FINAL REPORT:
STUDY OF ORGANIC EMISSIONS AND
POTENTIAL HEALTH EFFECTS
UPJOHN CHEMICAL COMPANY
NORTH HAVEN, CONNECTICUT
Contract No. 68-01-6312
Task Order 32
Prepared for:
U.S. Environmental Protection Agency
Region I
John F. Kennedy Federal Building
Boston, Massachusetts 02203
Project Manager:
Robert A. O'Meara
September 1984
9432.60/41A, 41B, 41C
Submitted by:
Engineering-Science
Two Flint Hill
10521 Rosehaven Street
Fairfax, Virginia 22030
-------
Qq
°6
vV
-------
FINAL REPORT:
STUDY OF ORGANIC EMISSIONS AND
POTENTIAL HEALTH EFFECTS
UPJOHN CHEMICAL COMPANY
NORTH HAVEN, CONNECTICUT
Contract No. 68-01-6312
Task Order 32
-------
TABLE OF CONTENTS
SECTION 1 EXECUTIVE SUMMARY 1-1
Background 1-1
Project Objectives 1-2
Duties of Project Participants 1-3
Summary and Conclusions 1-4
Process Vent and Lagoon Sampling Study 1-4
Ambient Air Quality Study 1-5
Dispersion Modeling Study 1-8'
Health Effects Study 1-9
SECTION 2 PROCESS VENT/LAGOON SAMPLING STUDY 2-1
Introduction 2-1
Background 2-1
Summary 2-2
Conclusions 2-3
EPA Screening Study 2-4
Process Monitoring 2-6
Process Analysis 2-6
Process Monitoring in the Field 2-6
Process Monitoring from Records 2-8
Procedures for Sampling Vents and Lagoons 2-9
VOC Sampling Procedure 2-9
Condensible Organics Sampling 2-11
Quality Control 2-12
Sample Handing/Chain-of-Custody 2-12
Analytical Procedures and Results for Process/
Vents/Lagoons Samples 2-13
Introduction 2-13
Analyses of Bag Samples from Process Vents 2-21
Gas Chromatographic Measurement Description 2-21
Results and Discussion of GS/FID Measurements 2-23
Quality Control for GC/FID Measurements 2-23
GC/MS Measurements Description 2-24
Results/Discussion of GC/MS Analysis 2-32
Analysis of Tenax® Tube Samples
Introduction 2-45
Analytical Procedures 2-45
Results and Discussion 2-49
Quality Control 2-49
ii
-------
Analyses of the Condensible Samples 2-53
Introduction 2-53
Analytical Procedures 2-53
Results/Discussion 2-55
Quality Control 2-63
Emission Rate Calculations 2-63
Volatile Organic Compound Emission Rates 2-63
Condensible Organic Compound Emission Rates 2-66
SECTION 3 AMBIENT AIR QUALITY STUDY 3-1
Introduction 3-1
Background 3-1
Summary 3-2
Phase 1: November 1980 Study 3-2
Phase 2: March 1981 Study 3-3
Phase 3: June 1981 Study 3-4
Conclusions 3-5
November 1980 Study 3-7
Introduction 3-7
Sampling Strategy 3-7
Equipment Preparation, Field Activities and
Analytical Procedures 3-9
Equipment Preparation 3-9
Field Activities
Analytical Procedures 3-11
Results and Conclusion 3-15
March 1981 Study 3-21
Introduction 3-21
Sampling Strategy
Equipment Praeparation, Field Activities and
Analytical Procedures 3-23
Equipment Pareparation 3-23
Field Activities
Integrated Air Samples 3-23
Instantaneous Air Samples 3-25
Analytical Procedures 3-29
Results and Conclusions 3-29
June 1981 Study 3-33
Introduction 3-33
Sampling Strategy 3-33
Equipment Preparation, Field Activities and
Analytical Procedures 3-34
Equipment Preparation 3-34
Field Activities 3-35
Integrated Air Samples 3-35
Instantaneous Air Samples 3-35
Analytical Procedures 3-42
Results and Conclusions 3-43
iii
-------
OCLC Connexion
Page 1 of 1
OCLC 1141769262 Held by EHA/LHR - no other holdings
Rec stat c Entered 20200224 Replaced 20200310
Type a ELvl K Srce d Audn Ctrl
BLvl m Form Conf 0 Biog MRec
Lang eng
Ctry vau
Cont b GPubf LitF 0 Indx 0
Desc i Ills a Fest 0 DtSt s Dates 1984 ,
040 EHA *b eng *e rda *c EHA *d EHA
050 4 TD883.5.C8 *b S78 1984
088 EPA 901-R-84-007
099 EPA 901-R-84-007
049 EHAD
245 0 0 Study of organic emissions and potential health effects, Upjohn Chemical Company, North Haven,
Connecticut: *b final report / *c prepared for: U.S. Environmental Protection Agency, Region I;
submitted by: Engineering-Science.
264 1 Fairfax, Virginia: *b Engineering-Science, *c 1984.
300 1 volume (various pagings): *b illustrations, tables ; *c 28 cm
336 text *b txt #2 rdacontent
337 unmediated ^b n *2 rdamedia
338 volume *b nc #2 rdacarrier
500 "Project manager: Robert A. O'Meara."
500 "September 1984, 9432.60/41 A, 41B, 41C."
504 Includes bibliographical references.
536 Prepared for U.S. Environmental Protection Agency, Region I *b contract no. 68-01-6312 *g task
order no. 32.
650 0 Air #x Pollution *z Connecticut #z North Haven.
650 0 Organic compounds *x Environmental aspects.
710 2 Engineering-Science Company. *e author.
710 1 United States. *b Environmental Protection Agency. *b Region I. *e sponsor.
Delete Holdings- Export- Label- Submit- Replace-C Report Error- Update Holdings- Validate-C
Workflow-In Process
about:blank
3/10/2020
-------
SECTION 4
DISPERSION MODELING STUDY
4-1
Introduction 4-1
Background 4-1
Summary 4-2
Conclusions 4-2
Input Data 4-2
Discussion of Key Model Options 4-8
Results 4-11
SECTION 5 HEALTH EFFECTS STUDY 5-1
Introduction 5-1
Background 5-1
Summary 5-2
Conclusions 5-2
Health Effects 5-3
Health Effects of Toluene 5-3
Health Effects - Other Compounds 5-4
Health Effects of Benzene 5-4
Impact of Upjohn Chemical Company Emissions on
Residents of North Haven 5-5
Short-Term Exposures - Benzene and Toluene 5-5
Long-Term Exposures - Toluene 5-5
Long-Term Exposures - Benzene 5-5
Estimated Leukemia Risk from Exposure to
Upjohn Company Emission 5-6
Estimated Leukemia Risk from Exposure to
Urban Factors 5-9
REFERENCES
APPENDIX A SAMPLE LOCATION MAP FOR AMBIENT STUDY
APPENDIX B ANALYTICAL EQUIPMENT SPECIFICATIONS AND OPERATING
CONDITIONS FOR AMBIENT STUDY
APPENDIX C LAGOON EMISSION RATE PREDICTION BY SHEN TECHNIQUE
APPENDIX D DERIVATION OF VOC EMISSION RATES FROM UPJOHN
COMPANY AERATION LAGOON VIA COMPARISON OF MODELED
AND MONITORED CONCENTRATIONS
APPENDIX E CARCINOGEN ASSESSMENTS GROUP'S FINAL REPORT ON
POPULATION RISK TO AMBIENT BENZENE EXPOSURES
iv
-------
DISCLAIMER
This report has been submitted to the U.S. Environmental Protection
Agency, Region I, and has been reviewed and approved for publication.
This report does not necessarily reflect the views and policies of the
U.S. Environmental Protection Agency, nor does mention of trade name or
commerical product constitute endorsement or recommendation for use.
-------
SECTION 1
EXECUTIVE SUMMARY
1.1 BACKGROUND
The Upjohn Company, Fine Chemicals Division, plant located in North
Haven, Connecticut, has been the subject of concern to many of the people
living in the immediate vicinity. Numerous citizen complaints and congres-
sional inquiries have been received by the Connecticut Department of Envi-
ronmental Protection and the U.S. Environmental Protection Agency (EPA)
concerning odorous emissions from the plant. The plant produces a variety
of organic chemicals.
In August of 1979, EPA Region I conducted some field sampling in the
area of the Upjohn facility in an attempt to assess the impact of volatile
organic emissions from the process vents and lagoons on ambient air qual-
ity. A preliminary survey of the area surrounding the plant was made with
a photo-ionization detector. This survey failed to identify any high con-
centrations downwind of the plant. It was believed that meteorological
conditions at the time of the survey (high wind speeds of 20-3 0 mph) may
have diluted the emissions to below the threshold response of the detector.
The inspectors also noted that odors in the area were barely detectable
during most of the survey (Reference 1).
To supplement the photo-ionization detector measurements, carbon col-
umn samples were taken at two locations downwind of the plant for a period
of approximately five hours. The sampling locations were north of the
plant on Route 40 just off the plant's property line. The first location
was 180 meters downwind of the plant's manufacturing facilities, and the
second location was 105 meters downwind of the plant's main outfall into
the waste treatment lagoon.
On August 28, 1979, EPA personnel revisited the area and found a
slight increase in odors. This increase was probably due to a much lower
wind speed, approximately five miles per hour. Ihe same two locations
were sampled for a period of approximately ten hours.
According to EPA, analysis of the four carbon column samples showed
organic contamination in quantities representative of automobile emissions
only. Results of the sampling, however, were judged by EPA to be in-
conclusive. It was felt that the samples might be biased low for one or
more of the following reasons:
o Adverse meteorological conditions (high wind speeds) during sam-
pling
1-1
-------
o Poor sampling locations
o Process variability
It was still unknown whether the Upjohn facility produced elevated
ambient levels of organic compounds and whether a health hazard was posed
through the emission of those compounds. Thus, a comprehensive, study was
undertaken by EPA Region I in 1980 to answer these questions. That study
is the subject of this report.
The assistance of experienced contractors was acquired to accomplish
this task since the Region did not have the necessary resources to perform
the evaluations and carry out its other mandates. Engineering-Science
(ES) and Ecology and Environment were retained to complete the study. Two
subcontractors, GCA Corporation and Dr. Margaret Hitchcock, were retained
by ES to help with the project. The duties of each participant are des-
cribed in Section 1.3.
1 .2 PROJECT OBJECTIVES
The basic objective of the project was to determine whether a health
hazard was being posed to the population in the vicinity of the Upjohn
facility through the emission of organic compounds from the plant processes
and wastewater treatment system to the ambient air. In order to satisfy
this basic objective, the study was to:
1. Identify sources at the facility responsible for organic compound
emissions, identify the compounds, and quantify their emission
rates.
2. Develop and implement an ambient sampling strategy to determine
the types of organic compounds in the ambient air and the
short-term ranges of concentrations to which the area population
may be exposed.
3. Determine annual meteorological conditions and develop a computer
model to predict the long-term ambient concentrations of organic
compounds to which the area population may be exposed.
4. Perform a health assessment based on both short- and long-term
organic concentration data.
Typically, in a chemical plant there are several processes with nume-
rous pieces of process equipment. Only some of the equipment is likely to
emit significant quantities of organic compounds. In order to identify
the potential emission sources, a tour of the plant, a review of process
information, discussions with Upjohn engineers, and material balance
calculations were planned. A portable organic vapor analyzer was to be
used to screen suspected emissions points. Sources of potentially sig-
nificant emissions were to be sampled using rigorous source testing proce-
dures. Collected samples were then to be analyzed by appropriate methods.
The processes were to be monitored during this sampling to help optimize
the use of sampling equipment and personnel and to insure that the pro-
cessess were operating normally. Using the sampling and analysis results,
1-2
-------
as well as data collected via the process monitoring, organic compounds
were to be quantified. Long-term concentrations needed for the health
effects assessment were to be obtained from dispersion model calculations
using data from the emission sampling and from the National Climatic
Center in Asheville, North Carolina.
Ambient sampling was to be conducted simultaneously with the process
vent/lagoon sampling activities. Short-term contaminant levels at speci-
fic locations on and surrounding the Upjohn site under various meteorolog-
ical conditions were to be obtained. Real-time or instantaneous samples
were to be taken to show the range in which contaminant concentrations
could be detected at any instant at locations on and surrounding the
Upjohn site.
The information obtained by the process vent/lagoon sampling and am-
bient sampling activities were to be used to assess the potential hazard
to the health of persons in the vicinity of the plant. A health effects
consultant was to be retained for this purpose.
1.3 DUTIES OF PROJECT PARTICIPANTS
This multi-faceted study of the organic chemical emissions from the
Upjohn plant involved several EPA contractors and subcontractors, as well
as EPA personnel. The organizations and individuals involved, their re-
sponsibilities, and acronyms used elsewhere in this report are:
o EPA Region I: Robert A. O'Meara, Project Manager - general pro-
ject organization and direction; Environmental Services Division,
Frank Gorry and Dr. Thomas Spittler - ambient air sampling and anal-
ysis; Air Management Division, Jon Pollack - dispersion modeling
study detailed in Appendix D of this report (EPA).
o Engineering-Science: Dr. Dennis Falgout, John Yates, and Randy
Patrick - source sampling, dispersion modeling, preparation of
project report (ES).
o GCA Corporation: Technology Division, Dr. Gary Hunt - analysis
of source samples (GCA).
o Ecology and Environment, Inc.: Paul Exner - source and process
characterization, general assistance to EPA (E&E).
o Dr. Margaret Hitchcock, independent consultant - assess health
effects of organic air contaminants.
The review of the process information, tour of the plant, and discus-
sions with Upjohn personnel to identify potential emission points and com-
pounds were completed by E&E and ES. EPA performed the source screening
with the organic vapor analyzer. Based on the results of this work, EPA,
ES, and GCA developed a sampling and analysis strategy during several
project meetings and telephone conversations. Actual process vent and
lagoon sampling was conducted by ES. During the sampling, E&E and EPA
1-3
-------
monitored the processes. GCA analyzed the samples according to methods
agreed to by EPA and ES. The ambient sampling and analysis were con-
ducted by EPA. Dr. Hitchcock conducted the health effects assessment.
ES performed the dispersion modeling and prepared this project report.
1.4 SUMMARY AND CONCLUSIONS
1.4.1 Process Vent and Lagoon Sampling Study
Twenty process vents and the wastewater treatment system aeration
and settling lagoons were identified as potentially significant sources
of organic emissions during the process data review, discussions with
plant personnel, plant tours, and organic vapor analyzer measurements.
Benzene, toluene, and chlorobenzene were determined to be the volatile
organic compounds (VOCs) emitted in the greatest quantities. Of the
twenty process vents, ten were sampled downstream of air pollution
control equipment (scrubber, fabric filter, carbon adsorber), six were
sampled downstream of vent condensers, and four vents were uncontrolled.
The two lagoons and seventeen of the process vents were sampled by
ES for volatile organic compounds. Five vents were from identical re-
actor units and only two of the five were sampled. It was assumed that
the emissions from the other three reactors would be the same as the two
sampled. Similarly, only two of four tanks were sampled. Samples were
withdrawn from the process vents into Tedlar® bags using a lung type
sampler. This technique eliminated the potential for loss or contami-
nation of the organic compounds by the internal parts of the sampling
pumps. The rate of sample withdrawal was controlled by needle valves
and measured by rotometers. Tenax® tubes were used to collect samples
of the gases that were emanating from the lagoons. A small wind tunnel
was fashioned to simulate wind speeds of from 3 to 10 mph acroas the
surface of each pond. The wind tunnel consisted of an open-bottom box
placed over a one-by-six foot area of pond surface. A forced draft fan
provided air flow. Flow measurements, by vane anemometer, and samples,
using Tenax® tubes, were taken in the six-inch ID exhaust port. Addition-
ally, in order to study the entire organic emission situation and to
search for odor sources, three process vents (S-303, S-1701, and S-1703)
were sampled for condensible organic compounds by a modified version of
EPA Reference Method 5. Although not in the original scope of work,
two sources (K1702 and K-1705) were also sampled for condensible
organic compounds using EPA Method 6 type equipment with 0.1 N potassium
hydroxide in the impingers.
All the samples were analyzed by GCA at their Bedford, Massachusetts
laboratory. Hie Tedlar® bag samples were analyzed for volatile organic
compounds by gas chromatography (GC) using a flame ionization detector.
Gas chromatography/mass spectroscopy (GC/MS) was used for confirmation
of volatile organic compound identification in those cases were ambiguity
existed. Hie Tenax® tube samples were analyzed for both volatile and
semi-volatile organic compounds by GC and GC/MS. Unfortunately, quality
control data showed that volatile organic emissions could not be ade-
quately assessed using the analytical protocols chosen for the program.
1-4
-------
GC/MS was also used to analyze samples collected by the Methods 5 and 6
equipment for condensible organic compounds.
Emission rates for the process vents were calculated frcm the sam-
pling and analysis results and information obtained from the process mon-
itoring. Because some of the processes were not continuous, estimated
average and maximum emission rates were calculated. Tables 1 .-1 and 1 .2
summarize the volatile and condensible organic compound emission rates
respectively. The major sources of benzene were S-202, S-205, S-206,
3-213, and S-301. Average benzene emissions from these vents ranged
from 12.2 to 59.5 grams per hour (g/hr) compared to essentially zero
for the other vents. Vent S-1701, with an average emission rate of
7040 g/hr, was overwhelmingly the major source of toluene. Only three
other sources emitted more than 10 g/hr of toluene. Very little chloro-
benzene was emitted compared to the quantity of benzene and toluene.
Only S-214 released more than 1 g/hr of chlorobenzene. Of the conden-
sible organic compounds identified, 2,4-dichloro-6-nitro-benzeneamine
was emitted at the highest rate, 125.5 g/hr on the average.
1.4.2 Ambient Air Quality Study
In order to determine the types of organic compounds in the ambient
air and the short-term ranges of concentrations to which the area popula-
tion may be exposed, an integrated air sampling strategy was formulated
to collect samples on the Upjohn property upwind and downwind of the
processing area and the lagoons and in the surrounding neighborhoods.
Ambient sampling was conducted by EPA concurrent with the process vent/
lagoon sampling during the period November 17-21, 1980. Tenax® and
charcoal were chosen as the sample collection media for VOCs. Due to
the nature of condensible organic compounds, it was felt that they
would not be present at measurable levels far from their emission points.
Samples were collected at 17 locations and later analyzed at EPA's New
England Regional Laboratory in Lexington, Massachusetts. A meteorological
station was set up on the Upjohn property to measure wind speed and
direction during the sampling period.
Results of the November sampling effort were disappointing. Evidence
was found that led EPA to conclude that the Tenax® data could not be
used with confidence. Also, the charcoal samples were stored for an
unrecommended period before analysis, which may have resulted in some
sample deterioration. As a result of these problems, the Tenax® data
were voided and the charcoal data could be used only to show approximate
pollutant levels. The charcoal data indicated upwind levels at five
locations of 0.7 to 2.3 parts per billion (ppb) benzene and 4.0 to 15
ppb toluene. Benzene levels immediately downwind of the processing
area were not significantly above background, but toluene levels up to
60 ppb were measured. On only one occasion was benzene measured at a
level significantly above background and on one other occasion the
toluene level was elevated in the downwind neighborhoods.
Because of the problems with the samples collected in November,
a second ambient sampling program was conducted during the period from
March 24-26, 1981. The sampling strategy was similar to that for Novem-
ber except that precautions were taken to insure that the November prob-
1-5
-------
TABLE 1.1
SUMMARY OF VOLATILE ORGANIC COMPOUND
EMISSION RATES FROM PROCESS VENTS
UPJOHN COMPANY, NORTH HAVEN, CONNECTICUT
Average Emission Rate for Maximum Emission Rate for
Two Batch Cycles (g/hr) Single Bag Sample (g/hr)
Emission
Sourcea
Benzene
Toluene
Chloro-
benzene
Benzene
Toluene
Chloro-
benzene
S-202
12.2
0
0
41.8
0
0
S-203
0.047
0.134
0.495
0.054
0.181
0.692
S-205
30.9
0
0
68.9
0
0
S-206
50.8
1 .23
0.413
197.
3.70
5.06
S-213
20.8
0
0.00042
32.3
0
0.00261
S-214
0
53.6
2.48
0
238
7.99
S-301
59.5
0
0
94.8
0
0
S-303
0
0
0
0
0
0
K-316
0.105
0
0.0323
0.158
0
0.0659
T-393/394
0.102
0
0
0.160
0
0
T-395/396
0
0
0
0
0
0
S-1701
0
7040
0
0
1 1900
0
S-1703
0
17.3
0
0
36.6
0
K-1701-5
0
1 .78
0.0967
0
8.60
0.502
K-1706
0.103
4.38
0.0174
0.0868
74.1
0.228
T-1718
0.114
56.8
0
0.594
383.
0
a The source identification scheme is as follows: generally "K" repre-
sents a reaction kettle, "S" a scrubber, "T" a tank vent. Hie leading
"2", "3", or "17" represents the building number, and the remaining
digits, the source number.
1-6
-------
TABLE 1 .2
CONDENSIBLE ORGANIC COMPOUND
EMISSION RATES FROM PROCESS VENTS
UPJOHN COMPANY, NORTH HAVEN, CONNECTICUT
Emission
Source
Condensible
Compound
Average Emission Rate
for Two Sampling Runs
(g/hr)
Maximum Emission Rate
for Two Sampling Runs
(g/hr)
S-303
S-1701
S-1703
K-1701
thru
K-1705
2,4,6-trichlorophenol 0.665
tetrachlorophenol 14.39
pentachlorophenol 4.15
chloro-nitro benzene 14.9
isomers
(incl. 1-chloro-3-nitro
benzene)
1,2-dichloro-3-nitro 6.16
benzene
trichloro-nitro benzene 8.71
isomers
2,4-dichloro-6-nitro 125.5
be nzene amine
2,6-dichloro-4-nitro 23.3
benzeneamine
trichloroaniline isomers 9.78
1,2-benzenedicarboxylic 0.670
acid 2-butoxylethyl butyl
ester
2-chloroanline 0.106
None Detected
chloronitrobenzene isomer 0.296
chlorobenzene 0.0115
2-chlorobenzeneamine 0.228
chloroaniline isomer 0.515
unidentified compound 0.0412
1.33
21 .6
5.13
15.4
9.24
1 2.3
186.5
29.7
17.4
1 .34
0.212
1.53
0.0143
1 .18
0.531
0.0511
1-7
-------
lems were not repeated. Tenax® and charcoal integrated air samples were
collected at nine locations and analyzed at EPA's New England Regional
Laboratory. Additionally, instantaneous samples were collected at 26
locations by direct injection of ambient air into a portable gas chroma-
tograph.
EPA concluded that the results of the March study were va-lid. The
single integrated sample taken immediately dowwind of the main Upjohn
processing area indicated 2 ppb benzene and 51 ppb toluene ambient
concentrations. Instantaneous samples collected in this area showed
that the benzene level varied between 7 and 20 ppb and the toluene
level varied from as low as 20 ppb to over 200 ppb. Upwind levels of
both benzene and toluene were consistent and low. Samples taken on
plant property indicated that the aeration lagoon is a significant
source of volatile organic compound emissions while the settling lagoons
are not. Samples taken at distant downwind locations showed that organic
contaminants were generally lower than in the vicinity of the emission
sources. The highest benzene level encountered was 5 ppb and the highest
toluene level was 20 ppb (both in instantaneous samples).
In order to generate air quality data on and off the plant property
during warm weather, another ambient air sampling study was conducted in
June, 1981 . EPA anticipated that meteorological conditions during the
summer could be different enough from those in November and March to have
an impact on ambient air contaminant levels. The sampling strategy was
identical to that for the March study. Tenax® and/or charcoal integrated
air samples were collected at twelve locations during the period from
June 30 to July 1, 1981 and analyzed at the EPA New England Regional Lab-
oratory. Additionally, instantaneous samples were collected at 46 loca-
tions. Integrated as well as instantaneous samples collected upwind of
the plant indicated benzene and toluene levels were at or below 1 ppb
during, the sampling period. Samples taken downwind of the aeration
lagoon confirmed that it is a measurable source of VOC emissions.
Samples taken in and around the main Upjohn processing area showed that
it is a significant source of VOC emissions. In this area instantaneous
sample benzene levels were as high as 100 ppb and toluene levels as high
as 500 ppb. Samples taken downwind of the processing area revealed the
intermittant nature of the processes. The instantaneous sample benzene
level varied from undetectable to 100 ppb and the toluene level varied
from undetectable to 800 ppb. Integrated samples taken at 7 locations
at a distance downwind of the identified volatile organic emission
sources showed the benzene concentration to range from less than 1 to 4
ppb and the toluene to range from less than 1 to 3 ppb. Instantaneous
levels ranged from undetectable to 10 ppb benzene and from undetectable
to 21 ppb toluene.
1.4.3 Dispersion Modeling Study
To analyze the health effects of potentially hazardous emissions
of volatile organic compounds from the Upjohn facility, ES applied
computer modeling techniques to estimate long-term ground level concen-
trations in the vicinity of the plant. Uie models used the average
emission rates of benzene and toluene as presented in Table 1.1.
1-8
-------
To predict these concentrations, ES utilized an EPA-approved dis-
persion model known as ISCLT, the long-term version of the Industrial
Source Complex model. The ISCLT version uses statistical wind summaries
to calculate annual ground level concentrations on a sector-averaged
basis. It is an advanced Gaussian plume model that requires detailed
information concerning the emission sources, meteorology, and the re-
ceptor layout.
The results of the modeling show that the maximum predicted off-
site annual average benzene concentration was 0.4 ppb. This concentra-
tion was calculated for a receptor on the plant property line to the
northeast in the wetlands of the Quinnipiac River and was primarily at-
tributable to emissions from the aeration lagoon. The contribution from
the remaining vent sources was less than 0.2 ppb regardless of location.
For toluene, the maximum annual average concentrations outside the
plant boundaries was 7 ppb, occurring near the expressway to the north.
In this case, the concentration was due almost entirely to emissions
from stack S-1701, by far the primary emitter of toluene. The contribu-
tion here attributable to other sources was less than 1 ppb. As for the
aeration lagoon, the maximum contribution to concentrations outside the
plant was 2 ppb, occurring on the property line to the northeast.
1.4.4 Health Effects Study
The purpose of this assessment was to evaluate the health effects
caused by the emissions of organic chemicals from the Upjohn Company
facility in North Haven, Connecticut. In addition to the studies per-
formed by EPA, ES, and GCA, SRI International's report on Definition of
Population-at-Risk of Environmental Toxic Pollutant Exposures (Volume
II, Appendices A-C, October, 1980) and the Carcinogen Assessment Group's
(EPA) Final Report on Population at Risk to Ambient Benzene Exposures
were used in performing the evaluation.
The health effects impact assessment translated the results of popula-
tion exposure to ambient concentrations of the emitted compounds into
measures of risk to the exposed population. This required that the ex-
posure be determined and that the health effects of the emitted com-
pounds be identified. The exposure was obtained from two sources. EPA's
instantaneous and 3-hours integrated sampling measurement data were used
to predict short-term exposures. Estimated annual average concentra-
tions generated by ES from dispersion modeling data were used to predict
long-term exposures.
Health effects data on the compounds emitted from the Upjohn facili-
ty were obtained from published information in the scientific literature.
Because ambient sampling by EPA on and surrounding the Upjohn facility
revealed that benzene and toluene were the only VOCs present in measur-
able quantities, the health effects study focused on these two compounds.
Benzene is a recognized human carcinogen associated with leukemia. Tolu-
ene is a narcotic for humans and as such may produce a variety of effects
such as altered psychomotor performance, irritability, disorientation,
and unconsciousness. Health data on chlorobenzene and the condensible
organics were reviewed to determine if any of these compounds were human
carcinogens. None of them are currently recognized carcinogens.
1-9
-------
Based on integrated and instantaneous ambient samples taken by EPA,
it can be concluded that short-term exposures to the measured levels of
toluene and benzene are unlikely to produce non-carcinogenic acute
health effects. Additionally, the annual average concentrations of
toluene outside of Upjohn property as predicted by ES via dispersion
modeling (maximum of 7 ppb) are unlikely to produce health effects.
Several condensible organic compounds were identified in Upjohn vent
emissions, some in concentrations similar to benzene and toluene. Due
to the nature of these compounds, it is unlikely that they would be
present in ambient air far from their emission points. Further, due to
the very low odor thresholds of some of the compounds, they would be
detectable by smell at concentrations far below those which might cause
health effects. {Compounds of this type tend to be skin irritants but
are not particularly toxic systemically.)
Exposure to ambient benzene increases the risk of leukemia. In
the United States as a whole, the expected number of leukemia deaths
per year from 'urban factor1 benzene is 57.87. The 'urban factor1 ex-
posure for the New Haven metropolitan area is much lower than for the
U.S. as a whole. If the U.S. population were exposed to 'urban factor'
benzene at the same level as the population of North Haven, the ex-
pected number of benzene caused leukemia deaths per year would not ex-
ceed 3.73. Therefore, the risk ratio of the residents of North Haven
to the entire U.S. population for 'urban factor' leukemia is 0.06:1.
That is, New Haven has a 94% lower cancer potential from urban benzene
sources than the U.S. average.
In the U.S. as a whole, the expected number of leukemia deaths per
year from general chemical manufacturing benzene exposures is 2.88. Ex-
posure to benzene from the Upjohn facility incrementally increases the
leukemia risk to certain residents of North Haven. As shown by ES, pre-
dicted annual average benzene concentrations outside of the Upjohn faci-
lity range from less than 0.03 to 0.4 parts per billion (ppb). If the
U.S. population were exposed to chemical manufacturing benzene at the
worst case level of 0.4 ppb,, the expected number of benzene caused leu-
kemia deaths per year would be 0.99. Therefore, on a specific source
basis, e.g. chemical manufacturing, the risk ratio of Upjohn emissions-
induced leukemia to the residents of North Haven to the total estimated
U.S. population exposed to benzene from chemical manufacturing leukemia
is 0.34:1. That is, New Haven has a 66% lower cancer potential from
chemical manufacturing than the U.S. average.
As discussed above, if the U.S. population were exposed to 'urban
factor' benzene at the same level as the population of North Haven, the
expected number of benzene caused leukemia deaths per year would not
exceed 3.73. Further, as discussed above, if the U.S. population were
exposed to the worst case Upjohn generated benzene level of 0.4 ppb,
the expected number of leukemia deaths per year would be 0.99. Therefore,
those residents of North Haven exposed to 0.4 ppb benzene from Upjohn
incur a leukemia risk 1.27 ( 3.73 + 0.99 ) times or 27% greater than
3.73
those exposed to 'urban factors' only. This factor decreases to 1.03
(3% greater) for resident exposed to 0.05 ppb of benzene.
1-10
-------
The risk of benzene induced leukemia in residents of North Haven
increases if a significant part of the day is spent in a high population
density central city location (e.g. working in New York City).
Finally, it must be recognized that there are uncertainties asso-
ciated with the data and procedures used to estimate the benzene cancer
risks presented above. One of the uncertainties is the inability to
precisely determine annual benzene emission rate. Further uncertainty
is introduced with the inherent limitations of the dispersion model to
perfectly describe the atmospheric dispersion of the emitted benzene.
There are similiar limitations associated with the linear non-threshold
model. There are also uncertainties concerning possible additive ef-
fects of multiple sources of benzene, synergistic or antagonistic health
effects, and heightened susceptibilities of some population groups.
Because of these uncertainties, the risk numbers presented above should
only be used as a rough estimate of the benzene induced cancer risk;
i.e. they should not be interpreted to represent the absolute magnitude
of risk.
1-11
-------
SECTION 2
PROCESS VENT/LAGOON SAMPLING STUDY
2 .1 INTRODUCTION
2.1.1 Background
This chapter discusses the process vent/lagoon sampling study. The
objectives of the sampling study were: (1) to identify sources within
the Upjohn plant that may emit significant quantitites of potentially
hazardous organic compounds to the air, and (2) to sample those sources
and quantify their emissions. Prior to the source testing, EPA and
ES established four criteria to insure that the results obtained by the
study would be useful in meeting the project objectives. The first of
these was that all significant emission sources be identified and
quantified. A second testing criterion was that vent sampling be
conducted only when each associated process was operated according to
reported procedures and when chemicals reported by Upjohn were actually
employed. The third criterion established in advance of the actual
field activities was that, because many pieces of process equipment were
exhausted to each emission point, a representative sampling of each point
would include emissions generated by each piece of process equipment.
This required sampling long enough so that each piece of process equip-
ment was operated at least once during each sampling period. The time
taken to carry out all associated process elements is referred to as the
"sampling cycle" for the particular emission point. A fourth criterion
established was to take a second sample over a second cycle where appli-
cable .
The sampling program was conducted by ES under the direction of
personnel from EPA. During the sampling program, personnel from E&E,
EPA, and ES monitored the plant operations to verify that each asso-
ciated process was operated according to reported procedures and that
chemicals reported by Upjohn were in fact being employed or produced in
each process. This task was also necessary to optimize the application
of testing equipment through timely selection of emission points for
sampling.
The samples collected during the effort were analyzed by the Lab-
oratory Analysis Department of GCA/Technology Division. These results
were then used by E&E to calculate the emission rates of volatile, semi-
volatile, and condensible organic compounds from the identified emission
sources.
2-1
-------
2.1.2 Summary
During the week of November 19, 1980, a comprehensive sampling pro-
gram was conducted to identify and quantify organic air emissions from
the Upjohn facility. This program involved E&E, EPA, and ES personnel.
It was initiated by a screening exercise conducted by EPA to identify
the compounds and emission points to be investigated. This task was
supplemented by a review of process data and plant layouts by E&E,
EPA, and ES. Twenty process vents and the aeration and settling lagoons
were identified as potentially significant emission points. Benzene,
toluene, and chlorobenzene were identified as the VOCs being emitted
in the greatest quantities.
ES chose seventeen of the vents for VOC sampling during the program.
Three of the twenty vents were not selected because they were identical
to vents included in the seventeen chosen. In order to study the entire
organic emission situation and to search for odor sources, EPA also
selected five vents to be sampled for condensible organic compounds.
Hie aeration and settling lagoons were chosen to be sampled for both
volatile and semi-volatile organic compounds.
ES collected VOC samples from the 17 process vents in Tedlar® gas
sampling bags. The samples were withdrawn from the vents using lung-
type samplers to eliminate the loss or contamination of VOC's by the
internal parts of the pump. Proportional sampling was necessary as the
flow from many of the vents was intermittent because the processes were
batch type. Flow was measured with a vane anemometer. The processes
were monitored throughout the sampling period by E&E, EPA, and ES per-
sonnel. This was done to ensure that the processes were operating
according to normal procedures and to optimize the use of sampling
equipment and personnel.
The collected VOC samples were transported to the GCA laboratory
in Bedford, Massachusetts for analysis. Gas chromatography was used
for both compound identification and concentration measurements. Con-
firmation of some of the more difficult compound identifications was
made by gas chromatography/mass spectroscopy (GC/MS). Results of these
analyses confirmed that benzene, toluene, and chlorobenzene were the
major VOC's emitted from the vents.
Using the analytical results and the flow rates measured by ES,
E&E calculated the emission rates of these three compounds from the
process vents. Results of these calculations are summarized in Table
1.1. It can be seen from Table 1.1 that, of the three compounds, tol-
uene is emitted in the greatest quantity. Nearly all of the 7176 g/hr
of toluene (98%) is emitted from one vent, s-1701. A total of 175 g/hr
of benzene is emitted from the vents. The major sources of the benzene
are vents S-301, S-206, S-205, S-213, and S-202. Relative to benzene
and toluene, only a small quantity (4 g/hr) of chlorobenzene is emitted
and 70% of this is released through vent S-214.
Upjohn constructed a small wind tunnel for use in quantifying the
emissions from the lagoons. This device was an open-bottom box with a
fan at one end and a duct in the other. It was floated on the lagoons
2-2
-------
to collect the samples. The fan blew air across the top of the enclosed
lagoon and out the duct at the other end. Tenax tubes were inserted
into the duct to collect the organic compounds volatilized by the move-
ment of the air across the surface. Flow through the duct was measured
with a vane anemometer. The fan inlet was partially occuded to simulate
different wind speeds.
The organic compounds collected on the Tenax® tubes were solvent
extracted and concentrated by evaporation. However, a study of the
analytical data indicated that the volatile organic emissions could not
be adequately assessed using the analytical protocols chosen for this
program. Thus, the VOC emissions from the lagoons could not be quantified
by this technique (see Section 4.2 for an alternative approach).
ES collected the condensible organic compounds using modified ver-
sions of EPA Method 5 and 6 trains. T5ie modification entailed using an
organic compound collection media (0.1 N KOH) in the impingers instead
of the media that is used in the standard procedures. After the samples
were collected, the collection media were transferred to glass containers
and sent to GCA for analysis. The Method 5 filters were desiccated and
weighed at the ES laboratory and then transported to the GCA laboratory
for extraction and condensible organic analysis. GC/MS was used for ana-
lyzing the impinger catches and the samples extracted from the filters.
Results of these analysis were used with the flow measurements obtained
during sampling to calculate the emission rates of the condensibles.
These rates are summarized in Table 1.2. The major condensible emitted
from the Upjohn facility is 2, 4-dichloro 6-nitro benzeneamine (DCNA).
The emission rate of this compound was 125.5 g/hr (from S-303).
2.1.3 Conclusions
The following conclusions are drawn from the results of process
vent/lagoon sampling and analysis efforts:
1. The Upjohn facility process vents are a source of VOCs in
air. An average VOC emission rate of over 7300 g/hr was cal-
culated from the data collected during this study.
2. Benzene, toluene, and to a much lesser extent, chlorobenzene
are the VOC's emitted in the greatest quantities from process
vents. The plant-wide average emission rates are 175, 7176,
and 4 g/hr for benzene, toluene, and chlorobenzene respectively.
3. Because some of the processes are batch rather than continuous,
VCX: emissions from some of the vents are intermittent. Con-
sequently, the maximum VOC emission rate could be greater than
that given in (2) above.
4. Vents identified as emission points for benzene include S-301,
S-206, S-205, S-213, and S-202.
5. Approximately 98% of the emitted toluene is released through
vent S-1701.
2-3
-------
6. Vent S-214 accounts for nearly 70% of the chlorobenzene emis-
sions.
7. Hie processes were operated according to normal procedures dur-
ing the sampling.
8. The Upjohn facility also emits a significant quantity of con-
densible organic compounds through the process vents sampled.
DCNA is the condensible compound released in the greatest
quantity.
9. The analysis method used for the lagoon samples was unsatisfac-
tory. (As a result, an alternative approach was chosen to quan-
tify the lagoon emissions. The alternative approach employed
the technique of Shen, and, was confirmed by using air quality
monitoring data generated by EPA as described in Section 4.)
