United States
Environmental Protection
Agency
Atmospheric Research and Exposure :, ;?>
Assessment Laboratory ~ ^ ' ^
Research Triangle Park NC 27711 / /, ^ x
Research and Development
EPA/600/S3-88/056 Aug. 1989
&EPA Project Summary
Development of Sampling
Methods for Source PM10
Emissions
Ashley D. Williamson, William E. Farthing, Sherry S. Dawes,
Joseph 0. McCain, Randal S. Martin, and James W. Ragland
This report describes an investi-
gation of the needs and available
techniques for in-stack PM10 samp-
ling. Discussion includes the con-
ceptualization, development, docu-
mentation, and testing of two
candidate methods. The first method,
Constant Sampling Rate (CSR), is a
procedural approach which adds
particle size separation to sampling
hardware that has been widely used
in EPA Methods S and 17 but modifies
the sampling protocol to accomplish
the PMto objectives. The second
method, Exhaust Gas Recycle (EGR),
is an equipment approach which ac-
complishes the PM10 objectives by
using a modified sampling train to
implement the concept of exhaust
gas ^circulation.
Six field studies indicated that
these techniques were practical and
compared well with one another and
with more labor-intensive ap-
proaches. Laboratory investigations
with monodisperse aerosols indi-
cated that commonly used geomet-
ries for sampling nozzles could
cause a decrease in the particle size
cut of a closely coupled inertia!
sizing device. Nozzle geometries
were also found which eliminated the
observed shifts in particle size cut
This Project Summary was devel-
oped by EPA's Atmospheric Research
and Exposure Assessment Laboratory,
Research Triangle Park, NC, to
announce key findings of the research
project that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).
Introduction
A size-specific PM10 ambient-air
particulate standard has been promul-
gated by EPA. The Quality Assurance
Division of the Atmospheric Research
and Exposure Assessment Laboratory
(AREAL) has initiated a research program
to develop cost-effective source meas-
urement techniques to support the PM10
standard. This report summarizes the
source PM10 method development work
performed at Southern Research Institute
(SRI) under EPA Contracts 68-02-3118
and 68-02-3696 and EPA Cooperative
Agreement CR-812274. Much of this
material is described more fully in other
reports, which are referenced in this
report.
The extensive particle size sampling
technology, developed as a result of
research efforts associated with Inhalable
Particulate (IP) matter from stationary
sources and particulate control devices,
provided valuable background informa-
tion for the PM10 efforts. The technical
difficulties in size-specific (PM10) particu-
late sampling are greater than, but similar
to, those of total particulate sampling by
EPA Reference Methods 5 or 17. In
Methods 5 and 17, potential sampling
biases exist due to variations in the
spatial distribution of particulate concen-
trations across the sampling plane
defined by the duct cross section. This
type of bias is limited by specifying the
minimum number of traverse points. Like-
wise, temporal variations due to process
-------
variations can also cause inaccurate or
unrepresentative emission measure-
ments. Thus, three traverses are required
to limit this type of error. Another
potential error in particulate measure-
ments is duct/nozzle sampling bias.
Unless the gas velocity entering the
sampling nozzle (plug velocity) equals
the local duct velocity, particulate matter
will be selectively depleted or enriched in
the sample gas stream because of inertial
separation at the nozzle entrance.
Percent isokinetic is limited to 100 ±
10% in Methods 5 and 17.
These potential errors are more difficult
to control in PM10 sampling because of
the additional requirement of aero-
dynamic size classification, which is
achieved by inertial techniques involving
aerodynamic drag on aerosol particles.
Any errors in the inertial cutoff diameter,
which is primarily determined by flow
rate through the size separator, will lead
to errors in the PM10 measurement by
misclassification of particulate matter in
the size range near 10 pm. Thus, this
flow rate must be held constant. Without
a sampling nozzle of continuously vari-
able cross-sectional area, this require-
ment for a fixed flow rate precludes
isokinetic sampling in the direct manner
of Method 5 or 17.
Previous work on this problem led to
the development of two candidate sam-
pling methods, the Exhaust Gas Recycle
(EGR) sampling train and the Constant
Sampling Rate (CSR) traversing protocol.
