United States
Environmental Protection
Agency
Atmospheric Sciences
Research Laboratory
Research Triangle Park NC 27711
J
/ ,
Research and Development
EPA/600/S3-86/063 Mar. 1987
&EPA Project Summary
Ambient Air/Source
Transport and Transformation
Relationships for Selected
Hazardous Air Pollutants
R. L. Brewer, R. J. Gordon, C.-C. Lin, R. J. Countess, T J. Peters, and
J. D. Walsh
The study was undertaken with the
goal of obtaining information that
might improve modelling techniques
for hazardous organic vapors in the at-
mosphere. It was proposed that both
source-receptor and dispersion models
be applied. Detailed examination of al-
ternative choices for pollutant, emis-
sion site and method of determination
was made. This resulted in the selec-
tion of the benzene/toluene/xylenes
class as pollutants, a waste disposal
site near Taft, California (in the San
Joaquin Valley) as source, and adsorp-
tion on Tenax cartridges from the air,
followed by thermal desorption and
gas chromatography as method of de-
termination. Cleanup procedures for
Tenax were developed that allowed
generally quite good agreement with
comparison measurements with refer-
ence mixtures supplied hi canisters by
the Atmospheric Sciences Research
Laboratory. Co-located canisters and
Tenax tubes were sometimes in agree-
ment and sometimes not. These dis-
agreements may be due in part to
interferences from other organic com-
pounds that are separated differently
on the packed GC column used for the
Tenax analyses and the capillary GC
column used for the canister analyses.
Collections of Tenax were made on sev-
eral occasions by means of a remotely
controlled system of samplers devel-
oped in this project. Poor meteorologi-
cal conditions and a sampling array
now seen to be too small prevented ob-
taining satisfactory data for modelling
purposes in 1985. In 1986 an expanded
network, better meteorology, and use
of an integrating anemometer greatly
improved data collection. It was found,
however, that the unexpected presence
of unidentified strong and irregular
sources of the target compounds pre-
vented consistent interpretation. Both
receptor and Gaussian dispersion mod-
els were examined in various forms, but
could not be applied to the collected
data.
Analyses of canister gases by capil-
lary GC at the ASRL showed a remark-
ably complex mixture of pollutants in
most of the samples. Over 200 com-
pounds were seen in some cases, and
non-methane hydrocarbons totalled as
much as 9.5 ppm carbon.
This Project Summary was devel-
oped by EPA's Atmospheric Sciences
Research Laboratory, Research Triangle
Park, NC, to announce key findings of
the research project that is fully docu-
mented in a separate report of the same
title (see Project Summary Report
ordering information at back).
Introduction
Atmospheric modelling of volatile
substances is usually done from the
source forward, by means of dispersion
models. If the concentration of pollu-
tants of interest decreases from causes
in addition to dispersion, the modelling
predictions of course are inaccurate.
Source-receptor modelling, in which
the ambient air composition is related
-------
to specific sources or source types by
means of its composition, is another ap-
proach of interest. The approach is par-
ticularly useful for certain hazardous
substances that are not found in the at-
mosphere except through emissions
from specific sources; also fingerprint
compositions identified at specific
sources are useful for this type of mod-
elling. The study was undertaken with
the aim of carrying out air sample col-
lections of one or more hazardous sub-
stances in such a way as to be applica-
ble to both types of modelling.
Phase I included the selection of pol-
lutants to be studied, the selection of
the study site, and the selection of the
sampling and analytical methodology.
Phase II included the actual sampling,
analysis, and data interpretation. Engi-
neering Science (ES), Berkeley, Califor-
nia, participated as subcontractor in
both phases; in site selection and devel-
opment of sampling strategy in Phase I,
and in meteorological forecasts and
measurements as well as data interpre-
tation in Phase II. The Atmospheric Sci-
ences and Research Laboratory (ASRL)
of the U.S. Environmental Protection
Agency (EPA) at Research Triangle Park
(RTP), North Carolina (NC), participated
by providing standard reference gas
mixtures, loans of canisters and pumps
for parallel gas sample collection, and
independent analyses of canister sam-
ples for quality assurance purposes.
The ASRL sample analyses were made
with capillary gas chromatography and
provided a very detailed hydrocarbon
breakdown for the canister samples.
Procedure
Phase I: Selection of Pollutant,
Site and Methods
The hazardous pollutant or pollutant
group was selected from the list of
"Chemicals Under Assessment" (as of
1983) in Section 112 of NATION EMIS-
SION STANDARDS FOR HAZARDOUS
AIR POLLUTANTS (NESHAPS). Ten of
these with very low volatility, or without
well-defined chemical composition,
were deleted. Ten others were added in
the process of selection, making a total
list of 48 compounds for consideration.