2.2 EPA SCREENING STUDY
In order to identify major emission sources, EPA conducted a compre-
hensive screening study in September 1980. All process vents in the
Upjohn plant were sampled. A portable photo-ionization detector was
used to screen 38 vents for volatile organic compounds. Grab-samples
were taken from all vents in which the instrument detected significant
concentrations of volatile organic compounds. Accurate quantification
was not attempted since the purpose of this initial sampling was simply
to identify the significant emission sources. The grab sampling was per-
formed by accumulating the samples in Tedlar® bags by means of a lung-
type sampler. Analysis was performed by EPA at the New England Regional
Laboratory using GC and GC/MS techniques. Uie GC was equipped with a
flame ionization detector (FID). Benzene, toluene, and chlorobenzene
were identified as the VOCs being emitted in the greated quantities.
Twenty emission points were identified as potentially significant
sources of VOC by the screening study. These are listed in Table 2.1.
Vents T-393 and T-394 were considered as a single source because they
are used alternately. The same reasoning was applied to T-395 and T-396.
In addition to the process vents listed in Table 2.1, the aeration lagoon
and settling ponds in the waste treatment facility were also identified
as potentially significant sources of volatile organic compound emissions.
Although it was not in the original scope of work, EPA requested ES
to attempt to identify and quantify emissions of condensible organic com-
pounds from sources within the Upjohn facility. Thus, as part of the
screening study several vents were investigated for condensibles. Pro-
cess, product, and raw material information obtained from Upjohn were
reviewed by EPA and ES to determine which vents would be candidates for
the qualitative screenings. Three vents, S-303, S-1701 and S-1703 were
selected for condensible organic compound sampling based on this review.
The sampling for condensible organics consisted of drawing sample gas
through 0.1 N potassium hydroxide-filled impingers in an ice bath. The
caustic solution was designed to retard reactions among the organics col-
lected and the low temperature was designed to promote efficient collec-
2-4
-------
TABLE 2.1
PROCESS VENTS SELECTED FOR EMISSION TESTING
BASED ON EPA SCREENING STUDY3
S-202
S-203
S-205
S-206
S-213
S-214
S-301
S-303
K-316
11. T-395/396
2.
3.
4.
5.
6.
7.
8.
9.
10.
T-393/394
12. S-1701
13. S-1703
14. K-1701
15. K-1702
16. K-1703
17. K-1704
18. K-1705
19. K-1706
20. T-1718
a Generally, the source identification scheme is as
follows: "K" represents a reaction kettle, "S",
a scrubber, "T", a tank vent. The leading "2",
"3", or "17" represents the building number, and
the remaining digits, the source number.
2-5
-------
tion. These samples were analyzed using GC/MS techniques at GCA's labora-
tory in Bedford, Massachusetts. Two additional sources, vents K-1702 and
K-1705, were selected for sampling by the EPA Project Officer during the
course of the field tests. These two vents were sampled with EPA Method
6 equipment. These samples were also analyzed by GCA using GC/MS tech-
niques. Basicially, the compounds detected were the raw materials
and/or products of the associated process.
2.3 PROCESS MONITORING
2.3.1 Process Analysis
Prior to source testing, the Upjohn Company provided EPA with con-
fidential process information (consisting mainly of block diagrams and
some process descriptions) for the four processes associated with emis-
sion points selected for sampling. The selection of the emission points
was based on screening data generated by EPA. Hie process information
was employed to create preliminary flow diagrams which focused on the
vent systems associated with the four preselected processes.
At the plant, immediately prior to setting up sampling equipment,
the Upjohn engineer responsible for each process was interviewed in
order to finalize the flow diagram. Subsequently, a detailed tour of
the processing equipment was made in order to locate potential process-
monitoring points and to verify the accuracy of the flow diagram.
Each of the four processes was then analyzed by chemical engineer-
ing principals for the elements which would result in the discharge of
vapors, in particular, organic- vapors through uncontrolled process
vents and vents controlled by air pollution control equipment. Examples
of process elements which result in stack discharges are:
1. Raw material additions to vessels (solids, liquids, and gases)
2. Batch transfers between vessels;
3. Gas purges of reactors; and
4. Distillations.
A total of ninety such elements were identified in the four pro-
cesses of concern. A through study of these process elements helped
determine the points of discharge, the sources of each discharge, the
duration of each discharge, and approximate stack exit flow rates.
Further, the process analyses proved invaluable when making field de-
cisions as to the type of sampling equipment, its placement, and the
schedule of testing.
2.3.2 Process Monitoring in the Field
During air sampling at the Upjohn facility, the four processes of
concern were monitored by an EPA or EPA designated engineer.
2-6
-------
One objective of the field monitoring effort was to optimize the
application of testing equipment through timely selection of emission
points for sampling while meeting the criteria for sampling which were
established in advance of actual field activities.
A major element of the monitoring effort was determining when a
representative sample had been taken. Sampling was initiated pn. each
emission point at the convenience of the sampler and process monitor.
From that point, sampling continued until at least one of each process
element from associated process equipment had been observed by the
monitor. Typically, for VOC sampling, this required the filling of a
number of gas sample bags over many hours. The criterion established
to take consecutive duplicate samples from each emission point required
that the monitor alert the sampling crew at the end of the first sam-
pling cycle to stop the sampling activity and to prepare for sampling
during the second, or duplicate cycle. At the end of the second cycle,
the monitor instructed the sampling crew to stop testing and to move
the sampling equipment to another emission point. (Due to production
scheduling, consecutive duplicate samples could not be generated from
one of the four processes.)
A second objective of the field monitoring effort was to verify
that, during sampling, each associated process was operated according
to reported procedures and that chemicals reported by Upjohn were, in
fact, being employed and/or produced in each process.
Verification of chemical use had to be accomplished mainly by in-
ference to avoid contact with hazardous materials. Based on a knowledge
of the physical properties of the applicable chemicals, however, enough
evidence was accumulated to reasonably ascertain that the materials, as
reported by Upjohn, were being fed to, generated in, and/or discharged
from processes at the North Haven facility. For example, phase separa-
tions were observed which are indicative of organic/aqueous mixtures;
acrid odors were detected from uncovered dilute acid tanks; samples were
drawn by operators to show organic material; and, in most cases, various
liquids and powders were observed during packaging.
In order to verify that the processes were being operated according
to reported procedures, a great number of observations were made and re-
corded in the field.
Examples of the types of observations made are:
1. Visually observed liquid in piping site glass between vessels
indicating that raw material was being introduced to process.
2. Observed decrease in weigh scale or level indicator reading
indicating that raw material was being introduced to process.
3. Observed rotameter or other flow indicator readout signifying
that liquid/gas was being introduced to process.
4. Observed increase in totalizer reading indicating that liquid
was being introduced to process.
2-7
-------
5. Visually observed level change in vessel site glass indicating
that liquid was flowing into or out of vessel.
6. Visually observed reaction batch through reactor observation
port indicating that process was being conducted.
7. Observed elevated remote temperature readout indicating that
process was being conducted.
8. Visually observed condensate in condenser return line indicat-
ing that distillation was being conducted.
9. Visually observed product drumming.
10. Visually observed manual charge of solid reactant from drums.
11. Felt drop in temperature of reactant feed pipe indicating that
expanding gas was introduced to process.
12. Physically heard material being introduced to vessel.
13. Visually observed agitator operating.
14. Visually observed product moving cn rotary vacuum filter.
15. Observed remote pressure readout indicating that process was
ope rating.
16. Observed amperage readout indicated that process was
operating.
17. Observed remote pump shaft speed readout indicating that pro-
cess was operating.
All of the process observations made on a 24-hour continuous basis
indicate that the Upjohn Company was producing product as indicated by
the process descriptions supplied to EPA.
2.3.3 Process Monitoring from Records
In addition to monitoring the preselected process in the field, an-
other objective of the monitoring activity wa^s to determine whether, dur-
ing source testing, every applicable process was operated according to re-
ported procedures in a manner consistent with past operations. To EPA's
benefit, it was discovered that Upjohn plant operators used "batch sheets"
to guide them through the manufacturer of each chemical product. These
batch sheets are "cock books" divided into discrete steps with space for
logging the starting and finishing times of each step and other pertinent
physical data such as the volume or weight of material added, the tempera-
ture, pressure, and pH of the product batch, etc.
Following the source testing week, visits were made to the Upjohn
facility to review batch sheets. Since EPA could not provide a moni-
tor at all processes operating concurrently at the Upjohn facility, a
2-8
-------
check on the processes not specifically monitored was made by monitoring
the completion of the batch sheets for those batches in process during
the source testing. Further, by comparing the sheets from batches pro-
cessed throughout the source testing period, a determination was rea-
sonably made as to whether there was a consistency to processing proce-
dures. Finally, by comparing batch sheet data from the source testing
week with historical batch sheet data for the same product, a deter-
mination was made as to whether vent emissions during the testing week
were comparable to, and representative of historical emissions (this
comparison must be used carefully since no information was obtained on
the installation history of stack gas cleaning equipment).
The batch sheet data indicates that all four processes of concern
were operated continuously on a 24-hour basis throughout the EPA source
testing period with only intermittant, short duration, shutdown periods.
Further, a close examination of the data shows that the batches processed
during the period were done so on a consistent basis. Finally, the data
shows that Upjohn had not significantly modified its processes during
the 1980 operating year.
2.4 PROCEDURES FOR SAMPLING VENTS AND LAGOONS
2.4.1 VOC Sampling Procedure
Based on a review of the screening study results, a total of 17
vents were sampled for emission rate quantification of 20 process vents.
The vents sampled were S-202, S-203, S-205, S-206, S-213, S-214, S-301,
S-303, K-316, T-393, T-396, S-1701, K-1702, S-1703, K-1705, K-1706, and
T-1718.
The samples were accumulated in Tedlar® bags by means of lung-type
samplers. That is, a Tedlar® bag was placed into an air-tight, rigid-
walled container. The inlet to the Tedlar® bag was connected to the
source to be sampled by a short length of Teflon® tubing. Air was then
pumped out of the rigid wall container at a known rate. This action
drew sample gas into the Tedlar® bag. The rate of sample accumulation
in the bag was measured with a rotameter.
Many of the Upjohn processes are batch rather than continuous. The
batch processes were sampled over one or more complete process cycles.
Continuous processes were sampled for a minimum of 2 hours. Multiple
sequential samples were taken at each vent tested. Early in the sampling
program it was discovered that the flow in many of the vents designated
for sampling varied significantly with time. This meant that the original
plan to control the sampling rate of the lung-type samplers with critical
orifices was inappropriate. It was necessary to utilize needle valves so
that the sample flow rate could be adjusted as the vent flow rate changed.
Upjohn's vents fall into three categories when defined in terms of
their exhaust flow characteristics:
1. Large diameter ducts with continuous flow;
2. Large diameter ducts with non-continuous flow; and,
3. Small diameter ducts with non-continuous flow.
2-9
-------
For purposes of this discussion, the division between large and small
diameter is 12 inches, the minimum diameter duct in which flow can be
measured accurately with an S-type pitot tube. Total gas flow in the
large diameter ducts with continuous flow was measured by the usual
pitot tube traverse methods (EPA Methods 1-4). The flow in the small
diameter ducts was measured with a vane anemometer that was calibrated
against a standard pitot tube in the ES wind tunnel. The anemometer
itself consisted of a low-friction propeller and a counter that counts
the number of times the propeller blade rotates. Each rotation of the
propeller is proportional to a known volume of gas passing through the
anemometer.
The small diameter ducts with non-continuous flow are, in general,
those having flame arrestors. All of the scrubber vents in Building 2
are also in this category. The flame arrestors are installed on low
pressure vessels and have pressure relief valves that only allow gases
to escape when a preset system pressure is reached. Escape of gases
through these vents is sporadic. ES fashioned hoods from Tedlar® bags
over these vents which collected the escaping gases and directed them
through an anemometer. The anemometer remained in place throughout the
entire period of sample accumulation.
Air flowed through the scrubber vents in Building 2 only during
the times that liquid was being transferred into a tank. During these
times the vapor in the tank was forced out through the scrubber. These
transfers typically were 15 to 30 minutes in duration. During the rest
of the cycle, the time during which reactions occurred, there was little
or no flow in these vents. ES recommended (while in the field) that
samples be accumulated from these vents only during those times when
there was an exhaust flow from them. The decision was made by EPA per-
sonnel to sample them continuously throughout the process cycle and to
record the total exhaust volume during the periods of flow.
The collected bag samples were transported each day by EPA person-
nel frcm the plant site to the GCA laboratory. Analyses of the bag
samples were conducted by GCA laboratories using GC/FID techniques as
is discussed in Section 2.5.2.
The aeration lagoon and settling pond, which are part of Upjohn1s
wastewater treatment facility, were also sampled. The sampling proce-
dure selected for these sources was different because of the nature of
these sources. The aeration lagoon receives liquid wastes from all parts
of the chemical plant. Therefore, it may contain a host of organic
compounds. The lagoon is a biological treatment device and is subjected
to vigorous aeration in order to provide oxygen to the microbes that
degrade the organic wastes. The portion of the air that is not consumed
by the microbes leaves the lagoon surface in the form of small bubbles.
Uiese bubbles probably strip some of the volatile organic material
from the lagoon water. The settling pond is not aerated. Thus, the
evolution of organic compounds from it is a function of the vapor pressure
of the compound, the temperature of the water, and the wind speed.
2-10
-------
To sample the lagoon and pond, it was decided to construct a minia-
ture wind tunnel to simulate wind speeds of from 3 to 10 mph across the
surface of both ponds. The wind tunnel consisted of an open-bottom
box placed over a one-by-six foot area of water surface with a forced
draft fan, an anemometer, and Tenax® tube sampler placed in the six-inch
ID exhaust port.
The lagoon and pond organic samples were collected on a porous
polymer sorbent resin (Tenax-GC®) rather than in plastic bags. Use of
this method eliminated the possible loss of marginally volatile compon-
ents on the walls of the bag, and also concentrated the samples for
easier detection. Two tubes were placed in series during the sampling
to assure complete recovery of the VOC. These samples were transported
to the GCA laboratory by EPA personnel on the day of their collection.
Two samples were taken at each location.
2.4.2 Condensible Organics Sampling
As was mentioned previously, the EPA Region I Laboratory had col-
lected condensible organic compound samples from several vents at the
Upjohn Company. The purpose of this effort was to provide some prelim-
inary information on the relative importance of the various vents and
the relative importance of condensible organic compound emissions. The
results of these screening tests indicated that some samples of conden-
sible organic compounds should be acquired and quantitatively analyzed
in order to understand the entire organic emission situation and to
search for odor sources.
Three sources, S-1701, S-1703, and S-303, were sampled for particu-
late emissions by a modified EPA Method 5 procedure. These tests were
performed in duplicate. In addition to these sources the EPA Project
Manager decided during the field sampling that two additional sources
(K-1702 and K-1705) should be sampled for condensible organics. Single
samples were collected at these two sources using EPA Method 6 type
equipment.
Samples of the gases collected for condensible organics analyses
were drawn through two impingers containing 0.1 N potassium hydroxide
followed by an impinger containing ethylene glycol. This final impinger
served as a backup to insure that no condensible organic escaped the
impinger train. The impingers were immersed in an ice bath. The low
temperature retarded reactions among the various components. These
two impinger solutions were transferred to separate borosilicate glass
containers for storage and transportation to the GCA laboratory. The
Method 5 filters were desiccated and weighed at the ES laboratory and
then transported to the GCA laboratory for extraction and condensible
organic analysis. The probes used in the condensible organic sampling
were rinsed with methylene chloride after each run. This rinse was
transferred to a glass container labeled "front half rinse". After
the KOH solution was transferred to sample bottles, the impingers were
washed with methylene chloride. This wash was transferred to a glass
sample bottle labeled "back half rinse". These solutions and the
rinses were sent to GCA for analysis.
2-11
-------
2.4.3 Quality Control
The importance of quality control cannot be over-emphasized for a
study such as this. Engineering-Science has a thorough quality control
program that has been carefully developed over the years. The equipment
used on this project was calibrated prior to use, and the results of
those calibrations are permanently recorded in special calibration data
notebooks. Specific instruments that were calibrated included the meter
boxes, nozzles, pitot tubes, and temperature devices used for the sam-
pling, and the anemometers used for flow measurement in the small vents.
Calibration of the equipment followed established EPA procedures. The
anemometers were calibrated in a wind tunnel against a standard pitot
tube.
Special care was used to insure that sample contamination was mini-
mized. The Tedlar® bags were purchased especially for this project and
were used only once. The Tenax-GC® sampling tubes were prepared fresh
for the study by GCA. The Tenax® was cleaned in a special two-step
process consisting first of Soxhlet extraction for 24 hours each with
methanol and hexane. The Tenax® was then dried and packed in the tubes
prior to the second clean-up which consisted of thermally desorbing the
tubes under helium for two 2-hour periods.
Other quality control procedures that were employed involved dupli-
cate analysis of three bag samples from the field sampling effort. The
bag samples were first analyzed by GCA, and were then given to the EPA
Laboratory for their analysis. Aliquots of the Tenax® tube extracts
from one run were given to the EPA Laboratory for independent analysis.
Similarly, aliquots from one Method 5 filter extract and one impinger
extract were submitted to the EPA Laboratory for analysis. These inde-
pendent analyses of split samples provided information on analytical
accuracy.
2.4.4 Sample Handing/Chain-of-Custody
The collected gas samples were transported each day by EPA personnel
from the plant site to the GCA laboratory. Prior to giving the samples
to the EPA personnel, ES labeled them to identify the process source
and date/time of sampling. This information was recorded on the field
data sheets. Each EPA Method 5 filter used for the condensible organics
sampling was assigned a unique identification number by ES prior to pre-
weighing at the McLean, Virginia laboratory. The identification numbers
were stamped in ink on the filter. Each filter was then placed in a
clean Pyrex® petri dish that was marked with the filter identification
number. The individual filter/petri dish sets were then weighed. These
numbers and the weights were recorded in a log book kept at the ES lab-
oratory. After the samples were collected, these filters were placed
in their respective tared petri dishes and shipped to the ES laboratory
for weighing. The filter identification numbers were recorded on EPA
Method 5 field data sheets as well as in the ES laboratory log book.
After weighing, the filters were returned to their respective petri
dishes and shipped to the GCA laboratory for extraction and condensi-
ble organic analysis. The impinger catches were transferred to boro-
silicate glass containers at the test site for storage and later trans-
2-12
-------
port to the GCA laboratory. The glass containers were labeled by ES
to identify the source, date, time, and run number.
Upon receipt of the samples, GCA logged them into their Master Log
Book and individual GCA Control Numbers were assigned to each. These
unique identification numbers were affixed to the respective sample con-
tainers and used during all further handling and analytical procedures
to ensure positive traceability. Sample custody procedures in the GCA
laboratory are maintained through the use of Custody Notebook and Sample
Custody Transfer records. At the time of receipt, a page for each sam-
ple was entered sequentially by GCA Control Number in the Custody Note-
book, and the samples were then placed in the locked Sample Bank for
storage at 4°C until the time of analysis. Subsequent handling of the
samples was documented by the recording of signatures and dates in the
Custody Notebook. In addition, the transfer of samples between analysts
within the laboratory was recorded on Sample Custody Transfer forms.
Table 2.2 is a listing of the 68 integrated Tedlar bag samples from the
17 process vents with their corresponding GCA control numbers. These
samples were received by the GCA/Technology Division Sample Bank in 8
separate submittals between November 18 and 24, 1980. Routine inspec-
tion on receipt revealed all samples to be in good condition and labeled
distinguishing the process line, date, and time of sampling.
Fourteen Tenax® tube samples were received by the GCA Sample Bank
in two separate submittals: the first four tubes were received on Novem-
ber 21, 1980, and the remaining 10 on November 22, 1980. A list of these
samples, with the corresponding GCA Control Numbers and descriptions is
given in Table 2.3.
Table 2.4 and 2.5 are the listings of the Method 5 train samples
received by GCA. Included in the tables are the assigned GCA Control
Numbers and submitted sample descriptions/identifications. ES submitted
these samples to GCA on three separate dates. All liquid samples (im-
pinger catches and train rinses) were received on December 9, 1980; the
related particulate filters were received on December 17, 1980; and the
appropriate filter blanks were received on March 5, 1981.
2.5 ANALYTICAL PROCEDURES AND RESULTS FOR PROCESS VENTS/LAGOONS SAMPLES
2.5.1 Introduction
In order to characterize and measure organic emissions from the
Upjohn facility, samples were taken from a variety of point sources and
analyzed at GCA/Technology Division in Bedford Massachusetts. As des-
cribed in Section 2.4, there were three different types of samples:
o Tedlar® bag gas samples taken from process vents for quantifica-
tion of volatile organics; e.g., benzene, toluene, and chloro-
benzene.
o Method 5 train samples collected for semivolatile/nonvolatile
organic compound identification and quantification; e.g., ani-
line and phenol derivatives.
2-13
-------
TABLE 2.2
LIST OF GAS SAMPLES
GCA
Sample Identification
Date
Control
Process
Date
Sampling
Received
No.
Line
Sampled
Interval
11/18/80
10434
S-214
11/18/80
0320-0650
10435
S-206
11/18/80
0130-0430
10436
S-214
11/17-18/80
2110-0015
10437
S-214
11/18/80
0055-0300
10438
S-206
11/18/80
0505-0725
10439
S-214
11 /1 7/80
1717-2030
10440
S-206
11/17-18/80
2135-0045
10441
S-214
11 /1 7/80
1419-1717
10442
S-206
11/18/80
0735-0950
10443
S-214
11/18/80
0715-1020
10444
S-206
11/17/80
1743-2115
11/19/80
10445
S-214
11/18-19/80
1900-0020
10446
S-214
11/19/80
0510-1015
10447
S-214
11/19/80
0040-0500
10448
S-214
11/18/80
1030-1325
10449
S-206
11/18/80
1320-1915
10450
S-214
11/18/80
1330-1900
10451
S-206
11/18/80
1000-1315
11/20/80
10556
S-202
11/19-20/80
1945-0035
10557
S-205
11/19-20/80
1045-0300
10558
S-206
11/19/80
1615-2358
10559
S-213
11/19/80
1500-2030
10560
S-206
11/19/80
1045-1615
11/20/80
10561
S-303
11/20/80
1438-1544
10563
S-213
11/19/80
2115-2250
10564
S-213
11/19-20/80
2256-0600
10565
S-303
11/19/80
1830-1939
11/21/80
10570
S-213
11/19/80
1045-1450
10571
S-202
11/19/80
1347-1945
10572
S-202
11/20/80
0046-1030
10573
S-203
11/20/80
1155-2120
10574
S-203
11/21/80
0550-1100
10575
S-206
11/20/80
0315-1345
10576
S-203
11/20-21/80
2135-0540
10577
K-316
11/20/80
1600-1800
10578
S-205
11/20/80
1105-1515
10579
S-206
11/20/80
1409-1515
10580
T-393
11/20/80
2210-2218
10581
T-393
11/20/80
2225-2300
10582
K-316
11/20/80
0935-1140
2-14
-------
Table 2.2 - Continued
GCA
Sample Identification
Date
Control
Process
Date
Sampling c
Received
No.
Line
Sampled
¦Interval
1 0583
S-206
11/20/80
0006-0304
10584
S-21 3
11/20/80
0600-1050
10585
S-205
1 1/20/80
0305-1 050
10586
S-202
11/20/80
1030-1525
11/22/80
10587
S-301
11/20-21/80
2110-0500
10588
S-301
11/21/80
0520-0910
10589
S-301
1 1/21/80
091 5-1 139
10590
T-396
1 1/21/80
1130-1315
10591
T-396
11/21/80
1315-1500
10592
S-1703
11/21/80
1614-1735
10593
S-301
11/21/80
1139-2205
10594
S-1701
11/21/80
1740-2235
10595
K-1705
1 1/21/80
1845-2345
10596
S-301
11/21-22/80
2205-0110
10597
T-1718
1 1/22/80
0030-0406
10598
S-1701
11/21-22/80
2245-0445
10599
K-1705
11/21-22/80
2355-0740
10600
K-1702
11/22/80
0055-0750
1 0601
K-1706
11/22/80
0800-0825
10602
S-1703-
¦R2b 11/22/80
1000-1130
1 0603
K-1705
1 1/22/80
0755-1 305
10604
K-1702
11/22/80
0800-1535
10605
S-1701
11/22/80
0445-1 415
10606
K-1706
11/22/80
1705-1717
10607
T-1718
11/22/80
1845-1955
10608
K-1705
11/22/80
1320-2130
11/24/80
10618
S-1701
11/22/80
1415-2140
10619
K-1702
11/22/80
1540-2234
a Time as recorded on a twenty-four hour clock
b R2 is run No. 2
2-15
-------
TABLE 2.3
LIST OF TENAX® TUBES
GCA Control No. Tenax® Tube No.a Sample Description
1 061 4
GCA-12-101
Aeration
Pond—Sample
2,
Tube
1
10615
GCA-12-107
Aeration
Pond—Sample
1 ,
Tube
2
1061 6
GCA-12-1 08
Aeration
Pond—Sample
2,
Tube
2
10617
GCA-12-109
Aeration
Pond—Sample
1,
Tube
1
10620
GCA-12-84
Blank
10621
GCA-1 2-85
Blank
10622
GCA-12-88
Blank
10623
GCA-1 2-90
Blank
10624
GCA-1 2-1 03
Blank
10625
GCA-1 2-110
Blank
10626
GCA-12-89
Settling
Pond—Sample
1/
Tube
2
10627
GCA-1 2-102
Settling
Pond—Sample
2,
Tube
1
10628
GCA-12-1 04
Settling
Pond—Sample
2/
Tube
2
10629
GCA-1 2-106
Settling
Pond—Sample
1 ,
Tube
1
a Tenax® tube identification provided by GCA prior to sampling.
2-16
-------
TABLE 2.4
LIST OF METHOD 5 FILTERS
Date
Received
GCA
Control No.
Submitted
Sample
Identification
Sample Sourcea
Stream Run
12/17/80
03/05/80
11090
11091
11092
11093
11094
11095
11096
11786
11787
11788
1 1789
R696
R697
R698
R702
R703
R709
R712
R865
R866
1b
S-303
S-1701
S-1703
S-1701
S-1701
S-303
S-1703
Blank
Blank
Blank
Blank
1
1
1
3
2
2
2
a Stream and run numbers for the seven filters received on 12/17/80
were forwarded from Engineering-Science on 01/23/81.
k These filters had received no field preparation, i.e., tare weights
and labels. They were submitted for use as blanks.
2-17
-------
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
TABLE 2.5
LIST OF IMPINGER AND RINSE SAMPLES
Submitted Sample Information5
Stream Run Bottle
Sample Type*3
Date
Reagent Sampled
CH2CI2 (methylene 11/19/80
chloride)
0.1N KOH (potassium 11/19/80
hydroxide)
0.1N KOH 11/19/80
Ethylene glycol 11/19/80
CH2C12 11/19/80
CH2C12 11/19/80
0.1N KOH 11/19/80
Ethylene glycol 11/19/80
ch2ci2
0.1N KOH
Ethylene glycol —
CH2C12 11/22/80
CH2C12 11/22/80
0.1N KOH 11/22/80
Ethylene glycol 11/22/80
CH2C12 11/2 2/80
S-303
S-303
S-303
S-303
S-303
S-303
S-303
S-303
S-1701
S-1701
S-1701
S-1701
S-1701
1
1
1
2
2
2
2
1
1
1
1
2
Front half rinse
Back half rinse
1st and 2nd impinger catch (2)
3rd impinger catch
Front half rinse
Back half rinse
1st and 2nd impinger catch (2)
3rd impinger wash
CH2C12 blank
KOH blank
Ethylene glycol blank
Front half rinse
Back half rinse
Back half (2)
3rd impinger
Front half rinse
-------
GCA
itrol
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
Table 2.5 - Continued
Submitted Sample Information3
Stream Run Bottle
Sample Type*3
Reagent
S-1701 2 A Impinger
S-1701 2 B Impinger
S-1701 3 C Front half rinse
S-1701 3 A Impinger
S-1701 3 B Impinger
K-1702 1 C Front half rinse
K-1702 1 A Impinger
K-1702 1 B Impinger
S-1703 1 — Front half rinse
S-1703 1 — Back half rinse (2)
S-1703 1 — 3rd impinger
S-1703 2 — Front half rinse
S-1703 2 — Back half rinse
S-1703 2 — 1st and 2nd impingers, BH
rinse (2)
S-1703 2 — 3rd impinger rinse
K-1705 1 C Front half rinse
0.1N KOH
Ethylene glycol
ch2ci2
0*IN OH
Ethylene glycol
ch2ci2
0.1N KOH
Ethylene glycol
ch2ci2
0.1N KOH
Ethylene glycol
ch2ci2
ch2ci2
0.1N KOH
Ethylene glycol
ch2ci2
-------
Table 2.5 - Continued
Submitted Sample Information3
GCA Date
Control No. Stream Run Bottle Sample Typek Reagent Sampled
11211 K—1705 1 A Impinger 0.1N KOH 11/23/80
11212 K-1705 1 B Impinger Ethylene glycol 11/23/80
11213 S-1703 1 — Back half rinse CH2Cl2 11/21/80
11214 - — CH2C12 blank CH2Cl2
11215 - — 0.1N KOH blank 0.1N KOH
11216 - — Ethylene glycol blank Ethylene glycol
a Information provided by ES.
k (2) indicates two containers received.
-------
o Tenax® tube samples taken to identify and quantify fugitive
organic emissions from the aeration lagoon and settling pond.
The Tedlar® bag samples were analyzed for the volatile organics
using approved gas chromatographic techniques (EPA Method 110). GC/MS
confirmation was provided on a number of these samples. State-of-the-art
capillary column GC/MS techniques were employed to analyze the Method 5
train samples for semivolatiles and nonvolatile organics. The Tenax®
tube samples were also analyzed by capillary GC/MS.
The analytical procedures employed and the results for each of the
above are summarized in this Section. Also included are the quality con-
trol protocols unique to each facet of the analytical program.
2.5.2 Analyses of Bag Samples from Process Vents
2.5.2.1 Gas Chromatographic Measurement Description
Based on the EPA Screening Study (Secion 2.2), analyses
of the process vent bag samples were conducted using standard
EPA protocols (Method 110) for benzene, toluene and chloroben-
zene. However, the presence of components eluting before
benzene and after chlorobenzene created an interest in the
stream-specific relative quantities of these unknown components.
At the request of the EPA project officer, results were provided
on those chromatographic peaks not identified as one of the
above three species. All measurements were performed using
conventional gas chromatographic (GC/FID) techniques. The
program scope also included provision for GC/MS analyses of a
select number of bag samples. The results of these analyses
are presented in Section 2.5.2.4 of this report.
Standard gas mixtures of benzene, toluene, and chloroben-
zene in nitrogen were prepared in Tedlar® bags according to
procedures outlined in EPA Method 110. Leak-checked, 50 liter
Tedlar® bags were filled with nitrogen gas and syringe-spiked
with increments of a benzene-toluene-chlorobenzene solution
to prepare concentrations of each between 5 and 50 ppm. Hie
barometric pressure, relative humidity, and temperature at the
dry gas meter and volume of solution added were recorded. The
actual concentration of each component was determined by the
following formula:
V(D/MW) 24.005 x 103
x = VmY (293/V (V760)
where: Cx = concentration of component "x" (ppm)
D a density of "x" (gm/cc)
MW = molecular weight of "x"
vm = volume measured by the dry gas meter (1)
Y = dry gas meter calibration factor
= absolute temperature of dry gas meter (°K)
2-21
-------
TABLE 2.6
GC/FID OPERATING CONDITIONS FOR
ANALYSES OF TEDLAR® BAG SAMPLES
Instrument
Column
Temperatures
Gas flow/pressure
Perkin-Elmer Model 3920 gas chromatograph with
flame ionization detector and a heated gas
sampling valve and 1 .0 ml sample loop
6' x 2 mm (ID) OV-101 on 80/100 mesh Chromosorb
W HP
Column 110°C
Sample loop 150°C
30 cc/min helium
22 psi hydrogen
50 psi air
Integrator
Spectra Physics Minigrator
2-22
-------
Pm = absolute pressure at dry gas meter (mm Hg)
V = volume of component "x" injected ( 1)
Instrument calibration was performed twice daily by the
duplicate injection of a series of three calibration mixtures
using the GC/FID operating conditions listed in Table 2.6.
Calibration curves were constructed by performing a linear re-
gression analysis on the detector response versus the concen-
tration of the standard mixture. The correlation coefficient
on the linear regression was 0.993 or greater.
Samples and standards were injected into the gas chromato-
graph by means of a heated gas sampling valve. The valve was
successively purged between injections; first with room air,
then with sample gas. Component identification in samples was
based on retention time matching with standards. The detector
response to samples was entered into the calibration curve for
that component in order to determine sample content. Samples
with component response outside the calibrated linear range of
response were diluted with air in a gas tight syringe and rein-
jected. All GC/FID analyses were completed within 72 hours of
sample collection.
2.5.2.2 Results and Discussion of GC/FID Measurements
As noted in the analytical protocol, analysis of each of
the Tedlar® bag samples was conducted using GC/FID techniques
as designated in EPA Method 110. Quantitative results were
provided for each of three (3) volatile species, benzene, tol-
uene, and chlorobenzene. Hie GC/FID results, categorized by
proces vent number, are summarized in Table 2.7. The reported
results present the replicate measurements (denoted as injec-
tion A and B) and the calculated mean (x) of each bag sample.
Table 2.8 shows the instrumental detection limits for benzene,
toluene, and chlorobenzene for each day of bag analysis. These
values were calculated by entering the minimum acceptable de-
tector response (200 area counts) into the daily calibration
curve for each component.
Representative chromatographic scans for a volatile cali-
bration mixture and a typical process vent sample (S-214) are
shown in Figures 2-1 and 2-2, respectively. The chromatographic
pattern shown in Figure 2-2 includes a number of peaks other
than benzene, toluene, or chlorobenzene. This pattern eluting
prior to the benzene peak was noted in several of the bag sam-
ples analyzed. Further discussions on the identity and approx-
imate concentration of these additional species are provided
later in this report.
2.5.2.3 Quality Control for GC/FID Measurements
In an effort to provide a check on the accuracy of GC/FID
quantitations, GCA/Technology submitted three previously ana-
lyzed bag samples for replicate analysis at another laboratory.
2-23
-------
The analyses were performed at the EPA New England Regional
Laboratory, Lexington, Massachusetts. The GC/MS (EPA) screen-
ing analysis of the submitted samples provided qualitative con-
firmation of the presence or absence of benzene, toluene, and
chlorobenzene indicated by the GC/FID analyses at GCA. Quanti-
tative confirmation was hot possible because similar analytical
techniques were not employed at both laboratories. However,
the GC/MS (EPA) screening results correlated well with the
GC/FID (GCA) data with respect to the quantities of benzene
found in the samples, and the ratio of component concentra-
tions between the samples.
Additional confirmatory qualitative GC/MS analyses were
performed at the GCA laboratory. A qualitative comparison of
the GC/MS (GCA) analyses and GC/FID (GCA) analyses is provided
in Section 2.5.2.5 of this report.
All bag samples were analyzed by GC/FID in duplicate in
order to fulfill the requirements of Test Method 110. All re-
plicate analyses were within 5 percent of each other. Table
2.7 presents the individual measurements obtained during the
replicate analyses of each sample. The data indicate that
replication was well within the 5 percent acceptance criteria.
GC/FID analysis was performed on all samples within 72
hours of sample collection. Test Method 110 stipulates a
maximum holding time of 96 hours on Tedlar® bag samples for
benzene analysis. In addition, all sample bags were protected
from light during storage.
2.5.2.4 GC/MS Measurements Description
Additional provisions for the analysis of volatile organic
emissions included the use of gas chromatography/mass spectro-
metry. This technique was applied to a number of bag samples
to provide a quality control function and to further supplement
the GC/FID measurements.
Results of the GC/FID analyses were used to prioritize bag
samples for further GC/MS analysis. Criteria for bag selection
included the presence of appreciable levels of benzene, toluene
or chlorobenzene as well as the presence of unidentified com-
ponents in the GC/FID chromatogram. Figure 2-2 illustrates a
GC/FID scan typical of those chosen for GC/MS analysis.
It was intended that the GC/MS analysis would provide use-
ful qualitative information on process gas emissions to supple-
ment the data on benzene, toluene and chlorobenzene. In addi-
tion to compound speciation, it was intended that the GC/MS
measurements would also serve to confirm the GC/FID quantita-
tive measurements. Both of these roles will be addressed in
more detail in the discussion of the GC/MS analysis results
presented in Section 2.5.2.5.
2-24
-------
TABLE 2.7
REPLICATE GC/FID RESULTS OF TEDLAR® BAG ANALYSES
Analytical Results (ppm)a
Benzene
Toluene
Chlorobenzene
Process
GCA
Vent
Control
Injection
Injection
Injection
No.
No.
A
B
X
A
B
X
A
B
X
K-316
10577
2.2
2.2
2.2
ND
ND
ND
to
•
O
2.0
2.0
10582
6.8
6.9
6.8
ND
ND
ND
ND
ND
ND
K-1702
10604
ND
ND
ND
510
510
510
3.3
3.3
3.3
10600
ND
ND
ND
280
280
280
2.2
2.2
2.2
10619
ND
ND
ND
1400
1500
1400
22
22
22
K-1705
10595
ND
ND
ND
140
140
140
4.5
4.6
4.5
10599
ND
ND
ND
36
36
36
3.1
3.1
3.1
10603
ND
ND
ND
64
66
65
8.3
8.3
8.3
10608
ND
ND
ND
100
100
100
67
67
67
K-1706
10601
5.1
5.1
5.1
3100
3200
3100
9.6
9.4
9.5
10606
ND
ND
ND
1800
1800
1800
2.6
2.6
2.6
S-202
10556
3000
3000
3000
ND
ND
ND
ND
ND
ND
10571
240
240
240
ND
ND
ND
ND
ND
ND
10572
1500
1500
1500
ND
ND
ND
ND
ND
ND
10586
1000
1000
1000
ND
ND
ND
ND
ND
ND
S-203
10573
99b
99b
99b
280
280
280
880
880
880
1 0574
88b
88b
88b
180
180
180
530
530
530
10576
6913
67b
68b
140
140
140
410
390
400
S-205
10577
2300
2300
2300
ND
ND
ND
ND
ND
ND
10578
9500
9800
9600
ND
ND
ND
ND
ND
ND
10585
8800
8600
8700
ND
ND
ND
ND
ND
ND
S-206
10435
470
470
470
6.2
6.0
6.1
ND
ND
ND
10438
110b
100b
100h
7.7
7.3
7.5
ND
ND
ND
10440
18
17
17
7.4
7.0
7.2
ND
ND
ND
10442
54b
53b
53b
6.0
6.1
6.0
ND
ND
ND
1 0444
37
37
37
ND
ND
ND
ND
ND
ND
10449
270
270
270
5.9
5.9
5.9
ND
ND
ND
10451
170
170
170
5.7
5.7
5.7
ND
ND
ND
10558
160b
160b
160b
ND
ND
ND
ND
ND
ND
10560
99
98
98
ND
ND
ND
ND
ND
ND
10575
21
21
21
ND
ND
ND
2.3
2.3
2.3
10579
12
11
11
1.5
1.6
1.5
10
9.7
9.8
10583
31
31
31
ND
ND
ND
ND
ND
ND
S-213
10570
170
170
179
ND
ND
ND
3.4
2.8
3.1
10559
2600
2600
2600
ND
ND
ND
ND
ND
ND
2-25
-------
Table 2.7 - Continued
Analytical Results (ppm)
a
Benzene
Toluene
Chlorobenzene
Process
GCA
Vent
Control
Injection
Injection
Injection
No.
