riser test

Upload: abraham-rojas

Post on 10-Mar-2016

47 views

Category:

Documents


2 download

DESCRIPTION

Riser fatigue

TRANSCRIPT

  • On the Test Parameters for Flexible

    Riser in-plane Bending Fatigue Testing

    E. Bendiksen, G. Agustsson C. Bergenstof Nielsen

    National Oilwell Varco, Denmark1 (NOV)

    www.nov.com

    ABSTRACT The fatigue design of the steel armour layers is an important factor in flexible riser design.

    Typically, the riser must be able to endure the loads of a field condition for a period of 20

    years. In flexible pipe design, this is analysed in global and local flexible pipe fatigue

    modelling. However, modelling needs to be supported by proper testing. One of the

    widespread test methods is the so-called in-plane bending fatigue testing which is one of the

    important qualification tests for flexible pipes. In spite of this, literature contains only limited

    guidance about the test procedures. Some guidance is found in API 17B/ ISO 13628-11. It is

    the aim of the present paper to provide more specific guidance on the selection of test

    parameters. A result of the review of the test parameters is a clear picture of the coverage of a

    specific test. This is important when judging whether a new in plane test is required for a

    specific pipe / project. Since there is no unique relation between field data and test

    parameters, one test programme will cover different fields and vice versa. In order to

    intelligently transfer the knowledge obtained in one test to different pipe designs and different

    fields it is necessary to understand the steps in the fatigue model, and these steps are

    schematized here-in. The focus is on the calculation of armour wire stress. NOV Flexibles

    has, over the past 12 years, built up knowledge on the modelling of the real strain response

    during cyclic bending and tension. This has been done using fibre bragg gratings embedded in

    the tensile armour layers. In this work it has been shown that NOV Flexibles is able to model

    the tensile armour stresses at low and high angles which is necessary to conclude from test to

    different field applications. The conclusions drawn in the paper are supported by examples

    from actual fatigue testing.

    1 Formerly known as NKT Flexibles

  • INTRODUCTION A study with the aim to establish the current state-of-the-art on flexible pipe integrity has

    been carried out by MCS Kenny in 2010 [3]. The study has compiled a global population

    database of flexible pipe usage and failure damage statistics. This data base includes 1900

    dynamic risers and 1400 static lines from over 130 production facilities. From the data base

    315 individual damages or failure incidents were reported. The incidents have been split up in

    damage mechanisms; see Figure 1. The histogram in Figure 1 shows 13 categories that

    represent the sum of 315 incidents. One category designated Other contains 25 incidents covered by 7 failure modes, one of which is armour wire failure. It is not possible to extract

    further information from the report with regard to the distribution within this group; however,

    if it is assumed that a seventh of the 25 incidents relate to armour wire failure, then of 315

    incidents approximately 3 wire failures have been reported. This corresponds to a failure

    percentage of approximately 1%.

    Figure 1 Failure/Damage mechanism distribution (source: State of the Art Report on

    Flexible Pipe Integrity, MCS Kenny, 2010, report number 2-4-5-013/SR01, [3].)

    The reason for this low failure percentage is probably that the fatigue design methodology of

    risers contains a number of steps that, depending on the level of knowledge within each step,

    comprises implicitly a level of conservatism/prudency. This level of conservatism adds up for

    each design step. It is of common interest to narrow the gap between the fatigue design

    methodology and real life operation of risers. In this way we get lighter designs that mitigate

    fatigue or, everything else being equal, we get increased lifetime.

    This is why it is of great interest to look closely at each design step of the fatigue design

    methodology.

    In this paper we have chosen to present our ideas on what to consider in connection with full

    scale in-plane bending fatigue testing (IPBT), and why we think that the results from such a

    test can be used to qualify pipes to more than just one application.

  • FATIGUE ANALYSIS OF FLEXIBLE PIPES

    Design methodology The fatigue design methodology for a flexible riser is based on the requirements in API 17J

    [1], the guidance in API 17B [2] and the ever evolving industry practice. The process is in

    nature iterative, but for ease of understanding it is described here as a sequence. This process

    may involve hundreds of engineering hours and computer simulation time. The description

    here is a simplification of the actual process. The process, as shown in Figure 2, can be

    described as a three step process:

    1. The first step involves collection of the environmental data. The environmental load on the flexible pipe comes through vessel movements and direct wave / current

    loading. Wind obviously plays a role in the loading but only in-directly. The vessel

    response is normally divided in a static off-set from its mean position (from wind and current), and a first order movement (from wave action). The environmental loads

    are given in statistical tables for the actual position of the vessel and the first step

    consists of condensing these wave scatter diagrams for different load directions into a

    table of load cases. In this table, load cases are described with a number of waves,

    wave height and wave period. This load case table needs to also describe how wave

    direction, off-sets, current and vessel loading are to be combined with the wave

    loading.

