United States
Environmental Protection
Agency
Region 4
345 Courtland Street, NE
Atlanta GA 30308
EPA 904/9-80-055
January, 1981
tsrERA Collection and Analysis
Of Nonmethane Hydrocarbon
Transport Data
Louisville, Kentucky and
Nashville, Tennessee
Ozone Study
-------
DCN #81-240-016-01-07
COLLECTION AND ANALYSIS OF
NONMETHANE HYDROCARBON DATA
From Upwind Ozone Monitoring Sites for
Louisville, Kentucky and
Nashville, Tennessee
DRAFT FINAL REPORT
December 3, 1980
By:
Robert D. Cox
Kenneth W. Lee
Gary K. Tannahill
Hugh J. Williamson
Radian Corporation
Austin, Texas 78766
EPA Contract No. 68-02-3513
Project Officer
Douglas C. Cook
EPA Region IV
Air Programs Branch
Atlanta, Georgia 30365
US ENVIRONMENTAL PROTECTION AGENCY
EPA Region IV, Air Programs Branch
Atlanta, Georgia 30365
-------
DISCLAIMER
This report is issued by Che Environmental Protection Agency to report
technical data of interest to a limited number of readers. Copies are avail-
able free of charge to Federal employees, current contractors and grantees,
and nonprofit organizations—in limited quantities—from the Library Services
Office (MG-35), U.S. Environmental Protection Agency, Research Triangle Park,
North Carolina 27711; or, for a fee, from the National Technical Information
Service, 5285 Port Royal Road, Springfield, Virginia 22161.
This report was furnished to the Environmental Protection Agency by Radian
Corporation, 8501 Mo-Pac Blvd., Austin, Texas 78766, in partial fulfillment
of Contract No. 68-02-3513. The contents of this report are reproduced
herein as received from Radian Corporation. The opinions, findings, and
conclusions expressed are those of the author and not necessarily those of
the Environmental Protection Agency. Mention of company or product names
is not to be considered as an endorsement by the Environmental Protection
Agency.
i
-------
ABSTRACT
The purpose of this project was to collect nonmethane hydrocarbon trans-
port data to be used in Level III: EKMA-OZIPP analysis for Louisville,
Kentucky, and Nashville, Tennessee. Ambient air samples were collected
during September, 1980, at sites normally upwind of these cities.
Integrated samples were collected at each site from 6:00 am to 9:00 am
daily for two weeks. Samples were collected in Tediar1® bags then transferred
to stainless steel canisters for shipment and analysis. Cryogenic trapping
with liquid oxygen was used to concentrate the hydrocarbon species and to
separate methane and nonmethane hydrocarbons. Nonmethane hydrocarbons were
thermally desorbed in a gas c'nromatograph and quantitated with a flame ioni-
zation detector (FID). This method was thoroughly tested and data in support
of it are presented within the report.
Nonmethane hydrocarbon (NMHC) concentrations (6-9 am average) near the
upwind monitoring site for Louisville, Kentucky, ranged from 0.11 to 0.96
ppmv-C with a mean value of 0.30 ppmv-C and a median of 0.22 ppmv-C. NMHC
concentrations near the upwind monitoring site for Nashville, Tennessee,
ranged from 0.06 to 0.34 ppmv-C with a mean value of 0.12 ppmv-C and a
median of 0.09 ppmv-C.
This report was submitted in fulfillment of Work Assignment No. 1 of
Contract Number 68-02-3513 by Radian Corporation under the sponsorship of
the U. S. Environmental Protection Agency. This report covers a period
from August 15, 1980, to October 20, 1980.
-------
TABLE OF CONTENTS
Section Page
ABSTRACT ii
LIST OF TABLES V
LIST OF FIGURES vi
1. INTRODUCTION , 1
2. CONCLUSIONS AND RECOMMENDATIONS 2
3. ANALYTICAL SYSTEM 5
Description 5
Calibration 8
Hydrocarbon Response 8
Flow Rate . 10
Moisture Effect 13
Detection Limit 13
Reproducibility and Linearity 15
Typical Hydrocarbon Responses 15
4. SAMPLING SYSTEM 17
Description 17
Tedlar Bags 19
Stainless Steel Canisters 19
Pumping System 20
Sampling Procedure 21
5. SAMPLE COLLECTION 23
Louisville Site 23
Nashville Site 25
6. SAMPLE ANALYSIS 27
7. RESULTS 28
Nonmethane Hydrocarbon Data 28
Weather Data 31
iii
-------
TABLE OF CONTENTS (Cont.)
Section Page
8. QUALITY CONTROL/QUALITY ASSURANCE 34
Initial Project Preparation 34
Analytical Technique 34
System Blank 34
Detection Limit 35
Precision 35
Linearity 35
Sampling Methods 36
Blanks 39
Quality Control During Sample Analysis 42
Sample and Data Handling 47
Analysis of the Components of Variance 47
Weather and Data Correlations 52
REFERENCES 54
APPENDIX A 55
iv
-------
LIST OF TABLES
Number Page
1 Calibration Slopes for ftydrocarbon Species 11
2 Ambient Nonmethane Hydrocarbon Levels Found
Near Louisville, Kentucky 29
3 Ambient Nonmethane Hydrocarbon Levels Found
Near Nashville, Tennessee 30
4 Weather Conditions During Ambient Air Sampling
Near Louisville, Kentucky 32
5 Weather Conditions During Ambient Air Sampling
Near Nashville, Tennessee 33
6 QC/QA Sampling Matrix for Louisville, Kentucky ... 37
7 QC/QA Sampling Matrix for Nashville, Tennessee ... 38
8 NMHC Sampling System Blanks From Louisville
and Nashville 40
9 Quality Control Sample Data 41
10 Variance Components for Source Contributions .... 50
v
-------
LIST OF FIGURES
Number ' Page
1 Schematic of Analytical System 6
2 Response to Various n-Alkanes 9
3 Effect of Flow Rate on Trapping Efficiency of
Propane and Ethane 12
4 Effect of Water on Response to Propane and Methane . .14
5 Sample Chromatograms 16
6 Ambient Air Sampling Train 18
7 Louisville Sampling Site in Shepardville, KY.,
Approximate Layout 24
8 Nashville Sampling Site in Spring Hill, TN.,
Approximate Layout 26
9 Quality Control Chart for y-Intercept of
Calibrations 44
10 Quality Control Chart for Correlation Coefficient
of Calibrations 45
11 Quality Control Chart for Standard Error of
Calibrations 46
Vi
-------
SECTION 1
INTRODUCTION
The purpose of this project was to provide information on the transport
of nonmethane hydrocarbons as ozone precursors into the cities of Louisville,
Kentucky, and Nashville, Tennessee. This information, in combination with
other collected data, will be used to prepare revisions to the relevant
State Implementation Plans (SIP) using Level III Analysis (1). The approach
used to collect the data is compatible with the needs of the EKMA-OZIPP
Model. Also, the publication "Guidance for the Collection and Use of
Ambient Hydrocarbon Species Data in Development of Ozone Control Strategies','
(2) was used as a guide for developing the Work Plan and Quality Control
strategies for this project. Finally, the publication "Guidance For Collec-
tion of Ambient Non-Methane Organic Compound (NMOC) Data for Use in 1982
Ozone SIP Development, and Network Design and Siting Criteria for the NMOC
and NO^ Monitors", (3) was used as a guide for selecting sampling sites in
the areas specified by the Environmental Protection Agency (EPA).
-------
-------
SECTION 2
CONCLUSIONS AND RECOMMENDATIONS
Nonmethane hydrocarbon (NMHC) concentrations (6-9 am average) obtained
at a sampling site south of Louisville, Kentucky, between Setpember 16, 1980,
and September 30, 1980, ranged from 0.11 to 0.96 pprav-C. The average of
these 6-9 am values was 0.30 ppmv-C and the median 0.22 ppmv-C. NMHC
concentrations obtained at a sampling site south of Nashville, Tennessee,
between September 19, 1980, and October 2, 1980, ranged from 0.06 to
0.34 ppmv-C, with a mean of 0.12 ppmv-C and a median of 0.09 ppmv-C. The
minimum detection limit for NMHC was determined to be 0.025 ppmv-C;
precision of duplicate analyses averaged ±6 percent; and the relative
accuracy using NBS standards averaged ±8 percent.
Samples were collected near existing upwind ozone monitoring sites.
Although upwind sites for both cities conformed to EPA sampling guidelines
(3) for hydrocarbon collection, several hydrocarbon sources near the
Louisville site potentially could have impacted the NMHC data. However,
temporal plots of NMHC concentrations and wind direction for this site did
not show any correlations which would indicate local point-source contami-
nation .
Several recommendations can be made concerning the use of Tedlar bags
and stainless steel canisters for collection of nonmethane hydrocarbons.
When initially purchased, Tedlar bags require a minimum of five 3-hour
purges with ultrapure air before they can be used for ambient air sampling.
Also, Tedlar bags should be protected from exposure to light at all times.
Bags used in this sampling effort were enclosed in cardboard boxes. Tedlar
bags which were exposed to sunlight, heat or grossly contaminated samples
2
-------
continued to leach hydrocarbon contaminants for up to two weeks after
exposure.
It is also recommended that sampling systems which are to be used for
collection of ambient air should be designed to allow the pump to operate in
a moderate flow range. A previous EPA report (2) recommends the use of a
critical orifice to reduce pump (Metal Bellows MB-158) flows to rates
required for integrated sampling. Experimentation conducted during this
project has shown that reducing the pump output in this manner for more than
30 min. causes excessive heating of the pujnp which results in low-level
(0.1-0.5 ppmv-C) contamination of the sample. For this project the pump
output was split and two flow restrictors were used to regulate the sample
collection and total pump flow. This system provided very constant sampling
flow rates and did not produce excessive pump heating. It is also recom-
mended in the EPA report mentioned above (2) that ultrapure air should be
pumped through the entire sampling system before and after sample collection,
and then analyzed. In the present case it was found that although 30 min.
sampling blanks produced no contamination, sampling blanks collected for
3 hours did show contamination; presumably due to excessive pump heating.
Therefore, it is also recommended that as a quality control measure, blanks
should be collected by pumping ultrapure air through the entire sampling
system for a length of time equal to that of sample collection.
Since the Level III analysis requires only total nonmethane hydrocarbon
data, hydrocarbon speciation is not required. However, when this project
was initiated the only method available for measurement of total nonmethane
hydrocarbons was continuous NMHC analyzers, which are unreliable at concen-
trations below 1.0 ppmv-C. It was recommended in a previous EPA report (2)
that this type of data should be obtained chromatographically, using two
separate chromatographic columns. This is a very time consuming and expen-
sive method for acquiring the type of data required. A method was developed
specifically for the determination of total nonmethane hydrocarbon content
as part of this work. This method proved to be very reliable and the short
3
-------
analysis Cimes allowed multiple analyses of the samples collected for this
project. The method discussed in this report provided an accurate, cost-
effective means of acquiring NMHC data and is recommended for future appli-
cations of this type.
4
-------
SECTION 3
ANALYTICAL SYSTEM
The Level III Analysis: City-Specific EKMA Model requires nonmethane
hydrocarbon data to provide information on upwind ozone precursors. Because,
of the imprecision of continuous THC/CHi» analyzers at ambient nonmethane
hydrocarbon levels, the EPA has recommended that the NMHC data be acquired
by collecting grab samples and analyzing these chromatographically (2) .
Since hydrocarbon speciation was not required, the analytical system was
designed without a chromatographic column. This provided rapid analysis
times resulting in an accurate, cost-effective method for acquiring the
NMHC data.
Description
A schematic diagram of the analytical system is presented in Figure 1.
