 4/5.

The ‘wave dominant’ condition may be further classified as follows: (i) wave superimposed bycurrent for α between 0 and 1/2; (ii) current superimposed by wave for α between 1/2 and 4/5.

The separation at α = 4/5 and further separations are arbitrary, merely based on the followingconsiderations:

— For α > 4/5, the flow–pipe interaction regime looks similar to conditions of interactionwith steady currents superimposed by a large turbulence.

— For α < 4/5, flow–pipe interaction regimes change considerably as a function of theparameter α, in particular: (i) for α between 0 and 1/2, the interaction is strictly linkedto wake reversal effects, considerable at α = 0 and low Keulegan–Carpenter number(KC, with KC = U0T/D, with U0 = amplitude of free stream mean velocity, T = period offluid oscillations and D = cylinder diameter); (ii) for α between 1/2 and 4/5, the wakereversal effect disappears and the interaction is strictly linked to the development ofvortex shedding under pulsating flow conditions; and (iii) for α still around 1/2, theinertial effects of vortical cells in the wake are significant and may influence the response.

Figure 3 gives an indication of the flow conditions and consequent implications of load regimes,whether characterized by a considerable oscillatory component, α between 0 and 1/2, orcharacterized by a unidirectional diffusion of vorticity, α > 1/2.

It is necessary to formulate one or more models capable of describing the correlation betweenflow condition and hydrodynamic loads that will impact on the pipeline, first keeping anyinterference with the seabed out of consideration.

Here below, load regimes are described as a function of α, assuming that the oscillatory flowsare harmonic. In real life, oscillatory flows due to waves are of random nature. For each sea state,the random sequence of wave heights corresponds to a random sequence of near-seabed orbitaloscillations. Therefore, it is difficult to make a choice about a characteristic α to be consideredin the superposition of steady currents over wave-induced velocities. The sequence of velocitycycles may affect the hydrodynamic loads acting on the pipeline. Either single oscillation-basedor characteristic oscillation-based α parameters are applicable. For design purposes, it can beproposed to use both values: (i) the single oscillation-based α is applicable for fatigue analysisbased on decomposition of sea states envisaged during the pipeline lifetime in single waves;(ii) the characteristic oscillation-based α, where a characteristic oscillation is assumed as thesignificant cyclic velocity associated with each sea state, is applicable for fatigue special analysis.



7


.........................................................

3.0

2.5

2.0

1.5

1.0

0.5

0

–0.5

–1.0

wave period (s)0.4 0.8 1.2 1.6 2.0

Uc+

Uw

(cm

s–1)

a = 2/3 a = 1/2 a = 0

Figure 3. Flow velocity (Uc + Uw) as function of wave period for variousα regimes.

The following can be assumed:

— For both wave and current dominant conditions, the impinging flow gives rise to bothcross-flow and in-line loads acting on the pipeline.

— For α ranging from 0 to 1/2, here above classified as pure wave and wave superimposedby current condition, the following hold.

(i) In-line loads can be described according to the Morison model, provided that dragand lift coefficients are relevant to test conditions for which the orbit of particleswas almost flat, as occurs near the seabed. Loads are due to the oscillations of therelative velocity between the impacting flow and the pipe response, and there is noor, better, negligible in-line excitation from vortex shedding.

(ii) Cross-flow loads are due to one main mechanism: the development and theasymmetric shedding of vortices during each half-cycle. Force or response modelsinterpreting such load and response have been developed and are available;however, no specific formulation has received a general consensus. For instance, DetNorske Veritas [37] supplies response amplitude curves as a function of α, reducedvelocity (VR) and KC numbers. However, the influence from the wall proximityeffects and the frequency of the excitation mechanism, as well as the influence ofrandom versus harmonic oscillatory flow conditions, are not clear.

— For α ranging from 1/2 to 2/3, here above classified as current superimposed by wavecondition, till α = 4/5, the concepts and models applied in the previous item (α rangingfrom 0 to 1/2) are still applicable, but with a considerable variation of relevant coefficientof force models or response amplitudes. Particularly the following must be considered.

(i) In-line loads are now unidirectional and more relevant to the velocity squared termof the Morison formula than to the acceleration component. Loads are still due tooscillation of the impacting velocity, and negligible in-line excitation from vortexshedding is foreseen.

(ii) Cross-flow loads due to vortex shedding are approaching a condition resemblingthe steady current response for α ∼ 2/3. Response amplitude curves for the relevantα, VR and KC parameters from Det Norske Veritas [37] are a unique reference.

— For α > 4/5, new considerations should be made, as follows.

(i) In-line loads comprise the following:— a steady component due to the drag force from the impacting current,— an oscillatory component due to the regular vortex shedding tuned on a natural

frequency of the free span.



8


.........................................................

Both can be dealt with by traditional approaches accounting for findings from jointindustry projects (JIPs) GUDESP [38] and MULTISPAN [29] (implemented in [37]),as far as response amplitudes and relevant parameters are concerned.

(ii) Cross-flow loads are cyclic due to vortex shedding. They can be dealt with using atraditional response model also including findings from MULTISPAN as for ‘onset’of oscillations, implemented in [37].

Here above, no consideration was given to the wall proximity effect. In general, the interferencebetween the flow separation from the cylinder wall and the boundary layer on the seafloor causesthe following:

— a significant reduction of the amplitude of vortex shedding-induced oscillations in steadycurrent conditions (α > 4/5) until they are cancelled at very small gap ratios (less than 0.3external diameter);

— significant vertical oscillations under oscillatory flow conditions (α < 1/2) particularlyfor gap ratio less than the external diameter, at a frequency twice the wave frequency,for which the variation of instantaneous lift force coefficients versus flow velocity is thedriving mechanism; and

— the increase in the streamwise forces due to both steady and oscillatory flows, because ofthe anticipation of flow separation from the top wall of the pipeline and the consequentincrease of drag coefficients as a consequence of the flow asymmetry created by the wall.

Moreover, the roughness of the sea bottom, which could be relevant where the flat bottom hasbeen modified due to sediment transport or the bottom is uneven, creates turbulence, which couldstrongly influence the velocity profile upstream of the pipeline and the consequent hydrodynamicloads acting on the pipeline. A comprehensive review of the subject topic is given in [12]. Inengineering practice, load regimes are described using equations or predictive models capable ofreproducing the loading process, which resemble at best records from field or laboratory tests.Predictive models are proposed in guidelines and codes. Purpose-developed predictive modelsand relevant coefficients are given in [12,28,37]. In particular:

— in-line loads (force model) due to steady or oscillatory flows are predicted using theMorison equation; recommended drag and inertia coefficients are given as a functionof the KC number, α and e/D (seabed proximity) ratio;

— cross-flow vibration (response model) due to steady or oscillatory flows is given as afunction of VR, KC number and α parameter; and

— in-line vibration (response model) due to steady flow (α > 4/5) is given as a function ofVR, free stream turbulence and e/D (seabed proximity) ratio.

4. Pipeline affected by local seabed erosionIn the case of pipelines crossing one another or seabed unevenness, the location where the spanoccurs are fixed and properly managed by routing analysis, but when the pipeline is laid onerodible seabed a span may exist in most locations. When a span develops due to scour, it maychange in location and time. Understanding of the scour mechanism, its generation (i.e. onset ofscour) and its development is critical in assessing the loads that a pipeline can experience duringits operative life.

