Premium Digest December 2010 Consequences of a Risk-based Approach for Natural Gas Pipelines

Embed Size (px)

Citation preview

  • 7/27/2019 Premium Digest December 2010 Consequences of a Risk-based Approach for Natural Gas Pipelines

    1/4

    19 PiPelines international digest | deCeMBer 2010

    teChniCal

    Consequences of a risk-basedapproach for natural gas pipelinesBy G.M.H.Laheij, National Institute for Public Health and the Environment, Netherlands; and,

    C.J.Theune, Ministry of Housing, Spatial Planning and the Environment, Netherlands

    In the Netherlands natural gas is transported through an underground pipeline network with a combined length of about14,000 km. In preparation of new legislation, in which zoning distances will be based on a probabilistic approach, thequantitative risk-analysis methodology for underground natural gas transmission pipelines has been revised to reectnew understandings in the risk scenarios, failure frequencies and effects.

    1. MinistryofVROM,2010.BesluitExterneVeiligheidBuisleidingen(inDutch,inprep.).

    In order to get a complete overview of third-party risks in

    the Netherlands, all pipeline owners are obliged to provide

    pipeline data to a national risk register. Based on these data,

    and taking into account the new zoning distances, an analysis

    of the consequences of these new zoning distances for land-useplanning is carried out. The aim of this analysis is to identify

    potential bottlenecks where dwellings are situated within the new

    zoning distances of these pipelines and to identify where, based

    on the spatial planning plans available up to 2030, future possible

    bottlenecks may appear.

    Additional measures should reduce the risk if, for example,

    dwellings are situated within the new zoning distances or if the

    guidance value for the societal risk is exceeded. As there are no

    measures available for reducing the eects of a pipeline rupture,

    the additional measures focus mainly on reducing the probability

    of pipeline ruptures. Because external interference is the main

    cause of pipeline ruptures, the additional measures focus onthis cause. Proposed measures for reducing the risk of high-

    pressure natural gas pipelines include the use of concrete slabs

    or warning tapes, and agreements with landowners about land

    utilisation. Also, the introduction of a statutory one-call system

    is an important generic measure for reducing the probability of

    pipeline failure.

    IntroductionIn the Netherlands, natural gas is transported through

    underground pipelines with a combined length of approximately

    14,000 km. In a circular letter issued in 1984, land-use planning

    guidelines and generic zoning distances were laid down for

    underground natural gas transmission pipelines. In preparationof new legislation, in which zoning distances will be based

    on a probabilistic approach, the quantitative-risk-analysis

    methodology for underground transmission pipelines for natural

    gas has been revised. New understandings in the risk scenarios,

    failure frequencies, and eects are now included.

    This article gives an overview of the reviewed risk methodology

    for natural gas pipelines. Also, the consequences for land-use

    planning will be given together with measures that can be taken

    to reduce the risk of these pipelines.

    Zoning policy

    The new zoning policy is part of a two-track policy forpreventing major accidents. Firstly, the frequency of accidents

    occurring and their eects when they do occur are reduced as

    much as reasonably possible by taking measures at the source

    of risk. Secondly, the number of persons exposed to eects,

    should an accident occur, is reduced by the zoning policy. Two

    measures are used in dening these policies: the individualrisk as a measure of the level of protection to each individual

    member of the public, and the societal risk as a measure of the

    disaster potential for the society as a whole. The individual risk

    is expressed as the risk of fatality per year. This is dened as the

    probability that an unprotected person residing permanently at

    a xed location will be killed as a result of an accident occurring

    at a source of risk. The societal risk is dened as the probability

    that a certain number of deaths will be exceeded during a

    single accident; it is expressed as the relationship between the

    number of people killed (N) and the frequency per year (F) that

    this number will be exceeded. For both the individual risk and

    societal risk, criteria limits will be set for pipelines

    1

    . For dwellingsand vulnerable buildings such as schools and hospitals, the

    individual risk limit is set at 10-6 per year. For less vulnerable

    building such as small oce buildings, restaurants, shops, and

    recreation facilities, the individual risk contour of 10-6 per year

    is a guidance value. The limit for the societal risk is an indicative

    limit. For transport routes, the limiting frequency (Flim) per

    kilometre of pipeline for the occurrence of an accident with N or

    more deaths is:

    Flim N2

    = 10-2

    (1)

    The number of deaths (N) must be larger than 10 to be

    incorporated into societal risk. In zoning policy, the individualand societal risks complement each other. The individual

    risk creates a distance between the source of risk and its

    surroundings. The societal risk limits the population density

    around the source of risk.