The chromatograph used was a Tracor 560 equipped with a flame ionization
detector. The signal from the chromatograph was processed with a Hewlett-
Packard® 3388 Integrating Computer. The system's plumbing was controlled
by a Valco® stainless steel 10-port valve which was mounted inside the
chromatograph oven. All flow lines were 1/8 in. o.d. stainless steel with
the exception of the line from the 10-port valve to the FID which was 1/16
in. o.d. stainless steel. Gas input to the system was controlled by two
3-way valves which allowed selection of gas standards, samples, or ultrapure
nitrogen purge gas. All flow through the system was by means of positive
pressure in the external gas cylinders. The sample trapping loop consisted
of 4 in. of 80/100 mesh glass beads packed in 1/8 in. o.d. stainless steel
tubing. Cryogenic trapping was achieved by cooling the trap in liquid
oxygen. Trapped hydrocarbon species were thermally desorbed with boiling
5
-------
0 • 30 psi
FLOWMETER HIGH PRECISION
GAUGE
GC
OVEN
SAMPLE
VENT
CLOSED
3-WAY
VALVE
VALCO
10 PORT
VALVE
NEEDLE
VALVE
ULTRAPURE
TO FID
SAMPLE
CANISTER
SAMPLE
LOOP
4'80-100 MESH
GLASS BEADS
3-WAY
VALVE
FLOW
CONTROLLER
PROPANE
STANDARO
ULTRAPURE
N2
SAMPLE LOAD
70-1966-1
Figure 1. Schematic of Analytical System.
6
-------
water. The volume of sample passed through the trap was measured by col-
lecting the gas in a 2.8 L reservoir and monitoring the pressure of the
reservoir with a high precision pressure gauge (Airco No. 60010). A flow
meter was situated between the 10-port valve and sample reservoir to monitor
flowrate through the loop. Sample flow was controlled with a needle valve
situated in the sample inlet line.
The analysis procedure was initiated with the 10-port valve in the
"analyze" position. In this position the ultrapure nitrogen carrier flowed
through the sample loop to the FID and sample flow was vented to the atmos-
phere. Prior to sample loading, the loop was cooled with liquid oxygen for
two minutes and the sample intake lines were purged. After two minutes the
10-port valve was switched to the "sample load" position and flow was routed
through the cryogenic trap and into the sample reservoir. When the desired
pressure was reached, the 10-port valve was switched back to "analyze" and
the computer program initiated. After a one minute delay period, the liquid
oxygen was removed and boiling water was placed on the trap. Runs were
allowed to proceed for five minutes. The purpose of the one minute delay
was to allow the FID time to stabilize from the pressure surge and to remove
any methane which may have collected on the trap.
Computer integration of peak areas was begun when the hot water was
placed on the trap. Upon termination of a run all peak areas were summed
and the final pressure was entered into the computer by the operator. The
total peak area was applied against the most recent calibration data by the
computer and a hydrocarbon value was calculated.
The 10-port valve and sample loop were situated in the GC oven to allow
rapid thermal cleaning. During use the oven was maintained at 60°C to pre-
vent condensation in the valve or tubing. When not in use, the oven was
maintained at 100°C with the valve in the "load" position and the sample
loop purged by ultrapure nitrogen at 15 mL/min. Occasionally, when samples
with high water content were analyzed, a small broad peak was observed to
7
-------
elute well after the hydrocarbon peaks. This peak was not reproducible and
appeared to be a material which collected in the system. This peak was not
integrated and when it was observed the system was purged for 15 minutes
with ultrapure nitrogen at 50 mL/min with the oven at 130°C. The oven was
then cooled and analyses continued.
Calibration
Calibration was achieved using a NBS traceable 0.094 ppmv propane
standard. Four point,calibration curves were prepared by analyzing 100 to
500 mL volumes of the standard. The data were then treated by at least
squares linear regression analysis and the slope, y-intercept, correlation
coefficient, and standard error were calculated and stored by the computer.
If the correlation coefficient was less than 0.99 the system was heated and
purged, then recalibrated. Calibrations were run before and after each set
of sample analyses.
Method Validation
The analytical system was tested to determine response to various carbon
compounds, efficiency of separation of methane, effect of flow rate through
the cryogenic trap, and effect of humidity. Parameters including detection
limit, reproducibility, and linearity were also evaluated.
Hydrocarbon Response
Response of the analytical system to different carbon compounds was
evaluated with standards of methane, ethane, propane, and a mixture of
C2-C6 n-alkanes. The system was calibrated with each of these by passing
volumes from 50 to 1000 mL of respective gas standards through the trap.
Calibrations for these compounds are shown in Figure 2. Linear calibrations
were obtained for up to one liter of propane or the C2-C6 n-alkane mixture.
A linear calibration was obtained for ethane at low analysis volumes but
8
-------
W 2000
200
400 600
Volume Sampled (mL)
1000
Figure 2. Response to Various n-Alkanes - Ethane
at 0.188 ppmv-C, Propane at 0.282 ppmv-C,
C2-C5 n-Alkane Mixture at 0.529 ppmv-C.
9
-------
some losses of this compound were observed when volumes above 400 mL were
used. This was probably not observed in the alkane mixture since ethane
comprised only 10 percent of the carbon content in this case. Methane gave
no response at volumes below 700 mL and only a slight response at 1000 mL.
The slopes of the hydrocarbon calibrations are presented in Table 1.
Also listed are the effective carbon number contributions, ECNC, for each
gas standard. These were calculated using the following equation:
Rx Cs
ECNC =
Rs Cx
where: Rx is the response (slope) to the standard which is propane in this
case, Rs is the response to the sample compound, and Cx and Cs are cancen-
trations of standard and sample, respectively. An equivalent FID response to
all carbon compounds will produce ECNCs of 1. In the present case, ECNC
values for the nonmethane hydrocarbons agreed to within four percent, which
justifies the use of a propane standard for quantitative evaluation of
total nonmethane hydrocarbon content.
Flow Rate
The effect of flow rate through the cryogenic trap on trapping effi-
ciency was evaluated using standards of ethane and propane, each at 0.094
ppmv.
Constant volumes of each gas were sampled at flow rates from 25 to 200
mL/min. Results of this study are presented in Figure 3. Different flow
rates in the range tested appeared to have no effect on trapping efficiency.
The coefficient of variation for responses obtained for ethane at various
flows was one percent and three percent for propane. Also, the average
response for propane was 1.6 times that of ethane (1.5 x theoretical), again
demonstrating the response to carbon content.
10
-------
TABLE 1. CALIBRATION SLOPES FOR HYDROCARBON SPECIES
Concentration
Compound (ppmv-C)
Methane 3.20
Ethane 0.188
Propane 0.282
C2 - C6 0.529
Slope
(area/L) ECNC1
90 0.004
1370 1.006
2040 1.0001
3670 0.960
1 = Effective Carbon Number Contribution using propane as reference
-------
1000
• Propane 0 094 ppm
^ Ethane 0.094 ppm
BOO
c
a
m
7 = 625± 20(3%)
600
-------
Moisure Effect
The effect of water on the analytical system was also studied. Water
could have several effects including clogging the cryogenic trap, increased
trapping of methane or decreased trapping of other hydrocarbons, and sup-
pression of FID response. Mixtures of gases were prepared by diluting a
0.94 ppm propane standard with ultrapure air, a percentage of which was
bubbled through the water. Water content of the gas mixtures was controlled
by proportionally varying the amount of air which was bubbled through the
water. Although moisture contents prepared in this matter are not exactly
analogous to relative humidity, any effects due to increasing moisture
content would be observed.
The results of this study are presented in Figure 4. Some variation
(CV = 10 percent) was observed due to the manner in which gas mixtures were
prepared, but the moisture content did not appear to have any effect on
propane trapping efficiency or FID response. Methane was included in the
gas mixture which was prepared at 60 percent humidity, and the response
obtained was equivalent to that for only propane; indicating that water did
not cause increased retention of methane. No significant reduction in flow
rate through the cryogenic trap was observed when samples containing high
moisture content were analyzed, demonstrating that clogging was not a
problem. In some cases of high moisture content a second peak was observed
to elute from the cryogenic trap approximately one minute after the hydro-
carbon elution zone. This was attributed to collection of polar material in
the trap and was not included in peak summation. It was concluded from these
studies that water had no significant effect on the analytical method and no
steps were included for the removal of water.
Detection Limit
The minimum detectable quantity (MDQ) of nonmethane hydrocarbon was
4 x lO-10 moles - C, which corresponded to approximately 6 ng of ethane or
13
-------
2000 -
1800 -
• Response to 0.094 ppm propane
also contained 16 ppm methane
«
cc
1600 -
1400 -
1200 -
in
c
a
o
o
<0
Q>
U 1000^
(A
c
o
a.
BOO -
600 -
400 -
200 -
y = 650 ±60 (10%)
2
10
20
30
40
50
60
70
80
go
100
% Humidity
Figure 4. Effect of Water on Response to Propane and Methane.
-------
propane. When using a 500 mL sample, the detection limit was 0.018 ppmv-C
(pL-C/L) or 0.006 ppmv-propane. Nonlinearity was less than 10 percent at
this level. The detection limit obtained was below the 0.05 ppmv-C level
quoted in the QC/OA document (4). The increase in sensitivity was beneficial
for accurate analysis of system blanks and ultrapure air.
Reproducibility and Linearity
Reproducibility on standard gas mixtures was generally within 3 percent
(CV) in the 0.1 to 1.0 ppmv-C range. Reproducibility on ambient samples
will be discussed with the data interpretation. Calibration was highly
linear providing correlation coefficients greater than 0.99. This will
also be discussed further in the Quality Control/Quality Assurance section.
Typical Hydrocarbon Responses
Typical response of the analytical system to hydrocarbon species are
illustrated in Figure 5. An initial noise spike was observed when liquid
oxygen was removed from the hydrocarbon trap. Peak integration was initiated
immediately following this spike. The propane standard produced a sharp,
rapid response (Figure 5-A). When a mixture of C1-C12 n-alkanes was anal-
yzed, a similar but more broad response was observed (Figure 5-D). Ambient
samples collected near Nashville and Louisville usually produced several
peaks (Figures 5-B and 5-C). With the present analytical system individual
peaks were not reproducible, although total peak areas were reproducible.
15
-------
A.
B.
C.
D.
8
ii
a>
E
II
a>
E
£
c
5
E
rim
E
i-
-------
SECTION 4
SAMPLING SYSTEM
The sampling system was set up as recommended in "Guidance for the
Collection and Use of Ambient Hydrocarbon Species Data in Development of
Ozone Control Strategies" (2), except for certain modifications which were
required during the course of this work. Sampling consisted of collection
of three-hour integrated air samples in Tedlar bags and transferring the
samples to stainless steel canisters for transport and analysis.
Description
A schematic diagram of the sampling system is presented in Figure 6.
A seven-micron stainless steel filter was used as the sampling probe to
remove particulate material. Rain water was prevented from entering the sys-
tem by a cone shaped aluminum shield which was formed around the filter.
The sampling probe was connected to the pump by 3 m of 1/4 in. o.d. stainless
steel tubing. Air flow through the system was effected with a Metal Bellows
model MB-151 pump. Since this pump produced an unrestricted flow of approxi-
mately 25 L/min and a flow of only 250-350 mL/min was required to collect
three-hour integrated samples, the output of the pump was split so that only
a fraction of the flow entered the collection bag. This was accomplished
by splitting the pump output with a stainless steel tee and two flow
restrictors. The flow restrictor between the tee and collection bag con-
sisted of a measured length of 1/16 in. o.d. x 0.01 in. i.d. stainless steel
tubing. A stainless steel fine metering valve (Nupro® no. SS-2MG) was used
as the second flow restrictor. Air flow through this restrictor was vented
to the atmosphere. The two restrictors were set so that the flow entering the
sampling bag was between 250 and 350 mL/min and a flow of 1.5 to 3.5 L/min
17
-------
S S. FILTER
1/4" S.S. TUBING
ALUMINUM
FOIL SHIELD
S.S NEEDLE
VALVE
1/4" S.S. TUBING
1/16" FLOW
RESTRICTOR
S.S. TEE JUNCTION
METAL BELLOWS
MB-151
PUMP
Nl VALVE
TEDLAR
BAG
CARDBOARD
BOX
70 1905 1
Figure 6. Ambient Air Sampling Train.