(a) The onset of scourA pipeline laid on an erodible seabed produces a pressure gradient between the upstream anddownstream sides of the obstacle. If the sea bottom is permeable, i.e. it is a sandy sea bottom, sucha pressure gradient drives a seepage flow in the bottom on which the pipe is laid. This seepage



9


.........................................................

u (c

ms–1

×10

–3)

0–5

0

5

2 4 6time (s)

8 10–5

0

5

time (s)0 2 4 6 8 10

(b)(a)

onset of scour

u (c

ms–1

×10

–3)

–5

0

5

0 2 4 6time (s)

8 10

(c)

Figure 4. (a–c) Stills from a videotaped sequence of onset of scour. (Online version in colour.)

flow tends to drag sand particles and, if sufficiently large, is able to wash away the sand beneaththe pipeline, originating the onset of scouring underneath the pipe itself. The onset of scour iswell described in [16]. Stills from a videotaped sequence of increasing wave passing above anerodible seabed are shown in figure 4.

The onset of scour occurs when the third wave passes above the pipeline. A video is availablein the electronic supplementary material (video 1. Animation of the onset of scour).

Scouring occurs when the pressure gradient between the two sides of the pipe exceeds thestabilizing forces that keep sediment in place, i.e. gravity and grain friction. Quite obviously, thepressure gradient increases for decreasing relative embedment, as the distance (in length) overwhich the pressure difference is exerted becomes smaller. One of the key parameters for scourinception is the critical embedment, above which no scour is expected. In the case of current theperiod for which the pipeline is exposed to critical pressure gradient forces is quite long, whereasfor waves the pressure gradient necessary for the onset of scour is available for a very short periodof time, O(0.5 s) [24].

The role of the vortex structure around the pipe has been discussed qualitatively in [16,39]and assumed to have a minor influence on erosion. The vortical structures that dominate the flowaround the cylinder generate sweeps and ejections. Ejections contribute to sustain the lift-off ofparticles from the seabed, while sweeps contribute to particles approaching the seabed [40].

In the initial stage of scour, the gap between the pipe and the seabed is very small, i.e. e � D(e is the pipe embedment). During this stage, a substantial amount of water moving underneaththe pipe causes water–sand mixture to divert to the gap (the so-called ‘tunnelling’). The scourunder the pipe occurs very violently considering that a factor of 4 increase in the bed shear stresscauses a factor of 8 increase in the sediment transport. The tunnelling progressively reduces untilan equilibrium condition is reached, i.e. the bed shear stress along the bed becomes constant andequal to the undisturbed value. The final stage of scour is basically characterized by the vortex



10


.........................................................

shedding, i.e. the so-called lee-wake erosion. Two videos showing the tunnelling and the lee-wakeerosion for a test case are available in the electronic supplementary material (video 2. Animationof tunnel erosion; video 3. Animation of lee-wake erosion) (animations have been extracted fromscour experiments conducted by Mattioli et al. [41]).

(b) Free-span developmentThe onset of scouring under a pipeline could lead either to rapid pipe self-lowering or to free-span formation, evolution and persistence up to several years. If the flow is largely overcritical,then the onset of scour could occur under long stretches of pipeline causing pipe self-lowering;if it is only locally overcritical, then the onset of scouring leads to free-span generation. Once thefree span is generated, it will propagate along the pipeline. This is due to the flow that is now ableto pass below the pipeline, is increased at the span ends and tends to remove sediment at the spanshoulders. The length increase in the free span causes a deflection of the unsupported section ofthe pipeline that tends to sag into the scour hole. Consequently, the pipe becomes more and moresheltered and its impact on the free flow decreases. This process ends when the decreasing flowbelow the pipeline is no longer able to increase the scour depth and the pipe finally touches thebottom of the scour hole. At this point, a backfilling of the scour hole starts, causing the burial ofthe previously unsupported section of the pipe.

The factors that help pipe self-lowering are a low pipe embedment into the soil, a shallowwater depth, i.e. a large wave action at the sea bottom, and a large sediment mobility, i.e. absenceof cohesive material (clay) and of very large grain size sediments. Conversely, a relatively lowwave and current action at the sea bottom and low sediment mobility favour the formationand the long duration of free spans. These two phenomena are often mutually exclusive anddetermine the typical evolution of the pipe–seabed configuration, i.e. the occurrence of a fastpipe self-lowering could cancel any free-span condition, while the presence of a free-spanningconfiguration is often associated with slow and limited pipe self-lowering.

The evolution of free spans can be divided into two large classes:

— Free spans that disappear within a single storm (scenario B(a) in figure 5), owing to failureof the support at the shoulders, without achieving full development. Owing to their shortlifetime, they do not experience significant fatigue accumulation and usually do not posedesign constraints. This scenario is typical of seasonal wave climate; the occurrence ofseasonal storm condition overlaps the free-span development.

— Free spans that develop until touchdown occurs, due to the deflection caused by theirown weight (scenario B(b)–C in figure 5). Given their length and persistence, they couldaccumulate fatigue damage with potential danger for pipeline integrity. It is observedthat this encompasses a rather wide class, ranging from the situation with a free spansurviving weeks to months (low-diameter pipeline in shallow water with significantwave activity) to the situation with such slow evolution—and alternations betweenbackfilling and erosion periods—that free spans are almost permanent (large pipelinesin deep, current-dominated environment; figure 6). This scenario is typical of moderatewave climate characterized by the occurrence of intense storms that induce scour andfree-span development.

Obviously, intermediate situations of transition between these two extremes occur. It should alsobe mentioned that both phenomena could occur at the same location at different stages of pipelinelife. Initially, when pipeline embedment is low, pipe self-lowering could dominate. This causesa progressive increase in pipe embedment up to a level such that the onset of scour can occuronly locally, leading to the formation of persistent free spans. The evolution of a free span canbe analysed by considering separately the transverse and longitudinal development of the scourhole.



11


.........................................................

wave climate

scenario A scenario B scenario C

free span

deflection

embedment

H

L

def. def. def.

H

L

H

L

time

time

time

time

time

time

time

time

time

absent

absent

fast self-lowering, no freespan with no risk for fatigue

possible self-lowering, freespan with limited to

medium risk for fatigue

very slow self-loweringlong free span with medium

to high risk for fatigue

max

max

max

ba

ba

max

max

max

Figure 5. Pipeline on erodible bottom—scheme of self-lowering versus free span development according to site waveconditions.

9

8

7

6

5

4

3

2

1

0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 760

90

80

70

60

50

40

30

20

10

0

time (months)

leng

th (

m)

Hs(m

)

Figure 6. Southern North Sea—example of the time evolution and disappearance of free span of a 48 inch pipeline in amoderate wave sea, model simulation. The dotted line shows the time series of significant wave height, whereas the solidline shows the length of the span.

The scour evolution in the transverse direction can be evaluated as an exponential trendtowards the maximum equilibrium scour depth. The time scale of this exponential evolutionin empirically estimated in [20,42], while the equilibrium scour depth is given in [22,24,43] forsteady current conditions, and in [24,44,45] for waves plus current conditions. The longitudinalscour development refers to the length evolution of the free spans, resulting from the progressiveerosion at the span shoulders. Empirical formulae for the longitudinal scour celerity are analysedin [44].



12


.........................................................

The unsupported sections of the pipeline tend to deflect and sag into the scour hole, owing tothe weight of the pipe itself. The deflection increases with increasing span length, until the pipetouches the bottom of the scour hole. From this point onward, the scour hole will be backfilledand the pipeline section undergoes a burial process. Formulae for pipe deflection are given in[28]. Pipeline sagging is the main mechanism leading to the disappearance of free spans, but thetime scale of its occurrence has a very wide range.

(c) Metocean conditions as driving factorMany of the developing scenarios could pose concerns about pipe fatigue depending on exposure,suspension time interval and site climate conditions. Hence, for a reliable pipeline design crossingerodible seabed areas possibly affected by onset of scouring, it is necessary to have a tool able toscreen these various alternatives.

