Tixier

7/24/2019 Tixier

1/13

Journal of Loss Prevention in the Process Industries 15 (2002) 291303

www.elsevier.com/locate/jlp

Review of 62 risk analysis methodologies of industrial plants

J. Tixier a,, G. Dusserre a, O. Salvi b, D. Gaston b

a Ecole des Mines dAles, Laboratoire Genie de lEnvironnement Industriel, 6 Avenue de Clavieres, 30319 Ales Cedex, Franceb INERIS-Direction des Risques Accidentels, Parc Technologique ALATA B.P. no. 2, 60550 Verneuil-en-Halatte, France

Abstract

For about 10 years, many methodologies have been developed to undertake a risk analysis on an industrial plant. In this paper,

62 methodologies have been identified, these are separated into three different phases (identification, evaluation and hierarchisation).

In order to understand their running, it seems necessary to examine the input data, methods used, obtained output data and to rankthem in several classes. First, all the input data are grouped together into seven classes (plan or diagram, process and reaction,

products, probability and frequency, policy, environment, text, and historical knowledge). Then, the methods are ranked in sixclasses based on the combination of four usual criteria (qualitative, quantitative, deterministic and probabilistic). And finally, theoutput data are classified into four classes (management, list, probabilistic and hierarchisation). This classification permits the

appraisal of risk analysis methodologies. With the intention of understanding the running of these methodologies, the connectionsbetween the three defined previously criteria (determinist, probabilistic and determinist and probabilistic) are brought to the fore.Then the paper deals with the application fields and the main limitations of these methodologies. So the hierarchisation phase is

discussed and the type of scale used. This paper highlights the difficulties in taking into account all risks for an industrial plantand suggests that there is not only one general method to deal with the problems of industrial risks. 2002 Elsevier Science Ltd.All rights reserved.

Keywords: Industrial hazards; Risk assessment; Explosions; Fires; Toxic gas dispersion; Hierarchisation

1. Introduction

The industrial risk problem and the diversification ofrisk types have increased concurrently with industrialdevelopment. In the same time, the risk acceptabilitythreshold of the population has decreased. In responseto this preoccupation, competent authorities and industri-alists have developed methodologies and tools for riskprevention and protection, as well as crisis management.

To cope up with major accidents, a previous analysesshould be done. The forward-looking risk analysis per-mits an exhaustive identification of potential hazardoussources to prevent accident scenarios and to assesspotential impact on human, environmental and equip-ment targets in order to propose prevention or protection(Lagadec, 1980). The risk analysis methodologiesfocuses on the main hazard sources. Two principalsources of risk can be brought to the fore: industrial

Corresponding author. Tel.: +33-4-66-78-27-53; fax: +33-4-66-

78-27-01.

E-mail address: [email protected] (J. Tixier).

0950-4230/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved.

PII: S0 9 5 0 - 4 2 3 0 ( 0 2 ) 0 0 0 0 8 - 6

establishment and transport of dangerous goods. Thesetwo types of sources are quite different. At first sight,the quantities involved are not really comparable, andthe environment is unsettled for an industrial site whilstthe opposite is true for the case of transport of danger-ous goods.

So to analyse and to manage safety aspects, variousapproaches are proposed, they focus on organisationaland technical features. Sixty-two risk analysis method-ologies are set out in the following.

2. Risk analysis methodologies

The management of major industrial risk should beone of the most important preoccupations for operators.To deal with this problem, many risk analysis method-ologies were developed by industrialists and competentauthorities.

2.1. Description of these methodologies

This work identified more than 60 risk analysis meth-odologies, which can include up to three main phases:

7/24/2019 Tixier

2/13

292 J. Tixier et al. / Journal of Loss Prevention in the Process Industries 15 (2002) 291303

An identification phase based on a site description(hazardous activities, products and equipment). Those

data are necessary to develop the processes of the

methodologies.

An evaluation phase to realise a quantification of therisk. There are two ways to lead thisa deterministic

approach and/or a probabilistic approach. This evalu-ation gives the previously found consequences of

scenarios and enables their impacts on the industrial

site or on its vicinity to be taken into account.

A hierarchisation phase which aims at ranking some

results, obtained through the two previous phases, in

order to put preponderant risks forward. Thanks to

this hierarchisation, the most important risks could be

solvedfirst.

The phase of risk identification is essential, becauseit establishes the bases of the risk analysis. Indeed, the

data of risk identification will be the input of the evalu-ation and/or hierarchisation phases. Therefore, it isnecessary to make an identification phase in an exhaus-tive way to get the best results.

The phase of risk evaluation should be realised

according to two different approaches: either by theevaluation of damage consequences (deterministic

approach) or by the evaluation of accident probability

(probabilistic approach). The phase of risk hierarchis-

ation ranks the risks obtained in the previous phase, in

order to implement modifications or corrective actionson the most severe risk systems.

