Table Of Content

Volume106,Number1,January–February2001 Journal of Research of the National Institute of Standards and Technology [J.Res.Natl.Inst.Stand.Technol.106,231–278(2001)] In Situ Burning of Oil Spills Volume 106 Number 1 January–February 2001 David D. Evans, George W. FormorethanadecadeNISTconducted turesandtrajectoryofwindblownsmoke Mulholland, Howard R. Baum, researchtounderstand,measureandpre- plumesintheatmosphereandestimatethe William D. Walton, and Kevin B. dicttheimportantfeaturesofburningoil groundlevelsmokeparticulateconcentra- onwater.Resultsofthatresearchhave tions.Predictionsusingthemodelwere McGrattan beenincludedinnationallyrecognized testedsuccessfullyagainstdatafrom guidelinesforapprovalofintentional large-scaletests.ALOFTsoftwareisbeing National Institute of Standards and burning.NISTmeasurementsandpredic- usedbyoilspillresponseteamstohelp Technology, tionshaveplayedamajorroleinestab- assessthepotentialimpactofintentional Gaithersburg, MD 20899-8640 lishinginsituburningasaprimaryoilspill burning. responsemethod.Dataaregivenforpool [email protected] fireburningrates,smokeyield,smokepar- Keywords: ALOFT;combustion;large [email protected] ticulatesizedistribution,smokeaging, eddysimulation;offshoredrilling;oil [email protected] andpolycyclicaromatichydrocarboncon- spills;poolfires;smokeplumes;smoke [email protected] tentofthesmokeforcrudeandfueloil sampling;smokeyield. [email protected] fireswitheffectivediametersupto17.2m. Newuser-friendlysoftware,ALOFT,was developedtoquantifythelarge-scalefea- Availableonline: http://www.nist.gov/jres 1. Introduction One of the risks of oil drilling and transportation is The1989oilspillfromtheExxonValdeztankeronto thataccidentscanoccurreleasingnaturalcrudeoilorits thewatersofPrinceWilliamsSoundinAlaskafocused refinedproductsinoilspills.Oilcontaminationofland nationalattentiononoilspills.Anestimated42million orwaterisanenvironmentalhazardtolife.Historically liters of oil were released from the ship into the water. oilspillresponsehasbeenlimitedtovariousmechanical Some of the oil, driven by winds and currents, was means of recovering the spilled oil from land or water depositedontheshorelineofPrinceWilliamsSound.At andthendisposingoforreprocessingthewaste.Gener- the time of that spill, NIST and others were already allylargeamountsofoilcontaminatedmaterialsneedto engagedintheevaluationofburningasaresponsetooil be removed and treated. Mechanical recovery of oil in spills. Industry was beginning to produce fire resistant areas such as rocky shorelines, marshlands, and in ice- boomsthatcouldbeusedtoconfineoilspilledonwater laden waterways is impractical. Industry needs to have inordertoburnitinplace.Itisalittleknownfactthat alternative technologies to mechanical recovery for oil using a fire resistant boom, approximately 57 000 L of spillresponse.Oneofthepossiblealternativesistoburn oilfromtheExxonValdezthathadbeeninthewaterfor the oil in place—in situ burning. nearlytwodayswasconfinedandburned.Theresulting fire lasting approximately 45 min consumed all but 1100 L of residue that remained in the boom [1]. 231 Volume106,Number1,January–February2001 Journal of Research of the National Institute of Standards and Technology Any response to oil spills includes considerations of buoyant flows were exploited to quantify wind driven oil containment, recovery, disposal and the logistics of smoke trajectories in the atmosphere and estimate the delivering adequate response equipment quickly to the downwind particulate concentrations. Finally, user spill site. Particularly in remote areas, the use of burn- friendlysoftwarewasdevelopedtoprovidethebenefits ingasaoilspillresponsemethodisattractive.Burning of this research to emergency responders and local au- requires a minimum of equipment, and because the oil thorities. is gasified during combustion, the need for physical Burning may be thought of as an emerging technol- collection,storage,andtransportofrecoveredproductis ogy for response to oil spills. NIST has been a major