Environmental risk assessment reports chapter 18

Air modeling analyses are used in risk assessment to evaluate three aspects of atmospheric releases: • The type of activity, including permitted normal or routine facility operations, or

CHAPTER 18

Air Toxics Dispersion and Deposition Modeling Richard A Rothstein

CONTENTS

I Introduction 370

A Regulatory Drivers Affecting Risk Assessment Modeling Studies 371

B Consultant Selection 372

II Overview of Air Modeling Process for Risk Assessment 373

A Reliability of Air Model Predictions 373

III Practical Air Modeling Considerations, Approaches, and Issues 374

A Basic Air Modeling Concepts 374

B Dispersion Modeling 376

C Deposition Modeling 378

IV Sources of Air Quality Models 380

V Sources of Data 380

A Air Quality and Meteorological Data 380

B Sources of Air Emissions Data 381

C Evaluating and Interpreting Air Emissions Data for Risk Assessment Modeling 382

VI “Cutting Edge” Air Modeling Issues for Risk Assessment A Air Pathway Fate and Transport Issues 383

for Contentious Multiphase Contaminants 383

B Atmospheric Fate And Deposition Modeling — Always Needed? 385

C Limitations of Deposition Modeling 385

D Micrometeorological Effects 386

VII Collection of Emissions Data Appropriate for Site-Specific, Multi-Pathway Risk Assessments 386 VIII Conclusion 387 References 388

I INTRODUCTION

Regulatory agencies increasingly require use of air dispersion and deposition mod-eling to evaluate the environmental risk of facility remediation, construction, or operation Mathematical models calculate air contaminant (plume) dispersion and deposition — the changes in concentration of substances from the source to some location at a given distance from the release point Typical air emission sources evaluated by regulatory agencies include superfund and hazardous waste sites under-going groundwater or soil remediation; municipal solid-waste incinerators and land-fills; industrial source operations that use various chemicals in the manufacturing process; industrial and municipal wastewater treatment facilities; and microelectron-ics industries which use specialty gases and chemicals

Air modeling analyses are used in risk assessment to evaluate three aspects of atmospheric releases:

• The type of activity, including permitted normal or routine facility operations, or unlikely or unavoidable malfunction of operation conditions

• The type of exposure, for example effects from predicted short-term (acute) and long-term (chronic) impacts from different exposure routes

• The exposed population, such as on-site workers and facility operators or on people off-site

Off-site exposures are often characterized as the potential impacts to the “rea-sonably” maximum exposed individual or as the “average” exposed individual within the modeled site region

Air emissions are also modeled from sources under consideration for air permits, environmental impact reports; facility engineering design; air monitoring network design; and input to exposure assessment and risk characterization studies, the focus

of this textbook In addition to their use in risk assessments, such air modeling results are also used to help properly site air-monitoring equipment for remediation projects Air dispersion and deposition modeling results are used to select technically feasible and commercially available state-of-the-art control technology so as to minimize source air emissions and the resulting exposure impacts

Air modeling for risk assessment can be broadly subdivided into two major categories: (1) those analyses conducted for stationary point sources, e.g., sources whose air emissions to the atmosphere come from a facility vent or stack; and (2) those conducted for near ground-level area type sources, e.g., an open area of emissions, such as a solid or hazardous waste landfill site, or a lagoon Depending

on the source category, the air quality analyst needs to ensure that models are properly selected and applied to provide for reliable exposure assessments and risk characterization predictions

A Regulatory Drivers Affecting Risk Assessment Modeling Studies

Over the past two decades, facilities involved with the generation, treatment, storage, and disposal of hazardous waste have been affected by U.S EPA regulations devel-oped to minimize and maintain air emissions at safe levels These rules include those developed under the Resource Conservation and Recovery Act (RCRA) and the Comprehensive Environmental Response, Compensation, and Liability Act (CER-CLA), also referred to as the Superfund Act The siting and design of new treatment facilities, or cleanup of existing contaminated waste disposal sites, often triggers a myriad of state and federal environmental permitting and impact assessment require-ments to receive necessary approvals Depending on applicable agency rules, or when planned project actions have the potential to adversely affect human health and the environment, a risk assessment is conventionally performed Such assess-ment will evaluate potential multimedia impacts, and where applicable, ensure that appropriate risk management plans and mitigation measures are implemented in the facility design, construction, and operation