A risk analysis methodology does not necessarily con-

tain these three phases. It can be constituted by only thefollowing combinations: an identification phase; identi-fication and evaluation phases; or identification, evalu-ation and hierarchisation phases. Whatever method-

ologies are used, to carry out a risk analysis, three kinds

of elements are required: the expected output data; avail-

able input data; and the selected method. Indeed, users

propose some objectives to reach (output data expected),next they collect information concerning the studied sys-

tem (input data available), and finally they choose theapplied method according to the two previous elements.

2.1.1. Types of methodThe used methods can be described according to the

following four properties:

Deterministic

Probabilistic

Qualitative

Quantitative

These methodologies can be sorted out in two principal

groups, one qualitative and the other quantitative. Each

group can be divided into three categories: only deter-

ministic; only probabilistic; and a combination of deter-

ministic and probabilistic approach.

The deterministic methods take into consideration the

products, the equipment and the quantification of conse-quences for various targets such as people, environmentand equipment. The probabilistic methods are based on

the probability or frequency of hazardous situationapparitions or on the occurrence of potential accident.

The probabilistic methods are mainly focused on failure

probability of equipment or their components. On theone hand, probabilistic methods are used to lead an

analysis on a restricted part of a plant. On the other hand,

deterministic and combined deterministic and probabilis-

tic methods are used to analyse the whole industrial

establishment.

The classification of the methods is based on the typeof output data. In each category, methods can be ranked

from the simple, which comprises only one step to the

more complex ones that are based on the three steps

(identification, evaluation and hierarchisation phases).The complex methods are generally composed of mod-

ules issued from simple methods and other modules are

added in order to realise a more complete risk analysis

with easier results to analyse.

In Table 1, 62 methodologies are ranked according tothe four defined previously criteria. The great majority ofmethods are deterministic, because historically operators

and public organisations have initially tried to quantify

damages and consequences of potential accidents, before

to understand why and how they could occur.

2.1.2. Types of input dataInput data can be technical like process characteristics

or qualitative like safety policy. The analysis of 62

methods leads us to propose seven classes of input data.

For each class some information is given on the type ofinput data.

Plans or diagrams are related to the description of

the site, installation, units, fluid networks, safety bar-riers and the storages.

Process and reactions are related to operations andtasks description, physical and chemical features of

process, kinetic and calorimetric parameters,

operating conditions and normal functioning con-

ditions.

Substancesare related to the type of substance, physi-

cal and chemical properties, quantities and the toxico-

logical data.

Probability and frequency are related to the type of

failure, probability and frequency of failure, human

failure, failure rate and the exposure probability.

Policy and Management are related to the mainte-nance, organisation, safety policy, Safety Manage-

ment System, transport management and the equip-

ment cost.

7/24/2019 Tixier

3/13

293J. Tixier et al. / Journal of Loss Prevention in the Process Industries 15 (2002) 291303

Table 1

Classification of risk analysis methodologies

Risk analysis methodologies

No.a Qualitative No. Quantitative

Deterministic 1 Action Errors Analysis AEA (Rogers, 2000) 31 Accident Hazard Analysis AHI (Khan & Abbasi,1997b; Khan & Abbasi, 1998a)

2 Checklist Khan & Abbasi, 1998b 32 Annex 6 of SEVESO II Directive (La directive

Seveso II: Annexe 6, 1997)]

3 Concept Hazard Analysis CHA (Rasmussen & 33 Chemical Runaway Reaction Hazard Index RRHI

Whetton, 1997; Rogers, 2000) (Kao & Duh, 1998)

4 Concept Safety Review CSR (Rogers, 2000) 34 Dows Chemical Exposure Index CEI (American

Institute of Chemical Engineers, 1994)

5 Failure Mode Effect Analysis FMEA (Khan & 35 Dow Fire and Explosion Index FEI (American

Abbasi, 1998b; Nicolet-Monnier, 1996; Institute of Chemical Engineers, 1987; Khan &

Rogers, 2000) Abbasi, 1998a)

6 Goal Orinted Failure Analysis GOFA (Rogers, 36 Fire and Explosion Damage Index FEDI (Khan &

2000) Abbasi, 1998a)

7 Hazard and Operability HAZOP (Kennedy & 37 Hazard Identification and Ranking HIRA (Khan &

Kirwan, 1998; Khan & Abbasi, 1998b; Abbasi, 1997b; Khan & Abbasi, 1998b)

Nicolet-Monnier, 1996; Rogers, 2000;Tweeddale, Cameron, & Sylvester, 1992)

8 Human Hazard and Operability HumanHAZOP 38 Instantaneous fractionnal loss index IFAL

(Kennedy & Kirwan, 1998) (Khan & Abbasi, 1998a; Khan & Abbasi, 1998b)

