1-s2.0-s0950423012001702-main

9

Click here to load reader

Upload: errrrrrrrrrr

Post on 04-Jun-2018

215 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: 1-s2.0-S0950423012001702-main

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 19

Application of a multi-plant QRA A case study investigating the riskimpact of the construction of a new plant on an existing chemicalplantrsquos risk levels

Shahabaldin Baesi a Bahman Abdolhamidzadeh a Che Rosmani Che Hassan aMahar Diana Hamid a Genserik Reniers bc

a Department of Chemical Engineering Faculty of Engineering University of Malaya 50603 Kuala Lumpur Malaysiab Antwerp Research Group on Safety and Security (ARGoSS) University of Antwerp Prinsstraat 13 2000 Antwerpen Belgiumc Hogeschool-UniversiteitBrusselKULeuven Belgium

a r t i c l e i n f o

Article history

Received 10 April 2012

Received in revised form

25 October 2012

Accepted 6 November 2012

Keywords

Quantitative Risk assessment

Individual risk

Societal risk

Domino accidents

Chemical cluster

Process industries

a b s t r a c t

The construction of chemical clusters whereby a variety of chemical plants are located next to each other

provides great economic bene1047297ts However in such clusters due to the mere scale on which hazardous

materials are processed stored and handled the potential of various accidents is much higher than in

single companies Furthermore the close proximity of process installations and storage tanks in such

areas gives rise to the risk of domino effects Therefore land use planning and layout design has always

been a challenge within such clusters

In this paper a Quantitative Risk Assessment (QRA) is carried out and used as a decision making tool to

evaluate the acceptability of constructing a new chemical plant adjacent to an existing one For this

purpose standard parameters such as individual risk and societal risk were quanti1047297ed before and after

the new plant would come into operation Given the experience of past accidents in the process

industries the likelihood of domino accidents in the two neighboring plants has also been analyzed

2012 Elsevier Ltd All rights reserved

1 Introduction

Due to the 1047298ammable and toxic nature of substances which are

being handled in the oil- gas- and petrochemical industries

installations within chemical plants have a high potential to cause

substantial damages in terms of fatalities serious injuries property

damages and environmental degradation In addition large

inventories of hydrocarbons intense temperature and pressure

conditions and inherent congestion in process installations

whereby process equipment is often situated in close proximity to

one another increases the probability of catastrophic accidents andampli1047297es their potential consequences (Abdolhamidzadeh Abbasi

Rashtchian amp Abbasi 2010 Khan amp Abbasi 1999a 1999b Reniers

2010) For example in Toulouse France 30 people were killed and

2242 injured in 2001 dueto an ammonium nitrate explosion and in

Texas city USA 15 persons lost their lives and 170 others were

injured in 2005 due to a re1047297nery disaster (Dechy Bourdeaux

Ayrault Kordek amp Le Coze 2004 Kalantarnia Khan amp Hawboldt

2010) We refer eg to Lees (1996) Wells (1997) Kletz (1999a

2003) and Atherton and Gil (2008) for many other examples of

major accidents in the chemical industries Such catastrophic

events also happen in developing countries and especially in the

Middle East (Khan amp Abbasi1999a) This may be related to the ever

growing number of chemical clusters in this particular region due

to the existence of massive sources of energy in the Persian Gulf In

chemical clusters upstream activities such as production and

separation are increasingly carried out in chemical plants being

physically located nearby chemical companies with downstream

activities such as re1047297neries

One of the important issues in the construction of chemicalclusters is so-called land use planning Land use planning is in1047298u-

enced by economic operational and safety aspects The high

number of hazardous activities and substances in chemical clusters

in combination with high levelsof congestion demand an adequate

and solid decision making process for situating new plants in the

area whereby individual and group risks are taken into account

Usually so-called Quantitative Risk Assessment or abbreviated

QRA is used for this purpose Previous studies indicate that QRA is

a useful tool forlanduse planninglayout designand modi1047297cation in

the process industries Khan and Abbasi (1999b) have estimated the

overall individual risk posed by a chemical cluster to adjacent Corresponding author

E-mail address abdolhamidzadehumedumy (B Abdolhamidzadeh)

Contents lists available at SciVerse ScienceDirect

Journal of Loss Prevention in the Process Industries

j o u r n a l h o m e p a g e w w w e l s e v i e r c om l o c a t e j l p

0950-4230$ e see front matter 2012 Elsevier Ltd All rights reserved

httpdxdoiorg101016jjlp201211005

Journal of Loss Prevention in the Process Industries 26 (2013) 895e903

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 29

residential areas in order to compare the result with safety

criteria and eventually recommendations were given to make

modi1047297cations in the plants since the criteria were not met Jo and

Crowl (2008) speci1047297ed the minimum safety distances between

a high pressure gas pipeline and residential areas Risk and conse-

quence analyses of toxic chemicals in certain warehouses to the

nearby villages were analyzed by Rigas and Sklavounos (2002) to

investigate the extent of compliance to exposure threshold limits

The study by Papazoglou Nivolianitou Aneziris Christou and

Bonanos (1999) demonstrated that a high level of risk is exposed

to passengers crossing the highway adjacent to a speci1047297c re1047297nery

and thereforea new trajectory withthe safety distancecalculated by

QRAshould be constructed These area limited numberof examples

showing that land use planning is vital forlocating process facilities

near to public areas and a QRA could be an effective technique for

this purpose In this method in addition to identifying the hazards

also their associated consequences and the frequencies of occur-

rences are quantitatively estimated (Kletz 1999b) Furthermore

parameters such as lsquoindividual riskrsquo and lsquogroup riskrsquo (the latter also

being called lsquosocietal riskrsquo) are parameters which support the deci-

sion making process for land use planning Several other relevant

studies can be found elsewhere (Bubbico Maschio Mazzarotta

Milazzo amp Parisi 2006 Gharabagh et al 2009 Milazzo et al2002 Yet-Pole Chi-Min amp Ching-Hong 2009)

The proliferation of chain accidents suggests that the probability

of domino effects should also be considered as an important

parameterin land useplanning forchemical clusters A recent study

on past domino events indicated that the number of cascading

accidents in the process industries has increased globally in recent

decades (Abdolhamidzadeh Abbasi Rashtchian amp Abbasi 2011) In

the same study there are some accidents mentioned in which an

initiating event in a single plant has led to a catastrophe within

a clusterof chemical plants A recentexample couldbe Shazand Iran

in which 30 peoplewere killedand 38 others were seriously injured

in 2008due to an initial explosion that ledto other majorexplosions

and 1047297res in the neighboring plants The dominant accident

that caused the majority of fatalities was the major explosions thatwere triggered by the initial blast (Abdolhamidzadeh et al 2011)

As already mentioned in land use planning possible scenarios

are reviewed from different points of view one being the safety

perspective In many cases different chemical plants in a chemical

industrial area do not come into operation at the same point in

time Forexample it is common practice to build plants to consume

the product(s) of an existing plant It is obvious that the construc-

tion of new plant(s) adjacent to existing one(s) will affect the

overall risk of the industrial area regardless of the fact whether risk

assessments have been performed for individual plants

In the remainder of this article QRA has been applied to eval-

uate the risk-based effects of constructing a new chemical plant

(which is called BSPC in this article) adjacent to an existing one

(called AKPC in this paper) These two plants forming our case-study are located in one of the largest energy zones in the world

in southern Iran (called lsquoPETZONErsquo) Parameters such as the indi-

vidualrisk and the societal risk have been assessedbefore and after

the new plant came into operation Furthermore the possibility

and the likelihood of domino events in the two neighboring plants

were investigated We mainly focus on safety aspects in our article

as the feasibility of the planning from both an economic and an

operational perspective has been carried out in an earlier stage of

the land-use planning decision process

2 Quantitative Risk Assessment (QRA)

QRA along with other techniques have been used in one way or

another in the process industries for loss prevention from 1970

onwards (Abbasi Krishnakumari amp Khan 1998 Khan amp Abbasi

1998) QRA is a method that provides quanti1047297ed estimation for

the risk posed by a group of hazards (Kletz 1999) Hence this

technique enables risk mitigation methods to be evaluated in order

to bring the risk to tolerable levels without resorting to too costly

protective systems (Pula Khan Veitch amp Amyotte 2006) Although

QRAhas itsweaknesses anddrawbacksit is often theonlyapproach

to get grip on possible risks and to tame complexity as indicated by

Pasman Jung Prem Rogers and Yang (2009) QRA involves the

following main steps scenario selection frequency estimation

consequence assessment and risk quanti1047297cation (CCPS 2000

Khan Sadiq amp Husain 2002) We will discuss these steps hereafter

21 Scenario selection

Possible scenarios in the process industries may be different

modes of 1047297re explosion and toxic dispersion (Arunraj amp Maiti

2009 Markowski 2007) Neglecting any possible scenario can

lead to a risk underestimation and eventually affect the overall risk

values (Van Sciver 1990) Scenario selection is usually based on

expertrsquos opinion history of past accidents and safety reviews For

our study after selecting the credible scenarios based on a combi-

nation of mentioned methods Event Tree Analysis (ETA) wasapplied for scenario development The quantitative feature of ETA

was then used for incident frequency calculation

22 Frequency estimation

Once the 1047297nal outcomes and their sequences are predicted for

each scenario the scenario frequency should be determined For

this purpose the failure frequency of each initiating event and the

probability of every intermediate event such as immediate ignition

delayed ignition vapor cloud 1047297re etc are used The frequency

values which can be found in literature are mostly based on

historical data of previous incidents (Beerens Post amp Uijt de Haag

2006) The frequency of a scenario speci1047297es the number of occur-

rences of that scenario in a speci1047297ed timeframe (which is usuallyone year) The probability of an intermediate event is a dimension-

less value between0 and 1 which indicates the possibility of

occurrence of that event in a speci1047297c period of time For this paper

the failure frequencies for the initiating events and the intermediate

values were obtained from the Purple Book (CPR 1999) Application

of failure frequenciesin thePurple Book andthe BEVI Manualduring

the past years wasindustry standard Also softwaresuch as ARIPAR-

GIS (Spadoni Egidi amp Contini 2000) or DomPrevPlanning (Reniers

amp Dullaert 2007) use these values The following equation shows

how the initiating event frequency intermediate probabilities and

1047297nal outcome frequency are related to each other (CCPS 2000)

f i frac14 F I P oiP oci (1)

where f i is the frequency of the 1047297nal outcome of scenario i arising

from incident I (1yr)F I is the frequency of incident I which causes

different outcomes (1yr)P oi is probability of occurrence for an

intermediate outcome of incident I which causes a number of i 1047297nal

outcomes and P oci is the probability of occurrence for a 1047297nal

outcome which is a subset of outcomes arising from incident I

23 Consequences assessment

During the consequence assessment the effects and magnitude

of the 1047297nal outcome of each scenario should be estimated The

consequences of potential scenarios could be classi1047297ed as follows

(Lees 1996) 1047297re thermal radiation explosion overpressure and

fragment projection and toxic dispersion and exposure

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903896

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 39

Besides serious physical human harm and eventually casualties

property damage and environmental effects are other conse-

quences among the overall potential loss (Pintaric 2007)

The measure of damage in QRA is fatality since other types of loss

are more complex to assess in comparison to human casualties

(Pula et al 2006)

A chain of rigorous calculations is needed to estimate the

intensity of the physical harm posed by the mentioned scenario

outcomes in a chemical cluster Software packages for the proce-

dure of consequence assessment are used to estimate the numer-

ical value of thermal heat load explosion overpressure and toxic

concentration at various spots around the release point For this

study the well-known PHAST (Process Hazard Analysis Safety Tool)

software was used for consequence assessment Once the appro-

priate failure frequencies are given this particular software con-

taining 1047297ne discharge dispersion evaporation and rainout models

is a powerful tool for the prediction of effect zones (Pintaric 2007)

In addition to the results of consequence modeling fatality probit

equations were used to quantify the expected percentage of

lethality for the exposed population

24 Risk quanti 1047297cation

The frequency and severity (rate of fatality) of every speci1047297c

scenario are combined to obtain a measure of the corresponding

risk Risk quanti1047297cation results are presented in two conventional

categories which are known as lsquoindividual riskrsquo and lsquogroup riskrsquo

(also called lsquosocietal riskrsquo)

The frequency at which a particular individual being fatally

harmed when standing at a certain distance from a potential

hazard is known as ldquoindividual riskrdquo (Gooijer Cornil amp Lenoble

2012) The overall individual risk at any location ( x y) inside or

outside the industrial plant is the summation of all individual risks

at that speci1047297c point The individual risk at any location ( x y) is

calculated by the following equation (CCPS 200 0)

IR x y frac14Xn

ifrac141

IR x yi (2)

In other words the risk of any identi1047297ed scenario will be indi-

vidually calculated at a speci1047297c location ( x y) and subsequently all

risks are summed to estimate the overall risk at that speci1047297c point

To calculate the individual risk at location ( x y) arising from

scenario i the following equation is applied (CCPS 200 0)

IR x yi frac14 f iP f i (3)

where f i is the frequency of the 1047297nal outcome of scenario i arising

from an incident (1yr) andP f iis the fatality probability of the 1047297nal

outcome of scenario i at the geographical location ( x y) Individualrisk contours represent the 1047297nal results of this step The individual

risk value in every position when compared to universally or

regionally accepted values is one of the criteria in risk-based

decision making for land use planning

Societal risk provides a risk evaluation for a group of people

located in the vicinity of the accident location In other words the

number of people affected by all 1047297nal outcomes is estimated

(Renjith amp Madhu 2010) Similar to the individual risk the societal

risk is a function of frequency of occurrence and rate of fatality

Another important and determinant factor for calculating the

societal risk is the population density around the incident location

Thesocietal riskis presented in form of FeN (FrequencyeNumber

of fatalities) curves where the cumulative frequency of 1047297nal

outcomes is plotted against the number of fatalities arising from anoutcome in a logarithmic scale To calculate the number of fatalities

of each 1047297nal outcome the following equation is used (CCPS 2000)

N i frac14X

x y

P x y p f i (4)

whereN irepresents the number of fatalities of the 1047297nal outcome of

scenario i P x yis the number of individuals at the geographical

location ( x y) and p f i is the probability that the 1047297nal outcome of

scenario i causes death at the geographical location ( x y)

The results obtained above are used in Equation (5) to calculate

the ultimate data required for plotting the FeN curve This equation

which is known as the cumulative frequency equation is expressed

as below (CCPS 2000)

F N frac14X

i

F i for all final outcome case I which N i gt N (5)

where F N is the cumulative frequency for all 1047297nal outcomes which

result in fatalities of more than N personsF i is the frequency of the

Fig 1 Layout of the study area (case-study) showing the two neighboring plants AKPC and BSPC

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 897

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 49

1047297nal outcome of scenario i and N i is the number of fatalities for the

1047297nal outcome of scenario i Finally having the quantities of F N and

N the cumulative frequency of 1047297nal outcomes is plotted against the

number of fatalities The obtained FeN curved will be compared

against the intended criteria to evaluate the acceptability of the

potential societal risk

3 Case study

ldquoAKPCrdquo and ldquoBSPCrdquo are two neighboring complexes located in

the so-called PETZONE In this industrial region 15 petrochemical

plants are situated in 5 distinct sites within an area of approxi-mately 20 km2 and therefore it is known to be one of the biggest

energy zones in the world The PETZONE lies in the northern coast

of the Persian Gulf and expands to the southwestern city of Mah-

shahr in southern Iran

Before and during the construction of many plants in the PET-

ZONE until present only economic and operational factors have

been taken into account leaving safety aspects to be largely

neglected in the layout design of those plants This was also the

case for our case study of the two plants AKPC and BSPC The ole1047297n

unit of AKPC produces ethylene and propylene and these

substances are later used in the polymerization unit to produce

high density polyethylene low density polyethylene and poly-

propylene Regarding BSPC its main products are naphtha p-

xylene and acetic acid

These two plants present an interesting case study due to their

high number of storage tanks and huge inventory of chemicals As

one of these plants provides a portion of the other one rsquos feed

constructing them close together seemed a wise choice from an

economic and operational point of view However the compliance

of risk criteria had not been an item of consideration After occur-

rence of some minor process accidents with limited consequences

in these plants the risk of accident escalation from one plant to

another was highlighted more than ever Therefore the idea of

constructing AKPC and BSPC in proximity of each other which

seemed a defendable choice once has been challenged In the

present study the two adjacent chemical complexes AKPC and

BSPC are both subjected to one QRA (treating both plants as one

plant) allowing us to analyze the effects of the construction of a

petrochemical plant adjacent to an existing one on the overall riskThe purpose was to verify whether it was indeed a wise choice to

construct these two plants this close to each other (for production

purposes) or not (for safety reasons) Fig 1 represents the layout of

the two petrochemical plants

Based on a safety review and on a safety screening carried out

as a preliminary step the major hazard sources of the study area

(as displayed in Fig 1) appear to be the storage tanks of these two

plants Although there are obviously other sources of hazards

within this area their contribution in the overall risk could be

regarded negligible (compared with the present storage tanks)

AKPC houses 5 major operational storage tanks containing

ethylene and propylene while BSPC contains 10 major atmo-spheric storage tanks containing naphtha p-xylene and acetic

acid Tables 1 and 2 provide a list of the chemical substance

inventories in AKPC and BSPC Although there were some other

storage tanks or process equipment no one passed the screening

step for scenario selection due to their low inventory of hazardous

materials and due to the lesser inherent hazardousness of the

materials

The atmosphericdata used in this study were established by the

Iranian Meteorological Organization (IMO) for the Mahshahr port

from 1988 to 2006 (IRIMO 2006) The result of the meteorological

data analysis indicates that two prevailing weather conditions (hot

season and cold season) can be intended for the QRA study during

daytime and nighttime to cover almost all of the probable condi-

tions Table 3 summarizes the average meteorological data whichare used for the consequence analysis

31 Accident scenarios

As stated previously the storage tanks have been identi1047297ed to

lead to the most hazardous scenarios in the case under study

Table 1

List of AKPC storages in this study

TK-item Compound Type Temperature

(C)

Pressure

(barg)

Volume

(m3)

Inventory

(tonne)

TK-101 Propylene Spherical 40 25 2200 1044

TK-102 Propylene Spherical 40 25 2200 1044

TK-103 Propylene Spherical 40 25 2200 1044

TK-104 Propylene Spherical 40 25 2200 1044

TK-105 Ethylene Cylindrical

104 Atm 14000 7968

Table 2

List of BSPC storages in this study

TK-item Compound Type Temperature

(C)

Pressure

(barg)

Volume

(m3)

Inventory

(tonne)

TK-201 Naphtha Cy lindrical 30 001 331188 24839

TK-202 Naphtha Cy lindrical 30 001 331188 24839

TK-203 Naphtha Cy lindrical 30 001 331188 24839

TK-204 Naphtha Cylindrical 30 001 16272 12204

TK-205 Naphtha Cylindrical 30 Atm 16272 12204TK-206 P -Xylene Cy lindrical 30 At m 170856 14693

TK-207 P -Xylene Cy lindrical 30 At m 170856 14693

TK-208 P-Xylene Cylindrical 40 0003 1500 1267

TK-209 Acetic acid Cylindrical 45 0014 1500 1554

TK-210 Acetic acid Cylindrical 45 0014 1500 1554

Table 3

Prevailing weather conditions

Category Atmospheric stability Average wind

speed (ms)

Average ambient

temp (C)

Average relative

humidity ()

Hot season (daytime) Neutral (D class) 5 40 50

Hot season (nighttime) Stable (F class) 15 25 60

Cold season (daytime) Neutral (D class) 5 20 80

Cold season (nighttime) Stable (F class) 15 10 90

Table 4

Accident scenarios and the relevant failure frequencies

Scenario

no

TK -i tem n o C ontai nmen t Scena ri o typ e Frequency

(1yr)

1e4 TK-101102103104 Propylene Leakage 1 105

5e8 TK-101102103104 Propylene Rupture 5 107

9 TK-105 Ethylene Leakage 1 105

10e14 TK-201202203

204205

Naphtha Leakage 1 105

15e17 TK-206207208 P-Xylene Leakage 1 105

18e19 TK-209 210 Acetic acid Leakage 1 105

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903898

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 59

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 69

Fig 4 A B individual risk contours before (for AKPC) and after (for cluster composed of AKPC and BSPC) constructing the new plant

