Dispersion of Large Spills of Denser-than-Air

7/28/2019 Dispersion of Large Spills of Denser-than-Air

1/15

Atnmsphmicwironmentol. 3,No.4,pp. 31-745,989. ooo16981/893.at+o.o0PrintednGnatBritain. pqmm pnas lc

A REVIEW OF RECENT FIELD TESTS ANDMATHEMATICAL MODELLING OF ATMOSPHERICDISPERSION OF LARGE SPILLS OF DENSER-THAN-AIRGASES

RONALD P. KOOPMAN, DONALD L. ERMAK and STEVENS T. CHANLawrence Livermore National Laboratory, Box 808, Livermore, CA 94550, U.S.A.

(First received 15 February 1988 and injnul form 2 Nouember 1988)Abstract-Large-scale spills of hazardous materials often produce gas clouds which are denser than air. Thedominant physical processes which occur during dense-gas dispersion are very different from thoserecognized for trace gas releases in the atmosphere. Most important among these processes are stablestratification and gravity Sow. Dense-gas flows displace the ambient atmospheric flow and modify ambientturbulent mixing. Thermodynamic and chemical reactions can also contribute to dense-gas effects. Somematerials flash to aerosol and vapor when released and the aerosol can remain airborne, evaporating as itmoves downwind, causing the cloud to remain cold and dense for long distances downwind. Dense-gasdispersion models, which include phase change and terrain effects have been developed and are capable ofsimulating many possible accidental releases. A number of large-scale field tests with hazardous materialssuch as liquefied natural gas (LNG), ammonia (NH,), hydrofluoric acid (HF) and nitrogen tetroxide (N204)have been performed and used to evaluate models. The tests have shown that gas concentrations up to tentimes higher than those predicted by trace gas models can occur due to aerosols and other dense-gas effects.A methodology for model evaluation has been developed which is based on the important physicalcharacteristics of dense-gas releases.Key word index: Atmospheric dispersion, dense gas, accidental release, dispersion modelling, hazardousmaterials, gravity flow, aerosol, flashing, complex terrain.

INTRODUCIIONResearch into the atmospheric dispersion of denser-than-air gases was begun in the early 197Os,promptedby accidents such as the cyclohexane explosion atFlixborough (U.K.) in 1974, the pesticide release inSeveso (Italy) and the ammonia tanker truck spill inHouston (U.S.), both in 1976. Since then, the tragicmethyl isocynate (MIC) release in Bhopal, India (1984)killed over 2000 people and the LPG explosions inMexico City (1984) killed over 400 people. Thesetragic accidents indicate that further research intothe possible consequences of accidental release ofhazardous materials coupled with the development ofprediction, mitigation and emergency responsecapabilities is still needed. The purpose of this paper isto review the major advances that have been madeover the last 10 years in our understanding of theatmospheric dispersion of large-scale spills and ourability to predict the consequences of an accidentalrelease.

Dense gas dispersion phenomena, which are oftencharacteristic of these spills, have been investigatedintensively both through theory and experiment(Koopman, 1987). The emphasis of the research atLawrence Livermore National Laboratory (LLNL)has been on gaining a physical understanding of themany processes involved such that a physically de-tailed and accurate quantitative predictive capability

could be developed. This approach involves work inthree complementary areas:

(1) mathematical modeling based on physical lawsin the form of conservation equations and sub-models of the important physical processes,(2) field experiments to validate models and dis-cover important phenomena,

(3) scaled simulations in wind tunnels, laboratoryor water flume.

Work at LLNL has concentrated on the first two ofthese areas.

The dominant physical processes which occur dur-ing dense gas dispersion are very different than thoserecognized for trace gas releases. Some dense gasdispersion models now include these processes andcan be evaluated against good quality data from avariety of well-instrumented dense gas dispersion ex-periments.

CAUSES OF DENSE GAS BEHAVIORThe accidental release of hazardous gases into the

environment often results in what has come to beknown as dense gas behavior, even when the gasesthemselves may be nominally less dense than theatmosphere into which they are released. This densegas behavior can dominate the consequences of theaccidental release making them either worse or better


2/15

132 RONALD P. KOOPMAN et ~1.

than might otherwise be expected and making predic-tion of the consequences of the release more difficult.Materials released during an accident can be denser-than-air due to one or a combination of the following:(1) they are cold or cool from evaporation whenreleased, (2) their molecular weight is higher than thatof air; (3) they are contained at elevated pressure andflash to aerosol and vapor upon release; (4) chemicalreactions.

The first two categories, low temperature and highmolecular weight, are largely obvious and will not bediscussed further. The consequences of flashing, on theother hand, are not so obvious. Many materials, suchas NH, and HF are kept at elevated pressure so thatthey may be efficiently stored as liquids or efficientlyparticipate in a chemical process. If these materials areaccidentally released, flashing will result in a mixtureof cooled vapor and cooled liquid aerosol. Theamount of vapor produced by flashing, the flashfraction, is a function of the initial storage temperatureand pressure and the thermodynamic properties of thereleased material, and can be calculated from basicthermodynamics. The amount and size distribution ofthe liquid aerosol entrained in this vapor plume is noteasily calculated and is currently the subject of re-search. Several competing theories are being examinedbut it appears to this author that the small droplet sizepredicted by the boiling breakup theory (Crowe andComfort, 1978) is most consistent with the pm-sizeparticles observed in the sparse existing data. Betterdata on HF aerosol size should be available soon andwill help identify the correct approach to modelingaerosols. The presence of aerosols causes severaldense-gas effects. First, the liquid droplets make thecloud denser than air even though the vapor may belighter. Second, the evaporation of the liquid dropletscools the surrounding air/vapor mixture, which couldfurther increase the cloud density. Third, turbulence issuppressed within the dense cloud, reducing mixingwith ambient air.

The importance of this effect was demonstratedduring large-scale NH, releases performed at theNevada Test Site in 1983 for the U.S. Coast Guardand The Fertilizer Institute (Goldwire, 1985,1986) andduring HF tests (Goldfish series) performed in 1986 forAmoco Oil Co. (Blewitt, 1987a). In both cases approxi-mately 20% of the liquid flashed to vapor and theremaining liquid (80%) formed a fine aerosol whichremained airborne. Figure 1 shows the temperaturedepression measured 20 m and 60 m downwind for thefirst three Goldfish HF spill tests (Blewitt, 1987a).Flashing has created a cold, aerosol-laden cloud, andthe aerosols evaporate as they travel downwind, caus-ing the cloud to remain cold and dense. For the largestspill rate, Test 1, the cloud temperature is nearly thesame at 60 m (Fig. 1b) downwind as it is at 20 mdownwind (Fig. la). The HF was stored at approxi-mately lOYF, close to ambient temperature.