No.
A
B
X
A
B
X
A
B
X
S-213
10563
440
440
440
ND
ND
ND
ND
ND
ND
10564
9100
9100
9100
ND
ND
ND
ND
ND
ND
10584
4800
4800
4800
ND
ND
ND
ND
ND
ND
S-214
10434
21
23
22
210
210
210
ND
ND
ND
10436
c
c
c
280
280
280
7.6
7.8
7.7
10437
c
c
c
260
260
260
7.5
7.8
7.6
10439
c
c
c
300
300
300
7.9
7.9
7.9
10441
32b
32b
32b
21
21
21
12
12
12
10443
c
c
c
27
27
27
8.2
8.2
8.2
10445
c
c
c
42
41
41
7.7
7.7
7.7
10446
c
c
c
12
12
12
6.7
6.7
6.7
10447
c
c
c
23
23
23
5.9
5.9
5.9
10448
c
c
c
30
31
30
7.1
7.1
7.1
1 0450
c
c
c
41
39
40
4.6
4.6
4.6
S-301
10587
1.7
1.7
1.7
ND
ND
ND
ND
ND
ND
10588
0.98
1 .0
0.99
ND
ND
ND
ND
ND
ND
10589
4.3
4.3
4.3
ND
ND
ND
ND
ND
ND
10593
4.3
4.3
4.3
ND
ND
ND
ND
ND
ND
10596
ND
ND
ND
ND
ND
ND
ND
ND
ND
S-303
10561
ND
ND
ND
ND
ND
ND
ND
ND
ND
10565
ND
ND
ND
ND
ND
ND
ND
ND
ND
S-1701
10594
ND
ND
ND
110
110
110
ND
ND
ND
10598
ND
ND
ND
190
190
190
ND
ND
ND
10605
ND
ND
ND
120
120
120
ND
ND
ND
10618
ND
ND
ND
36
36
36
ND
ND
ND
S-1703
10592
ND
ND
ND
3.
3 3.3
3.3
ND
ND
ND
01602
ND
ND
ND
ND
ND
ND
ND
ND
ND
T-393
10580
2.7
3.3
3.0
ND
ND
ND
ND
ND
ND
10581
1.4
1.4
1.4
ND
ND
ND
ND
ND
ND
T-396
10590
ND
ND
ND
ND
ND
ND
ND
ND
ND
10591
ND
ND
ND
ND
ND
ND
ND
ND
ND
T-1718
10597
25
25
25
8400
8800
8600
ND
ND
ND
10607
16
15
15
8300
81 00
8200
ND
ND
ND
a
b
ND - Not detected; see Table 2.8 for appropriate detection limits.
Tailing peak.
Interference.
2-26
-------
TABLE 2.8
DETECTION LIMITS FOR GC/FID ANALYSIS
Process
Vent
No.
GCA
Control
No.
Benzene
Detection Limits (ppm)
Toluene
Chlorobenzene
K-316
10577
10582
0.11a
2.3a
0.31
2.1
1 .2a
1 .2
K-1702
10604
10600
10619
3.0
3.0
3.0
3.1a
3.1a
3.1a
2.2a
2.2a
2.2a
K-1705
10595
10599
10603
10608
2.9
3.0
3.0
3.0
2.8a
3.1a
3.1a
3.1a
1 .9a
2.2a
2.2a
2.2a
K-1706
10601
10606
3.0a
3.0
3.1a
3.1a
2.2a
2.2a
S-202
10556
10571
10572
10586
2.3a
2.3a
0.11a
0.11a
2.1
2.1
0.31
0.31
1 .2
1 .2
1 .2
1.2
S-203
10573
10574
10576
0.11a
0.1 1a
0.11a
0.31a
0.31a
0.31a
1.2a
1.2a
1 . 2a
S-205
10557
10578
10585
2.3a
0.11a
2.3a
2.1
0.31
2.1
1 .2
1.2
1 .2
S-206
10435
10438
10440
10442
10444
10449
10451
10558
10560
10575
10579
10583
2.1a
2.1a
2.1a
2.1a
2.1a
5.9a
5.9a
2.3a
2.3a
0.11a
0.11a
2.3a
5.5a
5.5a
5.5a
5.5a
5.5
4.9a
4.9a
2.1
2.1
0.31
0.31a
2.1
5.5
5.5
5.5
5.5
5.5
2.7
2.7
1 .2
1.2
1.2a
1.2a
1.2
2-27
-------
Table 2.8 - Continued
Process
Vent
No.
GCA
Control
No.
Detection Limits
(ppm)
Benzene
Toluene
Chlorobenz ene
S-213
10570
2.3a
2.1
1.2a
10559
2.3a
2.1
1.2
10563
2.3a
2.1
1.2
10564
2.3a
2.1
1.2
10584
0.11a
0.31
1.2
S—214
10434
2.1a
5.5a
5.5
10436
2.1a
5.5a
5.5a
10437
2.1a
5.5a
5.5a
10439
2.1a
5.5a
5.5a
10441
2.1a
5.5a
5.5a
10443
2.1a
5.5a
5.5a
10445
5.9a
4.9a
2.7a
10446
5.9a
4.9a
2.7a
10447
5.9a
4.9a
2.7a
10448
5.9a
4.9a
2.7a
10450
5.9a
4.9a
2.7a
S-301
10587
0.11a
0.31
1 .2
10588
0.11a
0.31
1 .2
10589
2.9a
2.8
1.9
10593
2.9a
2.8
1.9
10596
2.9
to
•
00
1 .9
S-303
10561
2.3
2.1
1 .3
10565
2.3
2.1
1 .3
S-1701
10594
2.9
2.8a
1 .9
10598
3.0
3.1 a
2.2
10605
3.0
3.1 a
2.2
10618
3.0
3.1 a
2.2
S-1703
10592
2.9
2.8a
1 .9
10602
3.0
3.1
2.2
T-393
10580
0.11a
0.31
1.2
10581
0.11a
0.31
1 .2
T—396
10590
•
CM
2.8
1 .9
10591
2.9
2.8
1 .9
T-1718
10597
2.9a
2.8a
1 .9
10607
3.0a
3.1a
2.2
a Sample contained detectable quantities of component. See Table 2.7
for reported concentration.
2-28
-------
— CD
o» m
o z
"© N
•o m
3 *
~-m
M
VO
GC/Ft D Cond i t ions
Gas Chromatograph: Perkin Elmer
Model 3920 with heated
gas sampling valve (150°C)
and Flame Ionization
Detector
Column: 6' x 2mm (ID) with 10% 0V-I0I on
80/100 Chromosorb W HP
Co Iumn t emp.: II0°C
Carrier gas: 30 cc/min. of helium
Attenuat ion: I x 64
1
in
-4
>
7)
- o
o«"
-rTc
3 z
o
X
r~
— O
UI 30
at o
XJ OB
x> m
3 z
— Nl
m
z
rn
3
5'
o>
in
3
5"
ii
o<
ui
Figure 2-1. GC/FID chromatogram of standard mixture of benzene,
toluene and chlorobenzene.
-------
GC/FIO Cond i 11ops
Gas Chromatograph: Perkin Elmer
Model 3920 with heated
gas sampling valve ((50°C)
and Flame Ionization
Detector
Column: 6' x 2mm (10) with 10% 0V-10! on
80/100 Chromosorb W HP
Co 1 umn t amp.: 110°C
Carrier gas: 30 cc/min. of helium
Attenuation: 1 x 8
o
~ m
U>
H
>
3
3
*¦><
9
II
o
CD
N
X 23
-t H
3 3
S" 5"
ii n
o r
0) o
Ul M
en
a
u>
N
Figure 2-2. GC/FID chromatogram of Tedlar bag sample from the S-214
Process Stream
-------
TABLE 2.9
TEDLAR® BAG SAMPLES CHOSEN FOR GC/MS ANALYSIS
GCA
Control
No.
Process
Vent
No.
Date
Sampled
Sampling a
Interval
10437
S-214
11/18/80
0055-0300
10448
S-214
11/18/80
1030-1325
10449
S-206
1 1/18/80
1320-191 5
10595
K-1705
11/21/80
1845-2345
10600
K-17 02
11/22/80
0050-0750
a Time as recorded on a twenty-four hour clock.
2-31
-------
Upon completion of GC/FID measurements, all bag samples
were repackaged in boxes to minimize light contact, and stored
at room temperature to await further disposition. Five indivi-
dual samples were removed for additional analysis. Table 2.9
provides a listing of the samples chosen for GC/MS analysis.
Aliquots of each of the selected Tedlar® bag samples were
withdrawn for GC/MS anlaysis via a 100 liter gas tight syringe.
Analyses were conducted via direct syringe injection techniques
using a Hewlett-Packard 5985 GC/MS system. Analyses were con-
ducted using the instrumental conditions listed in Table 2.10.
Library searches were conducted on each bag sample for up
to 15 volatile organics including benzene, toluene, and chloro-
benzene. Component identification was accomplished using a for-
ward search of the EPA/NIH/NBS Mass Spectral Data Base. Library
identifications were confirmed by manual comparison of component
spectra with library spectra.
2.5.2.5 Results/Discussion of GC/MS Analysis
Mass spectral analyses were performed on five Tedlar® bag
samples as noted earlier. The bag selection criteria for GC/MS
analysis was based on the EPA Method 110 GC/FID results.
It was anticipated that the mass spectral analysis as desig-
nated in the program scope of work would provide the following:
o Confirmation of EPA Method 110 quantitative results
for benzene, toluene, and chlorobenzene.
o Qualitative and semiquantitative results on volatile
organics other than the three species noted above.
Each of these items was provided by a thorough examination of
the total ion chromatogram obtained for each of the five bag
samples.
Chromatographic data typical of the selected bags are shown
in Figure 2-3. Subsequent analysis of each chromatogram included
manual specral matching techniques. Component identifications
were provided by comparison of peak spectra with computer library
spectra as shown in Figures 2-4 and 2-5. This technique was em-
ployed on each of the designated Tedlar® bag samples for the
qualitative confirmation of benzene, toluene, and chlorobenzene.
A qualitative comparison of the GC/FID and GC/MS results
shown in Table 2.11 provides additional confirmation of the pre-
sence of toluene and benzene. Positive confirmation of chloro-
benzene was not possible due to the small concentrations pre-
sent in each of the five samples selected. It should be further
noted that chlorobenzene levels generally were lower than either
benzene or toluene as evidenced by the results listed in Table
2.7.
2-32
-------
TABLE 2.10
GC/MS OPERATING CONDITIONS FOR TEDLAR® BAG ANALYSIS
Instrument
GC Conditions
Column
Temperature program
Injection
Volume
Temperature
Sweep time
Column flow
MS Conditions
Emission
Electron energy
Scan time
Mass interval
Hewlett-Packard 5985
SE-54, 30 m fused silica
capillary column
25°C held for 2 min, then
5°/min to 100°C and held
Splitless
50 yl
275 °C
0.5 min
UHP helium, 0.5 ral/min
300 ya
70 eV
1.0 sec/scan
35-170 amu
2-33
-------
GCA 10600 TEDLAR BAG FROM K-1702 PROCESS LINE
1
i
I
1
1
2
3
AIR
TOLUENE
OICHLOROBENZENE
I
1
1
2
1
I
3
J
TI
p
I i I
5
0 1*5
GCA 10449 TEDLAR BAG FROM S-206 PROCESS LINE
i
1 AIR
\ Z HEXANE
1
I
i
I
3 BENZENE
3
r«,
\ 2 \
i» I i
' V' 1 •
TI
o
; 2 s ; *> fe 7 e
Figure 2-3. GC/MS scan of selected bag samples from Process Vents
K-1702 and S-206.
2-34
-------
SAMPLE SPECTRUM
GCA 10600 (TEDLAR BAG FROM K-t702 PROCESS LINE)
100.0%
100
LIBRARY SPECTRUM
FRN 3002 SPECTRUM 665 MW = 92 C7H8
BENZENE, METHYL - (9Ct)
1 r 1 i 1 , 1 r '1 1 ¦ r"1 11
30 40 5 0 60 70 80 90 100
Figure 2-4. Comparison of sample and library mass spectra for
toluene (methylbenzene).
2-35
-------
o. BENZENE
SAMPLE SPECTRUM
GCA 10449 (TEDLAR BAG FROM S-206 PROCESS LINE)
I i i 11 I I
100.0%
40 SO 60
•>
LIBRARY SPECTRUM
FRN 3002 SPECTRUM 308 MW=78C6h6
BENZENE(8CI9C1)
100.0%
b. I, 3-0ICHL0R0BEN2ENE
SAMPLE SPECTRUM
GCA 10600 iTEDLAR BAG FROM K-1702 PROCESS LINE)
In u n i ijlHu m) [ii ill In i|ii ill n n [i i nun i [in 11III f 1111 ii 11 ii ||ii mi 11 [i n ii ii || |||| || Ttj
100.0%
SO 60 70 80 90 100 110 120 130 140 ISO
LIBRARY SPECTRUM
FRN 3002 SPECTRUM 4595 MW=I46 C6H4CI2
BENZENE, 1,3-OlCHLORO-(9CI)
HI
100.0%
UI If 11111II11II111 |l If ffll 11 |l III ll II11111IIII111II11 III 1111 fill II11 fin i nil 11| H n I 111 I mi II ill I
30 60 70 80 90 100 110 120 130 140 150
Figure 2.5. Comparison of sample and library mass spectra for
benzene and 1,3-dichlorobenzene.
2-36
-------
TABLE 2.1 1
VOLATILE ORGANIC ANALYSIS — QUALITATIVE COMPARISON
OF GC/FID AND GC/MS RESULTS FOR BENZENE, TOLUENE, AND CHLOROBENZENE
GCA
Control
No.
Process
Vent No.
GC/FIDa
GC/MS
10437
S-214
Chlorobe nzene (7.6
Toluene (260 ppm)
ppm)
Toluene
10448
S-214
Chlorobenzene (7.1
Toluene (30 ppm)
ppm)
Toluene
10449
S-206
Toluene (5.9 ppm)
Benzene (270 ppm)
Toluene
Benzene
10595
K-1705
Chlorobenzene (4.5
Toluene (140 ppm)
ppm)
Toluene
10600
K-1702
Chlorobenzene (3.2
Toluene (280 ppm)
ppm)
Toluene
a ppm values in parentheses obtained from GC/FID measurements listed
in Table 2.7.
2-37
-------
As noted in the GC/MS sample selection criteria discussed
earlier in this section, further analytical efforts included
speciation of a number of volatile organics not identified as
benzene, toluene, or chlorobenzene. Again this included the
manual comparison of sample and computer library spectra as
noted in Figures 2-4 and 2-5. This technique was successful
in identifying additional volatile organics in each of the
five bag-samples chosen. Sample results summarized in Table
2.12 show that GC/MS analysis successfully identified a number
of volatile species, including Cg and C7 hydrocarbons. Spec-
tral matching efforts as shown in Figure 2-6 were successful
in identifying both hexane and cyclohexane in the S-214 process
vent. Figure 2-7 presents the GC/MS scan of a second sample
from the S-214 process line, noting 2-methylpentane and methyl-
cyclopentane as major components. A summary of these results
with semiquantitative (ppm) data for the additional volatile
species is shown in Table 2.12. Since standard materials were
unavailable at the time of analysis for the additional species,
reported concentrations are based on the ratio of the peak area
(total ion count) for each of these components relative to the
peak area for toluene in each of the samples. The latter com-
pound was chosen due to its presence in all of the bags selec-
ted for GC/MS analysis. This fact, combined with the lack of
an internal standard, make it the obvious choice for estimating
concentrations of unknown species. These calculations were
performed in the following manner:
(ppm) unknown = Pea^ Area A x q
Peak Area B
where: A = total ion count of unknown
B = total ion count of toluene
C = ppm value for toluene derived from GC/FID results.
As noted earlier, the results are summarized in Table 2.12.
In addition to the aliphatic hydrocarbon species discussed ear-
lier, a dichlorobenzene isomer was observed in two of the five
bags selected.
Table 2.13 presents a summary of the analytical data for
all Tedlar® bag samples. Component identifications were as-
signed by correlating available vent-specific GC/MS data and
process information with results of the GC/FID analyses on in-
dividual bag samples. Since some standard reference materials
were not available at the time of analysis, results are based
on the calibration curves constructed for either benzene, tol-
uene or chlorobenzene, as appropriate.
2-38
-------
TABLE 2.12
VOLATILE ORGANIC ANALYSIS — ADDITIONAL COMPOUND RESULTS
GCA
Control Process Concentration3
No. Vent No. Component (s) (ppm)
10437 S-214 Hexane 1,000
Methylcyclopentane 600
2-me thylpentane 2,000
Toluene 260
10448 S-214 Cyclohexane 300
2, 2-dimethylbutane 3,000
Hexane 6,000
Methylcyclopentane 3,000
2-methylpentane 3,000
Toluene 31
10449 S-206 Hexane 55
Cg hydrocarbons 23
10595 K-1705 Dichlorobenzene 65
10600 K-1702 Dichlorobenzene 70
a Calculated from the following equation:
(ppm) unknown = area unknown x (ppm) toluene
area toluene
Note that the values presented here may differ slightly from those
presented in Table 2.13 due to differences in analytical methods.
2-39
-------
0. HEXANE
SAMPLE SPECTRUM
6CA 10448 (TEDLAR BAG FROM S-214 PROCESS LINE)
100.0%
LIBRARY SPECTRUM
FRN 4002 SPECTRUM 486 MW» 86 C6HI4
HEXANE (8CI9CI)
100.0%
I 11 I I I 11 111 11 I 11 I I | "' I 11 1 I I 11 1 I 11 ' I I | I I 11 I 11 I I | I 11 I I 11 11|11 I I I IT
20 30 40 50 60 70 80 90
b. CYCLOHEXANE
SAMPLE SPECTRUM
CCA 10448 (TEOLAR BAG FROM S-214 PROCESS LINE)
M I I I I I I I | I I I I'l I I I I | I I I I I I I I I | I I I I I I I I I | I I I I I I I I I | I I I I I I I I I |
30 40 50 60 70 80 90
LIBRARY SPECTRUM
FRN 3002 SPECTRUM 398 MW = 84 C6HI2
CYCLOHEXANE (8CI9CI)
A
I I I I I I I I I I
100.0%
i11 111 II11 I I I I 111) 1111 I 11 I 111 I I I
100.0
Figure 2-6, Comparison of sample and library mass spectra for
hexane and cyclohexane.
2-40
-------
GC-A 10437 [TEOLdR BAG FROM .S-ZI4 PROCESS LINE)
TI
!,V
'¦A
L AIR/ETHYL NITRITE
Z. 2-METHYLPENTANE
5. HEXANE
4. METHLCYCLOPENTANg
5. TOLUENE
5
JV__.
PEAK 2 2-METHYLPENTANE
30
40
SO
60
70
80
100.0%
- I
~l , 1
, i—T-
90
PEAK 4 METHYLCYCLOPENTANE
. ,l
lid „„l
J
I I
SO
60
70
80
100.0%
90
Figure 2-7 . GC/MS 9can of Tedlar bag sample from process line S-214.
2-41
-------
TABLE 2.13
CONCENTRATIONS (PPM) OF COMPONENTS IN TEDLAR® BAG SAMPLES
GCA Component
Process Control Ethyl a Cg Hydro- a Cyclo- Chloro- Dichloro-
Vent No. No. Nitrite carbons3'*- Hexane Benzene'3 hexane3 Toluene benzene3 benzene0
K-316
10577
10582
ND
ND
2.2e
6.9e
e
e
e
e
ND
ND
ND
ND
2.0
ND
ND
ND
K-1702
10600
10604
10619
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
280
510
1400
2.2
3.3
22
84
ND
ND
K-1705
10595
19599
10603
10608
ND
ND
ND
ND
ND
ND
7.6e
ND
ND
ND
e
ND
ND
ND
e
ND
ND
ND
ND
ND
140
36
65
100
4.6
3.1
8.3
67
160
ND
ND
ND
K-1706
10601
10606
ND
ND
5.2e
ND
e
e
e
ND
ND
ND
3200
1800
9.5
2.6
ND
ND
S-202
10556
10571
10572
10586
4.1
ND
ND
ND
17e
20e
19e
2.9e
e
e
e
e
3300
240
1500
1100
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
S-203
10573
10574
10576
250
180
96
470e
500e
440e
e
e
e
99
88
68
ND
ND
ND
280
180
140
880
530
400
ND
ND
ND
S-205
10557
10578
10585
ND
4.8
3.3
ND
ND
2.7e
ND
ND
ND
2300
9700
8700
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
-------
Table 2.13 - Continued
GCA Component
Process Control Ethyl a C6 Hydro- a Cyclo- Chloro- Dichloro-
Vent No. No. Nitrite carbonsa'f Hexane Benzeneb hexanea Toluene benzene3 benzene0
S-206
S-213
S-214
10435
ND
1400e
e
e
ND
6.1
ND
ND
1 0438
2.9
1 200e
e
e
ND
7.5
ND
ND
10440
ND
29e
e
e
ND
7.2
ND
ND
10442
ND
690e
e
e
ND
6.1
ND
ND
1 0444
ND
53e
e
e
ND
ND
ND
ND
10449
ND
430e
e
e
ND
5.9
ND
ND
10451
ND
440e
e
e
ND
5.7
ND
ND
10558
14
72 0e
e
e
27
ND
ND
ND
10560
120
180e
e
e
ND
ND
ND
ND
10575
ND
22e
e
e
ND
ND
2.3
ND
10579
86
21e
e
e
ND
ND
9.9
ND
10583
2.6
72e
e
e
ND
ND
ND
ND
10570
ND
2.8e
e
170
ND
ND
3.1
ND
10559
2.5
2.4e
e
2600
ND
ND
ND
ND
10563
4.6
ND
ND
440
ND
ND
ND
ND
10564
13
8.1e
ND
9100
ND
ND
ND
ND
10584
2.6
4. 2e
e
5000
ND
ND
ND
ND
10434
>5000
>5000e
210
ND
ND
210
ND
ND
10436
>5000
>5000e
190d
ND
36
280
7.7
ND
10437
>5000
>5000e
39 0d
ND
45
260
7.6
ND
10439
>5000
ND
>5000d
ND
31
300
7.9
ND
10441
21
>5000e
ND
ND
240
21
12
ND
10443
41
>5000e
360d
ND
ND
27
8.2
ND
10445
33
ND
>5000d
ND
53
41
7.7
ND
10446
11
ND
>5000d
ND
49
12
6.7
ND
10447
7.8
53e
14
ND
ND
23
5.9
ND
10448
40
ND
>5000d
ND
180
30
7.1
ND
10450
9.1
64e
15
ND
ND
40
4.6
ND
-------
Table 2.13 - Continued
Process
Vent No,
GCA
Control
No.
Ethyl a
Nitrite
Component
C5 Hydro-
carbons3
Hexane
Benzene1
Cyclo- Chloro- Dichloro-
hexanea Toluene benzene3 benzene0
S-301
10587
10588
10589
10593
10596
ND
ND
ND
ND
ND
1 .7e
1 .0e
4. 3e
4.3e
ND
e
e
e
e
ND
e
e
e
e
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
S—303
10561
10565
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
S-1701
10594
10598
10605
10618
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
110
190
120
36
ND
ND
ND
ND
ND
ND
ND
ND
S-1703
10592
10602
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
3.3
ND
ND
ND
ND
ND
T-393
10580
10581
ND
ND
3.1e
1 . 4e
e
e
e
e
ND
ND
ND
ND
ND
ND
ND
ND
T-396
1 0490
10591
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
T-1718
10597
10607
ND
ND
39e
29e
e
e
e
e
ND
ND
8600
8200
ND
ND
ND
ND
a Quantitation based on benzene calibration curve,
k Mean measured value as reported in Table 2.7.
c Quantitation based on chlorobenzene calibration curve.
d Could possibly contain ethyl ether in addition to hexane.
e All individual components of assigned boiling point range cannot be identified in this sample with
the process and analytical data presently available.
f Cg hydrocarbons not specifically identified as hexane, benzene, or cyclohexane.
ND Less than 15 ppm
-------
2.5.3 Analysis of Tenax® Tube Samples
2.5.3.1 Introduction
The analytical procedures employed for the Tenax® tubes
were designed for the quantitative analysis of semivolatile
organics and a qualitative screen for volatile organics. Qual-
itative and quantitative analyses were conducted using capillary
GC/MS techniques. The results of the screening study described
in Section 2.2, combined with existing knowledge of Upjohn pro-
cess chemistry, indicated that the sampling and analysis efforts
should focus on a number of key components as listed in Table
2.14. Due to the anticipated complexity of the lagoon and
settling pond samples, analytical efforts were not limited to
these species alone, but included provisions for the identifi-
cation and quantitation of up to 10 additional components in
each adsorbent cartridge sample. All identifications were made
using computerized spectral matching techniques. Computer spec-
tral matches were further verified using manual spectral match-
ing techniques. Component quantitations were made relative to
the appropriate standard reference materials. For components
other than those listed in Table 2.14, quantitations were pro-
vided relative to the designated internal standard present in
each sample extract.
2.5.3.2 Analytical Procedures
All Tenax® tubes were spiked with a surrogate standard solu-
tion to contain 100 ug of d5~phenol and 93 ug of dg-naphthalene.
Additionally, seven of the tubes were randomly selected to be
spiked with 100 yg each of dg-benzene and dg-toluene. The tubes
were then extracted with pentane for 16 hours in a soxhlet appa-
ratus. The apparatus was allowed to cool and the solvent ex-
tract transferred to a rotary evaporation unit for solvent re-
duction. The extract was concentrated to 2.0 ml, using a 35°C
water bath. The extract was then diluted to 4.0 ml with hexane
for GC/MS analysis to include the condensable organics listed in
Table 2.14 and a volatile organics screen for benzene, toluene,
and chlorobenzene. GC/MS analysis followed the instrumental
conditions listed in Table 2.15. All extracts were spiked with
a d^g-anthracene internal standard immediately prior to GC/MS
analysis.
Identification of compounds for which standard reference
materials were available was based on a comparison of both the
retention time and mass spectral characteristics between sample
and standard injections. Refer to Figure 2-8 for a typical GC/
MS scan of a condensable calibration mixture. Quantitation of
these compounds was achieved by establishing a relative response
factor (RRF) for each compound versus the d-|Q-anthracene inter-
nal standard.
Additional compounds were identified via a GC/MS library
search. The library search program compares unknown spectra
2-45
-------
TABLE 2.14
ORGANICS OF INTEREST—LAGOON/
SETTLING POND STUDY
acetoxyanisole
benzene
m-chloroaniline
o-chloroaniline
chlorobenzene
chlorophenol
o-chlorophenol
2,5-di chloroaniline
dichloronitroaniline
di e thoxyace tophenone
dimethyIformamide
p-ni troaniline
o-nitroanisole
o-ni trochlorobe nzene
saturated amines
toluene
p-toluene sulfonylisocyanate
trichlorodipenylethane
2-46
-------
TABLE 2.15
GC/MS OPERATING CONDITIONS FOR CONDENSABLE ORGANICS ANALYSES
Instrument
Hewlett-Packard 5985
GC Conditions
Column
SE-54, 30-m fused silica
capillary
Temperature program
25°C held for 2 min, then
10°/min to 260°C and held
Injector temperature
275 °C
Injection volume
Typical 1 y1
Column flow
UHP helium, 0.5 ml/min
MS Conditions
Emission
300 ya
Electron energy
70 eV
Scan time
0.8 sec/scan
Mass interval
41-350 amu
2-47
-------
TI
^
6
8 9
•j i
1.
2.
3.
4.
5.
6.
7.
8.
9.
2-CHLOROPHENOL
0-CHLOROANILINE
m-CHLOROANILINE
2,3 - DICHLOROANtLINE
0-NITROAN1SOLE
p-NITROANILINE
p-TOLUENE SULFONYLISOCYANATE
2.6-DICHLORO-4-NITROAHILINE
^ 10ANTHRACENE (INTERNAL STANDARD)
' ' ' 1 To r
i
_L5_
Figure 2-8. GC/MS scan of standard condensable mixture.
-------
with known spectra from the NIH library and lists the library
spectra most similar to the unknown spectra. Confirmation of
the spectral match between the unknown and its assigned iden-
tification was then conducted manually. Semiquantitative mea-
surements of these additional components were accomplished by
assigning an RRF of 1.0 and calculating a concentration based
bn comparison of the unknown peak area with that of the inter-
nal standard.
Three of the sorbent tubes (settling pond, lagoon) were
re-extracted with a 1:1 ethyl ether:hexane solvent system in an
attempt to investigate the presence of any unknown organics not
extracted previously by the aliphatic solvent system. Prior
to extraction, each tube was respiked with d5-phenol and dg-
naphthalene. Tubes were soxhlet-extracted overnight with 1:1
ethyl ether:hexane. The sample extracts were concentrated us-
ing Kuderna-Danish evaporators and analyzed via GC/MS as before.
2.5.3.3 Results and Discussion
Results of organic analysis of Tenax® sorbent tubes are
summarized in Table 2.16. All reported results have been blank
corrected. As noted in that table, small quantities of o-
chloroaniline were detected on adsorbent tubes from both the
aeration and the settling ponds. This particular component was
not detected on either the method blank or on any of the tubes
submitted as sample blanks. A GC/MS scan characteristic of
these samples is shown in Figure 2-9.
Quality control results below document the validity of the
protocols chosen for the analysis of the semivolatile organics;
e.g., substituted anilines and phenols collected on the sorbent
tubes. The sorbent tube samples were also screened for volatile
organics including benzene and toluene which were thought to
be associated with the onsite lagoon and settling pond. These
components were not detected in the GC/MS analysis of Tenax®
extracts (Table 2.16). However, a study of the surrogate re-
covery data presented in the following discussion of quality
control procedures, indicates that volatile organic emissions
could not be adequately assessed using the analytical proto-
cols chosen for this program.
2.5.3.4 Quality Control
The quality control protocols for this project included
the pretest cleanup of Tenax® sorbent tubes by GCA/Technology
personnel. The Tenax® adsorbent was sequentially extracted
with methanol and then hexane in a soxhlet apparatus. The
extraction period for each solvent was 24 hours. The Tenax®
was then dried and packed into glass tubes. The Tenax® was
secured in the tube with plugs of pre-extracted glass wool.
Each tube was then desorbed twice with helium for a period of
2 hours. Tubes were assigned code numbers as indicated pre-
viously in Table 2.16.
2-49
-------
TABLE 2.16
RESULTS OF GC/MS ANALYSIS OF TENAX® EXTRACTS
Sample
Description
GCA
Control
No.
Compound
Identified
CAS No.
Quantity3 Quantity
(ug) (ug)
Aeration Pond—
Sample 2, Tube 1
Aeration Pond—
Sample 1, Tube 2
Aeration Pond—
Sample 2, Tube 2
Aeration Pond—
Sample 1, Tube 1
Blank
Blank
Blank
Blank
Blank
Blank
Settling Pond—
Sample 1, Tube 2
Settling Pond—
Sample 2, Tube 1
Settling Pond—
Sample 2, Tube 2
Settling Pond—
Sample 1, Tube 1
1061 4 None detected
10615 None detected
10616 Silane, [1, 3, 5-
benzene-triyltris
(oxy)] tris-
(trimethyl)-
10617 o-chloroaniline
10620 Hone detected
10621 None detected
10622 None detected
10623 None detected
1062 4 Hydroperoxide,
1-methylbutyl
1,2-Benzenediol,
4-(2-hydroxyethyl)
1 062 5 None de tected
10626 None detected
10627 o-chloroaniline
10628 None detected
10629 o-chloroaniline
0010586-12-6
0014018-58-7
0010597-60-1
<10
<10
56
<10
<10
<10
<10
56
52
<10
<10
<10
4.0
4.5
18
a Based on comparison of total ion area of unknown to that of internal standard,
b Quantitation using component-specific calibration curves.
2-50
-------
1. d5 PHENOL ISPIKE)
2. 0-CHLOROANIL1 HZ
3. dgNAPHTHALENE (SPIKE]
4. d,0ANTHRACENE [INTERNAL STANDARD]
3
I
I
,
i
TI
1,1 ¦
«
. 1
'0 15
-V "I 1 "J—-—] T- -r
?o ?«i
Figure 2-9. GC/MS scan of typical Tenax tube extract.
2-51
-------
TABLE 2.17
QUALITY CONTROL RESULTS—SEMIVOLATILES RECOVERED FROM TENAX®
Total ug Percent
Compound Actual Observed Recovery
o-chloroaniline 210 53 25
m-chloroaniline 130 51 39
2-chlorophenol 190 70 37
2-52
-------
Two blank Tenax® tubes were prepared for analysis concur-
rently with the program samples. One was analyzed to provide
a laboratory method blank. The other was spiked with two
chlorinated anilines and a chlorinated phenol to provide re-
covery information through the extraction procedure. Results
of the analysis of the spiked tube are presented in Table 2.17.
An additional quality control measure utilized was the sur-
rogate spiking of samples. All Tenax® tube samples were spiked
with 100 yg of ds-phenol and 93 ug of dg-naphthalene. Recovery
data for these compounds are shown in Table 2.18. Seven of the
program samples were randomly chosen to receive additional sur-
rogate spiking. To those samples was added 100 ug each of dg-
benzene and dg-toluene from solutions supplied by EPA. Recovery
data on these spikes indicated that benzene and toluene levels
remaining in the sample concentrations were not detectable.
2.5.4 Analyses of the Condensible Samples
2.5.4.1 Introduction
The results of the screening efforts combined with existing
chemical knowledge of selected Upjohn processes indicated that
analysis efforts should focus on a number of key components as
listed earlier in Table 2.14. The analytical efforts, however,
were not limited to these species but included provisions for
the identification and quantification of up to 10 additional
components not listed in Table 2.14. The anticipated complexity
of process chemistry coupled with the need to assess potential
health effects of unknown vent gas emissions dictated the need
for comprehensive mass spectral analyses.
As was the case with the analysis of Tenax® cartridge sam-
ples, preliminary component identifications were made using
computer based spectral matching techniques. These were fur-
ther verified using manual spectral matching procedures.
Analyses were conducted on samples received from a number
of process vents including the following designated streams:
o S-303 — process vent gas and particulate matter emissions,
o S-1701 — vent gas and particulate matter emissions,
o K-1702 — vent gas only.
o S-1703 — vent gas and particulate matter emissions,
o K-1705 — vent gas only.
Analytical results on each of the above streams and a dis-
cussion on quality control protocols are provided in Sections
2.5.4.3 and 2.5.4.4 of this report.
2.5.4.2 Analytical Procedures
The EPA Method 5 train used to sample condensable organics
provided for the generation of five different sample components:
2-53
-------
TABLE 2.18
QUALITY CONTROL RESULTS—SURROGATES RECOVERED FROM TENAX®a
di;-phenol
da-naphthalene
GCA
Control No.
Total
Added
Observed
Percent
Recovery
Total
Added Observed
Percent
Recovery
10614
10615
10616
10617
10620
10621
10622
10623
10624
10625
10626
10627
10628
10629
Method blank
Spike
10617b
10627b
10629b
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100c
100°
100
47
30
35
33
48
61
0.1
0.1
39
19
30
24
49
96
52
24
110
77
69
47
30
35
33
48
61
0.1
0.1
39
19
30
24
49
96
52
24
110
77
69
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93c
93c
93
50
37
46
43
68
91
0.4
1 .0
46
30
25
28
65
1 26
52
45
58
84
53
54
40
49
46
73
98
0.1
1 .0
49
32
27
30
70
1 30
56
48
62
90
57
a Sample concentrates did not contain detectable levels of dg-benzene and
dg-toluene surrogates.
b Tubes reextracted with 1:1 ethyl ether: hexane.
c Tubes could have additional surrogate remaining from first spiking.
2-54
-------
Particulate filters were surrogate-spiked with dg-naphtha-
lene and d5~phenol and soxhlet extracted with methylene chloride
(200 ml) for a period of 16 hours. The apparatus was allowed
to cool prior to transfer of the solvent extract. The extract
was concentrated via rotary evaporation (40°C) and the final
volume adjusted to 2.0 ml with methylene chloride. Each sample
was then refrigerated awaiting combination with the appropriate
front-half rinse.
A 50 percent aliquot of each submitted methylene chloride
front-half train rinse was taken for analysis. The solvent was
dried with anhydrous sodium sulfate and concentrated to 2.0 ml
using a Kuderna-Danish evaporator. A 1.0 ml portion of the con-
centrate was combined with a 1.0 ml portion of the corresponding
particulate filter extract for GC/MS analysis.
A 50 percent aliquot of each of the KOH impingers was trans-
ferred to a 1000 ml separatory funnel and surrogate-spiked with
ds-phenol and o-cresol. The pH was adjusted with 1N H2SO4 to be
between 2 and 4. The sample was then extracted three times with
20 percent (v/v) methylene chloride. The extract was then com-
bined with a 50 percent aliquot of the corresponding back-half
rinse and dried with anhydrous sodium sulfate. The combined
samples were then concentrated using a Kuderna-Danish apparatus.
The extracts were then analyzed by GC/MS using the instru-
ment conditions previously listed in Table 2.15. The qualitative
identification procedures used, as well as the judgment criteria
and spectral searching techniques, were the same as those used
for the Tenax® tube analysis. Quantitative concentrations are
provided on those compounds for which calibration curves from
injections of standard reference materials were prepared. The
reported concentrations of other components identified by the
library search were computed by comparing total ion area of the
component to that of the d^g-anthracene internal standard.
2.5.4.3 Results/Discussion
Nonvolatile and semivolatile process vent emissions were
collected using Method 5 and Method 6 type trains fitted with
alkaline (KOH) impinger systems. As noted earlier in this sec-
tion, analyses were conducted on samples from a number of pro-
cess vents. Qualitative and quantitative analyses were con-
ducted using GC/MS techniques as outlined above. The analyti-
cal results for all of the samples are summarized in Table 2.19.
As shown, the analytical efforts were not limited to the pro-
cess chemicals listed in Table 2.14, but included provisions
for the identification and quantitation of up to 10 additional
organic components. As noted in Table 2.19, the complexity of
the vent samples was quite variable. S-1701 (both runs) and
S-1703 (both runs) vent samples contained very little extract-
able organic material. Both of the particulate filter/rinse
extracts contained isooctyl phthalate, typically attributable
2-55
-------
TABLE 2.19
GC/MS ANALYSIS OF CONDENSIBLE SAMPLES
Process Run
Vent No. No.
Train
Fraction
GCA
Control
Nos.