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 79

the leakage case and in the rupture case Moreover based on the

plant layout review and due to the low level of equipment

congestion in this particular area under study the probability of

a 1047298ash 1047297re is higher than that of a Vapor Cloud Explosion (VCE)

since higher congestion is known to increase the probability of

a VCE rather than that of a 1047298ash 1047297re

4 Result and discussion

In order to evaluate the risk-based effects of constructing BSPC

adjacent to AKPC calculations were made for both individual risk

and societal risk before and after the new plant came under oper-

ation adjacent to the existing one The quanti1047297ed risk can then be

Fig 5 A B societal risk (Fe

N curve) before (A) and after (B) constructing the new plant

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 901

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 89

compared against the criteria to assess the acceptability of the risk

levels The likelihood of domino effects needs to be assessed as

well

41 Calculation of individual risk

Using the results of frequency estimation and a detailed

consequence assessment for all the possible scenarios the indi-

vidual risk contours were obtained and are shown on a satellite

view of the study area Key outputs of the consequence modeling

are presented in Table 5 to give more insight in the two selected

scenarios Fig 4A and B demonstrate the individual risk levels

before (A) and after (B) the second plant came operational The

numerical value of each risk contour represents the frequency at

which a particular individual is fatally harmed when standing

within the contour boundary

The results of individual risk calculations reveal that inside the

chemical cluster composed of both plants (AKPC and BSPC) as well

as within the single plant AKPC the individual risk values do not

exceed the tolerable magnitude of 105yr before and after the new

plant came under operation Fig 3A and B also demonstrate that for

public areas (such as the main roads around AKPC and BSPC) the

risk level does not exceed the magnitude of 106yr except fora short interval of 100 m at the southern main road of AKPC

Nevertheless this issue was already present even before the

construction of BSPC In general we can state that in terms of

individual risk constructing the new plant adjacent to the existing

one did not affect the acceptability of risk levels

42 Calculation of societal risk

Although the previous section indicates that individual risk

levels do not exceed the tolerable levels the impact of any potential

accident on operators and public should also be investigated

Fig 4A and B provide the societal risk calculation results before and

after the new plant came into operation respectively As it can be

seen in Fig 5A the societal risk posed by AKPC individually falls inthe ALARP (As Low As Reasonably Practicable) region as de1047297ned by

HSE (UK)Fig 5Bshows that after constructing the new plant the

societal risk curve gets closer to the maximum risk criteria but still

remains in the ALARP region Therefore constructing the new plant

BSPC adjacent to the existing one AKPC seems to have only limited

impact upon the change of tolerability of the societal risk in our

case study

43 Domino effects assessment

The distance from the storage tanks located in AKPC to the

closest operational storage tank in BSPC is about 800 m Based on

the values reported by Cozzani Gubinelli and Salzano (2006) this

distance suggests that potential cascading accidents between theneighboring plants caused by thermal radiation are highly

improbable Therefore the only factor that can trigger a secondary

accident in the other neighboring plant is the overpressure caused

by an uncon1047297ned Vapor Cloud Explosion Considering the volatility

of the materials involved the operational conditions of storage the

average ambient temperature and the prevailing wind direction

(west) we may conclude that only materials stored in AKPCrsquos

storages are capable of forming an effective vapor cloud

There are different values reported for damage threshold due to

overpressure in literature but the overpressure of 1bar would

surely destroy any target equipment (Salzano amp Cozzani 2005)

This threshold has been chosen to evaluate the possibility of

domino accidents on the storage tanks in BSPC Table 6 provides

the magnitude of overpressure caused by different AKPC scenarios

(1e9) at the location of BSPCrsquos storage tanks These values are

based on performing a detailed consequence modeling

As it can be seen in Table 6 the closest storage tanks to AKPC

(TK-201 and TK-202) are possibly damaged due to overpressurecaused by a propylene explosion after a leakage or tank rupture

eventually triggering secondary accidents In other words the

accidental release of propylene in AKPC either continuous

(scenarios 1e4) or instantaneous (scenarios 5e8) has the capa-

bility of causing cascading accidents in BSPC This suggests that if

decision-makers insist on not changing the new plantrsquos layout (eg

due to large economic and operational bene1047297ts) countermeasures

should be taken into consideration in order to minimize the

heightened risk of domino accidents

Both choices for the Purple Book and the UK risk tolerance

criteria are individual choices for this study and they should be

seen and respected as such We are aware that failure frequencies

are subject to constant optimization and that the Purple Book has

its limitations (Pasman 2011) Nonetheless for this study of twoplants situated in Iran using the frequencies reported in this well-

known and much-used work was a justi1047297able choice

5 Conclusions

The risk-based effects of constructing a new chemical plant

adjacent to an existing one forming a chemical cluster is analyzed

and discussed in this paper For this purpose a QRA was carried out

to assess different parameters of risk Individual and societal risks

were evaluated before and after the new plant came under oper-

ation The results indicate that the new plant did not have a major

impact on the risk levels since the risk levels stay within the

tolerability limits Although no major impact was observed there

was a signi1047297cant increase in the societal risk Moreover a dominoeffect analysis indicates that some storage tanks in the new plant

entail a high potential of being affected by events originating in the

existing plant

This research and this case study illustrate that in order to make

an objective plant lay out decision for a chemical cluster with

regard to operational risks not only conventional risk assessments

should be carried out but a domino effects analysis should be

performed as well since otherwise some risks may be under-

estimated and overlooked

It is undoubtedly true that themore accurateinformationis used

(eg concerning probabilities) the more accurate risk calculations

can be made However one should always remember that basically

calculated risks arerelativerisks andthat absolute valuesare prone

to variability and uncertainty As referencedby Pasman et al(2009)

Table 6

Target equipment affected by domino accident (due to overpressure)

scenario

Magnitude of overpressure caused by different AKPC scenarios at

target equipment (bar)

T K - 2 0 1

T K - 2 0 2

T K - 2 0 3

T K - 2 0 4

T K - 2 0 5

T K - 2 0 6

T K - 2 0 7

T K - 2 0 8

T K - 2 0 9

T K - 2 1 0

1-4 02 01 01 008 008 017 015 015

5-8 1 04 02 02 014 014 025 023 022

9 005 005 005 005 005 005 005 005 005 005

Affected by Dominoaccident

Not Affected by Dominoaccident

1 1

1

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903902

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 99

EU-studies have shown that a factor of 10e100 both ways in risk

results is not uncommon Hence future research might include risk

distributions and con1047297dence limits allowing risk analysts to even

better understand the uncertainties accompanying the results

References

Abbasi S A Krishnakumari P amp Khan F I (1998) Hot topics Global warming acidrain ozone hole hazardous waste industrial disasters disinfection New DelhiOxford University Press

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2010) A newmethod for assessing domino effect in chemical process industry Journal of Hazardous Materials 182(1e3) 416e426

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2011) Dominoeffect in process-industry accidents e an inventory of past events and identi-1047297cation of some patterns Journal of Loss Prevention in the Process Industries

24(5) 575e593Arunraj N S amp Maiti J (2009) A methodology for overall consequence modeling

in chemical industry Journal of Hazardous Materials 169 556e574Atherton J amp Gil F (2008) Incidents that de 1047297ne process safety Hoboken New

Jersey USA WileyBeerens H I Post J G amp Uijt de Haag P A M (2006) The use of generic failure

frequencies in QRA the quality and use of failure frequencies and how to bringthem up-to-date Journal of Hazardous Materials 130 265e270

Bubbico R Maschio G Mazzarotta B Milazzo M amp Parisi E (2006) Riskmanagement of road and rail transport of hazardous materials in Sicily Journalof Loss Prevention in the Process Industries 19(1) 32e38

Center for Chemical Process Safety (CCPS) (2000) Guidelines for chemical processquantitative risk analysis (2nd ed) New York American Institute of ChemicalEngineering

CozzaniV GubinelliG amp SalzanoE (2006)Escalationthresholds inthe assessmentof domino accidental events Journal of Hazardous Materials 129(1e3)1e21

CPR 18E (1999) lsquo Purple bookrsquo guidelines for quantitative risk assessment Part oneEstablishments The Hague RVIM

Dechy N Bourdeaux T Ayrault N Kordek M A amp Le Coze J C (2004) Firstlessons of the Toulouse ammonium nitrate disaster 21st September 2001 AZFplant France Journal of Hazardous Materials 111(1e3) 131e138

Gharabagh M J Asilian H Mortasavi S B Mogaddam A Z Hajizadeh E ampKhavanin A (2009) Comprehensive risk assessment and management of petrochemical feed and product transportation pipelines Journal of LossPrevention in the Process Industries 22(4) 533e539

Gooijer L Cornil N amp Lenoble C L (2012) An international comparison of fourquantitative risk assessment approaches e a benchmark study based on a 1047297cti-tious LPG plant Process Safety and Environmental Protection 90(2) 101e107

IRIMO (2006) Climatologically normal for the period 1987 e

2006 Mahshahr TehranIR of Iran Meteorological Organization Data Processing Center

Jo Y amp Crowl D (2008) Individual risk analysis of high-pressure natural gaspipelines Journal of Loss Prevention in the Process Industries 21 (6) 589e595

Kalantarnia M Khan F amp Hawboldt K (2010) Modelling of BP Texas City re1047297neryaccident using dynamic risk assessment approach Process Safety and Environ-mental Protection 88(3) 191e199 (Institution of Chemical Engineers)

Khan F I amp Abbasi S A (1998) Techniques and methodologies for risk analysis inchemical process industries Journal of Loss Prevention in the Process Industries11 261e277

Khan F I amp Abbasi S A (1999a) Major accidents in process industries and ananalysis of causes and consequences Journal of Loss Prevention in the ProcessIndustries 12 361e378

Khan F I amp Abbasi S A (1999b) Assessment of risks posed by chemical indus-triesdapplication of a new computer automated tool MAXCRED-III Journal of Loss Prevention in the Process Industries 12(6) 455e469

Khan F I Sadiq R amp Husain T (2002) Risk-based process safety assessment andcontrol measures design for offshore process facilities Journal of HazardousMaterials 94(1) 1e36

Kletz T A (1999b) The origins and history of loss prevention Transaction IChemE77 109e116

Kletz T A (1999a) What went wrong Case histories of process plant disasters andhow they could have been avoided Houston USA Gulf Professional Publishing

Kletz T A (2003) Still going wrong Case histories of process plant disasters and howthey could have been avoided Houston USA Gulf Professional Publishing

Lees F P (1996) Loss prevention in the process industries (2nd ed) Oxford LondonButterworth Heinemann

Markowski A S (2007) exLOPA for explosion risks assessment Journal of Hazardous Materials 142(3) 669e676

Milazzo M F Lisi R Maschio G Antonioni G Bonvicini S amp Spadoni G (2002)HazMat transport through Messina town from risk analysis suggestions forimproving territorial safety Journal of Loss Prevention in the Process Industries15(5) 347e356

Papazoglou I A Nivolianitou Z Aneziris O Christou M D amp Bonanos G (1999)Risk-informed selection of a highway trajectory in the neighborhood of an oil-re1047297nery Journal of Hazardous Materials 67 (2) 111e144

Pasman H J (2011) History of Dutch process equipment failure frequencies and thepurple book Journal of Loss Prevention in the Process Industries 24 208e213

Pasman H J Jung S Prem K Rogers W J amp Yang X (2009) Is risk analysisa useful tool for improving process safety Journal of Loss Prevention in theProcess Industries 22(6) 769e777

Pintaric Z (2007) Assessment of the consequences of accident scenariosinvolving dangerous substances Process Safety and Environmental Protection85(1) 23e38

Pula R Khan F I Veitch B amp Amyotte P (2006) A grid based approach for 1047297reand explosion consequence analysis Process Safety and Environmental Protec-tion 84(2) 79e91

Reniers G L L (2010) An external domino effects investment approach to improvecross-plant safety within chemical clusters Journal of Hazardous Materials 177 167e174

Reniers G L L amp Dullaert W (2007) DomPrevPlanning user-friendly software forplanning domino effects prevention Safety Science 45(10) 1060e1081

Renjith V R amp Madhu G (2010) Individual and societal risk analysis and mappingof human vulnerability to chemical accidents in the vicinity of an industrialarea International Journal of Applied Engineering Research Dindigul 1 135e148

Rigas F amp Sklavounos S (2002) Risk and consequence analyses of hazardouschemicals in marshalling yards and warehouses at IkonioPiraeus harbourGreece Journal of Loss Prevention in the Process Industries 15(6) 531e544

Salzano Eamp Cozzani V (2005)The analysis of dominoaccidents triggered byvaporcloud explosions Reliability Engineering amp System Safety 90(2e3) 271e284

Spadoni G Egidi D amp Contini S (2000) Through ARIPAR-GIS the quanti 1047297ed arearisk analysis supports land-use planning activities Journal of Hazardous Mate-rials 71(1e3) 423e437

Van Sciver G R (1990) Quantitative risk analysis in the chemical process indus-tries Reliability Engineering amp System Safety 29 55e68

Viacutelchez J A Espejo V amp Casal J (2011) Generic event trees and probabilities forthe release of different types of hazardous materials Journal of Loss Preventionin the Process Industries 24(3) 281e287

Wells G (1997) Major hazards and their management Rugby UK Institution of Chemical Engineers

Yet-Pole IChi-MinS amp Ching-Hong C (2009)Applicationsof 3D QRAtechnique tothe 1047297reexplosion simulation and hazard mitigation within a naphtha-crackingplant Journal of Loss Prevention in the Process Industries 22 (4) 506e515

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 903

Page 2: 1-s2.0-S0950423012001702-main

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 29

residential areas in order to compare the result with safety

criteria and eventually recommendations were given to make

modi1047297cations in the plants since the criteria were not met Jo and

Crowl (2008) speci1047297ed the minimum safety distances between

a high pressure gas pipeline and residential areas Risk and conse-

quence analyses of toxic chemicals in certain warehouses to the

nearby villages were analyzed by Rigas and Sklavounos (2002) to

investigate the extent of compliance to exposure threshold limits

The study by Papazoglou Nivolianitou Aneziris Christou and

Bonanos (1999) demonstrated that a high level of risk is exposed

to passengers crossing the highway adjacent to a speci1047297c re1047297nery

and thereforea new trajectory withthe safety distancecalculated by

QRAshould be constructed These area limited numberof examples

showing that land use planning is vital forlocating process facilities

near to public areas and a QRA could be an effective technique for

this purpose In this method in addition to identifying the hazards

also their associated consequences and the frequencies of occur-

rences are quantitatively estimated (Kletz 1999b) Furthermore

parameters such as lsquoindividual riskrsquo and lsquogroup riskrsquo (the latter also

being called lsquosocietal riskrsquo) are parameters which support the deci-

sion making process for land use planning Several other relevant

studies can be found elsewhere (Bubbico Maschio Mazzarotta

Milazzo amp Parisi 2006 Gharabagh et al 2009 Milazzo et al2002 Yet-Pole Chi-Min amp Ching-Hong 2009)

The proliferation of chain accidents suggests that the probability

of domino effects should also be considered as an important

parameterin land useplanning forchemical clusters A recent study

on past domino events indicated that the number of cascading

accidents in the process industries has increased globally in recent

decades (Abdolhamidzadeh Abbasi Rashtchian amp Abbasi 2011) In

the same study there are some accidents mentioned in which an

initiating event in a single plant has led to a catastrophe within

a clusterof chemical plants A recentexample couldbe Shazand Iran

in which 30 peoplewere killedand 38 others were seriously injured

in 2008due to an initial explosion that ledto other majorexplosions

and 1047297res in the neighboring plants The dominant accident

that caused the majority of fatalities was the major explosions thatwere triggered by the initial blast (Abdolhamidzadeh et al 2011)

As already mentioned in land use planning possible scenarios

are reviewed from different points of view one being the safety

perspective In many cases different chemical plants in a chemical

industrial area do not come into operation at the same point in

time Forexample it is common practice to build plants to consume

the product(s) of an existing plant It is obvious that the construc-

tion of new plant(s) adjacent to existing one(s) will affect the

overall risk of the industrial area regardless of the fact whether risk

assessments have been performed for individual plants

In the remainder of this article QRA has been applied to eval-

uate the risk-based effects of constructing a new chemical plant

(which is called BSPC in this article) adjacent to an existing one

(called AKPC in this paper) These two plants forming our case-study are located in one of the largest energy zones in the world

in southern Iran (called lsquoPETZONErsquo) Parameters such as the indi-

vidualrisk and the societal risk have been assessedbefore and after

the new plant came into operation Furthermore the possibility

and the likelihood of domino events in the two neighboring plants

were investigated We mainly focus on safety aspects in our article

as the feasibility of the planning from both an economic and an

operational perspective has been carried out in an earlier stage of

the land-use planning decision process

2 Quantitative Risk Assessment (QRA)

QRA along with other techniques have been used in one way or

another in the process industries for loss prevention from 1970

onwards (Abbasi Krishnakumari amp Khan 1998 Khan amp Abbasi

1998) QRA is a method that provides quanti1047297ed estimation for

the risk posed by a group of hazards (Kletz 1999) Hence this

technique enables risk mitigation methods to be evaluated in order

to bring the risk to tolerable levels without resorting to too costly

protective systems (Pula Khan Veitch amp Amyotte 2006) Although

QRAhas itsweaknesses anddrawbacksit is often theonlyapproach

to get grip on possible risks and to tame complexity as indicated by

Pasman Jung Prem Rogers and Yang (2009) QRA involves the

following main steps scenario selection frequency estimation

consequence assessment and risk quanti1047297cation (CCPS 2000

Khan Sadiq amp Husain 2002) We will discuss these steps hereafter

21 Scenario selection

Possible scenarios in the process industries may be different

modes of 1047297re explosion and toxic dispersion (Arunraj amp Maiti

2009 Markowski 2007) Neglecting any possible scenario can

lead to a risk underestimation and eventually affect the overall risk

values (Van Sciver 1990) Scenario selection is usually based on

expertrsquos opinion history of past accidents and safety reviews For

our study after selecting the credible scenarios based on a combi-

nation of mentioned methods Event Tree Analysis (ETA) wasapplied for scenario development The quantitative feature of ETA

was then used for incident frequency calculation

22 Frequency estimation

Once the 1047297nal outcomes and their sequences are predicted for

each scenario the scenario frequency should be determined For

this purpose the failure frequency of each initiating event and the

probability of every intermediate event such as immediate ignition

delayed ignition vapor cloud 1047297re etc are used The frequency

values which can be found in literature are mostly based on

historical data of previous incidents (Beerens Post amp Uijt de Haag

2006) The frequency of a scenario speci1047297es the number of occur-

rences of that scenario in a speci1047297ed timeframe (which is usuallyone year) The probability of an intermediate event is a dimension-

less value between0 and 1 which indicates the possibility of

occurrence of that event in a speci1047297c period of time For this paper

the failure frequencies for the initiating events and the intermediate

values were obtained from the Purple Book (CPR 1999) Application

of failure frequenciesin thePurple Book andthe BEVI Manualduring

the past years wasindustry standard Also softwaresuch as ARIPAR-

GIS (Spadoni Egidi amp Contini 2000) or DomPrevPlanning (Reniers

amp Dullaert 2007) use these values The following equation shows

how the initiating event frequency intermediate probabilities and

1047297nal outcome frequency are related to each other (CCPS 2000)

f i frac14 F I P oiP oci (1)

where f i is the frequency of the 1047297nal outcome of scenario i arising

from incident I (1yr)F I is the frequency of incident I which causes

different outcomes (1yr)P oi is probability of occurrence for an

intermediate outcome of incident I which causes a number of i 1047297nal

outcomes and P oci is the probability of occurrence for a 1047297nal

outcome which is a subset of outcomes arising from incident I

23 Consequences assessment

During the consequence assessment the effects and magnitude

of the 1047297nal outcome of each scenario should be estimated The

consequences of potential scenarios could be classi1047297ed as follows

(Lees 1996) 1047297re thermal radiation explosion overpressure and

fragment projection and toxic dispersion and exposure

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903896

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 39

Besides serious physical human harm and eventually casualties

property damage and environmental effects are other conse-

quences among the overall potential loss (Pintaric 2007)

The measure of damage in QRA is fatality since other types of loss

are more complex to assess in comparison to human casualties

(Pula et al 2006)

A chain of rigorous calculations is needed to estimate the

intensity of the physical harm posed by the mentioned scenario

outcomes in a chemical cluster Software packages for the proce-

dure of consequence assessment are used to estimate the numer-

ical value of thermal heat load explosion overpressure and toxic

concentration at various spots around the release point For this

study the well-known PHAST (Process Hazard Analysis Safety Tool)

software was used for consequence assessment Once the appro-

priate failure frequencies are given this particular software con-

taining 1047297ne discharge dispersion evaporation and rainout models

is a powerful tool for the prediction of effect zones (Pintaric 2007)

In addition to the results of consequence modeling fatality probit

equations were used to quantify the expected percentage of

lethality for the exposed population

24 Risk quanti 1047297cation

The frequency and severity (rate of fatality) of every speci1047297c

scenario are combined to obtain a measure of the corresponding

risk Risk quanti1047297cation results are presented in two conventional

categories which are known as lsquoindividual riskrsquo and lsquogroup riskrsquo

(also called lsquosocietal riskrsquo)

