Estimation of safety distances in the vicinity of fuel gas pipelines

Estimation of safety distances in the vicinity of fuel gas pipelinesSpyros Sklavounos,Fotis Rigas*School of Chemical Engineering,National Technical University of Athens,15700Athens,GreeceReceived2November2004;received in revised form19April2005;accepted10May2005AbstractIn this paper,safety distances around pipelines transmitting liquefied petroleum gas and pressurized natural gas are determined considering the possible outcomes of an accidental event associated with fuel gas release from pressurized transmission systems.Possible outcomes of an accidental fuel gas release were determined by performing the Event Tree Analysis approach.Safety distances were computed for two pipeline transmission systems of pressurized natural gas and liquefied petroleum gas existing in Greece using real data given by Greek Refineries and the Greek Public Gas Enterprise.The software packages CHETAH and BREEZE were used for thermochemical mixture properties estimation and quantitative consequence assessment,respectively.Safety distance determination was performed considering jetfire and gas dispersion to the lowerflammable limit as the worst-case scenarios corresponding to immediate and delayed cloud ignition.The results showed that the jetfire scenario should be considered as the limiter for safety distances determination in the vicinity of natural and petroleum gas pipelines.Based on this conclusion,the obtained results were further treated to yield functional diagrams for prompt safety distance estimation.In addition,qualitative conclusions were made regarding the effect of atmospheric conditions on possible events.Thus,wind velocity was found to dominate during a jetfire event suppressing the thermal radiation effect,whereas gas dispersion was found to be affected mainly by solar radiation that favors the faster dissolution of fuel gas below the lowerflammable limit.q2005Elsevier Ltd.All rights reserved.Keywords:Safety distance;Event Tree Analysis;Consequence modeling;Jetfire;Gas dispersion;Pipeline safety1.IntroductionIn recent years,a great deal of effort has been dedicated not only to accident prevention,but also to the mitigation of accident consequences.In Directive96/61/EEC,article3.e, as a general mitigation principle,the necessary measures are indicated that must be taken to limit accident consequences, while in the Directive96/82/EEC(December2001 Amendment)the compliance with appropriate distances between establishments of dangerous substances and residential areas or areas of public use is imposed(article 12.1),in order to limit the effects of a potential accident on people and human property(EEC,1996a,b).Indeed,USA Federal Office of Pipeline Safety(OPS)is going to issue new regulations for gas pipeline integrity management in high consequence areas,which will identify the vulner-ability zones along the length of pipeline installations (DeWolf,2003).Petroleum gas(PG)was always of great importance, used in chemical processes or for domestic service,while natural gas(NG)use has increased rapidly replacing ordinary fuels and electricity not only for environmental but also for economical reasons.As a result,new storage units are constructed and larger amounts of petroleum and especially natural gas are stored and transported worldwide. In addition to sea transport,fuel gases are transmitted through pipeline systems,which is the point of interest of this work.Pipelines transporting fuel gases for commercial purposes have many times been involved in major accidents (IChemE,2000;Khan&Abbasi,1999).Indeed,pipe failure rates assessed from recent European data range from2.1! 10K4(for small diameters)to7.7!10K4(for large diameters)per km per year,being much greater than the standard acceptable failure probability of10K6(Taylor, 1994).It is worth mentioning that a recent work of Yuhua and Datao(2005)based on Fault Tree AnalysisconcludesJournal of Loss Prevention in the Process Industries19(2006)24–31/locate/jlp0950-4230/$-see front matter q2005Elsevier Ltd.All rights reserved.doi:10.1016/j.jlp.2005.05.002*Corresponding author.Address:School of Chemical Engineering,National Technical University of Athens,15780Athens,Greece.Tel.:C302107723267;fax:C302107723163.E-mail address:rigasf@central.ntua.gr(F.Rigas).that an approximate failure probability of a transmission pipeline is6.25!10K2.Safety distances estimated in this work are based on real data given by Greek Refineries and the