From the above description, the evaluation of free-span formation and evolution on an erodiblebottom requires due consideration of the following main phenomena: (i) tunnelling and/or onsetof scouring, (ii) pipe self-lowering, (iii) free-span formation/development, (iv) backfilling of thescour hole, (v) unsupported pipeline sagging, (vi) pipeline self-burial, (vii) upstream backfillingand (viii) leeside erosion.

In this, a realistic sequence of metocean (meteorological and oceanographic) conditions at thesite is the driving factor. The availability of measured wave and current time series is a goodstarting point. Anyway, it is now widely recognized that different sequences of metocean eventscan lead to a certain variability of the reference scenario. This issue can be faced by reconstructingrealistic time series having the same overall statistics as the original one but where the sequenceof events is stochastically determined. In the past decade, long hindcasted time series havingthe statistical properties of the site under consideration have become available for most of theworld ocean. This time series can be statistically analysed and used to produce a few years longtime series for the probabilistic assessment of the possible scouring/free-span scenario and of itspossible variability in the near future.

A procedure for the reconstruction of time series has been developed [46]. It is based onthe subdivision of the waves and residual current time series into a sequence of storms andcalm events. The tidal current component is extracted from total current by performing aharmonic analysis. Marginal extreme distributions of intensity and conditional probability ofduration and direction are defined for storms and calms. Time series are reconstructed byrandom extraction from the distributions obtained from site-specific time histories of storm/calmpersistence and sequences. Then the tidal current is added again to the residual current. Theuse of a number of time series is adopted to allow for a statistical prediction of the outcomeof such a nonlinear process. The impact of metocean events on the time scale of processes maybe relevant. A semi-empirical model implementing the above phenomena has been developedwith the aim of providing a screening assessment of the possible occurrence and evolution ofscouring phenomena for a pipeline crossing sandy erodible areas [46]. The flow chart of the abovemethodology is given in figure 7a.

Once the time history of the free span is evaluated with the above approach, the fatigueanalysis can be carried out considering the effects of VIV and the oscillatory wave action,as indicated in [28]. A fully integrated approach taking into account (i) eigenvalue analysis,(ii) response analysis and (iii) long-term fatigue damage calculation has been developed by Dragoet al. [46]. The flow chart is shown in figure 7b, whereas figure 8 reports an example of the resultsobtained for a 48 inch pipeline in the Southern North Sea.

In the following, some results of the model are reported for a large-diameter pipeline(1.27 m) laid on a sandy sea bottom, in different water depths ranging from 20 to 50 m. Forthe environmental conditions, a 6-years long hindcasted time series of waves and current fora 40 m water depth location of the North Sea have been used. The soil is medium dense sandwith D50 = 0.15 mm (median diameter) and 0% clay content. Twenty 6-years long reconstructedtime series have been used to account for the stochastic variability of the environment. The value



13


.........................................................

wave and current

pipe soilanalysis

pipe soil results:time step

structural andmodal analysis:

modal parameters

scour hole/pipe configuration(seabed proximity and trench

factors)

response analysis(VIV and direct

wave action)

vibration amplitude and frequency(cross-flow and in-line)

time step contribution to fatigue damage:cross-flow and in-line

mid span and span shoulder

free spanlength

scour holedepth

pipedeflection

pipe loweringyes

no

no

yes

initial conditionspipe soil configuration

pipe and soil data

Is there a free span?

time stepwave and current data

upstreambackfilling

andleeside erosion

upstreambackfilling

andleeside erosion

Generalizedonset of

scouring?

Generalizedonset of scouring?

pipe loweringat shoulder

pipe deflection

Touchdown?

Localizedonset of scouring?

Free spandisappear?

(touchdown)

2D scour:scour hole

depth

3D scour:length offree span

yes

no

yes

no

no

yes

no

yes

(a)

(b)

Figure 7. (a) Flow chart of the free span onset and evolution. (b) Flow chart of the free span analysis. (Online version in colour.)

averaged over the number of simulations (1 for hindcasted, 20 for reconstructed) of minimum(Emin) and maximum (Emax) embedment and the value averaged over the number of occurredfree spans (N) of free-span length (L) and total duration (Dur) are reported in table 1.

A rather good agreement is found and only a slight overestimation of the minimumembedment Emin appears using the reconstructed time series. This demonstrates that ‘on average’the reconstructed time series produce results in accordance to the hindcasted time series, but



14


.........................................................

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

90

80

70

60

50

40

30

20

10

0

span

leng

th/w

ave

heig

ht (

×10

)

fatig

ue d

amag

e

time (h)0 500 1000

fatigue damage Hsfree-span length

1500 2000 2500

Figure 8. Integrated free-span analysis: time series of wave height, free-span length and accumulated fatigue damage (unitycheck, i.e. the ratio between the actual number of stress cycles and the number of cycles to failure). Model simulation.

0

5

10

15

20

25

30

35

40

total >10 m >20 m >30 m >40 m >50 m >60 m >70 m >80 m >90 m

time

of p

ersi

sten

ce (

mon

ths)

free-span threshold level (m)reconstructed mean reconstructed max hindcasted mean hindcasted max

Figure 9. Comparison of free-span total duration and duration over selected thresholds for hindcasted/reconstructed timeseries (depth 35 m).

reconstructed time series provide a larger test case database on which extreme or low-probabilityoccurrences can be detected.

Figure 9 shows total duration of free spans and durations of free spans over selected lengththresholds for the water depth 35 m obtained with the reconstructed time series and comparedwith the results obtained with the hindcasted ones. It can be noted that, even if the results withthe hindcasted series are close to the average values obtained by the reconstruction, the rangeof variability of the reconstruction is much wider and provides a statistically more significantsample.

Differently from permanent free spans caused by sea bottom unevenness, the evaluation ofthe fatigue accumulated by an evolving free span should be calculated considering the actualenvironment causing its development. In fact, as can be seen in figure 8, the main part of thefatigue is accumulated during the last part of its life when its length is maximum and caused



15


.........................................................

1000

100

10

1

0.10 100 200 300 400

free-span duration (days)0 50 100 150 200

persistence above 60 m (days)

fatig

ue d

amag

e

(b)(a)

Figure 10. Scatter diagrams of accumulated fatigue damage versus (a) free-span total duration and (b) persistence above 60 mlength. Depth 20 m.

Table 1. Comparison of mean values (number of free spans occurred, N), hindcasted and reconstructed minimum (Emin) andmaximum (Emax) embedment, the free-span length (L) and total duration (Dur). WD=water depth.

hindcasted reconstructed

WD Emin Emax N L Dur Emin Emax N L Dur(m) (m) (m) (m) (months) (m) (m) (m) (months)

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

20 0.34 0.89 3 76 7 0.39 0.89 67 75 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

30 0.29 0.89 3 78 14 0.34 0.89 54 76 14. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

35 0.28 0.89 3 78 21 0.33 0.89 43 77 19. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

40 0.27 0.89 2 78 28 0.32 0.89 32 77 27. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

50 0.27 0.89 1 80 50 0.30 0.79 22 78 46. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

by the same storm causing its disappearance, i.e. fatigue accumulation and storm occurrenceare strictly correlated in time. Hence, fatigue is not directly correlated to free-span persistence,and applying a pure statistical climate during the whole free-span lifetime, which would providelarger fatigue for longer persistent free spans, could result in fatigue overestimation for slowlyevolving free spans [46]. On the contrary, shorter-duration free spans, but subject to largehydrodynamic loads during the latter part of their life, could be more critical than long-durationfree spans in mild conditions and fatigue damage in this case could be underestimated.