The frequency at which a particular individual being fatally

harmed when standing at a certain distance from a potential

hazard is known as ldquoindividual riskrdquo (Gooijer Cornil amp Lenoble

2012) The overall individual risk at any location ( x y) inside or

outside the industrial plant is the summation of all individual risks

at that speci1047297c point The individual risk at any location ( x y) is

calculated by the following equation (CCPS 200 0)

IR x y frac14Xn

ifrac141

IR x yi (2)

In other words the risk of any identi1047297ed scenario will be indi-

vidually calculated at a speci1047297c location ( x y) and subsequently all

risks are summed to estimate the overall risk at that speci1047297c point

To calculate the individual risk at location ( x y) arising from

scenario i the following equation is applied (CCPS 200 0)

IR x yi frac14 f iP f i (3)

where f i is the frequency of the 1047297nal outcome of scenario i arising

from an incident (1yr) andP f iis the fatality probability of the 1047297nal

outcome of scenario i at the geographical location ( x y) Individualrisk contours represent the 1047297nal results of this step The individual

risk value in every position when compared to universally or

regionally accepted values is one of the criteria in risk-based

decision making for land use planning

Societal risk provides a risk evaluation for a group of people

located in the vicinity of the accident location In other words the

number of people affected by all 1047297nal outcomes is estimated

(Renjith amp Madhu 2010) Similar to the individual risk the societal

risk is a function of frequency of occurrence and rate of fatality

Another important and determinant factor for calculating the

societal risk is the population density around the incident location

Thesocietal riskis presented in form of FeN (FrequencyeNumber

of fatalities) curves where the cumulative frequency of 1047297nal

outcomes is plotted against the number of fatalities arising from anoutcome in a logarithmic scale To calculate the number of fatalities

of each 1047297nal outcome the following equation is used (CCPS 2000)

N i frac14X

x y

P x y p f i (4)

whereN irepresents the number of fatalities of the 1047297nal outcome of

scenario i P x yis the number of individuals at the geographical

location ( x y) and p f i is the probability that the 1047297nal outcome of

scenario i causes death at the geographical location ( x y)

The results obtained above are used in Equation (5) to calculate

the ultimate data required for plotting the FeN curve This equation

which is known as the cumulative frequency equation is expressed

as below (CCPS 2000)

F N frac14X

i

F i for all final outcome case I which N i gt N (5)

where F N is the cumulative frequency for all 1047297nal outcomes which

result in fatalities of more than N personsF i is the frequency of the

Fig 1 Layout of the study area (case-study) showing the two neighboring plants AKPC and BSPC

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 897

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 49

1047297nal outcome of scenario i and N i is the number of fatalities for the

1047297nal outcome of scenario i Finally having the quantities of F N and

N the cumulative frequency of 1047297nal outcomes is plotted against the

number of fatalities The obtained FeN curved will be compared

against the intended criteria to evaluate the acceptability of the

potential societal risk

3 Case study

ldquoAKPCrdquo and ldquoBSPCrdquo are two neighboring complexes located in

the so-called PETZONE In this industrial region 15 petrochemical

plants are situated in 5 distinct sites within an area of approxi-mately 20 km2 and therefore it is known to be one of the biggest

energy zones in the world The PETZONE lies in the northern coast

of the Persian Gulf and expands to the southwestern city of Mah-

shahr in southern Iran

Before and during the construction of many plants in the PET-

ZONE until present only economic and operational factors have

been taken into account leaving safety aspects to be largely

neglected in the layout design of those plants This was also the

case for our case study of the two plants AKPC and BSPC The ole1047297n

unit of AKPC produces ethylene and propylene and these

substances are later used in the polymerization unit to produce

high density polyethylene low density polyethylene and poly-

propylene Regarding BSPC its main products are naphtha p-

xylene and acetic acid

These two plants present an interesting case study due to their

high number of storage tanks and huge inventory of chemicals As

one of these plants provides a portion of the other one rsquos feed

constructing them close together seemed a wise choice from an

economic and operational point of view However the compliance

of risk criteria had not been an item of consideration After occur-

rence of some minor process accidents with limited consequences

in these plants the risk of accident escalation from one plant to

another was highlighted more than ever Therefore the idea of

constructing AKPC and BSPC in proximity of each other which

seemed a defendable choice once has been challenged In the

present study the two adjacent chemical complexes AKPC and

BSPC are both subjected to one QRA (treating both plants as one

plant) allowing us to analyze the effects of the construction of a

petrochemical plant adjacent to an existing one on the overall riskThe purpose was to verify whether it was indeed a wise choice to

construct these two plants this close to each other (for production

purposes) or not (for safety reasons) Fig 1 represents the layout of

the two petrochemical plants

Based on a safety review and on a safety screening carried out

as a preliminary step the major hazard sources of the study area

(as displayed in Fig 1) appear to be the storage tanks of these two

plants Although there are obviously other sources of hazards

within this area their contribution in the overall risk could be

regarded negligible (compared with the present storage tanks)

AKPC houses 5 major operational storage tanks containing

ethylene and propylene while BSPC contains 10 major atmo-spheric storage tanks containing naphtha p-xylene and acetic

acid Tables 1 and 2 provide a list of the chemical substance

inventories in AKPC and BSPC Although there were some other

storage tanks or process equipment no one passed the screening

step for scenario selection due to their low inventory of hazardous

materials and due to the lesser inherent hazardousness of the

materials

The atmosphericdata used in this study were established by the

Iranian Meteorological Organization (IMO) for the Mahshahr port

from 1988 to 2006 (IRIMO 2006) The result of the meteorological

data analysis indicates that two prevailing weather conditions (hot

season and cold season) can be intended for the QRA study during

daytime and nighttime to cover almost all of the probable condi-

tions Table 3 summarizes the average meteorological data whichare used for the consequence analysis

31 Accident scenarios

As stated previously the storage tanks have been identi1047297ed to

lead to the most hazardous scenarios in the case under study

Table 1

List of AKPC storages in this study

TK-item Compound Type Temperature

(C)

Pressure

(barg)

Volume

(m3)

Inventory

(tonne)

TK-101 Propylene Spherical 40 25 2200 1044

TK-102 Propylene Spherical 40 25 2200 1044

TK-103 Propylene Spherical 40 25 2200 1044

TK-104 Propylene Spherical 40 25 2200 1044

TK-105 Ethylene Cylindrical

104 Atm 14000 7968

Table 2

List of BSPC storages in this study

TK-item Compound Type Temperature

(C)

Pressure

(barg)

Volume

(m3)

Inventory

(tonne)

TK-201 Naphtha Cy lindrical 30 001 331188 24839

TK-202 Naphtha Cy lindrical 30 001 331188 24839

TK-203 Naphtha Cy lindrical 30 001 331188 24839

TK-204 Naphtha Cylindrical 30 001 16272 12204

TK-205 Naphtha Cylindrical 30 Atm 16272 12204TK-206 P -Xylene Cy lindrical 30 At m 170856 14693

TK-207 P -Xylene Cy lindrical 30 At m 170856 14693

TK-208 P-Xylene Cylindrical 40 0003 1500 1267

TK-209 Acetic acid Cylindrical 45 0014 1500 1554

TK-210 Acetic acid Cylindrical 45 0014 1500 1554

Table 3

Prevailing weather conditions

Category Atmospheric stability Average wind

speed (ms)

Average ambient

temp (C)

Average relative

humidity ()

Hot season (daytime) Neutral (D class) 5 40 50

Hot season (nighttime) Stable (F class) 15 25 60

Cold season (daytime) Neutral (D class) 5 20 80

Cold season (nighttime) Stable (F class) 15 10 90

Table 4

Accident scenarios and the relevant failure frequencies

Scenario

no

TK -i tem n o C ontai nmen t Scena ri o typ e Frequency

(1yr)

1e4 TK-101102103104 Propylene Leakage 1 105

5e8 TK-101102103104 Propylene Rupture 5 107

9 TK-105 Ethylene Leakage 1 105

10e14 TK-201202203

204205

Naphtha Leakage 1 105

15e17 TK-206207208 P-Xylene Leakage 1 105

18e19 TK-209 210 Acetic acid Leakage 1 105

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903898

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 59

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 69

Fig 4 A B individual risk contours before (for AKPC) and after (for cluster composed of AKPC and BSPC) constructing the new plant

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 79

the leakage case and in the rupture case Moreover based on the

plant layout review and due to the low level of equipment

congestion in this particular area under study the probability of

a 1047298ash 1047297re is higher than that of a Vapor Cloud Explosion (VCE)

since higher congestion is known to increase the probability of

a VCE rather than that of a 1047298ash 1047297re

4 Result and discussion

In order to evaluate the risk-based effects of constructing BSPC

adjacent to AKPC calculations were made for both individual risk

and societal risk before and after the new plant came under oper-

ation adjacent to the existing one The quanti1047297ed risk can then be

Fig 5 A B societal risk (Fe

N curve) before (A) and after (B) constructing the new plant

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 901

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 89

compared against the criteria to assess the acceptability of the risk

levels The likelihood of domino effects needs to be assessed as

well

41 Calculation of individual risk

Using the results of frequency estimation and a detailed

consequence assessment for all the possible scenarios the indi-

vidual risk contours were obtained and are shown on a satellite

view of the study area Key outputs of the consequence modeling

are presented in Table 5 to give more insight in the two selected

scenarios Fig 4A and B demonstrate the individual risk levels

before (A) and after (B) the second plant came operational The

numerical value of each risk contour represents the frequency at

which a particular individual is fatally harmed when standing

within the contour boundary

The results of individual risk calculations reveal that inside the

chemical cluster composed of both plants (AKPC and BSPC) as well

as within the single plant AKPC the individual risk values do not

exceed the tolerable magnitude of 105yr before and after the new

plant came under operation Fig 3A and B also demonstrate that for

public areas (such as the main roads around AKPC and BSPC) the

risk level does not exceed the magnitude of 106yr except fora short interval of 100 m at the southern main road of AKPC

Nevertheless this issue was already present even before the

construction of BSPC In general we can state that in terms of

individual risk constructing the new plant adjacent to the existing

one did not affect the acceptability of risk levels

42 Calculation of societal risk

Although the previous section indicates that individual risk

levels do not exceed the tolerable levels the impact of any potential

accident on operators and public should also be investigated

Fig 4A and B provide the societal risk calculation results before and

after the new plant came into operation respectively As it can be

seen in Fig 5A the societal risk posed by AKPC individually falls inthe ALARP (As Low As Reasonably Practicable) region as de1047297ned by

HSE (UK)Fig 5Bshows that after constructing the new plant the

societal risk curve gets closer to the maximum risk criteria but still

remains in the ALARP region Therefore constructing the new plant

BSPC adjacent to the existing one AKPC seems to have only limited

impact upon the change of tolerability of the societal risk in our

case study

43 Domino effects assessment

The distance from the storage tanks located in AKPC to the

closest operational storage tank in BSPC is about 800 m Based on

the values reported by Cozzani Gubinelli and Salzano (2006) this

distance suggests that potential cascading accidents between theneighboring plants caused by thermal radiation are highly

improbable Therefore the only factor that can trigger a secondary

accident in the other neighboring plant is the overpressure caused

by an uncon1047297ned Vapor Cloud Explosion Considering the volatility

of the materials involved the operational conditions of storage the

average ambient temperature and the prevailing wind direction

(west) we may conclude that only materials stored in AKPCrsquos

storages are capable of forming an effective vapor cloud

There are different values reported for damage threshold due to

overpressure in literature but the overpressure of 1bar would

surely destroy any target equipment (Salzano amp Cozzani 2005)

This threshold has been chosen to evaluate the possibility of

domino accidents on the storage tanks in BSPC Table 6 provides

the magnitude of overpressure caused by different AKPC scenarios

(1e9) at the location of BSPCrsquos storage tanks These values are

based on performing a detailed consequence modeling

As it can be seen in Table 6 the closest storage tanks to AKPC

(TK-201 and TK-202) are possibly damaged due to overpressurecaused by a propylene explosion after a leakage or tank rupture

eventually triggering secondary accidents In other words the

accidental release of propylene in AKPC either continuous

(scenarios 1e4) or instantaneous (scenarios 5e8) has the capa-

bility of causing cascading accidents in BSPC This suggests that if

decision-makers insist on not changing the new plantrsquos layout (eg

due to large economic and operational bene1047297ts) countermeasures

should be taken into consideration in order to minimize the

heightened risk of domino accidents

Both choices for the Purple Book and the UK risk tolerance

criteria are individual choices for this study and they should be

seen and respected as such We are aware that failure frequencies

are subject to constant optimization and that the Purple Book has

its limitations (Pasman 2011) Nonetheless for this study of twoplants situated in Iran using the frequencies reported in this well-

known and much-used work was a justi1047297able choice

5 Conclusions

The risk-based effects of constructing a new chemical plant

adjacent to an existing one forming a chemical cluster is analyzed

and discussed in this paper For this purpose a QRA was carried out

to assess different parameters of risk Individual and societal risks

were evaluated before and after the new plant came under oper-

ation The results indicate that the new plant did not have a major

impact on the risk levels since the risk levels stay within the

tolerability limits Although no major impact was observed there

was a signi1047297cant increase in the societal risk Moreover a dominoeffect analysis indicates that some storage tanks in the new plant

entail a high potential of being affected by events originating in the

existing plant

This research and this case study illustrate that in order to make

an objective plant lay out decision for a chemical cluster with

regard to operational risks not only conventional risk assessments

should be carried out but a domino effects analysis should be

performed as well since otherwise some risks may be under-

estimated and overlooked

It is undoubtedly true that themore accurateinformationis used

(eg concerning probabilities) the more accurate risk calculations

can be made However one should always remember that basically

calculated risks arerelativerisks andthat absolute valuesare prone

to variability and uncertainty As referencedby Pasman et al(2009)

Table 6

Target equipment affected by domino accident (due to overpressure)

scenario

Magnitude of overpressure caused by different AKPC scenarios at

target equipment (bar)

T K - 2 0 1

T K - 2 0 2

T K - 2 0 3

T K - 2 0 4

T K - 2 0 5

T K - 2 0 6

T K - 2 0 7

T K - 2 0 8

T K - 2 0 9

T K - 2 1 0

1-4 02 01 01 008 008 017 015 015

5-8 1 04 02 02 014 014 025 023 022

9 005 005 005 005 005 005 005 005 005 005

Affected by Dominoaccident

Not Affected by Dominoaccident

1 1

1

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903902

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 99

EU-studies have shown that a factor of 10e100 both ways in risk

results is not uncommon Hence future research might include risk

distributions and con1047297dence limits allowing risk analysts to even

better understand the uncertainties accompanying the results

References

Abbasi S A Krishnakumari P amp Khan F I (1998) Hot topics Global warming acidrain ozone hole hazardous waste industrial disasters disinfection New DelhiOxford University Press

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2010) A newmethod for assessing domino effect in chemical process industry Journal of Hazardous Materials 182(1e3) 416e426

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2011) Dominoeffect in process-industry accidents e an inventory of past events and identi-1047297cation of some patterns Journal of Loss Prevention in the Process Industries

24(5) 575e593Arunraj N S amp Maiti J (2009) A methodology for overall consequence modeling

in chemical industry Journal of Hazardous Materials 169 556e574Atherton J amp Gil F (2008) Incidents that de 1047297ne process safety Hoboken New

Jersey USA WileyBeerens H I Post J G amp Uijt de Haag P A M (2006) The use of generic failure

frequencies in QRA the quality and use of failure frequencies and how to bringthem up-to-date Journal of Hazardous Materials 130 265e270

Bubbico R Maschio G Mazzarotta B Milazzo M amp Parisi E (2006) Riskmanagement of road and rail transport of hazardous materials in Sicily Journalof Loss Prevention in the Process Industries 19(1) 32e38

Center for Chemical Process Safety (CCPS) (2000) Guidelines for chemical processquantitative risk analysis (2nd ed) New York American Institute of ChemicalEngineering

CozzaniV GubinelliG amp SalzanoE (2006)Escalationthresholds inthe assessmentof domino accidental events Journal of Hazardous Materials 129(1e3)1e21

CPR 18E (1999) lsquo Purple bookrsquo guidelines for quantitative risk assessment Part oneEstablishments The Hague RVIM

Dechy N Bourdeaux T Ayrault N Kordek M A amp Le Coze J C (2004) Firstlessons of the Toulouse ammonium nitrate disaster 21st September 2001 AZFplant France Journal of Hazardous Materials 111(1e3) 131e138

Gharabagh M J Asilian H Mortasavi S B Mogaddam A Z Hajizadeh E ampKhavanin A (2009) Comprehensive risk assessment and management of petrochemical feed and product transportation pipelines Journal of LossPrevention in the Process Industries 22(4) 533e539

Gooijer L Cornil N amp Lenoble C L (2012) An international comparison of fourquantitative risk assessment approaches e a benchmark study based on a 1047297cti-tious LPG plant Process Safety and Environmental Protection 90(2) 101e107

IRIMO (2006) Climatologically normal for the period 1987 e

2006 Mahshahr TehranIR of Iran Meteorological Organization Data Processing Center

Jo Y amp Crowl D (2008) Individual risk analysis of high-pressure natural gaspipelines Journal of Loss Prevention in the Process Industries 21 (6) 589e595

Kalantarnia M Khan F amp Hawboldt K (2010) Modelling of BP Texas City re1047297neryaccident using dynamic risk assessment approach Process Safety and Environ-mental Protection 88(3) 191e199 (Institution of Chemical Engineers)

Khan F I amp Abbasi S A (1998) Techniques and methodologies for risk analysis inchemical process industries Journal of Loss Prevention in the Process Industries11 261e277

Khan F I amp Abbasi S A (1999a) Major accidents in process industries and ananalysis of causes and consequences Journal of Loss Prevention in the ProcessIndustries 12 361e378

Khan F I amp Abbasi S A (1999b) Assessment of risks posed by chemical indus-triesdapplication of a new computer automated tool MAXCRED-III Journal of Loss Prevention in the Process Industries 12(6) 455e469

Khan F I Sadiq R amp Husain T (2002) Risk-based process safety assessment andcontrol measures design for offshore process facilities Journal of HazardousMaterials 94(1) 1e36

Kletz T A (1999b) The origins and history of loss prevention Transaction IChemE77 109e116

Kletz T A (1999a) What went wrong Case histories of process plant disasters andhow they could have been avoided Houston USA Gulf Professional Publishing

Kletz T A (2003) Still going wrong Case histories of process plant disasters and howthey could have been avoided Houston USA Gulf Professional Publishing

Lees F P (1996) Loss prevention in the process industries (2nd ed) Oxford LondonButterworth Heinemann

Markowski A S (2007) exLOPA for explosion risks assessment Journal of Hazardous Materials 142(3) 669e676

Milazzo M F Lisi R Maschio G Antonioni G Bonvicini S amp Spadoni G (2002)HazMat transport through Messina town from risk analysis suggestions forimproving territorial safety Journal of Loss Prevention in the Process Industries15(5) 347e356

Papazoglou I A Nivolianitou Z Aneziris O Christou M D amp Bonanos G (1999)Risk-informed selection of a highway trajectory in the neighborhood of an oil-re1047297nery Journal of Hazardous Materials 67 (2) 111e144

Pasman H J (2011) History of Dutch process equipment failure frequencies and thepurple book Journal of Loss Prevention in the Process Industries 24 208e213

Pasman H J Jung S Prem K Rogers W J amp Yang X (2009) Is risk analysisa useful tool for improving process safety Journal of Loss Prevention in theProcess Industries 22(6) 769e777

Pintaric Z (2007) Assessment of the consequences of accident scenariosinvolving dangerous substances Process Safety and Environmental Protection85(1) 23e38

Pula R Khan F I Veitch B amp Amyotte P (2006) A grid based approach for 1047297reand explosion consequence analysis Process Safety and Environmental Protec-tion 84(2) 79e91

Reniers G L L (2010) An external domino effects investment approach to improvecross-plant safety within chemical clusters Journal of Hazardous Materials 177 167e174

Reniers G L L amp Dullaert W (2007) DomPrevPlanning user-friendly software forplanning domino effects prevention Safety Science 45(10) 1060e1081

Renjith V R amp Madhu G (2010) Individual and societal risk analysis and mappingof human vulnerability to chemical accidents in the vicinity of an industrialarea International Journal of Applied Engineering Research Dindigul 1 135e148

Rigas F amp Sklavounos S (2002) Risk and consequence analyses of hazardouschemicals in marshalling yards and warehouses at IkonioPiraeus harbourGreece Journal of Loss Prevention in the Process Industries 15(6) 531e544

Salzano Eamp Cozzani V (2005)The analysis of dominoaccidents triggered byvaporcloud explosions Reliability Engineering amp System Safety 90(2e3) 271e284

Spadoni G Egidi D amp Contini S (2000) Through ARIPAR-GIS the quanti 1047297ed arearisk analysis supports land-use planning activities Journal of Hazardous Mate-rials 71(1e3) 423e437

