prugh model

7/24/2019 prugh model

1/17

http://jfe.sagepub.com/Engineering

Journal of Fire Protection

http://jfe.sagepub.com/content/3/1/9The online version of this article can be found at:

DOI: 10.1177/104239159100300102

1991 3: 9Journal of Fire Protection EngineeringRichard W. Prugh

Quantitative Evaluation of "Bleve" Hazards

Published by:

http://www.sagepublications.com

On behalf of:

Society of Fire Protection Engineers

can be found at:Journal of Fire Protection EngineeringAdditional services and information for

http://jfe.sagepub.com/cgi/alertsEmail Alerts:

http://jfe.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

What is This?

- Jan 1, 1991Version of Record>>

at ILLINOIS INST OF TECHNOLOGY on October 31, 2014jfe.sagepub.comDownloaded from at ILLINOIS INST OF TECHNOLOGY on October 31, 2014jfe.sagepub.comDownloaded from
http://jfe.sagepub.com/http://jfe.sagepub.com/http://jfe.sagepub.com/http://jfe.sagepub.com/content/3/1/9http://jfe.sagepub.com/content/3/1/9http://www.sagepublications.com/http://www.sagepublications.com/http://www.sfpe.org/http://jfe.sagepub.com/cgi/alertshttp://jfe.sagepub.com/cgi/alertshttp://jfe.sagepub.com/subscriptionshttp://www.sagepub.com/journalsReprints.navhttp://www.sagepub.com/journalsReprints.navhttp://www.sagepub.com/journalsPermissions.navhttp://www.sagepub.com/journalsPermissions.navhttp://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://jfe.sagepub.com/content/3/1/9.full.pdfhttp://jfe.sagepub.com/content/3/1/9.full.pdfhttp://jfe.sagepub.com/http://jfe.sagepub.com/http://jfe.sagepub.com/http://jfe.sagepub.com/http://jfe.sagepub.com/http://jfe.sagepub.com/http://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://jfe.sagepub.com/content/3/1/9.full.pdfhttp://www.sagepub.com/journalsPermissions.navhttp://www.sagepub.com/journalsReprints.navhttp://jfe.sagepub.com/subscriptionshttp://jfe.sagepub.com/cgi/alertshttp://www.sfpe.org/http://www.sagepublications.com/http://jfe.sagepub.com/content/3/1/9http://jfe.sagepub.com/


2/17

9

QUANTITATIVE EVALUATION OF "BLEVE" HAZARDS

by

Richard W.

PrughPresident, Hazard Reduction Engineering, Inc.Wilmington, DE*

SUMMARY

Rupture of a container in which a liquid is held above its atmospheric-pressure boilingpoint can result in explosive vaporization of a large fraction of the contents. This phe-nomenon is called a Boiling-Liquid Expanding-Vapor Explosion ("BLEVE").A method forestimating the TNT equivalence of a BLEVE is proposed, with descriptions of possiblecauses and consequences of the explosion and (if the contents are flammable) the con-

sequences of the resulting fireball.

I NTRODUCTIONA Boiling-Liquid Expanding-Vapor Explosion(BLEVE) is an unusual physical phenomenonwhich results from the sudden release from con-

finement of a liquid at a temperature above itsatmospheric pressure boiling pointl. Thesudden decrease in pressure results in explosivevaporization of a fraction of the liquid and acloud of vapor and mist, with accompanyingblast effects. If the material is flammable andan

ignitionsource is

present,a fireball

maybe a

further hazardous consequence of the BLEVE.

The potential for BLEVE creates a hazard forthe fire fighter and presents unusual designconsiderations for the fire protection engineer.The fire fighting hazard results from the blast,fireball, and missiles which can accompany pro-longed fire exposure of a tank of flammableliquid. The engineering aspects include protec-tion of flammable liquid and liquified gas tanksfrom

fire and mechanical damage, design ofremote-operated fire fighting facilities, and pro-tection of fire-water systems from the blast andmissiles.

The objectives of this paper are to:

1. Show how BLEVEs differ from other typesof explosions and explain the unexpectedseverity of the consequences;

* Address for correspondence: 501 Silverside Road,site 122,

Wilmington,DE 19809. Portions of this

paper were published as &dquo;Quantify BLEVEHazards&dquo; (Chemical Engineering Progress, 87, No.2, pp. 66-72, February 1991) and are reproducedwith permission of AIChe. 1991AIChe.

2. Discuss the history of BLEVEs, anddescribe the important causes;

3. Present a method for estimating the energypotential, or TNT equivalent, of a BLEVE;

4. Present methods for BLEVE prevention;and

5. Show how the causes become important

inputs into process safety analysis, particu-larly for toxic chemicals.

The broad definition of a BLEVE, as used by theNational Fire ProtectionAssociationl, Society ofFire Protection Engineers2, Lees3, and Kletz4 isused here; that is, it is assumed that any liquifiedvapor - flammable or non-flammable - can pro-duce a BLEVE.Also, a fireball is not part of this

definition, since a fireball would result only if thematerial released were flammable and ignitionoccurs. The historical fact is that most BLEVEs

involve flammable liquids, and most of theseBLEVE releases are

ignited bythe

surroundingfire, resulting in a fireballs. However, the methoddescribed in this paper for estimating the blast

effects ofBLEVEs holds for all liquids.

TANK, CONTAINER, ORPROCESS-VESSEL EXPLOSIONS

There are three types of &dquo;physical&dquo; or &dquo;boiler-

type&dquo; explosions which are of interest to the fire

protection engineer: bursting of a gas-filled con-tainer (or of a container of vapor at a tempera-ture above the critical point), bursting of a

liquid-filled container (at a temperature below

at ILLINOIS INST OF TECHNOLOGY on October 31, 2014jfe.sagepub.comDownloaded from
http://jfe.sagepub.com/http://jfe.sagepub.com/http://jfe.sagepub.com/http://jfe.sagepub.com/


3/17

10

the boiling point of the liquid), and bursting of acontainer in which the confined liquid is at a

temperature above the boiling point (and belowthe critical point). The last case is the BLEVE

scenario, but the first two cases are discussedhere for reference.

