EPA/600/A-85/104
TITLE OF SYMPOSIUM:
Symposium on methods for characterizing indoor sources and sinks
AUTHORS' NAMES:
Leslie E, Sparks1, Lars Molhave2, and Sten Dueholm3
TITLE OF PAPER":
Source testing and data analysis for exposure and risk assessment of indoor
pollutant sources
AUTHORS* AFFILIATIONS
1	Senior Chemical Engineer,, U. S. Environmental Protection Agency, Air and Energy
Engineering Research Laboratory, MD-54, Triangle Park, NC 27711
2	University of Aarhus, Aarhus, Denmark
3	Wesser and Dueholm, Copenhagen, Denmark

-------
2
Abstract
A major purpose of source testing is to provide information needed to determine the
impact of sources on the indoor environment. Both the source testing data and the data
analysis are important for meeting this need. Ideally the experiments would run long
enough to capture all of the emissions from the source. Unfortunately, such testing is not
practical for many types of sources such as pressed wood products and many building
materials. Then the source emission models and consequent risk assessment must be
based on incomplete knowledge of the total source emissions. The types of source data
and source data analysis needed for development of source risk assessments are discussed
in this paper. Suggestions for dealing with imperfectly characterized sources are made.
The paper discusses only those aspects of the risk assessment related to the source and
source emissions model. Aspects of risk assessment related to activity patterns and health
effects of pollutants are not discussed.
Keywords: Source testing, risk assessment, exposure, IAQ modeling, source modeling

-------
3
Introduction
Evaluation of the impact of indoor pollutant sources on indoor air quality (IAQ) and on
the risk to building occupants requires an understanding of several factors including the
source of the indoor pollutants, air exchange between the building and the outdoors, air
movement within the building, interactions of the pollutant with surfaces within the
building (i.e., sink effects), chemical or physical interactions affecting the pollutant
concentration, individual activity patterns, and effects of the pollutants at various doses.
A suggested process for carrying out these evaluations is shown in Figure 1, an extension
of ideas presented by Tichenor et al. [1].
The success of source evaluation depends in great part on the data provided by source
testing. These data must provide information necessary to develop a source model that
predicts the emission rate as a function of time over the lifetime of the source. The source
model must provide an adequate description of the peak emission rate, the long-term
emission rate, and the total amount of pollutant emitted from the source.
The design of the source testing program must consider the needs of all steps in the
process shown in Figure 1. For example, a source testing program that provides data
necessary to predict the total emissions from a source is of little use if the main effects of
the pollutant are due to peak exposure. On the other hand, a source test program that
provides good information on the peak emissions but provides little information on the
total emissions from the product is of little use if the effects of interest are due to total or
time integrated exposure.
The purpose of this paper is to review the source evaluation process with emphasis on the
types of information needed from source testing and source modeling. Examples using
data from testing real sources will be used to illustrate important points. The emphasis of
the paper is on source testing and source modeling; however, some discussion of general
IAQ modeling, exposure modeling, pollutant effects, and risk analysis is provided to place
the source testing and source modeling requirements in context.
Source Testing
"It is emphasized that small chamber evaluations are used to determine source emission
rates. These rates are then used in appropriate IAQ models to predict indoor
concentrations of the compounds emitted from the tested material. Consultation with IAQ
modelers may be required to ensure that the small chamber test regime is consistent with
the IAQ model assumptions. The concentrations observed in the chambers themselves
should not be used as a substitute for concentrations expected in full-scale indoor
environments." ASTM Standard guide for small scale environmental chamber
determinations of organic emissions from indoor materials/products ( D-5116-90).
As indicated by the quote from ASTM D-5116-90, the objective of source testing is to
measure source characteristics that, with subsequent mathematical modeling, can be used