Compound
Identified
Cas No.
Quantity3
(pg)
Quantity'3
(pg)
S-303 1 Particulate 11179,
filter/rinse 11090
benzeneamine, 2,4-
dichloro-6-nitro
benzeneamine, 2,6-
dichloro-4-nitro
1,300
0000099-30-9
240
Impinger/rinse
11180,
11181
pentachlorophenol
2,4,6-trichlorophenol
benzene, chloro-nitro
isomers
benzene, 1,2-dichloro-
3-nitro
trichloroaniline isomer
benzene, trichloro-
nitro isomer
phenol, tetra-chloro
isomer
benzeneamine, 2,4-
dichloro-6-nitro
benzeneamine, 2,6-
dichloro-4-nitro
0000087-65-5
0000088-06-2
0003209-22-1
0000099-30-9
1,400
300
1,700
500
700
5,000
310
130
1,400
-------
Table 2.19 - Continued
Process Run
Vent No. No.
Train
Fraction
GCA
Control
Nos.
Compound
Identified
Cas No.
Quantity3
(pg)
Quantity*3
(yg)
S—303
Particulate 11183,
filter/rinse 11095
benzeneamine, 2,4-
dichloro-6-nitro
1,2-benzenedicarboxylic 0033374-28-6
acid, 2-butoxyethyl
butyl esterc
210
130
benzeneamine, 2,6-
dichloro-4-ni tro
0000099-30-9
490
Impinger/rinse
11184,
11185
pentachlorophenol
0000087-86-5
500
benzene, 1-chloro-3- 0000121-73-3 600 -
nitro
benzene, chloro-nitro - 900 -
isomer
benzene, 1,2,dichloro- 0003209-22-1 900 -
3-nitro
trichloroaniline isomer - 210 -
benzene, trichloro-nitro - 1,200 -
isomer
tetrachlorophenol 0000935-99-5 2,100 -
benzeneamine, 2,4- - 18,000 -
di ch 1 or o-6-n i tr o
benzeneamine, 2,6- 0000099-30-9 - 2,400
dichloro- 4-nitro
-------
Table 2.19 - Continued
Process Run
Vent No. No.
Train
Fraction
GCA
Control
Nos.
Compound
Identified
Cas No.
Quantity3
(Mg)
Quantity
(pg)
S-1701
Particulate 11094,
filter/rinse 11194
None detected
160
Impinger
11195
2-chloroaniline
0000095-51-2
18
S-1701
Particulate 11197,
filter/rinse 11093
Impinger
11198
isooctyl phthalatec,d 0027554-26-3
280
None detected
240
K-1702
Impinger/rinse 11200,
11201
2-chloroaniline
benzene, chloro-nitro
isomer
0000095-51-2 200,000
260,000
chloroaniline isomer
90,000
S-1703 1 Particulate 11203, isooctyl phthalatec,c* 0027554-26-3 60
filter/rinse 11092
Impinger 11204 None detected - 160
S-1703 2 Particulate 11206, isooctyl phthalatec,d 0027554-26-3 31
filter/rinse 11096
Impinger/rinse 11207, None detected - 240
11208
-------
Table 2.19 - Continued
GCA
Process Run Train Control Compound Quantitya Quantity'3
Vent No. No. Fraction Nos. Identified Cas No. (pg) (pg)
benzene, chloro 0000108-90-7 14,000 —
50,000
benzene, chloro-nitro - 2,300
isomer
chloroaniline isomer - - 500,000
a Based on comparison of the total ion area of unknown to that of internal standard,
k Quantitation using component-specific calibration curves.
° Phthalates are common contaminants in environmental samples.
^ Present in laboratory blank.
e Component cannot be identified with process and analytical data presently available.
K-1705 Impinger/rinse 11210,
11211
-------
to artifact contamination during handling and analysis. Simi-
lar results were obtained for the impinger/rinse combinations
with the exception of S-1701 which contained a small quantity
(18 yg) of o-chloroaniline. The latter component was readily
identifiable from available Upjohn process chemistry.
Analysis of vent gas samples from K-1702 indicated signi-
ficant concentrations of several chloroaniline isomers includ-
ing o-chloroaniline. Particulate samples were not collected
from this vent stream. K-1705 vent gas samples also contained
significant quantities of an unidentified chloroaniline isomer.
Significant concentrations of a number of chlorinated benzenes
were also found including mono-chlorobenzene and the meta- and
para-dichlorobenzene isomers. These results in particular are
consistent with the available process chemical data on these
streams including the volatile emissions results noted earlier
in this report. As reported earlier, measurable concentrations
of mono-chlorobenzene were observed in vent K-1705. This same
vent also contained significant concentrations (55 ppm) of 1,
3-dichlorobenzene. Both of these earlier findings are further
documented by the results shown in Table 2.19.
Sample complexity, however, was most pronounced in the case
of the two S-303 samples as documented by the results listed in
Table 2.19 and the GC/MS scans shown in Figure 2-10. The
observed complexity of these samples indicated that a detailed
chemical analysis.was warranted. These observations were
further substantiated after consultation with appropriate
project personnel, including the EPA Project Officer, Mr.
Robert O'Meara. As was the case with all of the vent samples
examined, additional mass spectral analyses were conducted
using both computerized and manual spectral matching techniques.
Position confirmation of component spectra were provided after
comparison with available standard spectra as shown for 2,6-
dichloro-4-nitroaniline in Figure 2-11.
As noted in Table 2.19, both S-303 runs 1 and 2 contained
significant quantities of a number of chlorinated adjuncts of
aniline and phenol, in particular, 2,6-dichloro-4-nitroaniline
The condensable organics were generally contained in the alka-
line impinger extracts. There were, however, noticeable levels
of both dichlorophenol and 2,6-dichloro-4-nitroaniline assoc-
iated with the S-303 particulate filters. The percent dis-
tribution of DCNA between the gaseous and particulate phases
was calculated to be approximately 85:15 for both runs 1 and
2. It is interesting to note that despite the disparity in
total quantities observed in runs 1 and 2, the distribution
ratio between the particulate and gaseous phases remained
constant. As a result it can be concluded that the majority
of the organic emissions associated with the S-303 vent are
contained in the gaseous phase. In the case of both runs, 85
percent of the DCNA was collected in the alkaline impinger/
2-60
-------
1. d5 PHENOL (SPIKE)
2. 0-CRESOL (SPIKE)
3. BENZENE, CHLORONITRO ISOMERS
4. 2,4,6 - TRICHLOROPHENOL
5. 1,2-OICHLORO -3 - NITROBENZENE
6. TRICHLOROANIL IN£
7. BENZENE. TRICHLORONITRO ISOMER
8. TETRACHLOROPHENOL
9. 2,4- OICHLORO —6 — NITRO ANILINE
10. 2.6 -OICHLORO-4-NITRO ANILINE
11. PENTACHLOROPHENOL
12. d |0 ANTHRACENE (INTERNAL STANOARO)
to
I
C\
S-303 RUN I (GCA III80 11181)
•| ,,!S—V" Tr"
|! 4
Jl !'U
10
I J J
A i\J ¦>*
10
ii12
Ubi x
l—~r
is
T"
20
S-303 RUN 2 (GCA III84+IM8S)
• 2 I
^ 111 f» K ji
i—tTh t r—i— 1"" ' 'i
4
Jl
lO
'5
M
10
'J
12
20
Figure 2-1 °» GC/MS scan of impinger rinse extract from the S-303 Process Stream—Runs 1 and 2.
-------
STANOARD SPECTRUM
2,6 - DlCHLORO-4-NITRO ANILINE
100.0%
200 220
SAMPLE SPECTRUM DlCHLORO NlTRO ANILINE ISOMER
RUN Z, S-3C& PROCESS STREAM (PROBE RINSE/FILTER EXTRACT}
100.0%
LJ
'¦"lilt
III I
40
60
80
100
120
140
160
Jj
180
200
220
Figure 2.11. Comparison of standard and sample mass spectra
for dichoronitroaniline (DCNA).
2-62
-------
rinse and only 15 percent of the total was collected on the
filter/probe rinse.
In summary, the semivolatile/nonvolatile organic emis-
sions from the process vents tested are confined to a variety
of chlorinated aniline and phenolic derivatives. While the
organic emissions from vents S-1701 and S-1703 were below ana-
lytical detection limits, measurable quantities of chlorinated
organics were found in the S-303, K-1702 and K-1705 vent sam-
ples. These levels were primarily confined to the S-303 vent
as shown in Table 2.19. The K-1702 sample did, however, contain
quantities of o-chloroaniline and the K-1705 sample quantities
of a number of chlorobenzene isomers.
2.5.4.4 Quality Control
The quality control protocols for the condensible organic
compound sample analysis included the analysis of DCNA spikes,
method blanks, and surrogate spikes.
A typical Method 5 train filter (Reeve Angel Grade 900 AF)
was prepared by GCA/Technology for use as a matrix for a method
recovery study. The filter was spiked with 51 yg of DCNA and
analyzed along with the submitted sample filters. Results are
shown in Table 2.20. Similarly, a 500 ml portion of laboratory
water was spiked with 50 yg of DCNA and analyzed with the pro-
gram samples, with the results also shown in Table 2.20.
An additional quality control measure used was the surro-
gate spiking of filter and impinger samples. The filters were
spiked with 50 y g of d5-phenol and 56 yg of dg-naphthalene prior
to extraction. Impingers were spiked with 50 yg of d5~phenol
and 59 yg of o-cresol. Results of analyses for surrogate spikes
are presented in Table 2.21.
2.6 EMISSION RATE CALCULATIONS
2.6.1 Volatile Organic Compound Emission Rates
Seventeen sampling points were selected to characterize the twenty
emission sources listed in Table 2.1. As previously mentioned, T-393/
T-394 were considered as a single source because the two tanks were
used alternately. The same reasoning was applied to T-395/T-396. Only
two (K-1702 and K-1705) of five reactor stacks (K-1701, K-1702, K-1703,
K-1704 and K-1705) were sampled. The emission rates generated for those
two stacks were applied to all five reactor stacks.
Bascially, the emission rate, Ri, was calculated by employing the
formula:
Ri = (Ci) (MWi) (298°K) (P) (Q) (60 min/hr)
(24.45 liters/mole) (T) (760 mm Hg) (0.0353 cu. ft/liter)
2-63
-------
TABLE 2.20
QUALITY CONTROL RESULTS ~ DCNA RECOVERY
FROM FILTER/IMPINGER SAMPLES
Total yg Percent
Matrix Actual Observed Recovery
Particulate filter 51 90 180
Impinger 85 90 100
2-64
-------
TABLE
2.21
QUALITY CONTROL RESULTS—SURROGATES RECOVERED FROM CONDENSIBLE SAMPLES
ds-phenol
dR-naphthalene
o-cresol
GCA
Control No.
Total ug
Added
Observed
Percent
Recovery
Total pg
Percent
Total yig
Percent
Added Observed Recovery Added Observed Recovery
to
l
Oi
ui
11179/1
11180/1
11183/1
11184/1
11094/1
11195
11197/1
11198
11200/1
11203/1
11204
11206/1
11207/1
1090
1181
1095
1185
1194
1093
1201
1092
1096
1208
Particulate
Filter
Spikes
Impinger
Spike
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
30
69
49
43
70
74
63
24
110
99
110
59
88
120
75
60
140
98
86
140
150
130
48
220
200
220
120
180
240
150
56
56
56
56
56
56
56
37
50
94
72
130
69
140
66
89
170
130
230
120
250
59
59
59
59
59
59
59
60
69
68
63
82
60
78
100
120
115
110
140
100
130
59
56
95
-------
where: Ri = emission rate (grams/hour)
Ci = VOC concentration (ppm by volume)
MWi = VOC molecular weight (g/mole)
P = stack pressure (mm Hg)
Q = stack exit flow rate (ACFM)
T = stack temperature (°K)
For this study, it was assumed that the stack pressure for all emission
points was constant and approximately 760 mm Hg. Therefore:
Ri = (Ki) (Ci) (Q) / (T)
Using the subscript "B" for benzene, "T" for toluene, and "C" for chloro-
benzene:
Kb = 1.62 (MWb = 78.1 g/mole)
Kr = 1.91 (MWt = 92.1 g/mole)
Kc = 2.33 (MWC = 112.6 g/mole)
The VOC concentrations (Ci) were previously given in Table 2.7.
Stack temperatures (T) were obtained during discussions with Upjohn per-
sonnel. stack exit flow rates (Q) were calculated from actual stack
exit velocity data generated by Engineering-Science, from blower name-
plate data, and from process knowledge.
These calculations provide both average and maximum emission rates.
Volatile organic emission rates were generated for each bag sample by
employing the contaminant concentration found in the bag and the mean
flow rate during the bag sampling. The overall average emission rate
for each contaminant was calculated by averaging the values from all
bag samples over the entire sampling cycle. The maximum emission rate
for each contaminant was the highest single sample bag emission rate.
The results of the VOC emission rate calculations are presented in
Table 1.2. The major sources of benzene were S-202, S-205, S-206, S-213,
and S-301. Average benzene emissions from these vents ranged from 12.2
to 59.5 g/hr. compared to essentially zero for the other vents. Vent
S-1701, with an emission rate of 7040 g/hr, was overwhelmingly the major
source of toluene. Very little chlorobenzene was emitted compared to
the quantity of benzene and toluene.
2.6.2 Condensible Organic Compound Emission Rates
Condensible organic compound emission rates were calculated for
eight emission sources at the Upjohn facility. Five sampling points
were selected to characterize the eight emission sources. A list of
these sources is provided in Section 2.2. In one of the processes, only
two of five reactor stacks (K-1702 and K-1705) were sampled. The emis-
sion rates for those two stacks were applied to all five reactor stacks.
2-66
-------
Basically, the emission rate, Ri, was calculated by employing the
formula:
Ri _ (Wi) (298°K) (P) (Q) (60 min/hr)
(Vs.) Tt) (760 mm Hg) (10° yg/g)
where: Ri = emission rate (grams/hour)
Wi = weight of condensible organic compound in sample (micro-
grams, ug)
P = stack pressure (mm Hg)
Q = stack exit flow rate (ACFM)
Vs = volume of sample (standard cubic feet, SCF)
T = stack temperature (°K)
For this study, it was assumed that the stack pressure for all emis-
sion points was constant and approximately 760 mm Hg. Therefore:
Ri = (0.0179 ) (Wi) (Q) / (Vs) (T)
The weights of condensible organic compound (Wi) were previously
given in Table 2.19. Stack temperatures (T) were either measured by
Engineering-Science in the field or were obtained during discussions
with Upjohn personnel. Sample volumes (Vs) were measured by ES. Fur-
ther, stack exit flow rates (Q) were calculated by ES for the stacks
sampled by EPA Method 5 procedures. For the remaining stacks, flow
rates were estimated based on process knowledge.
These calculations provide both average and maximum emission rate.
Condensible emission rates were generated by employing the weight of
contaminant found in each sample and the mean stack flow rate during the
sample run. The overall average emission rate for each contaminant was
calculated by averaging the values from all sample runs over one or two
complete sampling cycles. Hie maximum emission rate for each contaminant
was the highest single sampling run emission rate.
The results of the condensible emission rate calculations are pre-
sented in Table 1.2. The condensible compound emitted at the highest
rate was 2,4-dichloro-6-nitrobenzeneamine, 125.5 g/hr on average.
2-67
-------
SECTION 3
AMBIENT AIR QUALITY STUDY
3.1 INTRODUCTION
3.1.1 Background
This section of the report describes the ambient air quality study
performed by EPA. The objective of the study was to supplement a data
base from which a health study could be performed to evaluate the
effects of emissions of organic compounds from Upjohn's process and
wastewater treatment facilities.
Upjohn Company's Fine Chemicals Division facility in North Haven,
Connecticut produces a variety of organic chemical products. In the
production of these chemicals, various organic compounds are used and
are emitted in measurable quantities to the ambient air from process
point sources and from the plant's wastewater treatment facility.
Emission rates from these sources were presented and discussed in Section
2. The ambient air sampling strategy was designed to determine the
types and ranges of concentrations of these compounds in the ambient
air (under various meteorological conditions) to which the area popula- ;
tion is exposed.
This ambient air sampling study was conducted in conjunction with
the ES point source sampling and analysis program described in "ection
2. Data from the point source study has been used in a computer model
to estimate annual ground-level contaminant concentrations under a variety
of meteorological conditions (Section 4). These estimates however, are
based on average emission rates which do not reflect short duration
variations, and further, are based on historical meteorological condi-
tions. As a result of the limitations of the point source study, the
results of the ambient air quality study should contribute to the overall
project goals in the following manner:
1. Three hour integrated sample results will show short term
contaminant levels at specific locations on and surrounding the
Upjohn site under various meteorological conditions.
2. Real-time or instantaneous sample results will show the
range in contaminant concentrations at locations on and surrounding
the Upjohn site.
3-1
-------
3.1.2 Summary
The ambient air quality study at the Upjohn Company was conducted
in three phases.
3.1.2.1 Phase 1: November 1980 Study
Initially, all ambient sampling was planned to coincide
with the point source sampling in November 1980. Prior to the
November study period, a screening exercise (described in Chap-
ter 2) was conducted to characterize the types and approximate
concentrations of the air contaminants emitted from the Upjohn
facility. Based on process data and the results of the screen-
ing survey, EPA determined that there were two major organic
compound emission areas at the Upjohn facility. These were:
1) the chemical processing area which included Buildings Nos.
2, 3, and 17; and, 2) the wastewater treatment facility which
included the settling ponds and aeration lagoon. EPA also
determined that volatile organic compounds (VOCs) such as
benzene, toluene, and chlorobenzene were the contaminants
emitted in the greatest detectable quantities from the Upjohn
facility and, therefore, would most likely be found in the
ambient air on and around plant property. Due to the nature
of condensible organic compounds, it was felt that they would
not be present at measurable levels in ambient air far from
their emission points. From this information, both Tenax® and
charcoal were chosen as sample collection media. Based on
preliminary vent flow rates and building and stack heights,
EPA's regional meteorologist performed some simple dispersion
modeling to predict where maximum pollutant ground-level con-
centrations would occur under varying wind speeds. Additionally,
Epa reviewed the history of odor complaints from people within
a 1.5 mile radius of the Upjohn property. This information
was used to select sampling sites within surrounding neighbor-
hoods. An air sampling strategy was formulated which included
the collection of upwind and downwind samples on or near Upjohn
property, and downwind samples in surrounding neighborhoods.
Tenax® and charcoal integrated air samples were collected at
seventeen locations during the period November 17-21, 1980 (See
Appendix A: Sample Location Map) and analyzed at EPA's New
England Regional Laboratory in Lexington, Massachusetts. The
analytical results showed that the only compounds present in the
samples in measurable quantities were benzene and toluene.
Evidence was subsequently found in the data and in the litera-
ture that there were some inherent problems with Tenax® collec-
tion material. Artifact formation, leaching of benzene and
toluene, breakthrough of benzene, and sample degradation appeared
to be prevalent. As a result, EPA determined that the November
1980 Tenax® data could not be used with confidence.
A close examination of the charcoal data reveals that the
upwind levels of both benzene and toluene measured at five
locations were consistent (0.7 to 2.3 ppb for benzene and 4.0
to 15.0 ppb for toluene). It was concluded from these results
3-2
-------
that either additional upwind sources of VOC's exist in the
vicinity of the Upjohn facility or, more likely, that the pollut-
ant levels represent VOC's which had remained on the cleaned
media. (A more detailed discussion of this issue is presented
in Section 3,2,4). Concerning downwind locations, benzene levels
immediately downwind of the processing area (3 locations) were
not significantly above background but toluene levels were
measured at up to 60 ppb. Downwind of the aeration lagoon (4
locations) benzene levels were detected as high as 49 ppb and
toluene levels up to 90 ppb.
Seven integrated sampling locations were established in down-
wind neighborhoods. On only one occasion wis benzene measured
at a level significantly above background and on one other
occasion the toluene level was elevated.
3.1.2.2 Phase 2i March 1981 Study
In Marcii 1981, an abbreviated ambient air sampling study
was conducted to improve on and assess the validity of the November
1980 study data through a new set of integrated samples and,
further, to generate real-time analytical data by direct injec-
tion of ambient air into a portable gas chromatograph. The
integrated air sampling strategy was similar to that for
November 19 80. However, precautions were taken to insure that
the November problems were not repeated. Tenax® and/or char-
coal integrated air samples were collected at nine locations
during the period March 24-26, 1981 and analyzed at EPA's New
England Regional Laboratory. Additionally, instantaneous
samples were collected at 26 locations. (See Appendix A: Sample
Location Map.) Overall, the analytical results showed that the
only measurable compounds were benzene and toluene.
A close examination of the data reveals that upwind levels
of both benzene and toluene measured at two locations were con-
sistent and low, supporting the conclusion that the charcoal sam-
ple media used in November was not thoroughly purged of VQCs
prior to its field application.
Samples from locations between the Upjohn processing area
and wastewater treatment facility showed elevated benzene and
toluene levels. The integrated sample level was 15 ppb for ben-
zene and 114 ppb for toluene. (Note that this toluene value is
the average of 27 ppb and 200 ppb. The limitation on the use of
this value is discussed further in Section 3.3.4.) Instantan-
eous sample levels were measured higher than 400 ppb for benzene
and 200 ppb for toluene. Downwind of the aeration lagoon (2 loca-
tions), integrated sample levels were 7 to 9 ppb for benzene and
14 to 20 ppb for toluene. Instantaneous samples from three loca-
tions near the lagoon indicated that VOC levels varied from 6 ppb
to 50 ppb for benzene and from 20 to 90 ppb for toluene. Both
integrated and instantaneous sample results from the area down-
wind of the Upjohn settling lagoons indicated that they were not
a source of VOC air contamination.
3-3
-------
Both integrated and instantaneous air samples were collected
immediately downwind of the main Upjohn processing area. The lone
integrated sample revealed an elevated toluene level (54 ppb).
Instantaneous sample results from six locations indicated that the
benzene level varied between 7 and 20 ppb and the toluene level
varied from as low as 2 ppb to over 200 ppb.
Two integrated sample locations and nine instantaneous sam-
ple locations were established off plant property at a distance
downwind of the identified Upjohn air emission sources. Hie
highest benzene level encountered was 5 ppb and the highest
toluene level was 20 ppb.
3.1.2.3 Phase 3: June 1981 Study
In June 1981, another ambient air sampling study was con-
ducted to gather air quality data during warm weather. This was
accomplished by collecting integrated ambient air samples on
Tenax® and charcoal media and by collecting real-time data through
direct injection into a portable gas chromatograph. Hie sampling
strategy was nearly identical to that for March 1981. Even
greater precautions were taken to minimize the previously en-
countered problems associated with Tenax® and charcoal sample
collection media.
Tenax® and/or charcoal integrated air samples were collected
at twelve locations during the period June 30 to July 1, 1981
and analyzed at EPA's New England Regional Laboratory. Addi-
tionally, instantaneous samples were collected at 45 locations.
(See Appendix A: Sample Location Map.) Again, the only compounds
detected during sample analysis were benzene and toluene.
Upwind data showed that both benzene and toluene levels were
at or below one part per billion during the sampling period. This
was true for both integrated and instantaneous samples.
Integrated sampling from a location between the Upjohn pro-
cessing area and wastewater treatment facility showed elevated
benzene (11 ppb) and toluene (8 ppb) levels. Downwind samples
taken south and west of the processing area revealed low ben-
zene and toluene levels suggesting that there are no major VOC
emission sources in the southern portion of the Upjohn facility.
Downwind of the aeration lagoon (2 locations), instantaneous sam-
ple results confirmed the March 19 81 study conclusion that the
lagoon is a measurable source of VOC emissions (benezene levels
varied from 10 to 30 ppb and toluene levels varied from 20 to 60
ppb).
Both integrated and instantaneous air samples were collected
immediately downwind of the main Upjohn processing area both on
and immediately off site. Hie lone integrated sample showed only
low VOC levels (3 ppb benzene and 3 ppb toluene). Hxese results
were quite different from samples taken in the same area during
earlier study periods. Possible factors contributing to this
difference include unaccounted for meteorological conditions and
3-4
-------
chemical process variations. Supplementally/ a number of instan-
taneous ground level air samples were taken in and immediately
around the main processing area. These showed benzene levels as
high as 100 ppb and toluene levels as high as 500 ppb. Addi-
tional instantaneous air samples taken downwind of the
processing area revealed'the intermittent nature of the process
area VOC emissions. The benzene level at these locations varied
from undetectable to 100 ppb and the toluene level varied from
undetectable to 800 ppb.
Seven integrated and twelve instantaneous sample locations
were established at a distance downwind of the identified Upjohn
VOC emission sources. The highest VOC levels were detected
immediately north of the plant near the expressway. Neighborhood
instantaneous sample results ranged from undetectable to 10 ppb
for benzene and from undetectable to 21 ppb for toluene.
Integrated sample results ranged from less than one to 4 ppb
for benzene and from less than one to 3 ppb for toluene.
3.1.3 Conclusions
Based on the results of this study, a number of conclusions can be
drawn concerning the air quality on and surrounding the Upjohn facility.
Table 3.1 depicts the range of pollutant levels that can be expected
at different sampling locations based on the sampling data from November
1980, March 1981, and June 1981. The table is only for discussion
purposes since many factors which may affect the dispersion of volatile
organic compounds in air have not been thoroughly studied. The conclu-
sions drawn from the ambient measurements were:
1. The volatile organic compounds toluene and benzene are the
most measurable in the ambient air on and surrounding the
Upjohn facility.
2. Background levels of VOCs in the vicinity of the Upjohn
facility are low.
3. One source of VOCs is Upjohn's main processing area. Ihree
hour integrated sample results range from 1-5 ppb for benzene
and 3-50 ppb for toluene immediately downwind of the area.
Due to variable meteorological conditions and the intermit-
tent nature of Upjohn1s processes, instantaneous sample re-
sults range from undetectable to 100 ppb for benzene and from
undetectable to 800 ppb for toluene.
4. Another source of VOCs is Upjohn's aeration lagoon. (This
is not true for the settling lagoons.) Three hour integrated
sample results ranged from 3-10 ppb for benzene and 1-30 ppb
for toluene immediately downwind of the lagoon. Instantaneous
sample results ranged from undetectable to 50 ppb for benzene
and from undetectable to 90 ppb for toluene.
5. At distant locations downwind from the main Upjohn emission
sources (including surrounding neighborhoods), VOC levels are
3-5
-------
TABLE 3.1
GENERALIZED AIR CONTAMINANT LEVELS
ON AND SURROUNDING
THE UPJOHN COMPANY
Integrated Pollutant Instantaneous Pollutant
Concentration Range Concentration Range
(ppb,v/v)a (ppb,v/v)a
Location Benzene Toluene Benzene Toluene
Upwind (background)
K1-1
K1-1
ND—2
ND-4
Immediately downwind
of main processing
area
1-5
3-50
ND-L100
ND-800
Immediately downwind
of aeration lagoon
3-10
1-3 0
ND-50
ND-90
In downwind
neighborhoods
K1-5
K1-2 0
ND-3 5
ND—150
a All values are in parts per billion (ppb) on a volume/volume basis in
ambient air. K1 reads "less than one". ND reads "not detectable".
L100 reads "greater than 100".
3-6
-------
generally lower than those measurable in the vicinity of the
pollutant sources. Three-hour integrated sample results ranged
from less than one to 5 ppb for benzene and from less than one
to 20 ppb for toluene. Instantaneous sample results ranged
from less than one to 35 ppb for benzene and from undetectable
to 150 ppb for toluene. ¦
3.2 NOVEMBER 1980 STUDY
3.2.1 Introduction
The following is a discussion of the activities associated with
EPA's ambient air sampling study of November 1980.
Prior to the actual sampling, EPA conducted a screening exercise to
optimize the use of manpower and equipment toward project goals. Addi-
tionally, since ambient air sampling for organic compounds is a relatively
new procedure, many sampling and analysis" protocols were developed during
this phase of the study. Those protocols found to be inadequate were
modified during later sampling programs. Actual sampling was conducted
during the period from November 17 to 21, 1980.
3.2.2 Sampling Strategy
A screening exercise was initially conducted to characterize the
types and approximate concentrations of air contaminants emitted from
the Upjohn facility as described in Section 2.2. This was necessary to
better define the type of sampling equipment and the number of sampling
stations which would best meet the objectives of the program. In addition
to those factors discussed in Section 2.2, this screening exercise also
consisted of preliminary measurements of building dimensions and other fac-
tors affecting ambient dispersion of pollutants.
From the screening exercise, it was determined that volatile organic
compounds such as benzene, toluene, and chlorobenzene were the compounds
emitted in the greatest detectable quantities from the Upjohn facility
and, therefore, would most likely be found in the ambient air on and
around plant property. Due to the nature of condensible organic compounds,
it was felt that they would not be present at measurable levels in ambient
air far from their emission points. A literature search was conducted
to determine the most effective medium on which to collect these compounds.
It was decided that both Tenax® and charcoal would be employed to collect
duplicate integrated air samples at each sampling point.
Based on preliminary vent flow rate measurements and building and
stack heights, EPA's regional meteorologist performed some simple dis-
persion modeling to predict, as a function of wind speed, where maximum
ground level concentrations of pollutants would occur. Hiese results
are present in Table 3.2. From this modeling, it was determined that
maximum ground level concentrations would likely be found on the Upjohn
property or at its perimeter. A sampling strategy was subsequently
formulated where an integrated air sampling station would be located
upwind of the emission sources, a second station would be located
downwind near the emission sources, and finally, a third station would
3-7
-------
TABLE 3.2
LOCATION OF MAXIMUM POLLUTANT GROUND CONCENTRATIONS
AS A FUNCTION OF WIND SPEED:
MODELING RESULTS FOR UPJOHN STUDY
Distance Downwind of
Wind Speed (mph) Emission Source (meters)
5 700
10 300-400
15 270-400
22 250-400
3-8
-------
be located farther downwind on a line with the first two near the point
of predicted maximum ground level concentration. Portable meteorological
equipment would be installed on site to aid in the positioning of sampl-
ing equipment. This strategy was designed to intercept the highest
ground level concentration and to provide an estimate of the diffusion
gradient.
Additionally, EPA reviewed the history of odor complaints from people
living within a 1.5 mile radius of the Upjohn facility. In response to
these complaints, EPA decided to locate sampling stations within the
neighborhoods to better define the types and quantities of airborne pollu-
tants. Based on sufficiently documented odor complaints in Upjohn's
files, the following neighborhood sampling locations were chosen:
1. An empty lot on Homewood Avenue,
2. Stiles Street near State Street,
3. Elm Street near William Street,
4. Several locations on Sackett Point Road,
5. A cul-de-sac at the intersection of Forest Street
and Ash Street,
6. Elm Street near Bailey Road.
3.2.3 Equipment Preparation, Field Activities and Analytical Procedures
3.2.3.1 Equipment Preparation
Table 3-3 lists the major pieces of field equipment which
were checked and used during the November 1980 ambient air
study. Additionally, sample traps and media were prepared at
EPA's Regional Laboratory as follows:
1. Empty stainless steel tubes were heated to 300°C in a
Century programmed thermal desorber (PTD). A number
of the steel tubes were then checked for cleanliness
with a Century Systems gas chromatograph (GC). (See
equipment details in Appendix B.)
2. The clean tubes were then filled with either Tenax® or
charcoal, gently tapping the sides to insure a tight
pack. A small piece of glass wool was packed in the
end, and a fritted metal plug was pressed into place.
3. The Tenax® and charcoal-filled tubes were then heated to
170°C in a Century PTD and checked on a Century GC.
This desorption cycle was repeated until no significant
GC peaks were observed. A number of tubes were cycled
up to six times.
4. When each tube was ready, the plastic caps provided by
the manufacturer were placed on the ends until the
tube was used.
5. Sample tubes were stored at room temperature until used.
3-9
-------
TABLE
FIELD SAMPLING EQUIPMENT
AMBIENT AIR STUDY AT
3.3
USED FOR NOVEMBER 1980
THE UPJOHN COMPANY®
1. Battery operated sampling pumps
2. Five-foot sampler mount stands
3. Hastings flow meter
4. Climatronics portable meteorological station
5. Belfort hand-held meteorological instrument
6. Tenax® adsorption tubes
7. Charcoal adsorption tubes
a Mention of trade names does not constitute an endorsement or recommenda-
tion for use.
3-10
-------
3.2.3.2 Field Activities
Integrated ambient air sampling was conducted on and around
the Upjohn facility during the period from November 17-21, 1980.
A portable meteorological station was first established on Upjohn
property to determine wind speed and direction. That data was
then used to locate the air sampling stations. Data was obtained
from this station throughout the sampling period.
The basic sampling station arrangement is depicted in Figure
3-1. Charcoal and Tenax® tubes were attached to the inlet lines of
air pumps which were mounted on sampler stands. The tube inlets
were situated approximately six inches apart and four to five feet
above ground (breathing height).
The sampling consisted of drawing known volumes of ambient
air through duplicate parallel sampling lines. Each line con-
sisted of two stainless steel tubes in series connected by a short
piece of Teflon® tubing. The second tube in the sampling line was
a back-up tube to detect breakthrough in the first. The tubes of
one sampling line were packed with charcoal, those of the other
line were packed with Tenax.®
The pumps were turned on, the start time was noted, and ini-
tial flow rates established using a Hastings bubble meter.
Acceptable flow rates and total volumes were precalculated based
on potential sample breakthrough and sample quantity necessary to
yield a low detectable limit. The flow rates typically ranged
from about 100cc/min to 700cc/min with most flow 300cc/min or
less. The sampling equipment was attended at all times and the
flow rate was checked periodically throughout the sample period
which was approximately three hours in duration. When the samp-
ling period was completed, a final flow check was made and the
exact stop time recorded. The samples were placed in appropriate
containers and labeled with date, time, and sample.number. Addi-
tionally, the samples were kept cold after exposure until re-
ceived by the laboratory. Chain-of-custody procedures were fol-
lowed at all times.
During the November 1980 study period, 24 Tenax® and 20 char-
coal samples were collected from 24 sample stations at the 17 dif-
ferent locations described in Table 3.4. At 19 sample stations
both Tenax® and charcoal samples were collected simultaneously.
At one station (8212) duplicate Tenax® samples were collected to
check for consistency of results.
3.2.3.3 Analytical Procedures
The 44 integrated air samples were analyzed at EPA's New England
Regional Laboratory in Lexington, Massachusetts according to the
procedure outlined below. More detail on equipment and procedures
is presented in Appendix B.
3-11
-------
POMP .
CHARCOAL TUBE
/ i
FOUR FEET
V
Figure 3-1 Typical Sampler Arrangement Used for Integrated Air Sampling
during Upjohn Study
~ PUMP"
TENAX TUBE
b-
APPROXIMATELY 6 INCHES
3-12
-------
TABLE 3.4
UPJOHN AMBIENT AIR SAMPLING STUDY:
NOVEMBER 1980 INTEGRATED SAMPLE LOCATIONS
Location
I. D.
Location Description
Station
Number
Sample
Date
Sample Time
NA-1
On Stiles Street West
8171
11/1 7
2:16
pm
5: 1 6
pm
of Upjohn facility
8183
11/18
2:28
pm
—
5:35
pm
NA-2
15 yards from Stiles
Street-Upjohn entrance
8213
11/21
10:30
am
-
12:30
pm
NA-3
In Upjohn administration
West parking lot
8201
11/20
5:25
am
—
8:25
am
NA-4
In Upjohn administration
East parking lot
8202
11/20
5:55
am
-
8:55
am
NA-5
At EPA meteorological
8194
11/19
2:15
pm
-
5:1 5
pm
station
8211
1 1/21
9:16
pm
-
12:16
pm
NA-6
On Upjohn property 100°
8199
11/19
8:1 5
pm
-
11:15
pm
from aeration lagoon**
82010
11/20
3:20
pm
-
6:00
pm
NA-7
On Upjohn property
8198
11/19
7:58
pm
-
1 0:58
pm
from aeration lagoon
NA-8
On Upjohn property 120°
8173
11/17
2:00
pm
-
5:00
pm
from aeration lagoon**
8182
11/18
2:30
pm
-
5:30
pm
8196
11/19
1:58
pm
-
4:58
pm
8212
11/21
9:30
am
—
noon
NA-9
On Upjohn property 160°
8192
11/19
8:10
am
-
11:15
am
from aeration lagoon**
NA-10
On Sackett Point Road 195°
8181
11/18
2:35
pm
—
5:35
pm
from Upjohn property*
NA-11
On RR tracks 150° from
8195
11/19
2:00
pm
_
5:00
pm
Upjohn property*
NA-1 2
On RR tracks 145° from
8193
11/19
8:20
am
11:30
am
Upjohn facility*
NA-1 3
Near cul-de-sac at Forest
8197
11/19
8:25
pm
_
11 :00
pm
and Ash Streets
NA-14
At intersection of Elm
8204
11/20
10:52
am
-
1:52
pm
and Baily Roads
3-13
-------
Table 3.4 - Continued
Location Station Sample
I. D. Location Description Number Date Sample Time
NA-15 At intersection of William
and Elm Streets
8203
11/20
10:25
am -
1:25
NA-16
On Devine Street
8191
11/19
8:58
am -
1 1 :58
NA-17
On Homewood Avenue
8172
11/17
2:50
pm -
5:50
82011
11/20
3:35
pm -
6:17
* Compass readings are measured from the intersection of Center Road and B
Street on Upjohn's property.
** Compass readings are measured from the center of the aeration lagoon.
3-14
-------
1. Each sample tube was desorbed at 200°C in a pro-
grammed thermal desorber and a syringe sample withdrawn.
2. The syringe sample was injected into a Tracor 560 gas
chromatograph with photoionization detector.
3. Random samples were injected into a Finnigan 3200 gas
chromatograph/mass spectrometer to verify the results of
the Tracor 560.
4. Most tubes were analyzed several times to prevent
random syringe error. Further, quality control in-
cluded laboratory blanks.
3.2.4 Results and Conclusion
The analytical results of all integrated air samples collected during
the November 1980 study are presented in Table 3.5. Basically, the only
measurable compounds were benzene and toluene. A summary of all the
meteorological data collected during the sampling period is presented
in Table 3.6.
Before an interpretation of the results is presented, a discussion of
the adequacy of the sampling and analysis protocol is required. Evidence
has been found in the data and in the literature that suggests some in-
herent problems with Tenax® collection material. Artifact formation,
leaching of benzene and toluene, and sample degradation are problems that
appear to be prevalent. An evaluation of the Tenax® data collected in
November indicates that these phenomena may have occurred. As a result,
EPA feels that the Tenax® data cannot be used with confidence in this
study.