reduced to the few percent of the original spill volume contributor to the science and technology used for safe that remains as residue after burning. andeffectiveinsituburningofoilspills.Amajorfactor Oil spilled on water begins to spread naturally away aiding the NIST research effort has been the relation- fromthesource.Wind,wavesandcurrentsmovetheoil shipsbuiltupovertheyearswithanumberoforganiza- overthewatersurfaceandalsocontributetoemulsifica- tions. Since 1985, NIST has enjoyed long term and tion.Thinlayersofoilonwatercannotbeburned.Soin substantialfundingfromtheMineralsManagementSer- order to ignite and sustain the burning of oil spills, the vice, U.S. Department of the Interior. NIST has also oilneedstobeconfined.Insomecasesnaturalconfine- worked closely with partners including the United ment such as ice leads provides the confinement. In StatesCoastGuard,U.S.DepartmentofTransportation, general,respondersneedtoprovideameanstothicken the Technology Development and Technical Services theoillayerandconfinetheburning.Todothis,artifi- BranchofEnvironmentCanada,theEnvironmentalRe- cial confinement is needed within fire resistant booms. sponse Team of the U.S. Environmental Protection Burningoilspillsin-placenormallyproducesavisi- Agency, Alaska Clean Seas (an oil industry coopera- blesmokeplumecontainingsootandothercombustion tive), and the National Research Institute for Fire and products produced in the burning. Lack of knowledge Disaster in Japan. abouttheextentoftheareaaffectedbythesmokeplume producedbyburningofcrudeoilspillsandthepossibil- ity of undesirable combustion products carried in the 2. Background plume have led to public concerns over the effects of intentional burning large crude oil spills. Unresolved Theinsituburningofoilspillshashistoricallybeen questions about personnel and equipment safety from regarded as a response method of last resort. The dy- the heat and thermal radiation produced by large fires namicsofignitionandsustainedburningofoilspillshas has also hampered application of burning to oil spills. not been understood. An early attempt to analyze the In the decision process for approval of intentional process of oil spill burning and to set down guidelines burningofoilspills,localauthoritiesneedtohavetools forwhenitwouldandwouldnotbesuccessfulwasthe toquantifythelikelybenefitsoftheburningintermsof workofThompsonetal.in1979[2].Priortothatwork, oilremovalandthelikelyconsequencesintermsofthe testing had been performed largely for the demonstra- fire generated smoke plume. The in situ oil spill re- tion of various products being developed to ignite and search program at the National Institute of Standards promote the burning of spills. Thompson reviewed the andTechnology(NIST)wasdesignedtodevelopquanti- useofinsituburninginresponsetospillaccidents.His tativeinformationandsoftwaretoolstoaidauthoritiesin review provides a perspective on the technology in use makinginformeddecisions.Thelackofthisinformation inthe1970sandthemixedresultsinpracticerespond- wasseenasanimpedimenttotheacceptanceanduseof ingtomajoroilspillincidents.Ingeneralsinceburning this emerging technology. was regarded as a response method of last resort, oil In order to do this NIST designed a comprehensive couldhavebeeninthewaterfordaysbeforeburningwas program of integrated measurements and predictions. attempted. The longer oil stays in the water allowing Understanding the process of burning oil on water and volatilecomponentstoevaporate(weathering)andwater itsconsequencesinvolvedfireexperimentsinNISTlab- to emulsify with the oil by wave action, the harder the oratories,infireresearchfacilitiesinJapan,atnewfield spill is to ignite and burn. For in situ burning to be facilities specifically constructed for this research in widelyeffective,itneededtobeconsideredasoneofthe Mobile, Alabama, and at large scale oil burn experi- primary oil spill response methods. ments in Alaska and offshore Newfoundland, Canada. In 1985, NIST began studies of oil spill burning on New measurement