    2. The second step consists of global analyses of the riser. This can be done in commercial programs like Orcaflex, Flexcom 3D, and Riflex. The inputs to these

    analyses in addition to the load case table are:

    Information about the vessel (RAO, vessel draught etc., geometry of interface between vessel and riser system)

    Information about the riser system itself (length of pipe, buoyancy, etc.)

    Geometry and material data of the bend limiting device (bend stiffener or bell mouth)

    Properties of the flexible pipe (weight, diameter, stiffness, etc.)

    The outcome of the global analyses is tables of corresponding values for tension and

    angle or curvature of the flexible pipe. The values will normally be given for the

    fatigue critical areas: Hang off, hog/ sag bend and touch down point.

    3. The final step is a local analysis of the flexible pipe for these global loads. Local refers to the fact that it is an analysis of a limited length of the flexible pipe. The

    analysis can be more or less simplified, see for example [4], [5] and [6]. In the

    analysis, the global loads are applied to a model of the pipe. This means that the

    analysis needs input on the detailed geometry including material data for the flexible

    pipe cross section and the bend limiting device (bend stiffener or bell mouth).

    Furthermore, a very important parameter is the pressure that the pipe is exposed to.

    Temperature also plays a role because the material properties are temperature

    dependent.

    Using these data, the angles for each load case are transformed to curvature.

  • Secondly the cross section(s) are analysed for the curvatures and the armour wire

    deflections and stresses are determined (mean value and range).

    The next step in the local analysis is to calculate the damage for the stresses. The link

    between these two is the armour wire SN-curve. Derivation of SN-curves is a subject

    on its own, see [7]. A very important parameter in SN-curve selection for flexible pipe

    armour wires is the annulus environment. The SN-curve is found by testing in the

    actual environment. Another aspect to consider is that there is a mean stress and a

    variation over time and both have an influence on fatigue performance. The mean

    stress can be taken into account through for example a Gerber correction of the stress

    variation, [7], or by doing SN-testing at the actual mean stress.

    The last step of the local analysis is to sum up the damage of the contribution from all

    load cases. This is normally done using the Palmgren-Miner rule, ref API 17J [1].

    The maximum allowable damage over the lifetime is 0.1, i.e. there is a safety factor of

    10 on the design life.

    The analysis can be carried out as a regular approach or an irregular approach. The

    description above covers the regular approach. However, the scheme for the irregular

    approach is very much the same except the analyses differ and some additional

    complexity is added (mainly in selection of load case but also in counting of number

    of cycles and computational efforts).

    Figure 2 The fatigue design procedure schematically

    No a_mean a_range

    mean

    Tension

    Tension

    range

    TDP

    Environmental data

    Selection of global fatigue

    load cases in-house programs

    Global analysis

    ORCAFLEXLocal analysis

    BFLEXAngle to curvature

    Curvature to stress

    Stress to damage

    Mean stress

    Damage summation

    Vessel and mooring data

    Configuration

    Bend stiffener

    Pipe

    Operational data

    Permeation

    analysis

    SN-test

    TDP Damage = 0.1HO Damage = 0.1Hog/ sag Damage = 0.1

    HO Damage = 0.1

    Hs/Tp 0-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10-11 11-12 12-13 13-14 14-15 15-16 16-17 17-18 18-19 19-20 >20 Sum

    0-1 5 37 108 176 197 172 126 82 49 28 15 8 4 2 1 1010

    1-2 2 35 170 402 585 612 506 353 218 123 65 32 16 7 3 2 1 1 3133

    2-3 3 38 158 343 475 473 371 243 139 72 35 16 7 3 1 2377

    3-4 3 30 113 232 304 286 209 126 66 31 13 5 2 1 1421

    4-5 3 24 83 159 195 170 114 62 29 12 4 1 856

    5-6 3 18 56 99 113 90 55 27 11 4 1 477

    6-7 2 12 35 57 59 43 23 10 3 1 245

    7-8 1 7 20 30 28 18 9 3 1 117

    8-9 1 4 11 14 12 7 3 1 53

    9-10 1 2 5 6 5 2 1 22

    10-11 1 2 2 2 1 8

    11-12 1 1 1 1 4

    12-13 0

    13-14 0

    14-15 0

    Sum 0 7 75 319 769 1265 1594 1637 1429 1084 722 426 224 106 43 17 4 1 1 9723

    Hs/Tp 0-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10-11 11-12 12-13 13-14 14-15 15-16 16-17 17-18 18-19 19-20 >20 Sum