Explosion of a Gas-Filled VesselThe bursting of a gas-filled container pre-dictably results in an outrush of gas, at a highinitial pressure and with the blast pressure

decreasing with distance from the burst con-tainer. The energy is usually calculated byassuming an adiabatic expansion of the con-

tents, as the pressure decreases from the initialconfined pressure toward

atmospheric6,7.The

initial shock-wave pressure is not equal to theinitial confined pressure but can be estimated

rather easily. It is also later shown that theblast effects of bursting gas-filled containers canbe assessed from the calculated TNT equiva-lence and a &dquo;virtual distance&dquo; from an &dquo;explo-sion center&dquo;.

Explosion of a Vessel Containing Cold

SquidThe bursting of a liquid-filled container pro-duces blast effects which are orders of magni-tude smaller than those of gas-filled containers,if the temperature of the liquid is below theatmospheric-pressure boiling point8. The energypotential is that of a &dquo;spring&dquo; which has beenexpanded by the internal pressure, and this

energy appears as a shock wave propelled bythe escaping liquid. Failure of a vessel beingsubjected to a hydrostatic test is illustrative ofthis type of &dquo;explosion&dquo;.

Explosion of a Vessel ContainingSuperheated Liquid (BLEVE)Based on experience with cold liquids, it mightbe expected that the blast energy available froma container almost completely filled with liquid- as in a railroad tank car - could be estimated

by only considering the vapor space as being agas-filled container. However, experience hasdemonstrated that the energy of a bursting con-

tainer of superheated liquid has far greaterenergy than would be estimated from such a

vapor-space calculation. Sections of essentially-

full tank cars have been propelled to distances

up to 400 meters, indicating that there must be

another important source of energy in a

BLEVE. It will be demonstrated that this

energy source is that of explosively-flashingsuperheated liquid.

BLEVE HISTORYAND INCIDENTS

(49 INCIDENTS)BLEVEs Caused by Exposure to Fire

(17 incidents)Certainly the most important cause of BLEVEsis fire. The typical incident scenario is that of

fire surroundinga container of

liquid,with heat

input to the container resulting in an increasein liquid temperature and an increase in inter-

nal pressure. If the container is fitted with a

relief valve, the valve will open when the inter-

nal pressure reaches the relief-valve setting,and some boiling will occur while the reliefvalve is open. If the relief valve is large enoughand the setting is sufficiently below the burst

pressure of the container, all of the liquid mayboil out of the container without further serious

aggravation of the fire situation (asin a

pres-sure cooker).

However, in many BLEVE situations, the settingof the relief valve is quite high (particularly for

liquefied gases).Also, the tensile strength of thecontainer material is typically reduced by expo-sure to fire, such that the container may not be

capable of withstanding pressure equal to therelief-valve setting. Thus, even though the relief

valve is open and venting vapor at a rapid rate

(frequently with the vented vapors afire), thecon-

tainer may burst. Table I presents brief descrip-tions of seventeen accidents in which fire caused

overpressure and subsequent BLEVE. The mate-

rials involved were: butane 9~1o ethylene oxidell,

gasoline 12-14 hydrogenls, propane~~-24~ and

vinyl chloride17,29.

An important fire-exposure test was conducted bythe U. S. Department of Transportation in

197426.A BLEVE occurred in an uninsulated

125-cubic-meter tank car containing about 6,000kilograms of liquified propane, after 24.5 minutesof exposure to a JP-4 jet-fuel fire. Rupture



4/17

11

occurred at an internal pressure of about 155

kiloPascals (kPa), and a bulk temperature near69C, despite the venting of vapor through a 39-kPa, 8.1 cm-diameter, relief valve at rates up to33 kg per second for 16 minutes and before a very

large relief device could be opened to prevent therupture.At the moment of rupture, about 34,000kg of liquified propane remained in the tank car.[The average heat flux was 105 kilowatts persquare meter of wetted surface.] This tank-car sit-uation is used as an example for the BLEVE cal-culations below.

BLEVEs Caused by Mechanical Damage(11 incidents)

Another important cause of BLEVE is damageto the liquefied-vapor container. Container fail-ure can be caused by corrosion, as occurred in a

fire-extinguisher incident27, or by impact, asoccurred during movement of a tank car atWaverly28. In the latter incident, the tank carruptured explosively as it was being moved (50hours after a derailment). The contents ignitedalmost immediately and burned as a fireball.

Collisions also have resulted in nine additional

explosive ruptures of vehicle or ship containers.These incidents involved: ammonia29,30~ butane3l,chlorine32,33, propane23,29,34, and propylene1?. Ineach of these incidents, the container was &dquo;laidopen&dquo; by the BLEVE, and parts of the containerswere propelled to great distances.

A series of tests conducted to evaluate the haz-ards of unconfined vapor cloud explosions39,3sshowed that mechanical damage to a steel tankcontaining superheated propane could cause a

BLEVE, accompanied by significant blast over-pressure (prior to ignition of the cloud). Themaximum pressure recorded was about 75 kPa

(gauge) at a distance of about 10 meters from anexploding tank containing 450 kg of propylene.

BLEVEs Caused by Overfilling(10 incidents)There may have been as many as ten incidents inwhich a BLEVE occurred without provocation

(Table I).A 15-ton chlorine tank burst in 1952 inGermany, destroying the building in which it washoused, and resulting in 7 fatalities37; an addi-

tional three explosive container ruptures haveoccurred with chlorine38, but all were more than35 years ago. Similar explosive ruptures of over-filled containers have occurred with butadiene39,butane9, carbon dioxide40, ethyl ether39, and

methyl bromide4l, and it is likely that the SanCarlos da Rapita incident - involving propylene -also was a BLEVE caused by overfilling42.