instruments were developed to im- open waters and in channels formed by broken ice in plement successful laboratory techniques into robust support of the Safety in Offshore Drilling program of air-deployableself-poweredmeasurementandsampling the Minerals Management Service, Department of the packages. New computational methods for fire driven Interior. The original intent of the program was to find 232 Volume106,Number1,January–February2001 Journal of Research of the National Institute of Standards and Technology means to generalize the experimental results of Brown from 0.085 m to 0.6 m in diameter. The smallest fires, andGoodman[3]andSmithandDiaz[4]fortheburn- 0.085 m diameter, were conducted in the Cone ingofoiliniceleads.SmithandDiazburnedoilonthe Calorimetertodeterminetheeffectiveheatofcombus- watersurfaceconfinedbyiceblocks(simulatingbroken tion for the crude oils and evaluate smoke yield using ice formations) in the EPA OHMSETT tank facility in three different measurement methods. The Cone Leonardo, NJ. The results of the experiments were en- Calorimeter,showninFig.1,ismoreformallyknownas couragingwithrespecttotheefficiencyofinsituburn- Standard Test Method for Heat and Visible Smoke Re- ingastypicallybetterthan50%oftheoilspilledinthe leaseRatesforMaterialsandProductsUsinganOxygen iceformationcouldberemovedbyburningandinsome Consumption Calorimeter [10]. The name of the ap- cases over 90% was possible. paratus,ConeCalorimeter,isderivedfromtheshapeof It was quickly realized that in order to gain accep- theheaterusedtoirradiatesamples.Theheatercoilsare tanceforinsituburningofoilspill,beingabletoquan- formed along the inner surface of a truncated cone. By tifytheamountandeffectsofthesmokefromtheburn imposing additional thermal radiation on a small sam- wasofgreaterimportancethanquantifyingtheamount ple, the sample is made to burn as if it were in the ofoilremovedfromthewaterbyburning.Thelatterwas interiorofalargerfire.Themajormaterialflammabil- thefocusofmanyofthestudiesfundedbyindustry.To itycharacteristicscanbeevaluatedusingthislaboratory address the burning, smoke production, and smoke apparatus. These include: rate of heat release, effective transport issues, it was clear that expertise in combus- heatofcombustion,totalheatrelease,ignitibility,mass tion, fire dynamics, computational fluid flow, particle loss rate, smoke specific extinction area, and yields of measurement, chemical analysis, and large-scale fire various gaseous species and particulate. measurementswouldbeneeded.NISThadthecapabili- Alargercalorimeterapparatuscapableofaccommo- tiestoassemblethisinterdisciplinaryteamandthefacil- dating samples up to 0.6 m in diameter was used to ities to support its experimental and computational ef- provide NIST data about the amount and properties of fort. smokeasthediametersofthefiresincreased.Extensive AspartoftheNISTresearchprogram,insituburning instrumentation and sampling hardware were added to technologieswerereviewedandinputonresearchneeds the exhaust flow from the hood as shown in Fig. 2. wasgatheredperiodicallythroughworkshopsconducted Samplesdrawnfromtheexhausthoodductwereusedto byNIST[5,6,7].Othersourceshavealsoreviewedthe quantify the amount of each major combustion product technologyofoilspillburningandotherresponsemeth- generatedperkilogramofcrudeoilburned,thechemi- ods [2, 8, 9]. calcompositionofthesmokeincludingpolycyclicaro- matic hydrocarbon (PAH) content, the particulate size distributionofbothfreshandagedsmoke,andtheoxy- 3. Experimental Facilities genconsumedinthecombustionprocess.Oxygencon- sumptioncalorimetryisusedtomeasuretheheatrelease Tounderstandandquantifytheimportantfeaturesof rate of the fire, which is the primary quantity used to insituburningitwasnecessarytoperformthreescales characterize burning intensity. To further characterize of experiments. Laboratory tests furnished property thecombustionprocess,additionalinstrumentationwas data,experimentsutilizinglarge-scaleoutdoorburnfa- usedtomeasureradiantheatfluxfromtheflameandthe