Van Sciver G R (1990) Quantitative risk analysis in the chemical process indus-tries Reliability Engineering amp System Safety 29 55e68

Viacutelchez J A Espejo V amp Casal J (2011) Generic event trees and probabilities forthe release of different types of hazardous materials Journal of Loss Preventionin the Process Industries 24(3) 281e287

Wells G (1997) Major hazards and their management Rugby UK Institution of Chemical Engineers

Yet-Pole IChi-MinS amp Ching-Hong C (2009)Applicationsof 3D QRAtechnique tothe 1047297reexplosion simulation and hazard mitigation within a naphtha-crackingplant Journal of Loss Prevention in the Process Industries 22 (4) 506e515

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 903

Page 3: 1-s2.0-S0950423012001702-main

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 39

Besides serious physical human harm and eventually casualties

property damage and environmental effects are other conse-

quences among the overall potential loss (Pintaric 2007)

The measure of damage in QRA is fatality since other types of loss

are more complex to assess in comparison to human casualties

(Pula et al 2006)

A chain of rigorous calculations is needed to estimate the

intensity of the physical harm posed by the mentioned scenario

outcomes in a chemical cluster Software packages for the proce-

dure of consequence assessment are used to estimate the numer-

ical value of thermal heat load explosion overpressure and toxic

concentration at various spots around the release point For this

study the well-known PHAST (Process Hazard Analysis Safety Tool)

software was used for consequence assessment Once the appro-

priate failure frequencies are given this particular software con-

taining 1047297ne discharge dispersion evaporation and rainout models

is a powerful tool for the prediction of effect zones (Pintaric 2007)

In addition to the results of consequence modeling fatality probit

equations were used to quantify the expected percentage of

lethality for the exposed population

24 Risk quanti 1047297cation

The frequency and severity (rate of fatality) of every speci1047297c

scenario are combined to obtain a measure of the corresponding

risk Risk quanti1047297cation results are presented in two conventional

categories which are known as lsquoindividual riskrsquo and lsquogroup riskrsquo

(also called lsquosocietal riskrsquo)

The frequency at which a particular individual being fatally

harmed when standing at a certain distance from a potential

hazard is known as ldquoindividual riskrdquo (Gooijer Cornil amp Lenoble

2012) The overall individual risk at any location ( x y) inside or

outside the industrial plant is the summation of all individual risks

at that speci1047297c point The individual risk at any location ( x y) is

calculated by the following equation (CCPS 200 0)

IR x y frac14Xn

ifrac141

IR x yi (2)

In other words the risk of any identi1047297ed scenario will be indi-

vidually calculated at a speci1047297c location ( x y) and subsequently all

risks are summed to estimate the overall risk at that speci1047297c point

To calculate the individual risk at location ( x y) arising from

scenario i the following equation is applied (CCPS 200 0)

IR x yi frac14 f iP f i (3)

where f i is the frequency of the 1047297nal outcome of scenario i arising

from an incident (1yr) andP f iis the fatality probability of the 1047297nal

outcome of scenario i at the geographical location ( x y) Individualrisk contours represent the 1047297nal results of this step The individual

risk value in every position when compared to universally or

regionally accepted values is one of the criteria in risk-based

decision making for land use planning

Societal risk provides a risk evaluation for a group of people

located in the vicinity of the accident location In other words the

number of people affected by all 1047297nal outcomes is estimated

(Renjith amp Madhu 2010) Similar to the individual risk the societal

risk is a function of frequency of occurrence and rate of fatality

Another important and determinant factor for calculating the

societal risk is the population density around the incident location

Thesocietal riskis presented in form of FeN (FrequencyeNumber

of fatalities) curves where the cumulative frequency of 1047297nal

outcomes is plotted against the number of fatalities arising from anoutcome in a logarithmic scale To calculate the number of fatalities

of each 1047297nal outcome the following equation is used (CCPS 2000)

N i frac14X

x y

P x y p f i (4)

whereN irepresents the number of fatalities of the 1047297nal outcome of

scenario i P x yis the number of individuals at the geographical

location ( x y) and p f i is the probability that the 1047297nal outcome of

scenario i causes death at the geographical location ( x y)

The results obtained above are used in Equation (5) to calculate

the ultimate data required for plotting the FeN curve This equation

which is known as the cumulative frequency equation is expressed

as below (CCPS 2000)

F N frac14X

i

F i for all final outcome case I which N i gt N (5)

where F N is the cumulative frequency for all 1047297nal outcomes which

result in fatalities of more than N personsF i is the frequency of the

Fig 1 Layout of the study area (case-study) showing the two neighboring plants AKPC and BSPC

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 897

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 49

1047297nal outcome of scenario i and N i is the number of fatalities for the

1047297nal outcome of scenario i Finally having the quantities of F N and

N the cumulative frequency of 1047297nal outcomes is plotted against the

number of fatalities The obtained FeN curved will be compared

against the intended criteria to evaluate the acceptability of the

potential societal risk

3 Case study

ldquoAKPCrdquo and ldquoBSPCrdquo are two neighboring complexes located in

the so-called PETZONE In this industrial region 15 petrochemical

plants are situated in 5 distinct sites within an area of approxi-mately 20 km2 and therefore it is known to be one of the biggest

energy zones in the world The PETZONE lies in the northern coast

of the Persian Gulf and expands to the southwestern city of Mah-

shahr in southern Iran

Before and during the construction of many plants in the PET-

ZONE until present only economic and operational factors have

been taken into account leaving safety aspects to be largely

neglected in the layout design of those plants This was also the

case for our case study of the two plants AKPC and BSPC The ole1047297n

unit of AKPC produces ethylene and propylene and these

substances are later used in the polymerization unit to produce

high density polyethylene low density polyethylene and poly-

propylene Regarding BSPC its main products are naphtha p-

xylene and acetic acid

These two plants present an interesting case study due to their

high number of storage tanks and huge inventory of chemicals As

one of these plants provides a portion of the other one rsquos feed

constructing them close together seemed a wise choice from an

economic and operational point of view However the compliance

of risk criteria had not been an item of consideration After occur-

rence of some minor process accidents with limited consequences

in these plants the risk of accident escalation from one plant to

another was highlighted more than ever Therefore the idea of

constructing AKPC and BSPC in proximity of each other which

seemed a defendable choice once has been challenged In the

present study the two adjacent chemical complexes AKPC and

BSPC are both subjected to one QRA (treating both plants as one

plant) allowing us to analyze the effects of the construction of a

petrochemical plant adjacent to an existing one on the overall riskThe purpose was to verify whether it was indeed a wise choice to

construct these two plants this close to each other (for production

purposes) or not (for safety reasons) Fig 1 represents the layout of

the two petrochemical plants

Based on a safety review and on a safety screening carried out

as a preliminary step the major hazard sources of the study area

(as displayed in Fig 1) appear to be the storage tanks of these two

plants Although there are obviously other sources of hazards

within this area their contribution in the overall risk could be

regarded negligible (compared with the present storage tanks)

AKPC houses 5 major operational storage tanks containing

ethylene and propylene while BSPC contains 10 major atmo-spheric storage tanks containing naphtha p-xylene and acetic

acid Tables 1 and 2 provide a list of the chemical substance

inventories in AKPC and BSPC Although there were some other

storage tanks or process equipment no one passed the screening

step for scenario selection due to their low inventory of hazardous

materials and due to the lesser inherent hazardousness of the

materials

The atmosphericdata used in this study were established by the

Iranian Meteorological Organization (IMO) for the Mahshahr port

from 1988 to 2006 (IRIMO 2006) The result of the meteorological

data analysis indicates that two prevailing weather conditions (hot

season and cold season) can be intended for the QRA study during

daytime and nighttime to cover almost all of the probable condi-

tions Table 3 summarizes the average meteorological data whichare used for the consequence analysis

31 Accident scenarios

As stated previously the storage tanks have been identi1047297ed to

lead to the most hazardous scenarios in the case under study

Table 1

List of AKPC storages in this study

TK-item Compound Type Temperature

(C)

Pressure

(barg)

Volume

(m3)

Inventory

(tonne)

TK-101 Propylene Spherical 40 25 2200 1044

TK-102 Propylene Spherical 40 25 2200 1044

TK-103 Propylene Spherical 40 25 2200 1044

TK-104 Propylene Spherical 40 25 2200 1044

TK-105 Ethylene Cylindrical

104 Atm 14000 7968

Table 2

List of BSPC storages in this study

TK-item Compound Type Temperature

(C)

Pressure

(barg)

Volume

(m3)

Inventory

(tonne)

TK-201 Naphtha Cy lindrical 30 001 331188 24839

TK-202 Naphtha Cy lindrical 30 001 331188 24839

TK-203 Naphtha Cy lindrical 30 001 331188 24839

TK-204 Naphtha Cylindrical 30 001 16272 12204

TK-205 Naphtha Cylindrical 30 Atm 16272 12204TK-206 P -Xylene Cy lindrical 30 At m 170856 14693

TK-207 P -Xylene Cy lindrical 30 At m 170856 14693

TK-208 P-Xylene Cylindrical 40 0003 1500 1267

TK-209 Acetic acid Cylindrical 45 0014 1500 1554

TK-210 Acetic acid Cylindrical 45 0014 1500 1554

Table 3

Prevailing weather conditions

Category Atmospheric stability Average wind

speed (ms)

Average ambient

temp (C)

Average relative

humidity ()

Hot season (daytime) Neutral (D class) 5 40 50

Hot season (nighttime) Stable (F class) 15 25 60

Cold season (daytime) Neutral (D class) 5 20 80

Cold season (nighttime) Stable (F class) 15 10 90

Table 4

Accident scenarios and the relevant failure frequencies

Scenario

no

TK -i tem n o C ontai nmen t Scena ri o typ e Frequency

(1yr)

1e4 TK-101102103104 Propylene Leakage 1 105

5e8 TK-101102103104 Propylene Rupture 5 107

9 TK-105 Ethylene Leakage 1 105

10e14 TK-201202203

204205

Naphtha Leakage 1 105

15e17 TK-206207208 P-Xylene Leakage 1 105

18e19 TK-209 210 Acetic acid Leakage 1 105

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903898

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 59

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 69

Fig 4 A B individual risk contours before (for AKPC) and after (for cluster composed of AKPC and BSPC) constructing the new plant

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 79

the leakage case and in the rupture case Moreover based on the

plant layout review and due to the low level of equipment

congestion in this particular area under study the probability of

a 1047298ash 1047297re is higher than that of a Vapor Cloud Explosion (VCE)

since higher congestion is known to increase the probability of

a VCE rather than that of a 1047298ash 1047297re

4 Result and discussion

In order to evaluate the risk-based effects of constructing BSPC

adjacent to AKPC calculations were made for both individual risk

and societal risk before and after the new plant came under oper-

ation adjacent to the existing one The quanti1047297ed risk can then be

Fig 5 A B societal risk (Fe

N curve) before (A) and after (B) constructing the new plant

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 901

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 89

compared against the criteria to assess the acceptability of the risk

levels The likelihood of domino effects needs to be assessed as

well

41 Calculation of individual risk

Using the results of frequency estimation and a detailed

consequence assessment for all the possible scenarios the indi-

vidual risk contours were obtained and are shown on a satellite

view of the study area Key outputs of the consequence modeling

are presented in Table 5 to give more insight in the two selected

scenarios Fig 4A and B demonstrate the individual risk levels

before (A) and after (B) the second plant came operational The

numerical value of each risk contour represents the frequency at

which a particular individual is fatally harmed when standing

within the contour boundary

The results of individual risk calculations reveal that inside the

chemical cluster composed of both plants (AKPC and BSPC) as well

as within the single plant AKPC the individual risk values do not

exceed the tolerable magnitude of 105yr before and after the new

plant came under operation Fig 3A and B also demonstrate that for

public areas (such as the main roads around AKPC and BSPC) the

risk level does not exceed the magnitude of 106yr except fora short interval of 100 m at the southern main road of AKPC

Nevertheless this issue was already present even before the

construction of BSPC In general we can state that in terms of

individual risk constructing the new plant adjacent to the existing

one did not affect the acceptability of risk levels

42 Calculation of societal risk

Although the previous section indicates that individual risk

levels do not exceed the tolerable levels the impact of any potential

accident on operators and public should also be investigated

Fig 4A and B provide the societal risk calculation results before and

after the new plant came into operation respectively As it can be

seen in Fig 5A the societal risk posed by AKPC individually falls inthe ALARP (As Low As Reasonably Practicable) region as de1047297ned by

HSE (UK)Fig 5Bshows that after constructing the new plant the

societal risk curve gets closer to the maximum risk criteria but still

remains in the ALARP region Therefore constructing the new plant

BSPC adjacent to the existing one AKPC seems to have only limited

impact upon the change of tolerability of the societal risk in our

case study

43 Domino effects assessment

The distance from the storage tanks located in AKPC to the

closest operational storage tank in BSPC is about 800 m Based on

the values reported by Cozzani Gubinelli and Salzano (2006) this

distance suggests that potential cascading accidents between theneighboring plants caused by thermal radiation are highly

improbable Therefore the only factor that can trigger a secondary

accident in the other neighboring plant is the overpressure caused

by an uncon1047297ned Vapor Cloud Explosion Considering the volatility

of the materials involved the operational conditions of storage the

average ambient temperature and the prevailing wind direction

(west) we may conclude that only materials stored in AKPCrsquos

storages are capable of forming an effective vapor cloud

There are different values reported for damage threshold due to

overpressure in literature but the overpressure of 1bar would

surely destroy any target equipment (Salzano amp Cozzani 2005)

This threshold has been chosen to evaluate the possibility of

domino accidents on the storage tanks in BSPC Table 6 provides

the magnitude of overpressure caused by different AKPC scenarios

(1e9) at the location of BSPCrsquos storage tanks These values are

based on performing a detailed consequence modeling

As it can be seen in Table 6 the closest storage tanks to AKPC

(TK-201 and TK-202) are possibly damaged due to overpressurecaused by a propylene explosion after a leakage or tank rupture

eventually triggering secondary accidents In other words the

accidental release of propylene in AKPC either continuous

(scenarios 1e4) or instantaneous (scenarios 5e8) has the capa-

bility of causing cascading accidents in BSPC This suggests that if

decision-makers insist on not changing the new plantrsquos layout (eg

due to large economic and operational bene1047297ts) countermeasures

should be taken into consideration in order to minimize the

heightened risk of domino accidents

Both choices for the Purple Book and the UK risk tolerance

criteria are individual choices for this study and they should be

seen and respected as such We are aware that failure frequencies

are subject to constant optimization and that the Purple Book has

its limitations (Pasman 2011) Nonetheless for this study of twoplants situated in Iran using the frequencies reported in this well-

known and much-used work was a justi1047297able choice

5 Conclusions

The risk-based effects of constructing a new chemical plant

adjacent to an existing one forming a chemical cluster is analyzed

and discussed in this paper For this purpose a QRA was carried out

to assess different parameters of risk Individual and societal risks

were evaluated before and after the new plant came under oper-

ation The results indicate that the new plant did not have a major

impact on the risk levels since the risk levels stay within the

tolerability limits Although no major impact was observed there

was a signi1047297cant increase in the societal risk Moreover a dominoeffect analysis indicates that some storage tanks in the new plant

entail a high potential of being affected by events originating in the

existing plant

This research and this case study illustrate that in order to make

an objective plant lay out decision for a chemical cluster with

regard to operational risks not only conventional risk assessments

should be carried out but a domino effects analysis should be

performed as well since otherwise some risks may be under-

estimated and overlooked

It is undoubtedly true that themore accurateinformationis used

(eg concerning probabilities) the more accurate risk calculations

can be made However one should always remember that basically

calculated risks arerelativerisks andthat absolute valuesare prone

to variability and uncertainty As referencedby Pasman et al(2009)

Table 6

Target equipment affected by domino accident (due to overpressure)

scenario

Magnitude of overpressure caused by different AKPC scenarios at

target equipment (bar)

T K - 2 0 1

T K - 2 0 2

T K - 2 0 3

T K - 2 0 4

T K - 2 0 5

T K - 2 0 6

T K - 2 0 7

T K - 2 0 8

T K - 2 0 9

T K - 2 1 0

1-4 02 01 01 008 008 017 015 015

5-8 1 04 02 02 014 014 025 023 022

9 005 005 005 005 005 005 005 005 005 005

Affected by Dominoaccident

Not Affected by Dominoaccident

1 1

1

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903902

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 99

EU-studies have shown that a factor of 10e100 both ways in risk

results is not uncommon Hence future research might include risk

distributions and con1047297dence limits allowing risk analysts to even

better understand the uncertainties accompanying the results

References

Abbasi S A Krishnakumari P amp Khan F I (1998) Hot topics Global warming acidrain ozone hole hazardous waste industrial disasters disinfection New DelhiOxford University Press

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2010) A newmethod for assessing domino effect in chemical process industry Journal of Hazardous Materials 182(1e3) 416e426

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2011) Dominoeffect in process-industry accidents e an inventory of past events and identi-1047297cation of some patterns Journal of Loss Prevention in the Process Industries

24(5) 575e593Arunraj N S amp Maiti J (2009) A methodology for overall consequence modeling

in chemical industry Journal of Hazardous Materials 169 556e574Atherton J amp Gil F (2008) Incidents that de 1047297ne process safety Hoboken New

Jersey USA WileyBeerens H I Post J G amp Uijt de Haag P A M (2006) The use of generic failure

frequencies in QRA the quality and use of failure frequencies and how to bringthem up-to-date Journal of Hazardous Materials 130 265e270

Bubbico R Maschio G Mazzarotta B Milazzo M amp Parisi E (2006) Riskmanagement of road and rail transport of hazardous materials in Sicily Journalof Loss Prevention in the Process Industries 19(1) 32e38

Center for Chemical Process Safety (CCPS) (2000) Guidelines for chemical processquantitative risk analysis (2nd ed) New York American Institute of ChemicalEngineering

CozzaniV GubinelliG amp SalzanoE (2006)Escalationthresholds inthe assessmentof domino accidental events Journal of Hazardous Materials 129(1e3)1e21

CPR 18E (1999) lsquo Purple bookrsquo guidelines for quantitative risk assessment Part oneEstablishments The Hague RVIM

Dechy N Bourdeaux T Ayrault N Kordek M A amp Le Coze J C (2004) Firstlessons of the Toulouse ammonium nitrate disaster 21st September 2001 AZFplant France Journal of Hazardous Materials 111(1e3) 131e138

Gharabagh M J Asilian H Mortasavi S B Mogaddam A Z Hajizadeh E ampKhavanin A (2009) Comprehensive risk assessment and management of petrochemical feed and product transportation pipelines Journal of LossPrevention in the Process Industries 22(4) 533e539

Gooijer L Cornil N amp Lenoble C L (2012) An international comparison of fourquantitative risk assessment approaches e a benchmark study based on a 1047297cti-tious LPG plant Process Safety and Environmental Protection 90(2) 101e107

IRIMO (2006) Climatologically normal for the period 1987 e

2006 Mahshahr TehranIR of Iran Meteorological Organization Data Processing Center

Jo Y amp Crowl D (2008) Individual risk analysis of high-pressure natural gaspipelines Journal of Loss Prevention in the Process Industries 21 (6) 589e595

Kalantarnia M Khan F amp Hawboldt K (2010) Modelling of BP Texas City re1047297neryaccident using dynamic risk assessment approach Process Safety and Environ-mental Protection 88(3) 191e199 (Institution of Chemical Engineers)

Khan F I amp Abbasi S A (1998) Techniques and methodologies for risk analysis inchemical process industries Journal of Loss Prevention in the Process Industries11 261e277

Khan F I amp Abbasi S A (1999a) Major accidents in process industries and ananalysis of causes and consequences Journal of Loss Prevention in the ProcessIndustries 12 361e378

Khan F I amp Abbasi S A (1999b) Assessment of risks posed by chemical indus-triesdapplication of a new computer automated tool MAXCRED-III Journal of Loss Prevention in the Process Industries 12(6) 455e469

Khan F I Sadiq R amp Husain T (2002) Risk-based process safety assessment andcontrol measures design for offshore process facilities Journal of HazardousMaterials 94(1) 1e36

Kletz T A (1999b) The origins and history of loss prevention Transaction IChemE77 109e116

Kletz T A (1999a) What went wrong Case histories of process plant disasters andhow they could have been avoided Houston USA Gulf Professional Publishing

Kletz T A (2003) Still going wrong Case histories of process plant disasters and howthey could have been avoided Houston USA Gulf Professional Publishing

Lees F P (1996) Loss prevention in the process industries (2nd ed) Oxford LondonButterworth Heinemann

Markowski A S (2007) exLOPA for explosion risks assessment Journal of Hazardous Materials 142(3) 669e676