Since large storm events cause the greatest component of fatigue damage and also tendto rapidly lengthen the free span, the overall duration of the free span could not be directlycorrelated to the amount of fatigue damage accumulated. This is demonstrated in figure 10, whichfor the 20 m water depth case shows the scatter diagram of the accumulated fatigue versus thefree-span total duration (a) and the persistence above 60 m length (b). It can be noted that there isnot a definite correlation between the total free-span duration and the fatigue damage (figure 10a).On the other hand, the main part of the free spans that accumulated unacceptable fatigue damagehad persistence above 60 m shorter than 30 days, while those with persistence longer than 100days accumulated relatively low fatigue damage (figure 10b). This indicates that if a relativelymild climate occurs when a free span reaches a length of about 50–60 m, it can persist for a longtime without accumulating large fatigue damage. Conversely, if large storm events occur, thenthe free span could accumulate unacceptable fatigue damage in a very short time and disappearfor self-lowering during the same storm event.

The usual assessment of the fatigue damage for a free span is performed by evaluating howlong a free span of a certain length can remain exposed to wave and current action, consideredwith their occurrence probability, before accumulating unacceptable damage. This methodology,applied to scouring-induced free spans, would obviously lead to the assessment that the freespans with longer persistence over a certain length are those that will accumulate greater damage.



16


.........................................................

mea

n re

turn

per

iod

in y

ears

WD = 30 m

0.10 0.15 0.20 0.30 0.40 0.50 0.60 0.80 1.00 1.50 5.00 6.000.01

0.10

0.20

0.30

0.40

0.50

0.600.70

0.80

0.90

0.99

0.999

0.99990.99999

cum

ulat

ive

prob

abili

ty W

eibu

ll sc

ale

fatigue damage

weibull 3 par. distributionP(x) = 1 – exp – [(x + C)/A]B

A = 0.213 B = 0.623 C = –0.170linear correlation = 0.99871

data per year 0.550data in sample 55.00

3

5

10

100

extreme values3 year(s) = 0.245 year(s) = 0.39

10 year(s) = 0.67100 year(s) = 2.14

Figure 11. Example of three-parameter Weibull distribution of fatigue damage for free span in 30 m water depth. (Onlineversion in colour.)

Therefore, it can be concluded that the assessment of fatigue damage for scouring-induced freespans should be carried out in parallel with the calculation of their evolution.

A correct approach should be to statistically analyse the accumulated fatigue of all the freespans observed in the simulations and to extrapolate the probability to reach unacceptable fatiguedamage, which should be compared with a reference acceptability threshold defined for theproject (which can depend on many factors, such as remoteness of the location, possibility ofinspection, etc.). For the analysed test case, a three-parameter Weibull distribution (figure 11) hasbeen fitted to the accumulated fatigue of all the occurred free spans in the twenty 6-years longsimulations.

From the distribution, it can be assessed that the resulting probability of exceeding theallowable fatigue limit of 1 in the 20 years design life of the pipeline is about 10%. Comparingthis result with the maximum accumulated fatigue of 0.25 obtained with the hindcasted timeseries, it can be deduced that the adopted methodology is effective for investigating theissue of pipeline fatigue damage for scour-induced free spans, helping the engineer in thedesign concept and decisions and being a valuable tool for a rational and adequate inspectionsurvey planning.

5. Pipeline free-span design approachFree-span management starts at the early project phases including pre-engineering survey, designand construction, and continues throughout the whole operating life. The pre-engineering surveyand its interpretation are the first steps of a rational, technical and cost-effective free-span integritymanagement. A detailed three-dimensional representation of the seabed and a characterizationof the shallow sediments are factors for a reliable assessment of the free-span configuration. In



17


.........................................................

addition, the collection of current and wave data, the investigation of sand-wave mobility (whenapplicable) and the assessment of scouring potential are also to be covered in the pre-engineeringand early engineering phases.

The design phase is the core of the free-span integrity management. The design objective is toassess the free-span scenario by using state-of-the-art predictive models and define the actions tobe undertaken during construction and during the operating life to ensure integrity according tothe target safety requirements.

During construction, a number of actions should be undertaken, to ensure the as-installedpipeline and as-built intervention works satisfy the design requirements. This includes specificrequirements of lay tension control, tight positioning and touchdown point monitoring, andintervention work construction tolerances.

Free-span integrity management continues throughout the whole operating life. A properinspection and management plan needs to be developed as part of and following the designphilosophy. When free spans are generated by the seabed roughness and no particular tendencyof seabed modification is envisaged, the inspection is just supposed to check for any trace ofaccidental events that might jeopardize the integrity of the free-spanning pipeline. Conversely, theinspection can play an active role in those cases where the seabed or pipe/seabed configurationare prone to modifications, such as in the presence of mobile sand waves or scouring phenomena.Throughout all the integrity management, state-of-the-art survey and analysis tools are the keyfactor for a reliable assessment of the actual free-span condition and evaluation of free-span stateof stress and residual fatigue life.

The Det Norske Veritas Recommended Practice [28] reflects the results from the various R&Dstudies and JIPs performed in the past three decades, providing the basis for the developmentof accurate structural response models referring to three-dimensional bottom roughness dataand shallow soil characterization. Today, there are a considerable number of free-spanningpipelines in satisfactory operation, testifying to the adequacy of design prediction versus actualpipe configuration, and thus proving the robustness of design criteria, analysis methods andengineering tools as well.

The outcome of the R&D efforts and project experiences of more than 30 years as concernsdesign criteria and analysis methodology have been implemented into a full set of computerprograms, covering all the relevant aspects of the free span. These programs constitute a packageof interfaced modules with fully compatible input/output. The main modules are the following:(i) pipeline routing module, (ii) on-bottom roughness module, (iii) modal analysis module,(iv) pipeline dynamic response, and (v) fatigue assessment module.

(a) Route selection and optimizationA detailed route optimization is necessary to minimize free spans and stress on the pipelineand reduce the need for mitigation works. Key factors in an efficient search for the optimumroute are a detailed mapping of the seabed, and efficient tools for the preparation andprocessing of a digital terrain model, drawing curved routes and in-place stress analyses ofthe pipeline.

Routing software has been developed with the aim at providing efficient assistance to thepipeline designer to handle bathymetric data and other data delivered from the survey, and todesign and manage route alignment and intervention works. The morphological information ofthe sea bottom acquired in a digital form is used to create a data base to be used for furtherprocessing activities. The stored data allow, after a pre-selection of possible route alternatives,the generation of automated profiles for the subsequent stress analysis (figure 12). Owing to thehigh degree of automation, several route alternatives can be compared in a short time. Maps andthree-dimensional drawings of the seabed, including bathymetric lines, morphological elementsand pipeline routes relevant to each alternative, as well as pipe–seabed profiles and alignmentsheets, are automatically produced at the requested scale.



18


.........................................................

C16.5 eastC16.5 westC14.0

Natura 2000

Finland TW

Finland/Estonia EEZ

Kalbadagrund corridor

Figure 12. Pipeline route optimization along the Kalbadagrund corridor in the Finnish EEZ of the Nord StreamPipeline, adaptedfrom [47]. (Online version in colour.)

(b) Bottom roughness analysis and free-span configurationNotwithstanding any route optimization effort, free-spanning sections of the pipeline inducedby the seabed roughness can often be unavoidable. In these cases, free-span assessment becomesone of the main design steps. The assessment of free-spanning pipeline integrity requires thefollowing criteria to be addressed:

— strength check, according to the applicable codes, against the stress and strain resultingfrom the combination of permanent (from seabed roughness), functional (flow assurance)and environmental (current and waves) loads; and

— fatigue damage check, to be assessed considering load cycles and dynamic response dueto vortex shedding and direct surface wave actions as well as the applicable S–N (stress–number of cycles) curve for the girth welds.

