5.3.1 air dispersion and emission modeling x s... · emission modeling emission modeling is a...

5-25

Draft Encyclopedia X April 2002

· If monitoring is technically impracticable,modeling is preferable to no action. Forexample, because of the unconfined nature of airreleases from such units, permit applicantshistorically have had difficulty in capturing theentire plume from OB/OD units through theexclusive use of air monitoring.

· Permit applicants that propose the exclusive useof monitoring should be required to conductmodeling to verify that the full extent of releasesfrom a unit are captured at the monitoringlocations selected.

· Permit applicants that propose the exclusive useof modeling should, where feasible, be requiredto conduct monitoring to provide a comparisonto the results of air modeling.

Presented below are specific details aboutmonitoring and modeling techniques for eachenvironmental medium.

5.3.1 Air Dispersion and Emission Modeling

The models that simulate the transport anddispersion of air contaminants from the point ofrelease to potential receptors use known data on thecharacteristics of a contaminant release and theatmospheric conditions as input to calculate airconcentrations and deposition values at almost anylocation specified by the user. Dispersion modelscan be used when monitoring is impractical orinfeasible. Models also can be used to supplementair monitoring programs by filling in data gaps orinterpreting monitoring results, or to assist indesigning an air monitoring program. Dispersionmodeling is an important tool for determiningpotential exposure by the air pathway.

Although dispersion modeling is a valuable tool foran air assessment, the permit writer should recognizethe considerable limitations that exist whenevaluating a modeling analysis. The accuracy of themodels is limited by the ability of the modelalgorithms to depict atmospheric transport anddispersion of contaminants and by the accuracy and

5-26


validity of the input data. For example, most refinedmodels require the input of representativemeteorological data from a single measuring station.In reality, a release will encounter highly variablemeteorological conditions that change constantly asthe release moves downwind. EPA’s Guideline onAir Quality Models - Revised (EPA 2001)describes two types of uncertainty related tomodeling. Inherent uncertainty involves deviations inconcentrations that occur even if all data used for themodel are accurate. Reducible uncertainty isassociated with the model and the uncertain inputvalues that will affect the results. While it isimportant to represent actual conditions accuratelyby selecting the right model and using accurate andrepresentative data, it should be recognized that theresults of all modeling are subject to uncertainty.Nevertheless, models generally are consideredreasonably reliable in estimating the magnitude ofhighest concentrations that result from a release,although the estimate will not be necessarily time-and space-specific (EPA 2001). When appliedproperly, air dispersion models typically are accurateto ± 10 to 40 percent and can be used to develop abest estimate of concentrations of air pollutants(EPA 2001).

In general, a modeling analysis should follow closelythe EPA modeling guidelines presented in GuidelineOn Air Quality Models, as well as informationpresented in user’s guides and EPA risk assessmentdocuments (e.g., Human Health Risk AssessmentProtocol for Hazardous Waste CombustionFacilities). Permit writers should refer to thesedocuments when evaluating an approach tomodeling. The permit applicant should identifyclearly and justify any deviations in the applicationfrom the guidelines. Other helpful resources that aidin reviewing a modeling approach include:

· EPA. 1994. Air/Superfund National TechnicalGuidance Study Series. Volume V -Procedures For Air Dispersion Modeling AtSuperfund Sites. Office of Air Quality Planningand Standards. Research Triangle Park, NorthCarolina. February.

http://www.epa.gov/ttn/scram/guidance/guide/appw_01.pdf




http://www.epa.gov/combustion/risk.htm



http://www.epa.gov/scram001/tt25.htm#guidance



5-27


· EPA. 1994.

(EPA530-R-94-021). Office ofSolid Waste and Emergency Response. April.

· EPA. Suggestions for Auditing AssessmentAir Modeling Studies following the 1998 U.S.Office of Solid Waste Human Health andEcological Risk Protocols. U.S. EPA Region 6Center for Combustion Science andEngineering.

The following sections discuss criteria for theselection of the model, the data that are required fora modeling analysis, and evaluation of the results ofmodeling. These sections describe for the permitwriter how to evaluate a modeling analysis andprovide information about recommended airdispersion models. Each section addresses therequirements for both screening air modelinganalyses and refined air modeling analyses.

Emission Modeling

Emission modeling is a method that uses knowninformation and assumptions about an emissionsource to predict the emission rate of a givencontaminant. The information and assumptions usedin emission modeling are incorporated into emissionfactors or emission equations that then are used tocalculate emissions. Often, the factors andequations are based on monitoring and modelingresults from several similar sources. Emissionmodeling should not be confused with dispersionmodeling. Unlike dispersion modeling, whichestimates concentrations and deposition rates ofcontaminants, emission modeling (or emissionfactors and equations) estimates the rate of releaseof contaminants from a source, in units of mass pertime. Emission factors and equations have beendeveloped for a wide variety of emission sourcesand a wide variety of release conditions. Mostemission factors and equations include a built-in biastoward conservativism, so that estimated emissionrates will represent the worst-case scenario. The

http://www.epa.gov/cgi-bin/claritgw?op-Display&document=clserv:epa-cinn:4385;&rank=4&template=epa




http://www.epa.gov/earth1r6/6pd/rcra_c/protocol/audit.pdf


5-28


permit writer should verify that any emission factorsor emission equations used by an applicant arecredible and result in conservative estimates.