Milazzo M F Lisi R Maschio G Antonioni G Bonvicini S amp Spadoni G (2002)HazMat transport through Messina town from risk analysis suggestions forimproving territorial safety Journal of Loss Prevention in the Process Industries15(5) 347e356

Papazoglou I A Nivolianitou Z Aneziris O Christou M D amp Bonanos G (1999)Risk-informed selection of a highway trajectory in the neighborhood of an oil-re1047297nery Journal of Hazardous Materials 67 (2) 111e144

Pasman H J (2011) History of Dutch process equipment failure frequencies and thepurple book Journal of Loss Prevention in the Process Industries 24 208e213

Pasman H J Jung S Prem K Rogers W J amp Yang X (2009) Is risk analysisa useful tool for improving process safety Journal of Loss Prevention in theProcess Industries 22(6) 769e777

Pintaric Z (2007) Assessment of the consequences of accident scenariosinvolving dangerous substances Process Safety and Environmental Protection85(1) 23e38

Pula R Khan F I Veitch B amp Amyotte P (2006) A grid based approach for 1047297reand explosion consequence analysis Process Safety and Environmental Protec-tion 84(2) 79e91

Reniers G L L (2010) An external domino effects investment approach to improvecross-plant safety within chemical clusters Journal of Hazardous Materials 177 167e174

Reniers G L L amp Dullaert W (2007) DomPrevPlanning user-friendly software forplanning domino effects prevention Safety Science 45(10) 1060e1081

Renjith V R amp Madhu G (2010) Individual and societal risk analysis and mappingof human vulnerability to chemical accidents in the vicinity of an industrialarea International Journal of Applied Engineering Research Dindigul 1 135e148

Rigas F amp Sklavounos S (2002) Risk and consequence analyses of hazardouschemicals in marshalling yards and warehouses at IkonioPiraeus harbourGreece Journal of Loss Prevention in the Process Industries 15(6) 531e544

Salzano Eamp Cozzani V (2005)The analysis of dominoaccidents triggered byvaporcloud explosions Reliability Engineering amp System Safety 90(2e3) 271e284

Spadoni G Egidi D amp Contini S (2000) Through ARIPAR-GIS the quanti 1047297ed arearisk analysis supports land-use planning activities Journal of Hazardous Mate-rials 71(1e3) 423e437

Van Sciver G R (1990) Quantitative risk analysis in the chemical process indus-tries Reliability Engineering amp System Safety 29 55e68

Viacutelchez J A Espejo V amp Casal J (2011) Generic event trees and probabilities forthe release of different types of hazardous materials Journal of Loss Preventionin the Process Industries 24(3) 281e287

Wells G (1997) Major hazards and their management Rugby UK Institution of Chemical Engineers

Yet-Pole IChi-MinS amp Ching-Hong C (2009)Applicationsof 3D QRAtechnique tothe 1047297reexplosion simulation and hazard mitigation within a naphtha-crackingplant Journal of Loss Prevention in the Process Industries 22 (4) 506e515

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 903

Page 4: 1-s2.0-S0950423012001702-main

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 49

1047297nal outcome of scenario i and N i is the number of fatalities for the

1047297nal outcome of scenario i Finally having the quantities of F N and

N the cumulative frequency of 1047297nal outcomes is plotted against the

number of fatalities The obtained FeN curved will be compared

against the intended criteria to evaluate the acceptability of the

potential societal risk

3 Case study

ldquoAKPCrdquo and ldquoBSPCrdquo are two neighboring complexes located in

the so-called PETZONE In this industrial region 15 petrochemical

plants are situated in 5 distinct sites within an area of approxi-mately 20 km2 and therefore it is known to be one of the biggest

energy zones in the world The PETZONE lies in the northern coast

of the Persian Gulf and expands to the southwestern city of Mah-

shahr in southern Iran

Before and during the construction of many plants in the PET-

ZONE until present only economic and operational factors have

been taken into account leaving safety aspects to be largely

neglected in the layout design of those plants This was also the

case for our case study of the two plants AKPC and BSPC The ole1047297n

unit of AKPC produces ethylene and propylene and these

substances are later used in the polymerization unit to produce

high density polyethylene low density polyethylene and poly-

propylene Regarding BSPC its main products are naphtha p-

xylene and acetic acid

These two plants present an interesting case study due to their

high number of storage tanks and huge inventory of chemicals As

one of these plants provides a portion of the other one rsquos feed

constructing them close together seemed a wise choice from an

economic and operational point of view However the compliance

of risk criteria had not been an item of consideration After occur-

rence of some minor process accidents with limited consequences

in these plants the risk of accident escalation from one plant to

another was highlighted more than ever Therefore the idea of

constructing AKPC and BSPC in proximity of each other which

seemed a defendable choice once has been challenged In the

present study the two adjacent chemical complexes AKPC and

BSPC are both subjected to one QRA (treating both plants as one

plant) allowing us to analyze the effects of the construction of a

petrochemical plant adjacent to an existing one on the overall riskThe purpose was to verify whether it was indeed a wise choice to

construct these two plants this close to each other (for production

purposes) or not (for safety reasons) Fig 1 represents the layout of

the two petrochemical plants

Based on a safety review and on a safety screening carried out

as a preliminary step the major hazard sources of the study area

(as displayed in Fig 1) appear to be the storage tanks of these two

plants Although there are obviously other sources of hazards

within this area their contribution in the overall risk could be

regarded negligible (compared with the present storage tanks)

AKPC houses 5 major operational storage tanks containing

ethylene and propylene while BSPC contains 10 major atmo-spheric storage tanks containing naphtha p-xylene and acetic

acid Tables 1 and 2 provide a list of the chemical substance

inventories in AKPC and BSPC Although there were some other

storage tanks or process equipment no one passed the screening

step for scenario selection due to their low inventory of hazardous

materials and due to the lesser inherent hazardousness of the

materials

The atmosphericdata used in this study were established by the

Iranian Meteorological Organization (IMO) for the Mahshahr port

from 1988 to 2006 (IRIMO 2006) The result of the meteorological

data analysis indicates that two prevailing weather conditions (hot

season and cold season) can be intended for the QRA study during

daytime and nighttime to cover almost all of the probable condi-

tions Table 3 summarizes the average meteorological data whichare used for the consequence analysis

31 Accident scenarios

As stated previously the storage tanks have been identi1047297ed to

lead to the most hazardous scenarios in the case under study

Table 1

List of AKPC storages in this study

TK-item Compound Type Temperature

(C)

Pressure

(barg)

Volume

(m3)

Inventory

(tonne)

TK-101 Propylene Spherical 40 25 2200 1044

TK-102 Propylene Spherical 40 25 2200 1044

TK-103 Propylene Spherical 40 25 2200 1044

TK-104 Propylene Spherical 40 25 2200 1044

TK-105 Ethylene Cylindrical

104 Atm 14000 7968

Table 2

List of BSPC storages in this study

TK-item Compound Type Temperature

(C)

Pressure

(barg)

Volume

(m3)

Inventory

(tonne)

TK-201 Naphtha Cy lindrical 30 001 331188 24839

TK-202 Naphtha Cy lindrical 30 001 331188 24839

TK-203 Naphtha Cy lindrical 30 001 331188 24839

TK-204 Naphtha Cylindrical 30 001 16272 12204

TK-205 Naphtha Cylindrical 30 Atm 16272 12204TK-206 P -Xylene Cy lindrical 30 At m 170856 14693

TK-207 P -Xylene Cy lindrical 30 At m 170856 14693

TK-208 P-Xylene Cylindrical 40 0003 1500 1267

TK-209 Acetic acid Cylindrical 45 0014 1500 1554

TK-210 Acetic acid Cylindrical 45 0014 1500 1554

Table 3

Prevailing weather conditions

Category Atmospheric stability Average wind

speed (ms)

Average ambient

temp (C)

Average relative

humidity ()

Hot season (daytime) Neutral (D class) 5 40 50

Hot season (nighttime) Stable (F class) 15 25 60

Cold season (daytime) Neutral (D class) 5 20 80

Cold season (nighttime) Stable (F class) 15 10 90

Table 4

Accident scenarios and the relevant failure frequencies

Scenario

no

TK -i tem n o C ontai nmen t Scena ri o typ e Frequency

(1yr)

1e4 TK-101102103104 Propylene Leakage 1 105

5e8 TK-101102103104 Propylene Rupture 5 107

9 TK-105 Ethylene Leakage 1 105

10e14 TK-201202203

204205

Naphtha Leakage 1 105

15e17 TK-206207208 P-Xylene Leakage 1 105

18e19 TK-209 210 Acetic acid Leakage 1 105

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903898

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 59

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 69

Fig 4 A B individual risk contours before (for AKPC) and after (for cluster composed of AKPC and BSPC) constructing the new plant

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 79

the leakage case and in the rupture case Moreover based on the

plant layout review and due to the low level of equipment

congestion in this particular area under study the probability of

a 1047298ash 1047297re is higher than that of a Vapor Cloud Explosion (VCE)

since higher congestion is known to increase the probability of

a VCE rather than that of a 1047298ash 1047297re

4 Result and discussion

In order to evaluate the risk-based effects of constructing BSPC

adjacent to AKPC calculations were made for both individual risk

and societal risk before and after the new plant came under oper-

ation adjacent to the existing one The quanti1047297ed risk can then be

Fig 5 A B societal risk (Fe

N curve) before (A) and after (B) constructing the new plant

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 901

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 89

compared against the criteria to assess the acceptability of the risk

levels The likelihood of domino effects needs to be assessed as

well

41 Calculation of individual risk

Using the results of frequency estimation and a detailed

consequence assessment for all the possible scenarios the indi-

vidual risk contours were obtained and are shown on a satellite

view of the study area Key outputs of the consequence modeling

are presented in Table 5 to give more insight in the two selected

scenarios Fig 4A and B demonstrate the individual risk levels

before (A) and after (B) the second plant came operational The

numerical value of each risk contour represents the frequency at

which a particular individual is fatally harmed when standing

within the contour boundary

The results of individual risk calculations reveal that inside the

chemical cluster composed of both plants (AKPC and BSPC) as well

as within the single plant AKPC the individual risk values do not

exceed the tolerable magnitude of 105yr before and after the new

plant came under operation Fig 3A and B also demonstrate that for

public areas (such as the main roads around AKPC and BSPC) the

risk level does not exceed the magnitude of 106yr except fora short interval of 100 m at the southern main road of AKPC

Nevertheless this issue was already present even before the

construction of BSPC In general we can state that in terms of

individual risk constructing the new plant adjacent to the existing

one did not affect the acceptability of risk levels

42 Calculation of societal risk

Although the previous section indicates that individual risk

levels do not exceed the tolerable levels the impact of any potential

accident on operators and public should also be investigated

Fig 4A and B provide the societal risk calculation results before and

after the new plant came into operation respectively As it can be

seen in Fig 5A the societal risk posed by AKPC individually falls inthe ALARP (As Low As Reasonably Practicable) region as de1047297ned by

HSE (UK)Fig 5Bshows that after constructing the new plant the

societal risk curve gets closer to the maximum risk criteria but still

remains in the ALARP region Therefore constructing the new plant

BSPC adjacent to the existing one AKPC seems to have only limited

impact upon the change of tolerability of the societal risk in our

case study

43 Domino effects assessment

The distance from the storage tanks located in AKPC to the

closest operational storage tank in BSPC is about 800 m Based on

the values reported by Cozzani Gubinelli and Salzano (2006) this

distance suggests that potential cascading accidents between theneighboring plants caused by thermal radiation are highly

improbable Therefore the only factor that can trigger a secondary

accident in the other neighboring plant is the overpressure caused

by an uncon1047297ned Vapor Cloud Explosion Considering the volatility

of the materials involved the operational conditions of storage the

average ambient temperature and the prevailing wind direction

(west) we may conclude that only materials stored in AKPCrsquos

storages are capable of forming an effective vapor cloud

There are different values reported for damage threshold due to

overpressure in literature but the overpressure of 1bar would

surely destroy any target equipment (Salzano amp Cozzani 2005)

This threshold has been chosen to evaluate the possibility of

domino accidents on the storage tanks in BSPC Table 6 provides

the magnitude of overpressure caused by different AKPC scenarios

(1e9) at the location of BSPCrsquos storage tanks These values are

based on performing a detailed consequence modeling

As it can be seen in Table 6 the closest storage tanks to AKPC

(TK-201 and TK-202) are possibly damaged due to overpressurecaused by a propylene explosion after a leakage or tank rupture

eventually triggering secondary accidents In other words the

accidental release of propylene in AKPC either continuous

(scenarios 1e4) or instantaneous (scenarios 5e8) has the capa-

bility of causing cascading accidents in BSPC This suggests that if

decision-makers insist on not changing the new plantrsquos layout (eg

due to large economic and operational bene1047297ts) countermeasures

should be taken into consideration in order to minimize the

heightened risk of domino accidents

Both choices for the Purple Book and the UK risk tolerance

criteria are individual choices for this study and they should be

seen and respected as such We are aware that failure frequencies

are subject to constant optimization and that the Purple Book has

its limitations (Pasman 2011) Nonetheless for this study of twoplants situated in Iran using the frequencies reported in this well-

known and much-used work was a justi1047297able choice

5 Conclusions

The risk-based effects of constructing a new chemical plant

adjacent to an existing one forming a chemical cluster is analyzed

and discussed in this paper For this purpose a QRA was carried out

to assess different parameters of risk Individual and societal risks

were evaluated before and after the new plant came under oper-

ation The results indicate that the new plant did not have a major

impact on the risk levels since the risk levels stay within the

tolerability limits Although no major impact was observed there

was a signi1047297cant increase in the societal risk Moreover a dominoeffect analysis indicates that some storage tanks in the new plant

entail a high potential of being affected by events originating in the

existing plant

This research and this case study illustrate that in order to make

an objective plant lay out decision for a chemical cluster with

regard to operational risks not only conventional risk assessments

should be carried out but a domino effects analysis should be

performed as well since otherwise some risks may be under-

estimated and overlooked

It is undoubtedly true that themore accurateinformationis used

(eg concerning probabilities) the more accurate risk calculations

can be made However one should always remember that basically

calculated risks arerelativerisks andthat absolute valuesare prone

to variability and uncertainty As referencedby Pasman et al(2009)

Table 6

Target equipment affected by domino accident (due to overpressure)

scenario

Magnitude of overpressure caused by different AKPC scenarios at

target equipment (bar)

T K - 2 0 1

T K - 2 0 2

T K - 2 0 3

T K - 2 0 4

T K - 2 0 5

T K - 2 0 6

T K - 2 0 7

T K - 2 0 8

T K - 2 0 9

T K - 2 1 0

1-4 02 01 01 008 008 017 015 015

5-8 1 04 02 02 014 014 025 023 022

9 005 005 005 005 005 005 005 005 005 005

Affected by Dominoaccident

Not Affected by Dominoaccident

1 1

1

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903902

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 99

EU-studies have shown that a factor of 10e100 both ways in risk

results is not uncommon Hence future research might include risk

distributions and con1047297dence limits allowing risk analysts to even

better understand the uncertainties accompanying the results

References

Abbasi S A Krishnakumari P amp Khan F I (1998) Hot topics Global warming acidrain ozone hole hazardous waste industrial disasters disinfection New DelhiOxford University Press

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2010) A newmethod for assessing domino effect in chemical process industry Journal of Hazardous Materials 182(1e3) 416e426

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2011) Dominoeffect in process-industry accidents e an inventory of past events and identi-1047297cation of some patterns Journal of Loss Prevention in the Process Industries

24(5) 575e593Arunraj N S amp Maiti J (2009) A methodology for overall consequence modeling

in chemical industry Journal of Hazardous Materials 169 556e574Atherton J amp Gil F (2008) Incidents that de 1047297ne process safety Hoboken New

Jersey USA WileyBeerens H I Post J G amp Uijt de Haag P A M (2006) The use of generic failure

frequencies in QRA the quality and use of failure frequencies and how to bringthem up-to-date Journal of Hazardous Materials 130 265e270

Bubbico R Maschio G Mazzarotta B Milazzo M amp Parisi E (2006) Riskmanagement of road and rail transport of hazardous materials in Sicily Journalof Loss Prevention in the Process Industries 19(1) 32e38

Center for Chemical Process Safety (CCPS) (2000) Guidelines for chemical processquantitative risk analysis (2nd ed) New York American Institute of ChemicalEngineering

CozzaniV GubinelliG amp SalzanoE (2006)Escalationthresholds inthe assessmentof domino accidental events Journal of Hazardous Materials 129(1e3)1e21

CPR 18E (1999) lsquo Purple bookrsquo guidelines for quantitative risk assessment Part oneEstablishments The Hague RVIM

Dechy N Bourdeaux T Ayrault N Kordek M A amp Le Coze J C (2004) Firstlessons of the Toulouse ammonium nitrate disaster 21st September 2001 AZFplant France Journal of Hazardous Materials 111(1e3) 131e138

Gharabagh M J Asilian H Mortasavi S B Mogaddam A Z Hajizadeh E ampKhavanin A (2009) Comprehensive risk assessment and management of petrochemical feed and product transportation pipelines Journal of LossPrevention in the Process Industries 22(4) 533e539

Gooijer L Cornil N amp Lenoble C L (2012) An international comparison of fourquantitative risk assessment approaches e a benchmark study based on a 1047297cti-tious LPG plant Process Safety and Environmental Protection 90(2) 101e107

IRIMO (2006) Climatologically normal for the period 1987 e

2006 Mahshahr TehranIR of Iran Meteorological Organization Data Processing Center

Jo Y amp Crowl D (2008) Individual risk analysis of high-pressure natural gaspipelines Journal of Loss Prevention in the Process Industries 21 (6) 589e595

Kalantarnia M Khan F amp Hawboldt K (2010) Modelling of BP Texas City re1047297neryaccident using dynamic risk assessment approach Process Safety and Environ-mental Protection 88(3) 191e199 (Institution of Chemical Engineers)

Khan F I amp Abbasi S A (1998) Techniques and methodologies for risk analysis inchemical process industries Journal of Loss Prevention in the Process Industries11 261e277

Khan F I amp Abbasi S A (1999a) Major accidents in process industries and ananalysis of causes and consequences Journal of Loss Prevention in the ProcessIndustries 12 361e378

Khan F I amp Abbasi S A (1999b) Assessment of risks posed by chemical indus-triesdapplication of a new computer automated tool MAXCRED-III Journal of Loss Prevention in the Process Industries 12(6) 455e469

Khan F I Sadiq R amp Husain T (2002) Risk-based process safety assessment andcontrol measures design for offshore process facilities Journal of HazardousMaterials 94(1) 1e36

Kletz T A (1999b) The origins and history of loss prevention Transaction IChemE77 109e116

Kletz T A (1999a) What went wrong Case histories of process plant disasters andhow they could have been avoided Houston USA Gulf Professional Publishing

Kletz T A (2003) Still going wrong Case histories of process plant disasters and howthey could have been avoided Houston USA Gulf Professional Publishing

Lees F P (1996) Loss prevention in the process industries (2nd ed) Oxford LondonButterworth Heinemann

Markowski A S (2007) exLOPA for explosion risks assessment Journal of Hazardous Materials 142(3) 669e676

Milazzo M F Lisi R Maschio G Antonioni G Bonvicini S amp Spadoni G (2002)HazMat transport through Messina town from risk analysis suggestions forimproving territorial safety Journal of Loss Prevention in the Process Industries15(5) 347e356

Papazoglou I A Nivolianitou Z Aneziris O Christou M D amp Bonanos G (1999)Risk-informed selection of a highway trajectory in the neighborhood of an oil-re1047297nery Journal of Hazardous Materials 67 (2) 111e144

Pasman H J (2011) History of Dutch process equipment failure frequencies and thepurple book Journal of Loss Prevention in the Process Industries 24 208e213

Pasman H J Jung S Prem K Rogers W J amp Yang X (2009) Is risk analysisa useful tool for improving process safety Journal of Loss Prevention in theProcess Industries 22(6) 769e777

Pintaric Z (2007) Assessment of the consequences of accident scenariosinvolving dangerous substances Process Safety and Environmental Protection85(1) 23e38

Pula R Khan F I Veitch B amp Amyotte P (2006) A grid based approach for 1047297reand explosion consequence analysis Process Safety and Environmental Protec-tion 84(2) 79e91

Reniers G L L (2010) An external domino effects investment approach to improvecross-plant safety within chemical clusters Journal of Hazardous Materials 177 167e174

Reniers G L L amp Dullaert W (2007) DomPrevPlanning user-friendly software forplanning domino effects prevention Safety Science 45(10) 1060e1081

Renjith V R amp Madhu G (2010) Individual and societal risk analysis and mappingof human vulnerability to chemical accidents in the vicinity of an industrialarea International Journal of Applied Engineering Research Dindigul 1 135e148