Greek Public Gas Enterprise concerning two pipeline systems transmitting liquefied petroleum and natural gas,respectively.In order to indicate the plausible accident outcomes of an accidental gas release,the Event Tree Analysis method(CCPS,1992) was applied.Eventually,jetfire event and gas dispersion to the lowerflammable limit were considered purposeful for further examination.Aiming at the investigation of atmospheric conditions effect on probable outcomes,weather conditions were also taken into account in the computations considering the Pasquill–Gifford stability categories(CCPS,1995c).Since pipe diameter and internal pressure are the major factors that determine the magnitude of a potential release(Bartenev, Gelfand,Makhviladge,&Roberts,1996),these parameters were assumed as independent variables in the construction of diagrams suitable for providing direct safety distance putational results for the thermal radiation resulting from a jetfire,as well as for the Lower Flammable Limit(LFL)distance resulting from gas dispersion,would clarify the role of atmospheric conditions(wind speed and solar radiation)in the accident evolution.2.Historical survey of petroleum and natural gas pipeline accidentsSevere accidents associated with NG and PG trans-mission systems have occurred in the past.The main causes initiating a pipeline accidental event may be classified in five categories(Papadakis,1999):†External interference or third party activity†Corrosion†Construction defect and mechanical or material failure †Ground movement or generally natural hazards†Other or unknown causesIt is worth to mention that recent statistical data(OPS, 2005)reveal that during the last20years,natural pipeline accident rates still remain in the same level,in spite of increased safety measures and advanced safety systems applied in practice.Typical examples of such accidents with catastrophic effects on people and property are quoted below: 2.1.NG pipeline explosion andfireOn March241994,at New Jersey,USA,an explosion of an underground natural gas pipeline was followed by a crater of approximately50m diameter and massiveflames which could be seen more than80km away.The accident resulted in1death and50injuries.Subsequent investi-gations revealed that the pipeline had been damaged by excavation works.Probably,a mechanically induced crack grew to size as a result of enhanced fatigue leading to material failure.2.2.NG pipeline explosionOn September281993,at Las Tejerias,Venezuela,an explosion of a natural gas pipeline occurred underneath a highway.The line ruptured while a state telephone company was installingfiber optic cables.The result was40injured people and50dead.2.3.PG pipeline explosionOne of the most severe chemical accidents that ever happened took place on June41989,at Siberia,Russia.A PG pipeline was commissioned in1985to carry mixed PG (propane,butane,pentane,methane and ethane)to feed an industrial city.Subsequently,it was reported that there had been a leakage for several days and that a heavy smell of gas had been reported a few hours before the explosions and fire.Instead of investigating the complaints,the responsible engineers had responded by increasing the pumping rate in order to maintain the required pressure in the pipeline.The leakage point was found about1/2mile away from the side of a railway.The smell of escaping gas was reported from valley habitants in the area and it is also stated that the escaping liquefied gas formed large pockets in low lying areas along the railway line.The gas cloud is reported to have drifted a distance of5miles away.Some hours later, two passenger trains traveling in opposite directions, approached the area.The imminent turbulence due to their motion mixed up PG mist and vapor with the overlying air to form aflammable cloud mixture.One train sparked off the cloud causing an initial explosion.Two explosions took place in quick succession followed by a wall offire that was about1mile wide.A considerable part of each train was derailed,while trees wereflattened and windows were broken within a radius of2.5and8miles,respectively. Totally462people died and706were injured.3.Methodology3.1.Event Tree AnalysisEvent Tree Analysis(ETA)used in this work is a formal technique and one of the standard approaches when performing industrial incidents investigation as well as pipeline risk assessment(Muhlbauer,1996).ETA is a logic sequence that graphically portrays the combination of events and circumstances in an accident