Under certain circumstances, emission modeling maybe used instead of emission monitoring to estimateemission rates from Subpart X units. When well-developed emission factors or equations areavailable for the specific type of unit and wastes,those factors may be used to estimate emissionsfrom a unit. Use of emission modeling may benecessary when monitoring would be difficult orimpossible. The most comprehensive collection ofemission factors and equations is found in EPA’sAP-42 Compilation of Air Pollutant EmissionFactors (EPA 1995c). Existing “BangBox” data forOB/OD operations as presented in EmissionFactors for the Disposal of Energetic Materialsby Open Burning and Open Detonation (OB/OD)are generally acceptable for use in estimatingemissions, as long as the composition of materialbeing burned or detonated at the unit is the same asthe composition of material for which the BangBoxdata were collected.

Another type of emission modeling technique is themass balance technique. Mass balance is ascreening technique that uses the mass of materialentering a system with the mass of material leavingthe system. The difference between those twoknown parameters is assumed to be the airemissions. This technique is applicable only toemission sources for which the mass of material bothentering and leaving the system is known. A permitapplicant may measure those values so that the massbalance technique can be applied.

Selection of the Dispersion Model

Selection of the proper dispersion model foranalyzing the release of a contaminant to theatmosphere is crucial to the success of the modelinganalysis. Dispersion models are developed forspecific types of sources, atmospheric conditions,terrain, locations of receptors, and chemical and

http://www.epa.gov/ttn/chief/ap42/

http://www.epa.gov/ttn/chief/ap42/

http://www.hnd.usace.army.mil/oew/tech/EmissFac/emissfac.pdf



5-29


physical processes involved. Only models that arecapable of assessing conditions against site-specificcriteria should be used in a modeling analysis.

Dispersion models are developed for eitherscreening or refined analyses. Screening models areeasier to use and require less site-specific data thanthose for refined analysis. Refined models requiremore data, but produce more realistic results. Table5.2 presents preferred screening dispersion modelsfor Subpart X units and outlines each model’scapabilities and features. Also provided below aresummary discussions of each preferred screeningmodel. It should be noted that Table 5.2 is notintended to be an exhaustive list of appropriatescreening models for Subpart X permitting, butprovides the most commonly used and mostaccepted screening models that may be applied to aSubpart X unit. Because of their versatility and easeof use, the SCREEN3 and TSCREEN models arethe most commonly used screening models.However, the models can simulate releases onlyfrom a single source; therefore, another screeningmodel or a refined model must be used to modelsites at which there are more than one source. TheCTSCREEN model is especially useful in cases inwhich complex terrain and multiple point sources arepresent.

Table 5.3 lists preferred refined dispersion modelsfor Subpart X permitting and outlines each model’scapabilities and features. Also provided below aresummary discussions of each preferred refinedmodel. As is true of the list of screening models, thelist of refined models in Table 5.3 is not intended tobe an exhaustive compilation of appropriate refinedmodels for Subpart X permitting.

It should be noted that a dispersion model has beendeveloped at the U. S. Army Dugway ProvingGrounds to specifically address release anddispersion characteristics from OB/OD sources.The model is a gaussian puff model, and is called theOpen Burn and Open Detonation Model(OBODM). EPA Region 4 recommends the use ofthe OBODM model for open burn and detonation

5-30


TABLE 5.2 PREFERRED SCREENING AIR DISPERSION MODELS

AND THEIR USES

MODEL CHARACTERISTIC

SCREEN31

TSCREEN2

CTSCREEN3

Source Types

Point, Area,

Volume, Flare

Numerous

Point

Terrain Types

Simple, Complex

Simple, Complex

Complex

Release Mode

Continuous

Continuous,

Instantaneous

Continuous

Averaging Time

1 Hour

15 Minutes to

Annual

1 Hour to Annual

Land Use

Rural or Urban

Rural or Urban

Rural or Urban

Contaminant Type

Gas or Particulate

Gas, Particulate

Gas or Particulate

Applicable Range

≤ 100 km

≤ 100 km

≤ 50 km

Generic or Real Meteorological Data?

Generic

Generic

Generic

Model Chemical Reactions?

No

No

No

Model Building Wake Effects?

Yes

Yes

No

Dry Deposition Calculations?

No

No

No

Wet Deposition Calculations?

No

No

No

Model Negatively Buoyant Gases?

No

Yes

No

Single or Multiple Sources per Simulation?