In general terms, the design approach required to verify the satisfaction of strength and fatiguecriteria of a free-spanning pipeline includes the following steps:

— a static analysis to obtain the static configuration and stress of the pipeline;— an eigenvalue analysis, which provides natural frequencies and corresponding modal

shapes for in-line and cross-flow vibrations modes; and— a response analysis, which evaluates the pipeline response to VIV and direct waves

and calculates stress amplitude for the evaluation of the environmental contribution tostrength check and for the calculation of fatigue damage.

The static analysis is nowadays always performed with a dedicated finite element model (FEM)program. The strength check against permanent and functional loads is then calculated by aproper post-processing (depending on the applicable design criteria and code) of the bendingmoment, axial force, stress and strain resulting from the static FEM analysis. Conversely, thedynamic contribution of the environmental loads to the strength check and to the fatigue checkmight still be assessed with different approaches, which basically differ from the methodologyapplied for the definition of the modal parameters (frequency and modal shape). Basically twodifferent approaches can be used depending on the pipeline free-span configuration:

— the parametric single-span approach; and— the multi-span approach based on FEM analysis on the actual bottom roughness [48,49].



19


.........................................................

C4.0 east route

–75

–70

–65

–60

–55

–50

–45

–40

216 800

dept

h (m

)

217 000 217 200 217 400 217 600 217 800 218 000 218 200

KP (m)

0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

unity

che

ckLC unity check

pipeline profileseabed profileintervention works

Figure 13. Example of bottom roughness analysis for a pipeline laid on an uneven sea bottom. Pipeline configuration withintervention works and local buckling unity check as per [50]. KP= kilometre post. (Online version in colour.)

(c) Static (on-bottom roughness) analysisAs the first stage, bottom roughness analyses are generally carried out by means of a dedicatednonlinear FEM-based program on the detailed profile extracted from the three-dimensionalmodel of the seabed. The FEM-based tool needs to include algorithms capable of modelling steelmaterial, geometric and pipe–soil interaction nonlinearities. For the geometric nonlinearities, theclassic beam theory of large rotations taking into account the contribution of the effective axialforces to the flexural stiffness of single beam elements is applied. The FEM program also needsto take into account the initial (as-laid) pipeline configuration when analysing the various loadconditions after laying, i.e. flooding, hydrotest and operating as well as profile modifications(post-trenching/artificial supports); see as an example figure 13.

Figure 13 shows the intervention works, realized by gravel heaps, that were necessary tocorrect the bottom–pipeline configuration in order to avoid unacceptable bending moment. Theintervention works have to be pre-lay in case of unacceptable stress found in as-laid configurationand can be post-lay if unacceptable stresses are found in flooding, hydrotest or operativeconditions. Additional gravel intervention works could be necessary to correct free spans thatdo not originate an unacceptable bending moment but are expected to develop excessive fatiguedamage by cyclic vibrations during the life of the pipeline.

In the past 15 years, based on the North Sea experience, a three-dimensional advanced FEMtool, based on standard structural codes such as ABAQUS, has been developed to properlymodel the upheaval, the lateral turning down and the lateral buckling development at theshoulders of free spans of pipelines subject to high pressure and high temperature conditions,as indicated in [51].

(d) Modal analysisThe modal/eigenvalue analysis is performed by linearizing the structural behaviour around thedeformed pipeline configuration with regard to the applied load conditions and the stiffness ofthe elements resembling the pipe–soil interaction. The stiffness of elements resembling pipe–soilinteraction is derived from the dynamic soil stiffness per unit length (axial, lateral and vertical)taking into account the effective pipe–soil contact length as evaluated from the static analysis.For each pipe element, the added mass coefficient is calculated accounting for the local value ofseabed proximity; see as an example figure 14.

The calculation of the eigenvalues and associated eigenvectors is based on the solutionof the classical homogeneous system. The eigenvalue analysis is performed separately in thevertical and horizontal planes, uncoupled. The method is based on the search for the eigenvaluethat annuls the determinant and uses the properties of the Sturm series to properly count theeigenvalues. This approach is particularly efficient due to the banded nature of the dynamicstiffness matrix of a pipeline. Special care needs to be paid to the potential occurrence of veryclose eigenvalues, especially as concerns the identification of correct eigenvalues and associated



20


.........................................................

99 100 99 200 99 300 99 400 99 500 99 600 99 700 99 800 99 900 100 000 100 100KP (m)

cross-flow vibration mode - F = 0.370

–70

–60

–50

–40

–30

–20

–10

WD

(m

)

(a)

–1.0

–0.8

–0.6

–0.4

–0.2

0

0.2

0.4

0.6

0.8

1.0

ampl

itude

99 100 99 200 99 300 99 400 99 500 99 600 99 700 99 800 99 900 100 000 100 100KP (m)

cross-flow vibration mode - F = 0.415

–70

–60

–50

–40

–30

–20

–10

WD

(m

)

(b)

–1.0

–0.8

–0.6

–0.4

–0.2

0

0.2

0.4

0.6

0.8

1.0

ampl

itude

Figure 14. Example of multi-span modal analysis in the vertical plane: two modal shapes of vibrations and associated naturalfrequencies (F). Green line (•), seabed profile; red line (�), pipe profile; blue line (—), intervention works; magenta (�),dimensionless amplitude (A/D) of vibration. (Online version in colour.)

eigenvectors, owing to the presence of many free spans that have quite close eigenvalues andeigenvectors. On this basis, Vitali et al. [48] implemented a program with advanced techniques,checking the zero of the determinant, using the orthogonal property of the eigenvectors andcalculating Rayleigh’s quotient.

After the modal analysis, some auxiliary parameters are also calculated for eachmode/eigenvector, in order to determine pipeline response to vortex shedding, such as theequivalent mass and the mode shape weighting factor [28]. The soil contribution to modaldamping is determined by inserting viscous dashpots in parallel with springs resemblingsoil–pipe contact.

(e) Free-span dynamic response to hydrodynamic loadsOnce the modal response of the free-spanning pipeline is known, the response to VIV or directwave action is normally assessed according to the response model (for VIV) and force model (for



21


.........................................................

direct surface wave actions) given in [28]. Response and force models can be applied to evaluatethe stress amplitude under extreme load conditions (for instance, 100 year return period of currentand wave) or under a long-term probability distribution of current and waves. The former is usedto calculate the environmental contribution for the strength check (local buckling ultimate limitstate, ULS); the latter is used for the assessment of long-term fatigue damage.

The dynamic analysis of a free-spanning pipeline should be performed according to theresponse models (VIV) and force model (direct surface wave action) as stated in [28]. A toolhas been developed by Vitali et al. [48] performing fatigue damage analysis, fatigue screeningcheck and calculation of environmental bending moment for ULS checking. The main structuralparameters adopted as input for the analysis are the natural frequency, the mode shape(normalized stress amplitude, mode shape weighting factor, etc.), the equivalent mass and themodal damping. It incorporates an interface that allows extracting and loading all the modalparameters for each vibration mode directly from the output of the modal analysis program. Assuch, it allows one to perform a full dynamic analysis accounting for isolated or interacting spans(multi-span), single-mode or multimode response, cross-flow, in-line and cross-flow induced in-line VIV and response to direct surface wave action. The accumulated fatigue damage (or themarginal fatigue life capacity) is evaluated accounting for the environmental loads described interms of scatter diagrams of the short-term sea states (wave significant height and peak period)and three-parameter Weibull distribution of steady current at the pipeline level.

The environmental bending moment for ULS and screening checks are instead calculated forthe input extreme values of sea state and current. The wave-induced velocity spectrum at thepipe level is obtained through a spectral transformation of the waves at sea level using a first-order wave theory and accounting for wave spreading and directionality. Wave and currents arecombined by integrating each sea state over the long-term current distribution.