Agencies also frequently require risk assessments for a variety of stationary combustion sources to confirm the necessary air-emission control levels These include municipal solid waste and medical-waste incinerators, hazardous-waste incin-erators, and boilers and industrial furnaces (BIFs) that burn hazardous wastes On May 18, 1993, the EPA Administrator issued a policy directive that included a draft combustion strategy intended to minimize toxic air emissions from new and existing hazardous-waste incinerators, as well as from BIFs The policy directive requires: (1) site-specific, comprehensive multipathway risk assessments to quantify potential risks to public health and the environment, and (2) facility-specific permit emission limits for dioxins/furans and particulate matter, to control unacceptable risks from trace organic compounds and hazardous metal emissions, respectively EPA’s Indus-trial Source Complex (ISC) dispersion and dry/wet deposition model can evaluate explicitly potential risks due to indirect exposures to combustor emissions

More recently, Title III, of the 1990 Clean Air Act Amendments, addresses control of 188 hazardous air pollutants (HAPs) that were identified initially by Congress EPA and states will be promulgating new air rules throughout this decade

to control HAP emissions from hundreds of major new and existing stationary source categories These include municipal, industrial, manufacturing, petrochemical, waste processing, and power generating facilities Hence, major sources of HAPs will need

to implement new control-technology measures, mainly during the next ten years,

to reduce HAP emissions EPA may later promulgate more restrictive emission control regulations for the affected HAP source categories based on the outcome of residual risk assessment studies

Unlike for RCRA, the HAP emission reduction rules that EPA is developing are mainly control-technology based rather than risk-assessment based State agencies may, nevertheless, require certain source owners and operators to continue to perform site-specific multipathway risk assessments This requirement may be part of the permit approval process for major or controversial projects to ensure that adequate levels of control will be used

Notwithstanding the regulatory drivers, air modeling to support the risk assess-ment process is an important tool for all affected parties to confirm the appropriate facility designs, remedial action cleanup levels, or source emission control technol-ogies to be employed

B Consultant Selection

This section summarizes the preferred education, experience, and special qualifica-tions that the air modeling practitioner should possess The criteria given below are germane to the project or task manager responsible for the air modeling This individual is responsible for managing and/or providing the model output which drives the exposure assessment and risk characterization studies, whether they be human health related or ecologically related

The art and science of air modeling is in selecting the proper model for a given situation, and then choosing scientifically credible model inputs It takes considerable scientific training and experience to ensure that the proper model data input are developed, and that model output and its implications for driving the risk assessment are properly interpreted Notwithstanding the continued advent of user-friendly computerized air dispersion models being readily available to the technical commu-nity via electronic bulletin boards and software vendors, air modeling for risk assessment should be performed by qualified and experienced individuals The individual (or firm) selected for the air modeling should have application experience in evaluating air emission impacts from (1) proposed and existing sta-tionary combustion or process emission sources; and (2) releases to the air, soil, ground water, and surface water from existing waste disposal sites, or from proposed waste remediation alternatives The diverse nature of risk assessment necessitates

an individual who is well-versed in technical, regulatory, and public health and environmental issues, with a particular sensitivity to public perception A basic understanding of both carcinogenic and noncarcinogenic risk assessment methodol-ogies pertaining to hazard identification, dose-response assessment, exposure assess-ment, and risk characterization is essential, so that the air models can be selected and applied properly

Technical knowledge and capabilities need to include an understanding of the physical, chemical, and toxicological properties of the contaminants in question, including proper identification and evaluation of the exposure pathways, transport, and fate of contaminants A basic understanding of both carcinogenic and noncar-cinogenic risk assessment methodologies (e.g., multistage linear models for assess-ing carcinogenic impacts; and hazard indices, quotients, and reference doses for assessing noncarcinogenic impacts) is also important, to ensure that modeling goals and objectives will be satisfied