-------
4
to predict the impact of the source. Source testing is not intended to be a small scale
physical simulation of the indoor environment. However, source tests must be conducted
to ensure that the source model developed from the tests can be scaled to deal with actual
indoor environments.
Procedures for dynamic testing in small chambers are discussed by Tichenor [2] and in
ASTM D-5116-90. These procedures provide a means of determining data for source
model development using dynamic tests in environmental chambers. In these tests the
pollutant source is placed in an environmental chamber. The chamber is then sealed, and a
constant flow of clean air (air that has been filtered to remove particulate and gaseous
contaminants) is passed through the chamber. The concentration of the pollutant leaving
the chamber is measured at various times until the end of the test. Because the air in the
chamber is assumed to be well mixed, the concentration leaving the chamber is the same as
the concentration in the chamber. The data provide the time history of the pollutant
concentration in the chamber. Procedures for obtaining an empirical source emission
model from these data are discussed by Tichenor [2] and ASTM D-5116-90. Similar
procedures can be used for large chambers.
Feigley et al. [3] discuss experimental methods for determining emission rates from
coatings. Matthews et al. [4] and others have described methods for studying the emission
of formaldehyde from wood products using large chambers. In these studies, the source is
generally treated as a steady-state source; that is, the emission rate is assumed to be
constant over the time period of interest.
Source testing of the long-term pollutant emissions from some sources, such as
formaldehyde and volatile organic compounds (VOCs) from building materials, has been
conducted by placing the materials in chambers for short periods of time, then calculating
the emission rate based on the assumption that the source emission rate is constant over
the time period, Mohave et al. [5], The materials are then placed in clean ventilated
storage for a period of time, and the process repeated. This type of testing provides
emission rates at various times. Because the emission rates of sources tested in this
manner change very slowly with time, the assumptions involved in developing emission
rates do not cause major errors.
Research is being conducted to develop source tests based on bioresponse instead of
chemistry [6] and [7]. In bioresponse testing a biological system is exposed to the
pollutant emissions. The response of the biological system is monitored over the emission
history of the source. This response is then used to develop a model to estimate the health
effects of the pollutants emitted from the source in actual indoor environments. The
process shown in Figure 1 can be applied to the results from bioresponse tests, if
bioresponse tests are to be used for risk assessment.

-------
5
Source models
The concentration-versus-time data from source testing must be processed and analyzed
to develop a source emission model. The general data analysis process is described by
Dunn and Tichenor [8], If the source has a constant emission rate, the emission rate, R
(mass/time per unit source size), can be calculated from
R = C(N/L)	(1)
where C is the steady state chamber concentration (mass/volume), N (air changes/time) is
the chamber air exchange rate, and L (chamber volume/source size) is the chamber
loading.
For those sources that have emission rates that decrease over time, for example wood
stain and other wet sources, the emission rate model is more complicated. For many
sources, a first-order decay model of the form
R(t) = Roe^	(2)
where R(t) is the emission rate at any time, t, Ro is the emission rate at time zero, and k is
a first-order emission rate decay constant (1/time), provides a good description of the
emission rate as a function of time. The total emittable mass is Ro/k.
Other sources can be described by second- or third-order decay models of the form
R(t) = Roe^ + R,ek + R2e^	(3)
where Ri, and R2, b and c are empirical constants. The constants for these and other
empirical models can be determined by a non-linear curve fit to the chamber concentration
versus time data. Tools for carrying out the necessary calculations are readily available.
The constants developed for empirical models are often affected by test conditions. If the
total emittable mass is increased, for example by heavy application of a wood stain, Ro
and/or k in the first-order decay model must change. If the mass transfer rate is limited by
gas-phase mass transfer, the empirical constants are affected by the air speed over the
source. Source testing should be conducted to provide scaling factors or under conditions
similar to those encountered in indoor environments.
In order to overcome the scaling problem, source models based on mass transfer processes
have also been developed. Tichenor et al. [9] proposed a mass transfer model for gas-
phase-limited mass transfer. Other examples of source models based on fundamental
processes include Christiansson et al. [10] who proposed a model for polyvinyl chloride
(PVC) flooring. Various models for drying of paint (e. g ., [11]) have also been proposed.
The long-term emissions of formaldehyde from pressed wood and other products have