9 Insurers involvement in risk reduction process 39 Methodology of domino effects analysis

(Sankey, 1998) (Dolladille, 1999)]

10 Manager (Pitblado, Williams, & Slater, 1990) 40 Methods of potential risk determination and

evaluation (Jager & Kuhnreich, 1998)

11 Optimal Hazard and Operability OptHAZOP 41 Mond Fire Explosion and Toxicity Index FETI

(Khan & Abbasi, 1997a; Khan & Abbasi, (Khan & Abbasi, 1998a; Khan & Abbasi, 1998b)

1998b)

12 Plant Level Safety Analysis PLSA (Toola, 42 SAATY methodology (Troutt & Elsaid, 1996)

1992)

13 Potential domino effects identification 43 Toxic Damage Index TDI (Khan & Abbasi,

(Delvosalle, Fievez, & Benjelloun, 1998) 1998a)

14 Preliminary Risks Analysis PRA (Nicolet-Monnier, 1996; Rogers, 2000;)

15 Process Risk Management Audit PRIMA

Hurst, Young, Donald, Gibson, & Muyselaar,

1996

16 Profile Deviation Analysis PDA (Korjusiommi,

Salo, &Taylor, 1998)

17 Safety related questions for computer

controlled plants (Chung, Broomfield, & Yang,

1998; Yang & Chung, 1998)

18 Seqhaz Hazard Mapping SHM (Korjusiommi

et al., 1998)

19 Sneak Analysis (Rogers, 2000)

20 Task Analysis TA (Rogers, 2000)

21 What if? Analysis (Khan & Abbasi, 1998b;

Nicolet-Monnier, 1996; Rogers, 2000)22 World Health Organisation WHO (Khan &

Abbasi, 1998b)

(continued on next page)

Environmentis related to the site environment, topo-

graphical data and the population density.

Text and historical knowledgeare related to the stan-

dards and regulations, and historical knowledge.

These input data are summarised in Table 2. The latter

gives connections between input data and the listed

methodologies.

Most of the methods are based on a general descrip-

tion of the site (Plans and Diagrams) and a few take

into account the Environment.The common input datasuch as:

plan or diagram;

process and reactions;

products

7/24/2019 Tixier

4/13


Table 1 (continued)

Risk analysis methodologies

No.a Qualitative No. Quantitative

Probabilistic 23 Accident Sequences Precursor ASP (Holmberg, 44 Defi method (Rogers, 2000)

1996)24 Delphi Technique (Rogers, 2000) 45 Event Tree Analysis ETA (Gadd, Leeming, &

Riley, 1998; Nicolet-Monnier, 1996; Rogers, 2000;

Tiemessen & van Zweeden, 1998;)

25 Earthquake safety of structures and 46 Fault Tree Analysis FTA (Khan & Abbasi, 1998b;

installations in chemical industries (Jezler, Nicolet-Monnier, 1996; Rogers, 2000)

1998)

47 Maintenance Analysis MA (Rogers, 2000)

48 Short Cut Risk Assessment SCRA (Rogers, 2000)

49 Work Process Analysis Model WPAM

(Davoudian, Wu, & Apostolakis, 1994)

Deterministic and 26 Maximum Credible Accident Analysis MCAA 50 AVRIM2 (Ham, van Kessel ,& Wiersma, 1998)

probabilistic (Khan & Abbasi, 1998b)

27 Reliability Block Diagram RBD (Rogers, 51 Facility Risk Review (Schlechter, 1996)

2000)

28 Safety Analysis SA (Khan & Abbasi, 1998b) 52 Failure Mode Effect Criticality Analysis FMECA(Rogers, 2000)

29 Safety Culture Hazard and Operability 53 IDEF3 (Kusiak & Zakarian, 1996; Larson &

SCHAZOP (Kennedy & Kirwan, 1998) Kusiak, 1996)

30 Structural Reliability Analysis SRA (Rogers, 54 International Study Group on Risk Analysis

2000) ISGRA (Khan & Abbasi, 1998b)

55 IPO Risico Berekening Methodiek IPORBM

(Tiemessen & van Zweeden, 1998)

56 Method Organised Systematic Analysis of Risk

MOSAR (Perhillon, 2000; Rogers, 2000)

57 Optimal Risk Assessment ORA (Khan & Abbasi,

1998b)

58 Probabilistic Safety Analysis PSA (Khan &

Abbasi, 1998b; Papazoglou, Noivolianitou,

Aneziris, & Christou, 1992)

59 Quantitative Risk Assessment QRA (Khan &Abbasi, 1998b; Leeming & Saccomanno, 1994;

Oien, Sklet, & Nielsen, 1998; Puertas, Sanz,

Vaquero, Marono, & Sola, 1998; Rogers, 2000)

60 Rapid Ranking RR (Larson & Kusiak,

1996Tweeddale et al., 1992)

61 Rapid Risk Analysis Based Design RRABD

(Khan & Abbasi, 1998)

62 Risk Level Indicators RLI (Oien et al., 1998)

a Each methodology is referred to by a number.

are generally used for deterministic methods, whiledeterministic and probabilistic methods use as mainly

input data:

probability and frequency;

plan or diagram;

products.