cilities provided mesoscale data and means to develop mass loss rate of the burning fuel. andevaluateinstrumentation,andfinally,actualburnsof spilledoilatseaprovideddataoninsituburningatthe 3.2 NRIFD Facilities anticipated scale of actual response operations. In this research program, there has been continued interaction Therelativelysmall,0.6mdiameter,firesprovideda between findings from measurements on small fire ex- meansofmeasuringfirecharacteristicsundercontrolled periments performed in the controlled laboratory envi- conditions,butaretoosmalltoprovideanadequatetest ronmentsofNISTandtheNationalResearchInstituteof of measurement equipment being developed for field FireandDisaster(NRIFD)inMitaka,Tokyo,Japan,and use.ThroughthecooperationofNRIFDjointstudiesof large fire experiments at facilities like the USCG Fire crude oil burning characteristics were conducted. SafetyandTestDetachmentinMobile,Alabamawhere NRIFD maintains a fire test facility in which crude oil outdoor liquid fuel burns in large pans are possible. pools up to 3 m indiameter are burned, with all of the combustion products collected in a large hood system. Figure3showsa2mdiametercrudeoilfireburningin 3.1 NIST Facilities the 24 m(cid:4)24 m(cid:4)20 m high test hall. This facility At NIST, two major facilities were used to perform could accommodate fires that are large enough so that measurements on crude oil pool fires ranging in size 233 Volume106,Number1,January–February2001 Journal of Research of the National Institute of Standards and Technology Fig.1. Conecalorimeterapparatus. sampling packages designed for mesoscale field tests Burnswereconductedinanominal15msquaresteel couldbeevaluatedbycomparisontotraditionallabora- burn pan constructed specifically for oil spill burning. torymeasurements.Todothistheexhaustsystemforthe Theburnpanwas0.61mdeepandwasconstructedwith buildingwasinstrumentedsothatmeasurementssimilar twoperimeterwallsapproximately1.2mapartforming to those performed in the NIST facility could be made an inner and outer area of the pan. The inside dimen- byeffectivelyusingtheentireNRIFDtestbuildingasa sions of the inner area of the pan were 15.2 m by 15.2 smoke collection hood. m.Thetwoperimeterwallswereconnectedwithbaffles andthespacebetweenthewalls,whichformedtheouter area of the pan, was filled with bay water during the 3.3 USCG Facility burns. The base of the pan was 6 mm thick steel plate The mesoscale burns of crude oil were carried out and the walls were 5 mm thick steel plate. The tops of under the direction of NIST at the United States Coast the walls were reinforced with steel angle to prevent Guard(USCG)FireandSafetyTestDetachmentfacility warping during the burns. The base of the pan was on Little Sand Island in Mobile Bay Alabama. Little located on ground level and was reinforced with steel Sand Island is approximately 0.2 km2 in size and in- beams on steel footers under the pan. Water fill pipes cludes decommissioned ships docked in a lagoon. The wereconnectedtoboththeinnerandouterareasofthe ships and facilities on the island have been used for a pan. Water was pumped directly from Mobile Bay into widevarietyoffull-scalemarinefiretests.Figure4isa boththeinnerandouterareasofthepan.Theinnerarea photograph of a burn in progress, and Fig. 5 is a plan ofthepanwasfilledwithapproximately0.5mofwater view of the portion of the island used for the oil spill andthecrudeoilwasaddedontopofthewater.Anoil burns. spill containment dike approximately 0.5 m high was constructed 4 m from the outer edge of the pan. 234 Volume106,Number1,January–February2001 Journal of Research of the National Institute of Standards and Technology Fig.2. NISTlargecalorimeterwithsmokesamplingequipmentinstalled. Threedifferentprimaryburnareaswereused.These in large laboratory tests indicated that the duration of areasconsistedofthefullinnerpanwithanareaof231 sampling would have to be nominally 10 min. Instru- m2 and partial pan areas of 114 m2 and 37.2 m2. The