RAC in Denmark & EU

download RAC in Denmark & EU

of 60

Transcript of RAC in Denmark & EU

  • 7/27/2019 RAC in Denmark & EU

    1/60

  • 7/27/2019 RAC in Denmark & EU

    2/60

    T he D anish Environmental Protection A gency will, when opportuni ty

    offers, publish reports and contributions relating to environmentalresearch and development projects financed via the D anish EPA .

    P lease note that publication does not signify that the contents of the

    reports necessari ly reflect the views of the D anish EPA .

    T he reports are, however, published because the D anish EPA fi nds that

    the studies represent a valuable contribution to the debate on

    environmental policy in D enmark.

  • 7/27/2019 RAC in Denmark & EU

    3/60

    3

    Contents

    P R E F A C E 5SU M M A R Y A N D C O N C L U SI O N S 71 EX P L A N A T I O N O F T ER M S 9

    1.1 I N T R O D U C T I O N 91.2 H A Z A R D , C O N S EQ U EN C E A N D R I SK 9

    1.2.1 Ri sk matr ices 91.2.2 Consequence distance and maximum consequence distance 101.2.3 Location-based (indivi dual) ri sk 111.2.4 Societal or group risk 121.2.5 Potenti al Loss of Li fe (PLL) 12

    1.3 L A N D U SE PL A N N I N G A N D M A JO R H A Z A R D E ST A BL I SH M EN T S 131.3.1 Safety distances 131.3.2 Ri sk anal ysis methods 14

    2 E A R L I E R D A N I SH ST U D I ES 172.1 EN V I R O N M EN T PR O JEC T 112 17

    2.1.1 Recommendations in Environment Project 112 forquanti tat ive ri sk acceptance criteria 18

    2.1.2 Recommendations in Envi ronment Project 112 for quali tati veacceptance cri teria 20

    2.2

    T H E T N D E R R E P O R T 22

    3 U SE O F R I SK A C C EPT A N C E C R I T ER I A W I T H I N T H E EU 253.1 L A N D U SE PL A N N I N G G U I D E L I N ES FR O M T H E EU R O P E A N

    C O M M I SSI O N 253.1.1 Ri sk analysis methods and risk criteria 263.1.2 Other techni cal issues 27

    3.2 EU R O P E A N C O M M I SSI O N EX A M P L ES O F R I SKA N A L Y SI S M ET H O D S ( R O A D M A PS ) 28

    3.3 R I SK A C C EP T A N C E C R I T ER I A I N F I N L A N D 293.4 R I SK A C C EP T A N C E C R I T ER I A I N FL A N D E R S 293.5 R I SK A C C EP T A N C E C R I T ER I A I N FR A N C E 303.6 R I SK A C C E PT A N C E C R I T E R I A I N T H E N ET H ER L A N D S 333.7 R I SK A C C EP T A N C E C R I T ER I A I N I C E L A N D 343.8 R I SK A C C E PT A N C E C R I T E R I A I N T H E U N I T ED K I N G D O M 353.9 R I SK A C C EP T A N C E C R I T ER I A I N G ER M A N Y 36

    4 D I SC U SSI O N O F R I SK A C C EPT A N C E C R I T ER I A 394.1 D I SC U S SI O N O F D E V EL O P M EN T S W I T H I N D EN M A R K 394.2 D I SC U SSI O N O F R EV I E W O F P R A C T I C E S W I T H I N T H E EU 40

    4.2.1 Quali tati ve versus quantitative cri teri a methods 404.2.2 Ensur ing consistent and uniform decisions 414.2.3 Comparison of quanti tat ive ri sk acceptance criteri a 414.2.4 Existi ng and new situati ons 424.2.5 Dealing with vulnerable objects (such as hospitals, schools,

    and infrastructure) 42

  • 7/27/2019 RAC in Denmark & EU

    4/60

    4

    4.2.6 Ri sk acceptance cri teria for env ironmental damage 434.2.7 Ri sk acceptance cri teria for personal injury 43

    4.3 G EN E R A L O B SER V A T I O N S 444.3.1 Indivi dual risk level and protecti on of vulnerable objects 444.3.2 Societal ri sk and ri sk aversion 454.3.3 Frequencies for reference accident scenarios and maximum

    consequence di stances 464.3.4 Ri sk acceptance criteri a for existi ng and new situati ons 464.3.5 Ri sk acceptance cri teria for env ironmental damage 474.3.6 Ri sk acceptance cri teria for personal injury 49

    5 C O N C L U SI O N S A N D R EC O M M EN D A T I O N SR EG A R D I N G U SE O F R I SK A C C EPT A N C E C R I T ER I A I ND E N M A R K 51

    5.1 ST A T U S I N D EN M A R K A N D T H E EU 515.2 R I SK A C C E PT A N C E C R I T ER I A R EQ U I R EM EN T S 515.3 I N C O R PO R A T I N G F REQ U E N C Y C R I T ER I A 525.4 PR O T EC T I O N O F V U L N ER A B L E O B JEC T S 525.5 R I SK A N A L Y SI S M ET H O D S 535.6 N E ED S FO R FU R T H E R W O R K 53

    6 G L O SSA R Y 557 R EFEREN C ES 59

  • 7/27/2019 RAC in Denmark & EU

    5/60

    5

    Preface

    I n connection wi th an envi ronmental and emergency-planning review ofmajor hazard establishments in D enmark , the D anish Emergency

    M anagement A gency, the A gency for Spatial and Environmental Planning,

    and the Envi ronmental P rotection A gency decided to investigate the use of

    acceptance cri teria for risk to third parties in other E U countries, and compare

    these with D anish cri teria. A task force has gathered the relevant information

    from a number of EU countries. T his report contains the results from a

    review of this information, and makes comparisons with the situation in

    D enmark. T he report concludes with some observations on how these

    experiences from other countries can be applied within D enmark . T he report

    thus serves as a supplement to previous contributions in this area.

    T he report is targeted at major hazard authoriti es in D anish local authorities

    and the regional environmental centres of the D anish M inistry of theEnvironment, and may also be of interest to the major hazard establishments

    themselves.

    M ajor hazard authori ties need ri sk acceptance criteria that can be used in the

    following situations:

    - when auditing environmental permits for existing major hazardestablishments,

    - when planning changes in land use ( in municipal or local plans) close toexisting major hazard establishments,

    - in connection with environmental impact assessment and environmentalpermit for expansion or changes to existing major hazard establishments,

    and

    - when establishing new major hazard establi shments.R isk acceptance criteria have to protect human li fe and health, as well as

    environmental resources and natural areas.

    Environment Project 112 ( T aylor et al., 1989) provides an important data

    basis by gatheri ng methods and data for ri sk assessment of major hazard

    establishments in D enmark . M ost of the considerations in Environment

    Project 112 are sti ll current. T his report may therefore be viewed as an

    update to the basis of E nvi ronment P roject 112, based on developments and

    experience within D enmark and a number of other E uropean countries since

    1989.

    T he report has the following structure:

    C hapter one reviews the relevant terms used in ri sk assessment, and provi desa brief description of two di fferent types of risk analysis method ( quantitative

    and qualitative) . A glossary at the end of the report contains a brief

    explanation of these and other relevant terms in the report.

    C hapter two reviews previous D anish studies. T his includes Environment

    Project 112, and a recent report reflecting changes in risk analysis and

    acceptance practices in D enmark.

    C hapter three reviews risk analysis and acceptance practices in the European

    U nion, based on documents prepared by the European C ommission, and

    special information obtained from selected countries ( Finland, Flanders,

    France, G ermany, I celand, the N etherlands and the U nited K ingdom) .

    C hapter four discusses and compares the information presented in chapters

    two and three to provide a picture of the current status of certain issues. T hechapter closes with a number of conclusions for the areas where the review

  • 7/27/2019 RAC in Denmark & EU

    6/60

    6

    indicates consensus, and provi des possible solutions where consensus is

    lacking for example, in relation to dealing with environmental damage.

    C hapter fi ve concludes with recommendations for how this experience might

    be applied in D enmark . T hese recommendations consist of a summary of a

    number of requirements for general ri sk acceptance cri teria and assessment

    methods, and a proposal for the general design of risk acceptance criteria and

    assessment methods in D enmark.

    C hapter six includes a glossary of the most important terms used in this

    report.

    T his report was prepared by N ijs Jan D uijm ( D T U M anagement

    Engineering, consultant and writer) between N ovember 2007 and A pri l

    2008.

    A task force consisting of:

    A llan T homsen ( D anish Emergency M anagement A gency) A nne C hristine Bryderup ( D anish Emergency M anagement A gency) G ert Johansen ( D anish A gency for Spatial and Environmental

    P lanning) N anna R rbech ( D anish Envi ronmental Protection A gency) A nders Skou (D anish Envi ronmental Protection A gency) and A xel Bendtsen ( D anish Environmental P rotection A gency,

    coordinator)

    has served as a monitoring group.

    T he conclusions of the report do not necessarily represent the opinions of the

    D anish Emergency M anagement A gency, the Agency for Spatial and

    Environmental P lanning, or the Environmental Protection A gency.

  • 7/27/2019 RAC in Denmark & EU

    7/60

    7

    Summary and conclusions

    T his report describes the use of risk acceptance criteria in order to identifyconflicts between major hazard establi shments ( establi shments covered by the

    major hazard ( Seveso ) directives) and surroundi ng land use with respect to

    protection of human life and environment.