This list was screened on the criteria of
usage or occurrence in California, emis-
sion potential and volatility. The addi-
tional criteria of estimated ambient
half-life and estimated toxicity or car-
cinogenicity were also considered. A
further requirement was a suitable
methodology for collection and analy-
sis. The final selection of the benzene
group (benezene, toluene, and o-, m-
and p-xylenes) was based upon both
site availability and suitable methodol-
ogy for collection and analysis.
An early decision was made to con-
fine the study to the California area, be-
cause of the greatly increased cost of
sampling elsewhere in the country, and
the risk of sample deterioration during
shipping long distances back to the
Global Geochemistry Corporation
(GGC) laboratory. Potential emissions
from the chemical industry in the state
were examined, but the use of a landfill
source appeared more promising.
Moreover there was interest on the part
of both EPA and local agencies for infor-
mation on landfill emissions. The char-
acteristics of active landfills in the state
were examined. A class 11-1 landfill able
to accept all but the most toxic wastes
was found in the southern San Joaquin
Valley within a hundred miles of the
GGC laboratory. This facility has good
terrain and, based on historical records,
suitable meteorological conditions.
Small-scale pilot trials indicated the site
released suitable organic compound
levels for sampling.
The choice of sampling method was
narrowed down to either canisters or
Tenax tubes. The Tenax tubes were cho-
sen both because of their lower cost and
because they are more readily deployed
in sampling arrays, due to their light
weight and operation with battery-pow-
ered pumps. Canisters provided by
ASRL were also used, and served as
valuable means of checking on quality
control. Some development was re-
quired to determine how to clean and
prepare Tenax traps for sample collec-
tion, storage and analysis. It was found
effective to seal off the ends of the glass
cartridges containing the Tenax, after
initial conditioning and again after sam-
pling, in order to minimize contamina-
tion. Contents of the tubes were ther-
mally desorbed directly onto a packed
GC column combined with a photoion-
ization detector sensitive to C6 to C8 aro-
matics. Aliquots from the ASRL sample
canisters were passed onto Tenax tubes
for analysis, and the canisters were then
shipped to ASRL for a more detailed
analysis by capillary column GC with
flame ionization detection. Interlabora-
tory comparisons were also made by
means of EPA reference mixtures of
benzenoid hydrocarbons provided by
ASRL and analyzed by both laborato-
ries.
GGC also developed a remote radio-
controlled system for operating as
many as sixteen Tenax tube samplers
individually from one transmitter point,
to allow either simultaneous or sequen-
tial sampling as desired.
Phase II: Sampling and
Analysis
The project subcontractor, Engineer-
ing Science, advised on the meteoro-
logical conditions required and the
sampling layout needed for the mod-
elling studies. Both source-receptor and
Gaussian dispersion models were to be
utilized. Consequently, it was intended
that data collection serve both pur-
poses. Ten samplers were deployed at
an array of seven to eight downwind
sites, on site near one to two source
ponds and at one upwind site. The sam-
plers were operated using the remote
control radio network switching system.
The dimensions of the array were ad-
justed according to the windspeed and
estimated atmospheric stability class
for adequate coverage of assumed
Gaussian dispersion.
Sampling efforts were performed on
July 9-11 and September 4-5, 1985,
based on favorable meterological fore-
casts. However, on both occasions the'
winds were quite variable during the
sampling periods. Good conditions
were experienced for only one sam-
pling period and that during the Sep-
tember sampling effort. None of the
data sets obtained from these runs was
adequate for Gaussian or receptor mod-
elling purposes. During both episodes,
canisters obtained from ASRL were
used for parallel sampling. These canis-
ters were used both at co-located and
independent sites. After sample
aliquots were taken for analysis by
GGC, the canisters were sent to RTP for
more detailed analyses by capillary GC.
Only limited opportunities occurred in
the fall and winter of 1985-86 for addi-
tional sampling at the site. Better condi-
tions were expected in May, 1986. Fur-
ther sampling with the array enlarged to
sixteen tubes was carried out from the
21st to the 25th of May. Three canisters
runs with five canisters each and eight
runs with 16 Tenax tubes were carried
out. For these runs an integrating
anemometer was used to give much
better definition of wind conditions.
Results and Discussion
GC analysis by thermal desorptioii
from Tenax tubes used to sample canifl
-------
ters of EPA reference mixtures agreed
well with the ASRL analyses and the
known concentrations. Blanks and repli-
cate samples showed low background
levels and good reproducibility. The use
of a photoionization detector permitted
highly specific sensitivity for the ben-
zene group despite the presence of
large amounts of other hydrocarbons
shown by ASRL's capillary GC-FID
analyses of the same canister samples.