Rigas F amp Sklavounos S (2002) Risk and consequence analyses of hazardouschemicals in marshalling yards and warehouses at IkonioPiraeus harbourGreece Journal of Loss Prevention in the Process Industries 15(6) 531e544

Salzano Eamp Cozzani V (2005)The analysis of dominoaccidents triggered byvaporcloud explosions Reliability Engineering amp System Safety 90(2e3) 271e284

Spadoni G Egidi D amp Contini S (2000) Through ARIPAR-GIS the quanti 1047297ed arearisk analysis supports land-use planning activities Journal of Hazardous Mate-rials 71(1e3) 423e437

Van Sciver G R (1990) Quantitative risk analysis in the chemical process indus-tries Reliability Engineering amp System Safety 29 55e68

Viacutelchez J A Espejo V amp Casal J (2011) Generic event trees and probabilities forthe release of different types of hazardous materials Journal of Loss Preventionin the Process Industries 24(3) 281e287

Wells G (1997) Major hazards and their management Rugby UK Institution of Chemical Engineers

Yet-Pole IChi-MinS amp Ching-Hong C (2009)Applicationsof 3D QRAtechnique tothe 1047297reexplosion simulation and hazard mitigation within a naphtha-crackingplant Journal of Loss Prevention in the Process Industries 22 (4) 506e515

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 903

Page 5: 1-s2.0-S0950423012001702-main

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 59

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 69

Fig 4 A B individual risk contours before (for AKPC) and after (for cluster composed of AKPC and BSPC) constructing the new plant

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 79

the leakage case and in the rupture case Moreover based on the

plant layout review and due to the low level of equipment

congestion in this particular area under study the probability of

a 1047298ash 1047297re is higher than that of a Vapor Cloud Explosion (VCE)

since higher congestion is known to increase the probability of

a VCE rather than that of a 1047298ash 1047297re

4 Result and discussion

In order to evaluate the risk-based effects of constructing BSPC

adjacent to AKPC calculations were made for both individual risk

and societal risk before and after the new plant came under oper-

ation adjacent to the existing one The quanti1047297ed risk can then be

Fig 5 A B societal risk (Fe

N curve) before (A) and after (B) constructing the new plant

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 901

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 89

compared against the criteria to assess the acceptability of the risk

levels The likelihood of domino effects needs to be assessed as

well

41 Calculation of individual risk

Using the results of frequency estimation and a detailed

consequence assessment for all the possible scenarios the indi-

vidual risk contours were obtained and are shown on a satellite

view of the study area Key outputs of the consequence modeling

are presented in Table 5 to give more insight in the two selected

scenarios Fig 4A and B demonstrate the individual risk levels

before (A) and after (B) the second plant came operational The

numerical value of each risk contour represents the frequency at

which a particular individual is fatally harmed when standing

within the contour boundary

The results of individual risk calculations reveal that inside the

chemical cluster composed of both plants (AKPC and BSPC) as well

as within the single plant AKPC the individual risk values do not

exceed the tolerable magnitude of 105yr before and after the new

plant came under operation Fig 3A and B also demonstrate that for

public areas (such as the main roads around AKPC and BSPC) the

risk level does not exceed the magnitude of 106yr except fora short interval of 100 m at the southern main road of AKPC

Nevertheless this issue was already present even before the

construction of BSPC In general we can state that in terms of

individual risk constructing the new plant adjacent to the existing

one did not affect the acceptability of risk levels

42 Calculation of societal risk

Although the previous section indicates that individual risk

levels do not exceed the tolerable levels the impact of any potential

accident on operators and public should also be investigated

Fig 4A and B provide the societal risk calculation results before and

after the new plant came into operation respectively As it can be

seen in Fig 5A the societal risk posed by AKPC individually falls inthe ALARP (As Low As Reasonably Practicable) region as de1047297ned by

HSE (UK)Fig 5Bshows that after constructing the new plant the

societal risk curve gets closer to the maximum risk criteria but still

remains in the ALARP region Therefore constructing the new plant

BSPC adjacent to the existing one AKPC seems to have only limited

impact upon the change of tolerability of the societal risk in our

case study

43 Domino effects assessment

The distance from the storage tanks located in AKPC to the

closest operational storage tank in BSPC is about 800 m Based on

the values reported by Cozzani Gubinelli and Salzano (2006) this

distance suggests that potential cascading accidents between theneighboring plants caused by thermal radiation are highly

improbable Therefore the only factor that can trigger a secondary

accident in the other neighboring plant is the overpressure caused

by an uncon1047297ned Vapor Cloud Explosion Considering the volatility

of the materials involved the operational conditions of storage the

average ambient temperature and the prevailing wind direction

(west) we may conclude that only materials stored in AKPCrsquos

storages are capable of forming an effective vapor cloud

There are different values reported for damage threshold due to

overpressure in literature but the overpressure of 1bar would

surely destroy any target equipment (Salzano amp Cozzani 2005)

This threshold has been chosen to evaluate the possibility of

domino accidents on the storage tanks in BSPC Table 6 provides

the magnitude of overpressure caused by different AKPC scenarios

(1e9) at the location of BSPCrsquos storage tanks These values are

based on performing a detailed consequence modeling

As it can be seen in Table 6 the closest storage tanks to AKPC

(TK-201 and TK-202) are possibly damaged due to overpressurecaused by a propylene explosion after a leakage or tank rupture

eventually triggering secondary accidents In other words the

accidental release of propylene in AKPC either continuous

(scenarios 1e4) or instantaneous (scenarios 5e8) has the capa-

bility of causing cascading accidents in BSPC This suggests that if

decision-makers insist on not changing the new plantrsquos layout (eg

due to large economic and operational bene1047297ts) countermeasures

should be taken into consideration in order to minimize the

heightened risk of domino accidents

Both choices for the Purple Book and the UK risk tolerance

criteria are individual choices for this study and they should be

seen and respected as such We are aware that failure frequencies

are subject to constant optimization and that the Purple Book has

its limitations (Pasman 2011) Nonetheless for this study of twoplants situated in Iran using the frequencies reported in this well-

known and much-used work was a justi1047297able choice

5 Conclusions

The risk-based effects of constructing a new chemical plant

adjacent to an existing one forming a chemical cluster is analyzed

and discussed in this paper For this purpose a QRA was carried out

to assess different parameters of risk Individual and societal risks

were evaluated before and after the new plant came under oper-

ation The results indicate that the new plant did not have a major

impact on the risk levels since the risk levels stay within the

tolerability limits Although no major impact was observed there

was a signi1047297cant increase in the societal risk Moreover a dominoeffect analysis indicates that some storage tanks in the new plant

entail a high potential of being affected by events originating in the

existing plant

This research and this case study illustrate that in order to make

an objective plant lay out decision for a chemical cluster with

regard to operational risks not only conventional risk assessments

should be carried out but a domino effects analysis should be

performed as well since otherwise some risks may be under-

estimated and overlooked

It is undoubtedly true that themore accurateinformationis used

(eg concerning probabilities) the more accurate risk calculations

can be made However one should always remember that basically

calculated risks arerelativerisks andthat absolute valuesare prone

to variability and uncertainty As referencedby Pasman et al(2009)

Table 6

Target equipment affected by domino accident (due to overpressure)

scenario

Magnitude of overpressure caused by different AKPC scenarios at

target equipment (bar)

T K - 2 0 1

T K - 2 0 2

T K - 2 0 3

T K - 2 0 4

T K - 2 0 5

T K - 2 0 6

T K - 2 0 7

T K - 2 0 8

T K - 2 0 9

T K - 2 1 0

1-4 02 01 01 008 008 017 015 015

5-8 1 04 02 02 014 014 025 023 022

9 005 005 005 005 005 005 005 005 005 005

Affected by Dominoaccident

Not Affected by Dominoaccident

1 1

1

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903902

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 99

EU-studies have shown that a factor of 10e100 both ways in risk

results is not uncommon Hence future research might include risk

distributions and con1047297dence limits allowing risk analysts to even

better understand the uncertainties accompanying the results

References

Abbasi S A Krishnakumari P amp Khan F I (1998) Hot topics Global warming acidrain ozone hole hazardous waste industrial disasters disinfection New DelhiOxford University Press

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2010) A newmethod for assessing domino effect in chemical process industry Journal of Hazardous Materials 182(1e3) 416e426

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2011) Dominoeffect in process-industry accidents e an inventory of past events and identi-1047297cation of some patterns Journal of Loss Prevention in the Process Industries

24(5) 575e593Arunraj N S amp Maiti J (2009) A methodology for overall consequence modeling

in chemical industry Journal of Hazardous Materials 169 556e574Atherton J amp Gil F (2008) Incidents that de 1047297ne process safety Hoboken New

Jersey USA WileyBeerens H I Post J G amp Uijt de Haag P A M (2006) The use of generic failure

frequencies in QRA the quality and use of failure frequencies and how to bringthem up-to-date Journal of Hazardous Materials 130 265e270

Bubbico R Maschio G Mazzarotta B Milazzo M amp Parisi E (2006) Riskmanagement of road and rail transport of hazardous materials in Sicily Journalof Loss Prevention in the Process Industries 19(1) 32e38

Center for Chemical Process Safety (CCPS) (2000) Guidelines for chemical processquantitative risk analysis (2nd ed) New York American Institute of ChemicalEngineering

CozzaniV GubinelliG amp SalzanoE (2006)Escalationthresholds inthe assessmentof domino accidental events Journal of Hazardous Materials 129(1e3)1e21

CPR 18E (1999) lsquo Purple bookrsquo guidelines for quantitative risk assessment Part oneEstablishments The Hague RVIM

Dechy N Bourdeaux T Ayrault N Kordek M A amp Le Coze J C (2004) Firstlessons of the Toulouse ammonium nitrate disaster 21st September 2001 AZFplant France Journal of Hazardous Materials 111(1e3) 131e138

Gharabagh M J Asilian H Mortasavi S B Mogaddam A Z Hajizadeh E ampKhavanin A (2009) Comprehensive risk assessment and management of petrochemical feed and product transportation pipelines Journal of LossPrevention in the Process Industries 22(4) 533e539

Gooijer L Cornil N amp Lenoble C L (2012) An international comparison of fourquantitative risk assessment approaches e a benchmark study based on a 1047297cti-tious LPG plant Process Safety and Environmental Protection 90(2) 101e107

IRIMO (2006) Climatologically normal for the period 1987 e

2006 Mahshahr TehranIR of Iran Meteorological Organization Data Processing Center

Jo Y amp Crowl D (2008) Individual risk analysis of high-pressure natural gaspipelines Journal of Loss Prevention in the Process Industries 21 (6) 589e595

Kalantarnia M Khan F amp Hawboldt K (2010) Modelling of BP Texas City re1047297neryaccident using dynamic risk assessment approach Process Safety and Environ-mental Protection 88(3) 191e199 (Institution of Chemical Engineers)

Khan F I amp Abbasi S A (1998) Techniques and methodologies for risk analysis inchemical process industries Journal of Loss Prevention in the Process Industries11 261e277

Khan F I amp Abbasi S A (1999a) Major accidents in process industries and ananalysis of causes and consequences Journal of Loss Prevention in the ProcessIndustries 12 361e378

Khan F I amp Abbasi S A (1999b) Assessment of risks posed by chemical indus-triesdapplication of a new computer automated tool MAXCRED-III Journal of Loss Prevention in the Process Industries 12(6) 455e469

Khan F I Sadiq R amp Husain T (2002) Risk-based process safety assessment andcontrol measures design for offshore process facilities Journal of HazardousMaterials 94(1) 1e36

Kletz T A (1999b) The origins and history of loss prevention Transaction IChemE77 109e116

Kletz T A (1999a) What went wrong Case histories of process plant disasters andhow they could have been avoided Houston USA Gulf Professional Publishing

Kletz T A (2003) Still going wrong Case histories of process plant disasters and howthey could have been avoided Houston USA Gulf Professional Publishing

Lees F P (1996) Loss prevention in the process industries (2nd ed) Oxford LondonButterworth Heinemann

Markowski A S (2007) exLOPA for explosion risks assessment Journal of Hazardous Materials 142(3) 669e676

Milazzo M F Lisi R Maschio G Antonioni G Bonvicini S amp Spadoni G (2002)HazMat transport through Messina town from risk analysis suggestions forimproving territorial safety Journal of Loss Prevention in the Process Industries15(5) 347e356

Papazoglou I A Nivolianitou Z Aneziris O Christou M D amp Bonanos G (1999)Risk-informed selection of a highway trajectory in the neighborhood of an oil-re1047297nery Journal of Hazardous Materials 67 (2) 111e144

Pasman H J (2011) History of Dutch process equipment failure frequencies and thepurple book Journal of Loss Prevention in the Process Industries 24 208e213

Pasman H J Jung S Prem K Rogers W J amp Yang X (2009) Is risk analysisa useful tool for improving process safety Journal of Loss Prevention in theProcess Industries 22(6) 769e777

Pintaric Z (2007) Assessment of the consequences of accident scenariosinvolving dangerous substances Process Safety and Environmental Protection85(1) 23e38

Pula R Khan F I Veitch B amp Amyotte P (2006) A grid based approach for 1047297reand explosion consequence analysis Process Safety and Environmental Protec-tion 84(2) 79e91

Reniers G L L (2010) An external domino effects investment approach to improvecross-plant safety within chemical clusters Journal of Hazardous Materials 177 167e174

Reniers G L L amp Dullaert W (2007) DomPrevPlanning user-friendly software forplanning domino effects prevention Safety Science 45(10) 1060e1081

Renjith V R amp Madhu G (2010) Individual and societal risk analysis and mappingof human vulnerability to chemical accidents in the vicinity of an industrialarea International Journal of Applied Engineering Research Dindigul 1 135e148

Rigas F amp Sklavounos S (2002) Risk and consequence analyses of hazardouschemicals in marshalling yards and warehouses at IkonioPiraeus harbourGreece Journal of Loss Prevention in the Process Industries 15(6) 531e544

Salzano Eamp Cozzani V (2005)The analysis of dominoaccidents triggered byvaporcloud explosions Reliability Engineering amp System Safety 90(2e3) 271e284

Spadoni G Egidi D amp Contini S (2000) Through ARIPAR-GIS the quanti 1047297ed arearisk analysis supports land-use planning activities Journal of Hazardous Mate-rials 71(1e3) 423e437

Van Sciver G R (1990) Quantitative risk analysis in the chemical process indus-tries Reliability Engineering amp System Safety 29 55e68

Viacutelchez J A Espejo V amp Casal J (2011) Generic event trees and probabilities forthe release of different types of hazardous materials Journal of Loss Preventionin the Process Industries 24(3) 281e287

Wells G (1997) Major hazards and their management Rugby UK Institution of Chemical Engineers

Yet-Pole IChi-MinS amp Ching-Hong C (2009)Applicationsof 3D QRAtechnique tothe 1047297reexplosion simulation and hazard mitigation within a naphtha-crackingplant Journal of Loss Prevention in the Process Industries 22 (4) 506e515

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 903

Page 6: 1-s2.0-S0950423012001702-main

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 69

Fig 4 A B individual risk contours before (for AKPC) and after (for cluster composed of AKPC and BSPC) constructing the new plant

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 79

the leakage case and in the rupture case Moreover based on the

plant layout review and due to the low level of equipment

congestion in this particular area under study the probability of

a 1047298ash 1047297re is higher than that of a Vapor Cloud Explosion (VCE)

since higher congestion is known to increase the probability of

a VCE rather than that of a 1047298ash 1047297re

4 Result and discussion

In order to evaluate the risk-based effects of constructing BSPC

adjacent to AKPC calculations were made for both individual risk

and societal risk before and after the new plant came under oper-

ation adjacent to the existing one The quanti1047297ed risk can then be

Fig 5 A B societal risk (Fe

N curve) before (A) and after (B) constructing the new plant

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 901

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 89

compared against the criteria to assess the acceptability of the risk

levels The likelihood of domino effects needs to be assessed as

well

41 Calculation of individual risk

Using the results of frequency estimation and a detailed

consequence assessment for all the possible scenarios the indi-

vidual risk contours were obtained and are shown on a satellite

view of the study area Key outputs of the consequence modeling

are presented in Table 5 to give more insight in the two selected

scenarios Fig 4A and B demonstrate the individual risk levels

before (A) and after (B) the second plant came operational The

numerical value of each risk contour represents the frequency at

which a particular individual is fatally harmed when standing

within the contour boundary

The results of individual risk calculations reveal that inside the

chemical cluster composed of both plants (AKPC and BSPC) as well

as within the single plant AKPC the individual risk values do not

exceed the tolerable magnitude of 105yr before and after the new

plant came under operation Fig 3A and B also demonstrate that for

public areas (such as the main roads around AKPC and BSPC) the

risk level does not exceed the magnitude of 106yr except fora short interval of 100 m at the southern main road of AKPC

Nevertheless this issue was already present even before the

construction of BSPC In general we can state that in terms of

individual risk constructing the new plant adjacent to the existing

one did not affect the acceptability of risk levels

42 Calculation of societal risk

Although the previous section indicates that individual risk

levels do not exceed the tolerable levels the impact of any potential

accident on operators and public should also be investigated

Fig 4A and B provide the societal risk calculation results before and

after the new plant came into operation respectively As it can be

seen in Fig 5A the societal risk posed by AKPC individually falls inthe ALARP (As Low As Reasonably Practicable) region as de1047297ned by

HSE (UK)Fig 5Bshows that after constructing the new plant the

societal risk curve gets closer to the maximum risk criteria but still

remains in the ALARP region Therefore constructing the new plant

BSPC adjacent to the existing one AKPC seems to have only limited

impact upon the change of tolerability of the societal risk in our

case study

43 Domino effects assessment

The distance from the storage tanks located in AKPC to the

closest operational storage tank in BSPC is about 800 m Based on

the values reported by Cozzani Gubinelli and Salzano (2006) this

distance suggests that potential cascading accidents between theneighboring plants caused by thermal radiation are highly

improbable Therefore the only factor that can trigger a secondary

accident in the other neighboring plant is the overpressure caused

by an uncon1047297ned Vapor Cloud Explosion Considering the volatility

of the materials involved the operational conditions of storage the

average ambient temperature and the prevailing wind direction

(west) we may conclude that only materials stored in AKPCrsquos

storages are capable of forming an effective vapor cloud

There are different values reported for damage threshold due to

overpressure in literature but the overpressure of 1bar would

surely destroy any target equipment (Salzano amp Cozzani 2005)

This threshold has been chosen to evaluate the possibility of

domino accidents on the storage tanks in BSPC Table 6 provides

the magnitude of overpressure caused by different AKPC scenarios

(1e9) at the location of BSPCrsquos storage tanks These values are

based on performing a detailed consequence modeling

As it can be seen in Table 6 the closest storage tanks to AKPC

(TK-201 and TK-202) are possibly damaged due to overpressurecaused by a propylene explosion after a leakage or tank rupture

eventually triggering secondary accidents In other words the

accidental release of propylene in AKPC either continuous

(scenarios 1e4) or instantaneous (scenarios 5e8) has the capa-

bility of causing cascading accidents in BSPC This suggests that if

decision-makers insist on not changing the new plantrsquos layout (eg

due to large economic and operational bene1047297ts) countermeasures

should be taken into consideration in order to minimize the

heightened risk of domino accidents

Both choices for the Purple Book and the UK risk tolerance

criteria are individual choices for this study and they should be

seen and respected as such We are aware that failure frequencies

are subject to constant optimization and that the Purple Book has

its limitations (Pasman 2011) Nonetheless for this study of twoplants situated in Iran using the frequencies reported in this well-

known and much-used work was a justi1047297able choice

5 Conclusions

The risk-based effects of constructing a new chemical plant

adjacent to an existing one forming a chemical cluster is analyzed

and discussed in this paper For this purpose a QRA was carried out

to assess different parameters of risk Individual and societal risks

were evaluated before and after the new plant came under oper-

ation The results indicate that the new plant did not have a major

impact on the risk levels since the risk levels stay within the

tolerability limits Although no major impact was observed there

was a signi1047297cant increase in the societal risk Moreover a dominoeffect analysis indicates that some storage tanks in the new plant

entail a high potential of being affected by events originating in the

existing plant

This research and this case study illustrate that in order to make

an objective plant lay out decision for a chemical cluster with

regard to operational risks not only conventional risk assessments

should be carried out but a domino effects analysis should be

performed as well since otherwise some risks may be under-

estimated and overlooked

It is undoubtedly true that themore accurateinformationis used

(eg concerning probabilities) the more accurate risk calculations

can be made However one should always remember that basically

calculated risks arerelativerisks andthat absolute valuesare prone

to variability and uncertainty As referencedby Pasman et al(2009)

Table 6

Target equipment affected by domino accident (due to overpressure)

scenario

Magnitude of overpressure caused by different AKPC scenarios at

target equipment (bar)