mentplatformsevaluatedforthispurposeincludedtow- partialpanareaswereachievedbypartitioningacorner ers, manned aircraft, fixed winged and helicopter re- oftheinnerpanwith0.14mby0.14mtimberscovered motely piloted aircraft, and balloons or mini-blimps with sheet steel. Plywood skirts 0.3 m deep were at- [12].Itwasdecidedthatatetheredmini-blimpwouldbe tachedtothetimbersbelowthewatersurfacetoprevent the primary means of positioning instrumentation for the oil from flowing under the timbers. An effective soot collection in the smoke plume. A 5.6 m long and diameterwascalculatedforeachoftherectangularburn 2.3mdiametertetheredheliumfilledmini-blimpcould areas. The effective diameter is the diameter of circle be controlled by ground personnel. Control and instru- with the same area as the rectangular burn area used. mentpayloadtestsusing15mdiameterfireswerecar- The effective diameters for the three areas are 17.2 m, ried out at the Navy’s Farrier Fire Training Facility in 12m,and6.88m.Additionaldetailsoftheconstruction, Norfolk, Virginia to evaluate operational limits of the installed instrumentation, and operation of this major mini-blimp. This size mini-blimp can carry a 4 kg in- test apparatus for oil burns are given by Walton et al. strument package to one hundred meters above ground [11]. level. It can be readily moved from one location to an- Variousmeanswereexploredtoobtainsamplesfrom other. Launch and recovery procedures are simple, and the wind blown smoke plume. Experience with smoke littletimeisneededtolearnhowtosafelymaneuverthe samplingusingbatterypoweredlightweightinstruments blimp.Mini-blimpswereusedbothinthemid-scalepan 235 Volume106,Number1,January–February2001 Journal of Research of the National Institute of Standards and Technology Fig.3. Two-meterdiametercrudeoilfireinNRIFD,Japantesthall. burntestsandintheatseatestsforpositioningairborne cases, where the burn was of sufficient duration, two instrumentation packages. Tethered mini-blimps were packages were deployed sequentially. also used to position airborne weather stations about Thesamplingpackagesconsistedofbattery-powered 50mabovethetestsitetocontinuouslymeasureatmo- pumpsthatdrewsamplesthroughfiltersanddischarged sphericconditionsduringexperimentsandtransmitthat aportionofthegasintoacollectionbag.Filtersamples data via radio to a ground station. wereanalyzedinthelaboratoryforPAHandVOCcon- During burns at the Mobile, Alabama test site, air- centrations. Particulate size distribution was measured bornesampleswerecollectedforbothlaboratoryanaly- usingacascadeimpactor.Inaddition,smokeparticulate sisandanalysisonthegroundimmediatelyfollowingthe wascollectedonathermophoretictransmissionelectron burns.Allsamplingpackagesweresuspendedapproxi- microscopegrid(TEMgrid)andanalyzedusingatrans- mately 60 m below the mini-blimp, Fig. 6. The mini- mission electron microscope to determine particle blimp was positioned downwind from the fire with the shape. sampling package centered in the smoke plume. The elevationanddownwindpositionofthesamplingpack- 3.4 Offshore Experiments age varied with each burn as a function of the plume position.Typically,samplingpackagesremainedinthe Everything that was learned about measuring smoke plume for 600 seconds. That permitted an adequate properties in plumes from large fires on land was uti- sample to be collected and allowed the natural fluctua- lized in a large scale burn at sea burn experiment con- tionsintheplumetobeaveraged.Sincetheliftcapacity ducted under the direction of Environment Canada 25 of the mini-blimp was limited, in general only a single km off the coast of Newfoundland, Canada near St. samplingpackagecouldbedeployedatatime.Insome John’s on August 12, 1993. This experiment known as 236 Volume106,Number1,January–February2001 Journal of Research of the National Institute of Standards and Technology Fig.4. U.S.CoastGuardSafetyandFireTestDetachmentmesoscaleburnfacility. NOBE (Newfoundland Offshore Burn Experiment) in- initially increase with increasing pool diameter, but cluded the sponsorship and participation