-------
was exhausted to the atmosphere through the fine metering valve. Although
a range of flow rates was used for the two sites, the flow was set at a con-
stant rate at the beginning of each sampling period and this did not vary
by more than 15 percent during the 3-hour period.
Samples were collected in 70 L Tedlar bags (30 in. x 30 in. x 0.002 in.)
which were fitted with Roberts 30A-C nickel plated valves. The valves could
be rotated to open or shut off flow to the bags. The bags were enclosed in
cardboard boxes during sampling to shield them from light. When sampling
was completed, the collected air was, pumped from the Tedlar bag into 2.8 L
stainless steel canisters, which were pressurized to approximately 30 psi.
The canisters used were obtained from D & S Instruments, Pullman, Washington,
and had been electropolished by the SUMMA® process.
Tedlar Bags
Tedlar bags were tested by filling them with ultrapure air and measuring
the total nonmethane hydrocarbon levels after three hours. A 3-hour value
of less than 0.05 ppmv-C was considered acceptable. It was found that when
initially purchased, the bags required a minimum of five purges with ultra-
pure air before the acceptable level was reached. It was also found that
these bags should be completely protected from exposure to sunlight and heat.
Tedlar bags which were exposed directly or indirectly (through a dark
colored garbage bag) to sunlight emitted high levels of nonmethane hydro-
carbons. This emission continued for several weeks after exposure even
though bags were stored in the dark. During this project, efforts were made
to always keep the Tedlar bags cool and in the dark.
Stainless Steel Canisters
Stainless steel canisters were cleaned by heating to 100°C for 15
minutes with a slow flow (20-50 mL/min) of ultrapure air passing through
them. Following this, the canisters were cooled, pressurized to 30 psi, and
19
-------
after several hours the air within was analyzed. A nonmethane hydrocarbon
level of less than 0.05 ppmv-C was considered acceptable for canisters. It
was found that canisters were much easier to clean and keep clean than the
bags. All canisters were pressurized and analyzed when initially obtained,
although no contamination was found when canisters were purchased from the
aforementioned manufacturer. The cleanup procedure for canisters was re-
peated after each use.
Pumping System
The pump was tested for contamination by pumping ultrapure air into a
precleaned Tedlar bag for three hours. Following this, air from the bag was
pressurized into a precleaned stainless steel canister and analyzed. This
technique actually measured the total contamination produced by the pump,
bag, canister, and sample transfer. A level of 0.1 ppmv-C was considered
acceptable as a three-hour blank value for the entire sampling system.
It was found that the pumping system can be the major contributor to
hydrocarbon contamination of the sample and that it also may be the most
difficult component of the sampling system to clean and keep clean. Ini-
tially, the pumping system was set up as recommended by EPA (2). It is
recommended in this document that a flow restrictor be used to reduce the
output of the sampling pump from 25 L/min to the required 300 mL/min. When
the sampling system was constructed in this manner, acceptable blank values
were obtained for 30 minutes of operation. However, when the system was
operated for 3 hours, the pump became hot and nonmethane hydrocarbon levels
from 0.1 to 0.5 ppmv-C were observed. Excessive heating of the pump is not
desirable since it could cause increased emission of hydrocarbon residues
in the pump and the warm air produced could cause increased leaching of con-
taminants from bag surfaces.
20
-------
It was found that the pumps produced the least heat when a flow between
1.5 and 10 L/min was maintained. This was accomplished in the sampling sys-
tem by using two flow restrictors and splitting the pump output. Flow
through the sampling restrictor was approximately 300 mL/min and flow through
the exhaust restrictor was approximately 2.5 L/min. This was most easily
accomplished using a needle valve as the exhaust restrictor. In addition to
the pump operating at a cooler temperature, only 10 percent of contaminants
produced by the pump passed into the Tedlar bag when the sampling system was
set up in this manner.
Sampling Procedure
Prior to initiation of sample collection the sampling system was purged
with ambient air for 30 minutes. This allowed the removal of material which
may have condensed in the sampling system while not in use and allowed the
pump and all transfer lines to reach an equilibrium temperature. During
this time the bag (or bags) to be used on that day were purged with ultrapure
or ambient air. The pump was disconnected momentarily to empty the bag.
After the pumping system and bag had been purged, the exhaust and sampling
flows were set at their respective levels and the Tedlar bag was placed in
a cardboard box. Sampling was initiated daily at 6:00 am and continued for
a 3-hour period.
When sample collection was completed, the bag was disconnected from the
sampling line and closed by means of the nickel valve. The pump was allowed
to continue running until the sampling and exhaust flows were measured.
Once this was accomplished the exhaust restrictor was opened completely and
the sampling line was purged of any water which had collected in it. Mois-
ture was observed in the needle valve used for the exhaust restrictor when
samples were collected during conditions of high humidity or precipitation.
Moisture was not observed in the 1/16 in. restrictor leading to the collec-
tion bag or in the collection valve. The Tedlar bag containing the three-
hour integrated sample was then connected to the pump input and the sample
21
-------
was pumped into an evacuated stainless steel canister. This canister was
evacuated and filled two more times to assure a representative sample. On
the final filling, the canister was pressurized to approximately 30 psi and
the sample number and pressure was recorded.
Normally, two canisters were purged and filled from the Tedlar bag
sample collected on a given day. As a quality assurance measure, samples
were collected on several days in duplicate Tedlar bags. This was accom-
plished by decreasing the length of sample restriction tubing to double the
sample flow and splitting the flow into two Tedlar bags. In this case only
one canister was filled from each bag.
When sample collection was completed and the stainless steel canisters
filled, the remainder of the sample was pumped from the Tedlar bag and
the bag flushed twice with ultrapure air. Filters were backflushed several
times with high pressure surges of ultrapure air, then with a slow flow of
ultrapure air for 10 minutes. Four filters were used at each site. All
sampling lines were then reconnected with the appropriate flow rates set
and ultrapure air was pumped through the sampling system for 20 minutes
after which the sampling input and outputs were sealed.
Blanks were collected by pumping ultrapure air through the entire
sampling system and into a Tedlar bag for 3 hours. The air in the Tedlar
bag was transferred to a single stainless steel cylinder for transport
and analys is.
Since the ultrapure air used contained detectable quantities of non-
methane hydrocarbons, samples of this were also collected in stainless
steel canisters and shipped to the laboratory for analysis.
22
-------
SECTION 5
SAMPLE COLLECTION
The criteria used for site selection were those recommended for NMHC
collection in EPA reports (2,3). EPA stipulated that the collection sites
should be two existing ambient air monitoring laboratories in Louisville and
Nashville. Within this restriction sites were chosen to collect samples
which were representative of the average air composition in that area.
Sample collection at the Louisville site was initiated on September
16, 1980, and completed on September 30, 1980. Sample collection at the
Nashville site was initiated on September 19, 1980, and completed on
October 2, 1980. All samples and accompanying data sheets were shipped
directly to Radian's Austin laboratory via air express.
Louisville Site
The ambient air sampling site for Louisville was near a state-owned
monitoring station located approximately 20 km south of the city. The
exact location of the sampling probe was chosen to be approximately equi-
distant from surrounding gas stations, a school bus parking lot, and a
highway. A drawing of the site is presented in Figure 7. The site was
also chosen upwind of the monitoring station (by wind records of the previous
year) to avoid possible contamination from ozone monitors in the trailer.
The sampling probe was located 4.5 meters above ground level and 3 meters
above the support platform.
23
-------
Gas Stations
School
Bus
Parking
lot
300m
Sampling
Site
250m
Monitoring
Trailer
500m
School
Gas Stations
State Highway 44
Figure 7. Louisville Sampling Site in Shepardsville, KY - Approximate Layout.
24
-------
Approximately 15 school buses were situated 300 meters east of the samp-
ling site. These were started at 7:00 am five mornings a week and allowed to
warm up for 20 minutes. Two gas stations were located 600 meters northwest of
the sampling site and 3 more were located 500 meters south of the sampling site.
No tank truck off-loading was observed during the 6:00 am - 9:00 am sampling
periods. An interstate highway (1-65) ran nctth to south about 250 meters west
of the site. Travel on the highway from 6:00 am - 9:00 am was approximately
300 cars. A school was located southeast of the site.
Weather data during sampling were obtained from the Louisville airport
which was about 20 km north of the site. In addition, weather data collected
at the sampling site were obtained from the Division of Air Pollution Control
of the Kentucky Department for Natural Resources and Environmental Protection.
Nashville Site
The ambient air monitoring site for Nashville was on the University of
Tennessee Agricultural Testing Facility in Spring Hill, Tennessee; approxi-
mately 60 km south of Nashville. The state monitoring station was in a fairly
secluded area on the eastern edge of the testing facility. A diagram of the
area is presented in Figure 8. One possible source of contamination at this
site was the Bull Testing Labs located 50 meters east of the monitoring station.
Samples at this site were collected approximately 5.5 meters above ground level
and 1.5 meters above the roof of the monitoring trailer. The exhaust from the
ozone monitor was vented underneath the trailer. A 10 meter length of poly-
ethlyne tubing was connected to this and was always positioned so that the
exhaust was at ground level and downwind of the sampling probe.
Wind direction data were obtained from monitoring equipment operating
in the trailer. All other weather data were obtained from the U.S. Weather
Service at the Nashville Airport, which was about 70 km northeast of the
sampling site.
25
-------
4-Lane Highway 31
Variable Ozone Monitor
Exausl Line
cr>
U''
Monitoring
1
Trailer
• —
Residence
Storage
Barn
Bull Testing
Labs
Sampling
Probe
Access Rd.
Unused
Structures
Figure 8. Nashville Sampling Site in Spring Hill TN. Approximate Layout.
Figure 8. Nashville Sampling Site in Spring Hill, TN " Approximate Layout,
-------
SECTION 6
SAMPLE ANALYSIS
When shipments were received in the laboratory, the data sheets were
collected and samples logged in. All samples were given a code number
which was the only identification provided to the analyst. Prior to
analysis the pressure of each canister was checked and recorded.
Each canister was analyzed in duplicate or triplicate providing four
to six data measurements per day. Quantitation was based on'a linear
calibration prepared with a 0.094 ppmv NBS traceable propane standard
prior to analysis. Sample volumes were kept at or below 400 mL to assure
complete trapping of ethane and ethylene.
27
-------
SECTION 7
RESULTS
Data were obtained over a-15 day period for Louisville and over a 14
day period for Nashville. All NMHC and weather data will be reported in
this section. By means of a thorough QC/QA program, it was possible to
assess individual error contributions from various components of the
sampling and analytical methods arid calculate confidence limits for
the NMHC data. All statistical calculations will be discussed in the
QC/QA section of this report.
Nonmethane Hydrocarbon Data
The nonmethane hydrocarbon concentrations found in ambient air samples
collected near Louisville, Kentucky, are presented in Table 2 and those
concentrations found in samples collected near Nashville, Tennessee, are
presented in Table 3. NMHC concentrations are reported as ppmv-C (yL-C/L)
on a dry air basis. One sample at each site was collected daily (in one
or two Tedlar bags) and transferred to two stainless steel canisters. The
average NMHC value and analytical variation for each canister are
presented in Tables 2 and 3 as well as the daily averages for both canisters.