The more specific input data from Table 2 (Policy andManagement, environment, and texts and historical

knowledge) are principally employed in deterministic

methods.

2.1.3. Types of output dataOutput data can be qualitative like recommendations

or quantitative like index of risk level. From the review

of 62 methods, four classes of output data proposed are

as follows.

Managementis related to actions, recommendations,

modifications, and formation or operation procedures. Lists are related to lists of errors, hazards, domino

effects, causes/consequences, failures and damages,

critical activities, failure mode, accident initiators,

vulnerable place and major accident scenarios.

7/24/2019 Tixier

5/13


Table 2

Connections between input data and methodologies

Types Input data Methodologiesa

Deteministic Probabilistic Det and prob

Plan and diagramSite Qualitative 2, 3, 4, 5, 7, 8, 11, 13, 23 26, 27, 30

14, 15, 18, 19, 21, 22

Installations

Units

Fluid or gas networks

Functionning Quantitative 31, 34, 35, 36, 37, 38, 39, 45, 46, 47 50, 51, 52, 53,

40, 41, 42, 43 56, 57, 59, 61, 62

Safety barriers

Storages

Process and reactions

Operations description Qualitative 2, 3, 7, 8, 10, 11, 12, 14, 16, 19, 20, 21

Tasks description

Reactions and physical

and chemical featuresProcess characteristics

Kinetics and Quantitative 33, 35, 36, 37, 40, 41, 42, 43 54

calorimetrical parameters

Normal functionning

conditions

Operating conditions

Products

Products types, physical qualitative 11, 12, 13, 14, 16, 26

and chemical properties

Quantities

toxicological data Quantitative 31, 32, 33, 34, 35, 36, 38, 40, 41, 42, 43 51, 54, 55, 56, 58

Probability and frequency

Failure type Qualitative 1, 12, 23, 24 25 26, 27, 28, 29, 30Failure probability

Initiation and failure

frequencies

human failure Quantitative 44, 46 ,47, 48, 49 50, 51, 52, 53,

58, 59, 60

Failure rate

Exposure probability

Policy and Management

Maintenance Qualitative 2, 9, 10, 15, 17 29

Organisation

Safety policy

SMS Quantitative 36, 42 49 51, 55

Transport Management

Equipments cost

Environment

Site environment Qualitative 11, 19

Topographical data

population density Quantitative 31, 34, 36, 37, 42 51

Texts and historical knowledge

standards Qualitative 4, 5, 13, 14, 18

Regulations and Quantitative 31, 35, 39, 41, 42 51, 52, 59, 61

Documents

Historical knowledge

a The number refers to the methodologies presented in Table 1.

7/24/2019 Tixier

6/13


Probabilisticis related to failure rate, reliability, scen-

arios or damages probability, and accident frequency.

Hierarchisationis related to level risk index, severity

and criticality, fire, explosion, toxic leakage index,organisational index, classification according to thetype of risk.

The connections between output data and method-

ologies are presented in Table 3. Output data like Man-

agement and Lists are based on expert choices and givequalitative results while output data like Probabilistic

and Hierarchisation give quantitative results. The results

are mainly in the form of lists. The deterministic and

probabilistic methods provide several types of results, so

more information can be obtained with this approach.

Table 3

Links between output data and methodologies

Types Output data Methodologiesa

Deteministic Probabilistic Det and prob

Management

Actions Qualitative 3, 4, 5, 6, 7, 9, 10, 11,

15, 16, 20

Recommendations

modifications Quantitative 39 50, 51, 59, 60

Formation and

operation procedures

List

List of errors Qualitative 1, 2, 5, 6, 7, 8, 11, 12, 23, 24 26, 28, 29, 30

13, 14, 17, 19, 21, 22Estimation/list of risks

List of domino effects

List cause/consequence

failure, damage

List of installation

critical activities

List of failure mode Quantitative 39 46, 47, 48, 49 50, 51, 53, 54, 56, 57,

59, 60, 61

List accident initiators

List of vulnerable

place

List of major scenarios

Probabilistic b

Failure rate Qualitative 23, 24, 25 26, 27, 28, 29, 30

ReliabilityScenarios or damages Quantitative 44, 45, 49 53, 54

probability

Accident frequency

Hierarchisation

Risk index/level Qualitative 11, 18 25

Severity/criticity

Fire/explosion index

Toxic leakage index Quantitative 31, 32, 33, 34, 35, 36, 51, 52, 55, 56, 58, 61,

37, 38, 40, 41, 42, 43 62

Organisational risk

index

Type risk classification

a The number refers to the methodologies presented in Table 1.