The charcoal samples were prepared and handled in the same manner as
the Tenax® samples. No significant artifact formation nor contamination of
the charcoal collection media was noted. However, due to laboratory sched-
uling and equipment problems, many of the charcoal samples were stored
for an unrecommended length of time before analysis, which may have result-
ed in some sample deterioration. Additionally, the cleaned sample tubes
were checked on a Century GC (see Section 3.2.3.1). This instrument
is not as sensitive to low level contaminants as the GC with photoioniza-
tion detector used in the final sample analysis. Therefore, it is possible
that some small quantity of volatile organics was not thoroughly desorbed
from the sample tubes during equipment preparation because it was undetect-
able on the GC used for checking. These volatile organics may have ulti-
mately been reported as low level contamination during the final sample
analyses. No mechanism was available to estimate either the amount of
sample deterioration, if any, or the contaminant level in the cleaned
sample tubes.
As a result of the above mentioned problems, the Tenax® data were
voided and the charcoal data were used to show only approximate pollutant
levels. Further, the problems necessitated a second sampling program
which is discussed in Section 3.3.
3-15
-------
TABLE 3.5
UPJOHN AMBIENT AIR SAMPLING STUDY:
ANALYTICAL RESULTS OF NOVEMBER 1980
INTEGRATED SAMPLES
Pollutant Concentrations Pollutant Concentrations
on Charcoal Sorbent on Tenax® Sorbent**
(ppb, v/v)*** (ppb, v/v)***
Station Number Benzene Toluene Benzene Toluene
8171
J 1 .2
J 7.7
J39
J181
81 72
J 0.7
J 4.0
J56
J200
8173
J 3.6
J1 0.5
J 7
J 20
8181
J 7.5
J15.5
J21
J103
8182
J10.5
J1 4.0
J26
J107
8183
J 1.3
J 6.5
J68
J327
8191
J 2.3
J1 5.0
J24
J104
8192
J 0.7
J 5.5
J30
J1 29
8193
J 0.9
J 8.7
J27
J1 28
8194
*
*
J27
J109
8195
J 1 .1
J 5.5
*
*
8196
J 9.0
J16.0
J48
J2 23
8197
J 2.7
J52.0
J 28
J 87
8198
J24.5
J45.0
J76
J288
8199
*
*
J13
J 21
8201
J 1.3
J17.7
J21
J 70
8202
J 1 .7
J30.0
J24
J 86
8203
J 1.5
J17.0
J18
J 81
8204
J 1 .2
J 9.0
J17
J 52
82010
J 1.7
J1 2.0
J13
J 11
82011
J 2.0
J14.0
J48
J290
8211
J 1.8
J 8.0
J26
J 77
821 2A
*
*
J22
J103
8212B
*
*
J22
J 86
8213
*
*
J22
J 19
* Sample not collected.
** These data are not considered valid as explained in the text.
*** All results reported in parts per billion (ppb) on a volume/
volume basis in ambient air. J reads "approximately".
3-16
-------
TABLE 3.6
METEOROLOGICAL DATA:
NOVEMBER 19 80 STUDY PERIOD
Wind Direction* Wind Speed
Day Hour (degrees) (mph)
12:00
-
1:00
p.m.
130
5
1:01
-
2:00
p.m.
150
3
2:01
-
3:00
p.m.
18 0
3
3:01
-
4:00
p.m.
210
6
4:01
-
5:00
p.m.
170
6
5:01
-
6:00
p.m.
170
3
9:00
—
10:00
a.m.
20
13
10:01
-
11:00
a .m.
20
14
11:01
-
12:00
noon
20
14
12:01
-
1:00
p.m.
20
13
1:01
-
2:00
p.m.
10
12
2:01
-
3:00
p.m.
0
12
3:01
-
4:00
p.m.
0
12
4:01
-
5:00
p.m.
340
13
5:01
-
6:00
p.m.
340
13
6:01
-
7:00
p.m.
340
13
9:00
_
10:00
a.m.
340
15
10:01
-
11 :00
a.m.
350
15
11:01
-
12:00
noon
340
15
12:01
-
1:0 0
p.m.
360
14
1:01
-
2:00
p.m.
350
13
2:01
-
3:00
p.m.
350
14
3:01
-
4:00
p.m.
340
12
4:01
-
5:00
p.m.
340
8
5:01
-
6:00
p.m.
340
7
6:01
-
7:00
p.m.
330
6
7:01
-
8:00
p.m.
330
5
8:01
-
9:00
p.m.
330
4
9:01
-
10:00
p.m.
variable
2
10:01
-
11:00
p.m.
variable
3
11:01
-
12:00
variable
3
3-17
-------
Table 3.6 - Continued
Wind Direction* Wind Speed
Day Hour . (degrees) (mph)
5:00
-
6:00
3i . Ql*
variable
3
6:01
-
7:00
3. • m.
variable
2
7:01
-
8:00
& . m.
variable
3
8:01
-
9:00
d * nu
variable
3
9:01
-
10:00
variable
2
10:01
-
11:00
sl . m.
310
5
11:01
-
12:00
noon
300
6
12:01
-
1:0 0
p.m.
310
6
1:01
-
2:00
p.m.
310
6
2:01
-
3:00
p.m.
300
6
3:01
-
4:00
p.m.
300
6
4:01
-
5:00
p.m.
270
4
5:01
-
6:00
p.m.
240
3
10:00
_
11 :0 0
ci. m.
225
2
11:01
12:00
noon
200
4
* 0° is north; 90° east; 180° south; and 270° west.
3-18
-------
TABLE 3.7
SUMMARY OF ANALYTICAL RESULTS AS A FUNCTION
OF METEOROLOGICAL CONDITIONS FOR NOVEMBER 1980
INTEGRATED SAMPLES
Approximate Pollutant Concentration
Location
I.D.
Station
Number
Upwind/
Downwind
Wind Speed
(mph)
Day/
Night
(ppb,
Benzene
v/v)**
Toluene
NA-1
81 71
D
6
Day
J 1.2
J 7.7
8183
U
1 2
Day
J 1.3
J 6.5
NA-2
8213
D
4
Day
*
*
NA-3
8201
D
3
Day
J 1.3
J17.7
NA-4
8202
D
3
Day
J 1.7
J30.0
NA-5
8194
D
12
Day
*
*
8211
U
4
Night
J 1.8
J 8.0
NA-6
8199
D
3
Night
*
*
82010
D
5
Day
J 1.7
J12.0
NA-7
8198
D
3
Night
J24.5
J45.0
NA-8
8173
D
5
Day
J 3.6
J10.5
81 82
D
12
Day
J10.5
J14.0
8196
D
12
Day
J 9.0
J16.0
821 2A
U
2
Day
*
*
821 2B
U
2
Day
*
*
NA-9
8192
D
15
Day
J 0.7
J 5.5
NA-1 0
8181
D
12
Day
J 7.5
J15.5
NA-11
8195
D
12
Day
J 1.1
J 5.5
NA-1 2
8193
D
15
Day
J 0.9
J 8.7
NA-1 3
8197
D
3
Night
J 2.7
J52.2
NA-1 4
8204
D
6
Day
J 1.2
J 9.0
NA-15
8203
D
6
Day
J 1.5
J17.0
NA-16
8191
U
15
Day
J 2.3
J15.0
NA-17
81 72
U
6
Day
J 0.7
J 4.0
8201 1
U
3
Day
J 2.0
J14.0
* Only Tenax® data was available for the sample
** Based on charcoal results only. All results reported in parts per billion
(ppb) on a volume/volume basis in ambient air. "J" reads "approimately".
3-19
-------
Table 3.7 has been constructed from Tables 3.5 and 3.6 to provide
a format from which general conclusions can be drawn on upwind and
downwind pollutant concentrations.
For each integrated sample, an attempt has been made to generally
determine whether the sample station was upwind or downwind during sample
collection. This is based on the station location and measured wind
direction. Some caution should be used when applying this data since
sample locations were rarely either directly upwind or directly downwind
from Upjohn air pollutant sources. Table 3.7 also lists the approximate
average wind speed during the sampling period since, as was discussed
in Section 3.2.3, low wind speeds (less than 10 miles per hour) tend
to increase the distance from the emission source where the maximum
pollutant ground level concentration can be expected. Further, a day/
night determination has been made because stability factors affecting
pollutant distribution are different during daylight and dark hours.
Lastly, only pollutant concentration data from the charcoal samples are
reported.
A close examination of Table 3.7 reveals that upwind levels of both
benzene and toluene were consistent. Benzene concentrations ranged
from 0.7 to 2.3 ppb and toluene levels from 4.0 to 15.0 ppb for five up-
wind samples. It was concluded from these results that either additional
upwind sources of volatile organics exist in the vicinity of the Upjohn
facility or, more likely, that the pollutant levels represent volatile
organic compounds which still remained on the media after cleaning. It
was decided that more thorough sample trap preparation would be required
for future sampling programs and that a more sensitive GC should be used
to check for cleanliness.
Concerning downwind integrated samples, it should be noted that two
major organic compound emission areas were identified at the Upjohn
facility based on the emission screening study (See Section 3.2.2):
1. The chemical processing area which includes Buildings Nos. 2,
3, and 17 and,
2. The wastewater treatment facility which includes the settling
ponds and aeration lagoon.
During the November 1980 study, integrated sampling locations NA-1,
2, 3 and 4 were established in the vicinity of the processing area. In
general the analytical results show that the downwind benzene level in
this area was not significantly above background. However, the toluene
level was measured once under low wind speed conditions at 30 ppb.
Sampling locations NA-5, 6, 7, 8 and 9 were established in the immediate
vicinity of the wastewater treatment lagoons. Basically, downwind day-
time benzene levels were measured as high as 10.5 ppb and one nighttime
level at 24.5 ppb, significantly above background. Daytime toluene
levels were in the 8-16 ppb range which was similar to the background
range. However, a downwind nighttime sample at NA-7 showed 45 ppb of
toluene.
3-20
-------
Integrated sampling locations NA-10 through NA-17 were established in
surrounding neighborhoods. On only one occasion (8181 at NA-10) was ben-
zene measured downwind at a level significantly above background. On
one other occasion, an elevated toluene level was measured. This was
at night under low wind speed conditions when 52 ppb was detected.
In summary, it was concluded from the November 1980 sampling study
that the Upjohn facility was a contributor of benzene and toluene to the
ambient air within and beyond its perimeter, although under most
meteorological conditions, neighborhood contaminant levels were within
the background range. However, it was further concluded that additional
sampling would be required due to problems with sample media preparation
and analyses.
3.3 MARCH 1981 STUDY
3.3.1 Introduction
The following is a discussion of the activities associated with
EPA's ambient air sampling study of March 1981. The purpose of this
study was to improve on and check the validity of the November 1980
study data through a new set of integrated samples and, further, to
generate real-time analytical data by the direct injection of ambient
air into a portable gas chromatograph. These instantaneous data provided
in-the-field checks on the types and approximate levels of volatile
organic compounds present in the air and, further, provided EPA with
a tool to locate the high ground level concentration areas under varying
meteorological conditions for subsequent integrated sampling.
3.3.2 Sampling Strategy
The integrated air sampling strategy for the March 1981 program was
similar to that for November 1980. However, the strategy included that
laboratory and field sampling personnel take every precaution available
to minimize the problems encountered during the November 1980 program.
Those precautions necessitated a much more rigorous cleaning procedure,
sample storage in thoroughly cleaned glass and Teflon® containers,
sample cartridge refrigeration, and the reduction of storage time prior
to analysis. These procedures would be followed for both Tenax® and
charcoal samples. Further, the sampling strategy included quality
control checks such as duplicate Tenax® and charcoal samples for at
least one location and the taking of field blanks.
As in the November 1980 study, it was decided that integrated air
sampling stations would be located upwind and downwind of the emission
sources both on Upjohn property and in surrounding neighborhoods.
Portable meteorological equipment would be installed on site to aid in
the positioning of sampling equipment. Additionally, either a portable
photoionization or flame ionization gas chromatograph would be used to
collect real time data. This data would be used to help locate inte-
grated sampling stations.
3-21
-------
TABLE 3.8
FIELD SAMPLING EQUIPMENT USED FOR MARCH 1981
AMBIENT AIR STUDY AT THE UPJOHN COMPANY*
1.
Battery operated, Sierra and DuPont constant flow
pumps
2.
Five-foot sample mount stands
3.
Hastings flow meter
4.
Climatronics portable meteorological station
5.
Belfort hand-held meteorological instrument
6.
Tenax® adsorption tubes
7.
Charcoal adsorption tubes
8.
Photovac model 10A10 portable photoionization gas chromato-
graph with sample collection syringes
9.
Century Systems 128 portable flame ionization gas
graph
chromato-
* Mention of trade names does not constitute an endorsement or recommenda-
tion for use.
3-22
-------
3.3.3 Equipment Preparation, Field Activities and Analytical Procedures
3.3.3.1 Equipment Preparation
The major pieces of field equipment which were checked and
used during the March 1981 ambient air study are listed in
Table 3.8. Sample traps and media were prepared at EPA's New
England Regional laboratory as follows:
1. Empty stainless steel tubes were heated to 300°C in a
Century programmed thermal desorber (PTD). The steel
tubes were then randomly checked for cleanliness with a
Century Systems gas chromatograph. (See equipment details
in Appendix B.)
2. The clean tubes were then filled with either Tenax® or
charcoal gently tapping the sides to insure a tight pack.
A small piece of glass wool was packed in the end, and a
fritted metal plug was pressed into place.
3. All the traps were baked out at 280°C for at least 21 hours
at a nitrogen flow of 50 ml/min.
4. The traps were individually sealed in a 40 ml screw cap
glass vials.
5. Each trap was screened on a GC with photoioniztion detector
for its total benzene and toluene background level prior
to the Upjohn survey.
6. Sample tubes were refrigerated until used.
Additionally, laboratory tests were conducted on the portable
gas chromatographs to determine instrument sensitivity and
column retention times for the compounds of interest. Calibra-
tion standards of benzene, toluene, chlorobenzene and others
were prepared using a static vaporization technique.
3.3.3.2 Field Activities
Air samples were taken on and around the Upjohn facility
during the period March 24-26, 1981.
3.3.3.2.1 Integrated Air Samples
During the March 1981 study, nine sets of char-
coal and Tenax® samples were collected at nine loca-
tions (Table 3.9). A portable meteorological station
was first established on Upjohn property from which
data on wind speed and direction were obtained
throughout the sampling period. This data, in com-
bination with instantaneous sampling data (See Sec-
tion 3.3.3.2.2), was used to locate the integrated
air sampling station.
3-23
-------
TABLE 3.9
UPJOHN AMBIENT AIR SAMPLING STUDY:
MARCH 1981 INTEGRATED SAMPLE LOCATIONS
Location Station Sample
I.D. Location Description Number Date
Sample Time
MA-1 On Sackett Point Rd. 170°
from Upjohn property* D001
MA-2 80 feet from lagoons on
Upjohn property D002
MA-3 On Upjohn property between
settling ponds
MA-4 In rear parking lot of
UMC bldg. 210° from
Upjohn property*
MA-5 On Sackett Point Rd. 185°
from Upjohn property
MA-6 School Lane
MA-7 On Upjohn property Center
Road and D Street
MA-8 On Upjohn property north
of aeration lagoon
MA—9 In Upjohn administration
parking lot
03/24 Not recorded
03/25 10:40 a.m. - 12:40 p.m.
D003 03/25 11:25 a.m. - 3:30 p.m.
D006 03/25 Not recorded
U008 03/26 9:30 a.m. - 12:00 noon
U010 03/26 Not recorded
D005 03/26 2:30 p.m. - 5:00 p.m.
D007 03/26 8:30 a.m. - 10:00 a.m.
D009 03/26 3:00 p.m. - 4:50 p.m.
* Compass readings are measured from the intersection of Center Road and
B Street on Upjohn property.
3-24
-------
The basic sampling station arrangement was simi-
lar to that described in Section 3.2.3.2 and depicted
in Figure 3-1. One modification to the procedure
was the use of Sierra and Dupont constant flow
pumps which required only an initial and final flow
check. Additionally, all Tenax® and charcoal car-
tridges were stored in the Teflon®-capped, cleaned
glass vials as described in Section 3.3.3.1 and kept
refrigerated at all times from cartridge preparation
until analysis. As in November 1980, all samples
were labeled with date, time, and sample number and
were subject to standard chain-of-custody procedures.
For the purpose of quality control, eight field
blanks were collected which included six Tenax®
sample traps and two charcoal sample traps. Due
to time and equipment constraints, no duplicate
Tenax® or charcoal integrated samples were taken as
originally planned.
3.3.3.2.2 Instantaneous Air Samples
During the March 1981 study, 33 instantaneous
samples were collected at the locations listed in
Table 3.10 (off-site) and Table 3.11 (on-site).
Initially, the portable gas chromatographs were
positioned in a small Upjohn laboratory building
near the wastewater treatment facility. However, due
to room air contamination from the treatment plant,
the instruments were moved to an office in one of
the Upjohn facility buildings sufficiently distant
from the processing area to minimize analytical
equipment contamination. Air samples were taken by
EPA laboratory personnel at preselected locations
by pulling ambient air into syringes fitted with air-
tight valves. The syringes were returned to the GC
and the air samples injected directly into the in-
strument. Print-outs of chromatograms were obtained
within minutes of the sample injection.
By using calibration standards, EPA chemists
were able to establish the types and quantities of
contaminants present in each sample (See Tables 3.10
and 3.11).
For the purpose of field quality control,
approximately one of every five injections was a
standard. Additionally, every tenth sample injected
was a blank. The syringes used for sampling were
cleaned by flushing several times with air filtered
through activated charcoal and subsequently checked
for contamination prior to each use.
3-25
-------
TABLE 3.10
UPJOHN AMBIENT AIR SAMPLING STUDY:
MARCH 1981 OFF-SITE INSTANTANEOUS SAMPLE LOCATIONS
AND ANALYTICAL RESULTS
Location
I.D. Location Description
Pollutant Concentration
Sample Sample (ppb, v/v)**
Number Date Benzene Toluene
MB-1
350 State Steet
AA-1
AA-2
03/24
03/24
K1
K1
1
K1
MB-2 West of processing
area BB 03/24
CC (airbag) 03/24
K1
K1
MB-3
MB-4
Republic Kitchens
DD
On Sackett Point road
170° from Upjohn
property EE
(Same as MA-1)* FF (airbag)
03/24
03/24
03/24
10
6
MB-5 Industrial Complex
160° from Upjohn
property*
MB-6 On Sackett Point Road
195° from Upjohn
Property (Same as
MA-5)*
MB-7 Warehouse west of
Upjohn administra-
tion bldg.
MB-8 East of lagoon/ponds
MB-9 On Humphrey Chemical Co.
property
MB-10 Dump off of Sackett
Point Raod
GG 03/24 5 20
HH-1 03/24 K1 1
HH-2 03/24 K1 ND
II 03/24 K1 1
JJ 03/24 K1 ND
KK 03/24 K1 ND
LL 03/24 2 2
* Compass readings are measured from the intersection of Center Road and
B Street on Upjohn property
** All results reported in parts per billion (ppb) on a volume/volume basis
in ambient air. K1 reads "less than one". ND reads "not detected".
3-26
-------
.tic
D.
¦11
¦12
•13
¦14
¦15
¦16
•17
¦18
•19
¦20
•21
¦22
TABLE 3.11
UPJOHN AMBIENT AIR SAMPLING STUDY:
MARCH 1981 ON-SITE INSTANTANEOUS SAMPLE LOCATIONS
AND ANALYTICAL RESULTS
Location Description
Pollutant Concentration
Sample Sample (ppb, v/v)**
Number Date Benzene Toluene
On Upjohn property
between lagoons
(Same as MA-3)
North of settling
ponds
North of aeration
lagoon
North of aeration
lagoon
D Street opposite
Bldg. #2
Center Road at primary
settling lagoon
C Street opposite Bldg.
#18
East of primary
settling lagoons
North of primary
settling lagoons
B Street opposite
Bldg. #17
B Street at West
Road
C Street at Bldg. #4
26-1
03/26
26-2 03/26
26-3 03/26
26-4 03/26
26-5 03/26
26-6 03/26
26-7 03/26
26-8 03/26
26-9 03/26
26-10 03/26
26-11
26-11
-1
2 03/26
26-12 03/26
In Upjohn administration 26-13-1
West parking lot (Same 26-13-2
as MA-9) 26-13
South of primary
settling lagoons
3 03/26
26-14 03/26
40
L50
L400
K5
20
7
20
10
20
10
7
K2
K2
30
20
10
70
200
K9
80
20
40
20
L200
50
30
3-27
-------
Table 3.11 - Continued
Location
I.D. Location Description
Sample Sample
Number Date
Pollutant Concentration
(ppb, v/v)**
Benzene Toluene
MB-25 East of Upjohn main
gate
26-15 03/26
56
MB-26 On Upjohn property,
north of aeration
lagoon (Same as MA-8) 26-16 03/26
50
90
* High ppm level. This sample was screened on the Century GC and not the
Photovac GC because of the high level.
** All results reported in parts per billion (ppb) on a volume/volume basis
in ambient air. K2 reads "less than two". L50 reads "greater than
fifty". ND reads "not detected".
3-28
-------
3.3.3.3 Analytical Procedures
The Tenax® and charcoal integrated air samples as well as
eight blanks and two back-up traps were analyzed within ten days
of collection at EPA's New England Regional Laboratory in
Lexington, Massachusetts, according to the procedures outlined
in Section 3.2.3.3. In addition to the Tracor 560 gas chromato-
graphy the regional laboratory employed a Photovac 10A10 portable
gas chromatograph with a photoionization detector which had
been used in the field to collect instantaneous samples. More
detail on equipment and operating conditions is presented in
Appendix B.
3.3.4 Results and Conclusions
The analytical results of all integrated air samples collected during
the March 1981 study are presented in Table 3.12. Further, a summary
of all meteorological data collected during the sample period is pre-
sented in Table 3.13. Although some quality control problems were
encountered (See Section 3.3.3.2}, EPA believes that the sampling
and analysis procedures for both Tenax® and charcoal traps were sound
during the March 1981 study. Additionally, the analytical results of
all instantaneous air samples collected during the same period are pre-
sented in Tables 3.10 and 3.11. Basically, the only measurable compounds
in any sample were benzene and toluene.
Table 3.14 has been constructed from Tables 3.12 and 3.13 to provide
a format from which some general conclusions can be drawn on upwind and
downwind pollutant concentrations. For each integrated sample, an
attempt has been made to generally determine whether the sample station
was an upwind or downwind location during sample collection. This is
based on the station location and measured wind direction. Also listed
is the approximate average wind speed during the sampling period, whether
the sample was collected during the day or night and, lastly, the pollu-
tant concentrations based on the average of available Tenax® and charcoal
analytical results.
A close examination of Table 3.14 reveals that upwind levels of both
benzene and toluene were consistent. Benzene concentrations were less
than one ppb and toluene levels were measured at 4 ppb at two sample
locations (ma-5 and MA-6). These results support the theory that the
cleaned sample media used in November were contaminated with low levels
of benzene and toluene which were undetectable on the GC used to check
for cleanliness. Additionally, two instantaneous samples taken at the
MA-5 location (MB-6? HH-1 and HH-2) showed no greater than one part
per billion of either pollutant.
During the March 1981 study, integrated sampling location MA-7 was
positioned between the two major air contaminant emission areas (See
Section 3.2.4). An elevated benzene level (15 ppb) and toluene level
(114 ppb) were encountered. (It should be noted that the 114 ppb toluene
value is the average of 27 ppb detected on the charcoal media and 200
ppb detected on the Tenax media at the same sampling station. Due to
the poor consistency, this data should be used with caution. Although
3-29
-------
TABLE 3.12
UPJOHN AMBIENT AIR SAMPLING STUDY:
ANALYTICAL RESULTS MARCH 1981
INTEGRATED SAMPLES
Station Number
D001
D002
D003
D005
D006
D007
D009
DO 08
D0010
Pollutant Concentrations
on Charcoal Sorbent
(ppb,v/v)**
Benzene
K 1
8
K 1
12
K 1
K 1
K 1
Toluene
2
8
K 1
27
X 1
K 1
K 1
Pollutant Concentrations
on Tenax® Sorbent
(ppb, v/v)**
Benzene
1
10
K 1
18
*
1
2
2
K 1
Toluene
10
31
2
200
*
22
54
7
7
* Sample not collected.
** All results reported in parts per billion (ppb) on a volume/
volume basis in ambient air. K1 reads "less than one".
3-30
-------
TABLE 3 .1 3
METEOROLOGICAL DATA:
MARCH 1981 STUDY PERIOD
Wind Direction* Wind Speed
Day Hour (degrees) (mph)
1 1:01
12:00 noon
360
5
1 2:01
-
1:00 p.m.
300
8
1 s 01
-
2:00 p.m.
90
7
2:01
-
3:00 p.m.
70
4
3:01
-
4:00 p.m.
70
5
4:01
-
5:00 p.m.
360
7
5:01
-
6:00 p.m.
360
4
6:01
-
7:00 p.m.
350
3
7:01
-
8:00 p.m.
350
3
8:01
-
9:00 p.m.
40
3
9:01
-
10:00 p.m.
variable
3
10:01
-
11:00 p.m.
variable
2
1 1:01
-
midnight
variable
2
1 2:01
1:00 a.m.
variable
2
1 :01
-
2:00 a.m.
variable
2
2:01
-
3:00 a.m.
variable
2
3:01
-
4:00 a.m.
variable
2
4:01
-
5:00 a.m.
240
3
5:01
-
6:00 a.m.
240
3
6:01
-
7:00 a.m.
240
3
7:01
-
8:00 a.m.
225
3
8:01
-
9:00 a.m.
200
4
9:01
-
1 0:00 a.m.
225
5
10:01
-
11:00 a.m.
200
5
1 1:01
-
1 2:00 noon
180
5
1 2:01
-
1:00 p.m.
170
7
1:01
-
2:00 p.m.
170
10
2:01
-
3:00 p.m.
170
1 1
3:01
-
4:00 p.m.
180
10
4:01
-
5:00 p.m.
180
9
5:01
-
6:00 p.m.
180
1.0
* 0° is north; 90° east; 180° south; and 270° west.
3-31
-------
TABLE 3.14
SUMMARY OF ANALYTICAL RESULTS AS A FUNCTION
OF METEOROLOGICAL CONDITIONS FOR MARCH 1981
INTEGRATED SAMPLES
Location Station Upwind/
I.D. Number Downwind
Approximate
Wind Speed
(mph)
Pollutant Concentration
Day/ (ppb, v/v)**
Night Benzene Toluene
MA-1
MA-2
MA-3
MA-4
MA-5
MA-6
MA-7
MA-8
MA-9
D001
D002
D003
D006
U008
U010
D005
D007
DO 09
D
D
D
D
U
U
D
D
D
8
5
10
5
10
Day
Day
Day
Day
Day
Day
Day
Day
Day
1
9
K1
K1***
1
K1
15
7
2*
6
20
1
KI***
4
4
114
14
54*
* Only Tenax® data was available for the sample.
** Based on average of available Tenax and charcoal sample results. All re-
sults reported in parts per billion (ppb) on a volume/volume basis in
ambient air. K1 reads "less than one".
*** Only charcoal data was available for the sample.
3-32
-------
a case can be made that the 200 ppb value is uncharacteristically high
for an integrated sample, it will nevertheless be used here as a conser-
vative approach.) Instantaneous samples from four locations (MB-16,
17, 18 and 24) in the same general area confirm the integrated sample
results.
Sampling locations MA-2 and MA-8 were established in the vicinity of
the aeration lagoon. Downwind samples showed 9 and 7 ppb of benzene and
20 and 14 ppb of toluene respectively indicating that the lagoon is a
source of VOC emissions. Instantaneous sample results from three
locations (MB-13, 14 and 26) indicate that contaminant levels can vary
from 6 ppb to 50 ppb benzene and from 20 ppb to 90 ppb toluene.
Location MA-3 was established near the settling ponds and showed
very low VOC levels indicating that these ponds were not a source of
VOC air contamination. Instantaneous samples taken at two locations
(MB-11 and 12) confirm this conclusion.
Integrated sampling location MA-9 was established near the process-
ing area and revealed an elevated downwind toluene level of 54 ppb.
Instantaneous samples taken at six stations (MB-15, 17, 20, 21, 22 and
23) indicated that the processing area benzene levels varied from 7 ppb
to over 400 ppb and toluene levels varied from as low as 20 ppb to over
200 ppb. It can be concluded from this data that the area encompassing
Upjohn Buildings Nos. 2, 3 and 17 was a source of ambient VOC contamina-
tion during the March study period.
Integrated sampling locations MA-1 and MA-4 were established at a
distance downwind of the identified Upjohn air emission sources. Addi-
tionally off-site instantaneous samples were taken at nine distant down-
wind locations. The highest benzene level encountered was 5 ppb and the
highest toluene level was 20 ppb.
3.4 JUNE 1981 STUDY
3.4.1 Introduction
The following is a discussion of the activities associated with
EPA's ambient air sampling study of June 1981. The primary purpose of
this study was to gather air quality data during warm weather by collect-
ing integrated ambient air samples on Tenax® and charcoal media and by
collecting real-time analytical data by direct injection into a portable
gas chromatograph. EPA anticipated that meteorological conditions
during the summer could be different enough from those in November and
March to have an impact on ambient air contaminant levels.
3.4.2 Sampling Strategy
The sampling strategy for the June 1981 program was nearly identical
to that for March 1981. Even greater precautions were taken to minimize
the previously encountered problems associated with Tenax® and charcoal
sample collection media. Further, the sampling strategy included
quality control checks such as duplicate Tenax® and charcoal samples
from at least one location and the collection of field blanks.
3-33
-------
As in the previous two studies, it was decided that integrated air
sampling stations would be located upwind and downwind of the emission
sources, both on Upjohn property and in surrounding neighborhoods.
Portable meteorological equipment would be installed on site to aid in
the positioning of sampling equipment. Additionally, either a portable
photoionization or flame ionization gas chromatograph would be used to
collect real-time data. This data would be used to help locate inte-
grated sampling stations.
3.4.3 Equipment Preparation, Field Activities and Analytical Procedures
3.4.3.1 Equipment Preparation
The major pieces of field equipment which were checked and
used during the June 1981 ambient air study were the same as
those employed during the March 1981 study and are listed in
Table 3.8 of this report (See Section 3.3.3.1). Sample traps
and media were prepared at EPA's New England Regional Labora-
tory as follows:
1. Empty stainless steel tubes were heated to 300°C in a
Century programmed thermal desorber (PTD). Some of
the steel tubes were then checked for cleanliness
with a Century Systems gas chromatograph. (See equip-
ment details in Appendix B.)
2. The 60/80 mesh Tenax® was soxhlet extracted with
methanol for 45 hours followed by pentane for 25
hours. It was then dried under an infrared lamp for
one hour. The traps were then packed with 120 mg of
Tenax® and conditioned at 330°C for 48 hours under
nitrogen at a flow of 50 ml/min.
3. The cocoanut charcoal was ground to 18/40 mesh. After
packing, these traps were conditioned at 280°C and
350°C for 24 hours each under nitrogen flowing at 50
ml/min.
4. The glass wool used in the traps was ultrasonicated
successively with methanol and pentane for 30 minutes
each. It was then baked overnight at 550°C.
5. "Hie traps were individually sealed in 40 ml screw cap
glass vials. The glassware was baked overnight at
550°C. The Teflon® cap liners were rinsed with methanol
and pentane, then dried overnight at 110°C.
6. Each carbon and Tenax® trap was analyzed for benzene
and toluene prior to the Upjohn survey with a GC with
photoionization detector.
7. Sample tubes were refrigerated until used.
3-34
-------
Additionally, laboratory tests were conducted on the port-
able gas chromatographs to determine instrument sensitivity and
column retention times for the compounds of interest. Calibra-
tion standards of benzene, toluene, chlorobenzene and others
were prepared using a static vaporization technique. The
sample integrity of the syringes to be used was checked by
injecting a known level benzene air sample into the syringe
and allowing it to stand for 40 minutes before injecting it
into a GC. Less than ten percent loss was measured.
3.4.3.2 Field Activities
Air samples were taken on and around the Upjohn facility
during the period from June 30 to July 1, 1981.
3.4.3.2.1 Integrated Air Samples
During the June 1981 study, charcoal and/or
Tenax® samples were collected at the twelve locations
listed in Table 3.15. A portable meteorological
station was first established on Upjohn property
from which data on wind speed and direction were
obtained throughout the sampling period. This data
in combination with real-time sampling data (See
Section 3.4.3.2.2) was used to locate the integrated
air sampling stations.
The basic sampling station arrangement was simi-
lar to that described in Section 3.2.3.2 and depicted
in Figure 3-1. As in the March 1981 study, the EPA
used Sierra and Dupont constant flow pumps. Addi-
tionally, following exposure, all Tenax® and charcoal
cartridges were stored in glass vials, labeled,
refrigerated, and subjected to standard chain-of-
custody procedures.
For the purpose of quality control, three field
blanks were collected which included two Tenax®
sample traps and one charcoal trap. Duplicate
Tenax® field samples were taken at one sample station
and duplicate charcoal field samples were taken at
a different sample station.
3.4.3.2.2 Instantaneous Air Samples
During the June 1981 study, 59 instantaneous sam-
ples were collected at the locations listed in Table
3.16 (off-site) and Table 3.17 (on-site).
The portable photoionization gas chromatograph was
positioned in a government vehicle which served as
a portable laboratory. During the collection of most
off-site samples, the vehicle was stationed off
3-35
-------
TABLE 3.15
UPJOHN AMBIENT AIR SAMPLING STUDY:
MARCH 1981 INTEGRATED SAMPLE LOCATIONS
Location Station Sample
I.D. Location Description Number Date
Sample Time
JA-1 On Sackett Point Rd. 195°
from Upjohn property* 1-1
JA-2 Railroad track
intersection near Rte.
5
JA-3 Parking lot near Advanced 1-3
Bearing Company 1-13
JA-4 On Locust Avenue, southeast
of Upjohn property 1-4
JA-5 On State Street at Whitewood
Apartment Complex 1-5
JA-6 On expressway 0° from
from Upjohn property*
JA-7 On Upjohn property between
treatment lagoons
JA-8 Between Rte. 5 and Upjohn
property near City
Print Co.
06/30 10:20 a.m. - 1:50 p.m.
1-2 06/30 2:25 p.m. - 5:35 p.m.
06/30 11:40 a.m.
07/01 2:49 p.m.
8:25 p.m.
5:36 p.m.
06/30 3:15 p.m. - 5:57 p.m.
06/30 10:55 a.m. - 2:21 p.m.
1-6 06/30 2:58 p.m. - 5:58 p.m.
1-7 06/30 9:40p.m. - 12:11 a.m.
1-8 07/01 10:00 a.m. - 1:00 p.m.
JA-9 On Sherwood Drive north
of Upjohn property
1-9 07/01 2:40 p.m. - 5:10 p.m.
JA-10 Immediately west of Upjohn
property near F Street 1-10
JA-11 At Upjohn's Stiles Street
entrance 1-11
JA-12 Immediately east of
aeration lagoon
07/01 9:53 a.m. - 12:53 p.m.
07/01 2:44 p.m. - 5:28 p.m.
1-12 07/01 9:52 a.m. - 12:56 p.m.
* Compass readings are measured from the intersection of Center Road and B
Street on Upjohn property.
3-36
-------
TABLE 3.16
UPJOHN AMBIENT AIR SAMPLING STUDY:
JUNE 1981 OFF-SITE INSTANTANEOUS SAMPLE LOCATIONS
AND ANALYTICAL RESULTS
Location
I.D.
Location Description
Pollutant Concentration
Sample Sample (ppb, v/v)**
Number Date Benzene Toluene
JB-1 Henry's Diner parking
lot
06/30
ND
ND
JB-2 On Sackett Point Road at
Tilco Plant
JB-3 Near Upjohn main gate
JB-4 On Sackett Point Road
195° from Upjohn
property (Same as
JA-1)*
B
C
06/30
06/30
06/30
7
5
ND
ND
21
JB-5 At Upjohn property line
between JB-4 and emission
sources
06/30
10
JB-6 On State Street at White- F-1 06/30
wood Apartment Complex F-2 06/30
(Same as JA-5) F-3 06/30
1
2
2
2
2
2
JB-7 Under railroad bridge 145°
from Upjohn property
(upwind)* G
06/30
JB-8 Parking lot near Advanced 1-1 06/30
Bearing Company 1-2 06/30
(Same as JA-3)
K1
4
4
ND
2
2
JB-9 0.4 miles from overpass
off Route 5
JB-10 0.3 miles from overpass
off Route 5
JB-11 North Haven Train Station
JB-12 Perpendicular to Express-
way (Series Collections,
4 samples collected
simultaneously)
K
L
06/30
06/30
06/30
ND
10
ND
ND
1 2
Farthest from plant M-1 06/30
Next to barrels M-2 06/30
Between barrels and plant M-3 06/30
Closest to plant M-4 06/30
2
27
34
11
11
120
170
76
3-37
-------
Table 3.16 - Continued
Location
I.D.
Location Description
Pollutant Concentration
Sample Sample (ppb, v/v)**
Number Date Benzene Toluene
JB-13 In field west of D
Street
JB-14 In field west of B
Street
JB-15 In field west of
Mministration Building
Parking Lot
JB-16 In field west of
A Street
JB-17 At Stiles Street
entrance
JB-18 In field west of JB-1 4
JB-19 In field west of JB-15
19 07/01
20 07/01
21
22
23
24
25
07/01
07/01
07/01
07/01
07/01
10
***
20
ND
ND
100
ND
100
80
500
ND
ND
100
800
* Compass readings are measured from the intersection of Center Road
and B Street on Upjohn property.
** All results reported in parts per billion (ppb) on a volume/volume
basis in ambient air. K1 reads "less than one". ND reads "not
detected".
*** Laboratory error.
3-38
-------
TABLE 3.17
UPJOHN AMBIENT AIR SAMPLING STUDY:
JUNE 1981 ON-SITE INSTANTANEOUS SAMPLE LOCATIONS
AND ANALYTICAL RESULTS
Location
I.D.
Location Description
Pollutant Concentration
Sample Sample (ppb, v/v)**
Number Date Benzene Toluene
JB-20 Upjohn Administration H-1 06/30
Building west parking H-2 06/30
lot
JB-21 On Center Road opposite
primary setting lagoons 1 07/01
JB-22 On Center Road south of
P Street 2 07/01
JB-23 On Center Road at C
Street 3 07/01
JB-24 On C Street at
Building #18 4 07/01
JB-25 On C Street at
Building #17 5 07/01
JB-26 On C Street at
Building #4 6 07/01
JB-27 On Center Road at
D Street 7 07/01
JB-28 On D Street at
Building #2 8 07/01
JB-29 On D Street at
Building #3 9 07/01
JB-30 On D Street at West
Road 10 07/01
JB-31 On P Street between
West Road and Center
Road 11 07/01
JB-32 Between Building #1 2
and holding pond 12 07/01
JB-33 Between Building #9
and holding pond 13 07/01
2
16
30
ND
40
100
60
40
ND
ND
ND
ND
ND
ND
ND
ND
6
60
ND
80
40
ND
70
10
ND
ND
20
ND
ND
ND
3-39
-------
Table 3.17 - Continued
Location
I.D.