In all of the above incidents, absence of an over-

pressure-relief device was a major contributingfactor. With the improving standards of over-

pressure protection, the frequency of BLEVEscaused by overfilling could be expected todecline significantly.

BLEVEs Caused by Runaway Reaction

(6 incidents)A fourth important cause of BLEVE is runawayself-reaction of the contents, such that the con-

tainer is overpressured and the contents are

heated above the atmospheric-pressure boilingpoint. Contamination caused by back-flow of pro-cess materials has been implicated in several ofthese incidents. The materials involved were:

acrolein43,44, chlorine3l, chlorobutadiene45, ethy-

lene oxide25,46, and phosgene3l. If the materialinvolved is flammable and the temperatureattained prior to container rupture is above the

autoignition temperature, the BLEVE incidentcould be accompanied by a fireball. This may

greatly increase the likelihood of serious injury.

BLEVEs Caused by Overheating (3 inci-

dents)Excessive heat input with an inoperable relief

device can result in container rupture and thenBLEVE if the temperature of the contents is

above the boiling point. This combination of cir-cumstances apparently has occurred with

ammonia29, ethylene25, and water4?. In the lastincident, explosion of a water heater caused

great destruction of a school room and fatal

injury to seven children.

BLEVE Caused by Vapor-Space

Explosion (1 incident)Another possible cause ofBLEVE is rupture of avessel - containing a liquid at a temperature



5/17

12

above its boiling point - by explosion of gasesor vapors above the liquid. This probably was acause of at least one early chlorine BLEVE37,38,where the atmosphere above liquefied chlorinewas contaminated with hydrogen. If a chlo-

rine/hydrogen mixture were to be ignited, theresulting explosion could rupture the container,and the liquefied chlorine would then vaporizeexplosively. The resulting explosion might beconsiderably more severe than explosion of anequivalent container filled with a mixture ofchlorine and hydrogen gases.

BLEVE Caused by Mechanical Failure

(1 incident)

At least one BLEVE incident was caused byabnormal stresses in the container. In 1968, a

compartmented tank truck loaded with ammo-nia failed along a tank-divider weld line, andthe ammonia vaporized explosively as aBLEVE48.An ammonia tank failed from

mechanical stresses at Potchefstroom, South

Africa30, but the contents did not BLEVE, prob-ably because the hole in the end of the tank didnot provide the rate of depressurizationrequired for explosive vaporization27.

BLEVE BLAST-EFFECTS THEORY

Explosion of a Gas-Filled ContainerSince the energy accompanying a BLEVE usual-

ly is released unexpectedly, with consequencesmuch greater than would have been expected, amethod for estimating the blast-pressure haz-ards is needed.

An expression for calculating the energyreleased when a gas, initially having a volumeV (M3), expands in response to a decrease in

pressure from P (in kPa, absolute) to atmo-

spheric iS6,7,49,50:

where k is the specific heat ratio. To obtain theabove result in terms of kilograms of TNT, a

conversion factor of 1,000 J/kPa m3 and a heatof explosion of 4.2x106 J/kg for TNT (trinitro-toluene), are used to give:

kg of TNT

For example, the bursting of a 125 m3 tank carcontaining propane vapor (k = 1.13) at 2,400

kPa in equilibrium with a small amount of

liquid at 66C would release energy equivalentto 170 kg ofTNT:

I I I I BI

Since the tank car would contain about 7,300 kg ofvapor, this TNT equivalent would correspond toabout 0.023 kg ofTNT per kg ofpropane and about

1.4 kg of TNT per cubic meter of vessel volume.

If 10% of this energy (thus, 7 million kg-m) were to

be applied to two halves of the 22,000 kg tank-car

shell, their final positions would be about 300 m

apartsl (assuming that they would slide along the

ground with a coefficient of friction of 1.0).

Explosion of a Container of LiquefiedVaporOne approach to estimating the energy released

upon rupture of a container of liquefied vaporwould be to consider the liquid as an inert and

then base the TNT calculation on the volume of

the vapor space and the pressure in the contain-

er at the moment of rupture~1.52. This methodcannot be correct, since rupture of a container

completely filled with liquefied vapor would

then have essentially zero TNT equivalent; thisis refuted by the observed severe damage result-

ing from rupture of overfilled containers.

Another approach would be to estimate the

weight of liquefied vapor that would flash to

vapor if the pressure were to be reduced to

atmospheric, calculate the volume of vapor cor-

responding to this weight at the pressure within

the container, add this fictitious volume to the

vapor-space volume, and calculate the TNT

equivalent as above, using the total volume V*(in m3) and the container pressure P (in kPa,

absolute). Thus:



6/17

13-

where V* is given by:

where VT is the total volume of the container (incubic meters), WL is the weight of liquid in thecontainer (in kg), f is the flashing fraction, and

DL;T is the density of the liquid and DV,T is thedensity of the saturated vapor at the temperatureand pressure existing in the container at themoment of rupture (both in kg per cubic meter).Since the densities of both the liquid and vaporchange as the temperature increases (particularlyas the critical temperature is approached), densi-ty values should be taken from saturated-condi-tion thermodynamic tabless3~~.

Flashing FractionSudden release of a liquid stored at a tempera-ture above

its atmospheric-pressure boiling pointwill result in instantaneous and adiabatic vapor-ization of a fraction of the liquid3. This occursbecause a liquid at atmospheric pressure cannotremain at a temperature above the atmospheric-pressure boiling point and thus will cool to thistemperature. In doing so, the change in sensibleheat would provide the latent heat required to

vaporize a fraction of the liquid.