2.1.4. Links between input data, output data and

methods

Now, it is relevant to underline how links between

input data, method and output data are running. Table 4

can be used according to whether the user expects someresults or has some available data:

First, if industrialists need a certain type of results,

then they will read through the results (output data)

columns given in Table 4. So different types ofmethods are proposed and finally the necessary inputdata can be identified

Secondly, if only several input data are available, then

the user will read through the input data columns

given in Table 4. The combination of available input

7/24/2019 Tixier

7/13


Table4

Linksbetweeninputdata,methodsando

utputdataa

a

Thenumberreferstothemethodolog

iespresentedinTable1.

7/24/2019 Tixier

8/13


data permits the identification of methods which areconceivable to use in the risk analysis.

Table 4 is a synthesis of this study and a tool for an

identification of methods which could be used accordingto objectives and available input data.The analysis of

Table 4 highlights that many input data are necessary torealise a:

qualitative and deterministic risk analysis;

quantitative and deterministic risk analysis;

quantitative and deterministic and probabilistic risk

analysis.

Whatever qualitative or quantitative methods, results are

complete when both deterministic and probabilisticmethods are used.

Probabilistic methods need some input data, but they

do not take into account some specificities of the indus-trial establishment like Policy or Environment. Now, the

running of methodologies have been brought to the fore,

and it is necessary to discuss two important points: on

the one hand, the application fields of those method-ologies and on the other hand, their main limitations.

2.2. Application fields of methodologies

The applicationfield of these different methodologiescan be ranked into three categories (Table 5). First,

which is the most important in number of developed

methodologies, concerns industrial site. Generally, some

methodologies are developed for specific application orprocess and they are not transposable to different typesof industrial establishment. The second application fieldis the transportation of dangerous goods and the third

one permits to take into account human factors in a spe-

cific environment. Some methodologies can be used forvarious purposes and several application fields. Forexample among the most general methods, there are

What if and Safety Analysis.

2.3. Limitations of methodologies

The main limitations of those methodologies can be

summarised in the following points.

The more general the methodology is, the less it takes

into account the specificities of the studied case. On the contrary, if the methodology is too specific it

will be less transposable to another case.

Knowledge of people, who are participating in the

risk analysis, is quite important (different types of

competences and levels of people involvement).

For probabilistic analysis, the validity of data is a

decisive parameter.

The updating of data takes a lot of time work.

For some methodologies, the operational application

is difficult to realise because of the lack of descrip-tion. It is useful to provide a guide book to explain

how methodologies could be used.

The complexity of methods requires specific trainingfor their implementation.

It can be noticed that there is a great disconnectionbetween risk analysis methodologies and human fac-

tors.

A new trend highlights a need to have more quickly

exploitable results. So, the output data of evaluation

phase must be completed by an additional step, i.e. the

hierarchisation phase.

3. The risk hierarchisation

The recent evolution in risk analysis methodologies

shows that easily applicable methods are proposed with

a risk level index as a result. The hierarchisation consists

of the action to organise some elements, data, or events

in increasing (or decreasing) order with the help of

classification tabs, with the view to bringing out the mainpoints for analysis. These methodologies are simple andrapid to use, the specific hierarchisation rules in theshape of data table are usually provided by the authorsmethods for the hierarchisation phase. The association

between risk level and geographical zones on site is pri-

mordial to have a good safety management.

These different methodologies including hierarchis-

ation step are gathered in Table 6. Methodologies withhierarchisation phase are almost all quantitative and

deterministic. They are based on the development of a

risk level index, calculated for each element, unit or

studied area in order to rank them. This ranking providesthe way to clearly identify critical areas on industrial

site. Thus it permits to realise priority actions with the

view to decreasing the probability of occurrence

(prevention) or to reducing the consequences of an acci-

dent (protection including emergency response).

To calculate this risk level index, several parameterscharacterising the site and processes are identified. Thenthese parameters are ranked with the help of scales

which are provided in a guide and which can be based

on a deterministic or probabilistic approach.

The deterministic scales can be quantitative or quali-

tative for the following variables.