by more than reaches a plateau for large fires. Large-scale experi- 25 Canadian and U.S. government agencies and indus- ments were conducted using two different methods to tries.Thisexperimentprovidedtheopportunitytomake determinethefuelmasslossrateduringburning—mea- measurementsandevaluateequipmentperformanceata surementofinitialvolumeandburntimeandmeasure- burn at the anticipated scale of actual response opera- mentoffluidpressurechanges.Forapplicationtoactual tionsatsea,Fig.7.NISTtookontheresponsibilityfor response,onlythelargescaleexperiments(burndiame- particulate and gas sampling from the smoke plume. ters greater than 5 m) are of interest. These measure- Instrument packages developed by NIST were sus- ments were conducted at the Mobile, Alabama facility, pended below a helium filled mini-blimp tethered to a over a period of several years. Generally the measure- vesselstationedapproximately300mdownwindofthe mentsfromthelastseriesoftestarethemostreliablefor burning fuel contained in a fire resistant boom. The burningratedeterminationsassignificantadvanceswere elevation of the mini-blimp and its position could be made in measurement technique based on experience. adjusted to keep the instrument package positioned in The burning of the crude oil was observed to take the plume for sample collection, Fig. 8. place in four distinct phases. The four phases were; 1) spreading, 2) steady burning, 3) steady burning with boilingofthewaterbelowtheoillayer,and4)transition 4. Experimental Results to extinction. The spreading phase lasted from 80 s to 180sasflamesspreadoverthesurfacefromthesingle ignitionpointontheupwindsideofthepantocoverthe 4.1 Burning Rate entirefuelsurface.Oncetheentireoilsurfacewascov- The burning rate of oil on water was measured to eredwithflames,theburningcontinuedatasteadyrate quantify the removal rate potential of in situ burning. untilthewaterbelowtheoilsurfacebegantoboil.The Theburningrateperunitareaoffuelspillsisknownto onset of boiling was characterized by a noticeable 237 Volume106,Number1,January–February2001 Journal of Research of the National Institute of Standards and Technology recenttestsfromtheliquidlevelinthepanasmeasured bythepressuretransducer[11].Theoutputofthepres- sure transducer was calibrated in salt water and con- vertedtooildepthusingthespecificgravityoftheoil. The specific gravity of the oil was 0.846(cid:1)0.001 as measured using the mechanical oscillator technique with an accuracy of (cid:1)0.001. The salt content of the waterinthepanwasmeasuredbeforeeachtestusingthe sodium ion electrode method with an accuracy of (cid:1)0.01%. The oil surface regression rate was calcu- lated using a least squares linear fit of the pressure transducer output over the time from full pan involve- menttothebeginningofextinction.Thedatashowedno differenceintheburningratebeforeandduringboiling. The specific mass burning rate (rate of mass loss perunitarea)wascalculatedfromthesurfaceregression rate and the density of the oil. The heat release rate was determined by multiplying the mass loss rate by the effective heat of combustion for the crude oil (41.9 MJ/kg) [11]. Figure9isagraphofthesurfaceregressionrateasa functionoftheeffectiveburndiameter.Fromthisgraph it appears that for the range of diameters used in the mesoscale burns there is no dependency of surface regression rate on burn area. The mean value is (0.062(cid:1)0.003)mm/s.Themeanvaluefortheburning rateperunitareais(0.052(cid:1)0.002)kg/s/m2andforthe heat release rate per unit area is (2180(cid:1)100) kW/m2. The scatter in the regression, burning and heat release Fig.5. SiteplanofU.S.CoastGuardmesoscaleburnfacilityonLittle SandIsland,MobileBay,Alabama. rateswasdueinparttothevariablenatureoftheburns. The wind direction and speed contributed to the wide variationinextinctionbehaviorobservedalthoughitdid not appear to affect the average burning rate. 4.2 Smoke increase in fire generated sound which resembles siz- zlingandbubblesbreakingthroughtheoilsurface.Dur- Animportantelementofthisstudywasthecharacter- ing