A simple heat balance gives an exponential

expression for the flashing fraction f 55:

In the above equation, To and Tb are the initialtemperature and the boiling point (in C or K),C is the average specific heat of the liquid (inJ/kg-K) over the temperature interval To to Tb,and L is the average latent heat of vaporization

over this temperature interval (in J/kg).

If the storage temperature is only twenty or

thirty degrees above the boiling point, sufficientaccuracy can be obtained by using an approxi-mation form of the above equation; thus:

However, neither of the above equations ade-

quately expresses the flashing fraction when the

liquid temperature is midway between the boilingpoint and the critical temperature, or higher.Atthe critical temperature, the latent heat of vapor-ization is zero and the specific heat is infinite. Useof the boiling-point values for latent and specificheats, with a &dquo;Watson&dquo; correction for latent heatand a similar relationship for specific heat, for

higher temperaturess6, yields satisfactory agree-ment with flashing fractions obtained from phasediagrams. The equation is:

In the above equation, T, is the critical temper-ature (C or K), 0.38 is obtained from using theWatson latent-heat exponent of 0.38 and a value

of 0.24 for the specific-heat exponent, and 2.63is an integration coefficient: the reciprocal of(1 - (0.38+0.24)).

Data are provided in Table II for calculation of

flashing fraction for several common materials.- To aid in such calculations, the independentparameters have been combined as a dimen-

sionless &dquo;Vaporization Factor&dquo;:

Vaporization Factor = (Cb lLb) (T~ -Tb) (9)

where Cb and Lb are the specific heat of the

liquified vapor and the latent heat of vaporiza-tion, both evaluated at the atmospheric-pres-sure boiling point (in J/kg K and J/kg, respec-tively), and Tc and Tb are the critical tempera-ture and atmospheric-pressure boiling point (inC or K), respectively.

As an example, sudden release of liquifiedpropane initially at a pressure of 2400 kPa and



7/17

14

a temperature of 66C would result in flashingof a fraction of the liquid. This flashing fractionwould be calculated as:

where 2430 (J/kg K) is the heat capacity ofliquid propane at the boiling point, 427,000(J/kg) is the latent heat of vaporization at theboiling point, and 97 and - 42 are the criticaltemperature and boiling point (in C).

As an example of the TNT-equivalent calcula-tion, explosive vaporization of 34,000 kg of liqui-fied propane in a 125 m3 tank-car rupturing at2,400 kPa (flashing fraction of 59%) would becalculated from:

where 58 is the density of liquid propane at66C and 410 is the density of propane vapor at2,400 kPa and 66C (in kg/m3).

Thus, the TNT equivalent would be:

This is about 0.015 kg of TNT per kg of propane,

about 4.1 kg of TNT per cubic meter of vesselvolume, and about three times the energyequivalent of a tank car containing propanevapor alone.A 10% energy transfer would sepa-rate the halves of the tank-car shell by a dis-tance of about 650 m (almost one-half mile).

The pertinent temperatures, densities, and spe-cific heat ratios for several common chemicalsare presented in Table II, to facilitate calcula-tion of the TNT equivalent of a BLEVE.Also

shown are the calculated TNT equivalents (interms of kg of TNT per cubic meter of liquifiedvapor) for a particular &dquo;worst case&dquo; condition

where a container becomes liquid-full at themaximum temperature to which a liquid can beheated without boiling (the &dquo;Superheat Limit

Temperature&dquo;) and then bursts as a result ofunrelieved hydrostatic pressure.

Alternative Method for Calculating TNT

EquivalentThe TNT equivalent of a BLEVE also can be

computed through use of thermodynamic tablesof specific volumes, enthalpies57, and entropies,where these are available for the material of

interest53,~4, The flashing fraction f is first com-

puted as (1) the entropy difference for the initial

and final liquid states, divided by (2) the

entropy difference for the vapor and liquidphases at atmospheric pressure5l, or:

where the entropies S (all in consistent units)are obtained for liquid L and vapor V at the ini-tial pressure P (in kPa, absolute) and at the

final pressure (101 kPa).

As an example, the flashing fraction for liquifiedpropane at a temperature of 66C, equivalent toan internal pressure of 2,400 kPa) would be cal-

culated as:

where the above entropies54 are in terms of J/kg-K For comparison, the flashing fraction calculat-

ed using the latent-heatJheat-capacity methoddescribed by Equation 10 above was 0.59.

The work done by the adiabatic expansion is

then computed as the product of the mass of

liquified vapor WL and the change in internal

energy of the vapor V and liquid L phases, atthe initial pressure P and atmospheric pressure(101 kPa). Thus:

where, for the enthalpies H in J/kg, pressures



8/17

15

P in kPa, absolute, and specific volumes V inm3/kg:

where 1000 is the conversion factor from kPam3 to J.

For the above propane-tank-car example and

metric data54:

and the TNT equivalent would be:r , .

This TNT equivalent is comparable to the 525kg of TNT calculated by using the above flash-ing-vapor-volume method. The methods givegood agreement for flashing fractions less than50%. However, thermodynamic tables are notreadily available for many materials subject tothe causes of BLEVE, and liquid heat capaci-ties, latent heats of vaporization, critical tem-peratures and boiling points are readily avail-able for many materials of interest53,54,58,59.

Superheat Limit TemperatureAn original BLEVE concept required the tem-perature of stored liquid to attain or exceed the&dquo;Superheat Limit Temperature&dquo; (SLT) prior tovessel burst, in order for a BLEVE to occur59,60.

However, the results of accidents indicate that aBLEVE can occur for initial temperatures belowthe SLT.

z

It does appear that the maximum TNT equiva-lent for BLEVE of liquified vapor occurs nearthe SLT. For almost all materials, the values ofSLT lie within the range 88% to 92% of the criti-

cal temperature (in K). Experimental values of

this theoretical maximum limit to superheatingof liquid have been determined for a large vari-

ety of organic and inorganic liquids22,61,62,63.