Single

Single

Multiple

1 SCREEN3 dispersion model for a single source. 2 TSCREEN screening model for a single source. 3 CTSCREEN model for complex terrain.

5-31


units. For this reason, OBODM is included in thisdocument as a preferred dispersion model forSubpart X Permitting.

SCREEN3

The SCREEN3 model is a Gaussian, steady-statedispersion model used for making simple screeningevaluations for neutrally buoyant, continuousemissions from a single source. The model usesbuilt-in worst-case meteorological conditions topredict concentrations from either point, area,volume, or flare sources. The SCREEN3 modelcan simulate dispersion from only one source at atime. The model is capable of simulating dispersionof gases or particulates in simple or complex terrain.Only one-hour averaging periods are calculated, soif different averaging periods are desired, genericadjustment factors must be used. (Note thatreference doses and other health criteria do notrequire exposures of less than 1 year.) SCREEN3is recommended for simple screening evaluations ofa single, continuously emitting source.

SCREEN3 is available on EPA’s Support Center forRegulatory Air Models (SCRAM) bulletin board,which is part of the OAQPS Technology TransferNetwork:

Telephone Number:(919) 541-5742Baud Rate: 200, 9600, or 14.4K baudLine Settings: 8 data bits, no parity, 1 stop bitTerminal Emulation: VT100 or ANSIInternet TTN site: http://ttnwww.rtpnc.epa.govSCRAM site: http://134.67.104.12/html/scram/scram.htm

TSCREEN

TSCREEN is a screening modeling system for toxicreleases that consists of four different dispersionmodels: 1) SCREEN3 for neutrally buoyant,continuous releases; 2) PUFF for neutrally buoyant,non-continuous releases; 3) RVD for dense gas jetreleases; and 4) the Britter-McQuaid Model forcontinuous or puff dense gas area sources. Whenexecuting TSCREEN, the user enters parameters for

http://www.epa.gov/ttn/scram

http://www.epa.gov/ttn/scram

http://ttnwww.rtpnc.epa.gov

http://134.67.104.12/html/scram/scram.htm

http://134.67.104.12/html/scram/scram.htm

5-32


the source and receptors, and the appropriate modelis selected within the modeling program.TSCREEN uses generic, worst-case meteorologicaldata to calculate downwind concentrations. Themodeling system is versatile in its ability to simulatedispersion from many different types of toxicemission sources. As in the case of SCREEN3,only one source can be entered in the model persimulation. TSCREEN is recommended forscreening evaluations of single sources of toxic aircontaminants. TSCREEN is available on EPA’sSCRAM bulletin board. See the SCREEN3summary for details on access to SCRAM.

CTSCREEN

CTSCREEN is the screening mode of theCTDMPLUS model for calculating downwindconcentrations from point sources in complexterrain. CTSCREEN and CTDMPLUS areidentical, except that CTSCREEN uses generic,worst-case meteorological data rather than theextensive site-specific meteorological data used inCTDMPLUS. CTSCREEN can be used in ascreening analysis for point sources when complexterrain affects dispersion of contaminants. See theindividual listing below for information aboutCTDMPLUS. CTSCREEN is available on EPA’sSCRAM bulletin board.

ISC3

The Industrial Source Complex 3 (ISC3) model is aGaussian plume model that can predict short- orlong-term concentrations of pollutants fromcontinuous emissions of point, area, and volumesources. The model can simulate the downwasheffects of buildings on point sources, can simulatemultiple sources per run, and is appropriate for useto a distance of 50 kilometers. The model recentlyhas been modified to include dry and wetdeposition, an algorithm for complex terrain, and animproved algorithm for modeling area sources.

ISC3 is preferred for most refined modelingapplications when there are continuous emissions ofneutrally buoyant, nonreactive pollutants. The ISC3

5-33


model is not the best model in cases in which arelease of a pollutant is instantaneous or intermittentor those in which the pollutant is significantly heavierthan air. ISC3 treats chemical reactions only bysimulating exponential decay of a pollutant. Ifcomplex chemical reactions of a pollutant in theatmosphere are important, a different model may bemore appropriate.

ISC3 requires entry of detailed data on the sourceand receptors and preprocessed hourlymeteorological data. Depending on the featuresused, additional data are required, such asinformation about building dimensions and particlesize. Because ISC3 requires entry of complex datato use various model features, analyses should beperformed by an experienced modeler. The ISC3model is available on the SCRAM bulletin board.

RAM

The RAM model (Gaussian-plume multiple sourceair quality algorithm) is a steady-state Gaussianplume model capable of predicting concentrations ofcontaminants from point or area sources. Themodel assumes level terrain and can assessconcentrations for short-term averaging periods(from one hour to one day). RAM can estimateconcentrations in rural or urban areas, but isrecommended specifically for use in urban areas.Although use of the RAM model is acceptable,within those limiting conditions, the ISC3 modelgenerally is preferred because of its updated featuresand algorithms. RAM is available from the NationalTechnical Information Service (NTIS).