The daily NMHC values were corrected for sampling system contributions at
each site. These data, along with confidence limits, are also presented
in these tables.
Variability of the duplicate canister samples which were collected
daily at each site is illustrated graphically in Figures A-l and A-2 of
Appendix A.
28
-------
Date
TABLE 2. AMBIENT NONMETHANE HYDROCARBON LEVELS FOUND NEAR
LOUISVILLE, KENTUCKY
Sample
TyPe'
NMHC
Concentration2
(ppmv-C)
Daily Average3 Corrected Value"*
(ppmv-C) (ppmv-C)
Confidence
Limits5
(ppmv-C)
9/16
9/17
9/18
9/19
9/20
9/21
9/22
9/23
9/24
9/25
9/26
9/27
9/28
9/29
9/30
DC
DC
DC
DC
DC
DC
DC
DC
DB
DC
DB
DC
DB
DC
DC
0.380±0.010
0.461±0.014
0.211±0.010
0.185+0.004
0.460±0.017
0.446±0.017
0.308±0.008
0.328+0.026
0.416+0.003
0.416+0.026
0.155±0.013
0.160±0.023
0.189+0.020
0.292+0.035
0.200+0.006
0.161±0.014
0.284±0.024
0.264+0.006
0.262+0.001
0.248±0.005
0.205±0.007
0.204±0.001
1.003±0.024
1.030+0.088
0.422±0.022
0.305±0.003
0.634+0.027
0.654+0.004
0.189±0.019
0.200+0.022
0.420+0.057
0.198+0.018
0.453±0.009
0.318+0.014
0.416±0.000
0.158±0.004
0.240±0.073
0.180±0.028
0.274±0.014
0.255±0.010
0.204±0.001
1.016+0.019
0.364+0.083
0.644+0.014
0.194±0.008
0.368±0.045
0.146±0.028
0.404+0.045
0.266±0.045
0.364±0.045
0.106+0.029
0.188±0.044
0.128+0.028
0.222±0.039
0.203+0.045
0.152±0.039
0.964+0.045
0.312±0.039
0.592±0.045
0.142±0.027
0.278-0.458
0.091-0.201
0.314-0.494
0.176-0.356
0.274-0.454
0.050-0.162
0.099-0.277
0.072-0.184
0.144-0.300
0.113-0.293
0.074-0.230
0.874-1.054
0.234-0.390
0.502-0.682
0.087-0.197
1 DC - duplicate canister sample, DB - duplicate bag sample.
2 Average nonnethane hydrocarbon value in ppmv-C (by volume) obtained for
each canister ± analytical variation (standard deviation).
3 Daily average for each set of duplicate canisters ± standard deviation.
u Daily average minus sampling system blank value ± standard deviation
(see Sample and Data Handling in Section 8).
5 Corrected NMHC value ± 2x standard deviation (95% confidence limits).
29
-------
TABLE 3. AMBIENT NONMETHANE HYDROCARBON LEVELS FOUND
NEAR NASHVILLE, TENNESSEE
Date
9/19
9/20
9/21
9/22
9/23
9/24
9/25
9/26
9/27
9/28
9/29
9/30
10/1
10/2
NMHC
Sample Concentration2
Type
DC
DC
DC
DB
DC
DC
DC
DC
DB
DC
DC
DC
DC
DC
(ppmv-C)
Daily Average
(pnmv-C)
Conf idence
5
Corrected Value Limits
(ppmv-C) (ppmv-C)
0.316±0.009
0.399±0.OA 7
0.118+0.003
0.120+0.000
0.133±0.015
0.145+0.001
0.128+0.011
0.175+0.027
0.094±0.018
0.118+0.002
0.191±0.004
0.180+0.021
0.153+0.010
0.167+0.020
0.090±0.006
0.090±0.002
0.145±0.017
0.130±0.000
0.038±0.001
0.094+0.003
0.106±0.023
0.116+0.011
0.083+0.009
0.08210.002
0.090+0.012
0.098+0.006
0.110+0.018
0.113+0.013
0.358±0.059
0.119+0.001
0.139+0.008
0.152+0.033
0.106+0.017
0.186±0.007
0.160±0.010
0.090+0.000
0.138+0.010
0.088±0.008
0.111+0.007
0.082±0.001
0.094+0.006
0.112+0.002
0.336±0.046
0.097±0.029
0.117+0.028
0.130±0.024
0.084+0.028
0.164±0.029
0.138±0.028
0.068±0.029
0.116±0.024
0.066+0.029
0.089±0.028
0.060±0.029
0.072+0.029
0.090i0.028
0.245-0.427
0.040-0.154
0.060-0.174
0.083-0.177
0.027-0.141
0.107-0.221
0.082-0.194
0.011-0.125
0.068-0.164
0.009-0.123
0.032-0.146
0.003-0.117
0.015-0.129
0.336-0.146
1 DC - duplicate canister sample, DB - duplicate bag sample.
2 Average nonmethane hydrocarbon value in pprav-C (by volume) obtained for
each canister ± analytical variation (standard deviation).
1 Daily average for each set of duplicate canisters ± standard deviation.
* Daily average minus sampling system blank value ± standard deviation
(see Sample and Data Handling in Section 8).
5 Corrected KMHC value ± 2x standard deviation (95% confidence limits).
30
-------
Weather Data
The weather conditions existing during sample collection in Louisville
are presented in Table 4 and those existing during sample collection in
Nashville are presented in Table 5. With the exception of precipitation,
all other data for the Louisville site'were obtained from the National
Weather Service at the Louisville Airport. For the Nashville site, wind
direction and precipitation data were collected on site and all other data
were obtained from the National Weather Service at the Nashville Airport.
NMHC values for each day are included in these tables for reference.
Wind direction and wind velocity data were collected from the Louisville
sampling site by the Division of Air Pollution Control of the Kentucky
Department for Natural Resources and Environmental Protection. These data
were obtained from this agency and are presented in Appendix A. The wind
direction data obtained from the Kentucky Division of Air Pollution Control
(KDAPC) were in good agreement with those obtained from the National Weather
Service at the Louisville Airport, however, wind speed data obtained from the
KDAPC were always less than those obtained from the National Weather Service.
These data were used to determine if high NMHC values obtained on certain
days were a function of local contributions. Plots of NMHC as a function
of wind direction and speed are presented in Appendix A. From the plots
for Louisville, KY it appears that the high NMHC values occurred when the
wind direction was from the north-northeast (9/27, 9/29, 9/18, 9/20). This
tends to implicate the school bus lot as a source of contamination. How-
ever, two of these sampling dates (9/27/9/20) fell on weekends when the
buses were not in service. Since the sampling site is situated south of the
city of Louisville, KY, high NMHC values on days with northerly winds could
be a result of city emissions.
31
-------
TABLE 4. WEATHER CONDITIONS DURING AMBIENT AIR SAMPLING NEAR LOUISVILLE, KENTUCKY
Corrected
Date
Temperature
(°F)
Pressure
(in Hg)
Relative
Humidity
(%)
Wind
Speed
(mph)
Wind
Direction
Precipitation1
NMHC
Consent ration
(ppmv-C)
9/16
60
29.90
92
5
E
0.37
9/17
66
29.82
90
7
SW
R
0.15
9/18
58
29.60
92
6
NE
F
0.40
9/19
58
30.22
94
5
N
F
0.27
9/20
71
30.10
82
7
S
0.36
9/21
78
29.98
80
7
S
0.11
9/22
78
29.96
82
7
s
R
0.19
9/23
65
30.01
86
12
N
0.13
9/24
60
30.04
86
7
NE
0.22
9/25
63
30.00
95
3
S
F
0. 20
9/26
54
30.32
72
8
NW
0.15
9/27
51
30.35
78
6
N
0.96
9/28
55
30.20
83
5
NE
0.31
9/29
58
30.10
79
4
NE
0.59
9/30
65
29.97
82
6
E
0.14
1 R -
rain; F - fog
or mist
-------
TABLE 5. WEATHER CONDITIONS DURING AMBIENT AIR SAMPLING NEAR NASHVILLE, TENNESSEE
Corrected
Date
Temperature
(°F)
Pressure
(in Hg)
Relative
Humidity
(%)
Wind
Speed
(mph)
Wind
Direction
Precipitation1
NMHC
Concontrat" ion
(ppmv-C)
9/19
59
29.53
—
2
NW
0.34
9/20
73
—
—
2
S
R
0.10
9/21
72
—
—
3
SE
F
0.12
9/22
77
30.01
82
3
SE
0.13
9/23
70
29.38
97
8 •
NE
0.08
9/2 4
67
29.37
97
3
N
F
0.16
9/25
68
29.35
100
5
NW
F
0.14
9/26
57
29.64
78
8
N
0.07
9/27
53
30.30
83
6
N
0.12
9/28
58
30.18
87
3
E
R
0.07
9/29
63
30.05
90
7
NE
R
0.09
9/30
62
29.93
97
5
NE
R
0.06
10/1
59
29.93
100
4
NW
F
0.07
10/2
67
29.87
84
10
NW
F
0.09
1 R - rain; F - fog or mist
-------
SECTION 8
QUALITY CONTROL/QUALITY ASSURANCE
The QC/QA program was designed to ensure the reliability of NMHC
data collected for this project (4). Discussion of this program will be
divided into those aspects dealing with initial project preparation, sample
collection, sample analysis, and data handling.
Initial Project Preparation
Prior to initiation of data collection, all analytical and sampling
techniques were evaluated to assure accuracy and reliability of the data.
Analytical Technique
The following aspects of the analytical technique were specified in
the QC/QA document and will be discussed here.
System Blank
A system blank was run daily prior to sample analysis. The results
of this analysis was used as the zero hydrocarbon level in the calibration
for that day. A 500 mL volume of ultrapure nitrogen was passed through
all carrier lines and analyzed as a sample. If any response above the
detection limit (0.02 ppmv-C) was observed, the system was heated to
130°C and purged, Ultrapure nitrogen was used instead of ultrapure air
(as recommended in the QC/QA document) since the UP air contained detectable
quantities of nonmethane hydrocarbons which hinderd recognition of the
source of low levels of NMHC contamination.
34
-------
Detection Limit
The lowest quantity of nonmethane carbon which could be reliably
detected was A x 10 10 moles-C. This was determined using a mixture of
ethane in air, since the trapping efficiency for this compound was less than
that for other hydrocarbons of interest. The detection limit was defined
as that quantity of ethane which produced an area count over 50, which was
the level set for noise rejection. This corresponds to a detection limit
of 0.018 ppmv-C when using a 500 mL sample. Since it was demonstrated
that ethane losses occur above 400 mL, the detection limit for this sample
size (0.025 ppmv-C) should be quoted for ambient samples.
Precision
Reproducibility on standard gas samples of ethane or propane was
3 percent or better in the concentration range from 0.1 to 1.0 ppmv-C.
Reproducibility of duplicate ambient samples ranged from 0 to 21 percent (CV)
with a mean value of 6 percent. Using variance component analysis (see
Sample and Data Handling in this section) the mean analytical standard
deviation for ambient samples below 0.2 ppmv-C was 0.017 ppmv-C (<22 percent)
and that for ambient samples above 0.2 ppmv-C was 0.026 ppmv-C (<13 percent).
Precision was much better than the 25 percent limit (at 0.4 ppmv-C) which
was initially stated in the QC/QA document.