(f) Fatigue damage assessmentFatigue analysis of free spans usually involves two activities: the definition of load variationsin number and intensity, and the calculation of cumulative fatigue damage using Miner’s ruleand the appropriate S–N curve. For offshore pipelines or tubulars, synthetic load histories aregenerally built on the basis of the long-term distribution of sea states. In some circumstances,the load history may not be important, for instance for deep-water pipelines, where loads aregenerally due to vortex shedding caused by the impact of a steady near-bottom current. This givesrise to a sequence of blocks of stress cycles of equal amplitude. Therefore, it becomes necessaryto calculate the number and the time duration of each block from the long-term distribution ofnear-bottom currents.

In offshore pipeline technology, as indicated in [52], the F1 and D class S–N curves are usedfor the weld root and for the weld cap, respectively. These curves are based on experimental testsperformed in the 1970s for offshore tubulars of platforms and by recent full-scale tests performedfor steel catenary risers for deep-water applications.

Figure 15 shows an example of the accumulated fatigue damage calculated using Miner’s ruleand [53] S–N class curves.

6. Some experiences on pipeline inspectionPeriodic pipeline investigation is fundamental to prevent any risk of failure and to guarantee thesafe condition of the pipeline over its whole operating life. Routine inspection and monitoringaim to ensure that:

— pipelines do not became defective or damaged;— damage or defects are detected before they cause a serious problem.

Some examples of pipeline inspections, chosen as representative of typical scenarios describedabove, are now presented.



22


.........................................................

unity

che

ck

0

0.2

0.4

0.6

0.8

1.0

1.2

–70

–60

–50

–40

–30

–20

–10

99 100 99 200 99 300 99 400 99 500 99 600 99 700 99 800 99 900 100 000 100 100

WD

(m

)

KP (m)

cross-flow total damage

Figure 15. Example of multi-span fatigue analysis performed. Green line (•), seabed profile; red line (�), pipe profile; blueline (—), intervention works; magenta (�), dimensionless unity check. (Online version in colour.)

The gas Transmed pipeline system joining Algeria to Italy includes five lines crossing the SicilyStraits from Cape Bon (Tunisia) to Cape Feto (Italy). The Northern Route maximum slope is about40%, whereas the Southern Route is about 35%. The maximum water depth is about 600 m. Thebottom lithology in the platform areas mainly consists of sand sediments with clayey inclusions.Generally, the seabed sediment in the deep section consists of soft silt.

Figure 16a shows the bottom of pipe and seabed profiles for two pipeline inspections in2000 and 2008. The pipeline profile does not change since the nature of the soil is typical ofsoil with high bearing capacity. The free-span scenario is time-invariant since it was frozenshortly after laying by gravel intervention works installed to correct the sea bottom–pipelineconfiguration in order to avoid over-bending and to interrupt too long free spans. Interventionworks can be identified in figure 16a by the spikes arriving above the bottom of pipe level. Insuch cases, the whole range of operational conditions should be checked, as any combinationmay be governing for the free-span design. Figure 16b shows a typical scour-induced free spanfor a pipeline laid on a sandy sea bottom. Once the free span was generated, it propagated alongthe pipeline reaching a length of about 25 m (1997). In 2008, the span length was about 40 m.This scenario is typical of intermediate water with medium-to-low hydrodynamic action. Thedevelopment of free spans is very slow, and limited or absent self-lowering can be expected.In such conditions, the exposure time may be taken as the remaining operational lifetime orthe time duration until possible intervention works will take place. Owing to risk of failurefor accumulated fatigue at KP 9275 m, an intervention work was completed in 2011 as shownin figure 16c.

The Blue Stream project, the gas pipeline system joining Russia to Turkey, includes twopipelines crossing the Black Sea. Along the E1 and W2 pipeline routes, the investigation limitshave been subdivided into sections or sub-areas. The Turkish Continental Shelf (EKP 354.9–378.5 and WKP 361.9–385.5) is the most interesting for scour-induced free span. A terraced shelfslope follows the plateau between 100 and 12 m water depth. The slope has an angle varyingapproximately between 1◦ and 3◦.

Figure 17 shows the E1 pipeline profile between KP 380.26 and KP 380.36 at approximate waterdepth of 37 m. A free span generated in 2000 propagated along the pipeline, reaching a length ofabout 30 m (inspection of 2003, figure 17a). In such a situation, free-span length is governed byvarious effects: (i) changing flow conditions, (ii) changing soil conditions, (iii) sinking at spanshoulder and (iv) sagging of the pipeline into the scour hole.



23


.........................................................

–76.0

–76.5

–77.0

–77.5

–78.0

–78.0

–79.0

–79.5

–66

–67

–68

–69

–70

–71

–72

–73

–74

KP (m)

KP (m) KP (m)

7500

9250 9260 9270 9280 9290 9300 9250 9260 9270 9280 9290 9300

7520 7540 7560 7580 7600 7620 7640 7660

WD

(m

)

WD

(m

)

BOP 2000 seabed 2000 BOP 2008 seabed 2008

seabed 1997 seabed 2008 BOP 2008BOP 1997 seabed 2008 BOP 2008 seabed 2011 BOP 2011

(b)

(a)

(c)

Figure 16. Transmed 1N pipeline. (a) Bottom of pipe (BOP) and seabed profiles in 2000 and 2008 inspections. From KP= 7500to 7660 m. (b) Bottom of pipe and seabed profiles in 1997 and 2008 inspections. (c) Bottom of pipe and seabed profiles in 2008and 2011 inspections. From KP= 9250 to 9300 m.

–35.0

–35.5

–36.0

–36.5

–37.0

–37.5

–38.0

–38.5

–39.0

–35.0

–35.5

–36.0

–36.5

–37.0

–37.5

–38.0

–38.5

–39.0

KP (m)

KP (m)

WD

(m

)

WD

(m

)

380 260 380 280 380 300 380 320 380 340 380 360

380 260 380 280 380 300 380 320 380 340 380 360


BOP 2007 seabed 2007 BOP 2008 seabed 2008KP (m)

380 260 380 280 380 300 380 320 380 340 380 360


(b)

(a)

(c)

Figure 17. Blue Stream pipeline. (a) Bottom of pipe (BOP) and seabed profiles in 2001 and 2003 inspections. (b) Bottom of pipeand seabed profiles in 2007 and 2008 inspections. (c) Bottomof pipe and seabed profiles in 2011 inspections. FromKP= 380 260to 380 360 m.



24


.........................................................

–98

–99

–100

–101

–102

–103

–104267 500

WD

(m

)

267 550 267 600 267 650 267 700 267 750 267 800

TOP 2010 seabed 2010 seabed 2008BOP 2010KP (m)

Figure 18. Åsgard Trasnport pipeline. Top of pipe (TOP), bottom of pipe (BOP) and seabed profiles in 2010 inspection. FromKP= 267 500 to 267 800 m.

The sequence of free-span development in 2007 and 2008 is shown in figure 17b. The free spanis observed to remain almost stable during the last 5 years (from 2003 to 2007). It seems thatfree-span development has come to an end, probably because the pipe reached a non-erodiblebed area. A considered stable condition may evolve in unstable conditions for two main reasons:(i) change in operational conditions and (ii) very extreme sea state. The inspection of 2011 showsthat the pipeline touched the bottom of the scour hole, thus stopping further development.Figure 17c shows that when the pipe reached the bed the length of the free span was cut in half.As soon as the gap between the pipe and the bed was closed, the backfilling process starts. Thedownstream side of the pipe was filled with the sediment forced by lee-wake erosion. Pipe sinkingat span shoulder is caused by soil failure.