sequence.It is an inductive method,which begins with an initiating undesir-able event and works towards afinal result(outcome);each branch of the Event Tree represents a separate accidentS.Sklavounos,F.Rigas/Journal of Loss Prevention in the Process Industries19(2006)24–3125sequence (CCPS,1992).The general procedure for ETA includes the following steps:1.Determination of the initiating events that can result in certain types of accidents.2.Identification of the critical factors that may affect the initiating event evolution.3.Construction of the Event Tree taking into account the interaction between critical factors and the initiating event.4.Designation and evaluation of resulting accidental events.Generally,the ETA is very useful for providing scenarios on possible failure modes,while the final outcomes can be ranked on the basis of their severity by the use of consequence models.3.2.Worst case scenariosLarge amounts of flammable gases released in the open-air pose a significant hazard for the surroundings,due to their ability to yield disastrous fires and explosions.In this work,in which natural and liquefied petroleum gas pipelines were the point of interest,the purpose of safety distance estimation was achieved following the simplified method of Fig.1.The relevant Event Tree (Fig.2)was constructed to indicate all possible outcomes of an accidental fuel gas release (CCPS,1995a )on the basis of the main critical factors that affect substantially the accident evolution:the time of ignition of the resulting cloud and the degree of confinement provided by the surroundings.The former is related to the mixing of escaping fuel gas with the air.When immediate ignition occurs,gas cloud mixing with atmospheric oxygen is still limited;thus,the ignition takes place on the outer layer,which is between the flammable limits,whereas the inner core of the cloud istooFig.1.Logic diagram for safety distancedetermination.Fig.2.Event Tree Analysis adapted to accidental fuel gas release.S.Sklavounos,F.Rigas /Journal of Loss Prevention in the Process Industries 19(2006)24–3126rich in fuel to ignite.As buoyancy forces of the hot gases begin to dominate,the burning cloud rises and becomes more spherical in shape forming a ball intoflames.That elevation causes gradually further mixing of the gas with oxygen,which brings new volumes of gas intoflammable limits sustaining thefire.Fireballs of this kind have been recorded to travel hundreds of meters until fully burned.On the contrary,when delayed ignition occurs,the fuel cloud can be adequately mixed with air,so that after ignition it flashes back.It differs fromfireball since it proceeds faster and can burn from inner to outerflammable layers provided that a proper ignition source is found there.If sufficiently mixed with air,fast burning occurs initially with a medium increase of pressure.This subsonic burning is known as deflagration and is possible when the fuel–air mixture is withinflammability limits,yet far from stoichiometry.If considerable confinement exists and in addition the oxygen content within the cloud is around the Zero Oxygen Balance(stoichiometric fuel–air mixture),the flame propagation speed increases rapidly producing a blast wave(Phillips,1994).In this case,theflame front propagates at a supersonic velocity and a strong shock wave develops in the cloud,which is characterized by an abrupt high overpressure front.The general term for explosions in which a shockwave develops is called detonation.Especially for gaseous mixtures exploding in confined spaces the term Confined Vapor Cloud Explosion (CVCE)is used,in which either a detonation or a deflagration takes place.In a very poor or very rich fuel mixture,but one that is still withinflammable limits,the flame front travels the cloud in low velocity and insignificant pressure increase,a phenomenon known asflashfire.Consequently,buildings inside the LFL radius could be severely damaged due to the potential of a confined vapor cloud explosion.Moreover,if the escaping gas is not trapped and immediate ignition occurs,then a long-lasting jetfire would pose the most significant hazard,due to the high thermal radiation levels(Jo&Ahn,2002).These two outcomes were considered as the dominant hazards for a pipeline fuel gas release event and reckoned as requiring further investigation.putational toolsHydrocarbons are practically non-toxic;even for chronic occupational exposure,systemic effects have not been reported(Hathaway,Proctor,&Hughes,1996).Therefore, in