As an example of a vessel burst at the SLT, a125 m3 container of liquefied propane originallycontaining 111 cubic meters (89% full at 21C)would become liquid-full at the SLT of 53C,and the equilibrium vapor pressure at this tem-perature would be 1,800 kPa. If the container

were to rupture at this point, about 34% of the

propane would explosively vaporize, and theTNT equivalent of the BLEVE would be about

; 440 kg ofTNT (about 0.007 kg of TNT per kg of

propane, and about 4.0 kg of TNT per cubicmeter of vessel volume).

&dquo;Self-BLEVE&dquo;

The SLT increases with pressure P64:

where P and P, are the pressure in the contain-er and the critical pressure (in kPa, absolute),respectively. Thus, heating a liquefied vapor in

equilibrium with vapor above the liquid wouldcause the SLT to increase as the pressure

increases, with the SLT becoming equal to thecritical temperature at the critical point. Thus,spontaneous explosive vaporization or BLEVEwould not occur in confined liquefied vapor

under equilibrium conditions, as long as there isa vapor space. However, if (1) the vapor spaceabove the liquid is small, (2) heating causesexpansion to a liquid-full condition, and (3)there is no overpressure protection, then

hydraulic rupture of the container could lead toexplosive vaporization of the contents and aBLEVE.

BLEVE EFFECTS

Tests to determine the blast effects resultingfrom bursting containers65,66 and theoreticalconsiderationss7 indicate that the blast effects

of a BLEVE can be estimated using a TNT-



9/17

16

Prugh, Quantitative Evaluation of &dquo;BLEVE&dquo; Hazards



10/17

17

J. of Fire Prot Engr.,



11/17

18

Prugh, Quantitative Evaluation of &dquo;BLEVE&dquo; Hazards



12/17

19

equivalent method, but a &dquo;virtual distance&dquo; cor-rection for distance from an explosion center68should be used. Need for such correction is aresult of the fact that the blast pressure at thesurface of an exploding pressure vessel is not

equal to the pressure within the vessel prior torupture and also is not equal to the pressure atan equal radius from a point charge of TNT.

The blast pressure P, (in kPa, absolute) at thesurface of an exploding pressure vessel can beestimated from52,65:

In the preceding equation, Pb is the burstingpressure of the container (in kPa), and k, T, andM are the specific heat ratio, temperature (K),and molecular weight of the vapor in the con-tainer, assuming expansion into air at atmo-spheric pressure and 25C. The above equationcannot be solved explicitly for P., requiringgraphical or trial-and-error solution.

As an example of the use of this equation, theblast pressure at the (former) cylindrical surfaceof a tank car containing propane (k=1.13) andbursting at 2,400 kPa and 6fiC (339 K) wouldbe determined from:

The value of P. which satisfies this equation is410 kPa (absolute).

To determine the pressure effects of a BLEVE-

type explosion, the TNT-equivalence analogyallows use of well-known blast-effects relation-

shipS51,69. Three of the more-important rela-tionships for blast pressures and blast impulseare shown in Figure 1.

For the above BLEVE example involving 34,000

kg of liquified propane at the moment of tankburst, the energy equivalent would be about 525

kg of TNT, and the maximum blast pressurewould be about 310 kPa (gauge). The value ofthe &dquo;scaled distance&dquo; Z corresponding to thisblast pressure (for a free-air explosion) would beabout 1.6. The scaled distance Z is related to

TNT equivalent W (in kg) and distance R (in m)from the explosion center by:

Thus, the value of R for this example (W 1/3equal to 8.1) would be about 13 m. Since the

radius of the cylindrical tank originally wasabout 1.5 m, the &dquo;virtual distance&dquo; to be added

to distances for blast-effects evaluations would

be 13 minus 1.5 meters, or 11.5 m.

The blast pressures and impulses can be deter-

mined from the &dquo;surface explosion&dquo; curve on the&dquo;Z&dquo; graph. For example, the &dquo;reflected&dquo; blastpressures and impulse, at a distance of 30mfrom the center of the tank car (and about 42 m

from the center of an equivilentTNT explosion),would be evaluated from a Z of 5.3 (42/5251/3).For this example, they are 90 kPa, gauge, and1,200 kPa-ms, respectively. Such an explosionwould cause considerable damage to a residen-

tial-type structure.

Further evaluation of blast effects is beyond thescope of this presentation.A discussion of theeffects of blast on humans and structures is

given in reference 68.

Fireball Effects

Evaluation of fireball effects is also outside the

scope of this presentation, due to the complexi-ties of fireball size and duration, distance fromthe fireball center, emissivity, atmospherictransmission of radiation

(humidityeffects), and

the relationship between thermal radiation

intensity and injury. This subject is discussed indetail in other references31,~1,



13/17

20

APPLICATION OF BLEVE HISTORY

AND THEORY

Prevention of BLEVE from Fire ExposureThe approaches to prevention of BLEVE fromfire exposure are presented in References 1, 11,67, 70, 71, and 72.

1. Sloping the ground away from fixed contain-ers, at a slope not less than 1% (1 cm permeter), and leading any spilled material to asafe area.

2. Insulating the containers in which materi-als are held above their atmospheric-pres-sure boiling points, with the insulation heldin place with stainless-steel sheathing andstainless-steel bands.A better alternative is

steel-in-steel double containers, as is fre-

quently used for ethylene storage.

3. Water-spray protection or, in emergencies,unattended water-cannon &dquo;monitors&dquo; for

cooling of containers exposed to fire, at arate of 0.01 cubic meter per minute persquare meter (1 cm per minute) or more.