Chemical reactions can occur which increase ordecrease cloud density. For example, when N204 (a

-10OTest2-20 hTul3 _

-30

-a01 I I I I I I0 200 400 do0 owl low 1200

Tlm# from rplll (8%)50

(b)-501 I I I0 400 ml 1200 1MwTlmo from s@II (m)

Fig. 1. Temperature differences for GoldfishHF Tests 1,2 and 3 measured at (a) 20 m and(b) 60 m downwind of the spill point.

rocket fuel oxidizer) is released, it dissociates rapidlyto NO2 which undergoes a further rapid endothermicreaction with atmospheric water vapor to produce anHNO, mist (McRae, 1984, 1985). This mist behavesdifferently than does the ambient temperature NO,,cooling the cloud and increasing its density whichreduces vertical mixing As another example, cold HFdroplets cause water vapor to condense on them andreact, releasing heat. This increases the temperaturewhich could be expected to increase the evaporationrate and further cool the droplet. However, thewater-HF solution has lower volatility allowing thedroplets to persist longer downwind. The net result isthat cloud travel distances to concentration levels ofconcern are expected to increase with increasinghumidity up to about 70% for HF releases eventhough the HF-water reaction is exothermic. Work isplanned to include these effects in dense-gas dispersionmodels so that parameter studies and analysis can beperformed.Dense gas effects

The important dense-gas effects which are not ob-served in the dispersion of trace emissions includeturbulence damping and gravity spreading due to


3/15

Review of recent field tests 733density gradients in the horizontal direction. In somesituations, the dense cloud actually displaces the am-bient wind field (Koopman, 1982a, 1987) in much thesame way that wind flows over a solid body. Thisdisplacement results in a stably stratified dense gaslayer and an interface through which it is difficult forexternal turbulence to penetrate (Hunt et al., 1983).Further, in many situations, the kinetic energy of theturbulence within the cloud is not sufficient to givefluid elements the large, vertical displacements neededto mix upward with the ambient atmosphere. Thus,stable stratification damps the vertical component ofturbulence at the interface. Displacement of the am-bient wind field also causes the cloud to becomenearly decoupled from the ambient flow and to movedownwind at a rate slower than the ambient windspeed. All of these effects are most pronounced whenthe ambient wind speed is low and the atmosphericconditions are stable.

Figure 2 shows data from bivane anemometers bothupwind and downwind of the 1980 Burro 8 liquefiednatural gas (LNG) test. The wind speed recorded bythe anemometer in the cloud drops to near zero whenthe cloud is present indicating that the cloud isdisplacing the ambient wind flow. Figure 3 showsvertical turbulence data from a bivane anemometer asit experienced a cold dense LNG cloud during the1987 Falcon series tests (Brown et al., 1989). The effectof the cloud on turbulence is marked.

Turbulence damping is accounted for in the variousdense gas dispersion models in a number of differentways (Ermak, 1988a). Currently, FEM3 (Chan, 1988)uses a K-theory model to parameterlze turbulencewhich is dependent on cloud Richardson number, a

Fig. 2. Bivane anemometer data upwind anddownwind of the Burro 8 LNG test.Significant modification of wind speed due toLNG cloud displacing atmospheric flow wasobserved.

m0 (88~)Fig. 3. Bivane anemometer data from Falcon test seriesshowing modification of vertical fluctuations due topresence of LNG vapor cloud.

measure of the stable stratification in the cloud. As thecloud density approaches that of the ambient air, theRichardson number also approaches that of the am-bient atmosphere.

Gravity flow is driven by the excess hydrostaticpressure caused by the density difference between thecloud and the ambient atmosphere (Hunt, 1983). Thefluid motion is generally horizontal except near thefront where there is a recirculating vortex. Most of themixing occurs just behind the front due toKelvin-Helmholtz instability. For a continuous spillinto a steady wind, gravity spreading also producesvortices in the crosswind direction which entrain airinto the cloud at the edges. This can be seen in Figs 4and 5. Figure 4 is a crosswind vertical cross sectionshowing the flow field reproduced by a FEM3 com-puter simulation of the Burro 8 LNG test both withand without terrain compared to contours con-structed from data (Chan and Ermak, 1983). Gravityflow in the crosswind direction produces the bifurc-ation and lobe structure. Terrain effects enhance thelobe on the left. Figure 5 shows horizontal contoursfor a FEM3 simulation of Burro 8, with terrain effects,compared to contours constructed from data. Thegravity-flow-produced cloud bifurcation is even moreapparent in this view.

Instantaneous gravitydriven releases have beenstudied in detail both experimentally and theoreticallyduring the Thomey Island Trials (McQuaid, 1987).Gravity flow in the near field usually increases cloudsurface area and produces intense outward movingvortices, both of which tend to enhance the entrain-ment of air into the cloud. In the far field, gravity flowis much less significant, thus entrainment is reduceddue to the low cloud profile and stable stratification,Gravity flow causes shear at the ground/cloud inter-face and at the air/cloud interface. In addition, forcryogenic clouds, heat transfer from the ground intothe cloud, is enhanced, increasing mixing with air dueto increased buoyancy.Mathematical dispersion models

Over the last 10 years, many models of the disper-sion of gases from accidental releases have been

AE 23:bB


4/15

134 RONALD P. KOOPMAN er ul

Crosswinddii (m)Fig. 4. Burro 8 crosswind contour plots of concentration 140 mdownwind at t = 180 s: (a) experiment; (b) flat terrain simulation;(c) variable terrain simulation.

created. These models fall into three categories ofvarying physical completeness and numerical com-plexity: three-dimensional conservation equation,similarity profile and Gaussian-based. There havebeen at least four reviews of 6 dense-gas dispersionmodels during this time (Ermak, 1988~ Blackmore etal., 1982; Havens, 1982; Wheatley and Webber, 1984).Much of the information presented here has beensummarized from Ermak et 01. (1988a,b).

The models which provide the most physicallycomplete description of dense gas dispersion are thosethat are based on the three-dimensional, timedepen-dent conservation equations. Examples of this type ofmodel include FEM3, SIGMET, MARIAH andZEPHYR, which have recently been evaluated byHavens et al. (1987). At the intermediate level ofcompleteness and complexity are the similarity-profilemodels. These models use simplified forms of theconservation equations that are obtained by averag-ing the cloud properties over the crosswind plane.Quasi-three-dimensional solutions are obtained byassuming a functional form for the crosswind profile ofconcentration and other cloud properties. Examplesof this type of model include SLAB, HEGADAS andDEGADIS. At the simpkst level are the modifiedGaussian plume models. These models were designed

to simulate trace-gas releases into a normal atmos-phere. They are usually used to simulate continuousreleases and employ a variety of modifications whichattempt to include the effects of dense gas dispersionwithin the Gaussian framework. Box or top-hat mo-dels, which are used to simulate instantaneous andcontinuous releases, fall into either the intermediate orsimple category depending upon the complexity of themodel regarding the number of conservation equa-tions to be solved.