the subsequent investigation the gas concentration prediction during cloud dispersion is limited in gas concentration above the Lower Flammable Limit(LFL). The distance from the point of incident at which concentration value drops below the LFL(LFL distance) is a limiting distance,beyond whichfire or explosion hazard is eliminated.Consequently,a model is needed for predicting gas dispersion,in combination with a model suitable for performing source-term analysis.Furthermore, a quantitative model for calculating thermal radiation effect of jetfire is required.Petroleum and natural gas are not particular compounds but mixtures constituted from several chemical molecules. On the other hand,consequence modeling computations demand the knowledge of some thermochemical properties of the mixtures.For this purpose,another computer program capable of calculating gas mixtures properties was utilized. Each one of the above models is described below.4.1.CHETAH programCHETAH is a computing tool developed by ASTM for predicting both thermochemical properties and certain reactive substance hazards associated with a pure chemical, a mixture of chemicals,or a chemical reaction.This is accomplished through the knowledge of just the molecular structures of the components and mixture composition. CHETAH is useful for classifying materials for their ability to decompose with violence,estimating heats of combustion (used infire models),calculating LFLs(used in dispersion models)and providing hazard potential classification (CCPS,1995b;Shanley&Melhem,1995).4.2.EXPERT modelDispersion models require data related to the source of the release.Such data are the temperature and density at the orifice area and the release rate with which a gas or aerosol escapes into the atmosphere.The EXPERT model calculates these source-term parameters based on user-specified chemical property data of the escaping substance,type of storage equipment,storage conditions and opening size (Grosch&Miller,1998).4.3.SLAB and AFTOX modelSLAB computer model simulates the dispersion of denser-than-air releases.Therefore,it was used for LFL distance calculation resulting from an LPG release,in addition to high pressure(50bar)natural gas release,in which gas density was calculated to be significantly higher (1.75kg/m3)than that of air(1.19kg/m3).It is a slab model with properties averaged in the horizontal and vertical directions thus being a one-dimensional model.The model is based on a set of six differential equations,for the conservation of total mass,material released mass, momentum and energy(Lees,1996).It does not take into account the presence of obstacles in theflowfield assuming a non-sloping unobstructed terrain.AFTOX model is a Gaussian puff model for uniform terrain and wind conditions.It has been designed to simulate lighter-than-air releases and it assumes a non-sloping unobstructed dispersion terrain,while it was usedS.Sklavounos,F.Rigas/Journal of Loss Prevention in the Process Industries19(2006)24–3127for the LFL distance estimation resulting from intermedi-ate(19bar)and low(4bar)natural gas release.4.4.JET FLAME modelThe JET FLAME radiation model predicts the radiation flux of gaseous jetflames resulting from pressurized pipeline accidents,in which the escapingflammable gas is ignited.Thus,it calculates the distance in which a particular thermal radiation intensity value is developed.The model assumes that theflame shape is cylindrical,while it takes into account the internal pipe conditions(pressure and temperature),in addition to atmospheric conditions(CCPS,1995c).EXPERT,JET FLAME,SLAB and AFTOX models are included in BREEZE HAZARD software package,which has been effectively put in practice in previous consequence analysis procedures(Rigas&Sklavounos,2002),while SLAB and AFTOX have been approved by US EPA’s Risk management program for Offsite Consequence Analysis Guidance.However,all of them are based on simplified assumptions that strongly affect the results.Thus,SLAB model computes mass,momentum and energy balances only in one direction(downwind)and assumesflat terrain with no obstructions or slope.Likewise,JET FLAME model does not take into account obstacles intervening between flames and target that actually decrease the thermal radiation impact.Consequently,the results presented hereinafter should be considered conservative.5.Material properties computationEstimations of thermochemical properties of a gas mixture like heat of combustion