Linearity
The analytical system was calibrated before and after each set of sample
analyses. Calibrations were prepared in the range of 0.03 to 0.3 ppmv-C
assuming a 500 mL sample size. Fifty-nine calibrations, each consisting of
at least four points, were prepared during the course of this study. Of
these, only two gave a correlation coefficient less than 0.99. The
analytical system was purged and recalibrated on these occasions. The
average correlation coefficient for all calibrations was 0.997.
35
-------
Sampling Methods
Two types of sample containers were used in this project: stainless
steel canisters and Tedlar bags. These containers were cleaned according
to EPA recommended procedures (2). Each container was tested by filling
it with ultrapure air which was analyzed after three hours. All containers
tested below 0.05 ppmv-C before being used in the field.
A variety of sample types were collected to comply with the Quality
Control/Quality Assurance plan. Ambient samples were collected either in
a single Tedlar bag and transferred to two stainless steel canisters, or
were collected in two Tedlar bags, each being transferred to a single
canister. In addition, sampling blanks were collected to ascertain the
amount of contamination contributed by the sampling system, and the
concentration of nonmethane hydrocarbons in the ultrapure air used.
Quality control samples were prepared in the laboratory and shipped to
the field where the pressure was checked, then returned to the laboratory
for analysis.
An inventory of QC/QA samples collected in Louisville is presented
in Table 6 and for Nashville in Table 7. Five types of QC/QA samples
were used in order to determine error contributions from each component
of the sampling technique. All canisters were analyzed in duplicate or
triplicate instead of the initial 50 percent frequency for duplicate
analyses.
During the sampling procedure, air flow through the sampling system and
stainless steel canister pressure were monitored. Sampling flows did not
vary by more than 15 percent during any 3-hour sampling period. Also, no
significant losses of samples during shipment were observed.
36
-------
TABLE 6. QC/QA SAMPLING MATRIX FOR LOUISVILLE, KENTUCKY
Sample Type
Duplicate Duplicate Replicate Analyses Sampling System Ultrapure Air Quality Control
Canisters Bags (per day) Blank Blank Standards
9/10 1
9/11 1
9/12 1
9/16 2* 4 1
9/17 2 5
9/18 2 4
9/19 2 4 ° 1
9/20 2 4
9/21 2 4
9/22 2 6 1
9/23 2 4
9/24 2 4
9/25 2 4
9/26 2 4
9/27 2 4
9/28 2 4 2
9/29 2 4 1
9/30 2 6
*Number of S.S. canisters used to collect sample
-------
TABLE 7. QC/QA SAMPLING MATRIX FOR NASHVILLE, TENNESSEE
Sample Type
Date
Duplicate Duplicate Replicate Analyses Sampling System Ultrapure Air Quality Control
Canisters Bags (per day) Blank Blank Standards
9/19
2*
4
9/20
2
4
9/21
2
5
9/22
2
5
1
9/23
2
5
9/24
2
4
1
9/25
2
6
9/26
2
4
1
9/27
2
4
9/28
2
4
9/29
2
5
9/30
2
4
1
10/1
2
4
1
1
1
10/2
2
6
1
*Number of S.S. canisters used to collect sample.
-------
Blanks
It was initally stated in the QC/QA document that the entire sampling
system, including containers, should not contribute more than 0.1 ppmv-C
NMHC contamination to the samples. This was determined by collecting and
analyzing 3-hour sampling blanks with the same procedure which was used for
ambient air collection. The NMHC concentrations in sampling system blanks
which were collected in Louisville and Nashville are presented in Table 8.
Also given in this table are NMHC concentrations for- the ultrapure air used
in collecting blanks. Since there was an initial concentration of NMHC
present in the ultrapure air this value must be subtracted from the NMHC
blanks for the sampling system. Of seven blanks collected during this
project, only one produced an NMHC concentration above the QC limit;
although when the UP-air concentration was subtracted from this value, it
also fell within the limit.
e
The average blank values for Louisville and Nashville were tested
and found to be statistically different. Therefore, respective blank
values were applied to samples collected at each site.
QC Samples
Data for Quality Control samples which were sent into the field
and returned to the laboratory for analysis are presented in Table 9.
All QC samples were prepared with NBS traceable propane. Absolute errors
ranged from -0.029 to +0.021 ppmv-C while relative errors ranged from
0.5 percent to 16 percent. The amount of time expired between preparation
and analysis of a sample did not appear to have a significant effect on
sample loss or gain.
39
-------
TABLE 8. NMHC SAMPLINC SYSTEM BLANKS FROM LOUISVILLE AND NASHVILLE
NMHC
Average Site
Net Site
Sample
Concentration2
Concentration
Concentratic
Date
Site
Type1
(ppmv-C)
(ppmv-C)
(ppmv-C)
9/16
L
SB
0.094+0.014
9/19
L
SB
0.128±0.006
9/22
L
SB
0.088±0.010
9/29
L
SB
0.062±0.006
0.09310.027
0.05210.014
9/10
L
UAB
<0.050
9/24
N
SB
0.075+0.016
9/26
N
SB
0.049±0.001
10/1
N
SB
0.064±0.003
0.06310.013
0.02210.015
9/22
N
UAB
0.04910.009
10/1
N
UAB
0.03310.009
0.04110.Oil3
1 SB - Sampling System Blank (3-hr); UAB - Ultrapure Air Blank
2
Average nonmethane hydrocarbon concentration + analytical deviation
(standard deviation).
3 This value used for UHP air at botli sites
-------
TABLE
9. QUALITY CONTROL SAMPLE DATA
NMHC
NMHC
Concentration
Concentration
Absolute
Relative
Date1
Prepared
Observed
Error
Error
Delay 2
Site (ppmv-C)
(ppmv-C)
(ppmv-C)
(%)
(days)
9/11
L 0.244
0.258±0.033
+0.014
5.7
4
9/12
L 0.181
0.152±0.022
-0.029
16.0
4
9/28
L 0.213
0.218±0.006
+0.005
2.3
6
9/28
L 0.215
0.214±0.008
-0.001
0.5
7
9/30
N 0.207
0.221±0.002
+0.014
6.8
6
10/1
N 0.116
0.132±0.012
+0.016
13.8
4
10/2
N 0.141
0.162+0.001
+0.021
14.9
4
1 Date of
sample pressure check
in field
o
Number
of days between sample
preparation and
analysis
-------
Quality Control During Sample Analysis
The analytical technique was calibrated before and after each set of
sample analyses. Each set of calibration points was treated by a least
squares linear regression analysis and the slope, y-intercept, correlation
coefficient and standard error were calculated. Rejection limits were
set for each of these parameters and the analytical instrumentation was
checked whenever a parameter fell outside its rejection limit. In
addition, quality control charts were kept to monitor day to day
variation of the analytical technique. These charts, covering a period
from August 22 through October 8, 1980, are presented in Figures 9 through
12. Although sample analysis did not begin until September 17, 1980,
calibrations were run previous to this when determining sampling system
blanks.
Variation in the calibration slope during the course of.this study
is shown in Figure 9. During this program, the FID was cleaned and
realigned prior to analysis of samples. This resulted in increased sensi-
tivity and a change in the calibration slope. Although this had no effect
on the sample data, it did require a new set of units for the quality control
chart. After this realignment, the mean value for the slope was 7260
(counts/ppmv-C) with a standard deviation of 575 (8 percent). This type of
day-to-day variation is not unusual for a flame ionization detector.
Daily variation in the calibration y-intercept is shown in Figure 10.
After the FID realignment, the y-intercept shifted negative. The mean
y-intercept during the entire study was -12 with a standard deviation of 46.
The correlation coefficient was used as a measure of system linearity
in this study. The daily variation of this parameter is shown in Figure
11. The mean value for all correlation coefficient data was 0.997 with
a standard deviation of ±0.003.
42
-------
2500
2400
2300
2200
2100
2000
1900
1800
1700
1600
1500
^ AM Calibration
0 PM Calibration
Calibration
6/22 8/27 8/28 8/29 9/2 9/3 9/4 9/5 9/3 9/9 9/10 9/11 *
Date
9/12 9/13 9/15 9/16 9/17 9/18 9/19 9/20 9/22 9/23 9/24 9/25 9/26 9/27 9/29 10/1 10/2 10/3 10/8
* :
• •
~
4 ~
. • ~
~
~
-8000-
7800
7600
7400
7200
¦ 7000"
6800
6600
6400
6200
-6000-
A
• A
*_
A
*FID Cleaned
Figure 9. Quality Control Chart for Slope of Calibrations.
-------
AM Calibiallon
0 PM Calibration
Calibration Date
8/22 8/27 8/28 8/29 9/2 9/3 9/4 9/5 9/3 9/9 9/10 9/11 9/12 9/13 9/15 9/16 9/17 9/18 9/19 9/20 9/229/23 9/24 9/25 9/26 9/279/2910/1 10/2 10/3 10/8
~
~ •
~ ~
~ ~
A ( ^
CO
CM
9
o
00
• A
* FID Cleaned
Figure 10, Quality Control Chart for y-Intercept of Calibrations.
-------
AM Calibration
0 PM Calibration
Calibration
8/22 6/27 8/28 8/29 912 9/3 9/4 9/5 9/3 9f9 9/10 9/11*
0 990
0.991
0.992
m 0.993
0.994
0.995
0.996
0.997
0.998
0.999
1.000
Date
9/12 9/13 9/15 9/16 9/17 9/18^9/19 9/20 9/22 9/23 9/24 9/25 9/26 9/27 9/29 10/1 10/2 10/3 10/8
—•
••
_A_
A A
• • A
A
m
i i
d>
«?
*FID Cleaned
Figure 11. Quality Control Chart for Correlation Coefficient of Calibrations.
-------
The standard error was used to provide a measure of the scatter of
data points around the best-fit line for the calibration. The quality cont-
rol chart for this parameter is presented in Figure 12 . The mean standard
error was 57 (±28) . The mean relative standard error (obtained by dividing
the mean standard error by the mean slope and expressed as a percent) was
0.3 percent.
Sample and Data Handling
All pertinent data concerning sample collection and weather conditions
were recorded in permanent notebooks at the sampling sites. These data
were also recorded on individual sample log sheets which were shipped along
with the samples. When samples were received in the laboratory, the log
sheets were collected" and each sample was assigned a code number. To avoid
experimental bias, the analyst was only provided the sample code number.
Protocol for this type of analysis dictates that the time between
sample collection and analysis be seven days or less. In this project
every sample was analyzed within three days of collection.
Analysis of the Components of Variance
The NMHC data which were collected are subject to various sources of
random variation, including the following:
1) true day-to-day variation in the hydrocarbon levels,
2) variation in the hydrocarbon concentration between Tedlar bags,
3) variance among canisters, and
4) random analytical errors.
Additionally, for each site, a term was subtracted from all hydrocarbon
data to account for the system response to ultrapure air (see Section 8.2).
This step was performed to remove a possible constant additive bias in the
46
-------
20C
180
160
140
120
100
ao
60
40
20
0
A AM Calibration
# PM Calibration
Calibration
6/228727 8/28 6/29 9/2 9/3 9/4 9/5 9/8 9/9 9/10 9/1 f
Date
9/12 9/13 9/15 9/16 9/17 9/169/19 9/20 9/22 9/23 9/24 9/25 9/26 9/27 9/29 10/1 10/2 10/3 10/6
~
4 4 #
~
A
A A
A *
• • ¦
A
A
• ~
~ •
*FID Cleaned
Figure 12.
Quality Control Chart for Standard Error of Calibrations.
-------
measurements. The correction terms were derived from finite sets of
measurements of blank samples and the NMHC concentrations in the ultrapure
air samples and; thus, are subject to random errors. The error in the
blank correction term, then-, is a fifth source of variance.