Location Description
Pollutant Concentration
Sample Sample (ppb, v/v)**
Number Date Benzene Toluene
JB-34
Between Building #3
and holding pond
14
07/01
ND
ND
JB-35
Behind Building #5
15
07/01
L100
200
JB-36
At Railroad station
16
07/01
ND
100
JB-37
In Administration
Building parking lot
opposite Bldg. #22
ent.
17
07/01
ND
ND
JB-38
In Administration
Building parking lot
near Stiles St. gate
18
07/01
ND
ND
JB-39
Northwest of
aeration lagoon
26
07/01
10
20
JB-40
Northeast of
aeration lagoon
27
07/01
30
60
JB-41
At A Street and Center
Road
28
07/01
ND
ND
JB-42
On A Street at
Building #16
29
07/01
K10
50
JB-43
On A Street at
Building #17
30
07/01
K10
200
JB-44
On A Street at West
Road
31
33
34
35
07/01
07/01
07/01
07/01
ND
ND
10
ND
L300
500
L300
90
JB-45 South of primary
settling lagoons
32
07/01
L600
200
3-40
-------
Table 3.17 - Continued
Pollutant Concentration
Location Sample Sample (ppb, v/v)**
I.D. Location Description Number Date Benzene Toluene
JB-46 Hoof of Building No. 22
(Series collection, 4
samples collected
s imult aneously)
120
ft.
frcm
S-1701
N—1
06/30
20
30
100
ft.
from
S-1701
N-2
06/30
5
8
80
ft.
frcm
S-1701
N-3
06/30
5
40
60
ft.
from
S-1701
N—4
06/30
20
30
** All results reported in parts per billion (ppb) on a volume/volume
basis in ambient air. K10 reads "less than ten". L100 reads "greater
than one-hundred". ND reads "not detected".
3-41
-------
Route 5 near the intersection of Sackett Point Road.
For on-site sampling, the vehicle was stationed on
Upjohn property sufficiently distant from the process-
ing areas to minimize analytical equipment contamina-
tion. Air samples were drawn by EPA personnel into
syringes and returned to the GC for analysis. The
procedure was similar to that employed during March
1981 study (See Section 3.3.3.2.2).
3.4.3.3 Analytical Procedures
All Tenax® and charcoal samples, as well as three field
blanks, two duplicate field samples and fourteen back-up traps
(6 Tenax®, 8 charcoal) were analyzed within one and one-half
weeks of exposure at EPA's New England Regional Laboratory.
The general procedure is outlined below. More detail on equip-
ment and operating procedures is presented in Appendix B.
1. Each sample trap was desorbed at 200°C in a programmed
thermal desorber and a syringe sample withdrawn.
2. "Die syringe sample was injected into a Tracor 560 gas
chromatograph with photoionization detector.
3. System blanks were run at the beginning of each day and
periodically between samples, especially if a relatively
high level of either benzene or toluene was found.
4. Each desorbed sample was injected into the chromatograph
at least twice to decrease the random syringe error
margin. The resulting values were averaged.
5. Vapor phase standards were run at the beginning,
periodically throughout the chromatographic run, and
at the end of the analysis. The accuracy of these vapor
standards was checked daily for agreement within 10% with
a liquid standard of benzene and toluene in methanol.
6. Several traps were spiked with 2-3 ug of benzene and
toluene in the laboratory. The benzene breakthrough on
the Tenax® was observed at the 30 liter collection volume.
PERCENT RECOVERIES OF THE SPIKED TRAPS
Volume Collected % Recoveries
Trap (Liters) Benzene Toluene
Tenax® 1 85 73
6 66 71
30 90
Charcoal 29 96 71
3-42
-------
3.4.4. Results and Conclusions
The analytical results of all integrated air samples collected during
the June 1981 study are presented in Table 3.18. Further, a summary of
all meteorological data collected during the sample period is presented
in Table 3.19. EPA believes that the sampling and analysis procedures
for both Tenax® and charcoal traps were sound during the June 1981 study.
Additionally, the analytical results of all instantaneous air samples
collected during the same period are presented in Tables 3.16 and 3.17.
Basically, the only measurable compounds in any sample were benzene and
toluene.
Table 3.20 has been constructed from Tables 3.18 and 3.19 to provide a
format from which some general conclusions can be drawn on upwind and
downwind pollutant concentrations. For each integrated sample, an attempt
has been made to generally determine whether the sample station was an
upwind or downwind location during sample collection. This is based on
the station location and measured wind direction. Also listed is the
approximate average wind speed during the sampling period, whether the
sample was collected during the day or night, and lastly, the pollutant
concentrations based on the average of available Tenax® and charcoal analy-
tical results.
A close examination of Table 3.20 shows that upwind levels (JA-4 and
JA-12) of both benzene and toluene were measured at or below one part per
billion during the sampling period. Additionally, an instantaneous upwind
sample was taken at location JB-7 with similar results.
During the June 1981 study integrated sampling location JA-7 was esta-
blished between the two major VOC emission areas (See Section 3.2.4). A
nighttime sample collected under very low wind speed conditions showed
the highest pollutant concentrations of all integrated air samples collected
during the study period (11 ppb of benzene and 8 ppb of toluene). Inte-
grated sampling location JA-10 was established downwind, immediately out-
side of Upjohn property west of F Street. The low VOC levels indicate
that the area southeast of F Street was not a major source of VOC emissions,
integrated sampling location JA-11 was established downwind of the
processing area and showed only low VOC levels (3 ppb benzene and 3 ppb
toluene). These results were quite different from the November 1980
and March 1981 samples taken at the same location which showed low benzene
levels but elevated (30-60 ppb) toluene levels. Possible factors contri-
buting to this difference include unaccounted for meteorological conditions
and chemical process variations.
In addition to taking integrated air samples, an attempt was made by
EPA to better identify and characterize the VOC emission sources through
use of the portable GC. Instantaneous air samples were collected at 46
locations.
3-43
-------
TABLE 3.18
UPJOHN AMBIENT AIR SAMPLING STUDY:
ANALYTICAL RESULTS OF JUNE 1981
INTEGRATED SAMPLES
Station Number
Pollutant Concentrations
on Charcoal Sorbent
(ppb, v/v)**
Benzene Toluene
Pollutant Concentrations
on Tenax® Sorbent
(ppb, v/v)**
Benzene Toluene
1-1
2
2
2
3
1-2
2
K1
2
K1
1-8A
*
*
4
K1
1-8B
*
*
2
1
1-9
K1
K1
*
*
1-5
K1
K1
2
2
1-6
3
3
2
3
1-7
13
5
8
10
1-1 OA
2
K1
*
*
1-1 OB
2
K1
*
*
1-11
4
2
1
3
1-3
*
*
1
2
1-4
*
*
1
K1
1-1 2
K1
K1
K1
ND
1-13
2
K1
2
1
* Sample not collected
** All results reported in parts per billion (ppb) on a volume/volume
basis in ambient air. K1 reads "less than one". ND reads "not de-
tected".
3-44
-------
TABLE 3.19
METEOROLOGICAL DATA:
JUNE 19 81 STUDY PERIOD
Wind Direction* Wind Speed
Day Hour (degrees) (mph)
6/30 10:01 - 1 1:00 a.m. 45 6
11-01 - 12:00 noon 45 6
12:01 - 1:00p.m. Variable 6
1:01 - 2:00 p.m. 180 5
2:01 - 3:00 p.m. 200 4
3:01 - 4:00 p.m. 160 5
4:01 - 5:00 p.m. 120 8
5:01 - 6:00 p.m. 120 5
6:01 - 7:00 p.m. 160 5
7:01 - 8:00 p.m. N/A 0
8:01 - 9:00 p.m. N/A 0
9:01 - 10:00 p.m. N/A 0
10:01 - 1 1:00 p.m. N/A 0
11:01 - 12:00 midnight N/A 0
7/01 11:01 - 12:00 noon 130 8
12:01 - 1:00 p.m. 140 9
1:01 - 2:00 p.m. 140 9
2:01 - 3:00 p.m. 140 8
3:01 - 4:00 p.m. 130 8
4:01 - 5:00 p.m. 130 10
* 0° is north; 90° east; 180° south; and 270° west.
N/a reads "not applicable".
3-45
-------
TABLE 3.20
SUMMARY OF ANALYTICAL RESULTS AS A FUNCTIONS
OF METEOROLOGICAL CONDITIONS FOR JUNE 19 81
INTEGRATED SAMPLES
Approximate Pollutant Concentration
Location Station Upwind/ Wind Speed Day/ (ppb, v/v)**
I.D. Number Downwind (mph) Night Benzene Toluene
JA-1 1-1 D 6 Day 2 3
JA-2 1-2 D 5 Day 2 K1
JA-3 1-3 D 5 Day 1+ 2+
1-13 D 6 Day 2+ 1 +
JA-4 1-4 U 6 Day 1 K1
JA-5 1-5 - Varied Day 2 2
JA-6 1-6 D 6 Day 3 3
JA-7 1-7 - 0 Night 11 8
JA-8 1-8A D 8 Day 4+ K1+
1-8B D 8 Day 2+ 1+
JA-9 1-9 D 8 Day KI++ K1++
JA-10 1-1 OA D 8 Day 2++ K1++
1-1 OB D 8 Day 2++ K1++
JA-111-11 D 8 Day 3 3
JA-12 1-12 U 8 Day K1 K1
** Based on average of available Tenax® and charcoal sample results. All
results reported in parts per billion (ppb) on a volume/volume basis in
ambient air. K1 reads "less than one".
+ Tenax® tribes only
++ Charcoal tubes only.
3-46
-------
Samples from locations JB-28 through JB-34 showed little, if any, VOC
levels which, under the sampling period meteorological conditions, support
the contention that the areas immediately south and southeast of the main
processing area were not a major source of VOC emissions. Sampling loca-
tions JB-39 and JB-40 were established immediately downwind of the aera-
tion lagoon. The results of these samples support the March 1981 study
conclusions that the lagoon is a measurable source of VOC emissions
(benzene levels varied from 10 to 30 ppb and toluene levels varied from 20
to 60 ppb). Instantaneous sampling station JB-21 and JB-45 were located
along the main open sewer line between the process area and the wastewater
treatment facility. Results indicated that this flowing sewer was a
source of VOC emissions.
Thirteen instantaneous air samples (at location JB-23, 24, 25, 26, 35,
36, 41, 42, 43 and 44) indicate that the Upjohn processing area between B
and D Streets and West and Center Roads is a measurable source of VOC
emissions. The benzene level in this area was measured higher than 100
ppb and the toluene level as high as 500 ppb at ground level. Thirteen
more instantaneous air samples taken immediately downwind of the process
area (at locations JB-13, 14, 15, 16, 17, 18, 19, 37, 38 and 46) revealed
the intermittent nature of process area VOC emissions. The benzene level
at these locations varied from undetectable to 100 ppb and the toluene
level varied from undetectable to 800 ppb.
Integrated sampling locations JA-1, 2, 3, 5, 6, 8 and 9 were located
at a distance downwind of the identified Upjohn VOC emission sources.
Additionally, off-site instantaneous samples were taken at twelve distant
downwind locations. The highest VOC levels were detected at instantaneous
sample location JB-12 near the expressway north of the site (up to 34 ppb
benzene and 170 ppb toluene). However, it must be noted that the highest
VOC levels were measured near a barrel storage area not connected with the
Upjohn facility. Material in the barrels may have contributed to the air
contaminant level. Other instantaneous sample results ranged from undetect-
able to 10 ppb for benzene and from undetectable to 21 ppb for toluene.
Integrated sample results ranged from less than one to 4 ppb for benzene
and from less than one to 3 ppb for toluene.
3-47
-------
SECTION 4
DISPERSION MODELING STUDY
4.1 INTRODUCTION
4.1.1 Background
As part of the effort to analyze the health effects of potentially
hazardous emissions of organic compounds from the Upjohn Company's
facility, ES applied computer modeling techniques to estimate long-term
ground level concentrations in the vicinity of the plant. The models
used the average emission rates of benzene and toluene as given in
Table 1.1 and estimated rates of the same compounds as a function of
wind speed for the aeration lagoon.
The objective of the modeling study was to generate air quality data
which could be used with the data obtained through the ambient monitoring
programs described in Section 3 to assess the human health risks associated
with long-term and short-term chemical exposure. The representativeness
of monitoring data is limited by the length of the sampling period and
the specific meteorological conditions that occurred. No such restric-
tions apply to computer modeling. Using appropriate modeling techniques
one can estimate the average concentrations expected over the course of
a year, under a wide range of meteorological conditions.
A major disadvantage of dispersion modeling is that the input data
is generated by vent sampling over a relatively short time period.
Significant changes in process conditions could adversely affect the
usefulness of the computer generated values. However, after a thorough
analysis of Upjohn's process history (see Section 2.3), EPA believes that
the major VOC emission sources had been operating consistently for many
months prior to the source testing period, and therefore, the emission
rate data generated in Section 2.6 of this report and the resultant am-
bient contaminant concentrations generated here are generally represen-
tative of historical conditions in the study area.
To predict ground level concentrations of benzene and toluene near
the Upjohn facility, ES utilized an EPA-approved dispersion program
known as ISCLT, the long-term version of the Industrial Source Complex
model (Reference 2). The ISCLT version uses statistical wind summaries
to calculate annual ground level concentrations on a sector-averaged
basis. It is an advanced Gaussian plume model that requires detailed
information concerning the emission sources, meteorology, and the recep-
tor layout. Sections 4.2, 4.3, and 4.4 provide a summary of the data
used in the analysis, a discussion of key modeling options, and the
results respectively.
4-1
-------
4.1.2 Summary
Dispersion modeling was used to predict long-term (annual) average
concentrations of benzene and toluene in the vicinity of the Upjohn
facility in North Haven, Connecticut. Input data included vent VOC
emission rates derived from sampling, aeration lagoon VOC emission
rates estimated using the Shen technique (Reference 3) , and appropriate
meteorological data. The maximum annual average concentrations of
benzene and toluene were 0.4 and 7 ppb, respectively. These results
must be interpreted in terms of the assumptions that were made concern-
ing the temporal variation of emissions, dispersion characteristics,
and the effects of local terrain and meteorology. Nevertheless, the
values represent estimated upper limits of the concentrations to be
expected.
4.1.3 Conclusions
The following conclusions are drawn based on the work performed
during the dispersion modeling study:
1. ' VOC emission rates from Upjohn's aeration lagoon can be
reasonably predicted by the Shen technique. Uiese rates
are a function of wind speed.
2. Hie maximum predicted off-site annual average benzene con-
centration is 0.4 ppb. This concentration is calculated for
a receptor on the property line to the northeast in the wet-
lands of the Quinnipiac River and is primarily attributable
to emissions from the aeration lagoon. The contribution from
the remaining sources is less than 0.2 ppb regardless of loca-
tion.
3. The maximum annual average toluene concentrations outside
the plant boundaries is 7 ppb, occurring near the express-
way to the north and is due almost entirely to emissions
from stack S-1701. The contribution attributable to other
Upjohn point sources is less than 1 ppb. The maximum contri-
bution from the aeration lagoon to toluene concentrations
outside the plant is 2 ppb, occurring on the property line to
the northeast.
4. Due to modeling methods, all contaminant concentrations have
been conservatively predicted.
5. Due to the consistent operation of the processes and waste-
water treatment facility at the Upjohn Company, it is be-
lieved that the results of the dispersion modeling study are
generally representative of historical plant emissions.
4.2 INPUT DATA
During the source sampling and analysis phase of the project in
November 1980, 17 sampling points were tested representing 20 emission
4-2
-------
sources as explained in Section 2. Average emission rates of benzene,
toluene, and chlorobenzene were determined from the concentrations measured
in the exhaust gas stream, and the flow rates estimated from actual stack
gas velocity data, blower nameplate data, and process knowledge. This
study focuses on the dispersion of the VOCs benzene and toluene because
these were the organic compounds emitted at the highest rate and most
measurable in the ambient air on and surrounding the Upjohn facility.
The predicted ambient concentration for any other VOC discussed in Section
2.5.2.5 of this report would generally be proportionally less than that
for benzene and toluene. This may also be true for the compounds given
in Table 1.2. However, the condensible nature of these compounds suggests
that they would not travel far from their emission points. Other stack
parameters required for dispersion calculations (stack exit diameter,
release height, and exhaust temperature) were measured directly or obtained
from Upjohn engineers. A plot plan of the facility provided the relative
location of each emission source. Table 4.1 summarizes the input data
of the vent sources at the Upjohn facility.
Other potential sources of organic emissions identified are the
settling ponds and aeration lagoon. Although air samples were collected
on the surface of these lagoons as part of ES's source sampling study,
volatile organic emissions could not be adequately assessed using the
analytical protocols chosen for that program (see Section 2.5.3.3). An
alternative approach was devised which employed the air quality data
generated by EPA (see Section 3) and the technique of Shen (Reference 3).
Since EPA ambient sampling showed no appreciable VOC levels downwind
of the settling ponds, it is assumed that they have a very low emission
rate and need not be considered in this modeling study.' This is consistent
with the facts that the settling ponds are not aerated and are also used
only for emergency bypass of Upjohn's wastewater treatment facility.
To estimate emissions from the aeration lagoon ES used the technique
of Shen, in which the volatilization rate is calculated from the molecular
weight of the compounds of interest, the concentration of the compounds
in the wastewater, and an overall mass transfer coefficient. The emission
rates obtained are a function of wind speed and are presented in Table
4.2 for six wind speed classes. Volatilization rates are affected by
meteorological conditions occurring at the gas-liquid interface, by the
presence of floating materials, and the amount of turbulence, all of
which may vary with time. Nevertheless, the values are the best estimates
that could be obtained. A more detailed discussion of the technique and
the parameters used in the calculations are given in Appendix C.
To support the results of the Shen technique, EPA conducted a model-
ing exercise to derive lagoon emission rates of benzene and toluene by
comparing ambient monitoring data (see Section 3) to concentrations pre-
dicted by the model under the same meteorological conditions. The ambient
monitoring data was taken from integrated sampling stations located
downwind, in the vicinity of the aeration lagoon. For each pollutant at
each station an emission rate was calculated by dividing the observed
concentration by the computer predicted concentration for an assumed
emission rate of one u«? m~2 sec"1 .
4-3
-------
TABLE 4.1
EMISSIONS AND SOURCE CHARACTERISTICS FOR THE OPJOHN FACILITY
Source Characteristics
Location3 Elev. Average Emissions (g/hr) Height
(b
»)
(m,
(m,
Temp
Veloc.b
Diai
Ident.
X
y
MSL)
Benzene
Toluene
Chlorobenzene
AGL)
(°K)
(m/sec)
(m
S-202
5
-216
5
12.2
6.4
294
.078
S-203
7
-218
5
0.047
0.134
0.495
9.5
294
.051
S-205
9
-220
5
30.9
8.9
294
.062
S-206
36
-213
5
50.8
1 .23
0.4131
16.2
294
2.0
.152
S-21 3
12
-222
5
20.8
.00042
7.2
294
.062
S-214
43
-216
5
53.6
2.48
15.3
294
1 .2
.152
S-301
-12
-234
5
59.5
17.7
294
2.9
.915
S-303
-12
-254
5
15.2
312
.591
K-316
-14
-239
5
0.105
0.0323
3.5
294
.610
T-393/
394
0
-249
5
0.102
3.5
294
.610
T-395/
396
4
-249
5
4.8
294
.610
S-1701
57
-183
5
7040.0
13.3
294
11 .7°
.702
S-1703
67
-177
5
17.3
17.3
294
10.9
.305
K-1701
42
-174
5
1.78
0.0967
11.2
310
.127
K—1702
44
-170
5
1 .78
0.0967
11 .2
310
.127
K—1703
39
-1 69
5
1 .78
0.0967
11 .2
310
.127
K-1704
38
-172
5
1 .78
0.0967
11 .2
310
.127
K-1705
35
-1 67
5
1.78
0.0967
11 .2
310
.127
K-1706
33
-170
5
0.0103
4.88
0.0174
11 .4
310
.138
T-1718
41
-1 52
5
0.114
56.8
7.8
310
.127
a
b
c
Relative to plant's main entrance.
Blank denotes less than 0.1 m/sec, zero used in calculation of plume rise.
Stack has "rain cap", zero used in calculation of plume rise.
-------
TABLE 4.2
EMISSION RATES FOR AERATION LAGOON3
conc. in Emissions (g/m^/hr)a
Wastewater^ Wind Speed (m/sec)c
Pollutant (ppb) 0.75 2.5 4.3 6.8 9.5 12.5
Aeration Lagoon
Benzene 30 .006 .014 .020 .027 .034 .040
Tbluene 200 .037 .084 .120 .164 .205 .246
a Using technique of Shen (Reference 3).
k Surface area of aeration lagoon is approximately 6,500 m^.
c Wind speeds correspond to mean speeds of the six speed categories
used by the National Climatic Center (NCC) in classification of
meteorological data,
d Data obtained by EPA in March 19 81.
4-5
-------
Legend
A 0007
X B2010
P 8182
El 8192
*
A X 8198
* X 0
1 1 1 1 1 1 1
2 4 6 8 10 12 14
WIND SPEED (m sec"')
8196
8198
~ Shan Vatuas
Figure 4-1
Comparison of Benzene Emissions from Aeration Lagoon
Calculated from Ambient Air Monitoring Data to Those
Predicted by Shen Technique.
-------
70n
I
-J
eo
so
o
« 40
H
E
o>
3 30
o
20
10
X
/
/
/ *
X A
/
/
*
/
/ u
/
~r~
2
1 1 r~
6 a to
WIND SPEED (m sec"')
i
12
14
Legend
A P007
X 82010
P 8182
H 8192
H §196
X 8198
~ Shen Value*
Figure 4-2 Comparison of Toluene Emissions from Aeration Lagoon
Calculated from Ambient Air Monitoring Data to Those
Predicted by Shen Technique.
-------
The derived emission rates for benzene and toluene are plotted as
a function of average wind speed for six downwind ambient sampling sta-
tions in Figures 4-1 and 4-2. These figures also display the emission
rates (converted to units of yg m~2 sec"1) which were generated by the
Shen technique. A comparison of the derived emission rates with the
Shen values reveals that the estimates are always well within an order
of magnitude. This degree of agreement might be considered fairly good
given the level of uncertainty in some of the meteorological variables
and in the ambient air quality data. Additionally, the comparison of the
derived emission rates and the Shen values reveals that the latter are
generally higher, especially for toluene. Therefore, the use of the Shen
values represents a conservative approach which is desirable when dealing
with health effects. A more detailed discussion of the modeling proce-
dures and input parameters is presented in Appendix D.
The meteorological data required by the ISCLT model consists of a
joint frequency distribution of wind speed and direction categories clas-
sified with respect to atmospheric stability. These distributions are
compiled by the National Climatic Center (NCC) from data observed at
airports and National Weather Service stations across the country. Al-
though raw data records are available for the airport at New Haven, obser-
vations of wind speed, wind direction, and cloud cover (used in the clas-
sification of atmospheric stability) are recorded only four times a day
during daylight hours. These records were judged by NCC to be insufficient
to provide a representative summary of meteorological conditions around
the clock. A frequency distribution constructed from five years of data
(1965-1969) recorded at Bridgeport (approximately 20 miles to the south-
west) was therefore used in the analysis with the approval of the EPA
regional meteorologist. The frequency distribution of wind direction and
speed at Bridgeport appears in Figure 4.3.
Other user-specified meteorological data requirements include an
estimate of average mixing height and temperature. Because of the low
release heights and the lack of thermal plume rise at most sources in
this study, the model is not sensitive to these parameters; values of
1000 meters and 293°K were used. Also, because no better information
was available, ES used documented default values for vertical gradient
of potential temperature, air entrainment coefficients, and wind speed
power law coefficients.
4.3 DISCUSSION OF KEY MODEL OPTIONS
Preliminary estimates of ground level concentrations using the pro-
cedure of Turner (Reference 4) indicated that maximum values would occur
well within a 1 km radius due to relatively low emission release heights.
Therefore, a receptor grid was designed to include all territory within
a 1 km radius of the processing area. Receptors were located in a Carte-
sian grid format with a spacing of 200 meters. Discrete receptor points
around the plant boundary, along the expressway to the north, and along
the railroad to the east were also used. These had a spacing of about
15 meters.
4-8
-------
LEGEND
0-6 Knots
7-10 Knots
11-16 Knots
17-21 Knots
>21 Knots
Figure 4-3 Frequency Distribution of Wind Direction
and Speed, Bridgeport, CT 1965-1969.
-------
Complicating factors in the analysis were that the sources emitted
in close proximity to buildings and at a very low level with respect to
surrounding terrain, and that emissions from many of the sources were
variable and non-continuous. The following discussion describes how ES
addressed these problems.
Although there is a model option for the consideration of terrain, it
requires that all receptor elevations be less than the top of the lowest
stack modeled. This condition could not be met in the study area because
some receptors were located at sites with elevations that actually exceeded
the release heights of some sources. To handle this situation, a three-
step procedure was used. First, sources greater in height than the ter-
rain within the study area were modeled using the terrain option. Second,
low sources were modeled without regard for terrain. Basically, this was
equivalent to assuming that the plume centerline remained a constant dis-
tance above the ground. Third, contributions from all sources were summed
at each receptor point.
This approach to consideration of terrain effects is appropriate for
a number of reasons. First, EPA in the VALLEY, COMPLEX I, and COMPLEX II
screening techniques does not allow the plume centerline to approach ground
level receptors closer than 10 m. Second, the vertical dispersion coeffi-
cient, a z, is equal to 10 m beyond 600 m under all stability conditions.
Thus, the ground level concentrations predicted at such distances would
constitute a significant percentage of the plume centerline. Finally,
previous studies have indicated that in areas of complex terrain both the
horizontal and vertical dispersion coefficients, a y and a z, are enhanced.
In fact, all of the EPA screening techniques mentioned above allow for
enhancement of one or the other of these coefficients. Since the proce-
dure used in this analysis does not include the enhancement of any of
these coefficients, plume centerline concentrations and, as a result,
ground level concentrations predicted by the model would be higher than
might be expected. Thus, this procedure is a conservative treatment of
the terrain effects.
An option that was not exercised was the consideration of building
wake effects. This option was not used for several reasons. First, ES
attempted to predict conservative concentrations and the wake effects
option generally tends to decrease downwind concentrations due to
increased dispersion near the source. This increased dispersion, however,
can cause, in the immediate vicinity of the building, greater concentra-
tions than those that would occur had there been no building. But in a
comparison of model results using stack S-1701 (the primary contributor
to predicted ambient concentrations of toluene), increased concentrations
due to modeled wake effects were predicted only within 200 meters of the
stack under most conditions. Under very stable conditions this distance
increased to 350 meters, but the annual frequency of this condition is
only 6.5%. Most of the area that could be affected by these increased
concentrations is located within the plant boundaries.
Second, the ISC users guide states that the wake effects evaluation
procedures may not be strictly applicable to all situations because the
data upon which the procedures were formulated reflect a specific stabil-
ity, building shape and building orientation with respect to the mean wind
direction. Also, the procedures use a single effective building width
4-10
-------
(equal to the diameter of a circle with the equivalent area of the build-
ing) for all wind directions. As stated in the users guide, the use of
one value of building width for squat buildings (building width greater
than building height) with plume height to building height ratios less
than about 1.2 affects the accuracy of the calculations near the source
if the building length is large in comparison with the building width. At
Upjohn, the lengths of buildings 2, 3 and 17 (where the stacks are located)
are greater than 2 times the building widths. The procedures are most
appropriate for stacks in the center of the building. At the Upjohn
facility, however some of the stacks are located to the side of che build-
ing. In some cases the stack tops are actually lower than the building
height. Considering all the complications, the non-use of the wake effects
option to obtain conservative downwind concentrations appears justified.
To satisfy model requirements, point source emissions were entered as
continuous at the average rate. This assumption is reasonably valid in
that the emissions are not expected to be strongly correlated with mete-
orological conditions. If emissions were not random and occurred regular-
ly at a specific time of day, however, this assumption could be weakened
because meteorological conditions frequently vary with time of day. For
example, neutral atmospheric stability (Category D) occurs more frequently
during the night. Also, in a valley situation, night-time drainage winds
may influence wind direction frequencies. To properly address the problem
of emission variation and the effect upon long-term ambient concentrations,
records of emission releases and the accompanying meteorological conditions
would be required to model short-term concentrations, which would then be
averaged over an entire year, something beyond the scope of this study.
4.4 RESULTS
The maximum predicted off-site annual average benzene concentration
was 0.4 ppb (Figure 4-4). This concentration was calculated for a receptor
on the property line to the northeast in the wetlands of the Quinnipiac
River and was primarily attributable to emissions from the aeration lagoon.
The contribution from the remaining sources was less than 0.2 ppb regardless
of location.
For toluene, the maximum annual average concentrations outside the
plant boundaries was 7 ppb, occurring near the expressway to the north
(Figure 4-5). In this case, the concentration was due almost entirely to
emissions from stack S-1701, by far the primary emitter of toluene. The
contribution here attributable to other vent sources was less than 1 ppb.
As for the aeration lagoon, the maximum contribution to concentrations
outside the plant was 2 ppb, occurring on the property line to the northeast.
As with any model, final results are open to a certain degree of in-
terpretation. In this case, various assumptions were made that may or
may not reflect actual conditions. Already mentioned was the variability
in emission rates with time and the fact that emission rates at the aera-
tion lagoon were estimated. Other factors to consider are the representa-
tiveness of the meteorological data and the performance of the model in
predicting concentrations when emissions occur at a low height in the
vicinity of buildings.
4-1 1
-------
-/I* / ^ r.vrn^' ^,f '- F" :£
•>. * -:^=4s.: i )/ -L-st- - ¦'/. .jSr" /?•••• ..• j -7*^-1-'mm, i If --
Vi-*- Y"?=. j / <-/ ¦ /'xtmk
! if ,¦' r^L)
jIf
- Jir ^-s vf^ L' "CF
ClSOQ
iSSW'k
¦ '* ••• - K / I! ,*N^d' . *
iJ%>/ «
vj/»<
Sandpit
/*
Library,^
VUW.
tlond
00C FiET
: kilometef
Figure 4-4 Predicted Annual Average Benzene Concentrations
(ppb) in the Vicinity of the Upjohn Facility
4-12
-------
KILOMETER
Figure 4-5 Predicted Annual Average Toluene Concentrations
(ppb) in the Vicinity of the Upjohn Facility
4-1 3
-------
Although the Bridgeport meteorological data were measured at a dis-
tance of 20 miles from the Upjohn plant, the five-year frequency distri-
bution is expected to reasonably reflect the major components of the
macroscale wind field. However, the use of Bridgeport meteorological
data may not be strictly appropriate because of microscale differences
between the two sites. Bridgeport is located near the Long Island Sound
and may be influenced by sea breeze circulations set up during warm days
due to the differential heating of land and sea. North Haven is located
10 miles inland and may not be affected to the same degree. Also, the
terrain surrounding North Haven may have a modifying effect on wind di-
rection and wind speed due to channeling and surface roughness. As a
result of these influences, the location of the maximum annual average
concentration may be different than that predicted. The magnitude of
the annual average, however, is not expected to be substantially diffe-
rent.
Downwind concentrations are influenced by the amount of initial dis-
persion at the source. Most sources in this study are vents on the roof
of various buildings. One source actually vents to the inside of a build-
ing. The amount of initial dispersion at these sources is expected to
be greater (and downwind concentrations less) than for classical stack
(point) sources because of building wake effects. Although it has been
shown that the building wake effects option does increase downwind concen-
trations close to the stack, ES chose not to exercise the option because
of implicit uncertainties in predicting these concentrations. ES chose
to maximize concentrations further downwind where estimates are more
reliable.
If building wake effects were considered, they would probably be most
important for source S-1701, the primary emitter of toluene. The point of
maximum annual average toluene concentrations is slightly greater than 200
meters downwind of source S-1701. Without consideration of building wake
effects, this source is estimated to contribute 6.5 ppb of the 7 ppb total.
Additionally, ES employed the short-term version of the ISC model to spec-
fically assess building wake effects. Under extremely stable conditions,
and with consideration of building wake effects, short-term concentrations
reach a value 14 times the value estimated without consideration of build-
ing wake effects. The occurrence of stable conditions regardless of wind
direction and speed is only 6.5%. Likewise, under slightly stable condi-
tions, the model predicts short-term values 1.4 times higher with considera-
tion of wake effects, and these conditions occur 13.4% of the time. Under
all other meteorological conditions, or about 80% of the time, the model
in the wake effects mode predicts short-term values up to 25% less than
those predicted without wake effects. Taken together and averaged, these
differences result in an annual average contribution from S-1701 of only
2.8 ppb.
For benzene, differences are less important because the point sources
contribute so little to ambient concentrations compared to the contribu-
tion from the aeration pond. In addition, the release height of some ben-
zene sources is lower than that for source S-1701. The area of increased
concentrations due to wake effects would therefore be even smaller for
these sources than it would be for S-1701.
4-14
-------
SECTION 5
HEALTH EFFECTS STUDY
5.1 INTRODUCTION
5.1 .1 Background
The scope of the work is an evaluation of the health effects of the
organic chemical emissions from the Upjohn Company Fine Chemicals
Division, North Haven, Connecticut.
The following information was used in performing the evaluation:
The draft final report on EPA's Ambient Air Quality Study at the Upjohn
Chemical Company, (see section 3 of this report); GCA Corporation's
Final Report on the Organic Emissions Assessment Program, Upjohn Company,
(see Section 2 of this report); Engineering Sciences' Report on Dispersion
Modeling of Volatile Organic Compound Emissions at the Upjohn Facility
in North Haven, (see Section 4 of this report); SRI International's
report on Definition of Population-at-Risk to Environmental Toxic
Pollutant Exposures, Volume II - Appendices A-C, October, 1980; the
Carcinogen Assessment Group's (EPA) Final Report on Population at Risk
to Ambient Benzene Exposures (see Appendix E of this report); and
relevant publications in the scientific literature.
It must be emphasized that modeling data predict the level of
exposure to organic contaminants derived solely from the Upjohn facility.
Such predicted exposures would be in addition to exposures from other
sources. In contrast, ambient air measurements define the levels of
a given substance at a particular point in time at a specific location.
These levels will be the sum of contributions from a variety of sources.
Benzene, toluene and chlorobenzene were the dominant volatile
organic compounds (VOCs) measured by ES in the Upjohn process vent
emissions (See Section 2). Other VOCs were identified in significantly
lower concentrations. Benzene is the only VOC identified that is a
recognized human carcinogen. Examination of the list of condensable
organic compounds identified in vent emissions revealed the presence of
chlorinated adjuncts of benzene, phenol, and aniline. Ambient sampling
by EPA (see Section 3) on and surrounding the Upjohn facility focused
on the measurement of VOCs. Only benzene and toluene existed at measur-
able levels.
5-1
-------
5.1 .2 Summary
An evaluation was made of the impact of the Upjohn Company emis-
sions on the health status of residents of North Haven. EPA's instan-
taneous and 3-hour integrated sampling measurement data were used to
predict effects of short-term exposures. Estimated annual average
concentrations generated by ES from dispersion modeling data were used
to predict effects of long-term exposures. Published information in
the scientific literature was used to evaluate the effects of short-and
long-term exposures to toluene and condensible organics and short-term
exposures to benzene. Since benzene is a leukemogen, the Carcinogen
Assessment Group's model (Appendix E) was used to estimate the number
of benzene-induced leukemia deaths from long term exposure to Upjohn
emissions and 'urban factors' such as automobile and gasoline station
emissions. The relative leukemia risks to residents of North Haven and
the U.S. population as a whole were generated and discussed.
5.1.3 Conclusions
Based on the results of this assessment, a number of conclusions
can be drawn concerning the health effects of organic compound emissions
derived from the Upjohn facility:
1. The ambient levels of toluene and benzene as measured via in-
tegrated 3-hour and instantaneous samples taken by EPA result
in short-term exposures which are unlikely to produce non-car-
cinogenic acute health effects.
2. The annual average concentrations of toluene outside of the
Upjohn property as predicted by ES via dispersion modeling
(maximum of 7 ppb) are unlikely to produce health effects.
3. Several condensible organic compounds were identified in the
Upjohn vent emissions, some in concentrations similar to ben-
zene and toluene. These compounds were not measurable by the
ambient sampling methods employed by EPA. Due to the nature
of these compounds, it is unlikely that they would be present
in ambient air far from their emission points. Further, due
to the very low odor thresholds of some of the compounds, they
would be detectable by smell at concentrations far below those
which might cause health effects. None of these compounds are
recognized human carcinogens. Compounds of this type tend to
be skin irritants but are not particularly toxic systemically.
4. Exposure to ambient benzene increases the risk of leukemia. In
the United States as a whole, the expected number of leukemia
deaths per year from 'urban factor' benzene exposures is 57.87.
The 'urban factor' exposure for the New Haven metropolitan area
is much (94%) lower than for the U.S. as a whole. If the U.S.
population were exposed to 'urban factor' benzene at the same
level as the population of North Haven, the expected number of
benzene caused leukemia deaths per year would not exceed 3.73
because the urban benzene exposure in New Haven is 94% lower
than the national average. Therefore, the risk ratio of the
5-2
-------
residents of North Haven to the U.S. for 'urban factor'
benzene exposure is 0.06:1.
5. In the U.S. as a whole, the expected number of leukemia deaths
per year from general chemical manufacturing benzene exposures
is 2.88. Exposure to benzene from the Upjohn facility in-
creases the leukemia risk to certain residents of North Haven.
As shown in Figure 4-4, predicted annual average benzene con-
centrations outside of the Upjohn facility range from less
than 0.03 to 0.4 parts per billion (ppb). If the U.S. popula-
tion were exposed to chemical manufacturing benzene at the
worst case level of 0.4 ppb, the expected numbers of benzene
caused leukemia deaths per year would be 0.99. Therefore, the
risk ratio of the residents of North Haven to the U.S. for
chemical manufacturing leukemia is 0.34:1.
6. As discussed in 4 above, if the U.S. population were exposed
to 'urban factor' benzene at the same level as the population
of North Haven, the expected number of benzene caused leukemia
deaths per year would not exceed 3.73. Further, as discussed
in 5 above, if the U.S. population were exposed to the worst
case Upjohn generated benzene level of 0.4 ppb, the expected
number of leukemia deaths per year would be 0.99. Therefore,
those residents of North Haven exposed to 0.4 ppb benzene
from Upjohn incur a leukemia risk 1.27 ( 3.73 + 0.99 ) times
3.73
greater than those exposed to 'urban factor' only. This factor
decreases to 1.03 for residents exposed to 0.05 ppb of benzene.
7. The risk of benzene induced leukemia in residents of North
Haven increases if a significant part of the day is spent in a
high population density central city location (e.g. working in
New York City).