The EGR train maintains isokinetic flow of
gas into the sampling nozzle and
augments it with an adjustable amount of
filtered, recycled stack gas upstream of
the inertial sizing device. In this manner,
the total flow through the EGR inertial
sizing device is held to the constant value
required for classification of particles
larger and smaller than 10 urn. The CSR
protocol is an alternate PM10 technique
which uses existing sampling equipment
without special gas recycle adaptations.
In the EGR train, stack gas is iso-
kinetically extracted through the sample
inlet portion of the EGR mixing nozzle
into the size separation device of the
sampling train. After passing the size
separator and in-stack filter, the sample
gas passes through the probe and
condenser or impinger train and into the
EGR flow control module. As in con-
ventional Method 5 control modules, the
gas flow rate entering the control module
is controlled by coarse and fine control
valves at the entrance of the sealed
pump. At the exit of the pump and
absolute filter, the total flow is measured
with a laminar flow element. The gas
stream is then split into the recycle and
sample flow lines. The sample flow is
monitored in the normal manner by using
a dry gas meter and a calibrated orifice.
The partitioning between sample and
recycle gas is controlled by a third valve
located in the recycle flow line. The
recycle gas line, along with the sample
and pitot lines, passes through the
heated probe in which the recirculated
gas is reheated to the duct temperature.
Power to the heater is regulated by a
proportional temperature controller with a
thermocouple reference sensor located in
the recycle gas stream.
The CSR is a procedural approach
which simply adds a particle size sepa-
rator (cyclones or cascade impactors) to
the basic sampling train already in use.
The objective of the protocol is to limit
error due to anisokinetic sampling to the
approximate range expected from the
spatial and temporal variation of the
emissions. Anisokinetic sampling bias is
held within these limits in most sampling
situations by performing a full duct
traverse with a single nozzle. However, in
the very unusual situation of large
velocity variation within the sampling
plane, the traverse may be synthesized
from two or more partial traverses using a
different nozzle for each partial traverse.
Thus, the flow rate through the inertial
sampler is held at the level required for a
10-nm size cut over the full traverse. The
range of duct velocities over which a
given sample nozzle may be used is
such that the combination of nozzle inlet
diameter and PM10 flow rate results in
anisokinetic sampling errors less than
±20% for 10-nm particles. Since corre-
sponding errors for particle sizes less
than 10 nm are much smaller (decreasing
proportionately to particle size squared)
and since some of these errors are of
opposite sign, actual anisokinetic samp-
ling error for PM10 will be much less than
20% in magnitude.
It was decided that this program should
primarily address utilization of a single-
stage size separator. The largest cyclone
(Cyclone I) of an existing five-series
cyclone train was chosen. More detailed
equipment descriptions and operating
protocols for the EGR train and the CSR
procedure are also given in the project
report.
Field Studies
As a key part of the PM10 development
program, six field studies were conducted
at four emissions sources. In the course
of these field tests, the two candidate
methods were refined and tested, and
PM10 measurements were performed at
range of source conditions.
Since both PM10 methods ha\
hardware or procedural elements whk
are different from other source samplir
methods, the field studies were used i
means of obtaining basic data aboi
these new procedures, as well s
development, refinement, and validatic
of the overall methods. The first objectiv
of the field studies was to test and refin
the procedures and sampling hardware <
both PM10 methods. A second objectiv
was to obtain comparison measuremenl
of PM10 and total particulate concer
tration by each method and by the be;
available reference measurements. Th
third objective was to measure the pre
cision of each PM10 method at
common source and compare these pre
cision measurements to precision meas
urements using Method 17.
To meet the objectives describe!
above, careful attention to test desigi
was required. While the detailed desigi
of each test varied according to the pri
mary objective for the test and th<
specifics of the test site, certain element!
were common to several tests. These
include use of independent measure
ments as the best available reference or
the accuracy of each technique, contra
of external variables by maximum feas'
ible use of collocated and simultaneous
sample protocols, and site selection foi
significant challenge of the methods over
a range of source conditions.
Several conclusions may be drawn
from the field data summarized in Table
1, as well as the more complete data sets
given in the full project report. First, in
every instance the average concen-
trations measured by different techniques
agreed within the combined 95% con-
fidence intervals. Since these intervals for
some tests reflect a substantial degree of
variation, presumably due mostly to
source fluctuations, a more meaningful
comparison can be drawn from paired-
run analysis of the simultaneous
measeurements indicated in Table 1.