and physical quantities such as the lowerflammable limit provide useful input data for the operation of consequence models.The above requirements were obtained by using the CHETAH program described previously.In Table1,typical compositions of petroleum and natural gas are given.Since the ratio of secondary components in NG mixture is very low(approximately2%),NG physical properties may be identified with the properties of methane. In contrast,PG mixture consists of more than one component in considerable concentrations(propane and butane)that cannot be merged.Providing the appropriate data of the mixture molar composition,CHETAH code calculated PG combustion heat and lowerflammable limit as shown in Table2.6.Safety distance estimations6.1.Input variablesNominal and operating conditions,as well as the sizes of LPG and NG pipelines studied in this work are shown in Table3.The primary goal of hazard zones determination and hence safety distances assessment is accomplished using consequence analysis models for the worst case scenarios mentioned in Section3.2.Regarding the input data introduced in these models,it is of vital importance to highlight the following:†The wind and particularly its turbulent nature plus solar radiation are significant factors in gas dispersion.The vigorous whirling molecular motion dominating in the atmospheric environment results in gas agitation that may bring faster the cloud into theflammable range (Deaves,1992).Moreover,the shape and size of the jet fire(and hence the thermal radiation emitted)is strongly affected by the wind speed(Arnaldos,Casal,Montiel, Sanchez-Carricondo,&Vilchez,1998).Therefore, atmospheric conditions were taken into account in the computations in the form of the so-called Pasquill–Gifford stability categories.Three discernible classes were reckoned to be significant:A—unstable conditionsTable1Typical composition of natural and petroleum gasGas mixture Components Content(%v.v.)NG(Russian natural gas)Methane98 Ethane0.6 Propane0.2 Butane0.2 Pentane and heavier hydrocarbons0.1 Nitrogen0.8 Carbon dioxide0.1LPG(refinery product)Ethane0.8Propane20.1Butane78.6Pentane and heavierhydrocarbons0.5Table2Petroleum gas combustion heat and LFL calculated with CHETAH codeLPG property Calculated value Input dataCombustion heat45826kJ/kg JET FLAME(jetfiremodel)Lowerflammablelimit1.78%v/v SLAB and AFTOX(dispersion models)Table3Data on operating conditions and internal diameters of LPG and NGpipelinesFuelgasPipeline system Pressure(bar)Diameter(cm)Nominal OperatingNG Internationaltransportation705025–91Interstate transportation191924Regional distribution449–16LPG Regional distribution312436 S.Sklavounos,F.Rigas/Journal of Loss Prevention in the Process Industries19(2006)24–3128attributed to strong solar radiation and low wind speed;D—neutral conditions identified with moderate to strong winds with significant cloud cover and hence low solar radiation;F—stable conditions,corresponding to rela-tively light winds and nights of clear sky (no solar radiation).†For the purpose of safety distance determination concerning the jet fire scenario,maximal permissible levels for thermal radiation intensity must be set.Among relative data (CCPS,1994;Green Book,1989)the more conservative values were adopted (Table 4).With reference to cloud dispersion scenario,the LFL distance was determined.6.2.Results and discussionThe burning of a flammable gas issuing from a pipe or other orifice at the point of exit forms a jet fire.In fact,release sizes range from small leaks to the full loss of contents giving a wide range of gas release rates,which go up to some hundred of kilogram/second with regard to high pressure pipelines (Olorunmaiye &Imide,1993).Initially,release magnitudes were studied for the range of small cracks to full-bore pipe rupture.Typical example demon-strating safety distance change with crack size in different weather conditions (A,F,and D)for the high pressure (50bar)NG pipeline is presented in Fig.3.It is obvious that thermal radiation impact is mainly affected by the wind speed:the lower the wind speed,the greater the radius ofrisk.The conclusion still remains the same for the LPG pipeline (Fig.4),namely the safety distance is maximized for the air stability category A.In contrast,LFL distance is most affected by solar radiation (Fig.5).In particular,the lower the solar radiation,the greater the LFL distance.This conclusion may be qualitatively drawn from Fig.6,in which safety distance has been