It was assumed that a given measured NMHC concentration C can be
expressed as follows:
C = u + e + e + e + e + e
HC D B C A P
where
y = true, long-term average NMHC concentration,
HC
e^ = true variation from the average for this particular sampling day,
e = random error associated with all samples drawn from this particular
B
bag,
e^ = random error associated with this canister,
e^ = analytical error, and
ep = random error associated with the blank correction term.
Thus, an additive random term is associated with each source of vari-
ation. It was desired to quantify the variance for each separate error
source. The usual equation used to compute the variance -42 of a set of
n points is:
n
2 _ 2 (x. - x)2
4 ~ 1=1 /C
n - 1
where X is the sample mean. This equation is not adequate to compute the
variances of e , e and e , however, as is illustrated by the following
B C, A
examples.
Suppose that n samples were collected on a given day from n separate
bags. The variance among the n concentrations, then, would not be the
variance of e^, the random error associated with the bags. Differences
among the n measurements would also exist due to possible differential
48
-------
effects of the n cannisters used and due to the n random analytical measure-
ments errors involved. Similarly, concentration measurements of samples in
different cannisters from the same bag differ due not only to random canister-
to-canister effects (term e^,) , but also due to analytical errors.
Thus, isolating the sources of variance cannot be achieved by the
straightforward calculation of variances by the equation :
n
¦6* = Z (x. - x)2
-<=1
n - 1
A statistical technique called variance component analysis (see
References 5 and 6), however, can be used to transform estimates of combi-
nations of variances into estimates of variances of individual sources of
variation. This technique was used to quantify the error variances due
to bag, canister, and analytical effects (e , e , and e ). The variance
D L> A
of the error ep due to the blank correction term was computed separately
from the set of blank, measurements.
It was suspected that the variances would vary with concentration, and
so separate variance component analyses were performed for concentrations
less than 0.2 and greater than 0.2 ppmv-C. The size of the data base did
not justify further subdivision. The same blank correction term was used
for all data for a given site, and so the variance of this term was not
separated by concentration.
The variances and standard deviations (the square root of the variances)
for the different sources of random error are presented in Table 10 . The
largest variances were associated with the canisters, followed by the bag
variances, followed by the analytical variances. The random errors of the
blank corrections have smaller variances than any of the above. All of the
error components have standard deviations less than or equal to 0.0620 ppmv-C.
49
-------
TABLE 10. VARIANCE COMPONENTS FOR SOURCE CONTRIBUTIONS
Source
Site
NMHC
Cone.
(ppmv-C)
Variance
Component
Standard
Deviation
(ppmv-C)
Analytical
N,
L
<0.2
0.000291
0.0171
Analytical
N,
L
>0.2
0.000698
0.0264
Tedlar Bags
N,
L
<0.2
0.000495
0.0222
Tedlar Bags
N,
L
>0.2
0.00104
0.0322
Canisters
N,
L
<0.2
0.00384
0.0620
Canisters
N,
L
>0.2
0.00125
0.0354
Blank
N
All
0.00023
0.015
Blank
L
All
0.00019
0.014
50
-------
The variance components presented in Table 10 were used to compute
the standard errors of the average of the NMHC measurements for each day
for each site. The standard error 4 for a given day was calculated as
e
follows:
-6 =
e
4b
+ '1
+ *C
+ -6 2
h
n
n
2
p
SB
A
Where
= variance component for bags,
tt = number of bags used on that day (one or two)
OD
4^ = analytical variance component.,
n^ = total number of analytical determinations performed for both air
samples (four, five or six)
4 2 = variance component for canisters (two canisters were used each day),
C
and
4p = variance of the error is the blank correction term
The different variance components are divided by the appropriate fac-
tors to account for reductions in the variances due to averaging effects.
To illustrate, averaging replicate analyses of the air sample from a single
canister reduces the analytical uncertainty but does not affect the addi-
tive random effects due to canisters or bags. Since only one canister and
only one bag are involved, no "averaging" of the random bag-or-canister-
associated errors is achieved. The different divisions of the variance
components are designed to account for this kind of effect. Actually, two
canisters were used on each day and two bags were also used on some days.
Confidence limits for each day's average NMHC concentration were cal-
culated as ± twice the standard error. These values were presented in
Tables 2 and 3.
51
-------
Weather and Data Correlations
Statistical tests were run to determine if any correlations existed
between NMHC levels and weather conditions or data. These were performed
to identify any systematic errors associated with the data. Since this
study was intended to determine NMHC levels upwind of the cities of interest,
tests were run to determine if wind direction or wind speed impacted the
data. Plots were also prepared to determine if NMHC concentrations varied
during different days of the week, which might suggest local influences.
Correlation plots were also prepared to determine if other weather condit-
ions could impact atmospheric NMHC concentrations. These studies provide
further insight on the validity of the analytical data or on any factors
which might directly affect atmospheric NMHC levels.
Plots of NMHC concentration as a function of weather conditions and
data are presented in Appendix A. Data points from duplicate samples
were treated independently in these plots.
The correlation coefficients indicating possible relationships between
NMHC concentrations and temperature, humidity, wind speed and pressure are
presented in Appendix A. The correlation coefficient is a measure of the
strength of the linear relationship between two variables; its square is
interpretable as the fraction of the variance in either variable that can
be explained or predicted in terms of the other. In Louisville, slight
negative correlations were observed between NMHC concentration and temper-
ature and wind speed; while for Nashville slight negative correlations were
observed between NMHC concentration and wind speed and pressure. These
observations should be viewed with discrimination since a limited number
of data points were available for the study. Although correlation
coefficients could not be calculated for wind direction data (non-numerical
variable), plots of wind direction and NMHC concentrations for each site
were prepared and are presented in Appendix A. With the number of data
52
-------
points available it is not possible to recognize (from these plots) any
relationship between wind direction and NMHC concentrations.
53
-------
REFERENCES
1. State Implementation Plans: Approval of 1982 Ozone and Carbon Monoxide
Plan Revisions for Areas Needing an Attainment Date Extension. Federal
Register. 45(191): 64856-64861, 1980.
2. Singh, H.B. Guidance for the Collection and Use of Ambient Hydrocarbon
Species Data in.Development of Ozone Control Strategies. EPA-450/4-80-
008, U.S. Environmental Protection Agency, Research Triangle Park, NC,
1980.
3. Guidance for Collection of Ambient Non-Methane Organic Compound (NMOC)
Data for Use in 1982 Ozone SIP Development, and Network Design and
Siting Criteria for the NMOC and NO^ Monitors. EPA-450/4-80-008, U.S.
Environmental Protection Agency, Research Triangle Park, NC 1980.
4. Lee, K.W., QC/QA Document for Collection and Analysis of Nonmethane
Hydrocarbon Data From Upwind Ozone Monitoring Sites. DCN 80-240-016-01-
03. Radian Corporation, Austin TX, 1980.
5. SAS Users Guide, 1979 Edition. SAS Institute, Inc., P.O. Box 10066,
Raleigh, NC, 1979. SAS Nested Procedure, page 313.
6. Searle, S.R., Linear Models, John Wiley and Sons, Inc., New York, 1971.
54
-------
-------
APPENDIX A
NMHC: WEATHER CORRELATION PLOTS
-------
U1
l_n
1.0 ~
0.9
H
C 0.0
C
0 0.7
N
C
E 0.6
N
T
R 0.5
A
T
1 0.M +C
0
N
0.3
0.2
0.1 ~
C*
c*
c *
c*
c *
B*
V
c
c
C
L*
C*
c
1
1
1
1
2
2
2
2
2
2
2
2
2
2
4
0
0
6
7
8
9
0
1
2
3
H
5
6
7
a
9
U
1
2
S
S
S
S
S
S
S
S
S
S
S
S
s
S
5
u
a
e
E
t
E
E
fc
E
E
E
E
E
E
E
E
L
c
c
p
P
P
P
P
P
P
P
P
P
P
P
P
P
P
T
T
8
6
6
6
6
6
6
a
a
6
a
6
a
6
O
a
a
0
0
0
0
0
0
0
0
0
0
0
0
o *
0
u
0
0
note: "Data points coincided.
SAMPLING date
Figure A-l. NMHC Concentration Plotted as a Function of Sampling Date: Louisville Site
B - Duplicate Bag Sample, C-- Single Bag/Duplicate Canister Sample,
* - Data Points Coincided for That Day.
-------
0*40 +
0,35
Lf l
ON
0.30
0,25
0,20
0,13
0,10
0,05 ~
C*
c*
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
0
0
t
7
e
9
0
1
2
3
4
5
6
7
8
9
0
1
2
s
S
s
s
s
S
S
S
S
S
S
S
S
S
s
0
0
E
E
E
E
E
E
E
E
E
E
t
E
E
C
E
c
c
P
P
p
P
p
p
P
P
P
P
P
P
P
P
P
T
T
8
S
8
e
s
e
S
e
e
8
8
8
8
e
8
8
8
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
SAMPLING DATE
NOTE: *Data points coincided.
Figure A-2. NMHC Concentration Plotted as
B - Duplicate Bag Sample, C -
* - Data Points Coincided for
a Function of Sampling Date: Nashville Site
Single Bag/Duplicate Canister Sample,
That Day.
-------
1.1
Ui
-^J
1.0
A
V
E 0.9
R
A
g o.a
H 0.7
c
C 0.6
0
M
C 0.5
E
N
T 0.4
ft
A
T 0.3
1
0
N 0.2
0.1
0.0 +
- + -
N E
E SE
WIND DIRECTION
SW
•-*
NU
Figure A-3. NMHC Concentration Plotted as a Function of Wind Direction:
Louisville Site. A = 1 Observation, B = 2 Observations.
-------
0 .HO +
l~n
00
A
V
E
R
A
G
E
H
C
C
0
(J
C
E
N
T
R
A
T
1
0
N
0.35
0.30
0.25
0.20
0.15
0.10
0.05 ~
NE
r se
WIND DIRt-CT ION
SU
NW
Figure A-4. NMHC Concentration Plotted as a Function of Wind Direction:
Nashville Site. A = 1 Observation, B = 2 Observations.
-------
Ui
-------
0«<*0 ~
ON
o
0.35
0.J0
0,2*
0,20
0,15
0.10
0.05 ~
53
t) A
A
57 5* 61 63 65 fa 7 69 7i 73 75 7<
tcmhekaturc
Figure A-6* NMHC Concentration Plotted as a Function of Temperature: Nashville Site
A = 1 Observation, B = 2 Observations, C = 3 Observations.
-------
l.l
1 .u
£>.*
0.6
0.7
O.t
0.5
O.f
O.i
0,2
0,1
0 A
O.U ~
*2 75 1h 75
76
>.t.—t — .t.
77 ft 7?
HO bj 02
-------
O.HO ~
N3
0„45
A
V
C
l<
A 0.30
0
E
H
C 0,i>5
fj
C 0.20
t;
/j
T
R
A 0.15
T
I
o
u
0.10
0.03 ~
60
A
A A
66 AU 90 92
RELATIVE HUMIDITY
9H
1UU
ngt£: f, oas had valuls
Figure A-8. NMHC Concentration Plotted as a Function of Humidity: Louisville Site
A = 1 Observation, B = 2 Observations, C = 3 Observations.
-------
U>
1.1
l.J
A
V
C 0.*
R
A
G O.tt
H 0.7
C
C 0,6
0
'J
C 0,5
r o,4
ft
r o.j
i
o.i
7 *
UI NO SHLCn
10
12
Figure A-9. NMHC Concentration Plotted as a Function of Wind Speed: Louisville Site
A = 1 Observation, B = 2 Observations, C = 3 Observations.
-------
0 # HO t
0.35
O.iO
o.**
ON
0,20
o.io
O.uS ~
WIND SPCEO
Figure A-10.