    0-1 5 37 108 176 197 172 126 82 49 28 15 8 4 2 1 1010

    1-2 2 35 170 402 585 612 506 353 218 123 65 32 16 7 3 2 1 1 3133

    2-3 3 38 158 343 475 473 371 243 139 72 35 16 7 3 1 2377

    3-4 3 30 113 232 304 286 209 126 66 31 13 5 2 1 1421

    4-5 3 24 83 159 195 170 114 62 29 12 4 1 856

    5-6 3 18 56 99 113 90 55 27 11 4 1 477

    6-7 2 12 35 57 59 43 23 10 3 1 245

    7-8 1 7 20 30 28 18 9 3 1 117

    8-9 1 4 11 14 12 7 3 1 53

    9-10 1 2 5 6 5 2 1 22

    10-11 1 2 2 2 1 8

    11-12 1 1 1 1 4

    12-13 0

    13-14 0

    14-15 0

    Sum 0 7 75 319 769 1265 1594 1637 1429 1084 722 426 224 106 43 17 4 1 1 9723

    Hs/Tp 0-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10-11 11-12 12-13 13-14 14-15 15-16 16-17 17-18 18-19 19-20 >20 Sum

    0-1 5 37 108 176 197 172 126 82 49 28 15 8 4 2 1 1010

    1-2 2 35 170 402 585 612 506 353 218 123 65 32 16 7 3 2 1 1 3133

    2-3 3 38 158 343 475 473 371 243 139 72 35 16 7 3 1 2377

    3-4 3 30 113 232 304 286 209 126 66 31 13 5 2 1 1421

    4-5 3 24 83 159 195 170 114 62 29 12 4 1 856

    5-6 3 18 56 99 113 90 55 27 11 4 1 477

    6-7 2 12 35 57 59 43 23 10 3 1 245

    7-8 1 7 20 30 28 18 9 3 1 117

    8-9 1 4 11 14 12 7 3 1 53

    9-10 1 2 5 6 5 2 1 22

    10-11 1 2 2 2 1 8

    11-12 1 1 1 1 4

    12-13 0

    13-14 0

    14-15 0

    Sum 0 7 75 319 769 1265 1594 1637 1429 1084 722 426 224 106 43 17 4 1 1 9723

    No Height Period additional

    off-set

    direction

    current

    No a_mean a_range

    mean

    Tension

    Tension

    range

    Hog/ Sag

    No a_mean a_range

    mean

    Tension

    Tension

    range

    Hang Off

  • FATIGUE TEST

    Why make fatigue tests Above it is explained how the flexible pipe is designed against fatigue. IPBTs are made in

    order to demonstrate that the design procedure leads to a safe operation for the planned design

    life. This validation can be done in several ways: 1. A validation of the methodology for

    fatigue design or 2. A validation of an actual pipe for loads similar to the actual loading. API

    17B defines the two types of testing:

    1. For the service-life model validation, API 17B recommends the use of a single load block using approximately 400,000 cycles of one load block to arrive at a fatigue damage

    of 1.0 in the most fatigue susceptible layer.

    2. For qualification of a certain pipe design (for a given application), a service simulation is run to the actual calculated fatigue damage for the pipe in operation (or 0.1 if this is not

    available). For the service life simulation, API 17B recommends 2-4 million cycles

    divided in to load blocks based on the pipe application in the field.

    Some modifications are often done relative to the above definitions. In order to demonstrate a

    safety factor, the service simulation test can be extended to a higher damage than experienced

    in-service. The service-life model validation test is often divided in to a number of load

    blocks of different sizes. Finally, an IPBT is often a two stage test where the first stage is a

    service simulation test and the second part is a service-life model validation test, this

    possibility is also described in API 17B. The damage in the service simulation test is

    representative for operating condition, i.e. it is run to a damage of 0.1 in the most fatigue

    susceptible layer. Then the test is continued into a service life model validation test. This part

    runs to a damage of 1.0 or until pipe failure, whichever comes first.

    How are fatigue tests made In an IPBT a pipe sample of approximately 15 m is tested. The pipe sample is complete with

    all layers and end fittings. As shown in Figure 3, the pipe is pressurized, tensioned and bent

    back and forth a number of times. The pipe sample is normally equipped with a bend limiter,

    i.e. a bend stiffener or a bell mouth, in order to avoid very concentrated bending of the pipe.

  • Figure 3 Basic principle of an in-plane bending fatigue test

    The testing is done in a dedicated test rig, see Figure 4. The test can be carried out with the

    pipe in a horizontal or vertical layout.

    Usually the same pressure is kept throughout the entire test.

    For practical reasons, bending is typically applied at one end of the pipe (top), while the other

    end is tensioned (bottom). Normally, the top end is fixed against the twist and the other end is

    allowed to rotate freely.

  • Figure 4 A fatigue test rig, [2]

    Usually, the bending plane is kept the same throughout the testing. This simplifies the test

    setup and accelerates the test since fatigue damage is concentrated on a small section on the

    pipe.