-------
6
received considerable attention. The emission rates of these products are controlled by
source-phase mass transfer processes (that is, the rate limiting step is controlled by
processes occurring inside the pressed wood). Matthews et al. [4] presented a mass-
transfer-based model for formaldehyde emissions from wallboard,
IAQ and exposure modeling
As is indicated by Figure 1, the source model is combined with building characteristics in
an IAQ model to predict air concentrations inside the building. In general these
predictions should provide a room-by-room prediction of the concentration-time history of
the pollutant in the building. The predictions of the IAQ model are then used by an
exposure model to predict individual exposure.
Individual exposure is determined by the time spent at a given pollutant concentration.
Therefore, it is a function of both the building concentration time history and the
individual activity pattern-that is, where the individual is located at what time. Building
concentration can be predicted using mass-balance-based models such as EXPOSURE,
[12], Different activity patterns, for example, entering and leaving a building at different
times or moving from one room to another, result in different exposures to the same
building pollutant concentration time history. Sparks [13] discusses exposure modeling.
Two classes of exposure are of interest: instantaneous exposure, Ej,
Et = C(t)	(4)
where C(t) is the concentration to which the individual is exposed at time t; and
cumulative exposure, Ec, between times ti and t2
*2
EC = J C(t)dt	(5)
Which of these two classes of exposure is appropriate for a given situation depends on the
nature of both the pollutant and the effect. The peak exposure is the maximum of the
instantaneous exposure versus time curve. Calculation of exposure requires the value of
the pollutant concentration to which an individual is exposed and the time exposed to that
concentration. Both the value of the concentration and the time exposed to the
concentration depend on the individual activity pattern.
As is indicated by Figure 1, risk modeling is the final step in the process of determining
impact of sources on IAQ. Risk modeling integrates individual exposure, health effects,
uptake, dose, and dose response to estimate the individual (or population) risk from the
pollutant of interest [14], Because source testing and source modeling provide key inputs
to the risk analysis process, it is important to have a general idea of the types of effects
and exposures of interest. In this way the source test and source model programs can be
designed to provide the data necessary for risk analysis.

-------
7
The effects of interest may be classified as: chronic, acute, and irritation/odor. The
intensity and prevalence of these types of effects both depend to some extent on the
maximum concentration to which each is occupant is exposed, the exposure duration, and
its average concentration. Some effects may additionally have a delay between exposure
and occurrence of effects. For these the entire time profile of exposure is needed. In
general chronic effects are due to cumulative total exposure while irritation and odor
responses occur at concentrations above some threshold level.
The IAQ, exposure, and risk analysis models must handle all the effects. The source
testing requirements to provide the data necessary for these analyses are discussed below.
Source testing/source model requirements
Because source testing and the source model are closely related, the discussion of
requirements will include both. The major requirements for the idea! source model are:
•	It must describe the peak emission rate.
•	It must describe the long-term emissions in terms of average emission and the time
profile of the emission.
•	It must account for all the emittable pollutant mass in the source.
Each of these requirements, and its impact on source testing, is discussed below.
Peak emission rate
Source testing must provide data to ensure that the source mode! adequately describes the
peak emission rate. The type of data needed depends on the type of emission model. If
the emission model is empirical, source testing must provide sufficient data in the first
several hours to allow accurate curve fit to the data. Two to four data points per hour for
the first 4 to 6 hours are often necessary. Data requirements for mass-transfer-based
models are discussed in reference [9].
Failure to adequately describe the peak emission rate will make prediction of peak "
concentration and peak exposure impossible. The failure to adequately describe the peak
emission rate will reduce the reliability of the predictions of threshold effects. In many
scenarios the peak exposure is a major fraction of an individual's total cumulative
exposure. Thus failure to accurately predict the peak, can result in large errors in
prediction of total exposure. Finally, because the peak concentration often determines
how much material is adsorbed by re-emitting sinks, a poor estimate of the peak emission
rate can result in poor estimates of long-term exposure even for those individuals not
exposed to the peak.
Because emissions from sources with high initial emission rates are likely to be controlled
by gas-phase mass transfer, appropriate mass-transfer-based models can often be used to
predict initial emission rates. An example, using data from [9], is shown in Figure 2. The
data in Figure 2 are from an IAQ test house, and the model predictions are based on the