CTDMPLUS

CTDMPLUS is a refined point source Gaussian airquality model for use in all stability conditions forcomplex terrain applications. The model requiresentry of considerable surface and upper airmeteorological data, and uses extensive data onterrain to define the shapes of individual hills. Themodel associates each receptor with a particular hill.CTDMPLUS is recommended specifically forcontinuous, elevated point sources near terrain that

5-34


is higher than the top of the source’s stack.CTDMPLUS is available on EPA’s SCRAM bulletinboard.

INPUFF

INPUFF is a Gaussian integrated puff model forevaluating downwind concentrations or depositionfluxes from continuous or noncontinuous sources.INPUFF is capable of modeling multiple sources atas many as 100 receptors and for as many as 144meteorological periods. Moving or stationarysources may be simulated with puffs that disperseover a gridded wind field. The puffs from a sourceare released in a series of user-specified time steps.INPUFF usually is applied to noncontinuous sourcesand is the most common model for use for OB/ODoperations. INPUFF is a suitable model for OB/OD releases under most circumstances, but it doeshave significant limitations including: use ofdispersion parameters for long-term releases, ratherthan short-term releases; use of plume rise equationsfor continuous sources; and unrealistic simulation ofatmospheric turbulence. Unfortunately, there arefew alternative models available to address OB/ODreleases. The limitations of INPUFF should berecognized when evaluating a modeling plan thatuses the model.

When INPUFF is used to model OB/ODoperations, source parameters should be input intothe model to best fit the actual release characteristicsof the source. Because INPUFF is not able tospecifically address OB/OD type releases, the inputparameters must be modified to fit the inputrequirements of another source type, and still exhibitthe release and dispersion characteristics of the OB/OD operation.

Open burn operations are usually characterized aspoint sources so that buoyancy plume rise can betaken into account. When running the model foropen burn sources, the buoyancy-induceddispersion option should be selected. Input valueswill vary depending on the type of material beingburned, and the location and construction of theburn pan. As a guide, the checklist gives typical

INPUFF can be used for OB/OD releases butit was not developed to model OB/OD typeprocesses. The limitations of INPUFF shouldbe recognized when evaluating a modelingplan.

5-35


open burn values of 3700 °K for exit temperatureand 0.1 to 10 meters per second for exit velocity.An open burn operation may be considered acontinuous release if the burn lasts for a long time (1hour or longer), but will usually be considered ashort-term release.

Open detonation sources should be characterized asa volume source with initial lateral and verticaldimensions equivalent to the expected maximumextent of the blast cloud. There are several methodsfor identifying the extent of the cloud. Stoner andKirkpatrick (1995c) suggest one method fordetermining the cloud size by first calculating theinitial source volume using the POLU model, whichestimates total detonation gases and initialtemperature. These results are entered intoINPUFF as a ground-level source with plume rise.As part of this method, the cloud is limited to amaximum height using estimates made from high-explosive algorithms developed by the DefenseNuclear Agency. Other methods for determining thecloud extent may also be used. In general, anymethod used to determine the cloud dimensionsshould be well documented and justifiable.

When an OB/OD source is located in complexterrain, a model such as CALPUFF should be usedto properly address the terrain issues. However, forscreening analyses, INPUFF may be used ifconservative assumptions are incorporated into theanalysis to account for the complex terrain. Oneexample of this is to assume that the cloud height isground level and all the receptors are at groundlevel. INPUFF is available from NTIS.

CALPUFF

The CALPUFF model is a complex modelingsystem that can estimate concentrations of pollutantsfrom non-steady-state emission sources. This modelcan simulate the effects of meteorological conditionsthat vary according to time and space, chemicaltransformation, and physical removal. CALPUFF isalso capable of simulating building downwash andtransport over complex terrain and over water, orcoastal transport. It can be used for point, area,

5-36


volume, or line sources. The CALPUFF modelingsystem has several modules, each intended forperforming a separate operation. One recentlyadded module treats buoyant rise and dispersionfrom area sources. This module may be useful formodeling OB/OD sources. Because CALPUFF isa complicated modeling system, and because EPAhas not fully recommended its use, review of aCALPUFF analysis by experts is recommended.CALPUFF is available on EPA’s SCRAM bulletinboard.

OBODM

The Open Burn/Open Detonation Dispersion Model(OBODM) is intended for use in evaluating thepotential air quality impacts of the open air burningand detonation (OB/OD) of obsolete munitions andsolid propellants. OBODM uses cloud/plume rise,dispersion, and deposition algorithms taken fromexisting models for instantaneous and quasi-continuous sources to predict the downwindtransport and dispersion of pollutants released fromOB/OD operations. The model can be used tocalculate gravitational deposition for emissions frommultiple OB/OD sources for either a single event ofup to a year of sequential hourly source andmeteorological inputs. The program is designed foruse on IBM-compatible PCs using the MS-DOS(Version 2.1 of higher) operating system withkeyboard and optional mouse-control. It will alsorun under most WINDOWS environments.