    T he report explains relevant notions in ri sk assessment and a glossary is

    included. T he report describes different types of risk acceptance criteria based

    on individual or location-based risk, societal risk and potential loss of li fe. T he

    di fferent types of ri sk assessment methodologies are described.

    Earlier D anish studies are reviewed, including Environment Project 112 , in

    order to identify the background for the present practice in D enmark , and the

    practice for risk acceptance in a number of other European countries is

    summarised.

    C onclusions are drawn where European practice has converged towardsconsensus regarding risk acceptance criteria, viz. the order of magnitude for

    individual or location-based risk ( fatality risk of an individual shall be less than

    10-6

    per year for the protection of the general population) and societal risk ( the

    probabi li ty of an accident shall be less than 10-3

    per year for major accidents

    with up to 1 fatality, dropping with a factor 100 when consequences are ten

    times bigger) . T hese values are in the middle of the grey zone proposed by

    Environmental Project 112 .

    T he A L A R A ( A s L ow A s R easonably A chievable) principle states that all

    safety precautions should be implemented that are reasonable in view of

    technical and economical possibili ties. T he A L A R A principle should always

    be applied, and therefore it is not necessary to define separate intervals of risk

    where A L A R A has to be used.

    T he report concludes that in D enmark no guidance is available on how safety

    distances should be determined using the available qualitative risk analysis

    methods, nor is a method to assess envi ronmental damage available. I t is

    proposed to define accident classifications for environmental damage similarly

    to those used for quali tative assessment of societal risk, by using the size of the

    area affected by the accident.

    T he reports conclusions give some recommendations on risk acceptance

    cri teria. T hese cri teria have to fulfil requirements regarding:

    C onsistency, proportionality and transparency. A ll accident possibi li ties to be considered. Environmental damage to be considered. R easonable safety distances ( for land-use limitations) to be

    determined as well as the consequence distance for the worst-case

    accident ( for emergency planning) .

    Societal risk for land use outside the safety di stances to be assessed. Specifi c safety measures at the establi shment to be considered.

    I t i s necessary to consider the probabi li ty of events in the assessments, as

    probability is elementary in the notion of risk. I nclusion of probabili ty can be

    either quantitatively or qualitatively.

    L imitations in land use distinguishes between vulnerabili ty of di fferent objects

    such as residential areas, hospitals, schools or natural reserves; generally

    higher levels of risk are accepted in industrial areas as compared to residential

    areas, while hospitals and emergency support facilities should be given the

  • 7/27/2019 RAC in Denmark & EU

    8/60

    8

    best protection possible. T he same risk criteria should apply, i f needed after a

    transitional period, for existing land-use situations and new developments.

    T he use of quanti tative and qualitative risk analysis methods should be

    possible side by side, but efforts should be directed into making the results

    more comparable. T his means that among others there should be obtained

    consensus about the relation between verbal descripti ons and the numerical

    values of accident probabi li ties. T he French risk assessment method, which

    includes both quantitative and qualitative elements, could be used as a basis

    for a relatively simple and transparent risk assessment method in D enmark.

    Further work is required in order to develop the qualitative methods to

    determine safety distances, to develop criteria and methods to deal with risks

    for environmental damage, and to develop a practical guideline for the use of

    the A L A R A principle.

  • 7/27/2019 RAC in Denmark & EU

    9/60

    9

    1Explanation of terms

    1.1 IntroductionW ords such as risk , hazard and consequence are commonly used. Y et

    communication about risk can sometimes be diffi cult, as the precise

    definitions used by risk experts are not always understood by non-experts

    with their more subjective perception of these terms.

    T his chapter wi ll explain these terms as they are used by experts in the field of

    assessing risk of i ndustri al activi ties. Earlier publications exi st in which these

    terms are defined and discussed ( in D anish) , i ncluding Environment Project

    112 ( T aylor et al., 1989) , D anish Standard D S/I N F 85 ( D ansk Standard,

    1993) and a paper by the D anish Environmental R isk C ouncil ( C hristensen et

    al., 2002; C hristensen et al., 2003) . T his report adheres most closely to thedefinitions in D anish Standard D S/IN F 85. T here is a glossary at the end of

    the report containing the most important terms.

    1.2 Hazar d, consequence and r iskT he terms, hazard, consequence , and risk , are used in the risk analysis

    process.

    I n order to determine the risks an activity may entail, one must fi rst identify

    thehazardsinherent in the activi ty. A hazardis defined as a situation or statethat could lead to injury. I t thus refers to the possibilityof an accident, withoutconsidering probabili ty or consequences.

    A consequenceis the result of an undesired event ( an accident) , such as injuryto health, life, assets, or the environment.

    Riskexpresses a combination of the frequency ( or probabili ty) of anundesired event, and the scope of the consequences. R isk is used in order to

    be able to compare various events in terms of the highest or lowest risk. R isks

    must either be classified qualitatively or expressed using a quantitative value,

    if they are to be ranked. A common quantitative expression for risk is the

    consequence ( expressed in a particular unit, such as number of deaths, or

    financial loss) multiplied by the probability. T his expression of risk is also

    called the expected loss, but it is only one of many options for expressing risk asa combination of consequence and probability.

    W hen we discuss risk criteria for third parties at major hazard establishments

    ( i.e. people other than employees of the establishment) , there are various

    relevant ways to express risk. T he most common terms are indivi dual r iskandsocietal risk. T he concept ofconsequence distanceis useful when one is unableor unwi lling to calculate risk quantitatively. I t can also be used i n combination

    with a ri sk matri x. T hese terms are explained further in the following sections.I t should be noted that terms have not yet been developed for assessing the

    environmental impacts of major accidents. Environmental damage is only

    discussed qualitatively. T his will be further dealt with in section 4.3.5.

    1.2.1 Risk matrices

    R isk analysis produces a list of accident scenarios, i .e. various potentialaccidents involving fire, explosion and/or the emission of dangerous

  • 7/27/2019 RAC in Denmark & EU

    10/60

    10

    substances into the environment ( air, water, or soil) . F or each of these

    scenarios, one can assess:

    T he probability (expected frequency) T he magnitude or extent of the consequences

    T his allows the scenarios to be entered in a table ranking probabi li ty

    classifications on one axis ( e.g. from frequent to extremely rare) and

    consequence classifi cations on the other axis. Such a table is called a risk

    matrix. Each field in a risk matrix represents a particular risk, and you can

    identify the sections of the matrix showing major risks ( high probabili ty and

    major consequences) and minor ri sks ( low probability and minor

    consequences) , see T able 1.

    Tabl e 1. Exampl e r isk mat r ix.Risk in creases f r om l ow to hi gh as you move al on g t he diagon al f r om t he bot t om l ef tt o t he upper r igh t cor ner. The l et t er s (A, B, C, et c.) il l ustr ate how you can ent erexampl e accident scenar io s in t he t abl e.

    Accident magnitude

    Frequency classification

    Frequencyperyear

    Undesiredevent

    Minormaterialdamageonly

    Minoraccident

    Minoroccupationalinjuries

    onsite

    Seriousaccident

    Seriousoccupationalonsite

    Majoraccident

    fatalitiesonsite,

    injuriesto

    peopleoffsite

    Disaster

    fatalitiesonandoffsite

    FrequentWill happen several timesduring lifetime ofinstallation

    1-10-2B, D, F

    LikelyWill probably, but notnecessarily, happen

    10-2 10-4 H A, C

    Not l ikelyCould possibly happen

    10-4 10-6I

    E

    Very unlikelyAlmost unthinkable

    10-6 10-8

    Extremely unlikelyFrequency is under thelimit of reasonability

    < 10-8 G

    1.2.2 Consequence distance and maximum consequence distanceO ne can calculate the maximum distance within which fatalities or injury canoccur for each potential accident. Consequence distanceis generally defined asthe distance within which death or serious injury is expected ( most uni ts for

    measuring risk and risk criteria are based on fatality alone, but this problem

    will be discussed later) . T he consequence distance is either based on the

    distance within which a particular mortality rate would be expected ( a

    mortali ty rate of one per cent is used in many studies and methods, e.g. the

    distance to a concentration level of L C 1% 1) , or the distance to a particular

    end-point value for toxicity, heat radiation, or overpressure. T he spread of

    1L C x % : L ethal C oncentration, the concentration at which x per cent of the exposed

    population will die.

    H igh risk

    Low risk

  • 7/27/2019 RAC in Denmark & EU

    11/60

    11

    toxic or explosive gas clouds will depend on meteorological conditions. O ne

    can use the worst scenario ( e.g. scenario G in T able 1) and the worst case

    meteorological factors to determine themaximum consequence distancethatapplies to the establishment in question. T he establishment wi ll not represent

    a risk to human life outside the maximum consequence distance.

    1.2.3 Location-based (individual) risk

    T he term, indivi dual r isk, is often used in relation to quantitative risk criteria.D S/IN F 85 defines indivi dual risk as the risk an individual is exposed to,

    based on their distance from the risk source. T his person-linked definition is

    problematic in relation to land use planning for major hazard establishments,

    as it involves assumptions about the movements and presence of individuals,

    which are not significant to understanding the risk situation surrounding the

    establishment. I n the N etherlands and Flanders the term location-based ri sk2has therefore been introduced. L ocation-based risk is calculated as the risk

    that a person who is continual ly present and unprotected at a given locati on wi lldieas a result of an accident within the establi shment.