-------
8
IAQ model EXPOSURE [12]. Note that the mass transfer model provides a much better
description of the peak concentration than does the first-order decay model.
Long-term emissions
The source model must be able to account for all the mass emitted from the source. This is
the mass that determines the prevalence of effects such as cancer due to long-term
exposure to the emissions from the source. This requirement may be difficult to meet for
many sources. For example, formaldehyde emissions from many products have an
emission half life (i.e., the time required for the emission rate to decay to one half of its
initial value) of several years. Some sources appear to have a three-phase emission: a fast
emission with a half life of less than a day, a slow emission with a half life of about a year
or so, and a very slow emission with a half life of several years. An example of such a
source is shown in Figure 3. The source in Figure 3, a collection of furniture, is described
in [15], This figure also demonstrates the problems encountered in developing models for
sources with long-term emissions. Malhave et al. [15] believe that the three-phase model
is the best model for this source. However, without additional data, it is difficult to
choose between the two-phase model and the three-phase model. The three-phase model
was selected because it was believed to provide a better estimate of the very long-term (up
to 70 years) emissions.
If the emissions from a source such as that shown in Figure 3 are to be modeled using an
empirical model, the source testing must be conducted for a long time. In many cases it is
impractical to run the emission tests long enough to define the long-term emissions. If the
total emittable mass is known, an empirical model that accounts for all the emittable mass
can be developed from short-term testing. The total emittable mass can also be used to
develop a conservative estimate of the long-term risk.
Unfortunately, the total amount of emittable material is often not known. In this case the
long-term risk can be calculated by dividing the source emissions into two phases. Phase 1
covers the time covered by the source tests and uses the model developed from the source
test data. Phase 2 covers the time after the end of the source tests. Source emissions in
phase 2 are assumed to be constant at the emission rate predicted by the model at the time
the tests ended. This two-phase approach is likely to provide a conservative estimate of
the long-term risk.
A second method for dealing with the case of unknown total emittable mass is to begin
with the two-phase approach described above. If the estimated long-term risk is
acceptable, testing can stop. If the long-term risk is not acceptable, source testing is
continued for an additional time. The two-phase risk analysis is then repeated. If the risk
is acceptable, then testing can stop.
A third method for dealing with the case of unknown total emittable mass is to develop the
source emission model using the source test data. Then run the source test for an
additional period of time. If the model predicts this last emission rate accurately (the

-------
9
required accuracy should be defined before the experiment is run), the long-term risk can
be analyzed using the resulting emission model.
Total emittable mass
Source testing should be designed to provide an independent estimate of the total amount
of pollutant that can be emitted (the total emittable mass). An independent estimate of
total emittable mass provides an excellent quality assurance check to the emission model.
A simple mass balance can be calculated to determine if the emission model is adequate.
If dynamic chamber tests are being used, the mass balance of the mass emitted during
testing and that of the total emittable mass should be compared as a quality control step.
Note that the total emittable mass of pollutant is not always the same as the total amount
of pollutant present. For example, some of the pollutant can be bound to the source and
thus never emitted. Test methods must be designed to separate the emittable and non-
emittable pollutant masses.
As discussed above, total emittable mass is necessary to estimate the long-term
cumulative exposure. Also mass-transfer-based models often require the total emittable
mass as a model parameter [9], Total emittable mass can often be quantified by direct
analysis of the source (e.g., using ASTM methods described by Brezinski [16]). Good
methods for estimating total emittable mass do not exist for some sources, such as
formaldehyde emissions from particle board. The reasons for the lack of good methods for
determining the total emittable mass vary depending on the source. In the case of
formaldehyde emissions from particle board, test methods to determine total formaldehyde
can release formaldehyde that is bound to the board. Additional research to develop
methods for estimating total emittable mass from these sources would be useful.
Other considerations
Although source tests are not intended to simulate real building conditions, the tests
should be conducted under conditions that allow easy scaling of the test results to building
conditions. The requirements of the source emission model will determine how best to
achieve easy scaling. If dynamic chambers are used, the emission regime should be
consistent with typical indoor environments. For example, dynamic chamber tests of
sources governed by gas-phase mass transfer should be conducted at air speeds of 0.05 to
0,1 m/s near the surface of the source. (These air speeds are typical of those found in
indoor environments.) A small fan can be used to achieve such conditions. Velocity can
be measured with hot wire or film anemometers.
The results of source tests should be reported in terms of a source emission model. The
data reduction used to develop the emission rate model should be consistent with the
source behavior and the type of source model. A common error in analyzing source test
data is to calculate daily emission rates for time dependent sources by assuming that the
emissions are constant for each day. The source emission model produced by such an