The individual should be experienced in technical and regulatory criteria for properly selecting and applying approved EPA computerized air dispersion and deposition models To properly interpret the air model output, the individual should have an understanding and appreciation of the limitations and uncertainties of apply-ing models These uncertainties pertain to adequacy of source emission and mete-orological databases, and applicability and appropriateness of model algorithms to properly simulate the site and regional setting

The individual should possess strong project management and people skills as

he or she will be dealing with a wide variety of multidisciplinary specialties and interested parties The individual should possess a B.S degree in a scientific or engineering discipline (M.S or Ph.D preferred) with at least 10 years of direct air modeling experience for risk assessment applications Certification in an air quality, meteorological, or multidisciplinary environmental science or engineering discipline

is also preferred

II OVERVIEW OF AIR MODELING PROCESS FOR RISK ASSESSMENT

Air quality analysts are vital members of a risk assessment team whose task is to evaluate the transport and impact of substances released to the environment via the air release pathway From a list of contaminants of concern, air emission rates are calculated, based on media concentrations (e.g., soil, air) of air contaminants at their source (e.g., fugitive emissions, trans-media movement of chemicals), and the emis-sion flux to the atmosphere Air quality analysts also use physical source character-istics (e.g., stack height, volumetric flow rate) and emission data to predict what the contaminant concentrations will be at a receptor point some distance from the contaminant source location Air quality analysts use computer mathematical models designed to simulate environmental processes that are thought to occur in the atmo-sphere from the source to a receptor location They use their computer simulation capabilities to evaluate how different environmental conditions will affect a chem-ical’s concentration and environmental distribution over the study area Receptor-point air concentrations and deposition rates are provided to risk assessors for one

or more exposure case scenarios, where these predictions are used as inputs in exposure equations that are designed to calculate chemical intakes and uptakes for risk characterization

A Reliability of Air Model Predictions

Two important roles for air modeling for risk assessment include: (1) making rea-sonably accurate and reliable predictions about the transport and fate of air emissions and (2) satisfying technical, regulatory, and public perception concerns about poten-tial source air impacts

Reliable air quality modeling provides for more reliable exposure assessments and risk characterization predictions Model predictions are only as good as the model itself, and the quality of data input As such, an air quality model can only be as good

as the databases and assumptions that are incorporated into its application Hence, proper quantification of site and regional characteristics, source operation parameters, emission rates, and meteorological data is essential in any risk assessment

Regardless of how carefully one selects and applies air quality models, a number

of unknowns, data gaps, and technical uncertainties still remain about the myriad

of chemical reactions and physical processes actually taking place in the atmosphere that affect the transport and fate of air contaminants Many computerized air models have been developed over the years for risk assessment applications While models continue to be developed and refined, they are predictive tools They should not be

perceived as yielding “absolute” accurate numerical estimates for all air contami-nants of concern, and for all conceivable environmental circumstances encountered Modeling uncertainties normally are addressed by making simplifying or conserva-tive assumptions to avoid underestimating the potential risk

III PRACTICAL AIR MODELING CONSIDERATIONS,

APPROACHES, AND ISSUES

Air dispersion and deposition models are used to estimate the atmospheric transport, the ambient air concentrations, and the surface deposition flux of specific air con-taminants An overview of dispersion and deposition models, including model appli-cation concepts, is given in terms of “what,” “where,” and “how” to model

A Basic Air Modeling Concepts

Physical source parameters and emission characteristics of contaminants of concern describe the nature of the discharges to the atmosphere Contaminant emission rates can be calculated for point and area (nonpoint) sources These rates are input to air models whose outputs are used to predict ambient air concentrations or deposition rates to various surfaces such as vegetation, soils, and water bodies Receptor-point concentrations are used in exposure models to calculate exposure levels