5.2. HEALTH EFFECTS
5.2.1. Health Effects of Toluene
Inhaled toluene is a narcotic to humans and as such produces a
variety of effects such as altered psychomotor performance, irritability,
disorientation and unconsciousness. Symptoms such as fatigue, weakness
and confusion have been reported in humans exposed to 200 to 300 ppm in
air for eight hours (Reference 5). Higher concentrations produce more
severe symptoms. The threshold for mild subjective complaints for some
subjects is 50 ppm. Other than central nervous system depression, there
is no evidence that acute or chronic exposure to toluene produces speci-
fic target organ toxicity. However, animal studies have shown that
toluene alters benzene disposition with the amelioration of benzene
toxicity (Reference 6).
5-3
-------
5.2.2 Health Effects - Other Compounds
Several condensible organic compounds were identified in the Upjohn
emissions (Section 2) some of which existed in concentrations similar to
benzene and toluene. These include the chlorinated adjuncts of aniline
and phenol. None of these compounds have been identified as a human
carcinogen. Compounds of this type tend to be skin irritants but are
not particularly toxic systemically. Some of the compounds examined
have very low odor thresholds, e.g. 1-chloro-3-nitrobenzene and 2,4,6
trichlorophenol have odor threshold of 3 ppb and 1.2 ppb respectively.
These concentrations are far below those which might cause other non-car-
cinogenic effects.
5.2.3. Health Effects of Benzene
Acute exposures to benzene via inhalation results in effects which
are primarily on the central nervous system. In man, exposure to 20,000
ppm is fatal within 5 to 10 minutes (Reference 7). Death is due to
respiratory failure and circulatory collapse. Some deaths have been
reported to occur at concentrations of between 100 and 200 ppm (Refer-
ence 8). However, co-exposure to other agents may also have occured in
these cases. Other symptoms of benzene intoxication are drowiness, diz-
ziness and headache. These symptoms may persist for two to three weeks
following a single exposure to a sublethal concentration of benzene.
Approximately 50 percent of inhaled benzene is absored by the lungs and
retained by the tissues, (Reference 9) although this may increase as
the amount of adipose tissue increases. Thus individual responses to
fixed concentrations of benzene vary considerably.
In chronic benzene exposure, central nervous system effects may
occur but the most important toxic effects are related to the hematopoe-
tic system. Blood abnormalities which include myelocytic anemia, throm-
bocytopenia, and leukopenia have been reported in individuals occupation-
ally exposed to concentrations of 20-60 ppm (References 10, 11). The
development of blood abnormalities can occur in some individuals fol-
lowing cessation of benzene exposure.
There have been several case reports and epidemiological studies
linking benzene exposure with leukemia (References 12-17). It must be
emphasized that these studies were on occupationally exposed individuals
who may also have been co-exposed to high concentrations of other sol-
vents. However, the onset of leukemia is usually preceeded by many
observable effects on the hematopoetic system (Reference 9). Further-
more, it is not known whether benzene per se or a metabolite of benzene
is responsible for any of the chronic health effects.
Controlled experiments on animals have produced conflicting re-
sults. Benzene has shown no evidence of carcinogencity when tested in
mice by skin applications. Injection of benzene into laboratory animals
to produce leukemia gave inconclusive results (Reference 20 21). Ad-
ministration of benzene to rats by stomach tube resulted in carcinoma
of the zymbal gland, and enhancement of the mammary carcinoma and leu-
kemia (Reference 22). Also, one inhalation study by Snyder et al.
5-4
-------
(Reference 23) found an increased hematopoietic neoplasms in treated
over control C57 BL/GS mice.
The Carcinogen Assessment Group (CAG) of EPA used the data from
the epidemiological studies of Infante et al (Reference 13), Ott et al
(Reference 14) and Askoy et al (Reference 16, 17, 24) and a linear
non-threshold model to estimate the leukemia risk to the low average
levels of benzene to which the general population is exposed. Details
of the mathematical development of this model are given in Appendix E
of this report. Application of this model permits an estimation of the
carcinogenic hazard from exposure to benzene in the entire United States
population. In order to do this, lifetime average benzene exposures
were estimated and were combined with the non-threshold linear model of
risk as a function of lifetime exposure. The results demonstrate that
the expected number of identifiable source specific benzene caused leu-
kemia deaths per year in the United States is 61.67 out of a total of
14,417 or 0.43 percent (See Appendix E). Integration of exposures from
all sources increased the estimated benzene caused leukemia deaths per
year to 89.8 or 0.62 percent (See Appendix E).
5.3 IMPACT OF UPJOHN CHEMICAL COMPANY EMISSIONS ON RESIDENTS OF NORTH HAVEN
5.3.1 Short-Term Exposures - Benzene and Toluene
The highest off site instantaneous measurements revealed concentrations
of 100 ppb and 800 ppb respectively for benzene and toluene (Section 3).
These concentrations are considerably higher than those found in residental
areas and are much lower than the concentrations reported to produce non-
carcinogenic acute effects in man. Thus residents of North Haven are
unlikely to experience non-carcinogenic acute effects at these concentra-
tions .
5.3.2 Long-Term Exposures - Toluene
The predicted maximum annual average concentration of toluene
outside of the Upjohn Company plant boundaries is 7 ppb (Section 4).
There are no known effects on human health resulting from prolonged
exposure to such a low toluene concentration. Thus, residents of North
Haven are unlikely to suffer adverse health effects from exposure to
toluene resulting from Upjohn Company emissions.
5.3.3 Long-Term Exposures - Benzene
Because benzene is considered to be a leukemogen, the probability of
residents of North Haven developing leukemia as a result of exposure was
assessed. To do this, a strategy of comparing benzene exposures to
residents of North Haven with exposures in the United States population
as a whole was developed and is outlined below.
Studies by Mara and Lee (Reference 25), show that portions of the
United States population are exposed to the following sources of benzene
in the ambient air:
5-5
-------
a) Chemical manufacturing
b) Petroleum refineries
c) Coke ovens
d) Automobile emissions
e) Gasoline service stations
f) Self-service gasoline
Categories d, e, and f may be accumulated in a single category of
'urban area1 exposures. The distribution of the United States population
exposed as a function of benzene concentration and by source category is
given in Table 5.1 (Reference 25). The Carcinogen Assessment Group's
estimates of source specific benzene caused leukemia deaths per year in
the United States population (Appendix E) is given in Table 5.2. It can
be seen that for the United States as a whole, exposures from urban
area factors, mainly automobile emissions, account for most of the estimated
benzene-caused leukemia deaths per year.
Exposures from coke ovens and petroleum refineries are not source
specific categories of benzene to which the population of North Haven is
exposed. Thus chemical manufacturing and 'urban areas' exposures are
the major sources of benzene exposure. Thus in order to determine the
impact of the Upjohn Company emissions on the leukemia risk to residents
of North Haven it is necessary to partition exposures due to Upjohn
emissions and exposures due to urban factors.
5.3.4 Estimated Leukemia Risk from Exposure to Upjohn Company Emission
Based on modeling data obtained from measurement of benzene in
Upjohn Company emissions determined in November 1980 (Section 3) the
residents of North Haven would be expected to be exposed to an annual
average benzene concentration of 0.03 to 0.4 ppb. Use of these data
assume that the processess occuring in November 1980 are similar through-
out the year and for several years. In the United States as a whole,
an estimated 7,300,000 people are exposed to benzene resulting from
chemical manufacturing (Table 5.1). The distribution of the exposed
population by annual benzene concentration is as follows: 0.1-1.0 ppb
(82%): 1-4 ppb (13.7%); 4.1-10 ppb (2.7%); >10 ppb (1.1%).
In order to apply the Carcinogen Assessment Group's model for
leukemia risk, it is necessary to calculate the source specific
exposures in ppb-person years. The total number of people in the
United States exposed to benzene from chemical manufacturing is 7,300,000
and the estimated total exposure is 8.5 x 106 ppb-person years (Table
5.1). This was calculated by multiplying the number of persons exposed
by the estimated median annual average benzene concentration from chemical
manufacturing (Reference 25) as follows:
5-6
-------
TABLE 5.1
SUMMARY OP ESTIMATED POPULATION EXPOSURES TO
ATMOSPHERIC BENZENE FROM SPECIFIC BENZENE EMISSION SOURCES3
Number of People Exposed to Benzene Concentrations (ppb)
Source 8-hour Worst Case: 2.5-25.0
Annual Average: 0.1-1.0
Chemical Manufacturing 6,000,000
Coke Ovens 300,000
Petroleum Refineries 5,000,000
Solvent Operations
Storage & Distribution
of Gasoline
Automobile Emissions -
urban 69,000,000
Gasoline Service Stations -
urban 30,000,000
People Using Self-service
Gasoline
2.5.1-100.0
1.1-4.0
1,000,000
3,000
45,000,000
2,000,000
100.1-250.0
4.1-10.0
200,000
250.0
10.0
80,000
Total
7,300,000
300,000
5,000,000
110,000,000
32,000,000
37,000,000
Comparison
Among Sources
(106 ppb-
person-years)
8.5
0.2
2.5
150.0
19.0
1.6
Source: (Reference 25)
-------
TABLE 5.2
SOURCE SPECIFIC BENZENE CAUSED LEUKEMIA DEATHS/YEAR
BASED ON TABLE 5.1a
Source of Exposure
Chemical Manufacturing
Coke Ovens
Petroleum Refineries
Automobile Emissions
Gasoline Service Stations
Self Service Gasoline
Total
Exposure in
106 x ppb person years
8.5
0.2
2.5
150.0
19.0
1.6
181 .8
Expected No. of
Benzene caused
Leukemia deaths/year
2.88
0.07
0.85
50.89
6.44
0.54
61 .67
a Source: Appendix E
5-8
-------
No. of Persons
Exposed
Mean Annual
Average Benzene
Concentration (ppb)
Exposure
(ppb-Person Years)
6,000,000 0.5
1,000,000 2.5
200,000 7.0
80,000 20.0
Total 7,280,000 (7.3 x 106)
3.0 x 106
2.5 x 106
1 .4 x 106
1 .6 x 106
8.5 x 106
Applying the Carcinogen Assessment Group's model for benzene induced
leukemia deaths ND = 0.024074 x D x 10^/70.96 = 0.339262D where 0.024074
is the geometric mean of the slope parameter (see Appendix E); 70.96 is
the average expected life of a randomly drawn person living in the United
States based on 1973 vital statistics, ND is the expected number of leu-
kemia deaths and D is the exposure in ppb x 10® person years. Thus the
expected number of benzene induced leukemia deaths per year among non-
occupational exposed persons from chemical manufacturing in the United
States population is 8.5 x 0.339362 = 2.88.
If it is assumed that the United States population is exposed to
benzene from chemical manufacturing sources at the annual average con-
centration levels predicted for the sections of North Haven shown in Fig-
ure 4-2, then the estimated exposures calculated for each concentration
level level in ppb-person years and the expected number of leukemia
deaths for each level are as follows:
No. of
Persons
Exposed
7.3 x 106
7.3 x 106
7.3 x 106
7.3 x 106
Predicted
Annual Average
Concentration(ppb)
0.4
0.1
0.05
0.03
Exposure
(10® ppb-
Person Yrs.)
2.92
0.73
0.36
0.22
Expected
Leukemia
Deaths/Yr.
0.99
0.25
0.12
0.07
Relative Risk
North Haven:
All U.S.
0.344:1
0.086:1
0.043:1
0.026:1
5.3.5 Estimated Leukemia Risk from Exposure to Urban Factors
Estimation of the risk of leukemia deaths to residents of North
Haven resulting from "urban factor" benzene exposures requires an esti-
mate of the annual average concentration from such sources. The popula-
tion of the New Haven metropolitan area, which includes North Haven, is
less than 250,000 and the number of motor vehicles registered in this
area is approximately 150,000. These numbers are slightly less than
the corresponding numbers for Wichita and Harrisburg. Mara and Lee
(Reference 25) estimated the annual average benzene concentration from
automobile emissions in metropolitan areas of comparable size to these
5-9
-------
two cities at less than 0.1 ppb. Thus based on their work, the number
of automobiles registered in North Haven and adjacent communities and
the population of the area, the urban factor benzene concentration in
North Haven is estimated at 0.1 ppb. Applying the Carcinogen Assess-
ment Group's model and assuming the U.S. population exposed to urban
factor benzene (110 x 10®) is exposed to 0.1 ppb rather than the values
given in Table 5.1, then the expected number of leukemia deaths per
year (from urban factor exposures) would not exceed 110 x 0.339262
x 0.1 = 3.73.
In comparison, with the exposure distributed as listed in Table 5.1,
the expected number of leukemia deaths per year from urban factor expo-
sures is 57.57 for the United States population as a whole (see Appen-
dix E). A comparison of the national benzene induced leukemia deaths
estimated using the SRI U.S. benzene concentration distribution (Table
5.1) to that estimated using North Haven concentrations is given in
Table 5.3.
The estimates for the benzene-induced leukemia risk from both
Upjohn emissions and urban factor exposures assume lifetime exposure to
the estimated annual average benzene concentrations and that working,
recreation and sleeping occur within the vicinity of the respective
sources. These risks would change if part of the day, e.g. working,
occurred in an environment with greater or lesser annual average benzene
exposures. An example is given in Table 5.4. In these calculations,
the proportion of the United States population (110 x 10® persons)
which is estimated to be exposed to "urban factor" benzene, is assumed
to live either in a central city or in a suburb, but works in any of
the following environments: suburb, central city, or in the vicinity of
a point source, i.e., chemical manufacturing, petroleum refinery or
coke oven. The types of cities which would be considered to have 'cen-
tral city' exposures are those with a population in excess of 4 million
and include Chicago, Detroit, Los Angeles, New York and Philadephia.
It can be seen from Table 5.4 that the leukemia risk increases most
dramatically for a population which resides in a suburb but commutes to
a central city and only slightly if the same population worked in the
vicinity of point sources. The risk would also increase for individuals
residing in the vicinity of a chemical manufacturing source and working
in a central city and decrease for the same individuals if they worked
in a suburb.
It must be recognized that there are uncertainties associated with
the data and procedures used to estimate the benzene cancer risks
presented above. One of the uncertainties is the inability to precisely
determine annual benzene emission rate. Further uncertainty is intro-
duced via the inherent limitations of the dispersion model to perfectly
describe the atmospheric dispersion of the emitted benzene. There are
similiar limitations associated with the linear non-threshold model.
There are also uncertainties concerning possible additive effects of
multiple sources of benzene, synergistic or antagonistic health effects,
and heightened susceptibilities of some population groups. Because of
these uncertainties, the risk numbers presented above should only be
used as a rough estimate of the benzene induced cancer risk; i.e. they
should not be interpreted to represent the absolute magnitude of risk.
5-10
-------
TABLE 5.3
Source
EXPECTED NUMBER OF BENZENE CAUSED LEUKEMIA DEATHS
PER YEAR BY SOURCE CATEGORY
If U.S.
Population
Were Exposed
to Concentra-
tions As Listed
in Table 5.1
If U.S. Population were
Exposed to North Haven
Concentrations
Chemical Manufacturing
Coke Ovens
Urban Factors
Petroleum Refineries
Total
2.88
0.07
57.87
0.85
61 .67
0.07 - 0.99
0
3.73
0
3.80 - 4.72
5-11
-------
TABLE 5.4
EFFECT OF DIFFERENCE IN RESIDENTIAL AND WORKING ENVIRONMENTS ON ESTIMATED NUMBER OF
BENZENE INDUCED LEUKEMIA DEATHS
Reside
Work
Estimated^1)
Exposure Level
(ppb)
Total exposure
(106 ppb-person-yrs)
Estimated No. ^ ^
Benzene-Induced
Leukemia Deaths
Urban - Surburban Locations
Benzene concentation = 0.1-1.0 ppb
Suburb Suburb 1.1 121 41.05
Suburb Vicinity of Source^4) 1.3 143 48.51
Suburb Central City 1.9 209 70.91
Urban - Central City
Benzene concentration = 1.1-4.0ppb
Central City Central City 3.5 385 130.6
Central City Vicinity of Source 3.9 429 145.5
Central City Suburb 3.0 330 111.9
(1) Estimated annual average exposure from all sources
(2) Assumes population exposed = 110 x 10^
(3) Calculated from CAG model (Appendix E)
(4) Includes chemical manufacturing, petroleum refineries and coke ovens
-------
REFERENCES
EPA Memorandum on Upjohn Chemical Company Preliminary Ambient Air Survey,
August 1979.
J. F. Bowers, J. R. Bjorklund, and C.S. Cheney, 1979: Industrial Source
Complex (ISC) Dispersion Model User's Guide, H. E. Cramer Company, Salt
Lake City, Utah. EPA Report Number EPA-450/4-79-030, NTIS Accession
Number PB80-133044.
T. T. Shen, 1982: Estimation of Organic Compound Emissions from Waster
Lagoons. J. Air Pollution Control Assoc., 32, 79-82
D. B. Turner, 1970: Workbook of Atmospheric Dispersion Estimates, U.S.
Department of Health, Education, and Welfare. Public Health Service
publication Number 999-AP-26, NTIS Accession Number PB-191482.
von Oettingen, W.F., Neal, P.A. and Donahue, D.D. Toxicity and potential
dangers of toluene. Preliminary report. J.A.M.A. 118; 579-84 (1942).
Andrews, L.S., Lee, E.W., Witmer, C.M., Kocsis, J.J. and Snyder, R.
Effects of toluene on the metabolism disposition and hematopeotic
toxicity of [%] benzene. Biochem. Pharmacol. 26: 293-300 (1977).
Cornish, H.H. 'Salvents and Vapors' Ch. 14 in "Toxicology the Basic
Science of Poisons". Casarett and Doull., J. eds. MacMillan Publishing
Co. 1975 pp. 503-526.
Bowditch, M. and Elkins, H.B. J. Med. Hyg. and Tox 21: 321 (1939).
Srbova, J., Teisinger, J. and Skramovsky, S. Absorption and elimination
of inhaled benzene in man. AMA Arch. Ind. Hyg. Occup. Med. 2i 1-8,
(1950).
Hardy, J.L. and Elkins, H.B. Medical aspects of maximum allowable concen-
trations: benzene. J. Med. Hyg and Tox. 30: 196 (1948).
National Institute for Occupational Safety and Health. Criteria for a
recommended standard occupational exposure to benzene, HEW Publication
No. (NI0SH) 74-137 U.S. Government Printing Office, Washington, D.C.
Vigliani, E.C. Leukemia associated with benzene exposure. Ann. N.Y.
benzene workers. The Lancet 76-78, July 9, (1972).
Infante, P.F., Wagoner, J.K., Rinsky, R.A. and Young, R.J. Leukemia in
benzene workers. The Lancet 76-78, July 9, 1972).
Ott, M.G., Townsend, J.C., Fishbeck, W.A. and Langer, R.A. Mortality
among individuals occupationally exposed to benzene. Exhibit 154, OSHA
Benzene Hearings, July 9 - August 10, 1977.
-------
15. Ott, M.G., Townsend, J.C., Fishbeck, W.A. and Langer, R.A. Mortality
among individuals occupationally exposed to benzene. Arch. Environ Health
33: 3-10 (1978).
16. Askoy, M. and Erdem, S. Follow-up study on the mortality and the develop-
ment of leukemia in 44 pancytopenia patients with chronic exposure to ben-
zene. Blood 52: 285-292 (1978).
17. Askoy, M. Erdem, S. and Dincol, G. Leukemia in shoe-workers exposed
chronically to benzene. Blood 44: 837-841 (1974).
18. Snyder, R. and Kocsis, J.J. Current concepts of chronic benzene toxicity.
CRC Crit. Rev. Tox. 265-288 (1975).
19. Laskin, S. and Goldstein, B.D. Benzene toxicity: a critical evaluation.
J. Toxicol. Environ. Health Suppl. 1-148 (1978).
20. Ward, J.M., Weisberger, J.H., Yamamoto, R.S. Benjamin T., Brown C.A., and
Weisburger, E.K. Long-term effect of benzene in C57BL/6N mice. Arch.
Environ. Health. 30: 22-25 (1975).
21. National Academy of Sciences. Drinking water and health, Vol 3.
Safe DrinkingWater Committee, Board on Toxicology and Environmental
Health Hazards, Assembly of Life Sciences, National Research Council,
National Academy Press, Washington, D.C. 1980 415 pp..
22. Maltoni, C. and Scarnato, C. First experimental demonstration of the
carcinogenic effects of benzene. Med. Lavoro 70: 352-57 (1979).
23. Snyder, C.A. et al. The Inhalation toxicology of benzene: Incidence of
hematopoietic neoplasms and hematotoxicity in AKR/J and C57B2/6J mice.
Tox and Applied Pharm. 54: 323-331, 1980.
24. Akoy, M. Types of leukemia in chronic benzene poisoning. A study in
thirty-four parts. Acter Heamat. 55: 65-72 (1976).
25. Mara, S.J. and Lee, S.S. Human population exposures to atmospheric ben-
zene. In Definition of population-at-risk to environmental toxic pollu-
tant exposures. Vol II Apendix C 1980 SRI International Menlo Park, Ca.
-------
APPENDIX A
SAMPLE LOCATION MAPS FOR AMBIENT STUDY
-------
JB-16
JB-19
JB-15
ATI OMS)
u
'Jul
^TT^r-
DRAWING A-1
/UPJOHN AMBIENT AIR SAMPLING STUDY
SAMPLE LOCATIONS ON-SITE AND
IMMEDIATELY OF^SITE OF
£_ " [ UPJOHN PROPERTY.
-------
DRAWING A-2
UPJOHN AMBIENT AIR SAMPLING STUDY
SAMPLE LOCATIONS OFF-SITE OF
UPJOHN PROPERTY
SI!
1
V
si;
LOCATIONS)
/ Hlirh Sch
s
¦Sewage
I KILOMETER
-------
APPENDIX B
ANALYTICAL EQUIPMENT SPECIFICATIONS
AND OPERATING CONDITIONS FOR AMBIENT STUDY
-------
Detailed Analytical Equipment Specifications
and Operating Conditions
Instrumentation:
Finnigan 3200 gas chromatograph-mass spectrometer
Photovac 10A10 portable gas chromatograph
Century Systems 128 portable gas chromatograph
Tracor 560 gas chromatograph with photoionization detector
1. Trap Desorber:
Instrument Century Systems programmed thermal desorber 132A
Procedure Equilibrate 1 rain., desorb sample, expel 10 mis
Injection mode Syringe
PTD volume 282.4 ml
Desorption time 3 min. 39 sec.
Desorption temperature 200°C
2. Laboratory Gas Chromatograph:
Instrument Tracor 560
Column 5 ft. stainless steel (1/8 in OD) packed with 0.2%
carbowax 1500 coated on 60/80 mesh carbopack C
Temperature Program Isothermal 160*C
Inlet; detector temperature Off; 200°C
Detector HNu 52 photoionization
Input;Recorder attenuations 1; 16 and 32
Carrier Nitrogen
Flow rate 38 ml/min
3. Portable Photoionization Gas Chromatograph:
Instrument Photovac 10A10
Column 18 in. Teflon (1/8 in. OD) packed with 5% SE-30
coated on chromosorb G, 100/120 mesh
Temperature program Ambient
Detector Photoionization
Attenuations 2, 5, 10, and 20
Carrier Air
Flow Rate 30 ml/min
4. Portable Flame Ionization Gas Chromatograph:
Instrument Century Systems 128
Column 4 in. stainless steel (1/8 in. OD) T6
Temperature program Ambient
Detector Flame ionization
Attenuation 10
Carrier Hydrogen
Flow rate 20 ml/min
5. Laboratory Chromatograph-Mass Spectrometer:
Instrument Finnigan 3200
Chromatograph Conditions:
Column: Eight ft. stainless steel 1/8 in OD packed with
0.2% Carbowax 1500 coated on 60/80 mesh Carbopack C
Preceded by a 1 ft. stainless steel column (1/8 in
OD) packed with 3% Carbowax 1500 coated on 60/80
mesh Chromosorb W
Program: 60° to 160°C at 12#/min, held for a total time of
30 min.
Injector, Separator, Off, 200°C, 200°C
Transfer Temperatures:
Mass Spectrometer Conditions:
Electron Energy: 70 V
Mass Range: 20-27, 33-260 AMU
Emission Current: 1 ma
Scan Rate: 17 msec/AMU
-------
APPENDIX C
LAGOON EMISSION RATE PREDICTION
SHEN TECHNIQUE
-------
LAGOON EMISSION RATE PREDICTION BY
SHEN TECHNIQUE
The Shen technique (Reference 1 ) was used for estimating the volati-
zation rate of benzene and toluene from the aeration lagoon. The tech-
nique is based on the two resistance theory of mass transfer. Hie rate
of vaporization per unit area of pond is expressed as:
(ERP) . ,
i _ t a .. * ,->-o
"18x10" (Kqa) C±
where: (ERP)j_ = emission rate potential of a compound, g/sec
Kqa = overall mass transfer coefficient, g/mole/cm^-sec
A = lagoon surface area, cm2
Cj_ = concentration of the compound in mg/liter
For benzene and toluene (for which the Henry's law constant is
greater than 10"^), the mass transfer is liquid phase controlled and
gas phase may be ignored, i.e., Kqa = Ki/ where K]_ is the liquid phase
mass transfer coefficient, g/mole/cm2-sec. Shen gives an empirical ex-
pression for K]_ as follows:
Kl = 4.45 x 10"3 M"0,5 (1.024)t_2° u°*67 h"0,85
where: t = lagoon temperature, °C
U = wind speed, ft/sec
H = depth of the lagoon, ft
M = molecular weight of the volatile organic compound of inter-
est, g/g-mole
Substituting the equation for K]_ into the expression for the emis-
sion rate gives:
(BRP)j- = 8.01 x 10~8 M"0,5 (1 .024)t_2° uO'67 H>0.85 c
a i
From discussions with Upjohn personnel (Reference 2) the average
lagoon depth is estimated at four feet. EPA collected water samples from
the lagoon in March 1981. Analyses of these samples showed benzene
and toluene levels of 0.03 and 0.20 mg/liter, respectively (Reference
3). The molecular weights of benzene and toluene are 78 and 92 g/g-mole,
respectively. By assuming t = 25°C, substituting the above values into
the emission rate equation and simplifying, expressions for benzene and
toluene emissions are obtained as:
(ERP) = 9.4 x io-"1 1 U0-67, g/cm2-sec Benzene
A
(ERP) » 5.7 x 10-10 U0*57, g/cm2-sec Toluene
A
These expressions were used to generate Table C.1 which gives the
emission rates as a function of wind speeds.
-------
TABLE C.1
EMISSION RATES FOR AERATION LAGOON
Emissions (g/m^/hr)5
Wind Speed (m/sec)0"
Pollutant
Cone, in
Wastewater0
(ppb)
0.75
2.5
4.3
6.8
9.5 12.5
Aeration Lagoon
Benzene 30
Toluene 200
.006
.037
.014
.084
.020
.120
.027
.164
.034
.205
.040
.246
a Surface area of aeration lagoon is approximately 6,500 m2.
k Wind speeds correspond to mean speeds of the six speed categories
used by the National Climatic Center (NCC) in classification of
meteorological data.
c Data obtained by EPA in March 1981 (Reference 3).
-------
REFERENCES
1. T. T. Shen, "Estimation of Organic Compound Emissions from Waste
Lagoons". J. Air Pollution Control Assoc., 32, 79-82, 1982.
2. L. Joffe, Upjohn Chemical Company, North Haven, Connecticut, Private
Communications with R. R. Patrick of Engineering-Science, March 26,
1982.
3. Ambient Air Quality Study, Upjohn Company, North Haven, Connecticut,
(Preliminary Report), EPA Region I, Lexington, Massachusetts, Novem-
ber 1981 .
-------
APPENDIX D
DERIVATION OF VOC EMISSION RATES FROM
UPJOHN COMPANY AERATION LAGOON VIA
COMPARISON OF MODELED AND MONITORED CONCENTRATIONS
-------
DERIVATION OF VOC EMISSION RATES FROM
UPJOHN COMPANY AERATION LAGOON VIA
COMPARISON OF MODELED AND MONITORED CONCENTRATIONS
BACKGROUND
This brief report describes air quality modeling by EPA of the aera-
tion lagoon at the Upjohn Company facility in North Haven, Connecticut.
The purpose of this modeling exercise was to derive emission rates of
benzene and toluene by comparing ambient monitoring data to concentrations
predicted to occur under the same meteorological conditions. These emis-
sion rate estimates were to be compared to those used in another modeling
study performed by Engineering-Science (ES) for EPA.
Selection of Monitors and Episodes
Information provided by EPA's consultant was used to identify the
monitoring sites nearest the aeration lagoon. Only the integrated sampl-
ing locations were considered in this study. The integrated samples re-
present concentrations averaged over the monitoring period (usualy about
3 hours). The instantaneous sampling results were not used for two rea-
sons. First, since no information was available on the time of those
observations, the associated meteorological conditions could not be
specified with any certainty. Second, it was not clear how hourly or
longer average concentrations could be reliably determined from isolated,
instantaneous samples.
Concurrent on-site meteorological and ambient air quality monitoring
data, as reported in the Ambient Air Quality Study^1) for Upjohn Company,
were tabulated. The integrated monitoring locations and observed wind
directions were examined to select only those time periods during which
ambient monitors were downwind of the aeration lagoon. Furthermore,
those periods for which ambient air concentrations of benzene or toluene
were reported as "less than one" ppb were not considered. Six episodes
ranging in duration from 1 1/2 to approximately 3 hours were thus iden-
tified and are listed in Table 1. Figure 1 is a map showing the aeration
lagoon, its immediate surroundings, and monitor locations.
Specification of Meteorology
The on-site meteorological data included in the Upjohn Ambient Air
Quality Study provided information on wind speed and direction. In
order to perform the air quality modeling, it was necessary to specify
the Pasquill-Gifford stability class for each hour. The P-G stability
class was estimated by the objective scheme of Turnerwhich is used
for EPA models (e.g., the CRSTER model)• This method requires informa-
tion on wind speed, solar elevation angle, times of sunrise and sunset,
cloud cover, and ceiling height. Data on solar elevation angles and sun-
set and sunrise times for this location were generated via a computer
program based on the method of Woolf(4). Data on cloud cover and ceiling
height were estimated from Local C lima to logical Data(5,6) reported by the
National Weather Service Office at Sikorsky Memorial Airport in Bridge-
1
-------
port, Connecticut. These monthly summaries provide the required data at
three hour intervals; values for intervening hours were estimated by
linear interpolation or by assumptions of persistence based on
meteorological judgment. Wind data from Bridgeport were also used when
data were not available from the on-site meteorological station.
In one episode (at station 8198) the on-site wind direction was re-
ported as variable for the last two hours of the three hour monitoring
period. The integrated ambient concentration was relatively high, but
the reported wind direction for the first hour was virtually tangent to
the lagoon edge resulting in fairly low predicted concentrations. For
purposes of estimation, a wind direction which would blow directly over
the lagoon towards the monitor was assumed to see if this would account
for the observed high concentration.
Model Selection
The EPA PAL model was selected for use in this modeling study pri-
marily because of its treatment of area sources. For a full description
of this model, see the User's Guide for PAL(^. The calculation of con-
centrations from area sources is simulated by calculating impacts from a
number of crosswind lire sources located at various distances from the
receptor under consideration. Edge effects are taken into account. This
computational approach was judged to be mora desirable than the virtual
point source method used in many other candidate models. In addition,
area sources in PAL can be squares or rectangles (as opposed to just
squares) and may be given negative emission rates, so that complex area
source configurations can be approximated far more easily and accurately
than with other models. This was important in this application due to
the proximity of the monitors to the irregularly shaped edge of the
lagoon.
Furthermore, PAL contains no restrictions related to source size or
placement of receptors. Other candidate models contained restrictions
on receptor placement such that it would not have been possible to place
receptors at the monitors. Lastly, it should also be noted that while
the User's Guide for PALt) states that the model is designed to assess
the impacts on air quality of portions of urban areas, it does in fact
act as a rural model in that it contains rural dispersion coefficients
and does not modify the input stability class to account for urban
effects.
Input Data and Model Options
The PAL model calculates concentrations on an hourly basis. Since
the averaging period at the air quality monitors was not always an in-
tegral number of hours and since the averaging periods for the air quality
and meteorological monitors were not always in phase, it was not always
possible to model precisely the same period of time corresponding to each
episode. The meteorological data input to the model was chosen so that
it would correspond as closely as possible and be most representative of
the conditions during the ambient monitoring.
2
-------
The aeration lagoon at the Upjohn facility was closely approximated
in shape as the sum of ten area sources, six of which had negative emis-
sion rates. Since the model requires that area source boundaries be
oriented north-south and east-west, the reported wind directions were
transformed so that the area source would be oriented as required by the
model while the wind directions would be oriented correctly with respect
to the lagoon. A unit emission rate equal in magnitude to 1 yg m-^ sec-1
was used to model the area sources. A model receptor was placed at the
corresponding location of the appropriate air quality monitor for each
episode. Figure 2 displays the area source configuration modeled and the
receptor locations. Note that the shaded areas indicate area sources
that were modeled with negative emission rates to offset regions with no
emissions that were modeled with positive emissions as part of larger
area sources. In this manner, the irregular shape of the lagoon was
accounted for.
A receptor height of 1.2 meters above ground level was specified to
correspond to a probe height of about 4 feet, and a ground level area
source release height was used. Vertical wind shear was not considered
due to the low range of vertical heights under consideration. A mixing
height of 5000 meters was used but is of no importance in this problem
due to the extremely low emission height and the small downwind distances
involved. Table 2 lists the meteorological conditions for each episode.
Results
The model produced concentration estimates for each hour as well as
average concentrations for each episode. The average predicted concen-
trations in g m~3 were then converted to parts per billion of benzene
and toluene for standard conditions (temperature of 298°K and pressure of
1013.2 mb). The relationships used were that a concentration of 1 ug m"^
corresponds to 0.313 ppb benzene and 0.264 ppb toluene.
For each pollutant and each episode, an emission rate for the aera-
tion lagoon in ug m~2 sec"1 was derived by dividing the observed concen-
tration by the concentration predicted for the assumed emission rate of
1 ug m~2 sec"1. This linear scaling of the results is valid since concen-
trations are linearly proportional to emissions for inert pollutants.
For the small source-receptor distances involved here, an assumption of
chemical inertness is valid. Also implicit in this derivation is the
assumption that the concentrations of benzene and toluene measured at
the air quality monitors were entirely due to emissions from the aeration
lagoon. This may be a fairly reasonable assumption given the proximity of
the monitors to the lagoon. In any case, there was no reliable, consis-
tent way to account for background concentrations. However, if the effects
of background had somehow been included, the derived lagoon emission rates
would have been smaller, since the lagoon would then have to account for a
smaller portion of the monitored concentrations. Therefore, the derived
lagoon emission rates are conservative in this respect.
For each episode, an average wind speed was calculated to see if
there was any relationship between wind speed and emission rate. Such a
relationship might be expected since evaporation increases with wind
speed. The derived emission rates for benzene and toluene are plotted as
3
-------
a function of this average wind speed for each episode in Figures 3 and
4, respectively. These figures also display the emission rates (converted
to units of ug m"^ sec-1) which were used in a modeling study by ES^8^.
These emission rates had been obtained by ES using the method of Shen^9^
and will henceforth be referred to as the "Shen values." The discrete
values reported by ES have been connected in the plots via a spline inter-
polation so that the dependence on wind speed can be more clearly seen.
I
The emission rates derived via the comparison of modeling and moni-
toring were not available for as wide a range of wind speeds as those
obtained via the method of Shen. The wind speeds for the modeled episodes
were generally either low (about 2 m sec"1) or moderate (about 6 m sec"1).
While the derived lagoon emission rates do not show the smooth variation
with wind speed characteristic of the Shen values, they do on average
show an association of higher emission rates with higher wind speeds.
A comparison of the derived emission rates with the Shen values re-
veals that for each episode the estimates are always well within an order
of magnitude. This degree of agreement might be considered fairly good
given the level of uncertainty in some of the meteorological variables
and especially in the ambient air quality data in which concentrations
obtained from different methodologies sometimes yielded far different re-
sults. For more information on the latter, see the Upjohn Ambient Air
Quality Study(1^ for information on sampling strategy and analytical
procedures.
The comparison of the derived emission rates and the Shen values also
reveals that the Shen values are generally higher, especially for toluene.
The only episodes for which the derived lagoon emission rate was signifi-
cantly higher than the Shen value are for stations 8182 and 8196 for ben-
zene. A closer examination of these episodes indicates that they both
occurred at the same monitor location (NA-8) and were associated with
northerly winds (between 340° and 360°). A clarifier and effluent pump
station are located just upwind of this station under these wind direc-
tions, so that one possible explanation of the discrepancy in the two esti-
mates is that the clarifier and/or effluent pump station might be another
source of benzene. If this is the case, then the derived emission rate of
benzene from the lagoon should be somewhat lower since the lagoon would
have to account for only some portion of the observed benzene concentration.
Conclusion
The emission rates derived for these six episodes do not seem to allow
for a straight forward specification of either a single emission rate or a
consistent variation of emission rate with wind speed. However, the de-
rived emission rates generally support the use of the Shen values. The
Shen values show a smooth variation with wind speed which can be included
in subsequent modeling. The Shen values are also generally conservative
when compared to the derived values. Since the emission rates are to be
used as input to a modeling exercise dealing with health effects, it may
be desirable to have a measure of safety built into any subsequent
analysis.
4
-------
TABLE 1
STUDY EPISODES
Ambient Concentrations
Location ID
Station #
Date
Time (EST)
Benzene (ppb)
Toluene (ppb)
00
t
g
8182
11/18/80
1430-1730
10.5
14.0
NA-9
8192
11/19/80
0810-1115
0.7
5.5
NA-8
8196
11/19/80
1358-1658
9.0
16.0
NA-7
8198
11/19/80
1958-2258
24.5
45.0
NA-6
82010
11/20/80
1530-1800
1.7
12.0
MA-8
D007
3/26/81
0830-1000
7
14
5
-------
TABLE 2
METEOROLOGICAL DATA
Station Location Observed Transformed Wind Speed Stability
# ID Wind Direction Wind Direction (m sec"*1 ) Class
8182
NA-8
360
338
6.0
D
II
tl
360
338
6.0
D
tl
II
340
318
6.5
D
ft
It
340
318
6.5
D
8192
NA-9
340
318
7.5
D
tl
11
350
328
7.5
D
II
II
340
318
7.5
D
8196
NA-8
350
328
7.0
D
II
II
340
318
6.0
D
tl
II
340
318
4.0
E
8198
NA-7
330
308
2.0
F
II
II
297*
275*
1 .0
F
II
II
297*
275*
1 .5
F
82010
NA-6
300
278
3.0
D
II
11
270
248
2.0
E
II
II
240
218
1 .5
F
D007
MA-8
200
178
2.0
D
It
n
225
203
2.5
D
* Assumed
6
-------
REFERENCES
1. Ambient Air Quality Study, Upjohn Company, North Haven, Connecticut,
(Preliminary Report), EPA Region I, Lexington, MA, 4/5/82.
2. Turner, D.B., "A Diffusion Model for an Urban Area", Journal of
Applied Meteorology, 3: 83-91, Feb. 1964.