    Figu r e 1. Exampl e ISO r isk cur ves showin g t he dist r ibuti on of l ocati on -based(individual ) risk sur r oun ding an ent er pr ise.

    L ocation-based risk describes the geographic distri bution of ri sk for the

    establishment in question. I t is shown using ISO risk curves, and is notdependent on whether people or residences are present ( see Figure 1) .

    L ocation-based risk is used to assess whether indi viduals are exposed to more

    than an acceptable risk in the locations where they may spend time ( e.g.

    where they live or work) . I t does not directly provide information on potential

    loss of life. N or does it distinguish exposure affecting employees or the

    general population ( By only drawing risk curves outside the company the

    indication is made that employees will not be considered in the assessment) .

    I n order to maintain continui ty with the commonly used term, this report will

    use the term location-based ( individual) risk .

    2

    I n D utch: P laatsgebonden risiko , translated as locational risk or location-basedrisk in English. T he term, locational risk , can mean something different in other

    contexts.

  • 7/27/2019 RAC in Denmark & EU

    12/60

    12

    1.2.4 Societal or group riskSocietal riskexpresses the risk that a group of people is simultaneouslyexposed to the consequences of an accident. T his is expressed using an F-

    N curve as a relationship between the expected frequency of the accident,

    and the number of people who will die (or be injured) as a result of the

    accident. F is the ( cumulative) frequency of an accident involving more than

    N deaths3, see Figure 2. T he result expresses the total expected simultaneousloss to the community. C alculation of an F-N curve takes into account the

    probability of a number of accident scenarios, and an assessment of how

    many people might be exposed to consequences under these scenarios, based

    on population density, places of work, and local protection ( whether they are

    indoors or outdoors) . Practices differ regarding the inclusion of the

    establi shments workforce as well as employees at surrounding establi shments,

    or if only the general population is included. Figure 2 shows an example

    societal risk curve. Each step in the curve represents an accident scenario.

    C ri teria for societal risk and location-based ri sk are used to complement each

    other. C ri teria for location-based risk are used to determine areas ( risk zones)

    which may not be used for residential or similar purposes, and ensure thatindividual persons are not exposed to excessive risk. C ri teria for societal risk

    are used to ensure that locations where many people may assemble are not

    exposed to excessive risk of a major accident, even if they lie outside the risk

    zones. T his is explained further in section 1.3.1.

    Figur e 2. Exampl e societal r isk cur ve f or an establ ishment .The f ir st step on t he r igh t side of t he cur ve shows t he most sever e accident scenar io(esti mat ed t o l ead to appr ox. 400 fat al it ies, wit h a f r equency of 10-9 per year ). Thesecond step shows th e con t r ibut ion f r om the second-most sever e accident (esti mat ed

    t o l ead t o appr ox. 200 fatal it ies, wit h a fr equency of 310-9 per year ).

    1.2.5 Potential Loss of Life (PLL)

    T he concept ofpotential loss of li fe( PL L ) is occasionally used when discussingrisk acceptance for third parties. T his is an easily understood term used in

    3

    I n practice, the number of deaths is used as the injury parameter, but otherparameters can be used, such as the number of injured or the extent of environmental

    damage.

  • 7/27/2019 RAC in Denmark & EU

    13/60

    13

    relation to dangerous workplaces, such as oil rigs. T he potential loss of life is

    calculated by summi ng up ( i .e. using mathematical integration) the location-

    based ( individual) risks, multiplied by the population density for each location

    ( after assessing local protection due to people being indoors or outdoors) for

    the enti re area within the maximum consequence distance. T he result i s a

    simple fi gure ( expressed as the number of deaths per year) , expressing the

    total potential loss of li fe to the community. H owever, unli ke societal risk, it

    does not take into account whether the loss results from many small accidents,

    or a few large ones. Potential loss of li fe is used indi rectly in the U nited

    K ingdom when risk acceptance criteria are formulated as the maximum

    number of people to whom exposure to a particular location-based

    ( indi vidual) risk is acceptable.

    1.3 Land use pl ann in g and major hazar d est abl ishment s1.3.1 Safety distances

    C learly, the risk of being affected by an accident at a chemical plant or a

    warehouse stori ng dangerous substances is greatest close to the risk source.

    T he risk of injury to neighbouring residents and/or damage to conservation-

    worthy natural habitats due to accidents in major hazard establishments can

    be effectively managed by controlling the distance between such

    establishments and the objects to be protected. H owever, risk zones

    surrounding major hazard establishments that cannot be used for residential

    or business purposes represent a cost to society. I t i s often very expensive to

    locate major hazard establishments at a distance greater than the maximum

    consequence distance from populated areas and other vulnerable objects.

    People are exposed to other involuntary4

    risks deriving from human activity

    ( such as traffic) , and i t is therefore seen as acceptable to expose thepopulation to a certain minor risk from major hazard establishments.

    H owever, it is difficult to define how small this risk ought to be, i .e. to lay

    down risk acceptance criteria, and to develop a method to ensure compliance

    with these criteria in practice. Figure 3 illustrates the difference between the

    maximum consequence distance and the safety distance( the distance withinwhich limitations are placed on the movement of people, to prevent them

    from being exposed to excessive location-based ( individual) risk in relation to

    the agreed risk acceptance criteria) .

    Societal risk depends on the population density within themaximumconsequence distance. T he population density within the safety distance willoften be low ( or effectively zero) , so the societal risk is determined by the

    population density betweenthe safety distance and the maximum consequencedistance. A n assessment of societal risk is therefore complimentary to an

    assessment of safety distance. Safety distances are either determined on the

    basis of location-based ( individual) risks, or reference accident scenarios

    ( please refer to the next section) .

    4A n involuntary risk is defined as a risk a person is exposed to, even though they have

    no part in or direct advantage from the activity the risk is associated with

  • 7/27/2019 RAC in Denmark & EU

    14/60

    14

    Figur e 3. Il l ust r at ion of concept s rel at ing t o l and use pl ann ing f or major hazar dest abl ishment s and t heir sur r oun ding s.The r isk is zer o at any distance gr eater t han t he maximum con sequence dist ance. Thesaf et y dist ance indicat es th e poin t at which t he r isk f al l s bel ow t he r isk accept ancecr it er ia, i.e. at gr eat er dist ances th e r isk to i ndi vidual s is accept abl e. Iso r isk curvessho w th e geogr aphi c dist r ibut ion of l ocat ion -based (individual ) ri sk.

    1.3.2 Risk analysis methodsI n order for risk acceptance criteria to work, there must be a relationship

    between the criteria, and the information generated using the risk analysis

    methods that are to be compared with the criteria. W hen using quantitative

    risk acceptance criteria ( such as a parti cular value for location-based

    ( individual) risk) , it is necessary to perform quantitative risk analyses that

    produce location-based ( individual) risks under the same condi tions as were

    used as a basis for the definition of the cri teria. V arious practices in some EUM ember States are reviewed in chapter three. T his review shows that the way

    risk acceptance criteria are formulated is strongly correlated with the methods

    used for risk analysis.

    Figu r e 4. Diagr am of act ivit ies in vol ved in r isk anal ysis

    Fareidentifikation

    Analyse af uheldsscenarier

    Hyppighed Konsekvenser

    Risiko

    Hazard identification

    Analysis of accident scenarios

    Frequency Consequences

    Risk

    Fareidentifikation

    Analyse af uheldsscenarier

    Hyppighed Konsekvenser

    Risiko

    Hazard identification

    Analysis of accident scenarios

    Frequency Consequences

    Risk

  • 7/27/2019 RAC in Denmark & EU

    15/60

    15

    T he general activities performed during risk analysis are shown in Figure 4.

    T hese activities must be performed for both qualitative and quantitative

    analyses ( although the time required for each activity in the two types of

    analyses can vary greatly) . R isk analysis begins with the identi fication of all the

    hazards that exi st for a given major hazard establishment. A number of

    hazards are then analysed in detail by considering the scenario and possible

    types of consequences. T his detailed analysis must clari fy how frequenttheevent is, and what magnitude the consequenceswould have. T hese analysesmay be qualitative or quantitative ( although consequences are usually

    analysed using quantitative methods, such as simulation of fires, explosions

    and atmospheric dissemination) . T he results are combined into an expression

    of the total riskin a way that can be compared to one or more risk acceptancecriteria ( this comparison is often called risk assessment) .

    1.3.2.1 Quantitative ri sk analysisT he aim ofquantitati ve ri sk analysisis to generate numeric values for location-based risk and societal risk that include risk contributions from all possible

    accidents.T here is a clear understanding of the content of a quantitative risk assessment

    for non-nuclear land-based major hazard establishments. T he method has, for

    example, been explained in the D utch P urple Book ( C ommittee for the

    Prevention of D isasters, 1999) . T his method analyses all accident scenarios

    expected to have an impact outside the establi shments fence5. T he frequency

    for all these scenarios is determined quanti tatively. T he consequences are

    calculated in detail using consequence models for dissemination, explosion,

    fi re and/or toxic effects. T his includes calculation of the probabili ty of fatality

    ( or injuries/damage) within the accident consequence area ( e.g. the area

    covered by a toxic cloud) . T he risk distribution surrounding the establishment

    for a given scenario is calculated, taking into account probable wind direction

    and strength. T he risks for all accidents are summed up, and the totalrepresents the geographic distribution of location-based ( individual) risk for

    the establishment. Societal risk is calculated by returning to the individual

    scenarios and determining the frequency for impact on a particular population

    area, taking into account the scenario frequency and the probability of the

    necessary wind direction and strength occurring.