DEGADIS

The Dense Gas Dispersion Model (DEGADIS) usesmass and momentum balances and laboratory andfield scale data to simulate the release and transportof pollutants (EPA 1995). It is used for negativelyor neutrally buoyant releases of toxic, nonreactivegases or aerosols. It is applicable for ground-level,low-momentum area releases; or upwardly directedstack releases. The release may be instantaneous,continuous, or of finite duration, or may vary overtime. The model simulates only one set ofmeteorological conditions, so the modeled timeframe should not exceed one to two hours. Another

OBODM is available on EPA’s SCRAM BulletinBoard http://www.epa.gov/scram001

http://www.epa.gov/scram001

5-37


limitation affecting the model is that dispersion isassumed to take place over flat, unobstructedterrain. DEGADIS is not equipped to addressterrain that is complex or that has extensive surfaceroughness.

DEGADIS requires entry of the characteristics ofthe release and its chemical and physical properties,data on receptors, and standard meteorologicaldata. If an aerosol release is being modeled, thedensity of the release also must be entered into themodel. Although DEGADIS is appropriate for awide range of sources, it is particularly valuable incharacterizing releases of pollutants that are verydense compared with air. An external input file oran interactive computer program can be used to runDEGADIS. DEGADIS is available on EPA’sSCRAM bulletin board.

HGSYSTEM

HGSYSTEM is a computer program thatincorporates several different dispersion models forvarious types of toxic releases. The modelingpackage can estimate one or more consecutivephases between a spill of a toxic substance andnear-field and far-field dispersion of a pollutant. Thepollutant being modeled can be a two-phase,multicompound mixture of nonreactive compoundsor hydrogen fluoride. The modeling system cansimulate chemical reactions only for hydrogenfluoride. HGSYSTEM assumes flat, unobstructedterrain and can be used for continuous, finite-duration, instantaneous, and time-dependentreleases. HGSYSTEM can be used to determineshort-term (one hour or less) concentrations of toxicreleases under one set of ambient conditions.HGSYSTEM is available from the AmericanPetroleum Institute.

SLAB

The SLAB model is used for modeling thedispersion of dense gas releases from a ground-levelevaporating pool, an elevated horizontal jet, a stackor elevated vertical jet, or an instantaneous volumesource. If two or more different types of releases

5-38


require evaluation, they must be processed inseparate model simulations. The SLAB model usesonly one set of meteorological conditions, so onlyshort-term concentrations can be calculated. Themodel assumes that the release consists ofnonreactive dense gases or aerosols and that nodeposition occurs. SLAB calculates concentrationsby using numerical integration in space and time tosolve the basic conservation equations. SLAB isavailable on EPA’s SCRAM bulletin board.

Source Type Specification

In part, selection of the proper dispersion modeldepends on the type of emission source or sourcesthat must be modeled. Each source must beclassified as a point, area, volume, or line source.Some models allow for identification of other typesof sources that are subsets of the four types listedabove. An example of such a sub item is a flare,which is a type of point source. In addition, eachsource must be classified as a continuous,instantaneous, or intermittent source; as a vapor-phase or particulate emission source; and, whenmodeling gaseous contaminants, as neutrally buoyantor negatively buoyant. These determinations willaffect the selection of a model.

Releases from point sources are those from stacksor vents; they exhibit well-defined exit parameterssuch as temperature, flow rate, and stack height.Releases from area sources are emitted at or nearground level and over a given surface area. Areasource emissions are entered into a model in units ofmass per time per area. Releases from volumesources are those that occur over a given area (likearea sources), and also within a certain depth.Volume sources can be ground level or elevatedsources. When entering data for a volume source, amodel requires the initial lateral and verticaldimensions of the source. Releases from linesources are releases from roadways or othersources that emit over a long and narrow space.Some models simulate line sources with a series ofvolume or area sources adjacent to one another.

5-39


In general, a permit writer should evaluate thedescription of a source or proposed source anddecide whether an applicant’s representation of thesource in a modeling analysis is reasonable. As canbe anticipated, the choice of the type of source to beused sometimes can be left to professional judgmentand based on how a source best fits into thedefinition of a given type of source.

Sources must also be classified as continuous,instantaneous, or intermittent. The most commondispersion models are Gaussian, steady-statemodels, such as ISC3. These types of models cansimulate dispersion from continuous sources. Forinstantaneous or intermittent releases, a “puff” modelmay be used. This differentiation is of particularimportance for OB/OD operations, from whichemissions occur over a very short period during ODoperations and from a few minutes to one hourduring OB operations. TSCREEN incorporates apuff model into its screening system, and INPUFFand CALPUFF and OBODM are puff models thatcan be used for screening or refined analyses. Apuff model should be used when the travel time ofthe plume from the source to a receptor is longerthan the duration of the emissions.