4. Providing depressuring facilities, to reducethe pressure to 700 kPa (gauge) or one-halfthe design pressure within 15 minutes.

Remote-operated fireproof valves bypassingthe installed relief valve can be provided toserve this purpose.

Prevention of BLEVE from Mechanical

Damage

For fixed storage tanks, overhead suspendedloads or nearby stacks present the greatest haz-ards. Tanks containing materials stored abovetheir atmospheric-pressure boiling pointsshould be isolated from areas in which cranesand other construction equipment operates fre-

quently and should be installed outside a radiusequal to the height of stacks, towers, andcolumns.

For railroad tank cars, the present trends in pro-

viding shelf couplers and head shields can beexpected to reduce the frequency ofpunctures andpotential BLEVEs from mechanical damage.

Some tank trucks and tank cars carrying mate-rials with BLEVE potential are protected withdouble containers, with insulation in the annu-lar spaces, although the outer shell may not

provide much protection against mechanical

damage from collision or overturning. It wouldbe desirable to fabricate the outer container

from a material which would provide protectionfor the inner tankage.

Prevention of BLEVE from OverfillingOverfilling and overpressure incidents arebecoming less frequent because of higher stan-dards in filling and weighing, with properinstallation and testing of relief devices.Also,

the recognition of relief-device plugging prob-lems has led to &dquo;in series&dquo; installation of rup-

ture disks as pluggage protection under relief

valves and sometimes to &dquo;in parallel&dquo; installa-tion of relief valves with rupture disks as &dquo;last

resort&dquo; protection.

Prevention of BLEVE from RunawayReaction

Instrumentation should be provided for mea-

surement oftemperature

andpressure

within

all process equipment likely to contain self-reac-tive materials. Further, facilities should be pro-vided for counteracting overpressure or

overtemperature; this would include internal

cooling coils or external jackets, remote-con-trolled venting valves, inhibitor-injection sys-tems, and internal deluges, as well as high-tem-perature and/or high-pressure alarms for con-trol-room and field personnel73.

Prevention of BLEVE from Vapor-SpaceExplosionPrevention of explosion in process equipment is,of course, essential to process safety, but partic-ularly so in equipment containing liquifiedvapors or liquids above their boiling points.Preventing contamination with reactive materi-als (such as hydrogen with chlorine), inertingvapor spaces with nitrogen or other nonreactive

gas, and installing explosion-suppression sys-tems are methods of

avoiding explosionand pos-

sible aggravation of the incident with release of

toxic or flammable vapors.



14/17

21

Prevention of BLEVE from Mechanical

Failure

Proper design and pre-use testing of containerscan be expected to prevent distortion and possi-

ble rupture of containers. Periodic wall-thick-ness measurements, with internal inspectionwhere the contained material could be corrosiveand with acoustic emission testing where crack-ing of the container could occur, should be per-formed to assure the integrity of containers.Aphilosophy to be developed is that the informa-tion obtained during inspections and testsshould be sufficient to assure proper perfor-mance during the interval between such inspec-tions and tests (thus, &dquo;predictive&dquo; maintenance).

Inclusion of BLEVE Causes in Process

SafetyAnalysisIn assessing the safety or potential hazards of aprocess in which materials are handled abovetheir boiling points51,74, the following aspectsshould be considered:

1. Fire exposure, evaluating the potential forflammable-material spills and ignition.

2. Mechanical damage, particularly collisionand overhead hazards.

3. Overpressure and overfilling, assuring over-pressure protection of adequate size and

reliability.

4. Runaway reaction, including the effective-ness of alarms, operator response, inter-

locks, and overpressure-protection devices.

S. Vapor-space explosion, including contamina-

tion of vapors with incompatible or self-reactive materials, and effectiveness of

inerting or explosion-suppression systems.

6. Mechanical failure, including the effective-ness of prestartup and periodic inspectionsand tests.

In addition, if the potential hazards of storage ofliquefied toxic vapor are to be considered in aprocess safety analysis9, the BLEVE should be-included as a

potentialsource of a

&dquo;puff&dquo; releaseof toxic vapor, together with the possible causesof BLEVE.

CONCLUSION

This paper has presented the potential anddemonstrated hazards of BLEVEs and the

quantitative evaluation of BLEVE causes and

effects. The guidelines and suggestions present-ed here may assist in reducing the frequency

and_ consequences of this severely destructivephysical phenomenon.

REFERENCES

1. National Fire ProtectionAssociation, FireProtection Handbook, 16th Ed., pp.4-22, 4-25 ;5-43 to5-49;6-23 to6-25; 11-35, 11-45,11-46, 1986.

2. Society of Fire Protection Engineers, The

SFPE Handbook of FireProtection

Engineering, pp.2-74,2-75,2-103, and 2-

104, 1988.3. Lees, F.P., Loss Prevention in the Process

Industries, Butterworth, pp. 425, 589, 601,604, 654, 656, and 887, 1980.

4. Kletz, T.A., "Unconfined Vapor CloudExplosions",AIChE Loss Prevention, 11, p.50,Appendix 1, 1977.

5. National Fire ProtectionAssociation,"BLEVE" (videotape), Item No. GL-FL-43V.

6. Center for Chemical ProcessSafety,

American

Institute of Chemical Engineers, "Safety,Health, and Loss Prevention in ChemicalProcesses - Problems for UndergraduateEngineering Curricula - Instructors Guide",CCPS Publication G-16, Problem No. 37:

"Bursting Gas Containers", 1990.

7. Strehlow, R.A., and Baker,W.E., "TheCharacterization and Evaluation of

Accidental Explosions", NASA Report CR134779, p. 52, June 1975.

8. Ploentz, L.M., "Calculate the Energy in a

Pressurized Vessel", Chemical Engineering,87,25, p. 95, Dec. 15, 1980.