    1.3.2.2 Qualitative ri sk analysisT he expression, qualitative ri sk analysis, covers a range of different methodsthat do not use numeric values ( i.e. precise figures) for location-based

    ( individual) risk or societal risk. T he term is therefore less well-defined than

    quantitative risk analysis.

    Q ualitative methods are in use because it is di fficult to determine expected

    frequencies for rare accidents. T here can be major di fferences in results from

    various analysis groups. A factor of 100 is not uncommon due to the use of

    different data sources ( L auridsen et al., 2002) . A nother argument for using

    qualitative methods is that it is impossible for people to comprehend

    frequencies as low as 10-6

    per year.

    Q ualitative methods therefore focus pri mari ly on an accidents consequences,

    and the consequence models used in quali tative methods are the same as those

    5T he Purple Book contains two specifi c criteria for the inclusion of aLoss Of

    Containment (LOC) eventin the analysis: ( 1) the frequency must be greater than 10-8

    per year, and ( 2) there must be a possibil ity of deaths ( 1 % probability) outside the

    enterprises boundary line

  • 7/27/2019 RAC in Denmark & EU

    16/60

    16

    used in quantitative methods ( although often only the distance to a particular

    damage effect is used, whereas quanti tative methods use the enti re damage

    area) .

    Based on hazard identification ( see Figure 4) , theworst case accidentcan beselected. T his accident is determined largely by basic physical factors at the

    establishment, such as the total volume of a dangerous substance ( in storage

    tanks) or the mass flow ( in pipes) , and its pressure and temperature. T he

    worst case accident might be the total collapse of a tank, or complete rupture

    of a pipe, in combination with aggravating conditions such as low wind speed

    and delayed ignition ( only once the explosive gas cloud has reached

    maximum size) . T he maximum consequence distance is calculated for these

    scenarios. A purely qualitative method that does not assess probabili ty only

    results in an expression of the maxi mum consequence distance ( this ri sk

    analysis method is sometimes called deterministic).

    T his is an unsatisfactory situation ( see section 1.3.1) , and a less serious, but

    more probable scenario is often selected. T his is sometimes referred to as the

    worst credible accident . W e prefer the more neutral term, reference accidentscenario . T he reference accident scenario is used to determine the safetydistance. T he reference scenario is perceived to be the most serious of all

    accident scenarios with a frequency high enough to represent an unacceptable

    risk, while more serious scenarios ( including the worst case scenario) are

    thought to be of such a low frequency that the risk may be discounted.

    T his approach can lead to the following problems:

    C ri teria have not been specifi ed for selecting the reference scenario( i .e. risk acceptance cri teria for the qualitative approach) .

    T he worst case scenario, and hence the maximum consequencedistance, is excluded from the risk analysis. T he maximumconsequence distance is relevant to emergency planning, and as

    described in section 1.3.1, the region between the safety distance and

    the maximum consequence distance is pivotal to the assessment of

    societal risk.

    Some consideration of probabilities cannot be avoided, even when using a

    qualitative approach. M ost qualitative methods therefore use frequency

    classifications based on qualitative verbal descriptions, like those introduced in

    the risk matrix ( T able 1, column one) . W hen such methods begin to

    incorporate probability considerations ( li ke the effect of subsequent safety

    measures on accident frequencies) , they are called hybrid methods.T he safety-barrier diagram method used in D enmark , can also be viewed as

    a hybrid method. I n this method, initiating events and safety barriers are

    allocated points, depending on the frequency of the event and the barrier

    fai lure rate (see section 2.1.2) . T he method focuses on assessment of

    initiatives to prevent accidents and reduce consequences, in contrast to the

    quantitative method described in the Purple Book. T he latter is based on

    generic ( i.e. non-site specific) frequencies for emissions of dangerous

    substances.

  • 7/27/2019 RAC in Denmark & EU

    17/60

    17

    2Earlier Danish studiesT his chapter summarises earlier D anish studies reflecting developments in thepractice of risk analysis and acceptance in D enmark . T he most important

    study is known as Environment Project 112 ( T aylor et al., 1989) . T his study

    discusses considerations in relation to the choice of ri sk analysis methods and

    risk acceptance criteria, and concludes with recommendations in this area. I t

    has subsequently been found that Envi ronment P roject 112 does not offer a

    solution for how to delimit safety zones when using qualitative risk

    assessments. T his has led to further consideration in a report on delimitation

    of safety zones for an underground natural gas storage facility in T nder,

    D enmark ( the T nder R eport ) ( D anish Environmental Protection A gency,

    1996) .

    2.1 Envir on ment Pr oject 112Envi ronment P roject 112 contains a thorough review of risk analysis methods

    and acceptance criteria. I t recognizes a link between the choice of risk

    acceptance criteria and risk analysis methods. T he report concludes that two

    analysis or assessment methods can be readily used in practice a method

    based on qualitative analysis, and a method based on quantitative analysis. A

    standards-based method i s considered impractical due to the work involved in

    provi ding the necessary standards.

    T he conclusion notes that it has been shown possible to compare results

    from the quantitative and qualitative approaches, such that these can provide

    comparable results under certain conditions . T his is an important conditionfor regulatory authorities to be able to accept the use of various methods and

    criteria, in light of the consistency requirement formulated in the most recent

    European C ommission G uidelines ( European C ommission, 2006) ( see

    section 3.1) .

    T he report highlights the principle considerations forming a basis for

    acceptance criteria. T he three most important considerations, concerning risk

    to third parties, are repeated here:

    1. The natural r isk we are exposed to in dai ly li fe should not be signi ficantlyincreased by activi ti es, such as industr y, etc., created by others without our

    personal consent .

    2. Before process plant is establi shed, i t should be investi gated whether certainprocesses can be substi tuted by other processes with a smal ler inherent riskof accident.

    3. The resources avai lable for acti vities to promote safety should be primar i lyappli ed in ways that lead to the best overal l resul t.

    Parts of these pri nciples take effect in the following requirements for approval

    of plant:

  • 7/27/2019 RAC in Denmark & EU

    18/60

    18

    1. Plant must be organised based on the A L A R A principle6, i.e. allreasonable measures must be taken to reduce the risk of accident. T his

    includes drawing on accumulated experience within the industry,

    adherence to recognised standards, and implementation of safety

    measures to counter the potential risks of the plant.

    2. I t must be demonstrated that the plant does not expose individuals orsociety to unacceptable levels of ri sk.

    3. T he advantages to society deriving from the plant must be greaterthan the risk the plant represents to society.

    Q ualitative criteria and assessment methods particularly address the fi rst

    requirement, while quantitative criteria and methods address the second

    requirement. T he third requirement involves cost-benefit analyses. T hese are

    often difficult to perform and open to uncertainty in terms of comparisons

    made between various values ( life, the environment, the economy) . T hey will

    therefore not be considered further.

    T he following issues are not covered in Environment P roject 112:

    Q uestions relating to existing versus new plant, and developmentactivities in proximi ty to major hazard establishments.

    C riteria for environmental damage.2.1.1 Recommendations in Environment Project 112 for quantitative riskacceptance criteria

    Environment P roject 112 recommends the following criteria for the technical

    assessment of plant:

    A location-based ( individual) risk of death for the most at-riskneighbour of 10

    -6per year.

    Societal risk formulated as a risk of death of 10-4 per year for anaccident involving at least one fatali ty. W here societal risk falls withinthe shaded grey region above the minimum curve, the risk should be

    A s L ow A s R easonably A chievable ( A L A R A ) .

    T hese criteria should be supplemented with a requirement that risksbe reduced as far as reasonably possible ( the A L A R A principle) , and

    that consideration be given to serious or permanent damage, and

    damage with delayed onset.

    Environment P roject 112 only makes recommendations relating to

    quantitative criteria based on number of fatalities.

    A cceptance criteria for societal risk ( see also Figure 2) are shown in Figure 5.

    Figure 5 shows the grey zone within which the above A L A R A principle

    should be used, extending over a frequency factor of 100. I n other words, if

    the risk of an accident involving at least one fatality is less than 10-6 per year( 100 times less than the minimum cri teria of 10

    -4per year

    7) , no further safety

    measures are necessary.

    6A L A R A : A s L ow A s R easonably A chievable. R isks must be reduced using all

    reasonable means, i.e. taking into account the cost of such measures.T his report

    views A L A R A and A L A R P ( A s L ow A s R easonably Practical) as synonyms. A L A R A

    has been used, as this term has been used in Environment P roject 112.7

    Expected accident frequencies are usually expressed as powers of ten, i .e. 10-4

    per

    year means that the probability of an accident is 1 in 10,000 per year. T his is the same

    as saying that, on average, one accident is expected every 10,000 years at the given

    plant, or that if there were 10,000 similar plants, anaverageof one accident would beexpected each year at any of these plants. H owever, it should be borne in mind that

    the accident could occur at any time (or place) .

  • 7/27/2019 RAC in Denmark & EU

    19/60

    19

    T he argument for these acceptance cri teria for location-based ( individual) risk

    is that:

    T hey correspond to the ri sk of natural disaster. Plants with good safety measures can realistically fulfi l the criteria in

    practice.

    T hey only increase the risk of death due to other causes by a tinyfraction ( no more than one per cent for children aged approx. 10-15

    years8) .

    T he main issues relating to societal risk criteria relate to:

    T he slope of the curve. T he absolute level ( i.e. tr imming the curve for accidents involving

    only one death) .