    API 17B recommends that the pipe length between the bend limiter and top end-fitting should

    be at least one pitch of the outer tensile armour wires, and between bend stiffener and bottom

    end-fitting at least three pitches. These are rules of thumb; care should be taken and it should

    be considered to analyse the effects on wire stresses from end-fittings if the sample is short

    relative to these recommendations.

    It is important to extensively monitor the loading, the pipe response and its integrity. This is

    further discussed in a section on acceptance criteria. The monitoring is accompanied with

    extensive high frequency logging of the data and software for presenting it in an easy and

    accessible way.

  • TEST PARAMETERS, LOAD CASES AND ACCELERATION

    OF TEST

    General The test programme for fatigue testing is a compromise between keeping loading realistic and

    applying a sufficient load to obtain the specified fatigue damage within an affordable number

    of load cycles. This acceleration of damage can be done in several ways. The field data of one

    field can be transferred to a test programme in different ways so that different test

    programmes can be representative of a certain field. The reverse is also true that a certain test

    programme can be representative of several different fields, both with respect to production

    data and environmental data.

    The parameters involved in the test are:

    The pipe design

    The bend stiffener design

    Temperature

    The curvature of the pipe (mean value and amplitude)

    Applied tension (mean value and amplitude)

    Internal pressure

    The number of cycles

    The remainder of this section explains the selection of test parameters.

    Pipe design/ test sample A certain pipe design would normally be taken as covering another application when the ID is

    within plus/minus 2 inches, the layer sequence is the same, and the wire geometry is the same

    while wire size and grade may differ, ref. API 17B (Table 20 and section 9.4.4.1).

    Bending restriction device; bend stiffener, bell-mouth The bend limiting device serves the same purpose as during operation: Protection of the pipe

    against excessive bend radii, especially close to the end fittings. The bend limiting device

    plays a large role in the relation between test rig angle and pipe curvature.

    On some project specific IPBT the test has been run with the actual bend stiffener. However,

    this can be unpractical because the bend stiffener may be so big that the length limitations of

    the test rig means that it can be difficult to have a free section of one pitch length above the

    bend stiffener and a three pitch length below. Furthermore, the bend stiffener is one of the

    links between the rotational angle of the pipe ends and the curvature of the pipe and thereby

    between the angle and stress fatigue damage. Therefore, a very large bend stiffener will lower

    the damage or a smaller bend stiffener can be a way of accelerating the test.

    Temperature API 17B does not recommend any specific bore temperature for the test and the temperature

    is considered a secondary parameter whose effects can be analysed and tested separately. The

    temperature distribution in the pipe influences the bend stiffness and therefore the relation

    between applied angle and obtained curvature.

  • API 17B recommends that a thermal analysis is carried out in order to establish the

    temperature increase in different layers due to friction between them.

    Active cooling or heating can be coupled to the test specimen in order to control the

    temperature of the bore and/or different layers.

    Bore fluid Water or oil can be used as a bore medium. The impact on SN-curve from diffusion of the

    bore medium into the pipe annulus must be taken into account.

    Frequency Due to cost and time considerations, a test will usually be run at the highest possible

    frequency. The frequency can be limited by the test rig itself and because friction may build-

    up temperature in the pipe wall.

    Test program The four remaining parameters are the test parameters entering the test program. It is

    important that the loading is realistic in order to test for the true failure modes. However, we

    want to apply a damage of say 1.0 in 3 million cycles where in the field a damage of 0.1 in

    100 million cycles is normal (and in the field the SN-curve is normally worse due to corrosion

    fatigue where the test is traditionally run with non-corrosive annulus environment). As a

    consequence we need a damage rate in the test which is 333 times larger in the test than in

    reality. Some acceleration of the damage is therefore necessary. The acceleration can be done

    by increasing the load in terms of pressure, tension and/ or curvature. Most often a

    combination is used.

    Internal pressure

    API 17B recommends that the highest expected operational pressure be used. However, bore

    pressure is usually a means for acceleration of the test and it is therefore selected

    conservatively, usually near the design pressure of the pipe.

    Number of cycles

    The number of cycles recommended in API 17B is in the range of 2-4 million cycles for a

    service simulation test and 400,000 cycles for a service life model validation test. It is

    difficult to argue against API 17B, although in the common situation where the two test types

    are combined a distribution with half the cycles in the service simulation test and half in the

    service life model validation test could result in a more reasonable loading pattern.

    Tension

    According to API 17B the tension should be selected conservatively from the global load

    cases. It is often said that tension and pressure must be selected so that the loading of the

    actual application is covered. It is not necessary to cover the extreme loading as this is not

    representative for fatigue, but tension corresponding to the normal operation should be

    represented, although it is the authors opinion that it is more relevant to represent the stress and contact pressure levels. Tension can be used as an acceleration parameter for the test

    since a higher mean tension will increase contact pressure and raise the mean stress in the

    tensile armour wires and thus accelerate the fatigue damage.