plotted against wind speed and pipe diameter.It is evident that safety distance is maximized for intermediate wind speed,namely for atmospheric stability category F that corresponds to complete absence of solar radiation.In Fig.7a and b,where jet fire and dispersion scenarios are compared with regard to the resulting safety distances,it is clear that jet fire scenario entails higher safety distances than that of dispersion scenario.Indeed,safety distances are higher when related to human effect rather than to property damage.The latter was expected,since the threshold value for human injury is lower than that of property damage (Table 4).As a result,the jet fire scenario should be considered as the limiter for safety distance determination.The utilization of the results obtained through the jet fire safety distance calculation (considering atmospheric stab-ility category A,full-bore rupture and the pipe diameters and pressures shown in Table 3)yielded the diagramsTable 4Thermal radiation threshold limit values for acceptable damage on people and property Damaged entity Thermal radiation critical value (kW/m 2)Exposure duration Foreseen effectProperty4Not too short(R 30min)Rupture of glassPeople 1.560s1%probability of 1st degreeburnsFig.3.Safety distance for the 0.91m and 50bar NG pipeline related to the jet fire scenario and atmospheric stability categories A,F and D,calculated for small releases to full-borerupture.Fig.4.Safety distance for the LPG pipeline related to the jet fire scenario and atmospheric stability categories A,F and D,calculated for small releases to full-borerupture.Fig.5.Safety distance for the 0.91m and 50bar NG pipeline related to the cloud dispersion scenario and atmospheric stability categories A,F and D,calculated for small releases to full-bore rupture.S.Sklavounos,F.Rigas /Journal of Loss Prevention in the Process Industries 19(2006)24–3129illustrated in Figs.8and 9.Thus,from these Figures,safety distance may be directly determined,when internal pipe pressure and diameter are known.As one can see,safety distance increases when the internal diameter or the operating pressure of the pipeline increases.Nevertheless,safety distance seems to be more sensitive with pipe size rather than operating pressure.As an example,let us consider an internal pipe diameter equal to 60cm (Fig.8).Then for the operating pressure range from 20to 70bar,the safety distance varies between 450and 750m,namely a range of 300m.In contrast,considering a pressure value equal to 50bar,the safety distance variesbetween 325and 925m when the diameter changes from 25to 91cm,namely a range of 600m.This conclusion is to be found in agreement with that of Jo and Ahn (2002).7.ConclusionsFlammable gas transmission pipelines pose major risk for the surroundings,due to the probability of rupture and release of the content to the atmosphere with the potential of a fire or/and explosion.In this work,a methodology for estimating safety distances in the vicinity of petroleum and natural gas pipelines is proposed,with the perspective to find application in emergency response planning.In summary,the following remarks may bereferred:Fig.6.Safety distance plot against wind speed and pipe diameter for the 0.91m and 50bar NG pipeline according to cloud dispersion scenario for full-bore piperupture.Fig.7.Safety distance for both jet fire and dispersion scenarios assuming full-bore rupture of 50bar (a)and 4bar (b)NG pipelines.The stability categories adopted were:A for the thermal radiation safety distance and F for the LFLdistance.Fig.8.Safety distance isopleths (m)against operating pressure and pipe diameter for the 50bar pipeline,assuming atmospheric stability category A,1.5kW/m 2thermal radiation intensity and full-bore piperupture.Fig.9.Safety distance isopleths (m)against operating pressure and pipe diameter for the 4bar pipeline,assuming atmospheric stability category A,1.5kW/m 2thermal radiation intensity and full-bore pipe rupture.S.Sklavounos,F.Rigas /Journal of Loss Prevention in the Process Industries 19(2006)24–3130†The jetfire scenario should be considered as the limiter for safety distances determination in the vicinity of natural and petroleum gas pipelines.†Safety distances are more sensitive to pipeline size than operating pressure.