Internal hazard of substances and equipments:

reaction types (hydrolysis, oxidation, reduction,

polymerisation,); reaction parameters (stability, reactivity, exother-

mically, pressure or temperature of reaction,); physical and chemical properties of substances, their

7/24/2019 Tixier

9/13

7/24/2019 Tixier

10/13


Table5(continued)

Methodologies

Transport

Checklist(Khan&A

bbasi,1998b)

HazardandOperabilityHAZOP(Kennedy&Kirwan,1998;Khan&Abbasi,1998b;

Nicolet-Monnier,1996;Rogers,2000;Tweeddaleetal.,1992)

EventTreeAnalysis

ETA(Gaddetal.,1998;Nicolet-Monnier,

1996;Rogers,2000;IPORisicoBerekeningMethodiekIPORBM(Tiemessen&van

Zweeden,1998)

Tiemessen&vanZw

eeden,1998]

FailureModeEffect

AnalysisFMEA(Khan&Abbasi,1998b;Nicolet-Monnier,

QuantitativeRiskAssessmentQRA(Alonso&Gavalda,1998;

Khan&Abbasi,

1996;Rogers,2000)

1998b;Leeming&Saccomanno,1994;Oienetal.,1998;Puertasetal.,1998;

Papazoglouetal.,1992;Rogers,2000)

FaultTreeAnalysisFTA(Khan&Abbasi,1998b;Nicolet-Monn

ier,1996;Rogers,

Whatif?Analysis(K

han&Abbasi,1998b;Nicolet-Monnier,1996;Rogers,2000)

2000)

Human

ActionErrorsAnalysisAEA(Rogers,2000)

SafetyCultureHazardandOperabilitySCHAZOP(Kennedy&

Kirwan,1998)

HumanHazardandOperabilityHumanHAZOP(Kennedy&Kirwan,1998)

TaskAnalysisTA(R

ogers,2000)

Manager(Pitbladoetal.,1990)

WorkProcessAnaly

sisModelWPAM(Davoudianetal.,1994)

ProcessRiskManagementAuditPRIMA(Hurstetal.,1996)

7/24/2019 Tixier

11/13


Table 6

Risks analysis methodologies with hierarchisation step

Methodologies with hierarchisation phase

Accident Hazard Analysis AHI (Khan & Abbasi, 1997b; Khan & Abbasi, 1998a)

Annex 6 of SEVESO II Directive (La directive Seveso II: Annexe 6, 1997)

Chemical Runaway Reaction Hazard Index RRHI (Kao & Duh, 1998)Dows Chemical Exposure Index CEI (American Institute of Chemical Engineers, 1994)

Dows Fire and Explosion Index FEI (American Institute of Chemical Engineers, 1987; Khan & Abbasi, 1998a)

Earthquake safety of structures and installations in chemical industries (Jezler, 1998)

Facility Risk Review (Schlechter, 1996)

Failure Mode Effect Criticality Analysis FMECA (Rogers, 2000

Fire and Explsion Damage Index FEDI (Khan & Abbasi, 1998a)

Hazard Identification and Ranking HIRA (Khan & Abbasi, 1997b; Khan & Abbasi, 1998b)

Instantaneous fractionnal loss index IFAL (Khan & Abbasi, 1998a; Khan & Abbasi, 1998b)

Methodology of domino effects analysis (Dolladille, 1999)]

Methods of potential risk determination and evaluation (Jager & Kuhnreich, 1998)

Mond Fire Explosion and Toxicity Index FETI (Khan & Abbasi, 1998a; Khan & Abbasi, 1998b)

Potential domino effects identification (Delvosalle et al., 1998)

Probabilistic Safety Analysis PSA (Khan & Abbasi, 1998b; Papazoglou et al., 1992)

Risk Level Indicators RLI (Oien et al., 1998)

SAATY methodology (Troutt & Elsaid, 1996)Seqhaz Hazard Mapping SHM Korjusiommi et al., 1998

Toxic Damage Index TDI (Khan & Abbasi, 1998a)

toxicity with a correlation dose/effect and the studyof their incompatibility;

quantities of substances used and stored;

storage characteristics (pressure, temperature,).Severity of the consequences:

human damage types ( over pressure blast, thermal

flux, toxicity); equipment damage types;

environmental damage types;

financial loss on equipment or production.Layout and environment:

distances between dangerous units of an industrial

site;

population density inside and outside the industrial

site.

The probabilistic scales are quantitative and they are

made of variables which are:

occurrence frequency of hazardous events;

incident or accident frequencies, historical knowl-

edge;

consequences probabilities (fatalities, structure dam-

ages, ground or water pollution, ).

The various scales are developed for several types of

application according to the particularities of certainindustrial site in order to rank the risk in consistent man-

ner. To generalise hierarchisation scale, it is necessary

to elaborate an exhaustive list of influencing risk factors

from many industrial sites and to take into accounthuman, environmental and equipment damages.

In fact, the hierachisation phase gives more advanced

data processing in order to rank risks on the studied area

(risk level index). This ranking provides help for

decision-makers (industrialists and competent

authorities).