  • Curvatures and angles

    The angle is the load applied in the test rig, while the curvature of the pipe is the parameter

    giving the armour wire stress. The link between the two is the bend stiffness of the pipe and

    bend stiffener.

    Non-zero mean angles during fatigue testing may result in pig tailing of the relatively short

    test sample and should be applied with care.

    A guidance in API 17B is that the extreme angles of the actual application are represented in

    the test; however, this is not always practical because it may create very large loads in the

    test. Furthermore, as explained above, the relevant parameter seen from the pipe is more the

    curvature or the stress. The purpose of the fatigue test is fatigue validation and not extreme

    load validation. As a consequence, the industry often relaxes this API 17B recommendation.

    Furthermore, as explained above, API 17B allows a service life model validation test to be

    run at one load case only, without any requirement for this load case to look like in-field

    loads. Nevertheless, most people would consider a certain design qualified through service

    life model validation tests.

    During the traditional 20 year design life, a flexible riser will see in the order of magnitude

    100 million load cycles. The majority of these will result in very low angles and curvatures.

    Angles and curvatures which are difficult to represent with reasonable precision in a fatigue

    test rig and if it was possible they would result in a fatigue damage corresponding to that of

    the riser in-service (i.e. maximum 0.1).

    Normally a distribution where there are many cycles of the small load cases and fewer of the

    large ones is aimed at (similar to what is seen on the wave scatter diagram in Figure 2 upper

    left corner). Angles are selected so that they are practical and result in the required damage in

    a series of load cases. This is not to say that the angles should be selected far away from what

    is experienced in normal service. In fact, one should aim for load cases that are in the same

    order of magnitude as that of the field. In some examples later this is shown in terms of stress

    or angle. One could have chosen curvature also.

    API 17 B warns against artificially improving the fatigue performance of the armour wires by

    applying large load cases early in the test. We recommend to verify that the largest load case

    does not induce yielding of the armour wires but stays in the high cycle fatigue region and to

    mix the load cases in a logical way and mix large and small load cases from the beginning.

    ACCEPTANCE CRITERIA API 17B suggests several acceptance criteria for the pipe in IPBT, which ranges from carcass

    cracks to loss of up to 5% of the tensile armour wires. However, API 17B also emphasizes

    that the purchaser and the manufacturer should agree on these, and therefore a wide variety of

    acceptance criteria can be found in the industry for these kinds of tests. Here, unified criteria

    are proposed. The main purpose of the IPBT is to verify fatigue performance and only

    acceptance criteria for the armour layers are given here:

  • Test phase Acceptance criteria

    Service simulation (damage up to 0.1) Less than 5% broken tensile armour wires, and a successful pressure test to 1.25 times the design pressure with tensile load applied.

    Service-life model validation (damage up to 1.0 based on the design SN-curve)

    No loss of containment, and a successful pressure test to 1.25 times the design pressure with tensile load applied. Number of broken armour wires in agreement with statistical analysis, see below.

    Service-life model validation (damage up to 1.0 based on the mean SN-curve) or destructive test

    Number of broken armour wires in agreement with statistical analysis, see below.

    Test stop criteria Techniques to detect wire breaks during testing are improving. NOV is perfecting the wire

    break detection via optical monitoring. Alternatives include acoustic monitoring, acceleration

    response from pipe, and measurements of elongation of pipe during testing. Results for the

    optical monitoring technique are very promising. It is obvious that such methods are very

    interesting from a pipe integrity management point of view during in-service. The technology

    is also beneficial for application in fatigue testing: Using this methodology it is possible to

    run a service-life model validation test to failure consisting of a predefined number of wire

    breaks. In this way the final failure of the pipe (loss of pressure containment) can be avoided,

    which saves time and results in a dissection of a pipe that is un-damaged by the loss of

    containment and thereby learnings from the IPBT can be increased.

    Statistical method for determination of expected number of

    failures. A practical approach to estimate the number of broken wires at a certain calculated

    Accumulated Fatigue Damage (AFD) is the following:

    1) Estimate the actual number of wires in the fatigue zone that have a certain level of AFD.

    2) The AFD expresses the probability that a wire is broken. The number of expected wire breaks in a fatigue region is therefore calculated as the number of wires times the AFD

    times the used probability curve (0.5 for the mean curve and 0.024 for the design

    curve).

    An example: Using the mean SN curve, and running to AFD = 1.14 in a fatigue region

    containing 10 wires, we get that the expected number of wire breaks is 10 1.14 0.5 = 5.7 wire breaks.