If a gaseous contaminant cloud emitted by a sourcehas a significantly higher density than air, it will benegatively buoyant and should be modeled with adense gas model (DEGADIS, HGSYSTEM, orSLAB). When uncertain whether a vapor cloudshould be modeled with a dense gas model, thevapor cloud’s Richardson Number (Ri) should becalculated. A cloud that exhibits Ri > 10 should bemodeled with a dense gas model (Trinity 1996).

Contaminants emitted from Subpart X units mayinclude NOx, SOx, particulates (including metals),VOCs, and SVOCs. Most of the preferred modelslisted in this document are capable of simulatingtransport of both particulate matter and vapor-phaseemissions.

5-40


Table 5.3 and the model descriptions of modelsprovided in this document should be helpful to apermit writer in determining whether a permitapplicant has selected the appropriate model.

Meteorological Parameters

It is important that the permit writer ensure thatappropriate meteorological data have been includedin a modeling analysis. For screening analyses, theinformation usually is straightforward because mostscreening models use generic, worst-casemeteorological conditions. Usually, themeteorological conditions that produce the highestmodeled concentrations are low wind speeds andstable atmospheric conditions. For models thatrequire the entry of a single set of conditions, such asdense-gas models, the permit writer should verifythat reasonable worst-case conditions have beenentered. Reasonable worst-case conditions may bemodified to reflect proposed operating restrictions.For example, OB/OD operations may be confinedto daylight hours; therefore, worst-case stabilitymight be the worst-case daylight stability conditions,since the atmosphere tends to become more stableat night.

If a refined modeling analysis requires entry of realmeteorological data, either on-site meteorologicaldata for one year, or off-site data for five years areneeded for a refined analysis. If on-site data for fiveyears are available, all those data should be used.Off-site data can be obtained from nearby NationalWeather Service stations, military facilities, orindustrial facilities. The permit writer should examinethe location from which any off-site data werecollected to ensure that location resembles the sitebeing modeled. Parameters to review includedistance of the station from the site, unique featuresof the terrain that may change the wind flowpatterns, and the exact location of the monitoringequipment. National Weather Service data frommany stations nationwide are available on theSCRAM Bulletin Board System or from NTIS. Ofthe models listed in Table 5.3, those that use detailedmeteorological data include ISC3, RAM,CTDMPLUS, INPUFF, CALPUFF, and

Tip:Procedures for creating multi-year files can befound in the User’s Guide for the air dispersionmodel. Procedures for the ISCST3 model arealso presented in Section 3.7.4 of the HumanHealth Risk Assessment Protocol for HazardousWaste Combustion Facilitis (HHRAP).

5-41


OBODM. The dense-gas models (DEGADIS,HGSYSTEM, and SLAB) accept only one set ofambient conditions.

If existing representative data are not available, apermit applicant must collect data from the site.Those data should be collected in accordance withthe guidelines set forth in MeteorologicalMonitoring Guidance for Regulatory ModelingApplications (EPA 2000). More information aboutthe collection of on-site meteorological data ispresented below in the discussion of monitoring.

Locations of Receptors

Any modeling analysis must define the locations ofreceptors for which concentrations of contaminantswill be calculated. For Subpart X permitting, thepoint of compliance (POC) receptors must beevaluated in a modeling analysis. POC receptorsmust be chosen to evaluate both direct exposure andindirect exposure from an air release (indirectexposure may result when hazardous constituentsare present in soil or water through deposition ofparticulates or gases). Permit applicants shouldidentify locations of potentially exposed individuals;the potentially maximum exposed individual (MEI);potential ecological receptors, such as local plantsand animals; and other sensitive environments andendangered species.

Dispersion models vary in the amount of detail theyrequire in information about receptors. Somescreening models (for example, SCREEN3 andTSCREEN) do not require entry of an exactlocation, but only the distance to a receptor. Forsuch models, the direction is not important becausethe models conservatively assume thatmeteorological conditions will be such thatdispersion is in the exact direction of the chosenreceptor.

Other models allow a user to enter discrete locationsof receptors or a gridded receptor field. Modelsthat evaluate considerations related to terrain alsorequire entry of elevations for receptors. Modelinganalyses for Subpart X permitting should include

http://www.webmet.com/metmonitoring/table_of_contents.html

http://www.webmet.com/met_monitoring/1_11.html

http://www.webmet.com/metmonitoring/table_of_contents.html

5-42


receptor points at all POC locations. In somecases, when POC locations are uncertain or whenthe maximum concentration must be determined fora given region, a full receptor grid may be necessary.The permit writer should evaluate, on a case-by-case basis, whether the modeled receptor locationsare adequate to characterize potential effects tohuman health and the environment.