4. Conclusion

The use of risk analysis methodologies contributes to

the prevention of accidents and to the preparation for

emergency response. This work based on the review of

62 methodologies underlines the difficulty in taking intoaccount all risks for an industrial site. This paper high-

lights the different types of input data, methods, output

data and their links. A risk analysis methodology can besimple and only focus on the identification of hazardsor a combined risk analysis methodology. A combined

risk analysis methodology can be composed of several

simple risk analysis methodologies, with an identifi-cation, estimation and hierarchisation phases in order to

obtain a risk level index, for example. The recent devel-opment of risk analysis methodologies strives towards

easily applicable methods with a hierarchisation phase,

which is based on specific scales depending on the stud-ied installation. The application fields of methodologiesare industrial site, transport of hazardous goods andhuman factors. The human factor risk analysis is often

disconnected with classical risk analysis that is due to

the complexity of human risk analysis. The types of

7/24/2019 Tixier

12/13


results are recommendations, lists, risk level index, event

frequency and damage probability.

The 62 methodologies identified show that there is nota uniqueness of methods to realise a risk analysis. On

the contrary, it is necessary to combine several method-ologies. The application of these methodologies requires

experience to obtain good results. In fact, the acquiredknowledge through the analysis of the 62 methodologies

can constitute a starting point to elaborate a new method-

ology. To elaborate a methodology it seems appropriateto propose an initial draft of methods. Then its appli-

cation on real industrial sites would improve it. There-

fore, its transferability to other cases would be easier.

The whole remarks, presented in this study, aim to

propose a draft of an ideal risk analysis methodology.

First, the studied area must be put in four parts to leadthe risk analysis:

the source term (industrial establishment);

the flux (vector of propagation of accidents); targets (human, environmental and equipments);

control and management.

These four identified parts must be described in anexhaustive way and together with their interactions.Then, the mainspring of risk analysis should permit a

deterministic and probabilistic approach with a hier-

archisation phase and finally, the output data could beof two different types:

qualitative in order to provide recommendations;

quantitative in order to evaluate the main conse-quences

This methodology presents an overall process of risk

analysis in order to provide some ways of improvement

and help in decision-making.

References

Alonso, C., & Gavalda, J. (1998). A method to determine environmen-

tal risk in chemical process industries. In Proceeding from ninth

international symposium loss prevention and safety promotion in

the process Industries (pp. 12191227).DOWs Fire & Explosion Index Hazard Classification Guide. (1987).

American Institute of Chemical Engineers.

DOWs Chemical Exposure Index Guide. (1994). American Institute

of Chemical Engineers (1st ed.).

Chung, P. W. H., Broomfield, E., & Yang, S. H. (1998). Safety related

questions for computer-controlled plants: Derivation, organisation

and application. Journal of Loss Prevention in the Process Indus-

tries, 11, 397406.

Davoudian, K., Wu, J. S., & Apostolakis, G. (1994). The work process

analysis model (WPAM). Reliability Engineering and System

Safety, 45, 107125.

Delvosalle, C., Fievez, C., & Benjelloun, F. (1998). Development of

a methodology for the identification of potential domino effects in

Seveso industries. In Proceeding from ninth international sym-

posium loss prevention and safety promotion in the process indus-

tries (pp. 12521261).

Dolladille, O. (1999). Proposition dune methode danalyse deffet

domino. Preventique-securite, 44 , 6270.

Gadd, S. A., Leeming, D. G., & Riley, T. N. K. (1998). Transport

Riskat: The HSE quantified risk assessment tool for toxic and

flammable dangerous goods transport by road and rail in Great Bri-

tain. In Proceedings from ninth International symposium loss pre-vention and safety promotion in the process industries (pp. 308

317).

Ham, K. J. M., van Kessel, H. J. C. M., & Wiersma, T. (1998). Experi-

ences with a safety report according to Seveso II: A pilot project

in the Netherlands. In Proceeding from ninth international sym-

posium loss prevention and safety promotion in the process Indus-

tries (pp. 13261340).

Holmberg, J. (1996). Risk follow up by probabilistic safety assess-

ment-experience from a finish pilot study. Reliability Engineering

and System Safety, 53, 315.

Hurst, N. W., Young, S., Donald, I., Gibson, H., & Muyselaar, A.

(1996). Measures of safety management performance and attitudes

to safety at major hazard sites. Journal of Loss Prevention in the

Process Industries, 9(2), 161172.

Jager, P., & Kuhnreich, K. (1998). Approach to a systematic determi-nation and evaluation of risk potential. In Proceeding from ninth

international symposium loss prevention and safety promotion in

the process industries (pp. 393403).

Jezler, W. (1998). Earthquake safety of structures and installations in

chemical industry in the context of risk analysis. In Proceeding

from ninth international symposium loss prevention and safety pro-

motion in the process industries (pp. 414421).

Kao, C. S., & Duh, Y. S. (1998). Chemical runaway reaction hazard

index and risk assessment. In Proceeding from ninth international

symposium loss prevention and safety promotion in the process

industries(pp. 965975).

Kennedy, R., & Kirwan, B. (1998). Development of a hazard and

operability-based method for identifying safety management vul-

nerabilities in high risk systems. Safety Sciences, 30, 249274.