-------
10
analysis is incorrect. The methods discussed in [2] and ASTM D-5116-90 must be used to
develop the emission model.
Conclusions
Source testing is intended to provide data needed to evaluate the impact of the source on
the indoor environment. This evaluation is accomplished, not by trying to physically
simulate the indoor environment in the chambers, but by mathematical modeling. The
source tests provide data needed to build mathematical models of the source emissions.
The models must account for the time history of the emissions and accurately predict the
peak emission rate as well as the long-term emission rate. The model should also predict
the total emissions from the source. The design of the source test program depends on the
needs of the source model and the nature of the effects, exposures, and risks being
modeled.
Where significant source emissions occur for several years, source testing to define the
long-term emissions is difficult. The best way to deai with this situation is to obtain an
independent estimate of the total emittable mass. A risk assessment can then be conducted
using the available test data and the total emittable mass. If an estimate of the total
emittable mass is not possible, one of the methods below could be applied:
1.	Conduct the risk analysis based on the source emission model developed from the data
and the assumption that the source emission rate for all times after the last
measurement is constant with a value equal to that calculated for the last measurement
time.
2.	Conduct the risk analysis as above and if the estimated risk is acceptable stop testing.
If the risk is not acceptable, continue testing and repeat the risk analysis using the new
data.
3.	Use the data to develop a source emission model. Continue testing for at least 25
percent of the testing time, and use the model to calculate the emission rate for the
new tests. If the model prediction is accurate, use the model for the risk analysis.

-------
11
References
[1]	Tichenor, B. A., Sparks, L. E., White, J. B., and Jackson, M. D., "Evaluating Sources
oflndoor Air Pollution," Journal Air and Waste Management Association. 40 pp. 487-
492, 1990.
[2]	Tichenor, B. A., "Indoor Air Sources: Using Small Environmental Test Chambers to
Characterize Organic Emissions from Indoor Materials and Products," EPA Report EPA-
600/8-89-074 (NTIS PB 90-110131), Research Triangle Park, NC, 1989.
[3]	Feigley, C. E., Ehmke, F. M , Goodson, T. H., and Brown, J. R. "Experimental
Determination of Volatile Evolution Rates from Coated Surfaces," American Industrial
Hygiene Association Journal, 42, pp. 365-372, 1981.
[4]	Matthews, T. G , Wilson, P. L., Thompson, A. J., Mason, M. A., Bailey, S. N., and
Nelms, L. H. "Interlaboratory Comparison of Formaldehyde Emissions from
Particleboard Underlayment in Small-scale Environmental Chambers," Journal of the Air
Pollution Control Association, 37, pp. 1320-1326, 1987.
[5]	Malhave L., Sparks, L. E., Wolkoff, P., Clausen, P. A , Nielsen, P. A., and Bergsee,
N. C. "The Danish Twin-Apartment Study—Part II: Mathematical Modeling of the
Relative Strength of Sources oflndoor Air Pollution," Indoor Air In press, 1994.
[6]	Tucker, W. G , "Characterizing Emissions and Health Effects of Sources oflndoor
Air Contaminants," in Sources of Indoor Air Contaminants Characterizing Emissions and
Health Impacts, Tucker, W. G., B, P. Leaderer, L. Malhave, and W. S. Cain Editors,
Annals of the New York Academy of Sciences, Volume 641, pp. 1 -6, 1992.
[7]	MMhave L. "Controlled Experiments for Studies of the Sick Building Syndrome," in
Sources of Indoor Air Contaminants Characterizing Emissions and Health Impacts, -
Tucker, W. G., B. P. Leaderer, L. Malhave, and W. S. Cain Editors, Annals of the New
York Academy of Sciences, Volume 641, 46, 1992.
[8]	Dunn, J. E. and Tichenor, B. A. "Compensating for Sink Effects in Emissions
Test Chambers by Mathematical Modeling," Atmospheric Environment: 22, pp. 885-
894, 1988.
[9]	Tichenor, B. A., Guo, Z., and Sparks, L. E. "Fundamental Mass Transfer Mode! for
Indoor Air Emissions from Surface Coatings," Indoor Air, 3, pp. 263-268, 1993.
[10]	Christiansson, J., Yu, J., and Neretnieks, I. "Emissions of VOC's from PVC
Floorings-Models for Predicting the Time Dependent Emission Rates and Resulting