    Quantitative risk methodologyFor each scenario, important parameters that should be

    determined are the relevant failure modes and the failure

    frequencies. For ammable substances, the consequences are

    determined by the mass ow rate of each scenario, the probability

    of ignition, the characteristics of the subsequent re and the

    corresponding heat radiation prole. Also, the population inthe surroundings of the pipeline must be identied in order to

  • 7/27/2019 Premium Digest December 2010 Consequences of a Risk-based Approach for Natural Gas Pipelines

    2/4

    20 PiPelines international digest | deCeMBer 2010

    teChniCal

    determine the consequences in terms of the number of deaths.

    These parameters are discussed in this article. The methodology

    is described in more detail in references 2 and 3.

    Scenarios and failure frequenciesFor underground pipelines, all possible release scenarios are

    divided over two scenarios, namely a rupture and a leakage4.

    Since leakages do not signicantly contribute to the risk of

    pipelines with ammable substances they are not taken in to

    account in the risk calculations. The failure frequencies and

    consequences are therefore only determined for pipeline ruptures.

    Generic failure frequencies for underground pipelines are given

    in Part II of the Purple Book4, which describes quantitative risk

    assessment guidelines for transportation activities. The general

    failure frequency for steel pipelines has been set at 6.1 x 10-4

    per kilometre per year. The failure frequency is split into two

    scenarios, a leakage and a rupture, with a probability of 0.75 and

    0.25 respectively. Therefore, for a rupture the failure frequency

    will be 1.5 x 10-4

    per kilometre per year. For high-pressure naturalgas, it has been reviewed whether this failure frequency is

    still valid.

    For natural gas pipelines, the failure frequency of pipeline

    ruptures is determined by external interference3. The failure

    frequency is derived using the PipeSafe methodology5. Firstly, the

    probability (fd in km/a) that the pipeline is hit is determined as a

    function of the depth of cover (d in metres) of the pipeline6:

    fd = e-2.4 d-3.5 (2)

    This function was derived combining the number of incidents

    in a pipeline depth class with the overall years of experience

    in the depth class. For every metre of extra depth of cover, the

    hit frequency decreases by about a factor of 10. Secondly, using

    historical damage data and fracture mechanics the probability

    of a pipeline rupture is calculated7. Pipeline parameters used in

    the calculations are the diameter, pressure, depth of cover, wall

    thickness, yield strength and Charpy energy. Using this model,

    the probability of a pipeline rupture for the total cross-country

    pipeline network of Gasunie was calculated. The Gasunie network

    is about 12,000 km in length and represented in the calculations

    by about 1.2 million data points. From this analysis 0.7 rupture

    per year is predicted. Historical failure data for the Gasunie

    network were available for the period 19772005. In this period 12

    pipeline ruptures occurred, with no pipeline ruptures during thelast 11 years. Based on this data it was determined that there is a

    statistical signicant trend in the number of pipeline ruptures2.

    Therefore only the last 11 years (with no pipeline ruptures) were

    used to compare the model predictions with the historical data.

    Based on the last 11 years the upper bound of the 95 per cent

    condence interval is equal to 0.25 pipeline rupture per year. As

    the model predicts 0.7 pipeline rupture per year for the Gasunie

    network, it was decided to reduce the model predictions by a

    factor of 2.8 (= 0.7/0.25). The factor of 2.8 does not currently apply

    to other shippers of (raw) natural gas, as it is believed that the

    signicant trend is due to specic measures taken by Gasunie.

    Also, for pipelines with raw natural gas, the additional wall

    thickness included for internal corrosion has to be excluded from

    the calculation of the failure frequency.

    In this review the eect of a statutory one-call system on the

    failure frequency is also included8. This system laid down

    by law replaced the voluntary one-call system in 2008. The

    statutory one-call system requires not only that all digging

    activities are notied, but also additional rules for the follow-up

    of a notication are introduced. The National Institute for PublicHealth and the Environment (RIVM) has estimated the inuence

    of the statutory one-call system, and the derivation of this

    estimate is described in references 2 and 9. In co-operation with

    N.V. Nederlandse Gasunie (the Dutch natural gas transmission

    company), the voluntary one-call system was reviewed and

    investigations were made why, despite an activity was notied,

    incidents still occurred10. Based on this review the rules of the

    statutory one-call system were evaluated on how they aect

    the chance that a pipeline is hit due to external interference.