At both site 1 and site 4 in tests 1, 5,
and 6, the EGR train measured less total
particulate than Method 17 by a small but
significant amount. Mean differences
ranging from 5 to 13% were observed,
and in each case these differences were
larger than the 95% confidence limits.
The reason for this small bias is not clear;
however, since it does not exceed 15%
for any of the sites tested, this bias is not
considered detrimental. CSR total mass
measurements at sites 1, 3, and 4 in testil
2, 4, and 6 were not significantly different
from the paired total mass measurements
-------
Table 1.
Percentage Differences in Particulate Concentrations Measured During Test
Series8
Tesf Number
Number of
Replications
PM
10
Total
Concentration
1 EGR Initial Test: Site 1
EGR Cyclone - Isokinetic Cycloneb
EGR - Method 17*
-8.3 ±27
9.0+29
-11.5+8.3
2 CSR Initial Test: Site 1
CSR-Method 17*
CSfl - Isokinetic lmpactorsb
3 EGR/CSR Comparison Test: Site 2
EGR Cyclone - CSR Cycloneb
EGR - Isokinetic Impactors
CSR - Isokinetic Impactors
4 EGR/CSR Comparison Test: Site 3
EGR - CSfl"
EGR - Impactor
CSR - Impactor
CSR Impactor - Method 17
5 EGR Precision Test: Site 4
EGR-Method 17*
EGR-I
5
5-6*
Inlet
6
6-5=
6-5C
Outlet
-
-1.8 ±22
-15.5+6.5
-11+31
3.8 ±25
11 ±9.8
27 + 16
16 + 16
-2.4+4.9
-16+32
-14.0+65
-9.2 ±8.5
1.3+38
14+31
1.7+21
-9.8 ±16
-11+14
-7.4+23
-12.9+4.2
-0.9 ±4.3
6 CSR Precision Test: Site 4
CSR - Method 17*
EGR -Method 17*
EGR - CSfl*
CSfl, - CSfl2«>
9
7
9
9
-T5.8±7,8
6.6+3.8
0.4+6.3
-4.8 ±1.7
1.2+5.4
*AH differences and confidence intervals expressed as percentages of the mean value.
Confidence intervals represent 95% significant level.
b These comparisons were analyzed as pairs since the measurements were simultaneous.
cWhere two numbers of replications are given, the first number corresponds to the first listed
device and the second to the second device.
from Method 17 or other reference
isokinetic sampling trains. Since the CSR
technique is expected to be less accurate
for total mass, these results are encour-
aging. When total mass data using the
two techniques are compared, the results
are mixed. At site 2 in test 3 the 9%
EGR—CSR difference is marginally sig-
nificant at the 95% confidence level. At
sites 3 and 4 the EGR and CSR data are
essentially the same.
The PM10 values measured by the two
techniques differ at every site by more
than 10% but less than 20%. At sites 2
and 4, the EGR PM10 value is about 15%
less than the CSR value. At site 3, the
EGR value is 11% greater than the CSR
value. All three differences are statis-
tically significant at the 95% confidence
level. The results at site 3 reverse what
would appear to be a trend at the other
tiree sites for EGR PM10 values to be
lower by about 5-15% than the CSR
values, which are not significantly dif-
ferent from the individual isokinetic im-
pactors. No clear reason was found for
this test-to-test reversal.
At site 4, measurements with col-
located pairs consisting of two EGR trains
in test 5 and two CSR trains in test 6
indicated excellent reproducibility be-
tween the two trains. In only one instance
of CSR PM10 concentrations does the
mean difference in the measurements of
two nominally identical trains exceed
2.5%, and even that low bias of 6.5%
was found to be due to a systematic
difference in cyclone flow rate between
the two trains. For both PM10 trains, 95%
confidence intervals were on the order of
±5%. By this measure, the precision of
the PM10 trains was the same as that of
the paired Method 17 trains operated
during these tests.