-------
12
Concentrations in the Indoor Air," Proceedings of Indoor Air 93, Vol. 2, pp. 389-394,
Helsinki, 1993.
[11]	Hansen, C. M. "The Air Drying of Latex Coatings," Industrial Engineering
Chemistry, Product Research Development, 13 (2), pp. 150-153, 1974,
[12]	Sparks, L. E. "EXPOSURE Version 2: A computer model for analyzing the effects
of indoor air pollutant sources on individual exposure," EPA-600/S-91-013 (NTIS PB 91-
201095), Research Triangle Park, NC, 1991.
[13]	Sparks, L. E., "Modeling Indoor Concentrations and Exposures," in Sources of
Indoor Air Contaminants Characterizing Emissions and Health Impacts, Tucker, W. G.,
B. P. Leaderer, L. Molhave, and W. S. Cain Editors, Annals of the New York Academy of
Sciences, Volume 641, pp. 102-111, 1992.
[14]	Naugle, D. F. and Pierson, T. K. "A framework for risk characterization of
environmental pollutants," Journal of Air and Waste Management Association, 41: pp.
1298-1307, 1991.
[15]	Malhave L., Dueholm, S., and Jensen, L K. "Health Assessment and Risk
Evaluation of Emissions from Furniture' a Case Study," Indoor Air In press, 1994.
[16]	Brezinski, J. J., Editor Manual on Determination of Volatile Organic Compounds in
Paints, Inks, and Related Coating Products, ASTM Manual Series, MNL4, 1989,
Philadelphia, PA.

-------
Risk analysis process
Source Source
Testing Modeling
Source
Size
Emissions
Ventilation			
Building factors (e.g., sinks)	
Source usage	
Occupancy.
IAQ Exposure Risk
Modeling Modeling Modeling
Air
Concentration
Exposure
Occupant sensitivity,
Dose-response	
Figure 1. Principles of assessing risk due to indoor sources.

-------
14
400
350
300
1 250
E
1	200
+5
2
1 150
u
c
o
° 100 -
50
0
-Mass transfer model
First-order decay
model
¦ Data
10
15
20
Time (hr)
Figure 2. Use of mass-transfer-based model to improve prediction
of the initial concentrations.

-------
15
\
-p,
y.
Chamber data
-Phase 1
Phase 2
- Phase 3
0
500
1000
Time (hr)
1500
2000
Figure 3. Formaldehyde emissions showing
three-phase emissions [15].

-------
accdt T3 ioco TECHNICAL REPORT DATA
i Ie> Sparks (EPA), L. Molhave (Aarhus),
and S. Dueholm (Wesser and Dueholm)
8. PERFORMING ORGANIZATION REPORT NO.
9, PERFORMING ORGANIZATION NAME AND ADDRESS
University of Aarhus Wesser and Dueholm
Aarhus, Denmark Copenhagen, Denmark
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
NA (Inhouse)
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Published paper; 9/93-6/94
14. SPONSORING AGENCY CODE
EPA/600/13
13. supplementary notes AEERL project officer is Leslie E. Sparks, Mail Drop 54, 919/
541-2458. Presented at ASTM Symposium on Methods for Characterizing Indoor
Sources and Sinks, Washington, DC, 9/25-28/94.
is. abstract paper reviews the source evaluation process with emphasis on the
types of information needed from source testing and source modeling. A major pur-
pose of source testing is to provide information needed to determine the impact of
sources on the indoor environment. Both the source testing data and the data analy-
sis are important for meeting this need. Ideally, the experiments would run long
enough to capture all of the emissions from the source. Unfortunately, such testing
is not practical for many types of sources such as pressed wood products and many
building materials. Then the source emission models and consequent risk assess-
ment must be based on incomplete knowledge of the total source emissions. The
types of source data and source data analysis needed for development of source risk
assessments are discussed. Suggestions for dealing with imperfectly characterized
sources are made. The paper dicusses only those aspects of the risk assessment
related to the source and source emissions model. Aspects of risk assessment re-
lated to activity patterns and health effects of pollutants are not discussed.
17. KEY WORDS AND DOCUMENT ANALYSIS
2. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Pollution
Mathematical Models
Exposure
Data Processing
Pollu tion Control
Stationary Sources
Indoor Air
Source Testing
Risk Assessment
Source Modeling
13B
12 A
06S
09B
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)

-------