boiling the burning rate increased to a steady rate ization of the smoke particulate, since it is the particu- which was greater than the rate prior to boiling. When late that will ultimately lead to the health and environ- thefuelwasnearlyconsumed,thefirebeganatransition mental consequences. These impacts depend on the toextinction.Thiswascharacterizedbyareasoftheoil amountofsmokeproduced,theparticulatesizedistribu- surface with no visible flames. Frequently, there were tion, and the chemical makeup of the soot. NIST pro- oscillations in the burning behavior with increased and vided innovative measurement methods and new infor- decreasedburningareaandtransitiontoandfromboil- mationonallofthesetopics.TheSmokeYieldsection ing. The burning area decreased toward the downwind describes the development and application of the Car- side of the pan until extinction. bonBalanceMethodtodeterminethesmokeyieldfrom The burning rate or the rate at which the oil was smallandlargeoilpoolfires.Theresultsoftheaerody- consumed during burning was estimated in the most namic size distribution of the soot, which determines 238 Volume106,Number1,January–February2001 Journal of Research of the National Institute of Standards and Technology Fig.6. Mini-blimpusedtosupportinstrumentpackagesforsmokeplumegasandparticulatesampling. the deposition fraction in the respiratory tract, are re- carbon Emission. To allow a more complete impact portedfromarangeofcrudeoiltypesandfiresizesin analysis,dataisalsoincludedinthissectiononthePAH the Size Distribution Section. The chemical makeup of distribution of the original crude oil and of the burn thesootincludingtheorganicandgraphiticcomponents residue.Twoadditionalsectionsareincludedtoprovide and the specific polycyclic aromatic hydrocarbons, a more detailed characterization of the soot agglomer- some of which are known or suspected carcinogens, is atesandtheirproperties.Thesootparticlesareagglom- describedinthesectiononPolycyclicAromaticHydro- erates made up of nearly spherical primary particles. 239 Volume106,Number1,January–February2001 Journal of Research of the National Institute of Standards and Technology Fig.7. OilburnsinNOBEexperiment. (cid:1) (cid:2) The size distribution of these spheres and their depen- m (cid:3) = s (cid:9), dence on fire size is discussed in the Primary Sphere 1 m f Section.Theagglomeratesgrowasaresultofparticles colliding and sticking, and the effect of this process on wherem isthesmokemasscollectedonthefilterfrom s the size distribution is reported in the Smoke Aging an exhaust stack sample, m is the fuel mass consumed f Section. during filter collection, and (cid:9)is the ratio of mass flow of air through the stack to the mass flow through the sample filter. 4.2.1 Smoke Yield Thesecondmethodofdeterminingthesmokeyieldis Smokeyieldisdefinedasthemassofsmokeaerosol referred to the carbon balance method [14, 15]. This generatedpermassoffuelconsumed.Thesmokeaero- method required a determination of the ratio of the sol collected during these experiments contained both smoke mass in a given volume to the total mass of solid material (graphitic carbon) and condensable hy- carbon in the form of gas or particulate in the same drocarbonsfromthefireplume.Twomethodsfordeter- volume. This was accomplished by dividing the smoke mining smoke yield were used in this study. The first masscollectedonafiltertothesumofthesmokemass was the flux method which measured the smoke col- and the mass of carbon contained in the forms of CO lected on a filter and the mass loss from the burning andCO.Theequationforcalculatingsmokeyield(cid:3) as 2 2 specimen [13, 14]. This type of measurement worked expressed in terms of CO and CO concentrations is 2 wellinalaboratorytestenvironmentwherealltheprod- given by: uctsofcombustionwerecollectedanddrawnthroughan exhaust stack. The defining equation for smoke yield fm (cid:3) = s . based on the flux method (cid:3) is given by: 2 [m +0.012n((cid:1)(cid:9)(CO)+(cid:1)(cid:2)(CO)] 1 s t 2 240