Vapor Cloud Explosion AnalysisQuentin A.Baker3EHS.com 3EHS.com3EHS.comWilfred Baker Engineering Inc.,8700 Crownhill Blvd.,San Antonio,TX 78209Ming Jun TangNanjing University of Science and Technology.Nanjing,ChinaEphraim A.ScheierOccidental Chemical Corp.,5005 LBJ Freeway,Dallas,TX 752443EHS.comOccidental International Exploration Production Co.,1200 Discovery Drive,Bakersfield CA 93389This paper introduces a new metbod (now commonlywas to provide an objective and well founded methodol-referred to as the Baker-Streblow Metbod)for estimatingogy for determination of these two terms.pressure and impulse generated by vapor cloud explo-The following approach was taken to develop the newsions.Streblow's blast curves and concepts from theVCE prediction methodology.ergy are applied in this tecbnique.New correlations for1.Review published VCE prediction methods and select amaximum flame speed based on obstacle density,fuelmethod based on technical merits and suitability to obreactivity,and cloud confinement allow selection of tbejective decision making.appropriate blast curve.Application of these correla-Review published experimental data and develop corre-tions removes much of the subjectivity present in exist-lations and procedures for the determination of explo-ing explosion estimates.sion energy and reaction rate.3.Conduct case studies to evaluate the new methodologyand to provide examples of its implementation.The development of the new methodology is summa-rized in the following sections.The work described in this paper was undertaken out ofthe need for an objective method to predict airblast param-eters for vapor cloud explosions(VCE).It was recognizedReview and Selection of VCE Prediction Methodologythat the primary weakness of VCE predictions was the sub-This study did not attempt to develop new blast curvesjectivity of assumptions and judgements that the user wasfor VCEs;rather,the purpose was to develop objectiveforced to make.Among the most commonly employed VCEmethodologies for the implementation of presently avail-prediction methods are TNT equivalence,the Multi-Energyable blast curves.However,the selection of the blast curvesmethod,and Strehlow's spherical model.All three methodswas a critical portion of the program since the blast curvesrequire an estimate of the energy of the explosion,and onlyare the heart of the prediction methodology.It was es-the Multi-Energy method provides a rational basis for deci-sential that the blast curves be technically sound,wellsions concerning the energy term.The Multi-Energy andpublished and peer reviewed.Further,the input data forStrehlow methods require the user to select from amonguse of the blast curves had to lend themselves to objectivefamilies of blast curves based upon the potential severitydetermination.or flame speed of the explosion.No guidance,however,isLiterature was gathered and reviewed concerning VCEprovided in either method for the selection of the appro-prediction methods.Table 1 provides a listing of the meth-priate curve.Considerable disparities in the prediction ofods that were studied.Interested readers are referred toblast parameters stem from judgements required for theReferences [1]through [10]for the technical basis of eachexplosion energy and reaction rate terms(severity factor forof the methods listed.the Multi-Energy method,and flame speed for the StrehlowThe list was narrowed to two primary candidates:themethod).A primary objective of the work described herespherical model by Strehlow,and the TNO Multi-EnergySummer,1996Process Safety Progress (Vol.15,No.2)The combined method still suffered from lack of guid-TABLE 1.Prediction Methods of Blasts from VCEsance for selection of flame speed and determination of par-Empirical MethodTNT Equivalence Method [tial confinement.These deficiencies were resolved byShell TRC Method [2]developing correlations based on empirical data as will bediscussed in the section on flame speed.Self-Similar MethodExact Solution by Kuhl et al.B3]Figure 1 presents Strehlow's blast curves for scaled side.(60's and 70's)Simplification by Strehlow [4on overpressure P/po,where P:is overpressure (gaugeApproximation by Guirao et al.[5]pressure),while po is atmospheric pressure.The abscissaSpherical Modelof Figures 1 and 2 is scaled standoff distanceCalculationby Strehlow et al.[6,7]where R is the distance from the cloud center,and E is the(late 80's and 90's)TNO Hemispherical Model [8]explosion energy.The curves in both Figures I and 2 areTNO Multi-Energy Method [9]labeled with flame velocities,expressed in Mach number.2D NumericalPancake and HemiellipsoidFigure 2 presents scaled side-on specific impulse,Calculation(80's and 90's)bient sound speed.Figures 1 and 2 use Sach's scaling(sometimes called Energy scaling),which is dimensionless.Figures 1 and 2 are for free-air spherical VCEs.Explosionat or near ground level will have a strong ground reflec-method.Unfortunately,neither method was found to en-tion,which is taken into account by doubling the energy.tirely satisfy the requirements of this study.The approachadopted was a combination of the Strehlow and Multi-Flame SpeedEnergy methods.Strehlow's spherical model was chosenLiterature was reviewed during this study to compile ex-because blast curves are selected based on flame speed,perimental data as the basis for an objective determinationwhich affords the opportunity to use empirical data in theof flame speed M to be used in selecting a blast curveselection.Also,the availability of impulse curves weighedfrom Figures 1 and 2.The review of the data suggested thatin favor of Strehlow's spherical model.The procedures fromthe combined effects of fuel reactivity,obstacle density,andthe Multi-Energy method were adopted for determinationconfinement can be correlated to the flame speed.of the energy term.Specifically,confinement is the basis ofThe major factors addressed in experimental studies werethe determination for the size of the flammable vapor cloudthe effects of confinement,obstacles and fuel reactivity onthat contributes to the generation of blast overpressure,andflame acceleration and the subsequent pressure build-up.multiple blast sources can emanate from a single release.Fuel reactivity and obstacle density influence the reactionrate10PentoliteMw5.2Mw-4.0Sphere Burstm.3EHS.comPentoliteBursting Spl0Mw-0.125Msu0.0343ENSMw=4.0Mw0.0742=2.0Mw3EHSMsu-0.21Mw-0.037w10R*(Po/E)1/3R*(Po/E)I/3FIGURE 1Scaled side-on pressure curves for sphericaFIGURE 2Scaled side-on specific impulse curves forVCEs.spherical VCEs.Safety Progress (Vol.15.No.2)