NMHC Concentration Plotted as a Function of Wind Speed: Nashville Site
A = 1 Observation, B = 2 Observations, C = 3 Observations.
-------
1.1 ~
1.0 ~
0,9 ~
Q.t» ~
C.7 ~
0 • b ~
o.s ~
o.* ~
o.i ~
0.2 ~
0.1 ~
0.0 ~
B
A
A
A3
A
0 A
29.60 29 v 6b 2*.7f. 29,0«* a*.92 50.00 50.08 50,16 50,2** 30.3*
oakumctkic KRrssunt.
Figure A-ll. NHMC Concentration Plotted as a Function of Barometric Pressure
Louisville Site. A = 1 Observation, K = 2 Observations.
-------
O.hO ~
A
cr>
cr>
0.35
A
C
R
A 0.30
G
C
>1
c o.«
c
0
C 0«2Q
1
• J
T
ft
A 0.15
T
I
0
(4
0.10
0.0^ ~
29.35 2*.<*3 2*.51 29,t>9 29,67 29.75 29,03 29,*1 2^.99 30,07 J0»1S 30,2J
8AKUMETHIC fRESSURt
ijuTC:
H DOS HAD MJSSIIlG VAlUIS
Figure A-12. NMHC Concentration Plotted as a Function of Barometric Pressure:
Nashville Site. A = 1 Observation. B = 2 Observations.
-------
I
.1 ~
1.0
0.9
a¦>
•^j
0.7
0 . b
0.1
O.b ~
I.OTE :
CLUUO CUVER
U OHS I'Au mISMHO VAlULS UK WLHL UUT (J* RANGt
Figure A-13. NMHC Concentration Plotted as a Function of Precipitation: Louisville
Site. 1 = Clear, 2 = Fog, 3 = Overcast, A = Rain. A = 1 Observation,
B = 2 Observations, C = 3 Observations.
-------
A
0#05 ~
I
CuUUO CUVEft
Figure A-14. NMHC Concentration Plotted as a Function of Precipitation: Nashville
Site. 1 = Clear, 2 = Fog, 3 = Overcast, 4 = Rain; A = 1 Observation,
B = 2 Observations, C = 3 Observations.
-------
ON
x#0
A
V 0.9
t
ft
A 0,0
0
e
0,7
O.o
c
0
ri O.b
c
c
H 0.H
t
R
A 0,3
T
1
0 0,2
0.1
0.U ~
\
OAT Of WEEK
Figure A-15. NMHC Concentration Plotted as a Function of Day of Week: Louisville
Site. 1 = Sunday, 2 = Monday, 3 = Tuesday, 4 = Wednesday, 5 = Thursday,
6 = Friday, 7 = Saturday; A - 1 Observation, B = 2 Observations,
C = 3 Observations.
-------
A
0.03 ~
~» 2 j' t
OAT OF WEEK
,OTC: I OBS '">0 MlS^IflG VA|_UES OH k£KL out I* RANGE
igure A-16. NMHC Concentration Plotted as a Function of Day of Week: Nashville
Site. 1 = Sunday, 2 = Monday, 3 = Tuesday, A = Wednesday, 5 = Thursday,
6 = Friday, 7 = Saturday; A = 1 Observation, B = 2 Observations,
C = 3 Observations.
-------
TABLE A-l. CORRELATION COEFFICIENTS1 FOR RELATIONSHIPS BETWEEN
NMHC CONCENTRATIONS AND WEATHER CONDITIONS
Correlation
Site
Relative
Temperature
Humidity
Wind
Speed
Pressure
Pearson
L
-0.5362
-0.1942
-0.3285
-0.3313
Probability
L
0.0023
0.3039
0.0764
0.0737
Spearman
L
-0.6113
-0.0076
-0.5405
0.3040
Probability
L
0.0003
0.09681
0.0021
0.1025
Pearson
N
-0.0199
0.1300
-0.4302
-0.3768
Probability
N
0.9199
0.5642
0.0223
0.0695
Spearman
N
0.3381
0.0688
-0.4661
-0.3563
Probability
N
0.0785
0.7611
0.0124
0.0874
*The Pearson
defined as:
correlation coefficient
E(x-x)
for random
(V-if)
variable X
and y is
Z(X-X)2 E(£/-£/;
Where X. and y are the sample means of x and y respectively. Its square
is interpretable as the fraction of the variance of either of the variables
which can be explained or predicted in terms of the other.
The probability associated with each correlation in Table A-l is the
probability that as large a correlation as the one observed or larger would
have been observed if the two variables were actually unrelated. Thus, if
the probability is small (say, less than 0.05), then one can be reasonably
sure that the two variables are statistically related.
The Spearman correlation coefficient is the same as the Pearson co-
efficient except that the ranks of the variable, rather than the original
data values, are used. The Spearman correlation is much less sensitive to
extreme values than is the Pearson correlation. The fact that the Pearson
and Spearman correlations are very similar suggests that there are no
isolated extreme points which dominate the Pearson correlations.
71
-------
TABLE A-2. WIND DIRECTION DATA COLLECTED AT THE SHEPARDSVILLE, KY
DIVISION OF AIR POLLUTION CONTROL
SAMPLING SITE BY THE KENTUCKY
• 1 I.
UMlM MM'OHI. MM IKD UIJUMUM I98U
I HAUliA 11\ , V'<
¦III iCUUf1
in* i.'lK'.'jS
CtJUU f'
PC I U SI
KirutC i 1 < ui»r
I H I S
i n ri i liiir jMiLirjs* u n
1203 I f I tMLKTMin "Jt MOOl
OUl L I T T
MOttlM O'MIUfll
<05l 4;»'FCI»U 'j I UD I f 'j
vf u
so
98 0
OOO I/ '
tvix 'j M f ll>
'it u I I Ml 1 vHf
(U V IftOUMfcr
l|f», kit
at( r r> f mb< I *>;* u
~ Or, 1 H
mjuubi**
r • u • » <•« * »o<>. h t¦ <¦o
80 04 2 1
fER llUDf':
UN I f •» » COOE ) - \ 4 i UeOftrTE 1
KfCOHD PE»!UD jfKlfHi£K,
U|f D DlHcA MUft
nf IHOU ' t. ubt » M|>) vniir
rtviwAOjNC 11 rti '<.oi»r i;n# »«hi»
'i •
A R
» 1 »
o
M u U
«
K
c«;. >
Urt .
00
0
02
0 3
0'«
y5
06
07
08
09
1 0
\ )
' 2
» 3
1 H
i 5
1 6
1
• 8
1 9
20
2'
22
2 7 I.U-i
0 1
^^5
1 jO
5
30
75
' 20
225
225
225
225
2*«o
24^
240
225
24J
255
2*0
221
MO
*0
• 0
V ¦
ft.*
145
145
1 ^
30
$
45
60
.;S1
W 70
90
?I0
»91
2 70
300
3O0
315
3>5
3i1
130
3 30
330
5
JO
i <
SU
' •
45
5
60
5
M 5
45
90
75
00
60
75
1 01
• 01
90
t 35
t 31
t 35
•«5
60
60
30
41
60
i M
' J
OH
t >
J 10
IH-J
CO
30
30
60
60
0
3 JO
221
211
211
211
240
2 0
30
« 35
240
1 80
i 80
/ 10
3 5
60
»*•
.,
05
*40
300
300
i 35
' 20
> 20
i'85
31 5
270
240
221
125
225
270
300
3 1
330
3 iO
300
270
vo
2 '0
) 30
JM
250
()(,
145
30
45
' 05
1S5
3S5
5
1
315
330
240
270
110
330
3 11
341
330
330
;o
• 5
30
45
• 0
60
2"
CO
O
00
75
330
1H5
3h5
255
90
5
45
300
2iO
2»0
240
300
221
30
4 1
41
i JO
< 00
< OO
• 80
m5
CO
i*<
66
90
)0
00
60
60
bO
00
90
45
205
i 01
1 oo
30
240
225
240
251
270
*80
• 20
CO
75
60
5
2**
2 •
0 9
130
5
^15
130
«*H 0
285
L'U5
J 5
330
2H0
225
225
211
2 10
770
201
201
205
201
215
05
75
30
5
?'•
220
i «
30
JO
30
30
5
5
30
30
30
30
60
41
4 1
41
30
30
30
1
30
30
30
60
90
90
2m
)U
\ \
1 U
»1
345
M5
75
'5
4 5
90
45
60
71
i I 0
«01
90
71
41
00
>01
60
45
¦«5
>5
5
60
2m
60
t 2
JO
30
60
» 05
H5
30
4 5
4 5
240
i 00
1 05
221
221
2 0
240
240
240
211
f 40
300
»5
90
) ;o
Kv
2*
'62
1 1
1 1*
145
15
5
>5
30
5
345
300
225
225
221
240
240
240
240
221
22 5
» 95
~65
<65
• 95
' v
2 3
' fi 3
1
145
15
300
255
240
255
240
225
225
255
300
300
3uO
205
300
311
3 1 5
300
>45
1 30
125
1J0
5
f *
169
1 *,
5
ibO
2 to
1M5
J 30
to5
i 05
150
i 50
t 60
30
41
30
5
•>
41
00
41
41
30
45
4 5
90
60
2 ~
' b
vn
145
«
•il
345
JO
4 9
90
10
195
300
n5
*5
2 0
2 iO.
«95
» 95
2 0
t 95
' 50
• 20
« o'i
1 9
2 • O
*•»
• 5 '
1
>30
2«?5
3 5
zei
£85
240
1 95
240
240
2 2 5
255
2 70
300
3 5
3t5
315
315
330
330
330
130
•)
o
2'*
2 m Q
i *
75
3H*5
90
i 35
34 5
4 5
« 35
60
00
7 5
75
90
100
240
270
• 00
« 00
5
>45
75
>
45
9v
f4
) 1
1 9
75
60
H5
45
» 05
90
5
60
90
1 00
'01
HO
I CO
« 65
195
ISO
2 0
i 50
75
• 5
»5
t 65
• UO
2 )
ifK5
345
1*15
5
6U
60
4 5
45
75
2 U 5
240
255
20 5
2 2 5
5
240
255
2 tO
2 0
i bo
• OO
1C<>
10
Jm5
M
vu
»
145
^ fO
2^5
-MO
225
.725
0
210
240
2*»0
240
'240
240
240
225
240
225
225
225
2 2 5
225
2 0
0
259
k'M
2 34
<"•0
1 uo
1 11 >
\ ^5
i25
? 2"i
0
*25
225
2*5
? s»
/HO
2*15
? 0
2 1 0
225
225
jfsO
2 70
221
2 70
1 5
' CO
295
2 M
/»'0
<»5*>
•10
t i r>
<*55
»30
5
5
5
:io
i 5
41
60
6-J
5
0
5
15
5
**5
45
• 05
14 9
2M
9 «
<'¦*
«5
45
)'• 5
> 05
90
'•5
45
3D
14 5
;5
iH
75
s5
300
9 0
90
90
• .15
i 20
*/0
~ 35
JO
11
• SO
2m
' 0>)
^•)S
1 <0
IM5
i 10
">
60
GO
i 95
2 2 '.'•
241 "5
<•5 5
J » 5
100
2 70
i'n»
270
2 7 O
2 '0
255
2 >0
255
i.2 5
~ 0
/ 2 *
2*«
IM">
n«,
Ifeo
3'»5
14 5
i 5
J 5
0
115
5
30
30
5
5
160
14 5
5
J4 5
330
5
45
145
5
•;
2**
• 9 1
IT,
IS'.
»15
30
»<»s
1 2 0
*30
00
00
7 5
i 0 5
1 2 0
i 20
1 «' 0
t i'O
i 05
i 20
45
h5
00
M5
n
» 5
2 4
' 0 4
.'H
» ()'.