The three types of models differ considerably intheir approach to simulating the atmospheric disper-sion of a dense gas release. Perhaps the most obviousdifferences are related to the degree to which eachmodel type incorporates the basic conservation lawsand three-dimensional effects. The modified Gaussianplume model is based on the single conservation ofspecies equation and either neglects momentum andenergy transfer or attempts to include them in some adhoc manner. On the other hand, the intermediatesimilarity profile models include the conservationequations of mass, momentum and energy, in additionto the species equation, but only in an average way. InSLAB, variations in the properties of the vapor cloudin the crosswind plane are described by functions(Gaussian, for example) and treated by the code as


5/15

Review of recent field tests 735200. I ICl Expuinmnt ,/

-2001

200 I I I I(b) FEM3 (variable terrain)

loo 200 300 400Downwind distance(m)

Fig. 5. Burro 8 contour plots of gas concentration at 1 mabove ground at 180s. FEM3 compared to experiment(dashed lines show edge of instrument array).

crosswind average values which vary in the downwinddirection only. The conservation equation models,such as FEM3, include the most complete descriptionof the conservation laws by treating them explicitly inthree dimensions.Both the three-dimensional conservation equationmodels and the intermediate similarity profile modelsincorporate mathematical descriptions of thesephysics that cause the major heavy gas dispersioneffects. Of the two types, the three-dimensional con-servation equation models provide the more detailedand complete description of the physics involved indense gas flows. An example of this type of model is theFEM3 model.

FEM3 simulates the dispersion of a released gas bysolving the time-dependent, three-dimensional, con-servation equations of mass, momentum, energy andspecies along with the ideal gas law for the equation ofstate. In addition, it can treat flow over variableterrain and around obstructions such as cylinders andcubes. Turbulence is treated by using a K-theorysubmodel. Since it is fully three-dimensional, FEM3can simulate complicated cloud structures such as:

(1) the vortices that are typical of dense gas flows,(2) cloud bifurcation observed during heavy gas

releases under low wind speed, stable, ambientconditions (Figs 4 and 5),

(3) cloud deflection caused by sloping terrain.FEM3 is the only known 3D conservation equationmodel with an optional submodel for phase change.

Two versions of this submodel have been developed.The first version (Leone et al., 1985) is for atmosphericwater (ambient humidity-fog) and the second(Rodean et al., 1986) is generalized so it can be used fora variety of hazardous liquids. It is assumed in bothversions that the mixture components (air in the gasphase, liquid material and material vapor) are alwaysin thermodynamic equilibrium. Therefore, there isnever any supercooled vapor or superheated liquid,and there is no thermal lag in the system.

In principle, three-dimensional models are capableof modeling complex time-dependent phenomena andtaking into account complex boundary conditionsimposed by terrain and structures. While these modelsgenerally provide the most detailed and completedescription of heavy-gas flow, there are limitations totheir use, and often simplifications and compromisesare made due to the complexity of these models.Wheatley and Webber (1984) discussed several ofthese aspects including the following.

(1) The modeling of turbulence is possible only bymaking assumptions and approximations. Thenecessity of using some form of turbulence clos-ure in order to solve the system of equations forthe mean fields of velocity, temperature, andconcentration is the weakest part of 3D models.

(2) Other approximations (e.g. hydrostatic, Bous-sinesq and anelastic) are made in order tosimplify the computations and reduce their cost.

(3) The numerical solutions necessarily involve dis-crete approximations (finite-difference or finite-element) to continuous functions. The design ofa three-dimensional mesh for a numerical simu-lation involves compromises among several par-ameters including the largest and smallest scalesof the phenomena and environment to be rep-resented, the capacity of the computer, and thecost of machine time.

(4) As indicated above in (a)-(c), these models areexpensive and time-consuming to develop anduse.

An example of an intermediate similarity profilemodel is Steady State SLAB. The SLAB model solvesthe crosswind averaged equations for the conservationof mass, species, downwind and horizontal crosswindmomenta and energy along with an additional equa-tion for cloud width and the ideal gas law equation ofstate. The current version of SLAB includes thesteady-state assumption for continuous releases, but apuff version (finite duration release) is under devel-opment. Thus, the code is one-dimensional withdownwind distance being the independent variable;however, since cloud width and cloud height are also


6/15

736 RONU D P. KOOPMAN CI alcalculated, the model is, in this sense, quasi-three-dimensional. The crosswind concentration distribu-tion is determined by using similarity profiles basedupon the calculated crosswind height and width.Mixing of the cloud with the ambient atmosphere istreated by using the entrainment concept. The mainadvantage of the SLAB code over a 3-D code is its lowcomputing cost. Typical simulations require only afew seconds on a CDC 7600 computer or a fewminutes on an IBM micro-computer.

A unique feature of the SLAB model is that itcalculates only crosswind-averaged properties, andcharacterizes the cloud shape by the height, h, andhalf-width, B. The parameters B and h do not corre-spond to any particular concentration level. Rather,they can be considered to describe a surface whichencloses the bulk of the cloud, for example 90%.Consequently, the crosswind concentration distribu-tion is not uniquely defined. This causes difficulties inattempting to compare the predicted cloud shape fromthis model with contour plots obtained from exper-iments. To overcome this difficulty, one generallydefines B and h in terms of an assumed distribution(such as Gaussian, exponential or quadratic) for thecrosswind horizontal and vertical vapor cloud concen-tration.

There are other important differences and these arerelated to the manner in which each model type treatsthe effects of gravity and turbulence. As indicatedabove, the modified Gaussian plume models use adhoc formulas with empirical coefficients to describe thegravity spread and turbulent dispersion of the cloud.In contrast to this, the SLAB and FEM3 models useconservation principles to treat the effects of gravity.This is done in the FEM3 model by solving the threemomentum conservation equations, including thebuoyancy term and variable density, while the SLABmodel solves two layer-averaged momentum equa-tions and uses the hydrostatic approximation.These two models differ considerably in their ap-proach to turbulence. The SLAB model uses thesomewhat artificial concept of entrainment across thecloud-air interface and essentially neglects any ex-plicit treatment of turbulence within the vapor cloud.Air is entrained into the cloud at the surface and thenis assumed to mix rapidly in the cloud creating anearly uniform layer in the crosswind plane. Thus,there are two separate regions: the cloud and theambient atmosphere. Mixing between the two is as-sumed to occur at the interface and is governed by anentrainment velocity which depends on the localproperties of both the cloud and the surroundingatmosphere. The FEM3 model assumes that turbu-lence can be described as a diffusion process and uses acontinuous diffusion coefficient which depends on thelocal properties of the dense gas cloud. While theentrainment and diffusion concepts are peculiar to theSLAB and FEM3 models, respectively, the choice of aparticular entrainment or diffusion submodel is not anessential aspect of the models. Several submodels have

been proposed in the literature including higher orderturbulence submodels and these could be used with-out changing the whole model.