    This evaluation leads to a reduction factor of 2.5 for spillages

    caused by external interference. The Ministry of Housing, Spatial

    Planning and the Environment (VROM) decided to take this

    factor into account in the risk calculations as it commits itself to

    a result achievement. Whether in practice the factor of 2.5 will be

    established must be monitored in the forthcoming years. If the

    risk-reducing factor is not reached in practice, additional rules

    should be put in place. This is possible under the statutory one-

    call system.

    Release and effect calculationsIn the release scenarios for pipelines with ammable

    substances only pipeline ruptures are taken into account, as leaks

    dont contribute to the individual risk contour of 10-6

    per year. For

    pipelines with ammable substances the eects are determined

    by heat radiation. The probability of death due to the exposureto heat radiation is calculated with the use of a probit function.

    2. G.M.H.Laheij,A.A.C.vanVliet,andE.S.Kooi,2008.Achtergrondenbijherzienezoneringsafstandenhogedrukaardgastransportleidingen.RIVMreport620121001/2008(inDutch).

    3. M.Gielisse,M.T.Drge,andG.R.Kuik,2008.Risicoanalyseaardgastransportleidingen.GasuniereportDEI2008.R.0939(inDutch).

    4. CommitteeforthePreventionofDisasters,1999.Guidelinesforquantitativeriskassessment,CPR18E.

    5. M.R.Acton,P.J.Baldwin,T.R.Baldwin,andE.Jager,1998.ThedevelopmentofthePipeSaferiskassessmentpackageforgastransmissionpipelines.Proc.Int.PipelineConference,Calgary,ASMEInternational.

    6. E.Jager,G.R.Kuik,G.Stallenberg,andJ.Zanting,2002.AqualitativeriskassessmentofthegastransportservicespipelinesystemnetworkbasesonGISdata,ICTPrague.

    7. I.Corder,1995.Theapplicationofrisktechniquestothedesignandoperationofpipelines,IMechE,C502/016.

    8. Staatsblad2008,Wetvan7februari2008,houdenderegelsoverdeinformatie-uitwisselingbetreendeondergrondsenetten(Wetinformatie-uitwis-

    selingondergrondsenetten)Stb.2008,120,Sdu(inDutch).9. G.M.H.Laheij,G.R.Kuik,R.vanElteren,andA.A.C.vanVliet,2008.Inuenceofastatutoryone-callsystemontheriskofnaturalgaspipelines.PSAM9,9thInt.Conf.onProbabilisticSafetyAssessmentandManagement,HongKong,China,18-23May(Eds.Tsu-MuKao,EnricoZioandVincentHo).

    10. R.vanElteren,M.H.vanAgteren,K.H.Kutrowski,G.G.J.Achterbosch,G.R.Kuik,2004.BepalingeectiviteitKLIC-procestenaanzienvanaardgastrans-portleidingen,GasuniereportRT04.R.0694(inDutch).

  • 7/27/2019 Premium Digest December 2010 Consequences of a Risk-based Approach for Natural Gas Pipelines

    3/4

    21 PiPelines international digest | deCeMBer 2010

    teChniCal

    The probit function for death due to heat radiation is given by:4, 11

    Pr = -36.38 + 2.56 ln(Q4/3t) (3)

    where Q is the heat radiation (Wm-2

    ) and t is the exposure time (s).

    The maximum exposure time is 20 seconds. Overpressure eects

    dont contribute signicantly to the risk and are therefore not

    included in the calculations.

    A pipeline rupture of an underground natural gas pipeline

    results in a vertical jet. In the calculations, two separate jets are

    taken into account. The rst jet is based on the average release

    during the rst 20 seconds of the accident. The second jet is based

    on the release between 120 and 140 seconds. This approach is

    chosen as the maximum exposure time for ammable eects is

    set to 20 seconds, see Equation 34, 11. Also, from historical data

    it was determined that in 75 per cent of the incidents ignition

    takes place in the rst 30 seconds. In 25 per cent of the incidents

    ignition takes place after at least 120 seconds. In the release

    calculations also the eect of the crater on the momentum of thejet is included.