    Further

    3) The fatigue zone can be divided into regions of equal damage and the number of wire breaks can be assessed

    4) Ultimately the number of wire breaks can be summed. 5) In case of evaluating the fatigue damage level of a service simulation, normally less

    than 0.5 wire breaks should be found using the above methodology.

  • The coverage of validation test for other pipes and

    other applications We have sometimes been faced with the question of whether one fatigue test covers another

    pipe or the same pipe in another part of the world. As stated above the test parameters should

    be selected to be in the same order of magnitude as those for a real application and the

    distribution should be so that a curve of relative occurrence versus load is similar to reality,

    see Figure 5 and Figure 7. However, the relative occurrence will not be identical; the slope

    needs to be different so that the test can be performed within a reasonable number of load

    cycles. Therefore, the resemblance between the actual field in terms of angles / curvatures and

    the test loading is something that can always be argued but also something that can be

    questioned. It cannot be used to determine whether a test covers a certain application. In our

    opinion a test covers a pipe design when the internal diameter is within plus / minus two

    inches of the test pipe, and when stresses and contact pressures of the test pipe covers those of

    the in-field conditions for the pipe.

    Figure 5 Top angle distribution (smooth blue curve is fatigue test and step wise

    purple in-field.

    Another aspect that one should consider when evaluating the validation of a certain pipe in a

    certain field is the overall verification of the methodology. Other successful tests, although on

    quite different pipes, add comfort to the methodology.

    Furthermore, it is important to realise that since the fatigue testing is accelerated, it means that

    the design methodologies, mainly the local analysis methodology must be equally well suited

    to handle large and small load cases. For the BFLEX package this has been demonstrated

    using fibre bragg grating as explained in [6].

    0

    2

    4

    6

    8

    10

    12

    14

    16

    18

    0 10 20 30 40 50 60 70 80 90 100

    accumulated occurence (%)

    top

    an

    gle

    (d

    eg

    )

    angle range, Kristin

    angle range, Visund

  • Figure 6 Comparison between measured and predicted stresses in tensile armour at

    small and large load cases, extract from [6].

    FATIGUE TEST A CASE STUDY

    Fatigue Modelling National Oilwell Varco (NOV) has performed a number of IPBTs. In order to show the

    conservatism in the NOV fatigue design methodology, one case story, the IPBT of an 8 production riser, is presented here.

    The global analysis for this particular riser was done for a lazy wave configuration where a

    spread moored FPSO was used. The water depth was 1700 m and the global dynamic

    modelling was performed using Flexcom.

    Based on metocean data the dynamic model provides, among other, fatigue load cases in

    terms of tension angle data/plots for each load case. Knowing the pipe and bend stiffener

    stiffness we get all of the tension, shear forces, and bending moment ranges for the hang-off

    section for the whole range of waves with the corresponding number of cycles. The resulting

    load cases are presented in Table 1. These load cases have been used as input to the IPBT

    service life test program for the 8 riser.

  • Table 1 Fatigue load cases in hang-off as found in the global dynamic model.

    The load cases are transferred to the local fatigue 3D model (BFLEX) of the hang-off

    configuration that returns, among other, mean stress, stress range, mean stress corrected stress

    range (using Gerber), number of cycles, and damage per cycle. The resulting load case

    spectrum is presented in Table 2.

    The normalized accumulated fatigue is further presented as a function of stress range in

    Figure 7. The service life test program is based on relatively fewer stress ranges between 150

    MPa and 250 MPa and relatively more stress ranges above 250 MPa in such a way that a

    fatigue damage of 0.1 is obtained during 2 million cycles. Such a distribution is shown by the

    red line in Figure 7.

  • Table 2 Load cases expressed as stresses, damage and cycles as expressed by the local

    fatigue 3D model BFLEX, calculated in hang-off.

  • Figure 7 Accumulated fatigue damage as a function of stress range

    The resulting load cases all based on the operating pressure of 30 MPa needed to reach a

    Miner sum of 0.1 are shown in Table 3.

    Table 3 Resulting load cases to obtain 0.10 fatigue damage.

    Seq. N_cycle

    X 1000

    Rig Angle mean (deg)

    Rig Angle range (deg)

    Tension @ mean rig angle

    (kN)

    Tension range (kN)

    Stress Range (MPA)

    Mean Stress (MPA)

    Damage

    1 1395 2.41 1.7 1372 135 164 493 2.51E-02

    2 400 2.71 2.3 1372 183 222 492 3.95E-02

    3 100 3.13 2.9 1372 230 253 485 1.51E-02

    4 75 3.48 3.5 1361 278 269 484 1.53E-02

    5 30 3.84 4.1 1337 326 282 481 7.89E-03

    2000 0.10

    The resulting test program based on not violating the maximum, 25% fatigue damage per load

    case criterion is shown in Table 4.