Calculation of point source emissions, from stack and vent emissions data, are generally straightforward in that source test data, emission factors, or mass balance calculations can be used Point-source emission rates based on testing are normally derived from the flue gas concentration of the contaminant and the volumetric flue-gas flow rate Emission rates for continuous point source operations are normally expressed as mass per unit time (typically in g/sec for air modeling)

Point-source physical parameters include stack height, internal stack top diam-eter, flue-gas stack exit velocity or volumetric flow rate, and flue-gas stack temper-ature It is also important to specify dimensions of building in the vicinity of the stack For relatively short stack to building height ratios, the stack plume dispersion

in the near field can be dramatically affected by turbulent building-wake effects caused by winds blowing over and around the structure(s) Such effects can cause the magnitude of the concentration impact to be higher, and the location of maximum impact to be closer to the stack, than would otherwise be the case in the absence of such building wake effects

Other point-source configurations to be modeled may include exhaust fans and louver vents that discharge air contaminants to the atmosphere In these cases, the physical height of the emission point above ground is normally modeled, along with the specified building dimensions, to account for turbulent building-wake effects Area sources result from underground or aboveground sources, typically referred

to as “fugitive emissions,” since they do not emanate from a stack or vent Contam-inants in the subsurface can exist as a free product (pure compound), absorbed to soil or other deposited substances, as vapor, or as solutes in groundwater Air emis-sions from the subsurface can be quantified from flux chamber type measurements;

gas emission models; or “back-calculation” air modeling analyses that use site perim-eter ambient-air monitoring and meteorological data to quantify the source term in the model

Aboveground area sources are typically associated with storage piles, landfills, ponds, and lagoons Fugitive dust or vapor emission rates are quantified from air emissions modeling or monitoring that relies on chemical and physical properties

of the contaminant, the type of medium hosting the contaminant, and associated meteorological influences (temperature, wind speed) Area-source emission rates are normally expressed in mass/area/unit time (typically in g/m2/sec for air mod-eling)

Area source parameters to specify in the modeling include the area-source dimensions and the effective emission height above local grade If the distance separating the area source and nearby receptors is too small, particularly for large area sources with nearby fence-line receptors, the model may require that the area source be divided into smaller “squares” to predict impacts at the close-in receptors Contaminants of concern selected for the risk assessment modeling usually satisfy the following general criteria — they are known to be routinely emitted, or have been detected in the air emissions from the source category in question, and they are irritants or potentially toxic to humans and/or have a propensity to bioac-cumulate or bioconcentrate in the environment Quantifiable air emission data from representative source tests, or from other data sources exist for inclusion in air modeling analyses The actual number of contaminants of concern that are quanti-tatively evaluated throughout the risk assessment is a function of factors including report rigor, economics, and availability of actual or surrogate data sets for a par-ticular emissions source In many cases, relatively few air contaminants are routinely monitored at certain facility source categories As a result, chemical identity/source emission data gaps can limit the robustness of air modeling for risk assessment Air model selection and application depends on addressing several source and site-specific questions For example, is the release to the atmosphere (1) quasi-instantaneous, such as from gas cylinder or chemical tank ruptures, or sudden soil venting during remedial excavation or construction work; (2) intermittent, such as from fugitive dust emissions from remedial equipment operations or windborne effects, or vapor emissions from contaminated soils; or (3) continuous, such as from combustion or process vents and stacks? Is it a (1) point source, such as fuel combustion stacks, solid and liquid waste incinerators, storage tanks, soil and landfill venting operations, and air stripper columns; (2) an area source, such as aggregate storage piles, landfills and hazardous waste storage sites, ponds, and lagoons; or (3)

a line-type source, such as trenches from remedial excavation and cleanup, perimeter venting at landfills, and vehicular traffic operating on, or egressing from, contami-nated property? Moreover, are released substances reactive, non-reactive, vapors, particles, buoyant, neutrally buoyant/passive, or denser than air? Other consider-ations include defining the location and nature of land use at receptor locconsider-ations (e.g., on-site, at the fenceline, on complex terrain, in a high rise building); the type of meteorological data available (e.g., collected on-site data, representative off-site data, worst-case screening meteorological data); the appropriate modeling time frame (e.g., short or long-term impacts); and the type of exposure pathways to be considered