T K - 2 0 1

T K - 2 0 2

T K - 2 0 3

T K - 2 0 4

T K - 2 0 5

T K - 2 0 6

T K - 2 0 7

T K - 2 0 8

T K - 2 0 9

T K - 2 1 0

1-4 02 01 01 008 008 017 015 015

5-8 1 04 02 02 014 014 025 023 022

9 005 005 005 005 005 005 005 005 005 005

Affected by Dominoaccident

Not Affected by Dominoaccident

1 1

1

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903902

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 99

EU-studies have shown that a factor of 10e100 both ways in risk

results is not uncommon Hence future research might include risk

distributions and con1047297dence limits allowing risk analysts to even

better understand the uncertainties accompanying the results

References

Abbasi S A Krishnakumari P amp Khan F I (1998) Hot topics Global warming acidrain ozone hole hazardous waste industrial disasters disinfection New DelhiOxford University Press

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2010) A newmethod for assessing domino effect in chemical process industry Journal of Hazardous Materials 182(1e3) 416e426

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2011) Dominoeffect in process-industry accidents e an inventory of past events and identi-1047297cation of some patterns Journal of Loss Prevention in the Process Industries

24(5) 575e593Arunraj N S amp Maiti J (2009) A methodology for overall consequence modeling

in chemical industry Journal of Hazardous Materials 169 556e574Atherton J amp Gil F (2008) Incidents that de 1047297ne process safety Hoboken New

Jersey USA WileyBeerens H I Post J G amp Uijt de Haag P A M (2006) The use of generic failure

frequencies in QRA the quality and use of failure frequencies and how to bringthem up-to-date Journal of Hazardous Materials 130 265e270

Bubbico R Maschio G Mazzarotta B Milazzo M amp Parisi E (2006) Riskmanagement of road and rail transport of hazardous materials in Sicily Journalof Loss Prevention in the Process Industries 19(1) 32e38

Center for Chemical Process Safety (CCPS) (2000) Guidelines for chemical processquantitative risk analysis (2nd ed) New York American Institute of ChemicalEngineering

CozzaniV GubinelliG amp SalzanoE (2006)Escalationthresholds inthe assessmentof domino accidental events Journal of Hazardous Materials 129(1e3)1e21

CPR 18E (1999) lsquo Purple bookrsquo guidelines for quantitative risk assessment Part oneEstablishments The Hague RVIM

Dechy N Bourdeaux T Ayrault N Kordek M A amp Le Coze J C (2004) Firstlessons of the Toulouse ammonium nitrate disaster 21st September 2001 AZFplant France Journal of Hazardous Materials 111(1e3) 131e138

Gharabagh M J Asilian H Mortasavi S B Mogaddam A Z Hajizadeh E ampKhavanin A (2009) Comprehensive risk assessment and management of petrochemical feed and product transportation pipelines Journal of LossPrevention in the Process Industries 22(4) 533e539

Gooijer L Cornil N amp Lenoble C L (2012) An international comparison of fourquantitative risk assessment approaches e a benchmark study based on a 1047297cti-tious LPG plant Process Safety and Environmental Protection 90(2) 101e107

IRIMO (2006) Climatologically normal for the period 1987 e

2006 Mahshahr TehranIR of Iran Meteorological Organization Data Processing Center

Jo Y amp Crowl D (2008) Individual risk analysis of high-pressure natural gaspipelines Journal of Loss Prevention in the Process Industries 21 (6) 589e595

Kalantarnia M Khan F amp Hawboldt K (2010) Modelling of BP Texas City re1047297neryaccident using dynamic risk assessment approach Process Safety and Environ-mental Protection 88(3) 191e199 (Institution of Chemical Engineers)

Khan F I amp Abbasi S A (1998) Techniques and methodologies for risk analysis inchemical process industries Journal of Loss Prevention in the Process Industries11 261e277

Khan F I amp Abbasi S A (1999a) Major accidents in process industries and ananalysis of causes and consequences Journal of Loss Prevention in the ProcessIndustries 12 361e378

Khan F I amp Abbasi S A (1999b) Assessment of risks posed by chemical indus-triesdapplication of a new computer automated tool MAXCRED-III Journal of Loss Prevention in the Process Industries 12(6) 455e469

Khan F I Sadiq R amp Husain T (2002) Risk-based process safety assessment andcontrol measures design for offshore process facilities Journal of HazardousMaterials 94(1) 1e36

Kletz T A (1999b) The origins and history of loss prevention Transaction IChemE77 109e116

Kletz T A (1999a) What went wrong Case histories of process plant disasters andhow they could have been avoided Houston USA Gulf Professional Publishing

Kletz T A (2003) Still going wrong Case histories of process plant disasters and howthey could have been avoided Houston USA Gulf Professional Publishing

Lees F P (1996) Loss prevention in the process industries (2nd ed) Oxford LondonButterworth Heinemann

Markowski A S (2007) exLOPA for explosion risks assessment Journal of Hazardous Materials 142(3) 669e676

Milazzo M F Lisi R Maschio G Antonioni G Bonvicini S amp Spadoni G (2002)HazMat transport through Messina town from risk analysis suggestions forimproving territorial safety Journal of Loss Prevention in the Process Industries15(5) 347e356

Papazoglou I A Nivolianitou Z Aneziris O Christou M D amp Bonanos G (1999)Risk-informed selection of a highway trajectory in the neighborhood of an oil-re1047297nery Journal of Hazardous Materials 67 (2) 111e144

Pasman H J (2011) History of Dutch process equipment failure frequencies and thepurple book Journal of Loss Prevention in the Process Industries 24 208e213

Pasman H J Jung S Prem K Rogers W J amp Yang X (2009) Is risk analysisa useful tool for improving process safety Journal of Loss Prevention in theProcess Industries 22(6) 769e777

Pintaric Z (2007) Assessment of the consequences of accident scenariosinvolving dangerous substances Process Safety and Environmental Protection85(1) 23e38

Pula R Khan F I Veitch B amp Amyotte P (2006) A grid based approach for 1047297reand explosion consequence analysis Process Safety and Environmental Protec-tion 84(2) 79e91

Reniers G L L (2010) An external domino effects investment approach to improvecross-plant safety within chemical clusters Journal of Hazardous Materials 177 167e174

Reniers G L L amp Dullaert W (2007) DomPrevPlanning user-friendly software forplanning domino effects prevention Safety Science 45(10) 1060e1081

Renjith V R amp Madhu G (2010) Individual and societal risk analysis and mappingof human vulnerability to chemical accidents in the vicinity of an industrialarea International Journal of Applied Engineering Research Dindigul 1 135e148

Rigas F amp Sklavounos S (2002) Risk and consequence analyses of hazardouschemicals in marshalling yards and warehouses at IkonioPiraeus harbourGreece Journal of Loss Prevention in the Process Industries 15(6) 531e544

Salzano Eamp Cozzani V (2005)The analysis of dominoaccidents triggered byvaporcloud explosions Reliability Engineering amp System Safety 90(2e3) 271e284

Spadoni G Egidi D amp Contini S (2000) Through ARIPAR-GIS the quanti 1047297ed arearisk analysis supports land-use planning activities Journal of Hazardous Mate-rials 71(1e3) 423e437

Van Sciver G R (1990) Quantitative risk analysis in the chemical process indus-tries Reliability Engineering amp System Safety 29 55e68

Viacutelchez J A Espejo V amp Casal J (2011) Generic event trees and probabilities forthe release of different types of hazardous materials Journal of Loss Preventionin the Process Industries 24(3) 281e287

Wells G (1997) Major hazards and their management Rugby UK Institution of Chemical Engineers

Yet-Pole IChi-MinS amp Ching-Hong C (2009)Applicationsof 3D QRAtechnique tothe 1047297reexplosion simulation and hazard mitigation within a naphtha-crackingplant Journal of Loss Prevention in the Process Industries 22 (4) 506e515

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 903

Page 7: 1-s2.0-S0950423012001702-main

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 79

the leakage case and in the rupture case Moreover based on the

plant layout review and due to the low level of equipment

congestion in this particular area under study the probability of

a 1047298ash 1047297re is higher than that of a Vapor Cloud Explosion (VCE)

since higher congestion is known to increase the probability of

a VCE rather than that of a 1047298ash 1047297re

4 Result and discussion

In order to evaluate the risk-based effects of constructing BSPC

adjacent to AKPC calculations were made for both individual risk

and societal risk before and after the new plant came under oper-

ation adjacent to the existing one The quanti1047297ed risk can then be

Fig 5 A B societal risk (Fe

N curve) before (A) and after (B) constructing the new plant

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 901

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 89

compared against the criteria to assess the acceptability of the risk

levels The likelihood of domino effects needs to be assessed as

well

41 Calculation of individual risk

Using the results of frequency estimation and a detailed

consequence assessment for all the possible scenarios the indi-

vidual risk contours were obtained and are shown on a satellite

view of the study area Key outputs of the consequence modeling

are presented in Table 5 to give more insight in the two selected

scenarios Fig 4A and B demonstrate the individual risk levels

before (A) and after (B) the second plant came operational The

numerical value of each risk contour represents the frequency at

which a particular individual is fatally harmed when standing

within the contour boundary

The results of individual risk calculations reveal that inside the

chemical cluster composed of both plants (AKPC and BSPC) as well

as within the single plant AKPC the individual risk values do not

exceed the tolerable magnitude of 105yr before and after the new

plant came under operation Fig 3A and B also demonstrate that for

public areas (such as the main roads around AKPC and BSPC) the

risk level does not exceed the magnitude of 106yr except fora short interval of 100 m at the southern main road of AKPC

Nevertheless this issue was already present even before the

construction of BSPC In general we can state that in terms of

individual risk constructing the new plant adjacent to the existing

one did not affect the acceptability of risk levels

42 Calculation of societal risk

Although the previous section indicates that individual risk

levels do not exceed the tolerable levels the impact of any potential

accident on operators and public should also be investigated

Fig 4A and B provide the societal risk calculation results before and

after the new plant came into operation respectively As it can be

seen in Fig 5A the societal risk posed by AKPC individually falls inthe ALARP (As Low As Reasonably Practicable) region as de1047297ned by

HSE (UK)Fig 5Bshows that after constructing the new plant the

societal risk curve gets closer to the maximum risk criteria but still

remains in the ALARP region Therefore constructing the new plant

BSPC adjacent to the existing one AKPC seems to have only limited

impact upon the change of tolerability of the societal risk in our

case study

43 Domino effects assessment

The distance from the storage tanks located in AKPC to the

closest operational storage tank in BSPC is about 800 m Based on

the values reported by Cozzani Gubinelli and Salzano (2006) this

distance suggests that potential cascading accidents between theneighboring plants caused by thermal radiation are highly

improbable Therefore the only factor that can trigger a secondary

accident in the other neighboring plant is the overpressure caused

by an uncon1047297ned Vapor Cloud Explosion Considering the volatility

of the materials involved the operational conditions of storage the

average ambient temperature and the prevailing wind direction

(west) we may conclude that only materials stored in AKPCrsquos

storages are capable of forming an effective vapor cloud

There are different values reported for damage threshold due to

overpressure in literature but the overpressure of 1bar would

surely destroy any target equipment (Salzano amp Cozzani 2005)

This threshold has been chosen to evaluate the possibility of

domino accidents on the storage tanks in BSPC Table 6 provides

the magnitude of overpressure caused by different AKPC scenarios

(1e9) at the location of BSPCrsquos storage tanks These values are

based on performing a detailed consequence modeling

As it can be seen in Table 6 the closest storage tanks to AKPC

(TK-201 and TK-202) are possibly damaged due to overpressurecaused by a propylene explosion after a leakage or tank rupture

eventually triggering secondary accidents In other words the

accidental release of propylene in AKPC either continuous

(scenarios 1e4) or instantaneous (scenarios 5e8) has the capa-

bility of causing cascading accidents in BSPC This suggests that if

decision-makers insist on not changing the new plantrsquos layout (eg

due to large economic and operational bene1047297ts) countermeasures

should be taken into consideration in order to minimize the

heightened risk of domino accidents

Both choices for the Purple Book and the UK risk tolerance

criteria are individual choices for this study and they should be

seen and respected as such We are aware that failure frequencies

are subject to constant optimization and that the Purple Book has

its limitations (Pasman 2011) Nonetheless for this study of twoplants situated in Iran using the frequencies reported in this well-

known and much-used work was a justi1047297able choice

5 Conclusions

The risk-based effects of constructing a new chemical plant

adjacent to an existing one forming a chemical cluster is analyzed

and discussed in this paper For this purpose a QRA was carried out

to assess different parameters of risk Individual and societal risks

were evaluated before and after the new plant came under oper-

ation The results indicate that the new plant did not have a major

impact on the risk levels since the risk levels stay within the

tolerability limits Although no major impact was observed there

was a signi1047297cant increase in the societal risk Moreover a dominoeffect analysis indicates that some storage tanks in the new plant

entail a high potential of being affected by events originating in the

existing plant

This research and this case study illustrate that in order to make

an objective plant lay out decision for a chemical cluster with

regard to operational risks not only conventional risk assessments

should be carried out but a domino effects analysis should be

performed as well since otherwise some risks may be under-

estimated and overlooked

It is undoubtedly true that themore accurateinformationis used

(eg concerning probabilities) the more accurate risk calculations

can be made However one should always remember that basically

calculated risks arerelativerisks andthat absolute valuesare prone

to variability and uncertainty As referencedby Pasman et al(2009)

Table 6

Target equipment affected by domino accident (due to overpressure)

scenario

Magnitude of overpressure caused by different AKPC scenarios at

target equipment (bar)

T K - 2 0 1

T K - 2 0 2

T K - 2 0 3

T K - 2 0 4

T K - 2 0 5

T K - 2 0 6

T K - 2 0 7

T K - 2 0 8

T K - 2 0 9

T K - 2 1 0

1-4 02 01 01 008 008 017 015 015

5-8 1 04 02 02 014 014 025 023 022

9 005 005 005 005 005 005 005 005 005 005

Affected by Dominoaccident

Not Affected by Dominoaccident

1 1

1

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903902

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 99

EU-studies have shown that a factor of 10e100 both ways in risk

results is not uncommon Hence future research might include risk

distributions and con1047297dence limits allowing risk analysts to even

better understand the uncertainties accompanying the results

References

Abbasi S A Krishnakumari P amp Khan F I (1998) Hot topics Global warming acidrain ozone hole hazardous waste industrial disasters disinfection New DelhiOxford University Press

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2010) A newmethod for assessing domino effect in chemical process industry Journal of Hazardous Materials 182(1e3) 416e426

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2011) Dominoeffect in process-industry accidents e an inventory of past events and identi-1047297cation of some patterns Journal of Loss Prevention in the Process Industries

24(5) 575e593Arunraj N S amp Maiti J (2009) A methodology for overall consequence modeling

in chemical industry Journal of Hazardous Materials 169 556e574Atherton J amp Gil F (2008) Incidents that de 1047297ne process safety Hoboken New

Jersey USA WileyBeerens H I Post J G amp Uijt de Haag P A M (2006) The use of generic failure

frequencies in QRA the quality and use of failure frequencies and how to bringthem up-to-date Journal of Hazardous Materials 130 265e270

Bubbico R Maschio G Mazzarotta B Milazzo M amp Parisi E (2006) Riskmanagement of road and rail transport of hazardous materials in Sicily Journalof Loss Prevention in the Process Industries 19(1) 32e38

Center for Chemical Process Safety (CCPS) (2000) Guidelines for chemical processquantitative risk analysis (2nd ed) New York American Institute of ChemicalEngineering

CozzaniV GubinelliG amp SalzanoE (2006)Escalationthresholds inthe assessmentof domino accidental events Journal of Hazardous Materials 129(1e3)1e21

CPR 18E (1999) lsquo Purple bookrsquo guidelines for quantitative risk assessment Part oneEstablishments The Hague RVIM

Dechy N Bourdeaux T Ayrault N Kordek M A amp Le Coze J C (2004) Firstlessons of the Toulouse ammonium nitrate disaster 21st September 2001 AZFplant France Journal of Hazardous Materials 111(1e3) 131e138

Gharabagh M J Asilian H Mortasavi S B Mogaddam A Z Hajizadeh E ampKhavanin A (2009) Comprehensive risk assessment and management of petrochemical feed and product transportation pipelines Journal of LossPrevention in the Process Industries 22(4) 533e539

Gooijer L Cornil N amp Lenoble C L (2012) An international comparison of fourquantitative risk assessment approaches e a benchmark study based on a 1047297cti-tious LPG plant Process Safety and Environmental Protection 90(2) 101e107

IRIMO (2006) Climatologically normal for the period 1987 e

2006 Mahshahr TehranIR of Iran Meteorological Organization Data Processing Center

Jo Y amp Crowl D (2008) Individual risk analysis of high-pressure natural gaspipelines Journal of Loss Prevention in the Process Industries 21 (6) 589e595

Kalantarnia M Khan F amp Hawboldt K (2010) Modelling of BP Texas City re1047297neryaccident using dynamic risk assessment approach Process Safety and Environ-mental Protection 88(3) 191e199 (Institution of Chemical Engineers)

Khan F I amp Abbasi S A (1998) Techniques and methodologies for risk analysis inchemical process industries Journal of Loss Prevention in the Process Industries11 261e277

Khan F I amp Abbasi S A (1999a) Major accidents in process industries and ananalysis of causes and consequences Journal of Loss Prevention in the ProcessIndustries 12 361e378

Khan F I amp Abbasi S A (1999b) Assessment of risks posed by chemical indus-triesdapplication of a new computer automated tool MAXCRED-III Journal of Loss Prevention in the Process Industries 12(6) 455e469

Khan F I Sadiq R amp Husain T (2002) Risk-based process safety assessment andcontrol measures design for offshore process facilities Journal of HazardousMaterials 94(1) 1e36

Kletz T A (1999b) The origins and history of loss prevention Transaction IChemE77 109e116

Kletz T A (1999a) What went wrong Case histories of process plant disasters andhow they could have been avoided Houston USA Gulf Professional Publishing

Kletz T A (2003) Still going wrong Case histories of process plant disasters and howthey could have been avoided Houston USA Gulf Professional Publishing

Lees F P (1996) Loss prevention in the process industries (2nd ed) Oxford LondonButterworth Heinemann

Markowski A S (2007) exLOPA for explosion risks assessment Journal of Hazardous Materials 142(3) 669e676

Milazzo M F Lisi R Maschio G Antonioni G Bonvicini S amp Spadoni G (2002)HazMat transport through Messina town from risk analysis suggestions forimproving territorial safety Journal of Loss Prevention in the Process Industries15(5) 347e356

Papazoglou I A Nivolianitou Z Aneziris O Christou M D amp Bonanos G (1999)Risk-informed selection of a highway trajectory in the neighborhood of an oil-re1047297nery Journal of Hazardous Materials 67 (2) 111e144

Pasman H J (2011) History of Dutch process equipment failure frequencies and thepurple book Journal of Loss Prevention in the Process Industries 24 208e213

Pasman H J Jung S Prem K Rogers W J amp Yang X (2009) Is risk analysisa useful tool for improving process safety Journal of Loss Prevention in theProcess Industries 22(6) 769e777

Pintaric Z (2007) Assessment of the consequences of accident scenariosinvolving dangerous substances Process Safety and Environmental Protection85(1) 23e38

Pula R Khan F I Veitch B amp Amyotte P (2006) A grid based approach for 1047297reand explosion consequence analysis Process Safety and Environmental Protec-tion 84(2) 79e91

Reniers G L L (2010) An external domino effects investment approach to improvecross-plant safety within chemical clusters Journal of Hazardous Materials 177 167e174

Reniers G L L amp Dullaert W (2007) DomPrevPlanning user-friendly software forplanning domino effects prevention Safety Science 45(10) 1060e1081

Renjith V R amp Madhu G (2010) Individual and societal risk analysis and mappingof human vulnerability to chemical accidents in the vicinity of an industrialarea International Journal of Applied Engineering Research Dindigul 1 135e148

Rigas F amp Sklavounos S (2002) Risk and consequence analyses of hazardouschemicals in marshalling yards and warehouses at IkonioPiraeus harbourGreece Journal of Loss Prevention in the Process Industries 15(6) 531e544

Salzano Eamp Cozzani V (2005)The analysis of dominoaccidents triggered byvaporcloud explosions Reliability Engineering amp System Safety 90(2e3) 271e284

Spadoni G Egidi D amp Contini S (2000) Through ARIPAR-GIS the quanti 1047297ed arearisk analysis supports land-use planning activities Journal of Hazardous Mate-rials 71(1e3) 423e437

Van Sciver G R (1990) Quantitative risk analysis in the chemical process indus-tries Reliability Engineering amp System Safety 29 55e68

Viacutelchez J A Espejo V amp Casal J (2011) Generic event trees and probabilities forthe release of different types of hazardous materials Journal of Loss Preventionin the Process Industries 24(3) 281e287