Khan, F. I., & Abbasi, S. A. (1997a). OptHazopan effective and

optimum approach for Hazop study. Journal of Loss Prevention in

the Process Industries, 10(3), 191204.

Khan, F. I., & Abbasi, S. A. (1997b). Accident hazard index: A multi-

attribute method for process industry hazard rating. Process safety

and environmental protection, 75(4), 217224.

Khan, F. I., & Abbasi, S. A. (1998). Inherently safer design based

on rapid risk analysis. Journal of Loss Prevention in the Process

Industries, 11, 361372.

Khan, F. I., & Abbasi, S. A. (1998a). Multivariate hazard identi fication

and ranking system. Process Safety Progress, 17(3), 157170.

Khan, F. I., & Abbasi, S. A. (1998b). Techniques and methodologies

for risk analysis in chemical process industries. Journal of Loss

Prevention in the Process Industries, 11, 261277.

Korjusiommi, E., Salo, R., & Taylor, R. (1998). Hazard analysis forbatch processes and for special operations. In Proceeding from

ninth international symposium loss prevention and safety pro-

motion in the process industries (pp. 422431).

Kusiak, A., & Zakarian, A. (1996). Risk assessment of process models.

Computers & Industrial Engineering, 30(4), 599610.

La directive Seveso II: Annexe 6. (1997).edition legislative, bul 230,

69336947.

Lagadec, P. (1980). Politique, risque et processus de developpement:

Le risque technologique majeur, These de doctorat detat en

science politique, Universitedes Sciences Sociales de Grenoble II.

Larson, N., & Kusiak, A. (1996). Managing design processes: A risk

assessment approach. IEEE Transactions on Systems, Man, and

CyberneticsPart A: Systems and Humans, 26(6), 749759.

Leeming, D. G., & Saccomanno, F. F. (1994). Use of quanti fied risk

7/24/2019 Tixier

13/13


assessment in evaluating the risks of transporting chlorine by road

and rail. Transportation Research Record, 1430, 2735.

Nicolet-Monnier, M. (1996). Integrated regional risk assessment: The

situation in Switzerland. International Journal of Environment and

Pollution, 6(4-6), 441461.

Oien, K., Sklet, S., & Nielsen, L. (1998). Development of risk level

indicators for petroleum production platform. In Proceeding from

ninth international symposium loss prevention and safety pro-motion in the process industries (pp. 382393).

Papazoglou, I. A., Noivolianitou, Z., Aneziris, O., & Christou, M.

(1992). Probabilistic safety analysis in chemical installation. Jour-

nal of Loss Prevention in the Process Industries, 5(3), 181191.

Perhillon, P. (2000). Elements methodiques. Phoebus, la revue de la

surete de fonctionnement, Lanalyse de risques, 12, 3149.

Pitblado, R. M., Williams, J. C., & Slater, D. H. (1990). Quantitative

assessment of process safety programs. Plant/Operations Progress,

9(3), 169175.

Puertas, I., Sanz, J. C., Vaquero, C., Marono, M., & Sola, R. (1998).

Procedure for the review of quantitative risk assessment of the pro-

cess industries. In Proceeding from ninth international symposium

loss prevention and safety promotion in the process industries (pp.

283288).

Rasmussen, B., & Whetton, C. (1997). Hazard identification on plantfunctional modelling. Reliability Engineering and System Safety,

55, 7784.

Rogers, R.L. (2000). The RASE Project risk assessment of unit oper-

ations and equipment, http://www.safetynet.de/EC-Projects/. pp.

1-50.

Sankey, P. D. (1998). An insurers involvement in the risk reduction

process. In Proceeding from ninth international symposium loss

prevention and safety promotion in the process industries (pp.

441450).

Schlechter, W. P. G. (1996). Facility risk review as a mean to address-

ing existing risks during the life cycle of a process unit, operationor facility. International Journal of Pressure Vessels and Piping,

66(13), 387402.

Tiemessen, G., & van Zweeden, J. P. (1998). Risk assessment of the

transport of hazardous materials. In Proceeding from ninth inter-

national symposium loss prevention and safety promotion in the

process industries(pp. 299307).

Toola, A. (1992). Plant level safety analysis. Journal of Loss Preven-

tion in the Process Industries, 5(2), 119124.

Troutt, M. D., & Elsaid, H. H. (1996). The potential value of SAATY s

eigenvector scaling method for short-term forecasting of currency

exchanges rates. Siam Review, 38, 650654.

Tweeddale, H. M., Cameron, R. F., & Sylvester, S. S. (1992). Some

experiences in hazard identification and risk shortlisting.Journal

of Loss Prevention in the Process Industries, 5(5), 279288.

Yang, S., & Chung, P. W. H. (1998). Hazard analysis and support toolfor computer controlled processes. Journal of Loss Prevention in

the Process Industries, 11, 333345.