(e.g., concentration predictions for inhalation exposure; deposition predictions for dermal and ingestion exposures)

Air modeling requirements and approaches for risk assessment applications may differ between political jurisdictions and governmental agencies An air modeling protocol prepared at the onset of a project for approval by the regulatory permitting entity serves to establish the “bench mark” for the conduct of the air modeling study

If certain modeling assumptions or considerations later need to be revised or updated during the course of the study, it is easier for the analyst to justify such changes, to the state or EPA, via comparison to the previously approved modeling protocol Considerable project time and expense can be saved if an approved modeling pro-tocol is used

Air models are used to calculate air concentrations or deposition rates for specific receptor locations, to evaluate risks to human health and the environment Modeled receptors can be: (1) onsite to predict exposure to workers; (2) fenceline and off-site to predict exposure to the general public and environment; (3) over land to predict (concentration) inhalation impacts, and (deposition) dermal and ingestion impacts; (4) over water to predict (deposition) ingestion impacts); and (5) over elevated terrain to predict stack plume impaction concentration impacts Model outputs can cover broad areas or can focus on particularly sensitive locations such

as hospitals The study area varies based on regulatory agency requirements and case-specific determinations

B Dispersion Modeling

Air dispersion models are mathematical representations that approximate the phys-ical and chemphys-ical processes in the atmosphere governing the transport and dilution

of gaseous and particulate air contaminants between the source and receptor They serve by using the source emission rate to the atmosphere to calculate the resultant ambient-air concentration at specified downwind receptor locations (usually at ground-level) The basic model algorithms which treat the source emission releases, plume rise, transport, and atmospheric dilution have not changed significantly over the past several decades However, the computational features of models have advanced to the point of providing a significant amount of model input and output data being available to sift through This allows the model user a greater degree of resolution to conform with applicable modeling regulations, guidelines, and study objectives

Gaussian air dispersion models are often used in support of risk assessments When Gaussian models are applied, the atmosphere is assumed to be homogeneous, with the source and meteorological parameters being steady-state for the interval

of time that the air concentrations are predicted (e.g., one-hour average) This model assumes that maximum chemical concentration occurs at the center of the cloud

or along the plume centerline axis, and that the concentration drops off with increasing vertical or crosswind distance from the plume centerline, thus appearing like the familiar bell-shaped “normal distribution” statistical curve in the vertical and horizontal

Not all Gaussian models are the same, and their dissimilarities can generate quite different answers from the same input data Dispersion coefficients define the rate

of plume spread with distance in models, and depend on whether the study region

is considered urban or rural Selecting urban or rural scenarios results in changes in dispersion coefficients, wind profiles (e.g., rate of change in wind speed with increas-ing height above ground), and atmospheric mixincreas-ing height (depth of atmosphere that the plume readily disperses within)

Gaussian model outputs vary but are generally a concentration or deposition rate for a unit time interval (e.g., hour, day, annual average, etc.) at a given receptor point Standard model averaging times used for exposure assessment purposes range from 1-hour to 24-hours to evaluate acute impacts (irritants, systemic toxicants), and up to annual average to assess long-term chronic noncarcinogenic and carcino-genic impacts

Regulatory agencies typically require either one year of on-site, or five years of representative off-site meteorological data to be used in refined modeling analyses When more than one year of meteorological data is used for risk assessment mod-eling, the year producing the highest annual average impact within the five year data block is commonly used in the exposure assessment However, it is not unreasonable

to average the multiyear impacts, predicted at each modeled receptor, to derive a five-year average impact when performing long-term (e.g., 70 year lifetime) average carcinogenic and noncarcinogenic impact assessments