Wells G (1997) Major hazards and their management Rugby UK Institution of Chemical Engineers

Yet-Pole IChi-MinS amp Ching-Hong C (2009)Applicationsof 3D QRAtechnique tothe 1047297reexplosion simulation and hazard mitigation within a naphtha-crackingplant Journal of Loss Prevention in the Process Industries 22 (4) 506e515

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 903

Page 8: 1-s2.0-S0950423012001702-main

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 89

compared against the criteria to assess the acceptability of the risk

levels The likelihood of domino effects needs to be assessed as

well

41 Calculation of individual risk

Using the results of frequency estimation and a detailed

consequence assessment for all the possible scenarios the indi-

vidual risk contours were obtained and are shown on a satellite

view of the study area Key outputs of the consequence modeling

are presented in Table 5 to give more insight in the two selected

scenarios Fig 4A and B demonstrate the individual risk levels

before (A) and after (B) the second plant came operational The

numerical value of each risk contour represents the frequency at

which a particular individual is fatally harmed when standing

within the contour boundary

The results of individual risk calculations reveal that inside the

chemical cluster composed of both plants (AKPC and BSPC) as well

as within the single plant AKPC the individual risk values do not

exceed the tolerable magnitude of 105yr before and after the new

plant came under operation Fig 3A and B also demonstrate that for

public areas (such as the main roads around AKPC and BSPC) the

risk level does not exceed the magnitude of 106yr except fora short interval of 100 m at the southern main road of AKPC

Nevertheless this issue was already present even before the

construction of BSPC In general we can state that in terms of

individual risk constructing the new plant adjacent to the existing

one did not affect the acceptability of risk levels

42 Calculation of societal risk

Although the previous section indicates that individual risk

levels do not exceed the tolerable levels the impact of any potential

accident on operators and public should also be investigated

Fig 4A and B provide the societal risk calculation results before and

after the new plant came into operation respectively As it can be

seen in Fig 5A the societal risk posed by AKPC individually falls inthe ALARP (As Low As Reasonably Practicable) region as de1047297ned by

HSE (UK)Fig 5Bshows that after constructing the new plant the

societal risk curve gets closer to the maximum risk criteria but still

remains in the ALARP region Therefore constructing the new plant

BSPC adjacent to the existing one AKPC seems to have only limited

impact upon the change of tolerability of the societal risk in our

case study

43 Domino effects assessment

The distance from the storage tanks located in AKPC to the

closest operational storage tank in BSPC is about 800 m Based on

the values reported by Cozzani Gubinelli and Salzano (2006) this

distance suggests that potential cascading accidents between theneighboring plants caused by thermal radiation are highly

improbable Therefore the only factor that can trigger a secondary

accident in the other neighboring plant is the overpressure caused

by an uncon1047297ned Vapor Cloud Explosion Considering the volatility

of the materials involved the operational conditions of storage the

average ambient temperature and the prevailing wind direction

(west) we may conclude that only materials stored in AKPCrsquos

storages are capable of forming an effective vapor cloud

There are different values reported for damage threshold due to

overpressure in literature but the overpressure of 1bar would

surely destroy any target equipment (Salzano amp Cozzani 2005)

This threshold has been chosen to evaluate the possibility of

domino accidents on the storage tanks in BSPC Table 6 provides

the magnitude of overpressure caused by different AKPC scenarios

(1e9) at the location of BSPCrsquos storage tanks These values are

based on performing a detailed consequence modeling

As it can be seen in Table 6 the closest storage tanks to AKPC

(TK-201 and TK-202) are possibly damaged due to overpressurecaused by a propylene explosion after a leakage or tank rupture

eventually triggering secondary accidents In other words the

accidental release of propylene in AKPC either continuous

(scenarios 1e4) or instantaneous (scenarios 5e8) has the capa-

bility of causing cascading accidents in BSPC This suggests that if

decision-makers insist on not changing the new plantrsquos layout (eg

due to large economic and operational bene1047297ts) countermeasures

should be taken into consideration in order to minimize the

heightened risk of domino accidents

Both choices for the Purple Book and the UK risk tolerance

criteria are individual choices for this study and they should be

seen and respected as such We are aware that failure frequencies

are subject to constant optimization and that the Purple Book has

its limitations (Pasman 2011) Nonetheless for this study of twoplants situated in Iran using the frequencies reported in this well-

known and much-used work was a justi1047297able choice

5 Conclusions

The risk-based effects of constructing a new chemical plant

adjacent to an existing one forming a chemical cluster is analyzed

and discussed in this paper For this purpose a QRA was carried out

to assess different parameters of risk Individual and societal risks

were evaluated before and after the new plant came under oper-

ation The results indicate that the new plant did not have a major

impact on the risk levels since the risk levels stay within the

tolerability limits Although no major impact was observed there

was a signi1047297cant increase in the societal risk Moreover a dominoeffect analysis indicates that some storage tanks in the new plant

entail a high potential of being affected by events originating in the

existing plant

This research and this case study illustrate that in order to make

an objective plant lay out decision for a chemical cluster with

regard to operational risks not only conventional risk assessments

should be carried out but a domino effects analysis should be

performed as well since otherwise some risks may be under-

estimated and overlooked

It is undoubtedly true that themore accurateinformationis used

(eg concerning probabilities) the more accurate risk calculations

can be made However one should always remember that basically

calculated risks arerelativerisks andthat absolute valuesare prone

to variability and uncertainty As referencedby Pasman et al(2009)

Table 6

Target equipment affected by domino accident (due to overpressure)

scenario

Magnitude of overpressure caused by different AKPC scenarios at

target equipment (bar)

T K - 2 0 1

T K - 2 0 2

T K - 2 0 3

T K - 2 0 4

T K - 2 0 5

T K - 2 0 6

T K - 2 0 7

T K - 2 0 8

T K - 2 0 9

T K - 2 1 0

1-4 02 01 01 008 008 017 015 015

5-8 1 04 02 02 014 014 025 023 022

9 005 005 005 005 005 005 005 005 005 005

Affected by Dominoaccident

Not Affected by Dominoaccident

1 1

1

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903902

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 99

EU-studies have shown that a factor of 10e100 both ways in risk

results is not uncommon Hence future research might include risk

distributions and con1047297dence limits allowing risk analysts to even

better understand the uncertainties accompanying the results

References

Abbasi S A Krishnakumari P amp Khan F I (1998) Hot topics Global warming acidrain ozone hole hazardous waste industrial disasters disinfection New DelhiOxford University Press

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2010) A newmethod for assessing domino effect in chemical process industry Journal of Hazardous Materials 182(1e3) 416e426

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2011) Dominoeffect in process-industry accidents e an inventory of past events and identi-1047297cation of some patterns Journal of Loss Prevention in the Process Industries

24(5) 575e593Arunraj N S amp Maiti J (2009) A methodology for overall consequence modeling

in chemical industry Journal of Hazardous Materials 169 556e574Atherton J amp Gil F (2008) Incidents that de 1047297ne process safety Hoboken New

Jersey USA WileyBeerens H I Post J G amp Uijt de Haag P A M (2006) The use of generic failure

frequencies in QRA the quality and use of failure frequencies and how to bringthem up-to-date Journal of Hazardous Materials 130 265e270

Bubbico R Maschio G Mazzarotta B Milazzo M amp Parisi E (2006) Riskmanagement of road and rail transport of hazardous materials in Sicily Journalof Loss Prevention in the Process Industries 19(1) 32e38

Center for Chemical Process Safety (CCPS) (2000) Guidelines for chemical processquantitative risk analysis (2nd ed) New York American Institute of ChemicalEngineering

CozzaniV GubinelliG amp SalzanoE (2006)Escalationthresholds inthe assessmentof domino accidental events Journal of Hazardous Materials 129(1e3)1e21

CPR 18E (1999) lsquo Purple bookrsquo guidelines for quantitative risk assessment Part oneEstablishments The Hague RVIM

Dechy N Bourdeaux T Ayrault N Kordek M A amp Le Coze J C (2004) Firstlessons of the Toulouse ammonium nitrate disaster 21st September 2001 AZFplant France Journal of Hazardous Materials 111(1e3) 131e138

Gharabagh M J Asilian H Mortasavi S B Mogaddam A Z Hajizadeh E ampKhavanin A (2009) Comprehensive risk assessment and management of petrochemical feed and product transportation pipelines Journal of LossPrevention in the Process Industries 22(4) 533e539

Gooijer L Cornil N amp Lenoble C L (2012) An international comparison of fourquantitative risk assessment approaches e a benchmark study based on a 1047297cti-tious LPG plant Process Safety and Environmental Protection 90(2) 101e107

IRIMO (2006) Climatologically normal for the period 1987 e

2006 Mahshahr TehranIR of Iran Meteorological Organization Data Processing Center

Jo Y amp Crowl D (2008) Individual risk analysis of high-pressure natural gaspipelines Journal of Loss Prevention in the Process Industries 21 (6) 589e595

Kalantarnia M Khan F amp Hawboldt K (2010) Modelling of BP Texas City re1047297neryaccident using dynamic risk assessment approach Process Safety and Environ-mental Protection 88(3) 191e199 (Institution of Chemical Engineers)

Khan F I amp Abbasi S A (1998) Techniques and methodologies for risk analysis inchemical process industries Journal of Loss Prevention in the Process Industries11 261e277

Khan F I amp Abbasi S A (1999a) Major accidents in process industries and ananalysis of causes and consequences Journal of Loss Prevention in the ProcessIndustries 12 361e378

Khan F I amp Abbasi S A (1999b) Assessment of risks posed by chemical indus-triesdapplication of a new computer automated tool MAXCRED-III Journal of Loss Prevention in the Process Industries 12(6) 455e469

Khan F I Sadiq R amp Husain T (2002) Risk-based process safety assessment andcontrol measures design for offshore process facilities Journal of HazardousMaterials 94(1) 1e36

Kletz T A (1999b) The origins and history of loss prevention Transaction IChemE77 109e116

Kletz T A (1999a) What went wrong Case histories of process plant disasters andhow they could have been avoided Houston USA Gulf Professional Publishing

Kletz T A (2003) Still going wrong Case histories of process plant disasters and howthey could have been avoided Houston USA Gulf Professional Publishing

Lees F P (1996) Loss prevention in the process industries (2nd ed) Oxford LondonButterworth Heinemann

Markowski A S (2007) exLOPA for explosion risks assessment Journal of Hazardous Materials 142(3) 669e676

Milazzo M F Lisi R Maschio G Antonioni G Bonvicini S amp Spadoni G (2002)HazMat transport through Messina town from risk analysis suggestions forimproving territorial safety Journal of Loss Prevention in the Process Industries15(5) 347e356

Papazoglou I A Nivolianitou Z Aneziris O Christou M D amp Bonanos G (1999)Risk-informed selection of a highway trajectory in the neighborhood of an oil-re1047297nery Journal of Hazardous Materials 67 (2) 111e144

Pasman H J (2011) History of Dutch process equipment failure frequencies and thepurple book Journal of Loss Prevention in the Process Industries 24 208e213

Pasman H J Jung S Prem K Rogers W J amp Yang X (2009) Is risk analysisa useful tool for improving process safety Journal of Loss Prevention in theProcess Industries 22(6) 769e777

Pintaric Z (2007) Assessment of the consequences of accident scenariosinvolving dangerous substances Process Safety and Environmental Protection85(1) 23e38

Pula R Khan F I Veitch B amp Amyotte P (2006) A grid based approach for 1047297reand explosion consequence analysis Process Safety and Environmental Protec-tion 84(2) 79e91

Reniers G L L (2010) An external domino effects investment approach to improvecross-plant safety within chemical clusters Journal of Hazardous Materials 177 167e174

Reniers G L L amp Dullaert W (2007) DomPrevPlanning user-friendly software forplanning domino effects prevention Safety Science 45(10) 1060e1081

Renjith V R amp Madhu G (2010) Individual and societal risk analysis and mappingof human vulnerability to chemical accidents in the vicinity of an industrialarea International Journal of Applied Engineering Research Dindigul 1 135e148

Rigas F amp Sklavounos S (2002) Risk and consequence analyses of hazardouschemicals in marshalling yards and warehouses at IkonioPiraeus harbourGreece Journal of Loss Prevention in the Process Industries 15(6) 531e544

Salzano Eamp Cozzani V (2005)The analysis of dominoaccidents triggered byvaporcloud explosions Reliability Engineering amp System Safety 90(2e3) 271e284

Spadoni G Egidi D amp Contini S (2000) Through ARIPAR-GIS the quanti 1047297ed arearisk analysis supports land-use planning activities Journal of Hazardous Mate-rials 71(1e3) 423e437

Van Sciver G R (1990) Quantitative risk analysis in the chemical process indus-tries Reliability Engineering amp System Safety 29 55e68

Viacutelchez J A Espejo V amp Casal J (2011) Generic event trees and probabilities forthe release of different types of hazardous materials Journal of Loss Preventionin the Process Industries 24(3) 281e287

Wells G (1997) Major hazards and their management Rugby UK Institution of Chemical Engineers

Yet-Pole IChi-MinS amp Ching-Hong C (2009)Applicationsof 3D QRAtechnique tothe 1047297reexplosion simulation and hazard mitigation within a naphtha-crackingplant Journal of Loss Prevention in the Process Industries 22 (4) 506e515

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 903

Page 9: 1-s2.0-S0950423012001702-main

8132019 1-s20-S0950423012001702-main

httpslidepdfcomreaderfull1-s20-s0950423012001702-main 99

EU-studies have shown that a factor of 10e100 both ways in risk

results is not uncommon Hence future research might include risk

distributions and con1047297dence limits allowing risk analysts to even

better understand the uncertainties accompanying the results

References

Abbasi S A Krishnakumari P amp Khan F I (1998) Hot topics Global warming acidrain ozone hole hazardous waste industrial disasters disinfection New DelhiOxford University Press

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2010) A newmethod for assessing domino effect in chemical process industry Journal of Hazardous Materials 182(1e3) 416e426

Abdolhamidzadeh B Abbasi T Rashtchian D amp Abbasi S A (2011) Dominoeffect in process-industry accidents e an inventory of past events and identi-1047297cation of some patterns Journal of Loss Prevention in the Process Industries

24(5) 575e593Arunraj N S amp Maiti J (2009) A methodology for overall consequence modeling

in chemical industry Journal of Hazardous Materials 169 556e574Atherton J amp Gil F (2008) Incidents that de 1047297ne process safety Hoboken New

Jersey USA WileyBeerens H I Post J G amp Uijt de Haag P A M (2006) The use of generic failure

frequencies in QRA the quality and use of failure frequencies and how to bringthem up-to-date Journal of Hazardous Materials 130 265e270

Bubbico R Maschio G Mazzarotta B Milazzo M amp Parisi E (2006) Riskmanagement of road and rail transport of hazardous materials in Sicily Journalof Loss Prevention in the Process Industries 19(1) 32e38

Center for Chemical Process Safety (CCPS) (2000) Guidelines for chemical processquantitative risk analysis (2nd ed) New York American Institute of ChemicalEngineering

CozzaniV GubinelliG amp SalzanoE (2006)Escalationthresholds inthe assessmentof domino accidental events Journal of Hazardous Materials 129(1e3)1e21

CPR 18E (1999) lsquo Purple bookrsquo guidelines for quantitative risk assessment Part oneEstablishments The Hague RVIM

Dechy N Bourdeaux T Ayrault N Kordek M A amp Le Coze J C (2004) Firstlessons of the Toulouse ammonium nitrate disaster 21st September 2001 AZFplant France Journal of Hazardous Materials 111(1e3) 131e138

Gharabagh M J Asilian H Mortasavi S B Mogaddam A Z Hajizadeh E ampKhavanin A (2009) Comprehensive risk assessment and management of petrochemical feed and product transportation pipelines Journal of LossPrevention in the Process Industries 22(4) 533e539

Gooijer L Cornil N amp Lenoble C L (2012) An international comparison of fourquantitative risk assessment approaches e a benchmark study based on a 1047297cti-tious LPG plant Process Safety and Environmental Protection 90(2) 101e107

IRIMO (2006) Climatologically normal for the period 1987 e

2006 Mahshahr TehranIR of Iran Meteorological Organization Data Processing Center

Jo Y amp Crowl D (2008) Individual risk analysis of high-pressure natural gaspipelines Journal of Loss Prevention in the Process Industries 21 (6) 589e595

Kalantarnia M Khan F amp Hawboldt K (2010) Modelling of BP Texas City re1047297neryaccident using dynamic risk assessment approach Process Safety and Environ-mental Protection 88(3) 191e199 (Institution of Chemical Engineers)

Khan F I amp Abbasi S A (1998) Techniques and methodologies for risk analysis inchemical process industries Journal of Loss Prevention in the Process Industries11 261e277

Khan F I amp Abbasi S A (1999a) Major accidents in process industries and ananalysis of causes and consequences Journal of Loss Prevention in the ProcessIndustries 12 361e378

Khan F I amp Abbasi S A (1999b) Assessment of risks posed by chemical indus-triesdapplication of a new computer automated tool MAXCRED-III Journal of Loss Prevention in the Process Industries 12(6) 455e469

Khan F I Sadiq R amp Husain T (2002) Risk-based process safety assessment andcontrol measures design for offshore process facilities Journal of HazardousMaterials 94(1) 1e36

Kletz T A (1999b) The origins and history of loss prevention Transaction IChemE77 109e116

Kletz T A (1999a) What went wrong Case histories of process plant disasters andhow they could have been avoided Houston USA Gulf Professional Publishing

Kletz T A (2003) Still going wrong Case histories of process plant disasters and howthey could have been avoided Houston USA Gulf Professional Publishing

Lees F P (1996) Loss prevention in the process industries (2nd ed) Oxford LondonButterworth Heinemann

Markowski A S (2007) exLOPA for explosion risks assessment Journal of Hazardous Materials 142(3) 669e676

Milazzo M F Lisi R Maschio G Antonioni G Bonvicini S amp Spadoni G (2002)HazMat transport through Messina town from risk analysis suggestions forimproving territorial safety Journal of Loss Prevention in the Process Industries15(5) 347e356

Papazoglou I A Nivolianitou Z Aneziris O Christou M D amp Bonanos G (1999)Risk-informed selection of a highway trajectory in the neighborhood of an oil-re1047297nery Journal of Hazardous Materials 67 (2) 111e144

Pasman H J (2011) History of Dutch process equipment failure frequencies and thepurple book Journal of Loss Prevention in the Process Industries 24 208e213

Pasman H J Jung S Prem K Rogers W J amp Yang X (2009) Is risk analysisa useful tool for improving process safety Journal of Loss Prevention in theProcess Industries 22(6) 769e777

Pintaric Z (2007) Assessment of the consequences of accident scenariosinvolving dangerous substances Process Safety and Environmental Protection85(1) 23e38

Pula R Khan F I Veitch B amp Amyotte P (2006) A grid based approach for 1047297reand explosion consequence analysis Process Safety and Environmental Protec-tion 84(2) 79e91

Reniers G L L (2010) An external domino effects investment approach to improvecross-plant safety within chemical clusters Journal of Hazardous Materials 177 167e174

Reniers G L L amp Dullaert W (2007) DomPrevPlanning user-friendly software forplanning domino effects prevention Safety Science 45(10) 1060e1081

Renjith V R amp Madhu G (2010) Individual and societal risk analysis and mappingof human vulnerability to chemical accidents in the vicinity of an industrialarea International Journal of Applied Engineering Research Dindigul 1 135e148

Rigas F amp Sklavounos S (2002) Risk and consequence analyses of hazardouschemicals in marshalling yards and warehouses at IkonioPiraeus harbourGreece Journal of Loss Prevention in the Process Industries 15(6) 531e544

Salzano Eamp Cozzani V (2005)The analysis of dominoaccidents triggered byvaporcloud explosions Reliability Engineering amp System Safety 90(2e3) 271e284

Spadoni G Egidi D amp Contini S (2000) Through ARIPAR-GIS the quanti 1047297ed arearisk analysis supports land-use planning activities Journal of Hazardous Mate-rials 71(1e3) 423e437

Van Sciver G R (1990) Quantitative risk analysis in the chemical process indus-tries Reliability Engineering amp System Safety 29 55e68

Viacutelchez J A Espejo V amp Casal J (2011) Generic event trees and probabilities forthe release of different types of hazardous materials Journal of Loss Preventionin the Process Industries 24(3) 281e287

Wells G (1997) Major hazards and their management Rugby UK Institution of Chemical Engineers

Yet-Pole IChi-MinS amp Ching-Hong C (2009)Applicationsof 3D QRAtechnique tothe 1047297reexplosion simulation and hazard mitigation within a naphtha-crackingplant Journal of Loss Prevention in the Process Industries 22 (4) 506e515

S Baesi et al Journal of Loss Prevention in the Process Industries 26 (2013) 895e903 903