In the North Sea, the Åsgard Transport (ÅT) system consists of an export riser base (ERB) at theÅsgard B platform, a 73 m 28 inch diameter pipeline to the ERB spool, a 28 inch/42 inch transitionpiece, a 684.1 km 42 inch diameter offshore pipeline and a shore approach at Kalstø. Scouringoccurred during the operating life of the offshore pipeline and a free span developed. The spangrowth velocity was estimated in the order of tens of metres per year. Within about 10 years ofoperative conditions, the scour process would come to an end and the so-called backfilling processwould start. As an example, considering the results of the 2010 survey, the backfilling process isobserved for KP 267.595 (figure 18). Hence, it can be reasonably assumed that any free span in asandy seabed has already developed and disappeared.

7. ConclusionIn many deep-water pipeline projects, free spans are one of the major challenges. The combinationof uneven–erodible seabed, near-seabed hydrodynamics and sediment transport and related,costly span intervention works requires large effort towards free-span interpretation. In the case ofseabed unevenness, the location where a span occurs is fixed and a detailed route optimization isnecessary to minimize free spans and stress on the pipeline and to reduce the need for mitigationworks. When the pipeline is laid on an erodible seabed, a span may exist in most locations,changing in location and time.

The evaluation of the fatigue accumulated by an evolving free span should be calculatedconsidering the actual environment causing its development. Some experiences in the NorthSea revealed that the main part of the fatigue is accumulated during the latter part of its lifewhen its length is maximum and caused by the same storm causing its disappearance, i.e. fatigueaccumulation and storm occurrence are strictly correlated in time.

The outcome of the R&D efforts and project experiences of more than 30 years as concernsdesign criteria and analysis methodology has been discussed. The paper is meant to give a focuson some aspects within the design of pipelines with free spans. It illustrates various themes



25


.........................................................

related to free-span analysis: (i) the characterization of the environment as driving factor, (ii) thehydrodynamic loads and erosion around a pipeline laid on an erodible seabed, and (iii) the criteriaand analysis module for design, static and dynamic. Some examples of pipeline inspection,chosen as representative of typical scenarios, are presented. It is not the intention to give acomprehensive overview, but some of the information given should be beneficial for engineersinvolved in pipeline design.

References1. Breivik J, Bruschi R, Leopardi G. 1995 Troll Oljerør—a unique technical challenge. In Proc. 2nd

Offshore Mediterranean Conf., Ravenna, Italy, 15–17 March. Ravenna, Italy: OMC.2. Bianchi S, Pulici M. 2001 Blue Stream—a challenging project for gas transmission. In Proc. 6th

Offshore Mediterranean Conf., Ravenna, Italy, 28–30 March. Ravenna, Italy: OMC.3. Gjertveit E, Holme R, Bruschi R, Zenobi D. 2007 The Langeled Project. In Proc. 17th Int. Offshore

and Polar Engineering Conf., Lisbon, Portugal, 1–6 July. Cupertino, CA: ISOPE.4. Bruschi R, Cimbali W, Leopardi G, Vincenzi M. 1986 Scour induced free span analysis. In Proc.

5th Int. Conf. Offshore Mechanics and Arctic Engineering, Tokyo, Japan, 13–18 April. New York, NY:ASME.

5. Bijker R, Staub C, Silvis S, Bruschi R. 1991 Scour induced free spans. In Proc. 23rd OffshoreTechnology Conf., Houston, TX, 6–9 May. Richardson, TX: OTC.

6. Bearman PW, Zdravkovic MM. 1978 Flow around a circular cylinder near a plane boundary.J. Fluid Mech. 89, 33–47. (doi:10.1017/S002211207800244X)

7. Sarpkaya T. 1976 Forces on cylinders near a plane boundary in a sinusoidally oscillating fluid.J. Fluids Eng. 98, 499–505. (doi:10.1115/1.3448383)

8. Sarpkaya T. 1977 In line and transverse forces on cylinders near a wall in oscillatory flow athigh Reynolds numbers. In Proc. 9th Annual Offshore Technology Conf., Houston, TX, 2–5 May.New York, NY: American Institute of Mining, Metallurgical and Petroleum Engineers.

9. Williamson CHK. 1985 Sinusoidal flow relative to circular cylinder. J. Fluid Mech. 155, 141–174.(doi:10.1017/S0022112085001756)

10. Tatsuno M, Bearman PW. 1990 A visual study of the flow around an oscillating cylinderat low Keulegan–Carpenter numbers and low Stokes numbers. J. Fluid Mech. 211, 157–182.(doi:10.1017/S0022112090001537)

11. Sumer BM, Jensen BL, Fredsøe J. 1991 Effect of a plane boundary on oscillatory flow around acircular cylinder. J. Fluid Mech. 225, 271–300. (doi:10.1017/S0022112091002057)

12. Sumer BM, Fredsøe J. 1997 Hydrodynamics around cylindrical structures, vol. xviii. Singapore:World Scientific.

13. Jas E, O’Brien D, Fricke R, Gillen A, Cheng L, White D, Palmer A. 2012 Pipeline stabilityrevisited. J. Pipeline Eng. 12, 259–268.

14. Palmer AC. 1996 A flaw in the conventional approach to stability design of pipelines. In Proc.19th Offshore Pipeline Technology Conf., Oslo, Norway, 15–16 February. London, UK: IBC.

15. Chiew YM. 1990 Mechanics of local scour around submarines pipeline. J Hydraulic Eng. 116,515–529. (doi:10.1061/(ASCE)0733-9429(1990)116:4(515))

16. Sumer BM, Truelsen C, Sichmann T, Fredsøe J. 2001 Onset of scour below pipelines and self-burial. Coastal Eng. 42, 313–335. (doi:10.1016/S0378-3839(00)00066-1)

17. Hansen EA. 1992 Scour below pipelines and cables. A simple model. In Proc. 11th OffshoreMechanics and Arctic Engineering Conf., Calgary, Canada, 7–12 June. New York, NY: ASME.

18. Sumer BM, Fredsøe J. 1990 Scour below pipelines in waves. J. Waterway., Port Coastal OceanEng. ASCE, 116, 307–323. (doi:10.1061/(ASCE)0733-950X(1990)116:3(307))

19. Sumer BM, Fredsøe J. 1996 Scour around pipelines in combined waves and current. In Proc.15th Offshore Mechanics and Arctic Engineering Conf., Florence, Italy, 16–20 June. New York, NY:ASME.

20. Fredsøe J, Sumer BM, Arnskov MM. 1992 Time scale for wave/current scour below pipelines.Int. J. Offshore Polar Eng. 2, 13–17.

21. Fredsøe J, Hansen EA, Mao Y, Sumer BM. 1988 Three-dimensional scour below pipelines.J. Offshore Mech. Arctic Eng. 110, 373–379. (doi:10.1115/1.3257075)

22. Cheng L, Yeow K, Zang Z, Teng B. 2009 Three-dimensional scour below pipelines in steadycurrents. Coast. Eng. 56, 577–590. (doi:10.1016/j.coastaleng.2008.12.004)


http://dx.doi.org/doi:10.1017/S002211207800244X

http://dx.doi.org/doi:10.1115/1.3448383

http://dx.doi.org/doi:10.1017/S0022112085001756



http://dx.doi.org/doi:10.1061/(ASCE)0733-9429(1990)116:4(515)

http://dx.doi.org/doi:10.1016/S0378-3839(00)00066-1

http://dx.doi.org/doi:10.1061/(ASCE)0733-950X(1990)116:3(307)

http://dx.doi.org/doi:10.1115/1.3257075

http://dx.doi.org/doi:10.1016/j.coastaleng.2008.12.004


26


.........................................................