    W hether or not the curve should be cut off at a particular accident size( i.e. that accidents above this size are not permi tted) .

    T he argument for the selected acceptance criteria for societal risk is that:

    A slope of 2 on a logarithmic scale matches practical situations( observations of accidents and results of risk analyses) .

    A slope with a value greater than 1 places more stringent requirementson larger accidents, and thereby takes into account the extra burden

    larger accidents place on the community.

    I t can be argued that the value at one death ( 10-4 per year) does notconflict with the criteria for location-based ( individual) risk in most

    practical situations.

    Plants with good safety measures can realistically fulfi l the criteria inpractice.

    Figu r e 5. Accept ance cr it er ia f or societ al r isk, accor din g t o Envir on ment Pr oject 112.

    The pur pl e l in e in dicates t he min imum cr it er ia. The gr ey zon e in dicates where th eALARA pr in cipl e sho ul d be used.

    8

    Envi ronment Project 112 refers to foreign statistics. T he corresponding D anishstatistics are shown in Figure 8. T his fi gure shows that children aged 6-12 years have

    the lowest fatality risk in D enmark.

  • 7/27/2019 RAC in Denmark & EU

    20/60

    20

    2.1.2 Recommendations in Environment Project 112 for qualitative acceptancecriteria

    Envi ronment Project 112 defines qualitative acceptance criteria as criteria

    ensuring the safety measures in place are reasonable in proportion to the risk

    of accident. A consequence analysis is essential when using quanti tative

    methods in order to quantify in detail how the accident impacts the

    surrounding area. H owever, when using qualitative methods, a consequenceanalysis is used to qualify these accidents in terms of their potential to impact

    the surrounding area i.e. to qualify the seriousnessof a given accident.D ependi ng on the seriousness and expected frequency, requirements may be

    specified regardi ng the number and quality ( fai lure rate and effectiveness) of

    safety measures. Envi ronment P roject 112 recommends barrier di agrams9

    as a

    tool to help present the results of risk analysis. T hese diagrams show the

    possible sequences of events prior to an accident, and the safety measures

    ( hereafter referred to asbarriersor safety barr iers) that can prevent or mitigatethe accident.

    Figur e 6. Exampl e safety-bar r ier diagr am

    A nalysis procedure:

    1. A ssess the seriousness of the final event ( e.g. R eactor explodes inFigure 6) based on an analysis of the consequences this event would

    have for people, buildings, the environment, etc.

    2. D etermine the ( approximate) frequency of the initiating events ( Faultat mi xing plant and I ncorrect mixture of ingredients in Figure 6) .

    3. T he seriousness of the final event and the frequency of the initiatingevents, in combination, will determine the requirements to be placed

    on the intervening safety barriers ( three barriers in F igure 6) .

    Environment Project 112 proposes scales for the seriousness of the

    consequences ( consequence scale K , T able 2) , frequency ( frequency scale H ,T able 3) and failure rates for safety barri ers.

    Tabl e 2. Con sequence scal e K f or accident s pr oposed by Envir on ment Pr oj ect 112

    Consequencescale K

    Description of consequences

    0 No consequences events within normal plant operations that involve nodisruption or hazard

    1 Insignificant consequences minor disruption, but no hazard, and no greatimpact on production

    2 Noticeable consequences noticeable impact on production, but no injury tohumans or environmental damage, and only minor damage to equipment in

    9T he full name issafety-barr ier diagram.

  • 7/27/2019 RAC in Denmark & EU

    21/60

    21

    the vicinity of the accident3 Significant consequences less serious personal injury

    and/or significant damage to the environment or equipment in the vicinity ofthe accident

    4 Serious consequences on site events of a serious nature, but which do notaffect the plants surroundings. Plant is destroyed, and permanent injuries orfatalities occur among employees.Major accident with impact both on site and on its surroundings. Several

    permanent injuries and possibly fatalities and/or major destruction to plantwithin the enterprise, as well as impacts on the enterprises surroundings interms of permanent injuries to people and possibly fatalities, environmentaldamage, or material destruction. May be subdivided into 5.1 and 5.2:5.1 Potential for up to 10 fatalities off site and/or limited environmental

    damage

    5

    5.2 Potential for more than 10 fatalities off site and/or extensiveenvironmental damage

    Tabl e 3. Fr equency scal e H f or in it iat in g event s pr oposed by Envi r on ment Pr oject 11210

    Frequency scale H Qualitative description Magnitude (peryear)

    6 Frequent event, twice a week or more > 1005 Common event occurring one or more times a year, but

    less than twice a week1 - 100

    4 Uncommon event 0.01 - 13 Rare event

  • 7/27/2019 RAC in Denmark & EU

    22/60

    22

    Envi ronment Project 112 notes that the qualitative requirements are more

    restrictive than the quantitative requirements ( the above cri teria for i ndividual

    risk is a maximum of 10-7

    per year) . T his may be seen as a disadvantage for

    these criteria. H owever, Envi ronment Project 112 does not discuss how the

    qualitative method handles different accident scenarios that each contribute to

    risk separately. T his situation would make the qualitative cri teria less

    restrictive for the total plant, if applied per scenario.

    2.2 The Tnder Repor t I n 1996, a task force under the D anish Environmental Protection A gency and

    D anish Energy A gency, with representatives from the relevant major hazard

    authori ties, prepared a report the T nder R eport advising the regional

    authori ty ( known at that time as the C ounty of Snderjylland) about the

    location and design of a surface plant for a planned natural gas storage facility

    in T nder ( D anish Environmental Protection A gency, 1996) . T he

    recommendations made to the C ounty particularly related to the delimi tation

    of safety zones, and may therefore also be relevant in relation to other major

    hazard establishments.

    T he T nder R eport was prepared in order to highlight a number of safety

    issues related to an underground natural gas storage facility, without placing

    unnecessary restrictions on business development opportuni ties close to the

    facility.

    T he report focuses on an event considered to be a reference scenario for the

    safety distance ( restrained leakage from a 12 /16 pipe) . N o assessment was

    made of the consequences of an uncontrolled gas blow-out due to fire, as the

    probabili ty was perceived by the task force to be so small that i t should

    [not11

    ] serve as a reference event for the safety zones . T he report does not

    specify whether a) the consequences of blow-out would be greater than forrestrained leakage, or b) the probabili ty levels in question.

    Safety zones are defined as zones where rapid evacuation i s possible, and

    institutions that are difficult to evacuate may not be placed within them.

    R apid evacuation is not further explained. R eference is made to the town of

    Stenli lle, where an inner and outer safety zone have been defined. N o

    bui ldings may be erected in the inner safety zone, as is the case for safety

    zones adjacent to gas transmission pipes. T he report proposes that the inner

    safety zone for T nder be set to the consequence distance for the accident

    scenario, leakage from 12/16 pipe with restrained gas cloud. T he outer

    safety di stance has been set using the method used at the time to set the outer

    safety zone at a maximum of 200m adjacent to gas transmission pipes12

    .

    W e conclude that:1) T he inner safety distance is only based on consequences. A ccident

    frequency for the reference scenario has not been explicitly stated.

    2) T he outer safety distance is not explicitly based on eitherconsequences or ri sks, but follows a design standard.

    11T he original text omi ts the word not.

    12T he regulations relevant for the determination of safety zones can be found in

    D anish M inistry of Employment Statutory O rder no. 414/1988, supplemented by the

    W orking Environment A uthoritys supplementary provisions to the A SM E G uide

    for G as T ransmission and D istribution Piping Systems ( 1983) . A t the present time,refer to the D anish Working Environment A uthority G uide F.0.1 N aturgasanlg ,

    2001.

  • 7/27/2019 RAC in Denmark & EU

    23/60

    23

    Based on the minutes of a meeting attached to the R eport, we conclude that

    quantitative calculations were performed which support the choice of a safety

    distance of 100-200m as acceptable in comparison to E nvi ronment P roject

    112s societal risk cri teria ( the lower curve in F igure 5) .

  • 7/27/2019 RAC in Denmark & EU

    24/60

    24

  • 7/27/2019 RAC in Denmark & EU

    25/60

    25

    3Use of risk acceptance criteriawithin the EU

    T his chapter describes risk analysis methods and risk acceptance criteria in

    relation to land use planning for, or around, major hazard establishments in

    the European U nion. T he European C ommission has published two guides

    regarding practical implementation of the land use planning regulations in the

    Seveso I I D irective ( European C ommission, 1999; European C ommission,

    2006) . T he C ommission has also prepared an overview of available

    R oadmaps ( European C ommission, 2007) containing further technical

    details and implementation examples from a number of E U M ember States.

    T hese guides and the publi cation R oadmaps are described below, followed

    by a more detailed review of practices in seven selected countries ( Finland,

    Flanders, F rance, G ermany, I celand, the N etherlands and the U nitedK ingdom) .

    3.1 Land Use Pl annin g Guidel in es f r om the Eur opean CommissionT he European C ommission has prepared land use planning guidelines for

    major hazard establishments and surrounding areas ( European C ommission,

    1999; E uropean C ommission, 2006) . T hese guidelines show the various ways

    in which M ember States can fulfi l their obli gations in terms of ensuring the

    necessary distances exist between major hazard establishments and vulnerable

    objects. T his review focuses on the latest guidelines from 2006. T hese

    guidelines have been divided into parts A , B and C .