9. Carson, P.A., and Mumford, C.J., "AnAnalysis of Incidents Causing Major Hazardsin the Chemical Industry", Journal ofHazardous Materials, 3, p. 149, 1979.

10. Davenport, J.A., "Hazards and Protection ofPressure Storage of Liquefied PetroleumGases", 5th International Loss PreventionSymposium (Cannes), part 22, 1986.

11. Kletz, T.A., "Protect Pressure Vessels from

Fire", Hydrocarbon Processing, 56, 8, p. 98,Aug. 1977.12. Reid, R.C., "Possible Mechanism for

Pressurized-Liquid Tank Explosions or



15/17

22

BLEVEs", Science, 203, p. 1263, Mar. 23,1979.

13. Sharry, J.A., "Three Fire Fighters Die WhenExposure Hazard Ignored", Fire Command,44, 1, p. 22, Jan. 1977.

14. Walls, W.L., "Just What is a BLEVE?", FireJournal, 72, 6, p. 46, Nov. 1978.

15. Rogers, W.P. (chairman), "PresidentialCommission on the Space Shuttle Challenger

Accident", pp. 8, 67, and L-15, 1986.16. Davenport, J.A., "A Study of Vapor Cloud

Incidents -An Update", 4th International LossPrevention Symposium (Harrogate), p. 1, 1983.

17. Eisenberg, N.A., et al., "Vulnerability Model -Marine Spills", Enviro Control Report CG-D-136-75, pp. 301, 305, 309, and 313, June 1975.

18. Garrison, P.E., "100 Largest Losses", 7thEd., M&M Protection Consultants; also,Gardner, W. G., "One Hundred LargestLosses in the Hydrocarbon ChemicalIndustries -A Thirty-Year Review", IChemELoss Prevention Bulletin (058): p. 2, 1984.

19. Kletz, T.A., "What Went Wrong?", GulfPublishing Company, pp. 107, 109, and 199,1985.

20. Lathrop, J.K., "The Terrible Blast of aBLEVE", Fire Command, 41, 5, p. 14, May1974.

21. Pietersen, C.M., "Analysis of the LPGDisaster in Mexico City", 5th InternationalLoss Prevention Symposium (Cannes), part21, 1986.

22. Porteus, W., and Blander, M., "Limits ofSuperheat and Explosive Boiling of LightHydrocarbons, Halocarbons, andHydrocarbon Mixtures",AIChE Journal, 21,3, p. 560, May 1975.

23. Sharry, J.A., and Walls, W.L.,"LP-GasDistribution Plant Fire", Fire Journal, 68, 1,

p. 52, Jan. 1974.24. Washburn,A.E., and Harlow, D.W., Fire

Service Today, 50, 10, p. 11, Oct. 1983.25. Gugan, K., Unconfined Vapor Cloud

Explosions, Gulf Publications, pp. 28, 41, 44,and 83, 1979.

26. U. S. Department of Transportation,Federal RailroadAdministration, "TheEffects of a Fire Environment on a RailroadTank Car Filled with LPG", FRA ReportOR&D 75-31, Sept. 1974.

27. Kim-E, M.E., and Reid, R.C., "The RapidDepressurization of Hot, High PressureLiquids or Supercritical Fluids", Chapter 3

in Chemical Engineering at SupercriticalFluid Conditions,AnnArbor Science -Butterworth, 1982.

28. Lewis, D.J., "BLEVEs Dont Need Fires",Hazardous Cargo Bulletin, 4, 10, p. 34, Oct.

1983.29. Baldock, P.J., "Accidental Releases ofAmmonia: An Analysis of ReportedIncidents",AIChE Loss Prevention, 13, p.35, 1980; also, Lonsdale, H.,AIChE

Ammonia Plant Safety, 17, p. 126, 1975.30. Lonsdale, H., "Ammonia Tank Failure -

SouthAfrica",AIChEAmmonia PlantSafety, 17, p. 126, 1975.

31. Marshall, V.C., Major Chemical Hazards,John Wiley & Sons, pp. 81, 128-137, 142-167,

149, 165, 174, 184, 187, 189, 202, 270, 330,332, 334, 335, 337, 428, 435, and 512, 1987.32. Hazardous Cargo Bulletin, "The Years

Worst?", Sept. 1981.

33. Hazardous MaterialsAdvisory Bulletin,Aug. 1981; also, Chlorine Institute incidentreport.

34. Walls, W.L., "Update: Containing theBLEVE Hazard", Fire Journal, 76 (5), p. 14,Sept. 1982.

35. Giesbrecht, H., et al., "Explosion HazardAnalysis of Inflammable Gas Released

Spontaneously into theAtmosphere",Chemie Ingenieur Technik, 52, 2, 110, inGerman, 1980.

36. Maurer, B., et al., "Modeling of VapourCloud Dispersion and Deflagration after

Bursting of Tanks Filled with LiquefiedGas", 2nd International Symposium onLoss Prevention in the Process Industries

(Heidelberg), p. VI-305, Sept. 1977.

37. Harris, N.C., "Analysis of ChlorineAccidentReports", 21st Plant Managers Seminar of

the Chlorine Institute, p. 11-1,1978;also

described at the International Symposiumon Preventing Major ChemicalAccidents,Washington, DC, Feb. 4, 1987.

38. Solomon, K.A., et al., "On Risks from theStorage of Hazardous Chemicals", UCLA

Report ENG-76125, p. 39, Dec. 1976.

39. Lewis, D.J., "Beware the Heat of July",Hazardous Cargo Bulletin, 4, 8, p. 19,Aug.1983.

40. Voros, M., "Explosion of a Liquid CO2Storage Vessel", 1st Loss Prevention

Symposium (The Hague), p. 337, 1974.41. Hazardous Cargo Bulletin, "Infestation

Agent Blast", 4 (11): 23, Nov. 1983.