    The methodology was initially set-up for pipelines with

    processed natural gas3. It was also investigated whether for

    pipelines transporting raw natural gas the same methodology

    as for processed gas could be used12. Raw natural gas pipelines

    contain, among other byproducts, water and condensate. For raw

    natural gas pipelines the pipeline pressure drops over time. It

    was therefore rst investigated which production case gives the

    highest release rate and results in the highest heat of combustion

    in case of a pipeline rupture; the low production case with

    relatively high liquid hold-up or the high production case with

    relatively small liquid hold-up in the pipeline. Investigating a

    pipeline transporting gas with a high condensate gas ratio (CGR)

    of 80 [m3 condensate per million Nm3 of gas] and using computed

    uid dynamics it was concluded that the high production case

    results in the highest release rates. In both cases no rain-out

    of condensate droplets occurred. It was also concluded that

    the highest heat of combustion occurs in the high production

    case. Furthermore, the calculated eect distances for the high

    production case where almost equal to the eect distances of

    a similar pipeline with processed natural gas. Therefore it was

    concluded that for the release and eect calculations of pipelines

    with raw natural gas (CGR < 80) the same models as for processed

    gas can be used without the need to adapt them specically for

    raw natural gas.

    Probability of ignitionThe probability of ignition is subdivided into direct ignition and

    delayed ignition. For natural gas pipelines, ignition results in a jet

    re. From historical data it was determined that the probability of

    ignition (Pign) is related to the diameter and pressure of the pipeline3:

    Pign = a + bpD2 (4)

    where a, b are constants, p is the pipeline (barg), and D is the

    pipeline diameter (mm). The maximum ignition probability

    equals 0.8.

    For example, using Equation 4, for a 4 inch diameter natural

    gas pipeline at 40 bar the probability of ignition equals 0.08; in

    the case of a 48 inch diameter pipeline at 80 bar, the probability

    of ignition equals 0.8. As the inuence of the built-up area on

    the probability of ignition is most likely not included in Equation

    4, the contribution of the built-up area to the probability of

    ignition was evaluated separately2. From a literature study, it was

    concluded that the most important contribution of the built-up

    area to the ignition probability comes from two sources:

    Sparks induced by the impact of debris on house bricks; and,

    The ignition of gas inltrated into buildings.

    From release calculations it could be concluded that the built-

    up area can only inuence the ignition probability for releases of

    pipelines with a diameter smaller than 18 inches. For pipelines

    with a diameter equal to or larger than 18 inches, the ignitable part

    of the jet, dened by its 50 per cent lower-explosive-limit (LEL)

    contour, will not be present at a height lower than 20 m. For these

    pipelines it was therefore concluded that there is no contribution

    of the built-up area. For pipelines with a diameter smaller than18 inches it was estimated that ignition probability as a result of

    the impact of debris was increased by 0.07 and as a result of the

    inltration of gas in buildings by 0.032. For these pipelines the

    probability of ignition used in the calculations is now:

    Pign = a + bpD2 + 0.1 (5)

    Using Equation 5, for a 4 inch natural gas pipeline at 40 bar

    the probability of ignition equals 0.18; in the case of a 48 inch

    pipeline (80 bar), the probability of ignition still equals 0.8.

    ResultsUsing the risk methodology as described in the aboveparagraphs, both individual risk and societal risk calculations

    can be performed. For natural gas pipelines the PipeSafe program

    is used to calculate the individual and societal risk8. The risk is

    not only dependent on the diameter and pressure of the pipeline

    but also on the depth of cover, wall thickness, yield strength,

    and Charpy energy. Therefore, the distance to the individual risk

    contour of 10-6 per year lies between 0 m and the maximum eect

    distance of a pipeline. The pipelines maximum eect distance,

    dened as the distance to 1 per cent lethality, is only dependent

    on the diameter and pressure of the pipeline. For 48 inch

    diameter pipelines the maximum eect distance can be up

    to 600 m. Whether the indicative limit of the societal risk willbe exceeded, also strongly depends on the above mentioned

    parameters.

    Consequences for land-use planningIn order to get a complete overview of third-party risks in the

    Netherlands, pipeline owners are required to provide pipeline

    data to a national risk register13. Based on these data and taking

    into account the new zoning distances, an analysis of the

    consequences of these new zoning distances was carried out.