Optimization of PMio Cyclone
and Sampling Nozzles
One further element in the testing and
refinement of both candidate methods
was the inertia! sizing device itself. While
the candidate PM10 cyclone had been
used for several years in other appli-
cations, it had not been characterized
either in the laboratory or field under
conditions typical of PM10 operation. The
versions of the cyclone which are com-
mercially available have different exterior
dimensions and nozzle designs from the
SRI prototypes which were used on the
initial studies. These differences pre-
vented design of a single EGR nozzle
system suitable for both versions of the
cyclone. Several adaptations in both
versions were necessary for use as a
PM10 precollector for a single- or dual-
stage sizing train. Prior to this work, it
was also not clear how well a lO-jim cut
-------
could be predicted over a range of stack
gas conditions with the cyclone, either in
a gas recycle or a conventional nozzle
configuration. During the test series,
several of these potential difficulties were
clarified or resolved.
Calibrations of Cyclone I were per-
formed with a vibrating orifice aerosol
generator (VOAG). The VOAG provides
monodisperse dye aerosol of chosen
particle size at a rate of about 60,000
particles/s. After lofting and drying, the
aerosol is passed through the sampling
train which includes an absolute filter.
After the sampling run, all internal sur-
faces of the sampler and the filter are
carefully washed with a measured
volume of solvent. Spectrometry or fluor-
ometry techniques are then used to
determine the concentration of dye in the
wash solutions and thus the collected
aerosol mass for each surface and the
mass captured by the backup filter. The
dye particles utilized in this laboratory
investigation were composed of dry
ammonium fluorescein.
To simulate sampling from process
streams, an apparatus for the calibration
studies was designed that established a
sample flow stream substantially larger
than the diameter of the sample nozzles.
The sample flow stream had a uniform
velocity profile at (or near) the nozzle
inlet and resulted in only minimum
dilution of the VOAG aerosol. In addition,
to understand better the effect of the
nozzle geometry on the particle sizing
performance of Cyclone I, a system was
developed to obtain high-resolution
velocity profiles at the cyclone inlet for
each of the nozzle geometries calibrated.
To correlate these data with the collection
efficiency data, the velocity profile was
measured at conditions which simulated
each of the cyclone calibration con-
ditions. The velocity sensing device used
in the test section was a amall pitot made
of two hypodermic needles (0.03-in.
diameter) with beveled openings approxi-
mately 0.06 in. in length. For the purpose
of traversing the test section in known
increments, the pitot was mounted on a
horizontal positioner attached to a vernier
scale (reproducible to 0.001 in.).
In addition to the 1/2-in. nozzle, which
has the largest sampling diameter and
which was used as a reference, the other
existing nozzles used for Cyclone I were
classified into the following three types:
tapered nonrecycle, large expansion non-
recycle, and recycle. Test results for
Cyclone I collection, nozzle efficiency,
and velocity profile were presented by
nozzle type.
The tapered-nonrecycle nozzles have a
small angle of expansion from the inlet
diameter to the cyclone inlet diameter.
The behavior of Cyclone I may be slightly
different from that of the reference
nozzle, but the cut size is changed by
much less than 1 urn. Measurable nozzle
losses did, however, occur. Losses in the
larger nozzle increased with particle size
from 3% (at 4 urn) to about 20% (at 10
nm). With the smaller nozzle, losses of
about 22% were found, which decreased
only slightly for the smaller particle sizes.
Cyclone I collection efficiency was
measured for all nonrecycle large expan-
sion nozzles, in which a large expansion
angle within the nozzle is the sample
aerosol pathway. The EGR nozzles with
zero recycle air are included in this group
since they present an abrupt expansion
to the flow at the end of the nozzle
sample tube. The cyclone cut diameter
shifted down to about 6 urn with all of
these nozzles. The highest nozzle depo-
sition losses were incurred by the 0.138-
in. nozzle, 30% at 4-um particle size,
decreasing to 20% at 10 um. The 0.155-
in. nozzle had about a 10% loss at 4 pm
and 15% at 10 urn. Nozzle optimization
studies have minimized these problems.
Efficiencies were measured for recycle
(EGR) nozzles at multiple recycle rates.
In each instance, efficiency was higher
(or cut size smaller) for the lower recycle
rate. All three nozzle sizes caused cuts to
vary from about 6 pm at 0% or 10%
recycle to about 9 um at 75% recycle.
Nozzle losses with the 1/4-in. and 1/8-in.
EGR nozzles were insignificant (at the
<2% level) at all recycle rates studied.
For the 1/8-in. EGR nozzle, losses were
low (~3%) at the 75% recycle rate. At
0% and 10% recycle, the nozzle losses
at the 4-iim particle size were about 20%
and dropped to 2% for lO-pm particles.