16/17

23

42. Stinton, H.G., "Spanish Camp Disaster",Journal ofHazardous Materials, 7, 4, p. 393,March 1983.

43. Davenport, J.A., "A Survey of Vapor CloudIncidents",AIChE Loss Prevention, 11, p. 39

(1977); Chemical Engineering Progress, 73,9, p. 54, Sept. 1977.44. Quarantelli, E. L., "Evacuation Behavior:

Case Study of the Taft Louisiana ChemicalTank Explosion Incident", Ohio StateUniversity Report EMW-2-4186, May 1983.

45. The Times-Picayune (New Orleans), "21Injured by Explosion in Reserve", July 13,1983.

46. Troyan, J.E., and LeVine, R.Y, "EthyleneOxide Explosion at Doe Run",AIChE LossPrevention, 2, p. 125, 1968.

47. Best, R., and Walls, W.L., "Hot WaterHeater BLEVE in School Kills Seven", FireJournal, 76 (5), p. 20, Sept. 1982.

48. Medard, L., "Rupture of anAmmonia RoadTanker",AIChEAmmonia Plant Safety, 12,p. 17, 1970.

49. Strehlow, R.A., and Ricker, R.E., "The BlastWave from a Bursting Sphere",AIChE LossPrevention, 10, p. 115, 1976.

50. Weber, H.C., and Meissner, H.P., Thermo-dynamics for Chemical Engineers, JohnWiley & Sons, p. 122, 1957.

51. Center for Chemical Process Safety,American Institute of Chemical Engineers,"Guidelines for Chemical ProcessQuantitative Risk Analysis", CCPSPublication G-8, pp. 8, 87, 108, 114, 116, and123 -133. 1989.

52. Baker, W. E., et al., Explosion Hazards andEvaluation, Elsevier, pp. 53, 145, 148, 196,and 198, 1983.

53. Perry, R.H., Chemical Engineers Handbook,4th Ed., pp.3-147 to3-155,3-178, and 3-186, 1963.

54. Perry, J.H., and Green, D., Perrys ChemicalEngineers Handbook, 6th Ed., pp.3-46,3-62,3-111,3-160 to3-246, and3-275, 1984.

55. Clancey, V.J., "The Evaporation andDispersion of Flammable Liquid Spillages",Institution of Chemical EngineersSymposium Series No. 39a, p, 80, 1974.

56. Reid, R.C., The Properties of Gases andLiquids, 3rd Ed., pp. 149 and 629; 4th Ed.,

p. 137, 1987.57. Center for Chemical Process Safety,American Institute of Chemical Engineers,

"Safety, Health, and Loss Prevention inChemical Processes - Problems for

Undergraduate Engineering Curricula -Instructors Guide", CCPS Publication G-16,Problem No. 38: Rupture of Tank

Containing Superheated Liquid,1990.

58. Gallant, R.W., Physical Properties ofHydrocarbons, Volumes 1 and 2,1970 and 1984.

59. Reid, R.C., "Superheated Liquids",AmericanScientist, 64, 2, p. 146, Mar. 1976.

60. Reid, R. C., "Superheated Liquid:ALaboratory Curiosity and, Possibly, anIndustrial Curse", Journal of ChemicalEducation, 12, 2, p. 60; 12, 3, p. 108; 12, 4,p. 194, 1978.

61. Blander, M., and Katz, J.L., "BubbleNucleation in

Liquids",AIChE

Journal,21,

(5), p. 883, Sept. 1975.

62. Katz, D.L., "Superheat Limit Explosions",Chemical Engineering Progress, 68, 5, p. 68,May 1972.

63. Moore, G.R., "Vaporization of SuperheatedDrops in Liquids",AIChE Journal, 5, 4, p.458, Dec. 1959.

64. Reid, R.C., "Rapid-Phase Transitions",Volume I, Appendix A, in "MIT-GasResearch Institute LNG Safety andResearch Workshop",Aug. 1982.

65. Petes, J., "Watch Your Equivalent Weight -Pressure Vessels", 13thAnnual ExplosivesSafety Seminar (San Diego), p. 135, 1971.

66. Pittman, J. F., "Pressures, Fragments, andDamage from Bursting Pressure Tanks",14th Explosives Safety Seminar (NewOrleans), p. 1117, Nov. 1973.

67.American Petroleum Institute, "Fire-Protection Considerations for the Designand Operation of Liquefied Petroleum Gas(LPG) Storage Facilities",API Publication

2510A, pp. 2, 6, 22, 29,and

31,April1989.

68. Prugh, R. W., "Evaluation of UnconfinedVapor Cloud Explosion Hazards", p. 712 inInternational Conference on Vapor Cloud

Modeling (Boston), Nov. 1987.

69. U. S.Army, "Structures to Resist the EffectsofAccidental Explosions", Technical ManualTM5-1300, Figure 4-5, June 1969.

70. Walls, W. L. (National Fire Protection

Association), Liquefied Petroleum GasesHandbook, pp. 34, 123, 125, 138, 145, 148,172, 259, and 262 - 266, 1986.

71. Nazario, F. N., "Preventing or SurvivingExplosions", Chem. Eng., 95, 16, p. 102,

August 15, 1988.



17/17

24

72. Martinsen, W.E., Johnson, D.W., andTerrell, W.F., "BLEVEs: Their Causes,Effects, and Prevention", HydrocarbonProcessing, 65, 11, p. 141, Nov. 1986.

73. Center for Chemical Process Safety,American Institute of Chemical Engineers,"International Symposium on RunawayReactions", CCPS Publication P-60, pp. 503,551, 594, and 665, 1989.

74. Center for Chemical Process Safety,American Institute of Chemical Engineers,"Guidelines for Hazard Evaluation

Procedures", CCPS Publication G-1, 1985.

Documents

prugh model