    The aim of this analysis was to identify potential bottlenecks

    where dwellings are situated within the new zoning distances

    and to identify where, based on land-use plans available up

    11. RIVM.ReferenceManualBeviRiskAssessments.Version3.2.2009.

    12. K.Beijer,2009.Technicalnote:Mogelijkeverschillenin(externeveiligheid)risicotussendeoperatievannatgasendrooggastransportleidingsystemen,NAMEP200702210020,revision3.March(inDutch).

  • 7/27/2019 Premium Digest December 2010 Consequences of a Risk-based Approach for Natural Gas Pipelines

    4/4

    22 PiPelines international digest | deCeMBer 2010

    teChniCal

    to 2030, future possible bottlenecks may appear. The analysis

    has been performed using data from 11 pipeline owners with in

    total about 14,000 km of natural gas pipelines. The data used

    for the existing vulnerable objects were obtained combining

    several commercially available databases with coordinates of

    dwellings, building functions, and population numbers. The

    land-use plans of municipalities were taken from the New Map

    of the Netherlands14. Using these data, the following results

    were found. The total length of pipelines with already existing

    vulnerable objects within the individual risk contour of 10-6 per

    year is approximately 4050 km. Due to the land-use plans up to

    2030 additional 80 km of pipeline could become a bottleneck. It

    depends strongly on how these new plans are nally developed

    whether the potential identied bottlenecks appear or not.

    Additional measuresAdditional measures should reduce the risk if dwellings are

    situated within the new zoning distances or if the guidance

    value for the societal risk is exceeded. As there are no measuresavailable for reducing the eects of a pipeline rupture, the

    additional measures focus mainly on reducing the probability

    of pipeline ruptures. Because, for natural gas pipelines,

    external interference is the main cause of pipeline ruptures3,

    the additional measures for natural gas pipelines focus on this

    cause2, 15. Proposed measures for reducing the risk of natural

    gas pipelines can be categorised into two groups. Measures in

    the rst group prevent (partly) that the pipeline is actually hit

    during digging activities. Measures in the second group prevent

    or control digging activities in the neighbourhood of a pipeline.

    Proposed measures for reducing the risk of high-pressure

    natural gas pipelines are, for example, the use of concrete slabs

    or warning tapes and agreements with landowners about land

    utilization. For all measures, preconditions are dened and all

    preconditions should be met before the subsequent reduction

    factor can be used. In Table 1, proposed measures with their eect

    on probability of a pipeline rupture due to external interference

    are given. The eectiveness of the measures in practice must be

    monitored in the forthcoming years.

    Harmonisation with other transmission pipelinesIt is the objective to harmonise the methodologies for

    underground transmission pipelines with ammable liquids

    and other chemical substances. However, it is noted that other

    failure mechanism than external interference, such as corrosion

    or mechanical failure, are of importance, and the above measures

    are therefore only partly eective for risk reduction. The risk

    methodologies and risk reduction measures for these substancesare under development.

    ConclusionsIn preparation of new legislation, in which zoning distances

    and the limits for societal risk are based on a probabilistic

    approach, the quantitative risk analysis methodology for

    underground transmission pipelines for natural gas has been

    revised to reect new understandings in the risk scenarios, failure

    frequencies and eects. An outline of the new risk methodology is

    given together with the consequences for land-use planning and

    measures that can be taken to reduce the risk of these pipelines.

    The new legislation will lead to additional measures, both in

    design and operation, with a more optimised land use. It also

    will allow for tightened compliance checks by governmental

    competent bodies.

    13. J.P.vantSant,H.J.Manuel,A.vandenBerg,2008.Dutchregistrationofrisksituations.ESRELEuropeanSafetyandReliabilityConference,Valencia,Spain,22-25September(eds.S.Martorelletal.).

    14. NetherlandsInstituteforPlanningandHousing(NIROV),2009.DeNieuweKaartvanNederland,January.

    15. G.M.H.LaheijandA.A.C.vanVliet,2009.Measuresforreducingtheprobabilityofrupturesofhighpressurenaturalgaspipelines.EuropeanSafetyandReliabilityconference2009,Prague,7-10September.

    Measure Reduction factor

    Extra depth of cover See Equation 2

    Warning tape 1.67

    Concrete slab 5

    Concrete slab + warning tape 30

    Stringent supervision of digging activities 3

    Agreements about land utilization 1.6 100

    Table1:Eectofadditionalmeasuresontheprobabilityofa

    pipelinerupture.

    This paper was presented at the Pipeline Technology

    Conference held in Hannover, Germany, in April 2010, and

    organised by EITEP.