Further measurements were performed
to test approaches for eliminating the
observed shift in cyclone cut size at the
higher nozzle inlet velocities. The results
obtained with the original nozzles
indicated that the inertia of the higher
velocity aerosol streams was not dis-
sipated sufficiently to prevent additional
impaction in the cyclone. Therefore,
modified nozzles were tested which
reduced the sample gas velocity prior to
entering the cyclone. Two types of
modified nonrecycle nozzles were tested.
Both were extensions of nozzle lengths
beyond the original nozzles of the same
inlet diameter, one group having large
expansion angles, >45°, and the other
group having small tapered expansion
angles of 7° and 14" (total included
angle). One type of modification to the
EGR nozzles was tested extensively. This
was a simple extension of the nozzle
length so that more distance was
available for expansion.
The extended EGR nozzles gave
cyclone behavior identical to the refer-
ence nozzle. However, these modifiec
nozzles had substantially higher nozzle
losses than the unextended EGR nozzles.
The extended 1/8-in. EGR nozzle witr
expansion distance of 3.1 in. exhibitec
nozzle losses of 20% at the 4-iim particle
size and 35% at 8 um.
Nonrecycle 1/8-in. nozzles having large
expansion angles and expansion dis-
tances greater than 2.2 in. improved
cyclone efficiencies to those of the
reference nozzle. The shorter of the iwc
nozzles tested (2.2- and 3.Hn. expansion
distance) exhibited tower nozzle losses a1
the 8-um particle size, 36% compared tc
47%.
The tapered-nonrecycle nozzles were
compared to nozzles having large expan-
sion angles and the same inlet diameter.
The 0.16-in. nozzle with 7° angle elim-
inated the undesired effect on cyclone
behavior caused by the original 0.154-in.
nozzle that had an abrupt expansion
angle and short length. The 0.16-in.
nozzle with 7° angle had 5 to 10% highet
nozzle losses, ranging from 10% at 4-virrl
particle size to 20% at 10 iim.
Cyclone behavior was not affected by
the 1/8-in. nozzles with 7° or 14° tapers.
Both of these had a total length of 3.2 in.,
i.e., the 14° nozzle had a straight section
at its exit end. Nozzle tosses for these
two tapered nozzles were essentially the
same. In contrast, the nozzle losses for
the 1/8-in. nozzle with an expansion
distance of 3.1 in. and an abrupt nozzle
tube expansion were higher than losses
for the two tapered nozzles of the same
length, 11 and 13% higher at 4- and 8-
um particle sizes, respectively.
The laboratory data obtained in this
PM10 program have major importance for
PM10 methods in two ways. First, the
data establish a basis for using Cyclone I
as a PM10 size separator for a wide range
of operating conditions. The efficiency
curve for Cyclone I (D^o = 10 um) has an
acceptable geometric standard deviation
of 1.4. Although the slope of the
efficiency curve may decrease somewhat
at elevated temperatures where Reynolds
number ts lower, it is expected to retain
sufficient sharpness of cut to remain
quite acceptable.
Second, the laboratory data establish
the existence of, and point to a solutiM
for, a significant effect of existing samp?
-------
ling nozzles upon cyclone behavior. It is
expected that the solution found in this
investigation could be optimized further.
It is reasonable to assume that a similar
nozzle effect occurs to some degree in
all sizing devices used in process
streams with high velocities. The ob-
served effect of small nozzles on
behavior was a shift in cut point from 10
urn to as low as 6 pm, the shift generally
decreasing as nozzle inlet velocity de-
creased. If left uncorrected, this effect
would cause measured PM10 to be lower
than actual concentrations to a degree
which depends upon the aerosol size
distribution. The cause of the shift in D^
associated with some nozzles was attrib-
uted to high inertia associated with high-
velocity gas streams. The shift in D50 was
found throughout the data to correlate
closely with the velocity of the gas
entering the nozzle.
The effect of sampling nozzle on
cyclone behavior and, hence, measured
PM10 can be eliminated by causing the
sample gas to decelerate after entering
the nozzle and before entering the
cyclone. In this present work, extending
the nozzle inlet farther from the cyclone
inlet regained the basic cyclone perform-
ance. However, nozzle losses were
enhanced. The lowest losses occurred for
tapered nozzles with expansion angles of
7° or 14°. These differences in losses
between nozzle geometries were prob-
ably caused by flow separation accom-
panied by a region of flow recirculation
with the larger expansion angles.