  • Table 4 The service life test program in blocks, not violating the maximum of 25%

    fatigue damage per load.

    Internal pressure 30.0MPa

    Seq. Rigcase N_cycle

    Rig

    Angle

    mean

    (deg)

    Rig

    Angle

    range

    (deg)

    Tension

    @ mean

    angle_rig

    (kN)

    Tension

    range

    (kN)

    1 1 174375 2.4 1.7 1372 135

    2 2 50000 2.7 2.3 1372 183

    3 3 12500 3.1 2.9 1372 230

    4 4 9375 3.5 3.5 1361 278

    5 5 7500 3.8 4.1 1337 326

    6 4 9375 3.5 3.5 1361 278

    7 3 12500 3.1 2.9 1372 230

    8 2 50000 2.7 2.3 1372 183

    9 1 348750 2.4 1.7 1372 135

    10 2 50000 2.7 2.3 1372 183

    11 3 12500 3.1 2.9 1372 230

    12 4 9375 3.5 3.5 1361 278

    13 5 7500 3.8 4.1 1337 326

    14 4 9375 3.5 3.5 1361 278

    15 3 12500 3.1 2.9 1372 230

    16 2 50000 2.7 2.3 1372 183

    17 1 348750 2.4 1.7 1372 135

    18 2 50000 2.7 2.3 1372 183

    19 3 12500 3.1 2.9 1372 230

    20 4 9375 3.5 3.5 1361 278

    21 5 7500 3.8 4.1 1337 326

    22 4 9375 3.5 3.5 1361 278

    23 3 12500 3.1 2.9 1372 230

    24 2 50000 2.7 2.3 1372 183

    25 1 348750 2.4 1.7 1372 135

    26 2 50000 2.7 2.3 1372 183

    27 3 12500 3.1 2.9 1372 230

    28 4 9375 3.5 3.5 1361 278

    29 5 7500 3.8 4.1 1337 326

    30 4 9375 3.5 3.5 1361 278

    31 3 12500 3.1 2.9 1372 230

    32 2 50000 2.7 2.3 1372 183

    33 1 174375 2.4 1.7 1372 135

    After the completion of the service life test the pipe should be further tested to a Miner sum of

    1.0 in order to illustrate the safety factor of 10 in the fatigue model. This was done using

  • 316000 cycles of a specially designed load case with increased angle and tension range as

    well as increased internal pressure. The test program is shown in Table 5.

    Table 5 Test program to obtain a total Miner sum of 1.0

    Internal Pressure 39.0 MPa

    N Cycles

    X 1000

    Rig

    Angle

    mean (deg)

    Rig

    Angle

    range (deg)

    Tension

    @ mean rig

    angle

    (kN)

    Tension range

    (kN)

    Stress Range

    (MPA)

    Mean Stress

    (MPA)

    Damage

    316 5.79 8 1125 750 421 560 0.90

    After completion of the service life test, the pipe should be further tested using a test-to-

    failure test program. The fatigue calculations were therefore performed using the mean SN

    curve. In order to accelerate the fatigue damage the bore pressure was set to 39 MPa like the

    extension test program. The accumulated fatigue damage (AFD) was at the onset of the

    destructive test 0.09 using the mean SN curve. The remaining destructive testing program was

    set to 0.91 so that the AFD would be 1.0 using the mean SN curve at the end of the destructive test. The test program for the destructive test part is shown in Table 6.

    Table 6 Test to failure test program

    Internal Pressure 39.0

    MPa

    N_cycle x 1000

    Rig

    Angle mean

    (deg)

    Rig

    Angle range

    (deg)

    Tension

    @ mean

    rig

    angle (kN)

    Tension

    range

    (kN)

    Stress

    Range

    (MPA)

    Mean

    Stress

    (MPA)

    Damage

    590 1.03 10 1125 750 438 503 0.91

    The presented test programs derived from an actual field shows tension and angle ranges. The

    test programs also show the stress range, mean stress and the internal pressure.

    However, it is more interesting to observe the contact pressure (from tensile armour to anti-

    wear tapes below) and stresses on wires combined, since they contain information on wire

    movements and exposure to fatigue loading. High contact pressures and low stresses indicate

    little movement of the wires, hence a low exposure to fatigue loading. High contact pressures

    and high stresses indicate extensive loading and exposure to fatigue loading. Therefore these

    parameters are directly comparable between different pipes, different wave fields and

    different test programs and give more information of a pipe designs fatigue resistance.

    An overview of these numbers as calculated from the three test programs is presented in

    Table 7.

  • Table 7 Overview of some of the most important fatigue drivers, i.e. contact pressure

    and stresses, divided in the test program load cases.