23. Wu Y, Chiew YM. 2013 Mechanics of three-dimensional pipeline scour in unidirectionalcurrent. J. Pipeline Syst. Eng. Pract. 4, 1. (doi:10.1061/(ASCE)PS.1949-1204.0000131)

24. Sumer BM, Fredsøe J. 2002 The mechanics of scour in the marine environment. Singapore: WorldScientific.

25. Gravesen H. 1985 A case in pipeline design. In Advances in Offshore Oil and Gas PipelineTechnology (ed. RF de la Mare). London, UK: Gulf.

26. Bruschi R, Curti G, Tura F. 1991 Free spanning pipelines: a review. In Proc. 1st Int. Offshore andPolar Engineering Conf., Edinburgh, UK, 11–16 August. Golden, CO: ISOPE.

27. Det Norske Veritas. 1981 Rules for submarine pipeline system.28. Det Norske Veritas. 2006 Free spanning pipeline. Recommended Practice DNV-RP-F105.29. Mørk KJ, Vitali L, Verley R. 1997 The MULTISPAN Project: design guideline for free spanning

pipelines. In Proc. 16th Int. Conf. Offshore Mechanics and Arctic Engineering, Yokohama, Japan,13–17 April. New York, NY: ASME.

30. Mørk KJ, Fyrileiv O, Chezhian M, Nielsen FG, Søreide T. 2003 Assessment of VIV inducedfatigue in long free spanning pipelines. In Proc. 22nd Int. Conf. Offshore Mechanics and ArcticEngineering, Cancun, Mexico, 8–13 June. New York, NY: ASME.

31. Hansen EA, Bryndum M, Mørk KJ, Verley R, Sortland L, Nes H. 2001 Vibrations of freespanning pipeline located in the vicinity of a trench. In Proc. 19th Int. Conf. Offshore Mechanicsand Arctic Engineering, Rio de Janeiro, Brazil, 3–8 June. New York, NY: ASME.

32. Fyrileiv O, Chezhian M, Mørk KJ, Arnesen K, Nielsen FG, Søreide T. 2004 New free spandesign procedure for deepwater pipelines. In Proc. 16th Deep Offshore Technology (Int. Conf. &Exhib.), New Orleans, LA, 30 November–2 December. Waltham Abbey, UK: Penwell.

33. Fyrileiv O, Chezhian M, Mørk K. 2005 Experiences using DNV-RP-F105 in assessment of freespanning pipelines. In Proc. 24th Int. Conf. Offshore Mechanics and Arctic Engineering, Halkidiki,Greece, 12–17 June. New York, NY: ASME.

34. Chezhian M, Mørk KJ, Fyrileiv O, Nielsen FG, Søreide T. 2003 Assessment of deepwater multi-spanning pipelines—Ormen Lange experience. In Proc. 15th Deep Offshore Technology (Int. Conf.& Exhib.), Marseille, France, 19–21 November. Waltham Abbey, UK: Penwell.

35. Sleath DF. 1984 Sea bed mechanics. New York, NY: John Wiley & Sons.36. Soulsby R. 1997 Dynamics of marine sands. London, UK: Thomas Telford.37. Det Norske Veritas. 2010 On-bottom stability design of submarine pipelines. Recommended

Practice DNV-RP-F109.38. Tura F, Dumitrescu A, Bryndum MB, Smed PF. 1994 Guidelines for free spanning pipelines:

the GUDESP Project. In Proc. 13th Int. Conf. Offshore Mechanics and Arctic Engineering, Houston,TX, 27 February–3 March. New York, NY: ASME.

39. Mao Y. 1986 The interaction between a pipeline and an erodible bed. Series paper 39. Denmark:Technical University of Denmark.

40. Mattioli M, Alsina JM, Mancinelli A, Miozzi M, Brocchini M. 2012 Experimental investigationof the nearbed dynamics around a submarine pipeline laying on different types of seabed:The interaction between turbulent structures and particles. Adv. Water Res. 48, 31–46.(doi:10.1016/j.advwatres.2012.04.010)

41. Mattioli M, Mancinelli A, Brocchini M. 2013 Experimental investigation of the wave-induced flow around a surface-touching cylinder. J. Fluids Struct. 37, 62–87. (doi:10.1016/j.jfluidstructs.2012.10.002)

42. Hansen EA, Staub C, Fredsøe J, Sumer BM. 1991 Time development of scour induced freespans of pipelines. In Proc. 10th Offshore Mechanics and Arctic Engineering Conf., Stavanger,Norway, 23–28 June, vol. 5, pp. 25–31. New York, NY: ASME.

43. Kjeldsen SP, Gjorsvik O, Bringaker KG, Jacobsen J. 1973 Local scour near offshore pipelines. InProc. 2nd Int. Port and Ocean Engineering Under Arctic Conditions Conf., Reykjavik, Iceland, 27–30August. Reykjavik, Iceland: University of Iceland.

44. Klomp WHG, Hansen EA, Chen Z, Bijker R, Bryndum MB. 1995 Pipeline seabed interaction,free span development. In Proc. 5th Int. Offshore and Polar Engineering Conf., The Hague,The Netherlands, 11–16 June. Golden, Co: ISOPE.

45. Cheng L, Yeow K, Zang Z, Li F. 2014 3D scour below pipelines under waves and combinedwaves and currents. Coast. Eng. 83, 137–149. (doi:10.1016/j.coastaleng.2013.10.006)

46. Drago M, Pigliapoco M, Ciuffardi T. 2007 Analysis of pipeline fatigue damage for scourinduced freespans. In Proc. 17th Int. Offshore and Polar Engineering Conf., Lisbon, Portugal, 1–6July. Cupertino, CA: ISOPE.


http://dx.doi.org/doi:10.1061/(ASCE)PS.1949-1204.0000131

http://dx.doi.org/doi:10.1016/j.advwatres.2012.04.010

http://dx.doi.org/doi:10.1016/j.jfluidstructs.2012.10.002

http://dx.doi.org/doi:10.1016/j.jfluidstructs.2012.10.002

http://dx.doi.org/doi:10.1016/j.coastaleng.2013.10.006


27


.........................................................

47. Bruschi R, Pettinelli D, Pigliapoco M, Vitali L. 2013 Free spanning pipelines—Three decadesof design and integrity management. In Proc. Rio Pipeline Conf. and Exhibition, Rio de Janeiro,Brazil, 24–26 September, IBP 1475, 13. Rio de Janeiro, Brazil: Instituto Brasileiro de Petroleo.

48. Vitali L, Mørk KJ, Verley R, Malacari LE. 1997 The MULTISPAN Project: response models forvortex induced vibrations of submarine pipelines. In Proc. 16th Int. Conf. Offshore Mechanicsand Arctic Engineering, Yokohama, Japan, 13–17 April. New York, NY: ASME.

49. Vitali L, Marchesani F, Curti G, Bruschi R. 1993 Dynamic excitation of offshore pipelinesresting in very uneven seabeds. In Proc. Eurodyn, Trondheim, Norway, 21–23 June (eds T Moanet al.). Rotterdam, The Netherlands: AA Balkema.

50. Det Norske Veritas. 2012 Submarine pipeline system. Recommended Practice DNV-OS-F101.51. Det Norske Veritas. 2007 Global buckling of submarine pipelines. Structural design due to

high temperature/high pressure. Recommended Practice DNV-RP-F110.52. Det Norske Veritas. 2009 Riser interference. Recommended Practice DNV-RP-F203.53. Det Norske Veritas. 2010 Fatigue design of offshore steel structures. Recommended Practice

DNV-RP-C203.



Insightsonthedesignof...

Documents

Transcript of Insightsonthedesignof...