>5
10
60
90
4 5
145
60
45
i5
'iO
5
i 05
1 O'i
» 05
105
5
90
10
JO
90
10
ft
4 5
«?M
to
<•9
r.o
J5
'.O
^0
UO
60
00
GO
60
•«
«J0
1 i'O
i 11
1 05
9 0
» O'I
'5
90
1 15
105
240
205
60
• NO
2*«
• 0 1
k»
i m>
1 <>D
»1*V
<*0
60
60
60
S
1 15
1
'•}
*i 0
' '
105
9"
CO
105
i 20
« 50
' 15
40
h*
10
£ 1
4M
M
/ M
."i
<*y
IU
<"*
~ M
)0
Jo
«0
iO
10
10
li
10
10
:io
10
10
30
10
10
10
IO
24
MM
1 >1 >
1 /'I
j
1 VJ
" sv
1 »
I OH
1 'Jt
1 4 4
1 / 1
1 0 *
1 0 1
1 I. t
IS"
• /
t -tl
1 11
« 9 2
1 14
• 64
1 4b
» / '
' 1
' J4
?4*
<(.()
K.o
r«'»
,.,S
I4S
145
1 *>
< 4 •>
1 10
I'll)
11 5
J0>|
IMJ
t(»0
!'«•»
1 IO
IMS
» IO
IMS
1 10
IMS
1 iO
HI
itnx J 11(111 on 'it u vn i | on
-M m« in: • ill nit
"UMIH w III HUM UVA 1 I IIH'i
-------
TABLE A-3. WIND SPEED DATA COLLECTED AT THE SIIEPARDSVILLE, KY SAMPLING SITE BY THE KENTUCKY
DIVISION OF AIR POLLUTION CONTROL
~< ' w H»«U I'll 1 »
I r L (LlJUf
AtlUkt '»r>
( uuu r f
HFdlOH
PUOjc :: 1 ' tiDf
> (II1I1C M 1 S
IN.II f Uk |HU UiJAWUH < SHO
ifi II i SllO'Ht HU'jv 11.1 J
KUgY t L t ME H 1 AH Y SOltlOl
QUII I 1 I
NOHlM
<0*, FECIAL •itlJOlfj
Vf M
•t*o
vO" •' / '
I »'.« Mil It » oS ¦ « H
"j t rt 1 | OH c HfSlUN1'*,
f I L v fcgOVf » 7 f I f 1
wtn. ».n /»<».# h' * too ¦ * ¦»
IrATI f StA0t ISMl v no 0** J I
I'AMArtf If H 'CQDEi i 6 1 1 0 » ) UlHD SKEfU
'JHMS <«.00£). Mli.ES f*trt MUOH
HEiUHD HC»l(JD GFtMt'HKEtf. \ <*U0
fifli'OD 0
i
0
2 5
t
•
c
0
0
2"
I
t
OS
'
s
0
0
5
0
5
5
0
0
0
5
0
2
0
2
1
3
u
0
6
0
8 >>
6 0
3
' *
>«
V
u
0
u
2"<
/
«
00
0
0
0
%
0
5
0
i
0
0
0
•)
t
1
0
0
0
3
0
5
6
0
7 1
0
0
9 0
6 1
6
0
J C
0
0
0
0
4 '•
2
0
i)
r,
0
5
0
1
0
t
$
0
t
0
1
2
0
2
*
3
0
«4
0
7 0
0
0
5 0
2 1
V
5
• 5
3
;
5
f •«
2
1
Oli
1
I)
1
0
'
5
0
5
£?
0
I
5
1
0
2
0
0
2
&
0
M
0
0
2 *
•»
0
N 5
«• 0
0
0 5
0
0
0
o
V
! *
/
09
0
9
0
0
0
5
0
0
0
5
»
0
i
0
5
?.
0
J
1
0
&
*
G
7
5
8 5
•
0
7 0
1
•» 0
0
J
Q
*
!H
i
*
1 fl
8
0
•.
9
;
0
5
1
0
0
5
0
6
7
0
6
u
C
5
7
0
7 5
7
0
7 $
& 1
3
i o
2
0
i **
' '
0
0
»
•5
0
5
5
0
u
U
0
0
1
5
?
5
?
0
0
H
5
5
1
«« 0
3 1
2
0
0 5
0
0
0
0
0
; •<
2
»t>
u
9
ft
%
'
0
5
0
0
0
5
u
\
0
1
0
1
0
2
0
J
1
6
0
0
9
7 5
6 0
t 9
t
5
1 0
0
0
0
5
2>*
2
»
i
')
%
o
}
0
y
o
0
0
0
i
0
7
8
0
* 0
0
9
0
1
0
1 <>
Q
0
8 1
8 0
3
5
0 5
0
e
0
u
2H
)
I *4
1
0
0
9
1
0
(J
?
«;
0
¦i
0
0
6
0
7
*
6
0
7
i 0
0
i t
0
' <)
0
9 *
« 0 1
1
0
© 5
2
0
0
2*'
-,
1 s
6
0
3
1
s
s
\
0
%
0
0
1
0
1
0
0
6
6
5
6
0
0 0
5 5
J
0
1 0
2
c
t
»
0
2H
»
)
> b
0
1
%
0
5
(J
<>
s
t
0
1
0
\
0
t
0
£
0
>
3
*
2
3
5
J
1 0
0
0
9 0
5 o
3
0
i *
•
«
3
0
««
0
J
C>
1
1
0
t 1
0
a
5
h
0
5
&
0
H
6
0
7
5
6
0
>
0
5
5
H 5
5
5
i 5
/ 0
1
9
3 0
J
0
0
5
' 8
't
s
0
i
5
0
1
u
\
0
0
1
0
1
0
?.
J
5
?.
*
i
0
•i
0
J
•s
H 0
3
0
J 5
3 0
• 0
0
0
I
*•«
.
9
1 f
¦>
5
1
0
0
0
Ij
s
*
0
5
o
1
»
5
2
5
0
*
&
0
0
5 0
3
*
3 0
2
9
• 5
0
0
0
2 4
2
*
jy
i
't
<
0
o
*>4
•)
'
0
0
0
0
0
I
0
0
5
?
9
«*
8
0
*
0
9
0
* 0
8
5
8 1
0 u
J
' 0
5
0
0
*•«
3
9
t»
0
'»
'
*•
5
U *
*
)
0
t>
$
0
5
0
5
0
1
5
i 0
5
i 3
0
• 5
0
• 5
u
» w
0
1 0 0
0
0
i l 0
V o
»
0
5 1
N
0
2'<
»
1
0
0
i,
o
0
&
5
6
0
1
«,
s
6
0
1
0
/
0
« t 0
0
0
1 « 0
b 0
9
0
1 0
9
0
•«
0
5
0
*'•
t
o
u
•4
't
0
<•
H
0
6
7
(i
(J
6
5
6
0
>«
/
»
1 1
0
1 0
«« v
/
5
• 0
0
0
1
o
2*4
1
9
1
0
»
o
»
1
•>
5
1
0
»
3
5
)
0
»
0
0
2
0
2
u
2
0
» 5
1
0
J 0
«r 5
•
» 5
<
0
t
0
.
0
?*~
2
0
1
')
o
1
0
0
V
0
»
U
»
}
J
w
0
•4
1
0
1
0
•«
N 0
•4
0
5 0
5 0
i
2 5
I
3
2
0
j-
2
»
.'L
'*
41
(.
s
5
•t
'I
'
N
/
0
1
S
1
s
b
6
u
5 o
1
•>
1 0
» 5
0
5
• o
0
5
0
5
i
0
2-4
)
9
s '
'»
>J
0
1)
s
«
0
i
1)
0
0
«>
2
•>
0
1
5
S
5
1 1
1
0
/ 1
•« 0
)
5
i *>
0
I
0
2i
2
1
• >
1'
•»
1
<)
"
I
o
1
1
0
1
*
j
0
1
*)
s
* 0
•4
:i o
/ o
\
5
0 *
<
0
2
*
0
A
2*4
i
2
u
<«
0
»
«!
>
II
t
0
•i
0
I
«>
I
u
'<
o
•«
5
0
•1 1
•>
0
I s
I o
0
t ^
N
i
,»
>•,
2
b
'
"
'
0
z'
"
i}
''
t
"
*'
G
n
1
0
b
0
h
">
'j 0
i s
J 0
*
o
0
o
0
0
4>
2*4
2
*
..
iu
MJ
IU
in
Ml
»«»
!«'
w
<0
• 1
»0
(0
»0
0
IO
IO
IO
IO
IO
IO
JO
li>
n>
IO
M $4
'
H
rt
'
/
•
'
'
H
<
t
'
*)
K
H
1
(.
•<
i, '
h
t
U ,'
•« *
1
'
*.
H
•
1>
n«.
i
"
"
«
'
"
'
"
t.
'
"
•>
'¦-
"
1 S
»
"
I ,' 0
"
1 o
«o •>
>
"
f *
«*
u
-
0
'
••
nr< x rum
MR
)| IfYAf ||)N
> •> u
Reproduced from
best available copy.
NiiHHt v id iiHM vvn i | i
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completingj
1. REPORT NO. 2.
EPA 9 04/9-80-055
3. RECIPIENT'S ACCESSION NO.
/ v :> > \ ,•
4. TITLE AND SUBTITLE
COLLECTION AND ANALYSIS OF NONMETHANE HYDRO-
CARBON TRANSPORT DATA, Louisville, Kentucky
and Nashville, Tennessee Ozone Study
5. REPORT DATE
January, 19 81
6. PERFORMING ORGANIZATION CODE
7 AUTHOR(S)
Robert D. Cox, Kenneth W. Lee,
Gary K. Tannahill, Hugh J. Williamson
8. PERFORMING ORGANIZATION REPORT NO.
DCN 81- 240- 016- 01-07
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Radian Corporation
8501 Mo-Pac Blvd.
Austin, Texas 78766
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
EPA 68-02-3513
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Protection Agency
Region IV
345 Courtland Street
Atlanta, Georgia 30365
13. TYPE OF REPORT AND PERIOD COVERED
Final r Aiip- . - Dec. 1980
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
EPA Project Officer: Douglas C. Cook
16. A8STRACT
This document describes the collection of nonmethane hydrocarbon
transport data to be used in Level III: EKMA-OZIPP analysis for Louis-
ville, Kentucky and Nashville, Tennessee. Ambient air samples were col-
lected during September, 1980, at sites normally upwind of these cities.
Integrated samples were collected at each site from 6:00 am to 9:00
am daily for two weeks. Samples were collected in Tedlar® bags then
transferred to stainless steel canisters for shipment and analysis. Cryo-
genic trapping with liquid oxygen was used to concentrate the hydrocar-
bon species and to separate methane and nonmethane hydrocarbons. Non-
methane hydrocarbons were thermally desorbed in a gas chromatograph and
quantitated with a flame ionization detector (FID). This method was
thoroughly tested and data in support of it are presented within the
report.
Nonmethane hydrocarbon (NMHC) concentrations (6-9 am average) near
the upwind monitoring site for Louisville, Kentucky, ranged from 0.11 to
0.96 ppmv-C with a mean value of 0.30 ppmv-C and a median of 0.22 ppmv-C,
NMHC concentrations near the upwind monitoring site for Nashville,"Ten-
nesee ranged from 0.06 to 0.34 ppmv-C with a mean value of 0.12 ppmv-C
and a median of 0.09 ppmv-C
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Air Pollution
Atmospheric Sampling
Nonmethane Hydrocarbons
Transport
18. DISTRIBUTION STATEMENT
Release Unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
77
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (Rev. 4-77) previous edition is obsolete
------- |