This occurred recently when FEM3 was used tomodel the Thorney Island Phase 1 trials (Chan er ul..1987a). which revealed that the dense gas Richardsonnumber determined in a previous K-theory turbulencesubmodel was not adequate for regions where flowvelocities were significantly perturbed, either by thepresence of obstacles or a heavy-gas cloud. A newK-theory model was developed based on the gradientRichardson number concept.

with the assumptionsP-P~=(P-P.)I----xpC-(zlhcrl

u = ? In(z/z, ).k

This resulted in a local Richardson number definedby

&u2--!ia. +nk2(z/hlu:+w:)where u+ = friction velocity = u,+, where a desig-nates ambient,

w2 = in cloud convection velocity (Chan et al.,1987a),

k = von Karmans constant = 0.4,z = height above ground,

h, =characteristic cloud height,p = density,

Ri, = ambient Richardson number = z/L,g = acceleration of gravity,n =2 based on comparison to experiments

(McQuaid, 1976).z0 = surface roughness,L = Monin-Obukhov length scale.

Thus the Richardson number vertical profile has amaximum value near the top of the cloud where thedensity gradient is largest and approaches zero atground level. As cloud density approaches ambientdensity, the ambient value of Richardson number isrecovered. Although this new K-theory model per-forms much better, a higher-order turbulence modelwill probably be necessary for adequate general treat-ment of complex situations.

Finally, a three-dimensional, time-dependent, con-servation equation model is required in order todescribe the distribution of gas concentration in spaceand time from a heavy gas release into an atmosphericboundary layer with speed and directional wind shearin the presence of complex terrain and man-madestructures. This versatility of the three-dimensionalconservation equation model in treating more realisticsituations and providing a more detailed descriptionof the flow is somewhat balanced by the increased


7/15

Review of recent field tests 737

computer costs in running these models. Conversely,similarity models, while giving up some degree ofrealism and detail, are much faster to run on com-puters, Consequently, many researchers tend to viewthe three-dimensional conservation equation modelsmore as research tools that help them discover newthings about the flow and to view the intermediate, orsimilarity models, as more of an operational model forsituations where computing costs and time are of theessence. The 3-D models are also very useful for doingassessments at fixed sites where dense gas and terrainor structure effects are im~rtant.

FIELD EXPERIMENTS AND COMPARISON TO MODELS

Field experiments in dense gas dispersion have beenperformed since the early 1970s. Initial experiments weremostly with LNG with some early work also withLPG and NH,. The quantity and quality of dataobtained from these experiments has varied greatly.Most of the good quality data have been obtainedsince 1980 when, essentially simultaneously, Shell/NMI conducted a series of LNG and LPG trials atMaplin Sands in England and DOE/LLNL conduct-ed the Burro series of LNG tests at China Lake,CA. There have been several recent reviews of fieldprograms (Koopman, 1987; McQuaid, 1983a, 1983b;Puttock and Colenbrander, 1985; Vilain, 1986;Puttock et a/., 1982).

There are a number of reasons for doing field-scaleexperiments. The relative importance of the manyeffects which occur upon accidental release is oftenunknown. Much of the physics dominant in dense gasdispersion is highly nonlinear and may not even beobservable in small scale tests in the atmosphere.Stable atmospheric conditions allow dense gas effectssuch as turbulence damping and gravity spreading todominate cloud behavior. Nonisothermal effects (i.e.heat transfer, phase-change, etc.) can also be import-ant and are not easily scaled. Thus, one of the majorreasons for performing these tests is the determinationof the important effects and their scaling laws. Inaddition, the well instrumented tests allow quantitat-ive measurement of these effects so that correct de-scriptions can be incorporated into the mathematicaldispersion models. This is necessary if accurate predic-tions are to be made for circumstances different thanthose under which tests were performed.

Other reasons for conducting field tests includeaccident simulation or evaluation of mitigation equip-ment such as water or steam curtains. These may besituations which are simply too complicated, with toomany unknown contributions, to yield to mathemat-ical or physical model simulation. Other complicatingeffects which are difficult to model or scale includechemical reactions and certain irreversible thermo-dynamic effects such as flashing two-phase flow.

Perhaps the most common reason for the conductof field experiments has been to obtain basic data for

dispersion model evaluation. This requires extensive,carefully verified, quantitative data from well docu-mented and well instrumented experiments. The needfor such data was demonstrated by McQuaid (1976) ofthe British Health and Safety Executive (HSE), whenhe invited the mathemati~l modeling immunity topredict in advance the results of an instantaneousrelease of 2000 m3 of gas with an initial density 2 timesthat of air, Variations between models of two orders ofmagnitude were present even for this relatively small,isothermal release onto flat terrain without chemicalreactions, the~~yn~ic effects or other compli-cations. Models have improved since that time, but thetasks they are being given are much more difficult andinclude many more complicated effects. Thus, modelevaluation continues to be one of the major reasonsfor the conduct of field experiments. In spite of thenumber of tests which have been conducted, data formodel evaluation are still in short supply. Table 1gives an edited summary of recent large scale disper-sion tests. The emphasis is on the observation of densegas effects as defined in this paper. Consequently,small tests which do not show dense gas behaviorclearly and tests which were not well instrumentedhave been left off.

Each major set of field experiments has had aparticular emphasis and particular goals. This sectionprovides a brief summary of the major achievementsof each of the test series listed in Table 1 and selectedexamples of model comparisons to the data.

The Maplin trials were sponsored by Shell andconducted at Maplin Sands, England, by the NationalMaritime Institute (NMI) (Puttock et al., 1982). Thiswas a very ambitious well-instrumented experimentalprogram involving some 20 spills of 5-20 m3 of LNGand 14 spills of 13-31 m3 of propane onto water. Someof these trials were instantaneous releases from asubmersible barge, some were continuous and somewere ignited. Good data on dispersion over waterunder various meteorological conditions, flamepropagation in un~on~ned gas clouds, heat transferand humidity effects were obtained. Large-scale rapidphase transition (RPT) explosions occurred on one ofthese tests.Burro

The Burro series of tests were sponsored by the U.S.DOE and conducted by LLNL and Naval WeaponsCenter (NWC) personnel at China Lake, California,(Koopman, 1982a,b). This series involved eight largeLNG releases onto water, conducted under a varietyof meteorological conditions. Good quality data froma large array of instruments were obtained. Veryrevealing dense gas e&c&, showing that the cloudcaused displacement of the ambient atmospheric flowand modification of turbulence, were observed andmeasured on the Burro 8 test. An example of this effectis shown in Fig. 2. Good data on dispersion over land


8/15

738under a variety of met~#rolo~ca~ conditions and dataon heat transfer, turbulence, and humidity effects wereobtained. Large-scale RPT explosions with chamcter-istics unobserved before occurred on one of these tests.

The Coyote test series was a continuation of theBurro series into the areas of &ombustion and RPTs(Goldwire et al ., 1983; Rodean et al., 1984; Morgan etal., 1984; McRae, 1984b). Five combustion tests wereconducted involving 8-28 m3 of LNG spilled under avariety of meteoro~o~~a~ conditions with hardenedinstrumentation so that dispersion data, prior toignition, could also be obtained. Flame propagationwas measured for free-doud combustion with bothsoft (flare) and high-vel~ity (Same jet) ignition sour-ces. Thirteen spills of 3-14 m3 with varying composi-tions were performed to get basic data on the new typeof RPT observed during the Burro tests.