    Part A discusses general considerations to take into account when

    implementing land use planning policies for major hazard establishments and

    surrounding areas. T hese concern protection of human life, natural areas,

    surface water and groundwater. R obust land use planning policies for major

    hazard establishments and/or surrounding areas should be based on:

    C onsistency: to ensure that comparable determinations are reached i ncomparable situations.

    Proportionality: the scale of limitations ( such as safety distances)should increase in proportion to the extent of the risk.

    T ransparency ( in the decision process) .

    T hese elements have been expressed in a number of general principles,

    including the need for risk analysis methods ( to ensure consistency) and

    assessment cri teria based on damage or risk ( to ensure proporti onality) .

    Part B discusses technical issues such as:

    T ypes of risk analysis methods. R isk criteria. Selecting accident scenarios to use in the analysis and decision

    process.

    I nformation on the frequency of critical events. C onsequence modelling and damage impact.

    A number of these issues are further di scussed in sections 3.1.1 and 3.1.2below.

  • 7/27/2019 RAC in Denmark & EU

    26/60

    26

    Part C deals with envi ronment issues. M ost risk analysis methods ( and hence

    risk cri teria relevant to land use planning) focus on human li fe. T he guide

    concludes that methods to assess risk to the envi ronment are lacking, yet

    authorities are still under obligation to consider impacts to neighbouring

    sensitive natural areas, rare animal or plant species, and protected wetlands.

    I ndex methods are being used to a limited extent to quali fy the potential for

    damage by dangerous substances to the envi ronment ( using a hazard index) ,

    and to qualify the sensitivi ty of surrounding areas ( using a sensitivi ty index,

    such as sensitivi ty based on the speed pollution infi ltrate the soil) . T here have

    been attempts to define acceptance criteria based on the time required to re-

    establi sh the original state.

    3.1.1 Risk analysis methods and risk criteria

    T here are various approaches to ri sk analysis and risk cri teria within the EU .

    T he guidelines from 1999 and 2006 describe these different approaches,

    without recommending any single method. T he most recent guideline

    highlights four elements on two different axes:

    1. Q uantitative ( numeric) versus 2. Q ualitative ( non-numeric) 3. D etermi nistic ( safety is defined using specific consequence

    analyses, without considering probabili ty) versus 4. Probabili stic

    methods ( safety is defined using probabili ty distributions) .

    T hese elements exist in various combinations. Four of these combinations are

    further explained in the following sections.

    3.1.1.1 Consequence-based risk anal ysis methodsC onsequence-based risk analysis methods are also called deterministic risk

    analysis methods. C onsequence-based methods are based on an assessment of

    the (geographi c extent of) consequences of credible/conceivable accidentswithout explicitly quantifying their frequency. T he basic principle is the

    worst credibleaccident . T he philosophy is that if the necessary protection isprovided to counter the worst credible accident, this protection will also be

    adequate in the case of smaller accidents. T he probabili ty of a given accident

    is only assessed implicitly in the criteria used to determine the worst credible

    accident or reference scenario. T his assessment may be quali tative ( e.g. based

    on the number and type of safety barriers) or quantitative. M ore improbable

    events are excluded from the analysis ( see comments in section 1.3.2.2) .

    T he consequence distance for the reference scenario is calculated using one or

    more exposure threshold values ( e.g. one per cent deaths and hospi tal

    admissions) . T his method leads to delimitation of safety zones in the form of

    concentric circles.

    3.1.1.2 Ri sk- based ri sk analysis methodsR isk-based methods perceive risk as a combination of frequency and

    consequence, and are examples of probabilistic methods. T he consequences

    are analysed in the same way as in consequence-based methods, but an

    explicit assessment of the scenarios frequency is included. T hese can be

    combined with various levels of sophistication. T he most advanced methods

    are called quantitative risk analysis ( see section 1.3.2.1) . T hese methods sum

    the results from all the different accident scenarios, weighted in proportion to

    their frequency. Q uantitative risk analysis generally results in two expressionsof risk: location-based ( individual) risk, and societal risk in the form of an F-N

    curve ( see sections 1.2.3 and 1.2.4) . L ocation-based ( individual) risk is used

  • 7/27/2019 RAC in Denmark & EU

    27/60

    27

    to show the geographic distribution of ri sk, while societal risk assesses whether

    areas with high population density might be exposed to risk.

    3.1.1.3 State-of-the-A rt / Best Practi ce approachT his approach is not strictly a risk analysis method. T he underlying

    phi losophy is that the necessary safety measures must be in place to protect

    the population against the worst case accident. T his means that theestablishment needs to have considered the consequences of these worst case

    accidents, and taken the necessary preventati ve and accident-limiting steps,

    such that the ri sk outside the establishments fence is negligible ( zero-risk

    principle) . H owever, i t is recognised that it wi ll not always be possible to limi t

    accidents to the establishments own property, and therefore safety zones are

    laid out, based on an assessment of typical ( not necessari ly the worst credible)

    accident scenarios.

    3.1.1.4 H ybrid methodsR isk-based methods are mentioned as an example of hybri d methods. U nderthese methods, one of the elements (usually frequency) is assessed more

    qualitatively, i .e. using classes rather than continuous figures. U se of a ri sk

    matrix is a typical example.

    A nother hybrid method mentioned is the use of tables with fi xed distances as

    a simpli fi cation of the consequence-based method. T ables of f ixed safety

    distances are mostly used for minor or more routine situations ( e.g. F-gas

    vehicle fi lling stations) . T he guideline points out that tables are often

    conservative ( i .e. they employ relatively large safety distances) , and are mostly

    used to quickly assess which situations require more analysis.

    3.1.2 Other technical issuesT he C ommission guidelines from 2006 also discuss the data used as a basis

    for risk analysis. T he four most important elements are:

    Selecting accident scenarios. Selecting accident frequencies. M odelling end-point values. T echnical M easures as defined in article 12 of the Seveso I I

    D irective.

    T he C ommission maintains a database to assist wi th the selection of accident

    scenarios ( R isk H azard A ssessment D atabase R H A D13

    ) . T his database

    should contain information about relevant scenarios for each dangerous

    substance and activity, including frequency based on various conditions andpreventati ve measures, but so far the database only contains few relevant

    scenari os.

    T he guidelines list fi ve principles for scenario selection:

    R eference scenarios ( equivalent to reference accident scenarios inthis report) may be selected based on frequency and the severi ty of the

    consequences.

    W orst case accidents should not necessari ly be used as a basis forland use planning, but may be assessed in connection wi th emergency

    planning and evaluati on of whether the necessary measures ( Best

    13http://mahbsrv4.jrc.it/rhadnew-v3/adminindex.html

  • 7/27/2019 RAC in Denmark & EU

    28/60

    28

    P ractice) are in place to reduce the frequency of the worst case

    accident to a negligible level.

    C onsideration should be given to the time frame over which theconsequences take effect, i.e. whether there is time to activate

    preventati ve systems.

    T he effectiveness of safety barri ers should be included in theassessment.

    L and use planning is seen as both a preventative and mitigatingmeasure for establishments complying with good practices under the

    applicable standards.

    3.2 Eur opean Commissio ns exampl es of r isk anal ysis met hods(Roadmaps)

    T he C ommission guidelines from 1999 (European C ommission, 1999)

    contained a brief description of various example risk analysis methods for land

    use planning in a number of EU M ember States. T hey also contained a brief

    description of practices in A ustralia, C anada, R ussia, Switzerland and the

    U S A .

    Following the release of the guidelines in 2006, these descripti ons have been

    expanded in a separate document, called R oadmaps ( European

    C ommission, 2007) . I t is now clearly stated in the descriptions from which

    M ember State the examples derive. T here is also greater focus on the processof land use planning, with a description of the authorities and stakeholders

    involved and their powers.

    R oadmaps also contains a detailed introduction describing the various risk

    analysis methods in relation to acceptance criteria and end-point values.

    C ompared to earlier publications, there is much more focus on the use of risk

    matrices, wi th examples of a descriptive ( i .e. quali tative) consequence

    classification.End-point values are discussed for heat radiation, pressure from explosions,

    and toxicity. D istinction can be made between probit functions and fixed

    values. Probi t functions estimate the percentage of the exposed population

    that will suffer a particular injury at a particular concentration, radiation

    intensity, or overpressure. T hreshold values specify the concentration,

    radiation intensity, or overpressure that results in a pre-determined damage to

    health. T he document contains example threshold values for heat radiation

    and overpressure in the EU M ember States, and a comparison of various

    toxicity data ID L H , ER PG and A EG L14

    ( data shown in T able 15 in section

    4.3.6)

    L and use planning methods are reviewed for the following countries: U nited

    K ingdom, France, G ermany, I taly and the N etherlands. T hese countries, withthe exception of I taly, are also discussed further in the following sections. I t

    should be noted that the I talian method is very similar to F rench practices.

    14I D L H : I mmediately D angerous for L ife and H ealth ( values developed for

    emergency personnel) ; ER PG : E mergency R esponse P lanning G uidelines ( valuesdeveloped for workers, i .e. people in good health) ; A EG L : A cute Emergency

    G uidance L evel ( values being developed for the general population)

  • 7/27/2019 RAC in Denmark & EU

    29/60

    29

    3.3 Risk accept ance cr it er ia in Fin l andSource:

    Document wr i tten by Pivi Rantakoski and Leena Ahonen, Safety TechnologyAuthori ty (TUKES) 9.8.2007.