†The thermal effect of a jetfire event,as well as the distance that fuel gas travels from source to its lower flammable limit position,strongly depends on atmos-pheric conditions.Indeed,different weather conditions favor each type of accident:the impact radius thermal radiation is mainly affected by wind speed and increases when wind speed decreases,whereas gas dispersion in mainly affected by solar radiation and LFL distance increases when solar radiation decreases.†Safety distances in the vicinity of fuel gas pipelines may be plotted in diagrams against independent variables.These diagrams could be used in loss prevention applications(i.e.ignition sources elimination within the LFL distance in industrial sites),as well as in safer land-use planning.AcknowledgementsThis work was supported by the Ministry of National Education and Religious Affairs(Community Support Framework2000–2006)under the HERACLITUS research program.ReferencesArnaldos,J.,Casal,J.,Montiel,H.,Sanchez-Carricondo,M.,&Vilchez, J.A.(1998).Design of a computer tool for the evaluation of the consequences of accidental natural gas releases in distribution pipes.Journal of Loss Prevention in the Process Industries,11(2),135–148. Bartenev,A.M.,Gelfand,B.E.,Makhviladge,G.M.,&Roberts,J.P.(1996).Statistical analysis of accidents on the Middle Asia—Centre gas pipelines.Journal of Hazardous Materials,46(1),57–69.CCPS(1992).Guidlines for hazard evaluation procedures.New York: AIChE.CCPS(1994).Guidelines for evaluating the characteristics of vapor cloud explosions,flashfires and BLEVEs.New York:AIChE.CCPS(1995a).Chemical transportation risk analysis.New York:AIChE. CCPS(1995b).Guideliness for safe storage and handling of reactive materials.New York:PS(1995c).Undetstanding atmospheric dispersion of accidental releases.New York:AIChE.Deaves,D.M.(1992).Dense gas dispersion modelling.Journal of Loss Prevention in the Process Industries,5(4),219–227.DeWolf,G.B.(2003).Process safety management in the pipeline industry: Parallels and differences between the pipeline integrity management (IMP)rule of the office of pipeline safety and the PSM/RMP approach for process facilities.Journal of Hazardous Materials,104(1–3),169–192.EEC(1996a).Concerning integrated pollution and control.European Council Directive96/61,Brussels.EEC(1996b).On the control of major-accident hazards involving dangerous substances.European Council Directive96/82,Brussels. Green Book(1989).Methods for the determination of possible damage-CPR16E(2nd ed.).The Netherlands:TNO.Grosch,T.G.,&Miller,M.D.(1998).An expert system for source-term analysis and accidental release modelling.Proceedings of the91st Air and Waste Management Association Annual Meeting,San Diego. Hathaway,G.J.,Proctor,N.H.,&Hughes,P.J.(1996).Chemical hazards for the workplace(4th ed.).New York:Van Nostrand Reinhold. IChemE(2000).The accident database—CD form.Loughborough: IChemE.Jo,Y.D.,&Ahn,B.J.(2002).Analysis of hazard areas associated with high-pressure natural gas pipelines.Journal of Loss Prevention in the Process Industries,15(3),179–188.Khan,F.I.,&Abbasi,S.A.(1999).Major accidents in process industries and an analysis of causes and consequences.Journal of Loss Prevention in the Process Industries,12(5),361–378.Lees,F.P.(1996).Loss prevention in the process industries(2nd ed.).Oxford:Butterworth-Heinemann.Muhlbauer,W.K.(1996).Pipeline risk management manual(2nd ed.).Houston:Gulf.Olorunmaiye,J.A.,&Imide,N.E.(1993).Computation of natural gas pipeline rupture problems using the method of characteristics.Journal of Hazardous Materials,34(1),81–98.OPS(2005).Natural gas incident yearly summaries(1986–2004).US Department of Transportation,Office of Pipeline Safety,Washington. Papadakis,A.G.(1999).Major hazard pipelines:A comparative study of onshore transmission accidents.Journal of Loss Prevention in the Process Industries,12(1),91–107.Phillips,H.(1994).Explosions in the process industries(2nd ed.).Warwickshire:IChemE.Rigas,F.,&Sklavounos,S.(2002).Risk and consequence analyses of hazardous chemicals in marshalling yards and warehouses at Ikonio/Piraeus harbour,Greece.Journal of Loss Prevention in the Process Industries,15(6),531–544.Shanley,E.S.,&Melhem,G.A.(1995).A review of ASTM CHETAH7.0 hazard evaluation criteria.Journal of Loss Prevention in the Process Industries,8(5),261–264.Taylor,J.R.(1994).Risk analysis for process plants,pipelines and transport.London:E&FN Spon.Yuhua,D.,&Datao,Y.(2005).Estimation of failure probability of oil and gas transmission pipelines by fuzzy fault tree analysis.Journal of Loss Prevention in the Process Industries,18(2),83–88.S.Sklavounos,F.Rigas/Journal of Loss Prevention in the Process Industries19(2006)24–3131。