There were at times large discrepancies
between the analyses of field samples
in canisters and the co-located Tenax
tubes. These discrepancies were proba-
bly due to sampling rather than analyti-
cal differences. At times however, large
amounts of other organic compounds
co-eluted with the benzene class com-
pounds and interfered with accurate
analysis comparisons. A further striking
finding from the canisters was the large
number of compounds and the high
concentrations present in many of the
samples.
The highly variable atmospheric con-
ditions which prevailed during the 1985
sampling appeared in retrospect to
have made the sampler number and
placement less than optimal. Sample lo-
cations were either inadequate in down-
wind and crosswind directions, or were
inadequate in total number to provide a
clear picture of plume behavior from the
waste ponds. Although there is a con-
sistency in sampling results, there are
insufficient data from these runs from
which to draw any but the most general
conclusions. Overall, the limited quan-
tity and somewhat inconsistent nature
of the 1985 sampling results do not sup-
port either receptor or Gaussian disper-
sion models. Additional sampler loca-
tions and a reconfiguration of the
downwind network were implemented
in 1986 to improve sampling results and
possibly provide an improved basis for
model analysis. The data quality ap-
peared to be better, but it became ap-
parent that there were present large
sources of the benzenoid compounds
(and many other substances) which
were obviously not the known disposal
ponds, based on the wind patterns.
These again made it impossible to apply
either receptor or Gaussian dispersion
models to the data in any meaningful
way.
Results from the Tenax cartridge
samples collected the sampling during
July and September of 1985 indicate a
reasonable internal consistency in data;
Ihe relative magnitudes of the various
chemical species in a sample were con-
sistent, and downwind and crosswind
variations followed similar patterns.
The data, however are not sufficiently
detailed to permit substantive conclu-
sions. The measured concentrations are
not statistically significant enough to
allow estimation of sensitivity and veri-
fication of models. Small sampling size
and irreproducibility of results will
probably lead to unacceptably low lev-
els of confidence in data interpretation,
such that the statistical significance of
the data set is insufficient to adequately
define a fit with either Gaussian or re-
ceptor model hypotheses. The data set
obtained in 1986 was larger, both in
samples per run and numbers of runs,
but nevertheless was still not usable. In
some of the runs there were one or
more maxima in concentration that
could not have originated from the dis-
posal ponds, based on wind patterns,
and successive runs did not always re-
semble one another even when winds
were similar. It seemed apparent that
either there were bursts of emissions of
short duration from the ponds, or there
were other intermittent strong sources
in the area. Although the disposal site
planned for study is in an area of oilfield
activity, it was not realized beforehand
that there were other local emission
sources of the magnitude indicated by
the data.
Recommendations
1. For attempts to improve receptor
models, it appears that waste dis-
posal sites are not desirable
sources, because of the wide vari-
ety of substances emitted, unless
high resolution analytical tech-
niques (e.g. capillary gas chro-
matography) are used to discrimi-
nate among the many pollutants.
Moreover, for this purpose it is
also important to have the study
source isolated from other sources
with similar emissions.
2. Mounting samplers on short poles
along the downwind centerline
will determine that the plume is in-
deed a ground-level source, and is
not behaving as an elevated
source because of ground heating
or some other external factor.
Sampler elevations of approxi-
mately 5 to 10 meters will permit a
more definitive vertical profile of
the plume and a better grasp on
atmospheric conditions through a
fit of vertical dispersion (sigma-z)
parameters.
3. A better definition of source emis-
sions will be obtained through im-
proved sampler locations and use
of tracer, or non-reactive, emis-
sions.
4. The remote control switching sys-
tem proved to function well and
make the operation of a network of
at least 16 battery operated sam-
plers over distances of about on'
kilometer quite practical.
-------
R. L. Brewer, R. J. Gordon, C.-C. Lin, and R. J, Countess are with Global
Geochemistry Corporation, Canoga Park, CA 91393-2194; T. J. Peters and
J. D. Walsh are with Engineering Science, Berkeley, CA 94710.
William Lonneman is the EPA Project Officer (see below).
The complete report, entitled "Ambient Air/Source Transport and
Transformation Relationships for Selected Hazardous Air Pollutants," (Order
No. PB 87-129 730/AS; Cost: $30.95, subject to change) will be available
only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
V
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
U.S.OFFIC!ALMA!i
U.S.POSIAGE
r 0 .2 Z =
Official Business
Penalty for Private Use $300
EPA/600/S3-86/063
0000329 PS
------- |