    N ew legislation on land use planning for major hazard establishments is being

    drafted in Finland. T he guiding principle for planning a major hazard

    establishment is that an accident must not result in permanent injury or

    prevent people from evacuating their homes or present location. A ll

    conceivable accident scenarios must be assessed. A n accident scenario may

    only be discounted if the operator can show that he can prevent the accident

    in every case. Scenarios are selected using quali tative cri teria.

    T he T U K ES safety authority is preparing guidelines for selecting appropriate

    end-point values for the various effects ( shock waves, heat radiation, toxicity)

    and for selecting accident scenarios.

    T he consequences of the accident for people, the envi ronment and bui ldings

    are considered. T he method to be used to implement safety distances is under

    development. I n an initial case study, three zones were placed around an

    establishment, depending on the magnitude of the consequences. V ulnerable

    objects ( schools, hospitals and nursing homes) may only be placed outside the

    outer zone, while only other establishments may be located within the

    innermost zone.

    T he method only applies to new development, i.e. new construction i n the

    vicinity of existing major hazard establishments, or the location of new major

    hazard establishments. T he method is not being used to assess whether

    existing major hazard establishments have been located appropriately in

    relation to existing land use.

    3.4 Risk accept ance cr it er ia in Fl ander sSource:

    Een code van goede prakti jken inzake risicocriteria voor externe mensrisicos vanSeveso-inrichtingen

    Flanders uses quantitative risk criteria which are the same for both new and

    exi sting column-2 and column-3 major hazard establishments. T he criteria

    are based on three values for location-based risk ( see T able 4) and a curve for

    societal risk ( see Figure 7) . I f existing establishments fail to comply with the

    criteria, additional safety measures will be required ( such as a safetyinformation plan , see below) , their environmental permi t may not be issued,

    or the government may prohibit operation of ( parts of) the establishment.

    I solated residences ( including farm bui ldings) are not counted as residential

    areas. C hildcare centres and preschools are not counted as vulnerable objects.

    W hen the iso-risk curve for 10-5

    per year passes outside the establishments

    property boundary, a safety information plan has to be drawn up describing

    how the major hazard establishment and other establishments in the

    surrounding area have agreed to exchange information about risks arising

    from dangerous substances. L ow-tier establi shments are not obliged to

    perform a quantitative risk assessment under the Flemish implementation of

    the Seveso I I D irective, but F lemi sh local authorities may require a

    quantitative risk assessment as a basis for an environmental approval.

  • 7/27/2019 RAC in Denmark & EU

    30/60

    30

    Tabl e 4. Risk cr it er ia in Fl ander s.

    Location-based risk per year Land use

  • 7/27/2019 RAC in Denmark & EU

    31/60

    31

    Tabl e 5. Fr equency cl assif icati on s for accident scenar ios under t he Fr ench met hod

    Frequency classQualitative frequencyassessment

    Semi-quantitativefrequencyassessment

    Quantitativefrequencyassessment (peryear)

    A Frequent event

    B Likely event

    C Unlikely event

    D Very unlikely event

    EPossible event with anextremely low probability

    A hybrid modelpermitting risk-

    limiting measures tobe taken intoconsideration

    T he next step is to assess the consequences. For the slow scenari o , only

    irreversible damage to assets and the environment is assessed ( as people wi ll

    have been evacuated) . For each scenari o, the distance from the source atwhich the consequence will be very serious, serious, i rreversible, or ( whereappropriate) indirectis analysed, see T able 6. I n other words, a map can bedrawn for each scenario showing three or four circles one for each level of

    seriousness ( in the case of slow scenari os, there is only one circle, for

    irreversible damage) . T hese maps for the various scenarios are then summed.

    T oxicity, heat radiation, overpressure and the slow scenarios are summed

    separately. Substances that are both flammable and toxic will result in four

    geographic maps showing assessments for each of these types of consequence.

    Some very unlikely scenarios may be excluded from the amalgamation using a

    frequency fi lter . T hese are only events in frequency class E fulfi lling extra

    requirements in terms of passive safety or a minimum number of guaranteed

    safety barriers. Such scenarios are retained for the purposes of emergency

    planning.

    T he amalgamation takes place by summing frequency classes: Four scenarios

    of class E are wri tten as 4E . Scenarios are upgraded to the next class based

    on a factor of 10, such that 10 scenarios of class E count as one D .

    T he final geographic map showing consequence distances is created by:

    1. D etermining the most serious consequence at a given location on themap.

    2. Summing the frequency classes which lead to the most seriousconsequence at this location.

    3.

    Specifying the risk at this location using a risk matri x, such as T able7.

    10-5

    10-4

    10-3

    10-2

  • 7/27/2019 RAC in Denmark & EU

    32/60

    32

    Tabl e 6. Thr esho l d val ues fo r t he Fr ench met ho d f or del imit in g t he con sequencedistance

    Consequencedescription

    Toxicity Heat radiation Overpressure (mbar)

    Limit for very serioushazard to human li fe

    LC 5%1

    Greatest distance to:8 kW/m2

    or

    1800 [(kW/m2

    )4/3

    ].s

    200

    Limit for serioushazard to human li fe

    LC 1%

    Greatest distance to:5 kW/m2

    or1000 [(kW/m2)4/3].s

    140

    Limit for significanthazard to human li fe(irreversible damage)

    Limit for irreversiblepersonal injuries

    Greatest distance to:3 kW/m2

    or600 [(kW/m2)4/3].s

    50

    Limit for indirectdamage (brokenglass)

    - - 20

    Tabl e 7. Risk mat r ix f or l ocation -based (individual ) r isk under t he Fr ench met hod f orpr epar in g l and use pl ann in g r est r ict ion s.

    Maximumconsequence

    Frequency

    Very serious Serious Significant Indirect

    >D TF+ F+ M+ Fai

    Between 5E andD

    TF F M Fai

  • 7/27/2019 RAC in Denmark & EU

    33/60

  • 7/27/2019 RAC in Denmark & EU

    34/60

    34

    homes, hospitals, etc.) large ( centres containing) stores and hotels ( defined as

    more than 1500 m2

    of f loor space) and campgrounds for over 50 people.

    For vulnerable objects, a limit value for location-based risk of 10-6

    per year

    must not be exceeded. For objects with limited vulnerability, the same value

    applies as a target, but may be exceeded under certain conditions. For existingenvironmentally approved establishments, an interim acceptance criteria of

    10-5

    per year applies, but the general limit value ( i .e. the value used for

    vulnerable objects) of 10-6 per year must be complied with by 2010.

    Figu r e 7. Combin ed pr esent at io n of accept ance cr it er ia f or societal r isk in Denmar k inaccor dance wit h Envir on ment Pr oject 112, incl udi ng t he gr ey ALARA ar ea), Fl anders(sect ion 3.4), t he Net her l ands (sect ion 3.6) and t he Uni t ed Kin gdom (indicati on of anunaccept abl e societ al r isk, see sect io n 3.8).

    1.E-11

    1.E-10

    1.E-09

    1.E-08

    1.E-07

    1.E-06

    1.E-05

    1.E-04

    1.E-03

    1.E-02

    1 10 100 1000 10000

    C

    umulativefrequency(peryear)

    Number of deaths from an accident

    EP 112

    Flanders

    The Netherlands

    UK (indicative)

    1.E-11

    1.E-10

    1.E-09

    1.E-08

    1.E-07

    1.E-06

    1.E-05

    1.E-04

    1.E-03

    1.E-02

    1 10 100 1000 10000

    C

    umulativefrequency(peryear)

    Number of deaths from an accident

    EP 112

    Flanders

    The Netherlands

    UK (indicative)

    W hen assessing permits, the authorities must compare the calculated societal

    risk with the targets for risk acceptance shown in Figure 7:

    T he expected frequency of accidents involving more than 10 deathsmust not exceed 10

    -5per year.

    For accidents involving more than 100 deaths, 10-7 per year. For accidents involving more than 1000 deaths, 10

    -9

    per year.

    W hen calculating societal risk, a maximum consequence distance is used

    which, by definition, covers the region where the expected mortality rate is

    greater than one per cent for the worst case accident.

    3.7 Risk accept ance cr it er ia in Icel andI celand uses fi xed safety distances between major hazard establi shments or

    stores of dangerous substances, and other bui ldings and types of land use.

    T hese distances are largely based on fire protection. Explicit risk

    considerations are not used.

  • 7/27/2019 RAC in Denmark & EU

    35/60

    35

    Several statutory orders specify the distance from facilities storing or using

    dangerous substances, including the statutory order for F-gas storage and

    storage of flammable liquids. H owever these statutory orders only address

    relatively small amounts (25 tonnes of F-gas and 100 m3

    of flammable liquid) .

    T he statutory order on explosives is the only statutory order addressing

    Seveso establi shments. For example, thi s statutory order specifi es a distance

    from high-tier establishments ( referring to an establishment with a 50,000

    tonne store) as shown in T able 10.