Features of the Terrain

Incorporation of features of the terrain is animportant factor in modeling analyses, especiallywhen buoyant plumes are being modeled. Whenthere are significant terrain features in the vicinity of asite, a model should be used that can simulate aplume’s transport over or around such features. Formodeling a point source in complex terrain,CTSCREEN (for screening analyses) andCTDMPLUS (for refined analyses) are preferred.The two models require extensive information andsignificant sophistication on the part of the personoperating the model. If the permit writer wishes torerun the model to check results, he or she mayrequire assistance from staff experienced inoperating the models. The ISC3 model includes thecomplex terrain algorithm from the Valley model andcan be applied in areas of complex terrain. Whenusing a model that cannot address complex terrainan applicant may also choose to apply conservativeassumptions to account for such terrain. Modelinganalyses that make assumptions to account forfeatures of the terrain should be consideredscreening analyses. In any case, the permit writershould verify that a modeling analysis has addressedproblems related to complex terrain and that permitapplicant has used the best model practicable underthe circumstances.

If a facility is near a coastline or next to a large bodyof water, dispersion differs from that over land, anda model particularly suited for dispersion andtransport over water may be necessary. TheCALPUFF model incorporates algorithms foroffshore and coastal dispersion. Other models thataddress offshore or coastal dispersion that are not

5-43


listed in Table 5.3 include the Offshore and CoastalDispersion Model (OCD) and the ShorelineDispersion Model (SDM).

Deposition

Deposition of contaminants onto land or watersurfaces may result in indirect exposure and risk tohuman and environmental receptors. Depositionmay increase risk by exposure pathways other thanair. A refined model with deposition capabilities canbe used to model deposition, or modeledconcentrations can be multiplied by calculateddeposition velocities to estimate deposition. A thirdoption, which the permit writer must consider atoperating facilities on a case-by-case basis, is toestimate deposition by taking soil samples.

Chemical Transformation

Chemical transformation of contaminants after theyhave been released into the air is difficult to quantify;and most dispersion models do not address it,except in a limited fashion. Chemical reactions in theatmosphere from releases of contaminants dependon many different factors and cannot beincorporated easily into a modeling analysis.However, chemical transformations take time tooccur in the atmosphere, so the processes generallyare not considered significant when travel times arelimited to a few hours (EPA 1995a). One exceptionis in urban areas, where photochemical models areapplied to address complex chemical mechanisms.The models typically do not evaluate individualsources, but are used for regional modeling analyses.

Some of the models listed in Table 5.3 are capableof limited calculations of chemical transformations.ISC3 and RAM allow the user to enter anexponential decay factor to address breakdown ofchemicals. CALPUFF is able to model pseudo-first-order chemical reactions and is based onalgorithms from the MESOPUFF II model, which isa long-range puff model (EPA 1995b). Last,HGSYSTEM can calculate chemical transformationfor releases of hydrogen fluoride.

5-44


Transformation of NOx to NO2 can be estimated bypostmodeling calculations. Usually, a conservativeassumption is made that all the NOx converts toNO2 in the atmosphere. If a permit applicant hasincluded a transformation calculation from NOx toNO2 in the modeling analysis, the permit writershould refer to Guidelines On Air Quality Modelsfor details on review of this process.

Other available dispersion models estimate chemicalreactions in the plume (EPA 1995a) and may beused as determined appropriate on a case-by-casebasis. Modeling calculations that include chemicaltransformations should be reserved for refinedanalyses. Screening analyses should use worst-caseassumptions for chemical transformation.

Background Concentrations

Under the Subpart X permitting requirements listedin 40 CFR §264.601(c)(5), a permit applicant mustprovide information about existing air quality in thearea. The information must include the effects ofother sources of contamination. Other sources ofcontamination may be natural sources, nearbysources, or unidentified sources. The information isimportant in understanding the overall air quality atthe site and in its vicinity. When an air dispersionmodeling analysis is conducted, the existingconcentrations of air contaminants (or background)must be determined so that total effects on air qualitycan be evaluated. Modeled effects from individualsources are added to the background concentrationto obtain the total concentration of a contaminant ata given receptor location. In many cases, existingbackground concentrations measured in the vicinityof a Subpart X source may be obtained from localregulatory agencies, universities, or nearby industrialfacilities. If the Subpart X unit is an isolated, singlesource and no data exist for the area, a regionalbackground site may be used that is not nearby, butthat is affected by similar natural and distant sources.However, if the site at which regional backgrounddata were collected is not affected by similarsources, those data should not be used. In general,the permit writer should evaluate the backgrounddata submitted by an applicant carefully to determine


5-45


whether they adequately characterize the air qualityin the vicinity of the unit. In cases in which SubpartX units are located near other sources that areexpected to have a significant concentration gradientin the area, the nearby sources should be modeledexplicitly. The effect expected from all other sources(natural and distant sources) then should be addedto the results of modeling.

It is important that the background concentrationsthat are added to the results of modeling have thesame averaging period as those results. Forexample, if the eight-hour average concentration of acontaminant is modeled, an eight-hour averagebackground concentration should be added todetermine the total eight-hour concentration.