The Thorney Island trials were organized by theBritish HSE and conducted at Thorney Island (U.K.)by NM1 (M&&aid, 1987). They invotved large in-stantaneous isothermal dense gas (Freon+N2) re-ieases from a cohapsibie tent-like structure. Althoughthe quantity of gas released was relatively small(2000 m3) compared with most of the other test series(23,000 m3 ofgas for Burro) the release was instantan-eous, and the density was high, producing largeRichardson numbers for some of the trials. These Wialshad a strong experimental emphasis on creation of awe&defined model source and gravity driven flow. Inmost cases gravity flow dominated cloud behaviorover the entire measurement array, The triafs werewell instrumented, and high quality data were ob-tained. The first series (Phase If consisted of f 6 releasesover unobstructed terrain, with gas densities varyingbetween neutral (air) and 4 times air, with most doneat 2 times air. The second series (Phase II) consisted often releases into an obstacle field featuring an imper-meable wail, a permeable screen simulating fohage,and a cube simulating a building. A third series(Phase III) consisted of 17 continuous releases, some14 of which were for the U.S. Department of Trans-portation and involved releases into a vapor fencecontainment structure.

FEM3 was used to model Phase I trials 913 and 17(Chan, 1987a). Simuiat~on of the dispersion was per-formed in two stages. Initiafly, the flow field over andaround the source was calculated. This is shown inFig. 6 for Trial No. 9. A vertical plane of symmetry isassumed and only half of the structure and flow field isshown in Fig. 6(b). Then the constraints were removedand the gas cloud was allowed to slump and mix withthe complex ambient air Row caused by the structure.Figure 7 shows the resulting gas concentration con-tours and velocity vectors at 4 s after the release.

To demonstrate the effects of the different K-theory(Richardson number) models considered, Trial No. 13


9/15

Review of recent field tests

(b) lWkontalpb~atz=O.4m

-40 -20 0 20 40 60DownwInddbtubm (m)

Fig. 6. FEM3 velocity field prior to Thomey Island TrialNo. 9.

ICI nmst4s

40 -20 0 20 40DownwInd dhtanw (m)

Fig. 7. FEM3 predicted concentration contours andvelocities along the vertical symmetry plane forThomey Island Trial No. 9 at 4 s after release.

was simulated with the previous model (Ri peaks atground level), with ambient turbulence only (Ri = Ri ,= Z/L where L = ambient Monin-Gbukhov lengthscale) and with the present model where Ri peaks nearthe top of the cloud. The result is shown in Fig. 8.Without the density stratification term to damp outturbulence in the cloud, the ambient K model under-predicted the peak concentrations by a factor of nearly2 for the entire curve. The previous K model improvedagreement close to the source but actually causedpoorer agreement beyond about 1OOm downwind.The present K model produced the best agreement,but further improvement is needed to better simulatethe early phases of the release. This K model still doesnot adequately account for significantly perturbedvelocities due to the presence of obstacles or heavy gasclouds.

0 100 200 300 400Downwind diatma (m)

M)

Fig. 8. Peak concentrations from Thomey Island Trial No.13 compared to FEM3 with various turbulence models.

Desert Tortoi seThese large-scale NH, releases (Goldwire, 1985,1986) were performed by LLNL at the Nevada Test

Site under the sponsorship of the U.S. Coast Guardand the The Fertilizer Institute (TFI). The goal of thetests was to obtain basic dispersion data for two-phase(flashing) releases of pressurized ambient temperatureammonia under simulated worst case accident con-ditions. The instrument array is shown in Fig. 9.Detailed measurements of cloud shape, compositionand aerosol density were made at 100 m and 800 mdownwind (measurements at 1,3,8 m heights), with anarc of single level (1 m) measurements of gas concen-tration at 2800 m downwind. Four tests underPasquill-Gifford category D and E stability condi-tions were performed with good quality data obtainedon all of them. Aerosol effects dominated cloud behav-ior on these tests with approximately 80% of the massreleased being in the form of an aerosol, most of whichwas transported downwind. This created strong densegas effects which persisted for long distances down-wind.

The latest version of FEM3, with improved turbu-lence and phase change models, was used to simulatethe fourth Desert Tortoise test, a release of 60 m3 inthe form of a strong horizontal jet pointing downwind.Since a large pool of NH, was observed immediatelyafter the test and since mass flux measurements at100 m and 800 m downwind recorded only 70% of themass released in the main cloud, the source strengthwas reduced to 70% of the amount released. Figure 10shows FEMZpredicted cloud centerline temperaturecompared to measurements closest to the center. Sincethe current phase-change model does not include


10/15

740 RONALD P. KOOPMAN et al

N .c 225 :. f I

Dispwaion array -

. Gas sensor stationA Anmwmotw station

Fig. 9. Diagnostic instrument array for Desert Tortoise and Eagleseries experiments.

40

M-

E OI -20 -

I I I I

-80 1. I I I0 200 400 6a a00 1000

Fig. 10. Temperature along downwind distance at an elevation of 1 m for AmmoniaSpill Experiment No. 4 at t = 300 s. The experimental data are measurements closest tothe center of the vapor cloud. The FEM3 curve is based on predicted values along thecloud centerline.


11/15

Review of recent field tests 741water condensation and reaction with the NH, aer- of ten at 100 m and iow by a factor of ten at 2800 m.osol, the aerosol evaporates too fast causing the Presumably these differences are due mainly to thepredicted temperature to be too low at IOOm. Thii absence of aerosol and dense gas effects in the Gaus-also results in too little aerosol being present at 800 m sian plume model. Differences in concentration aver-in the model cloud and a predicted temperature which aging times between the data and the Gaussian modelis too high. Figure 11 shows the predicted cloud may also account for some of this difference. Thecenterline gas concentration from both FEM3 and a aerosol evaporated within a few hundred meters, butGaussian plume model compared to the peak gas dense gas effects persisted and the cloud remained coldconcentration data from the various arcs of gas sen- and compact.sons. FEM3 concentrations are too high at 100 m andtoo low at 800 m because not including reactions with gqfewater vapor causes the aerosol to evaporate too fast. The Eagle series of N,O, experiments (I&Rae,The high-speed jet source, which was not modeled by 1984a, 1985), was performed y LLNL under the spon-FEM3, may also account for some of the discrepan- sorship of the U.S. Air Force for the purpose ofties. The Gaussian prediction is high by nearly a factor determining source. conditions for confined and un-

loo1

10%

111

0.11

I I

Fig. 11. Comparison of measured peak ammonia concentrations as a function ofdownwind distance at an elevation of 1 m for Desert Tortoise Test 4 with FEM3 andGaussian plume calculations.