    Tabl e 10. Saf et y distances f r om h igh -t ier establ ishment s in Icel and

    Object Safety distanceHospitals, day-care homes, large assembly rooms and streets with densetraffic

    1100 m

    Residential areas 1100 mBusy roads, harbours 330 mOther buildings and public roads 330 m

    3.8 Risk accept ance cr it er ia in t he Unit ed Kin gdomSources:

    Risk cri teri a for land-use planning in the vi cin it y of major i ndustr ial hazards, UKHealth and Safety Executi ve, ISBN 11 885491 7 (1989)Proposals for revised policies to address societal risk around onshore non-nuclearmajor hazard instal lat ions, UK Health and Safety Executi ve, ConsultativeDocument CD112 (2007)

    T he U nited K ingdom uses the term, consultation distance ( C D ) , which is

    comparable to safety distance in practice. T hese consultation distances

    around each major hazard establishment are determined by the central

    authority, the H ealth and Safety Executive ( H SE) . T he H SE performs riskcalculations in each case based on information gathered via the local

    authori ties from the establishments permits ( H azardous Substances C onsent

    data, including information on quantities of dangerous substances, tank sizes,

    pressure and temperature, etc.) . T he calculations are probabilistic for toxic

    substances, but can be deterministic where the risk is due to the potential for

    fire or explosion.

    C onsultation distances are divided into three zones, so that the probabilities

    for exposure to a dangerous dose are 10-5

    ( inner zone) , 10-6

    ( middle zone) ,

    and 0.3 x 10-6

    ( outer zone) . T hese probabilities are approximately equivalent

    to individual risk or location-based risk.

    L ocal planning authorities must consult H SE regarding any development

    involving a major hazard establishment or a surrounding area ( i.e. within theouter consultation distance) . H SE has no authority to reject a permit, but may

    return a negative advi ce. H owever, the government guidelines state that

    in view of their acknowledged expertise in assessing the off- si te risks presented bythe use ofhazardous substances, any advi ce from HSE that planning permission should berefused for development for, at or near to a hazardous instal lat ion or pipeli ne shouldnot be overri dden wi thout the most careful considerati on.

    For the present, H SE only performs assessments based on indi vidual risk

    ( comparable to location-based risk) . H SE uses 10-6

    per year as the lower limit

    for i ndividual risk to the general population. A ny risk lower than this is not

    signifi cant in relation to everyday risks. A limi t of 0.310-6 per year is used for

    vulnerable people ( the aged, people vulnerable to exposure) . 10-5

    per year is

  • 7/27/2019 RAC in Denmark & EU

    36/60

    36

    used as the upper limit for acceptable involuntary risk ( involuntari ly exposed

    people include employees in surrounding establishments, etc.) .

    H SE has assessed the maximum number of people permi tted within the high-

    risk zone ( up to 10-5

    per year) . A number of more than 25 people is

    considered a signif icant ri sk. T able 11 shows which types of land use are

    considered acceptable within the various risk zones.

    Tabl e 11. Exist in g basis f or HSEs assessment of new devel opment pl ans

    A: residences, hotels, holidayaccommodation

    Negative advice when more than 25 people are exposed toindividual risk exceeding 10-5 per year, or more than 75 peopleare exposed to individual risk exceeding 10 -6 per year.

    B: workplaces, enterprises,parking areas

    A negative advice will only be given if the risk from the majorhazard establishment exceeds the normal risk for workplaceaccidents

    C: Shops, meeting rooms, sportand leisure.

    No set rules, but advice will be consistent with the principlesfor category A. Generally cut-off figures of 100 or 300 people (atpeak) are used in the two risk scenarios, respectively.

    D: Highly vulnerable or largefacili ties (hospitals, schools,large category C facil ities > 1000

    people)

    As for category A, but the risk criteria are set lower usually to

    1/3 of the acceptance criteria for category A (0.310-6 per year)

    H SE has not previously assessed societal risk, partly because it used to be

    difficult to calculate, but an easy method has now been developed requiring

    fewer resources. T here is a process underway in the U nited K ingdom to

    determine cri teria for societal risk. T here is also debate as to whether H SE s

    advisory role should be extended beyond the consultation distance defined

    above, as societal ri sk is relevant outside safety distances ( see sections 1.2.4

    and 1.3.1) . I n 2001, H SE published a report entitled, R educing R isks,

    Protecting P eople ( R 2P2) . T his report proposes that criteria for societal ri sk

    should ensure the following condition is met: the ri sk of an accident involvi ng50 or more deaths from a single event should be seen as unacceptable i f the expectedfrequency i s greater than once every 5000 years ( this condition has beenindicated in Figure 7) .

    T he approach described above is only used in connection with new

    development ( construction of new residences, enterprises, etc.) in proximity

    to existing establishments. I t is not used for approval of exi sting

    establishments.

    3.9 Risk accept ance cr it er ia in Ger manySources:

    SFK /TAA-GS-1: Strfall- K ommission technischer Ausschuss frAnlagensicherhei t, Lei tfaden.Empfehlungen fr Abstnde zwischen Betriebsbereichen nach der Strfal l-Verordnung und schutzbedrft igen Gebieten im Rahmen derBauleit planung - Umsetzung 50 BImSchG

    T he G erman Strfallkommission prepared guidelines in 2005 for

    implementing article 12 of the Seveso I I D irective on land use planning and

    ensuring the necessary distances between major hazard establishments and

    vulnerable objects. T he guidelines draw their authority from the G erman

    Envi ronmental Protection A ct. T he guidelines are not used to assess existing

    situations where an environmental permit has already been issued or wherecurrent general environmental regulations have been complied with.

  • 7/27/2019 RAC in Denmark & EU

    37/60

    37

    Special distance requirements apply to stockpiles of explosives and

    ammonium nitrate.

    D istinction is made between situations with or wi thout detailed knowledge.

    W here detailed knowledge of an enterprise is unavailable ( this is typical

    situation when planning new plant) , general distance requirements are

    recommended based on the declared substances stored or used by the

    establishment. T hese requirements are divided into four classes, with

    distances of: 200, 500, 900 and 1500m ( see T able 12) . I n D enmark these

    distances apply to establi shments that would be covered by section 4 of the

    Statutory O rder on R isk ( low tier establishments) , ( D anish Environmental

    Protection A gency, 2006) . T hese recommendations are based on

    consequence calculations using the following standard assumpti ons:

    L eakage of the dangerous substance from a 490 mm 2 hole( corresponding to a break in a D N 25 pipe

    17) .

    Flammable substances ignite immediately. A tmospheric dissemi nation calculated as described in V D I G uideline

    3783, using average meteorological conditions for industrial areas,

    such as a wind speed of 3 m/s. A threshold value of 1.6 kW /m2 for heat radiation. A threshold value of 0.1 bar for maximum overpressure for explosions

    ( average between 0.175 bar for hearing damage, and 0.05 bar for

    personal injury due to glass spli nters) .

    A threshold value for toxic substances equal to the EP R G 18 2 value.W here detailed knowledge of an industry is available ( for example, in relation

    to land use planning around existing establi shments) , consequence

    calculations are performed for the specific establishment. T hese calculations

    are based on the above assumptions, though certain freedom is permitted in

    selecting scenarios ( such as the size of a hole i t is recommended that a hole

    size greater than 80 mm2

    be used for the calculations) . W here dangeroussubstances are stored in tanks or cylinders, leakage from a single unit ( tank or

    cylinder) is assumed. A ccident-limiting measures are taken into account in

    consequence assessment. Fire, explosion and tox ic effects are assessed

    separately.

    Tabl e 12. Ger man r ecommended d ist ance requ ir ement s f or use when det ail edconsequence cal cul at ions ar e no t avail abl e

    Class 1, requireddistance: 200 m

    Class II, requireddistance: 500 m

    Class III, requireddistance: 900 m

    Class IV, requireddistance: 1500 m

    Ethylene oxideAcrylonitrile

    Hydrogen chlorideMethanolPropane (F-gas)Benzene

    Oleum 65% (Sulphurtrioxide)

    BromineAmmoniaHydrogen fluorideFluorine

    Sulphur dioxideHydrogen sulphide

    Formaldehyde(>90%)Hydrogen cyanide,HCN

    PhosgeneAcrolein

    Chlorine

    17A D N 15 pipe is assumed for phosgene.

    18EPR G : Emergency R esponse P lanning G uidelines ( see also note 14) . T hree

    concentrations are specif ied for each substance ( EPR G 1 to EP R G 3) . T he EPR G 2

    value for a substance is the airborne concentration that almost every person can beexposed to for one hour without experiencing permanent injury or effects that would

    prevent their escape.

  • 7/27/2019 RAC in Denmark & EU

    38/60

    38

  • 7/27/2019 RAC in Denmark & EU

    39/60

    39

    4Discussion of risk acceptancecriteria

    T his chapter discusses the studies presented in the preceding two chapters.

    Section 4.1 di scusses D anish studies and developments based on

    Envi ronment P roject 112. Section 4.2 discusses and compares European

    studies. Section 4.3 concludes with some general observations on choosing

    risk acceptance criteria based on an overall evaluation of the studies and

    existing practices in Europe.

    4.1 Discussio n of devel opment s wit h in Denmar kEnvironment P roject 112 provides a thorough examination of risk analysisand risk acceptance criteria issues.

    T he qualitative method developed, using safety barri er diagrams, has been

    found useful. T he method is widely used in D enmark and in some other

    countries19

    . D iscounting the fact that more experience with using various

    methods and criteria has since been gained, the work can only be criticised on

    the following points:

    1. T he qualitative method is not linked to an explicit assessment of thegeographic condi tions for the establishment in question ( distance to

    residences, extent of consequences) .

    2. T he qualitative method focuses on acceptance criteria for individualaccident scenarios, and makes no comment on how to sum risk

    contributions from various accident scenarios.

    T his has presumably been a factor in the later focus in D enmark on selecting

    a referenceaccident scenario to determine safety distances, so that the worstcaseaccid