If no representative background data are available,monitoring may need to be conducted to determinethe existing air quality. Section 5.2.1.2 discussescollection of on-site air quality data.

Evaluation of Selection and Application of theModel

Selection and application of a suitable air dispersionmodel is to a great extent dependent on theapplication of site-specific criteria. Several of theprincipal criteria for selecting a model werediscussed in preceding sections. They include typeof source, meteorological data, locations ofreceptors, features of the terrain, deposition,chemical transformation, and backgroundconcentrations. Permit writers should evaluate thedetails about the site, the available data, and theprocess by which the applicant selected the site todetermine whether the modeling analysis isappropriate.

In some cases, site-specific or source-specificcharacteristics of a Subpart X unit may be such thatno screening models are capable of simulating theireffects on the transport and dispersion of acontaminant. In such cases, a refined modelinganalysis must be required. The permit writer shouldevaluate the capabilities of the screening model usedin a permit applicant’s screening analysis and

5-46


compare those capabilities with the characteristics ofthe source and site to determine whether the modelselected is appropriate. In cases in which the permitwriter determines that the screening model selectedis inadequate, he or she should issue a NOD toindicate the reasons for such inadequacy.

The permit writer should evaluate carefully anyscreening models that are not in Table 5.1 or inAppendix A or B of Guidelines On Air QualityModels to determine whether such models aresuitable for the task at hand. In such cases, it isrecommended that the permit writer seek the adviceof modeling experts to determine whether thealternative model is suitable for the specified task.

If a permit applicant cannot demonstrate compliancewith appropriate standards through the use of ascreening model, or if the site-specific details requireuse of a more sophisticated model, the permit writershould issue a NOD to indicate that a refinedmodeling analysis must be conducted. Use of site-specific data will result in more accurate modelingresults. Since refined models use more detaileddata, the permit writer should verify that the modelused in a refined analysis is appropriate for thespecial features of the site and the data available.

Of the refined models listed in Table 5.3, ISC3 is themost commonly used and accepted for regulatoryapplications. Other models in Table 5.3 can beapplied for specific purposes. For example,releases from OB/OD units are usually intermittentor near instantaneous, and are not stack-typesources. In such cases, use of ISC3 would not beappropriate because it can simulate only continuousreleases. The INPUFF model has been used forOB/OD operations and its results have been foundacceptable. However, INPUFF has somelimitations, and other models may be better suitedfor OB/OD applications. The limitations ofINPUFF are discussed briefly in the model summarysection of this guidance. The CALPUFF model canbe used for OB/OD applications and has moreextensive capabilities than INPUFF, but the modelrequires additional data and is more difficult to use.As discussed in the previous sections, OBODM



5-47


was developed specifically to model OB/ODemissions. The OBODM algorithms address theunique dispersion characteristics associated withopen burn and detonation operations.

Evaluation of Results of Modeling

A permit writer must consider several factors whenevaluating results of modeling. Averaging time,background concentrations, and an overallperspective of the data entered and results producedmust be taken into account in interpreting results todetermine whether they make sense. The permitwriter should compare the model results with thedata entered to determine whether the results arerealistic.

Often, model analyses must estimate maximumshort-term as well as long-term effects. Somemodels calculate concentrations for only oneaveraging period (usually one hour), while otherscalculate concentrations for several averagingperiods. If a model is limited to one averagingperiod, permit applicants may use modeledconcentrations to estimate concentrations for otheraveraging periods. Adjustments may be made toreflect how long the unit emits hazardousconstituents, and for variations in meteorologicalconditions. Any averaging time factors used bypermit applicants should be well documented andjustified.

5.3.2 Groundwater Modeling

This section provides information regardinghydrogeological characterization and model selectionto assist permit writers in evaluating modeling resultssubmitted by Subpart X permit applicants.

Groundwater modeling can be used whenmonitoring is impractical or to supplement and verifymonitoring data. Groundwater modeling has severalapplications in the permitting process for Subpart Xunits. The groundwater model can be used (1) topredict conservative, “worst-case” scenarios duringa detailed groundwater assessment, (2) to assist in

Reviewing the Results of Air Modeling:Items to Check

• Spot check source characterization datainput files.

• Compare building down wash parametersto the output from BPIP.

• Spot check several modeled receptorelevation against USGS map.

• Review receptor lists or files to ensurethat the elevation array contains non-zerovalues.

• Check the anemometer height to ensurethat it is correct for the station and yearsused in the analysis.

• Ensure that the GEP stack heightdetermined by BPIP was not used in theair modeling analysis.

Additional guidance on reviewing airmodeling results can be found in EPA Region6’s “Suggestions for Auditing AssessmentAir Modeling Studies”.

5.3.1 air dispersion and emission modeling x s... · emission modeling emission modeling is a...

Documents