12/15

142 RONALDP. KOOPMAN t al.confined spills, making basic dispersion measure-ments, and providing emergency response trainingand equipment evaluation. The instrumentation arraywas similar to Desert Tortoise, Fig. 9. In addition,extensive reaction of NO* with atmospheric watervapor caused approximately half of the cloud to betransported downwind as an HNO, mist which dam-aged gas sensors and was not detectable. The net resultis that maximum gas concentrations had to be esti-mated from analysis of these data. Good data on cloudwidth and height at 800 m downwind were obtainedwhich show significant dense gas effects and have beenuseful for model data comparison. Data from two ofthese tests are summarized in Table 2 where they arealso compared with simple Gaussian-type dispersionmodels in use by the USAF at the time of the tests. Thedata indicate that gas concentrations 5 to 10 timeshigher than those predicted by the simple models weremeasured, that measured cloud heights were l/5 to1 lO those predicted by the models, and that measuredcloud widths were about half those predicted. Thus,the cloud remained low and compact due to its highdensity.Goldj ish

The Goldfish test series was performed in 1986 byLLNL and Amoco Oil Co. under Amoco sponsorship(Blewitt et al., 1987a,b,c; Chan, 1987b). A series of sixtests, four with releases of about 4 m3 (1000 gal) andtwo of about 2 m3 (500 gal) of HF were performed atvarious spill rates and with varying meteorologicalconditions. The first three tests were designed toobtain data on source characteristics and dispersionand the last three were to evaluate the effectiveness ofvarious water spray curtains. An extensive instrumentarray was fielded with rows of gas sensors at 300 m,1OOOm and 3000m. Prior to the tests, there wasconsiderable uncertainty about how much HF aerosolwould form and be transported downwind for HFreleased suddenly through an orifice at 40C and 115psig. Under these conditions, approximately 20% wasexpected to flash to vapor, and the remaining 80% toremain as liquid droplets, either raining out on theground or being transported downwind as an aerosol.

One of the major findings of the test series was that allof the material released, both vapor and liquid aerosol,was transported downwind as a heavy cloud. Simu-lations of these tests were performed using the FEM3,(Chan, 1987b) SLAB and DEGADIS (Blewitt et al.,1987~) models. Figure 12 shows SLAB calculations ofthe first three tests compared to peak concentrationdata. The major differences between the.tests were spillrate, which varied from 1.8 m3 min- on Test 1 to0.6 m3 min _ on Tests 2 and 3. The measured andSLAB-predicted plume crosswind concentration pro-files at 1000 m downwind are compared in Fig. 13 forTest 1.Falcon

The Falcon test series was performed by LLNL forthe Gas Research Institute and the U.S. Departmentof Transportation in 1987 (Brown et al., 1988). Thistest series involved five large-scale (20-66 m3) releasesof liquefied natural gas (LNG) into a 44 m x 88 mx 10 m-tall vapor containment fence. The tests were a

full-scale follow-on to the Thorney Island Phase IIItests and were designed to provide the detailed datanecessary to validate wind tunnel and mathematicalmodels of dispersion from vapor fences and evaluatetheir effectiveness. Data reduction and reporting arecurrently underway for these tests with analysis usingFEM3 to begin soon.Model evaluation

Model evaluation, always an important part ofdispersion modeling, has received more attention inthe past few years as industries and regulating agenciesbegin to rely on models for risk assessment, siting,emergency planning and other regulatory decisions.This has led to an increasing need to know howaccurate models really are and what are the limits oftheir applicability.

Various approaches have been developed to airquality model evaluation in recent years, but theemphasis has been on statistical methods (Fox, 1981,1983). This approach has recently come under criti-cism by several authors (Venkatram, 1986; Dennis,1985) because it does not promote understanding of

Table 2. Comparison of eagle test results at 800 m downwind with simple dispersion model predictions

Test resultsRecordedEstimatedModel predictionsOcean breeze/dry gulchGaussian plumeShellCharm(McRae, 1985)

Eagle 3 Eagle 6Peak Peakconcentration(3) (2)

concentration(ppm) (ppm) (2) (G)

>500 35 3.8 315 35 1.62275 575

165 _ _ 73 -.163 60 32 89 60 32199 56 27 112 56 27220 60 24 127 59 23


13/15

Review of recent field tests 143

TEST1 TEiT 2 TEST 3

Fig. 12. Comparison of observed and SLAB-predicted plume centerline concentrations at aheight of 1 m for the Goldfish HF test series.

Distance (m)Fig. 13. Comparison of observed and SLAB predictedplume crosswind concentrations for Goldfish HF Test 1(1 km downwind and 1 m height).

the underlying physical processes and often the statist-ics generated are meaningless. The importance ofdeveloping a physical understanding of a modelsperformance has been well-recognized within the at-mospheric sciences community (Knox and Walton,1984; McNaughton et al., 1986). In order to gain thisinsight, the evaluation must be tailored to the specificmodel under investigation, the available experimentaldata, and the specific application for which the modelis to be used. The steps usually followed in this processare given below.

(1) Evaluation of the accuracy of the inputmeteorological data and the gas concentrationdata used for comparison.

(2) Examination of model structure including theaccuracy of the mathematical framework, therealism of the model representation of import-ant physical processes, and the appropriatenessof assumptions used in the model when appliedto real situations.(3) Sensitivity analysis of the model to uncertaintiesin the input meteorological and source data.

(4) Testing of the model predictions against observ-ations including both laboratory and field-scaleexperiments.

Since dense gas dispersion is a relatively new field,little had been done to develop a methodology formodel evaluation until recently. Mercer (1986) con-ducted a review of the methods used to validate densegas dispersion models and concluded that there havebeen few, if any, attempts to objectively compare anumber of dispersion models using a common set ofcriteria. Authors have often verified their own modelsand in essentially all cases made only subjective,qualitative assessments on how well the model predic-tion compares with the data.

Sufficient good quality data from a variety of densegas releases now exist (Ermak et al ., 1987) such thatsystematic model evaluation is possible. A methodol-ogy for evaluating dense gas dispersion models hasrecently been developed (Ermak et al ., 1988b) whichrelies heavily on physical understanding of the pro-cesses characteristic of dense gas releases in the atmos-phere. Dense gases act as intrusions, with differentphysical properties, which significantly displace andalter ambient atmospheric flow. Dense gas dispersionis dominated by effects which are not observed in thedispersion of trace emissions; there is a reduction inturbulent mixing due to stable stratification of thedense layer and gravity flow keeps this dense layer lowand wide. It is important that the tests and com-parisons which make up the evaluation address those


14/15

744 RONALD P. KOOPMA~ et ul.processes which are dominant for dense gas releases. and the 1D cross-wind averaged SLAB model haveBased on analysis of these processes, four key plume been improved with the addition of phase-changeparameters emerge for comparison. submodels and evaluated against dense-gas data sets.

(1) Maximum gas concentration as a function ofdownwind distance from the source.