    Fatigue Testing During the year 2007 and 2008 the full scale dynamic fatigue testing of the NOV 8 flexible riser according to the modelling and test program given above was carried out.

    The phase 1 of the dynamic test program included a service life simulation test and an

    extension test.

    The service life simulation test program was designed to qualify the pipe design by simulating

    a predicted cumulative damage of 0.1 based on a selected service loading for deep water risers

    and the design SN curve for the tensile armour wires. The extension test was designed to

    demonstrate the reserve capacity of the pipe design and to validate the safety factor implicit in

    the design analysis methodology.

    The extension test included a total of 316 000 cycles, composed to generate a cumulative

    damage of 0.9 under service conditions based on the design SN curve for the tensile armour

    wires. Thus on completion of the service life simulation test and the extension test, a total

    cumulative damage of 1.0 was obtained. The successful test thus verified a safety factor of 10

    for the design analysis methodology for the present flexible riser. The planned dynamic test

    program was successfully completed without any observations of damage to the pipe, based

    on measured test parameters and visual external inspections. A hydrostatic pressure test was

    successfully performed at the end of the service life simulation test, demonstrating the

    structural integrity of the sample with respect to pressure holding capacity. The hydrostatic

    test was undertaken at a pressure of 1.25 times the design pressure of 345 bar, combined with

    an axial tensile load of 1500 kN.

  • A phase 2 test program was initiated to further investigate the residual life of the riser as well

    as possible failure modes and failure mode developments for the riser. The aim of the phase 2

    test program was to apply a number of cycles corresponding to a total cumulative damage of

    1.0 based on the mean SN curve for the tensile armour wire.

    The phase 2 test was planned with a total of 590 000 cycles in order to get to a total

    cumulative damage of 1.0 based on the mean SN curve for the tensile armour wire.

    The planned phase 2 of 590 000 cycles was successfully completed without any observations

    of damage to the pipe, based on measured test parameters and visual external inspections. For

    the above reason the testing was continued. Failure by leak of the sample was observed after

    an additional 92 000 cycles. The leak was noted through one of the venting holes of the top

    end fitting. Measured elongation of the pipe sample during the final phase of the test program

    indicated failure developments of some of the tensile armour wires.

    At the end of testing, the sample had been subjected to a total cumulative damage of 1.14

    based on the mean SN curve of the tensile armour wires.

    Following the fatigue testing the dynamic riser was dissected. The dissection showed that a

    number of tensile armours were broken in the fatigue zone. Further analysis of the test data

    showed that the wires broke during phase 2, and that the number of wire breaks was as

    expected when compared to the total cumulative damage of 1.14.

    Thus it can be concluded that the NOV fatigue methodology in this test has proven to be

    conservative.

    CONCLUSIONS The complex fatigue design procedure has been documented. Although complex, it is well

    described and statistics show few in-service failures due to fatigue.

    The In-plane bending fatigue test is a standard test for validation of the methodology and for the validation of actual pipes for their use.

    The In-plane bending fatigue test is defined in API 17B and the present paper provide further guidance to the selection of test parameters, test acceptance criteria and the

    common practice.

    Fatigue tests are an accelerated test and a test for one pipe in certain field conditions will cover for a range of other pipes and / or other applications as long as the internal diameter

    is the same within plus / minus two inches and the armour wires stresses and contact pressures are covered in the test.

    Therefore the following parameters are important when making a fatigue test:

    That the test pipe has the same ID as the actual pipe plus/minus 2 inch

    That the test pipe has the same cross sectional shape of armour wires not necessarily same size or material

  • That the load cases are selected so that contact pressure between armour wires, armour wire stress and armour wire stress range are of the same size as those for the actual

    application

    An example of fatigue testing validating the methodology is given. The example demonstrates the conservatism of the NOV fatigue design methodology.

  • REFERENCES [1] Specification for unbounded flexible pipe, API 17 J, July 2008 [2] Recommended Practice for Flexible Pipe, API 17B, July 2008. [3] State of the art report on flexible pipe integrity, report no. 2-4-5-013/SR01,

    http://www.oilandgasuk.co.uk/publications/viewpub.cfm?frmPubID=152

    [4] Pascal Estrier, Updated Method for the Determination of the Service Life of Flexible Risers. Proceedings of the First European Conference Marinflex, 1992

    [5] Svein Svik, On stresses and fatigue in flexible pipes, Dep. Of Marine Structures, NTNU, 1992

    [6] Svein Svik & Ragnar R. Igland, Calibration of a flexible pipe tensile armour stress model based on fibre optic monitoring, OMAE 2002-28092

    [7] S. Berge, E. Bendiksen, J. Gudme and R. Clements, Corrosion fatigue testing of flexible riser armour procedure for testing and assessment of design criteria, OMAE 2003 - 37327.