The analytical results based on these
laboratory data show clearly that for
further optimization studies the non-
recycle and recycle nozzles for Cyclone I
should be redesigned with a smooth
taper from the nozzle inlet diameter to
the 0.5-in. diameter of the cyclone inlet. It
appears that modifications of the EGR
nozzle should also include modifying the
recycle gas flow so that the recycle gas
will have a higher average velocity but
more uniform velocity profile. The data
obtained thus far indicate that nozzle
losses for these improved nozzle geo-
metries will be significant for small inlet
diameters or high stream velocities. Aver-
age losses for, particulate diameters
within the range studied here of 4 to 10
pm would be about 1% at 5 ft/s, to
approximately 13% between 30 and 60
ft/s, and 30% at 88 ft/s. The velocity
values relate to this laboratory study in
which PM10 flow rate (for Cyclone I) was
0.45 acfm. The PM10 flow rate and the
corresponding nozzle velocities are
typically 20 to 30% higher in field meas-
urements.
Conclusions and
Recommendations
Six field studies have been performed
to develop and characterize the methods.
As measured by a modified dual-probe
technique, the precision of each method
is better than ±5%, comparable to that
of EPA Method 17 at the same location.
Comparability of the EGR and CSR
techniques is within 16% at all sites
tested. The EGR measured lower PM10
concentrations than CSR and other refer-
ence samplers at two sites, and higher
than both at a third. All of these
differences were statistically significant at
the 95% confidence level.
Laboratory studies in this program
indicate that decreases in particle size
cut can occur for inertial sizing devices
when the sampling nozzle has a small
inlet diameter and is closely coupled with
the inertial separation stage. Such shifts
were observed to occur in Cyclone I, the
current PM10 sizing device, which was
tested with several of the current nozzles.
Shifts were observed in particular with
the three EGR nozzles and those non-
recycle nozzles which had an abrupt
expansion within a short distance from
the cyclone body. It is projected that this
effect probably occurs in other available
inertial samplers in this size range.
Optimization studies for sampling
nozzles for Cyclone I indicate that the
shifts in cyclone collection efficiency can
be eliminated by lengthening the expan-
sion zone in the nozzle. This lengthening,
however, increases particle deposition in
the nozzle. Nozzle losses averaged over
particle sizes of 4 to 10 nm were
observed to range from about 1% at low
velocity, to near 13% at medium velocity,
to 30% at high velocity. Although further
research should be directed at the nozzle
effects problem, the methodologies in
their present form are usable with accep-
table relative accuracy and precision for a
wide range of sampling situations.
In view of these conclusions, the high-
est priority recommendation for further
research is a more thorough design and
characterization study to optimize the
nozzles for use in a PM10 sampling
method, in particular the EGR nozzles.
While both EGR and CSR are usable in
their present form with no modifications
other than simple extensions of the
smallest nozzles, the current methods
appear likely to show a slight negative
bias in measured PM10, which increases
with increasing duct velocity. Nozzle
optimization and detailed specifications
on nozzle design will probably be useful
for measurement of PM10 at very high
duct velocities.
Recommendations for further research
of a somewhat lower priority can also be
made. Reduction of approximate setup
calculations for both methods to a form
suitable for programmable calculators
should be attempted. Extension of per-
formance data of the PM10 sampling
procedures to source conditions beyond
the range currently studied is desired.
Further field studies are suggested also
to test any new nozzles from the recom-
mended design study. Finally, extensive
field studies are recommended to extend
the number of source types tested by
these methods.
-------
Ashley D. Williamson, William E. Farthing, Sherry S. Dawes, Joseph D. McCain,
Randal S. Martin, and James W. Ragland are with Southern Research Institute,
Birmingham, AL 35255-5305.
Thomas £ Ward is the EPA Project Officer (see below).
The complete report, entitled "Development of Sampling Methods for Source PM10
Emissions," (Order No. PB 89-190 375/AS; Cost: $21.95, subject to change) will
be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC27711
United States Center for Environmental Research
Environmental Protection Information
Agency Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S3-88/056
------- |