(2) Average ground-level plume centerline gas con-centration as a function of downwind distance.

(3) Plume half-width.(4) Plume height.

Sufficient good quality dense-gas data now existsuch that the evaluation of dense-gas models can bedone in a systematic, quantitative and scientific fash-ion. A methodology based on the physical ckarac-teristics of dense-gas releases has been developed. Asystematic evaluation of all models in common use hasnot yet been performed.

The piume height and width comparisons allowassessment of the combined effects of gravity flow andmixing with the atmosphere. The recommended com-parison technique is based on developing ratios be-tween the four model-predicted parameters and theexperimentally-observed values. Ratios allow com-parison over a wide range of values from high concen-trations where gravity effects dominate to trace levels,where toxicity is still an issue. Ratio methods alsoaliow determination of the bounds of model error andthe limits of model appli~bility.

Acknow ledgement -Work performed under the auspices ofthe U.S. Department of Energy by the Lawrence LivermoreNational Laboratory under contract No. W-7405-ENG-48.

In addition, the characteristic averaging time of themodel must match that of the data. The model aver-aging time must be known and if possible adjusted tomatch that of the sensors, Ifmodel adjustments cannotbe made, then the data must be averaged to match themodel. A maximum concentration has meaningonly if referred back to the time constant characteristicof either the measurement or the model.

Disclaimer--This document was prepared as an account ofwork sponsored by an agency of the United States Govern-ment. Neither the United States Government nor the Univer-sity of California nor any of their employees, makes anywarranty, express or implied, or assumes any legal liability orresponsibility for the accuracy, completeness, or usefulness ofany information, apparatus, product, or process disclosed, orrepresents that its use would not infringe privately ownedrights. Reference herein to any specific commer~i~ products,process, or service by trade name, trademark, manufacturer,or otherwise, does not necessarily constitute or imply itsendorsement, recommendation, or favoring by the UnitedStates Government or the Universitv ofcalifornia. The viewsand opinions of authors expressed herein do not necessarilystate or reflect those of the United States Governmentthereof, and shall not be used for advertising or productendorsement purposes.

REFERENCESSUMMARY

Certain key dense-gas effects, such as stable strati-fication and gravity-driven flow have been identifiedas the factors making dense gases behave differentlywhen released into the atmosphere. Dense-gas ROWSdisplace the ambient atmospheric flow and modifyambient turbulent mixing.Chemical and physical processes, such as flashing toaerosol and vapor, which occur for many commonlyused hazardous chemicals contribute strongly todense-gas effects by creating heavy, cold, liquid- ladenclouds which behave very differently from traceemissions of the same substances. Chemical andthermodynamic effects are unique to each materialand can have a significant effect on dispersion dis-tance.

Blackmore D. R., Herman M. N. and Woodward J. L. (1982)Heavy gas dispersion models. J. Haz. Mat. 6, 106-128.Blewitt D. N., Yohn J. F., Koopman R. P. and Brown T. C.(1987a) Conduct of anhydrous hydrofluoric acid spillexperiments. AIChE lnternational Conference on VaporCloud Modeling Boston, Massachusetts.Blewitt D. N., Yohn J. F., Koopman R. P., Brown T. C. andHague W. J. (1987b) Effectiveness of water sprays onmitigating anhydrous hydrofluoric acid releases. AIChEInternational Conference on Vapor Cloud Modeling, Bos-ton, Massachusetts.Blewitt D. N.. Yohn J. F. and Ermak D. L. (1987~) Anevaluation of SLAB and DEGADIS heavy gasdiskionmodels using the HF spill test data. AIChE InternationalConference on Vapor Cloud Modeling, Boston, Mass-achusetts.

Gaussian and other trace-gas based models wereintended to describe the dispersion of a tracer in anormal atmosphere and are not adequate for descri-bing dense-gas releases. Comparison to experimentalresults for several materials show that errors of up to afactor of ten can occur if trace gas models are used fordense-gas problems.

Brown T. C., Cederwall R. T., Ermak D. L., Koopman R. P.,McClure J. W. and Morris L. K. (19891F&on Series DataReport, 1987LNG Vapor Barrier Verffi cation Field Tri als,UCID- , Lawrence Livermore National Laboratory,Livermore, Caliiomia (in preparation).Ghan S. T. (1988) FEM 3A-A Fin&e Element Model/or t heSimulat ion qfGps Transport und Di spersion: Users M anual.UCitL-21043. Lawrence Livermore Nationat Laboratory,Livermore, California.

Two different approaches to dense-gas moddinghave produced models capable of predicting cloudbehavior in three-dimensional, time-dependent detailin complex terrain or averaged over both time andspace. The 3D conservation equation model FEM3

Chart S. T. and Ermak D. L. (1983) Recent progress inmodeling the atmospheric dirpersion of heavy gases overvariable terrain using the threedimensional conservationequations. Proceedings of the IU TAM Symposbn on At -mospheri c D ispersion of H eavy Gases and Srrcal l Part icl es,Ddft Univ. of Technology, The Hague, The Netherlands.Chart S. T, Ermak D. L. and Morris L. K. (1987a) FEM3model simulations of sekctsd Thomey I&and Phase ITrials. J. Huz. M at. 16.


15/15

Review of recent field tests 745Chan S. T., Rodean H. C. and Blewitt D. N. (1987b) FEM3 Koopman R. P. (1987) Atmospheric dispersion of large-scalemodeling of ammonia and hydrofluoric acid dispersion. spills. Proceedings from the AI CHE Symposium Seri es,AIChE International Conference on Vapor Cloud 1986, Cryo. Prop., Processes A ppl . Vol . 82, No . 251, pp.Modeling, Boston, Massachusetts. 141-159.Crowe C. T. and Comfort W. J. III (1978) Atomizationmethods in single-component two-phase flows. 1st Inter-national Conference on Liquid Atomization and SpraySystems, 45-50.Dennis R. L. (1985) Issues, design and interpretation ofperformance evaluations: ensuring the Emperor hasclothes. In Air Pol lut ion Modeling and Its Application(edited by C. De Wispelaere, F. A. Schiermeier and N. V.Gillani), pp. 41 l-424: Plenum Press, New York.Ermak D. L.. Chaoman R.. Goldwire H. C., Jr. Gouveia F. J.

Leone J. M., Jr., Rodean H. C. and Chan S. T. (1985) FEM3Phase Change Model, UCID-20353, Lawrence LivermoreNational Laboratory, Livermore, California.McNaughton D. J., Atwater M. A., Bodner P. M. and WorleyG. G. (1986) Eval uat ion and A ssessment o f M odels forEmergency Response Planning, TRC Environmental Con-sultants, Inc., TRC Project No. 3088-R31, report preparedfor the Chemical Manufacturers Association.McQuaid J. (1976) Some experiments on the structure ofstably stratified shear flows. Technical Paper, p. 21, Safetyin M ines Research Establ ishment, Sheffield, U.K.McQuaid J. (1983a) Large-scale experiments on the disper-sion of heavy gas clouds. Presented at the IUTAM Sym-posium, Delft.