Gaussian dispersion models are relatively straight forward and easy to apply compared to other statistical and physical models They produce results that agree with experimental data as well as any model Hence, most of the air modeling formulations for risk assessment applications are Gaussian models The most popular and versatile Gaussian dispersion model, to develop air contaminant concentration and deposition predictions for use in risk assessments, is EPA’s ISC model The ISC model, originally developed in 1979, remains the “work horse” model for a wide variety of model applications in relatively “simple” terrain settings ISC can be used

to simulate dispersion from point, area, and line-type sources ISC is also the only EPA-approved dispersion model capable of estimating the effects, from building-induced downwash, on the distribution of downwind ground-level concentration impacts ISC can be used to calculate maximum 1, 3, 8, and 24-hour, monthly, calendar quarter, and annual average concentration impacts at each receptor location with a full year (8,760 hours), or for multiple years, of hourly meteorology data This model, along with numerous other Gaussian dispersion models, are described

in EPA’s Guideline on Air Quality Models (1993) Other EPA dispersion models are available to evaluate impacts in complex terrain settings where terrain height exceeds stack top height

Dispersion models can be used in either a refined or screening fashion depending

on the application Screening modeling produces “worst case” concentration esti-mates Screening modeling can be relatively quick to apply, less computer intensive, and more conservative The standard approach for screening modeling is to assume

a set of hourly meteorological data that represents a wide range of possible meteo-rological conditions (about two dozen combinations of hourly wind direction, wind

speed, and atmospheric stability class) Screening modeling can help to: (1) initially confirm which sources in a multisource region or complex may cause the greatest concentration impacts at key receptor locations, (2) confirm whether complex terrain models also need to be applied, and (3) confirm the receptor grid configuration for the refined dispersion modeling Screening modeling usually yields overly conser-vative results which are typically inappropriate for detailed risk assessment analysis purposes As discussed before, refined modeling uses at least a full year of hourly meteorological data

C Deposition Modeling

Deposition modeling is a method of accounting for the transfer of air contaminants from the ambient air to environmental surfaces Deposition modeling accounts for the concentration of the contaminant in ambient air that is subsequently deposited onto the surface feature at ground-level (e.g., vegetation, soil, lakes) This transfer, or deposition, affects the availability of air contaminants for human (or ecological) exposure via indirect pathways (e.g., dermal and ingestion exposure routes) rather than from direct inhalation The removal of pollutants from the atmosphere can be represented by two processes — dry and wet deposition Dry deposition modeling accounts for both gravitational settling and deposition due to other atmospheric pro-cesses, and hence, can be used for all particle sizes Dry deposition of particles is modeled as the result of several processes including gravitational settling, eddy motion (atmospheric turbulence), Brownian motion, and electrostatic attraction Wet deposi-tion of particles can account for precipitadeposi-tion washout from a dispersing stack plume The approach used in the ISC model is especially well-suited for predicting deposition of submicron particles for which deposition rate increases with decreasing particle diameter It is these finer particles in which certain trace organic compounds, such as PAH, PCB, and dioxins/furans, and heavy metals, such as lead, cadmium, and mercury, are assumed to be primarily associated with, such as from waste combustion sources Due to the greater ratio of the particle surface area to volume, these trace contaminants will preferentially adsorb or condense onto the finest-sized particulates Dry deposition model algorithms handle different particle sizes, in the analysis of the surface deposition of air contaminants, that are either bound to the particle surface or included as part of the matrix of the particle The particle surface-area fraction distribution is used in the analysis of air contaminants that are bound

to the particle surface, while the mass fraction distribution is used if the contaminants are part of the matrix of the particle The dry deposition rate is proportional to the ambient air contaminant concentration immediately above the ground surface Dry deposition modeling is generally based on applying a calculated particle deposition velocity which is based on particle size, particle density, wind speed, atmospheric stability, air temperature, and surface roughness parameters The par-ticle deposition velocity is multiplied by the predicted ambient air concentration at each modeled receptor, which results in a deposition rate to the ground or water body surface Compared to water surfaces, the calculated dry deposition rate is normally greater over land surfaces, due to the greater associated surface roughness, which increases the particle deposition velocity