, .and Rodean H. C. (1987) Heavy Gas Dispersion TestSummary Report, UCID-21102, Lawrence LivermoreNational Laboratory, Livermore, California.Ermak D. L. and Merry M. H. (1988b) A Methodology forEvaluating Heavy Gas Dispersion Models, UCRL-21025,Lawrence Livermore National Laboratory, Livermore,California.Ermak D. L., Rodean H. C., Lange R. and Chan S. T. (1988a)A Survey of Denser-t han-Ai r At mospheric Di spersion Mo-dels, UCRL-21024, Lawrence Livermore National Lab-oratory, Livermore, California.Fox D. G. (1981) Judging air quality model performance, asummary of the AMS workshop on dispersion modelperformance, Woods Hole, Massachusetts, 8-11 Sep-tember 1980. Bul l . Am . met. Sot. 62(5), 599609.Fox D. G. (1983) Uncertainty in air quality modeling. BullAm . met. Sot . 65(l ), 27-36.Goldwire H. C., Jr., Rodean H. C., Cederwall R. T., KansaD. J., Koopman R. P., McClure J. W., McRae T. G.,Morris L. K., Kamppinen L., Kiefer R. D., Urtiew P. A.and Lind C. D. (1983) Coyote Series Data Report,LLNL INWC 1981 LN G Spil l Tests Di spersion, VaporBurn, and Rapid-Phase Transit ion, Vols. land Z,UdD-19953 Lawrence Livermore National Laboratory, Liver-more, California.Goldwire H. C., Jr. (1985)Desert Tortoise Series Dat a Report:1983 Pressurized Ammoni a Spil ls, UCI D- 20526, LawrenceLivermore. National Laboratory, Livermore, California.Goldwire H. C.. Jr. (1986) Lame-scale ammonia soil1 tests. InChemi cal Engineeri ng Progi ess, pp. 3541. _Havens J. A. (1982) A review of mathematical models forprediction of heavy gas atmospheric dispersion. J. Chem.E. Symposi um Seri es No. 71.Havens J. A., Spicer T. 0. and Schreurs P. J. (1987)Eoaluat ionof 3-D hy drodynamic computer models for predicti on ofLN G vapor dispersion in the atmosphere. Final ReportGRI-87-0173.Hunt J. C. R., Rothman J. W. and Britter R. E. (1983) Somephysical processes involved in the dispersion of densegases. IUTAM Symposium on Atmospheric Dispersion ofHeavv Gases and Small Particles. Delft. The Netherlands.Knox J: B. and Walton J. J. (1984) Modeling regional toglobal air pollution. In Atmospheric Science and PowerProduction (edited by Randerson D.). Technical Informa-tion Center, Office of Scientific and Technical Information,U.S. Dept. of Energy.Koopman R. P., Cederwall R. T., Ermak D. L., GoldwireH.-C., Jr., Hogan W. J., McClure J. W., Mckae T. G.,Moraan D. L.. Rodean H. C. and Shinn J. H. (1982a)Analysis of Burro series 40-m3 LNG spill experiments. JIHaz. Mat. 6, 43-83, also published as Dense Gas Di sper-sion. Elsevier Scientific, Amsterdam.Koopman R. P., Baker J., Cederwall R. T., Goldwire H. C.,Jr., Hogan W. J., Kamppinen L. M., Kiefer R. D., McClureJ. W., McRae T. G., Morgan D. L., Morris L. K., SpannM. W., Jr. and Lind C. D. (1982b) Burro Series Dat a ReportLL NL /NW C 1980 LN G Spil l Tests, UCI D-19075, Law-rence Livermore National Laboratory, Livermore, Cali-fornia.

McQuaid J. (1983b) Observations on the current status offield experimentation of heavy gas dispersion. Presented atthe IUTAM Symposium, Delft.McQuaid I. (1987) Proceedings of the Symposium on HeavyGas Di spersion Trial s at Thorney Island (J. Haz. M at. 16).McRae T. G., Cederwall R. T., Goldwire H. C., Jr., HippleD. L., Johnson G. W., Koopman R. P., McClure J. W. andMorris L. K. (1984a) Eagle Series Data Report: 1983Ni tr ogen Tetroxi de Spil ls, UCI D- 20063, Lawrence Liver-more National Laboratory, Livermore, California.McRae T. G., Goldwire H. C., Jr. and Koopman R. P. (1984b)Anal ysis of Large-Scale LNG/ Wat er RPT Explosions,UCRL-91832, Lawrence Livermore National Laboratory,Livermore, California.McRae T. G. (1985)Anal ysis and M odel/D ata Comparisons ofLar ge-Scale Releases of Ni t rogen Tet roxi de, U CID -20388,Lawrence Livermore National Laboratory, Livermore,California.Mercer A. (1986) Methods of Vali dati ng M odels of Dense-Gas Dispersion: A Review. Health and Safety Executive,Research and Laboratory Services Division, SafetyEngineering Laboratory, Sheffield, U.K.Morgan D. L., Jr., Morris L. K., Chan S. T., Ermak D. L.,McRae T. G., Cederwall R. T., Koopman R. P., GoldwireH. C.. Jr.. McClure J. W. and Hoaan W. J. (1984)Phenom-enology and Modeling of Li quefied N atural Gas VaporDispersion, UCRL-53581, Lawrence Livermore NationalLaboratory, Livermore, California.Puttock J. S., Blackmore D. R. and Colenbrander G. W.(1982) Field experiments on dense gas dispersion. J. Haz.M at. 6, 1341.Puttock J. S. and Colenbrander G. W. (1985) Dense-gasdispersion-experimental research. Presented at the HeavyGas Workshop, Toronto.Rodean H. C., Hogan W. J., Urtiew P. A., Goldwire H. C., Jr.,McRae T. G. and Morgan D. L., Jr. (1984) Vapor BurnAnal ysis for the Coyot e Series LN G Spil l Experim ents,UCRL-53530, Lawrence Livermore National Laboratory,Livermore, California.Rodean H. C. and Chan S. T. (1986) Generalized Phase-Change Submodels in he FEM 3 M odel or Gas Di spersion,UCID-20817, Lawrence Livermore National Laboratory,Livermore, California.Venkatram A. (1986) Model Evaluation. In A Short Courseon Air Pol lut ion Model ing. American MeteorologicalSociety, Boston, Massachusetts.Vilain J. (1986) Two-phase flows in major technologicalhazards. Presented at the von Karman Institute for FluidDynamics Lecture Series.Wheatley C. J. and Webber D. M. (1984) Aspects of theDi spersion of Denser-Than-Ai r Vapours Relevant t o GasCloud Explosions, Contract Report SR/OO7/80/UK/H/EAEC/UKAEA No. X11/829/84-EN.

Dispersion of Large Spills of Denser-than-Air

Documents