3. User's Manual for Single-Source (CRSTER) Model, EPA Publication No.
EPA-450/2-77-013, Research Triangle Park, N.C., July 1977.
4. Woolf, H.M., On the Computation of Solar Elevation Angles and the
Determination of Sunrise and Sunset Times, NASA Technical Memorandum
NASA TM x-1646, Washington, D.C., September 1968.
5. Local Climatological Data, Monthly Summary, Bridgeport, Connecticut,
National Oceanic and Atmospheric Administration, National Climatic
Center, Asheville, N.C., November 1980.
6. Local Climatological Data, Monthly Summary, Bridgeport, Connecticut,
National Oceanic and Atmospheric Administration, National Climatic
Center, Asheville, N.C., November 1981.
7. User's Guide for PAL, EPA Publication No. EPA-600/4-78-013, Research
Triangle Park, N.C., February 1978.
8. Dispersion Modeling of Volatile Organic Compound Emissions at the
Upjohn Facility in North Haven, Connecticut, (Preliminary Draft
Report) Engineering-Science, McLean, Virginia, April 1982.
9. T.T. Shen, "Estimation of Organic Compound Emissions for Waste
Lagoons", J. Air Pollution Control Assoc., 32, 79-82, 1982.
7
-------
¦}/
I
-M&
s
S
rjf
IS
\
r
FIGURE 1
JC-J:
-------
N
©
r*
o
C4
tK
©
c*
ES
vJ
-------
FIGURE 3
30n
25-
20-
o
a
u
BENZENE
~
¦i
4
I
E
at
3
a
1S-
to-
5-
XX
(3
~r
2
~r
4
T~
6
~r~
s
i
to
~T~
12
WIND SPEED (m sec"')
14
Legend
A D007
X 82010
~ BIS2
(3 8192
H 8196
X 8198
~ Shan Values
-------
FIUIKE 4
TOLUENE
70->
60-
30-
o
«> 40-
u
E
o>
5 30 -
O
20-
10-
/
/
/
/ x
X A
/
D
/
B
H
T"
2
~T~
4
T"
8
"~1
to
T~
12
~I
14
WIND SPEED (m sec"')
Legend
A D007
X 82010
P B1B2
|3 8192
a 8196
X 8196
Shan Value»
-------
APPENDIX E
CARCINOGEN ASSESSMENT GROUP'S
FINAL REPORT ON POPULATION
RISK TO AMBIENT BENZENE EXPOSURES
-------
EPA-450/5-80-004
Carcinogen Assessment Group's
Final Report on Population
Risk to Ambient Benzene Exposures
by
Dr. Roy E. Albert. Chairman
Carcinogen Assessment Group
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air, Noise, and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
January 10, 1979
-------
DISCLAIMER
This report has been reviewed by the Strategies and Air Standards
Division cf the Office of Air Quality Planning and Standards, EPA,
and approved for publication. Mention of trade names or commercial
products is not intended to constitute endorsement or racorrmen-
dation for use. Copies of this report are available through the
Library Services Office (MD-35), U.S. Environmental Protection
Agency, Research Triangle Park, N.C. 27711, or from National
Technical Information Services, 5285 Port Royal Road, Springfield,
Virginia 22161.
-------
CONTENTS
I Summary ..... 1
II Introduction 3
III General Approach to Utilizing Epidemiological Studies
to Predict Lifetime Probability of Cancer Deaths Due
to Benzene 4
A. Mathematical Model Employed 4
B. Estimation of Lifetime Probability of Death Due to
Various Forms of Leukemia for a Member of the U.S.
Population 6
IV Epidemiological Studies Utilized . . 7
A. Infante (1977) ...... 8
B. Askoy (1974, 1976, 1977) 13
C. Ott, et al. (1977) 17
V Estimation of Expected Number of Leukemia Deaths Due
to Environmental Exposure to Benzene 21
VI Bibliography 30
VII APPENDIX - Mutagenic Risks of Benzene Exposure ... 31
111
-------
TABLES
1. Data Utilized to Estimate Lifetime Probability of
Death Due to Various Forms of Leukemia 24
2. Lifetime Probability of Death in U.S. Population
Due to Leukemia Type Upon Which Relative Risk in
Each of the Epidemiological Studies is Based 25
3. Summary of Data Used to Estimate Lifetime
Probability of a Leukemia Death Per ppm Benzene
Lifetime Exposure 26
4. Source Specific Benzene Caused Leukemia Deaths/Year
Based on Table 1-1 of SRI Benzene Exposure Document. . 27
5. Total Exposure of People Residing in Various
Locations, and Resulting Estimated Benzene Caused
Leukemia Deaths/Year - Based on Table 1-2 of SRI
Benzene Exposure Document .... 27
6. Confidence Limits on Total Benzene Caused Leukemia
Deaths/Year (Assumes "One-Hit" Model is the True
Dose Response Relationship) 29
iv
-------
CARCINOGEN ASSESSMENT GROUP'S FINAL REPORT ON POPULATION RISK
TO AMBIENT BENZENE EXPOSURES
I. Summary
There 1s substantial epidemiological evidence that
benzene 1s a human leukemogen. However, no validated animal
model has yet been developed for benzene as a carcinogen.
There are several large series of case reports Indicating a
high risk of leukemia in individuals who developed aplastic
anemia consequent to benzene exposure. In addition there
are a number of epidemiological studies in the rubber,
chemical and shoe industries that demonstrate an excess risk
of leukemia associated with benzene exposure.
Three of these epidemiological studies provide enough
Information about exposure to benzene and the occurrence of
leukemia to allow us to make crude quantitative estimates of
the leukemia risk associated with current general population
exposures to benzene 1n the United States. These studies
were conducted by Infante et al, (1977), Ott, et al., (1 977)
and. Askoy et al., (1977, 1976 , 1974).
The Infante study, which showed an excess incidence of
leukemia, is not yet completely analyzed by the authors.
Hence, some assumptions made about the average duration and
magnitude of exposures are necessary. The Ott study
indicated a marginal excess myelogenous leukemia risk with
-------
relatively wel1-documented exposures. The Askoy studies
"Indicated a marked increase in non-lymphatic leukemia to
individuals using benzene based adhesives in small shoe
making shops, however, the exposure data in this situation
was difficult tb evaluate.
A linear non-threshold model was used to estimate the
leukemia risk to the low average levels of about one part
per billion to which the general population is exposed. The
slope parameter of this model was taken as the geometric
mean of the slope parameter estimates obtained from the
three epidemiological studies. Using this extrapolation
model, we estimated that the number of cases of leukemia per
year in the general population due to ambient atmospheric
benzene is about 90 with a 952 confidence interval from 34
to 235 assuming a precision of within two fold in the
exposure estimate. This is from .232 to 1.622 of the total
leukemia deaths in the United States based upon 1973 vital
statistics.
The purpose of this calculation is to obtain a rough
estimate of the carcinogenic hazard to benzene in the entire
United States population. To do this lifetime averages of
benzene exposure were estimated and these were combined with
the non-threshold linear model of risk as a function of
lifetime average exposure. In this report no attempt has
been made to estimate the* risks to selected sub-populations
who may have greater or less than average exposure or
sensitivity to benzene although it is certain that such
X
groups exist.
-------
II. Introducti on
The Carcinogen Assessment Group (CAG) has been asked by
the Office of A1r Quality Planning and Standards (OAQPS) to
estimate the cardnogenic r1sk to the United States
population of ambient benzene concentrations. This type of
information is useful 1n judging the overall contribution of
benzene emissions to the national rates of cancer mortality,
and will be used by OAQPS in the decision whether to
regulate benzene.
As the basis for this estimation, the CAG is using three
epidemiological studies that show a relationship between
excess mortality due to leukemia and benzene exposure. Each
of these studies have strengths and weaknesses that will be
discussed, but taken together they represent convincing
evidence that benzene Is a -human carcinogen.
To date, no clear evidence exists implicating benzene as
a carcinogen from animal experiments. A study is in
progress at Hew York University that appears to suggest that
Inhaled benzene is causing leukemia in rats. At the present
time it is felt that it would be premature to base a risk
extrapolation on this preliminary data. However, at the
completion of this study the CAG will update the present
risk analysis to take account of this new information.
-3-
-------
Ill- General Approach to Utilizing Epidemiolo
-------
If we make the assumption that, "R", the relative risk
of leukemia for benzene exposed workers compared to the
general population 1s independent of the length or age of
exposure but depends only upon the total exposure, it
follows that
R 3 QQ-A + 8 (xi +
Pj^ A + Bxi
or RPl ¦ A + B (xjl + X2)
Pj, =* A + Bxi
so that B ¦ Pi(R-1)/x2
where: x^ « ambient level exposure to benzene
x£ 3 industrial level exposure to benzene
Pj, a the lifetime probability of dying of
leukemia with no or negligible benzene
exposure
To use this model estimates of R and must be obtained
from the epidemiological studies. The exposure values x^
are derived in the exposure study conducted by SRI dated May
1978 and will be discussed where they are utilized.
The estimate of the lifetime probability of death due to
different types of leukemia, Pj, is discussed in detail in
the next section.
-5-
-------
B. Estimation of Lifstime Probabi1ity of Death Due to
Various Forms of Leukemia for a Member of the U.S.
Population
The data utilized to estimate the lifetime probability
of death due to various forms of leukemia is shown in Table
1, which was taken from "Vital Statistics of the United
States 1973 Volume II - Mortality Part A." The second and
third columns (total deaths and total death rate in 1973)
were taken from page 1-184 and 1-8, respectively and
utilized to derive column four, total U.S. population in
each of the age classes.
The total number of deaths in 1973 due to each of the
types of leukemia listed by the 8th ICO code
204 - lymphatic
205 - myeloid
206 - monocytic
207 - other and unspecified
are shown in columns five through eight.
The age specific death rates for each type of leukemia
are estimated by dividing the total number of deaths due to
that type by the total number of people in that age class.
-6-
-------
In the appendix of a 1978 CAG document on population
risk due to coke ovens a method referred to as the "constant
segmented model" 1s derived that allows one to estimate the
lifetime probability of death due to a disease given the
age-specific incidence rates for the disease and all sources
of death. This model was employed using the data in Table 1
to obtain the lifetime probabilities of leukemia that
approximate as closely as possible the type of leukemia that
the relative risk estimates were based upon in each
epidemiological study. These lifetime probabi1ities are
shown in Table 2, and will be used subsequently to estimate
lifetime probabi!ities of leukemia death for each unit of
exposure to the general population for each of the
epidemiological studies.
IV. Epidemiological Studies Utilized
Each of the epidemiological studies is discussed in
general. The relative risks are modified in each of the
studies to represent a most likely rather than a
conservative lower limit as usually is the case where the
primary aim of epidemiological workers is to establish with
little doubt a "statistically significant" elevated relative
risk. Estimates of the average lifetime exposure are also
made using as much data as is available. This information
is then utilized employing the previously discussed
mathematical model to estimate the lifetime probability of
leukemia for each unit of exposure.
-7-
-------
A. Infante (1977)
1• Description of Infante Study (1977)
In a retrospective study of mortality in a cohort of 748
white male workers in two Ohio plants manufacturing a
natural rubber cast film product, Infante et al., (1977)
observed a statistically significant higher rate of leukemia
than in either of two control groups. The leukemia
raortality rate was 5.06 times higher than the general U.S.
white male population standardized for age ana time period
of the cohort exposure, and 4.74 times higher than a cohort
of 1447 white males employed at an Ohio fibrous-glass
'construction products factory. These results were based on
a 752> follow-up of the vital status of the workers. A total
of 160 deaths were observed and of these there were 7
leukemia deaths, four of which were acute myelogenous, one
chronic inyelogenous, and two monocytic leukemia.
As with virtually all epidemiological studies, the
Infante study has various strengths and shortcomings. Among
Its strengths are; (1) the worker exposures are said to have
been almost exclusively restricted to benzene, since it is
"used throughout the plant as the principal solvent in all
major processes; (2) the individuals in the cohort all
worked before 1950 and were followed until 1975, thus
allowing long latency diseases to be observed, and (3) acute
-8-
-------
myelogenous leukemia was observed, which is the same cell
type of leukemia observed in other studies where workers
have had known benzene exposures. The disadvantages of
relying on this study for determining general population
risks are: (1) the authors essentially give no estimate of
worker exposures except to say that the levels were -less
than the prevailing recommended occupational limits at the
time various monitoring surveys were made; (2) the members
of the cohort study actually worked at two separate plants
(Akron and St. Mary's, Ohio). Air monitoring information in
the former plant is almost non-existent (Baier, 1977), and
therefore the exposure to half of the members of the cohort
1s almost completely unknown.. However, 1t is known (Young
1977) that the crude rates (leukemia cases/total people in
the cohort) are similar in the two locations; (3) Warren et
al., (1977) claimed that over 400 workers known to be
exposed to low benzene levels were deliberately excluded
from the cohort. In spite of these problems, it is felt
that this study is the least flawed of the three utilized.
2. Estimation of the Relative Risk
In an update, published as a letter to the editor in
Lancet (Benzene and Leukemia, October 14, 1977) Infante et
al., note that:
-9-
-------
(1) Sakol (1977) has supplied additional
Information that at least two more cases of leukemia
known to exist, but not reported on death
certificates, were probably in Infante's cohort;
(2) Due to a more complete follow-up, the expected
number of deaths due to leukemia in their cohort was
reduced from 1.38 to 1.25.
Using this supplemental information, the new relative risk
due to total leukemia is estimated to be
R = (7+2J/1.25 « 7.20
3. Estimation of Average Occupational Exposure
Information about the plant benzene levels is contained
1n the Appendix to the testimony of Baler at the OSHA
benzene hearings (Baier, 1977). From the opening of the
factory in 1940 until 1946, no monitoring records were
available. Follov/1ng the installation of new ventilation
equipment in 1946, a survey showed that levels in "most
areas" in the plant ranged from 0 to 15 ppm and that all
areas had less than the maximum safe limit of 100 ppm which
prevailed at that time. From this information, one can
guess that the average exposure to all people in the plant
before 19.46 1s probably not much more than 100 ppm, and not
less than 15 ppm.
-------
Benzene levels were monitored after 1946 at various
plant locations but they were all instantaneous samples and
no reliable information is available about how many
man-hours were spent at those locations or whether
protective masks were worn. These are case reports of
exposures to 1000 ppm for short time intervals. Since the
average levels were generally close to the occupational
standard, we will make the assumption that the average
worker exposure was the same as the prevailing recommended
occupational limits. These are tabulated below along with
the time weighted average for the 36 years of the total
exposure period.
Time Mo. of Average Time-Weighted
Interval Cases Exposure (ppm) Average (ppm)
1957-68
1940-46
1947
1948-56
(7)
(1)
(9)
(12)
100-15
25
50
35
39.9-23.3
1969-75 (7)
10
-------
The actual levels to which the workers were exposed was
a subject of heated debate at the OSHA benzene hearings.
The CAG would like to see a realistic estimate of the
population weighted average exposure and its uncertainty
limits. The time-weighted averages for occupational
exposure must be converted to a continuous exposure lifetime
basis. It will be assumed that the maximum likely lifetime
exposure would result if a worker entered the factory in
1940 and was exposed for 35 years to the occupational limit
of benzene. This exposure would result in a time-weighted
average of 40.36 ppm for 35 years. The least likely
exposure 1s assumed to occur if a worker started in 1S5Q
and was exposed to the occupational limit, which results in
a time-weighted average of 23.7 ppm for 25 years. The
equivalent continuous lifetime exposures corresponding to
these work place exposure estimates are:
High estimate: 40.36 x 240 x 1 x 35 » 4.4ppm
^ 3 70
Low estimate: 23.7 x 240 x ^ x 25 = 1.8ppra
365 3 70
The geometric mean of the high-low exposures,
\/ 4.4x1.8 = 2.81, is taken to be the best estimate of the
lifetime average for workers in the cohort.
-------
Estimation of Lifetime Probability of leukemia
Per Unit of Exposure
The change in the leukemia rate per lifetime average ppra
In the atmosphere is derived from the previously discussed
equation:
B » Pi (R-1)/X2
which gives us an estimate
B a .006.732 x (7.20-1)/2~ 81 = .014854
B. Askoy (1974, 1976, 1977)
i. Description of Askoy Studies
Askoy (1977, 1976, 1974) has reported his observations
of the occurrence of leukemia and aplastic anemia cases at
two medical institutions in Istanbul over a period from 1967
to 1975. He has compared the types of leukemia seen in shoe
workers, who work with benzene solvents in small
unventilated shops, with the types of leukemia observed in
people with no known exposure to benzene. He has also
tabulated the exposure duration of patients with different
types of leukemia. He found that in shoe workers there were
34 cases of leukemia observed in the nine years from 1967 to
1975. Based on "official records" which show that in
Istanbul ..there are 28,500 workers in the shoe, slipper and
handbag industry, he calculates that the annual incidence
-13-
-------
pate of leukemia 1s 13 per 100,000, which is significantly
higher than 6 per 100,000, the rate in the general
population. The calculation is based on crude rates with no
age adjustment.
He also found that the types of leukemias occurring in
people exposed to benzene are different than for those with
no known exposure. In a sample of 50 non-exposed leukemia
patients, approximately 50 percent had chronic leukemia, but
1 n 40 benzene-exposed patients only 5 percent had chronic
leukemia. Also, in the exposed group preleukemia and acute
erytnroleukemia accounted for 34 percent of the cases,
whereas in the non-exposed group only 6* of the cases were
of those types.
The concentration of benzene to which the workers were
exposed was estimated only in terms of the maximum
concentrations existing at the times when benzene was being,
used in the shops. At the 0SHA benzene hearings in 1977
Dr. Askoy stated that the concentrations outside working
hours ranged between 15 and 30 ppm and reached a maximum of
between 150 and 210 ppm when adhesives containing benzene
were being used.
-14-
-------
2. Estimation of Relative Risk.
A total of 26 patients with leukemia were observed in
the 6 2/3 year period from 1966 to September 1973 in a group
of 28,500 Instanbul shoe workers exposed chronically to
benzene. This was felt to be an underestimate of the true
number of leukemia cases among the shoe workers during the
period with Askoy subsequently being aware of two additional
cases. However, three of the twenty-eight total cases were
lymphoblastic or lymphoid leukemia, not thought to be as-
sociated with benzene exposure. Eliminating these three
cases, an estimate of the yearly incidence rate is
I s {26+2-3) x IPS = 13.15 per 100,000 per year
28,500 x 6.67"
The total incidence rate of leukemia in Turkey 1s
thought to be about 2.5 to 3.0 per 100,000 Askoy (.1977).
Howe.ver from Askoy's non-exposed patient group we estimate
that 48% based on 24 out of HO are non-lymphoblastic or
lymphoid leukemia. In addition, the national rate which is
based on the total population was felt by Cooke {1954) to be
about twice that experienced for the relatively young group
of benzene-exposed shoe workers who had a average age at
diagnosis of 34.2 years. Using this Information we estimate
that the yearly incidence rate of non-lymphoblastic or
lymphoid leukemia in the Turkish population of the same age
structure as the benzene exposed shoe workers is
-------
I » (2.5+3.0) x 24 x 1 » .66 per 100,000
2 TO 2
An estimate of the relative risk for benzene exposed
shoe workers is thus
R » 13.15 * 19.92
75%
3. Estimation of Lifetime Average Exposure
It was notid that the benzene levels were 15 to 30 ppm
outside working hours and 150 to 210 ppm during working
hours when benzene was in use in the typical small shoe
manufacturing shop. We will assume that the average working
hour exposure to benzene was the geometric mean of the
midpoint of the two Intervals or
X? 3 \{ 15 + 30) x ( 150 + 210) = 63.6 ppm
W I 2
In addition we will assume:
(1) A ten hour working day
(2) A 300 day working year
(3) An average age at the end of the observation
period of 50 years
(4) An average of 9.7 years of exposure; this is
the average length of exposure for the
leukemia cases in Askoy's series.
-16-
-------
These assumptions lead to a lifetime average exposure
estimate of
x? 3 63.6 x (10) x (300) x (9.7) * 4.22 ppm
2? TtTo W
4. Estimation of Lifetime Probability of Leukemia
Per Unit of Exposure
The change in the leukemia rate per lifetime average ppm
1n the atmosphere is derived from the previously discussed
equation:
B » P]{R-l)/X2
which gives us an estimate
B * .004517 x (19.92-1)/4.22 * .020252
C. Ott, et al., ( 1977 )
1. Description of Ott Study
The long-term mortality patterns and associated exposure
estimation of a cohort of 594 workers exposed to benzene
were reported by Ott, et al., (1977). The workers were
employed in three production areas of the company, which had
been in operation for varying times since 1920.
Each job category was assigned an average exposure range
as accurately as the historical air monitoring data
permitted. The concentrations ranged from less than 2 ppm
(8-hour time weighted average) to greater than 25 ppm. The
analysis covered employees with known benzene exposure who
-17
-------
worked from January 1, 1940 through 1973. A total of 53
employees with known exposure to arsenicals, vinyl chloride
and asbestos in addition to their benzene exposure was omit-
ted from the formal cohort of people exposed to benzene.
The benzene exposure of each person was evaluated and
ex-pressed as the product of parts per million times months
of exposure. For the 91 deceased people with exposure to
benzene alone, 45% of them had exposures between 0 and 499
ppm-months and 35* had exposures greater than 1000 ppm-
months. The results of the analysis of mortality by cause
of death showed no statistically significant excess of
mortality compared to the U.S. white male age-specific
mortality rates. Three cases of leukemia were observed
where 0.8 cases were expected, a situation of borderline
statistical significance (p<0.047). All three were my-
elocytic leukemia, two of them acute, which latter is the
type associated with benzene exposure of shoe workers
(Askoy, 1976) and other occupations (Yigliani, 1376).
2. Estimation of Relative Risk
In Ott's cohort 3 deaths due to noo-lymphocytic non-
monocytic leukemia were observed with only .8 expected. An
estimation of the relative risk of non-lymphocytic -
monocytic leukemia is thus
R » 3/.8 = 3.75
-18-
-------
3. Estimation of Lifetime Average Exposure
Ott estimated the ppm-months of exposure of each
individual in his cohort from work history data and plant
hygiene benzene measurement surveys. The most complete
presentation of this data is given in Ott's (1977) table 7
which is used to estimate the averagie level of exposure.
It is assumed that the average exposure in each of the
exposure intervals is equal to the midpoint of the first two
intervals and is equal to the lower limit plus 1/2 of the
interval width for the open or third classification. The
total average ppm-months 1s obtained by taking the average
of the three classifications weighted by the expected value
of the number of deaths in each classification giving the
val ue
(250 x 65.1 + 750 x 16.2 + 1250 x 32.8)/(65.1 + 16.2 + 32.8)
= 608.46 ppm x months
The average lifetime exposure is obtained by using the
following assumptions:
(1) An*eight hour working day
(2) A 240 day working year
(3) An average age at the end of the observation
period of 65 years
which gives the lifetime estimate of
(608.46) x (8 ) x (240) x (1 ) = .171 ppm
( rr~) (24) (365) (65)
-------
4. Estimation of Lifetime Probability of Leukemia
Per Unit of Exposure
The change in the leukemia rate per lifetime average ppm
1n the atmosphere is derived from the previously discussed
equation:
8 » Pi(R-1)/X2
which gives us an estimate
B * .002884 x (3.75-1 .)/.17 * .04638
0. Summary of Results
The total leukemic response has been based on different
classifications of leukemia for the three studies. A
summary of the type of response utilized is given in Table
3. It would have been preferable to have applied a uniform
method of classification for all the studies. However, due
to the lack of specific detail in the presented papers this
was not possible.
Even with this added source of variability the resulting
slope estimates B, which have the physical meaning of the
total probability of deaths due to 1 ppm of benzene in the
air breathed over an Individual lifetime, were remarkably
consistent between studies. The geometric mean of the three
estimates ic
• B 014854x.020252x.046380" * .024074
-20-
-------
The estimated log mean 1s
log-joB = -1 .618453, with the estimated variance
of this mean being 0"log B » .021785
Y. Estimation of Expected Number of Leukemia Deaths Due to
Environmental Exposure to Benzene
The SRI in their exposure document expressed exposure to
the U.S; population In two ways.
The first method assumed a static population living
around the point sources and gave total ppb-person years for
each of the point source classifications in Table 1-1.
If exposure units in 10*> ppb-person years are denoted
as D> the expected number of leukemia deaths per year may be
estimated approximately by the relationship
Mq 3 .024074 x 0x10^/70.96 « .339262D
where .024074 is the geometric mean of the slope parameter
taken from the three studies, and 70.96 is the average
expected life of a randomly drawn person living in the
United States based on 1973 vital statistics.
Using Table 1-1 and the above equation the number of
leukemia deaths per year are estimated for the various point
sources and are shown in Table 4.
-21-
-------
The second method employed to estimate exposure did not
make the over-simplIfled assumption that the human
population was static* Instead an attempt was made to
follow a typical Individual through a typical day 1n order
to obtain his average exposure. The exposure estimates
derived on this basis are shown 1n Table 1-2 of the SRI
document and were utilized in conjunction with the above
equation to derive the estimated number of leukemia deaths
per year shown in Table 5.
We note that approximately a total of 90 cases of
leukemia per year could be expected due to benzene exposure
In a recent CAG document on POM's, a method was developed t
obtain confidence intervals for estimates based upon the
assumptions that each epidemiological study gave an unbiase
estimate of the true slope parameter and the estimates were
distributed log normally. Adding the additional assumption
that the exposure estimates are also log-normally
distributed we derive the relationship that the 95%
confidence interval for the log of the number of leukemia
deaths per year is
1 .953289 + JTo83689 + log2u
-------
where we are 95% confident that the true exposure is between
(u-1) x 100% and (u-l)-l x 100* of the exposure estimate.
The confidence limits derived from this relationship for
various assumed values of u are shown in Table.6.
J_
-------
TABLE 1
- Data Utilized to Hstimate Lifetime.Probability -of
Death Due to Various Porms of Leukeinin
f
ro
I
Total Dcatlis
Total Death
Rate x 105
Total People
To t a I
Number of Deaths Due to
I CD COUP. NllMIUiKS
In tcrval
TD
TDK
xl0b
204
205
200
207
0-1
55
581
1,805.2
30.7 894
17
7
3
13
1-5
10
843
79.5
130.3899
157
74
4
93
5-9
7
514
41.5
181.0002
308
04
4
12 5
20-14
8
408
4 0.0
208.5714
211
88
0
92
15-19
22
900
111.9
204.7185
174
175
8
98
20-24
20
549
140.8
180.8515
70
141
4
05
25-29
22
205
143.5
154 . 7387
43
194
13
50
30-34
21
512
105.7
129.8250
28
181
9
54
35-39
20
374
235.1
112.1821
28
211
11
55
40-44
40
913
355.2
115.1830
30
241
19
08
4 5-4 9
07
349
50 3.3
119.5015
70
327
24
90
50-54
98
007
032.8
118.4702
131
407
28
134
55-59
133
004
1,314.5
101.0387
250
502
29
190
60- 04
170
973
1,94 3.9
91 .0402
370
570
42
202
05-09
213
495
2,804.7
70.1204
477
722
49
317
70-74
241
100
4 ,302.7
5G.0499
515
794
7.0
370
75-79
203
251
0,722,4
39.1003
051
730
79
357
80-04
250
985
9,777.4
25.0099
570
522
39
311
8 5 +
284
•400
17,429.4
10.3173
„ 409
304
47
254
TOTAL
*71)9.8344 x
10°
-------
TABLE 2 - Lifetime Probability of Death in U.S.
Population Due to Leukemia Type Upon
Which Relative Risk in Each of the
Epidemiological Studies is Based.
Epidemiological
Study
Infante
Ott
As key
JCD Codes
Utilized
204-207
20 S
2.05-2.07
Type of
Leukemia
Total Leukemia
Myelogenous
Nan-Lymphatic
Lifetime Probability
of Death Due to
Type of Leukemia
.006732
.002834
~004317
-25-
-------
TABLE 3 -
Summary of Data Used to Estimate Lifetime
Probability of a Leukemia Death Per ppm
Benzene Lifetime Exposure
K x? P, B
Relative Risk Estimate Average Lifetime Probability Estimated Slope
Estimate Lifetime Exposure of Leukemia Parameter
7.20 2.81 .006732 .01/1854
19.92 4,22 .004517 .020252
3.75 .17 .002884 .046380
-------
TA3L£ 4 - Source Specific 3enrene Caused Leukemia
Deaths/Year 3ased on Table 1-1 of SRI Benzene
Exposure Document3
sflurce o£
Exposure
i
eciical
jwfacturing
sice Ovens
i
i
*
troleun
fineries
.tonobile
lis s ions
.soline Ser-
es Stations
elf Service
asoline
6 Exposure in
10 x ppQ-Person. Years
8.5
.2
2.5
150.0
19.0
1.6
Expected Number of Benzene Cau.
Leukemia Deaths/Year
2.83
.07
.35
50.89
6.44
.54
TOTAL
181.8
61.67
a,
Mara, Susan J. and Shonh S. Lee. Assessment of'Human Exposures to Atmospheric
Benzene. SRI International for U.S. Environmental Protection Agency, Research
Triangle Park, NC. Publication No. EPA-450/3-78-031. June 1978.
-97_
-------
TABLE 5 - Total Exposure of People Residing in Variou:
Locations, and Resulting Estimated Benzene-
Caused' Leukemia Deaths/Year - Based on
Table 1-2 of SRI Benzene Exposure Document3
Vicinity of Residence
Exposure in
lQ^ppk-Person
Years
Expected Number o£
Benzene-Caused
Leukemia Deaths/Ye:
Chemical Manufacturing
Coke Ovens
Petroleum Refineries
Urban Areas
TOTAL
10.0
.2
4.5
250.0
264.7
3.59
.07
1.53
84.80
89.80
Mara, Susan J. and Shonh S. Lee. Assessment of Human Exposures to Atmospheric
Benzene. SRI International for U.S. Environmental Protection Agency, Research
Triangle Park, NC. Publication No. EPA-450/3-78-031.
-28-
-------
TABLE.6 - Confidence Limits on Total Benzene
Caused Leukemia Deaths Per Year
(Assumes "One-Hit" Model is the True
Dose Response Relationship)
Level of Precision
Assumed for Exposure 95% Confidence Limits
Estimate * Lower Upper
(U-l) x 100% _ Limit Limit
OS*
46.1
174.8
10%
4-5.8
176.0
50%
41.2
195.9
O
O
34.3
234.9
1000%
7.5
1081.7
*Assumes no error' in exposure estimate.
-29-
-------
Bibliography
Cancer Risk of Benzene
Askoy, M. (1977) "Testimony of Mazaffer Askoy, M.D. to
Occupational Safety and Health Administration, U.S.
Department of Labor," July 13, 1977, 13 pp.
Askoy, M. (1976) "Types Leukemia in Chronic Benzene
Poisoning. A Study in Thirty-Four Parts," Acta Haemat,
55, 65-72.
Askoy, M., Erdem, S., and Dincol, S. (1974) "Leukemia in
Shoe-Workers Exposed Chronically to Benzene," B1ood, 44 (6),
837-341.
Askoy. M., Erden, S., Dincol, K., Kepyuksel, T., and Dincol,
6. (1974) "Chronic Exposure to Benzene as a possible
Contributary Etiologic Factor in Hodgkin's Disease," B1 ut,
28, 293-298.
Baier, E.J. "Statement of the National Institutes of
Occupational Safety and Health," Exhibit 84A, QSHA Benzene
Hearings, July 19-Aug 10, 1977.
Cooke, J.Y. (1954) "The Occurrence of Leukemia," B1ood 9,
34Q.
Goldstein, B.D.,. Snyder, C.A., Snyder, R., and R. and S. R.
Wolman, "Review of Benzene Toxicity," Prepared for EPA,
August 18, 1977.
Infante, P.F., Rinsky, R.A., Wagoner, J.K., and Young, R,J.
(1977) "Benzene and Leukaemia (Letter to the Editor),"
Lancet, 867-869, October 22, 1977.
Infante, P.F., Rinsky, R.A., Wagoner, J.K., and Young, R.J.
(1977) "Leukemia in Benzene Workers,' " Lancet, 76-78, July
9, 1977.
Ott, M.G., Townsend, J.C., Fishback, W.A., and Langer, R.A.
(1977) "Mortality Among Individuals Occupational1y Exposed
to Benzene," Exhibit 154, OSHA Benzene Hearings, July 9-
August 10, 1977.
Sakol, M.J. ( 1977) "Testimony of Marvin J. Sakol , M.D., to
Occupational Safety and Health Administration, U.S.
Department of Labor".
-30-
-------
VII. APPENDIX
Mutagenic Risks of Benzene Exposure
Summary
In addition to the risk from.leukemia, benzene exposure
is also likely to Induce inherited mutations. The magnitude
of this risk can not be estimated because of the uncertain
quantitative relationship between heritable mutations and
chromosome aberrations which have been consistently observed
in exposed workers.
Review of Experimental Results
Benzene was found to be non-rautagenic in the Ames' test
for point mutational effects (Simmon et a!., 1977; Shahin,
1977; and Lyon, 1975). However, it is possible that a human
metabolic activation enzyme system or a mammalian body fluid
activation system would cause it to be mutagenic.
Somatic chromosomal aberrations have been demonstrated
In animals and humans. In rabbits, Kissling and Speck, 1971
reported the induction of cytogenetic damage i^ vivo by
subcutaneous Injection of 0.2 ng/kg day benzene. The
frequency of metaphase spreads showing aberrations (mostly
gaps and breaks) Increased from 5.92 to 57.8% after an
average exposure interval of 18 weeks. Two months after
discontinuance of the benzene treatment, cytogenetic damage
was still observed.
-31-
-------
Dobrokhotov (1972) exposed rats to 0.2 g/kg day benzene and
0.8 g/kg day toluene, and found similar rates of chromosomal
aberrations in the two chemicals given separately, and an
additive effect when given together. Chromatic deletions in
metaphase chromosomes of bone-marrow cells have been found
1n rats given single doses of benzene subcutaneously at 2
ml/kg (Philip and Jensen, 1970). Deletions have also been
observed in rats given benzene at Ig/fcg day, subcutaneously,
for 12 days.
A dominant lethal and iji vivo cytogenetics combined
test has been performed with rats dosed intraperitoneal^
v/ith 0.5 ml/kg benzene (Lyon; 1975).. No dominant lethality
was found but increases were found in chromatic and
chromosomal aberrations. Lyon (1975) also found increased
micronuclei counts 6 hours after the final dosing of rats at
0.05 and 0.25 tal/kg/day after two days of dosing
Intraperitoneal^.
In patients with benzene-induced aplastic anemia,
lymphocyte chromosome damage has been found (Pollini and
Colombi, 1964). Pol 1ni et a!., (1964) later found a 70*
incidence of heteroploid chromosomal patterns in the blood
lymphocytes and bone marrow parenchyma cells of each of four
subjects with benzene-induced blood dyscrasias. Similar
patient studies of benzene exposed individuals with
-32-
-------
persistent chromosomal alterations associated with blood
dyscrasias have also been reported by others ("Forni and
Moreo, 1967, 1969; Hartwlch et al., 1969; Khan and Khan,
1973; Sellzel and Kelemen, 1971; Forni et al., (1971); Tough
and Court Brown, 1965).
Vigiliani and Forni (1969) found a significant increase
of chromosomal aberrations in pheripheral lymphocytes of
workers exposed to benzene, but not 1n those exposed to
xylene and toluene. Some of these aberrations persisted for
several years after recovery from benzene hemopathy. ¦ They
suggested that toxicity to the bone marrow might result 1n
cells with an abnormal number of chromosomes and that
proliferation of these cells could then give rise to an
advantaged leukemic clone. Forni et aU, (1971) examined
chromosomal aberrations in 34- workers in a rotogravure plant
and.compared these to 24 matched controls, and found a
significantly higher number of both stable and unstable
aberrations in 10 benzene-exposed workers but a number
comparable to controls in all of the 24 toluene-exposed
workers.
A recent report (Kilian and Daniels, 1978) on.52 workers
exposed to benzene for one month to 26 years (mean of 56.6
months) found chromosomal aberrations (chromosome breaks,
dicentric chromosomes, translocations and exchange figures)
-------
1n peripheral lymphocytes at 2-3 times the rates found in
controls. In this study, the 8 hour average tlnie-weighte
benzene exposure was 2-3 ppm, the average concentration
determined by 15 minute sampling was 25 ppm and the peak
concentration was 50 ppm.
The same laboratory reported on the monitoring of 471
peripheral lymphocyte cultures from 290 Texas Division
benzene workers between 1965 and 1978 (Benge et al.,.1978)
A group of 972 "preemployment examinees" who were judged,
the basis of the history taken at the time, to have had
negligible exposure to known chramosoma-breaking agents we;
used as controls. Rates of chromosomal abnormalities were-
found not to be increased in the exposed group over the
control group. The time-weighted average benzene
concentrations were estimated to have been below 50 ppm
prior to 1972 and well below 10 ppm from 1973 ta the presen
time.
A report by Picclano {1978} which is a further analysis
of the Kilian and.Daniel ( 1978) and Senge ( 1978) study
comparing the Information on benzene exposed Individuals to
a 44-person group seen for preemployment examination.
Workers exposed to three different levels of benzene at les:
than 10 ppm for several years showed a dose response
relationship. The types of aberrations detected are similar
to those reported for higher benzene exposures by (Tough et
-34-
-------
al., 1970). The workers were monitored for urinary
excretion of phenol which is a primary metabolite of
benzene. All workers had no detectable phenol which
indicated no recent exposure to benzene. Exposures for the
dose-response relationship were 0, less than 1, 1-2.5 and
greater than 2.5 ppm.
Fredga et al., (1978) performed a study on 65 workers,
occupationally handling motor fuels. A moderate, but
statistically significant, increase in frequency of chro-
mosome aberrations was found in road tanker drivers and
industrial workers, but net in ship tanker crews and gaso-
line station staff. The estimated exposure dose was 60 ppm
or less. The dose absorbed will be reported in a subsequent
study.
Conclusions
Ample evidence exists that benzene causes chromosomal
aberrations in animals and humans exposed to benzene. This
evidence was reviewed above. However, since this is a
somatic cell effect as opposed to a germinal cell effect it
1s difficult to estimate the heritable risk to future
generations from such evidence. These chromosomal aber-
rations probably involve breaks in DNA and therefore are
heritab!e" events if they occur in the germinal cells,
although the experiments to prove that point have not been
-------
It 1s generally recognized that rings, dicentrics,
trans!ocatlons and exchange figures are heritable, but
chromosome breaks could be caused from toxicity of scmat
cells and therefore may not be heritable. The former
lesions should be used as Indicators that genetic damage
future generations may have occurred. At the current ti
quantitative estimates of heritable genetic damage due t
benzene cannot be made from data on the frequency of som
mutations, although this damage may be occurring at
concentrations as low as 1 ppm in air.
-------
Bib!iography
Mutagenicity: Benzene
Benge, M.C., J.R. Yenable, D.J. Picciano and D.J.,Ki11ian
(1977) "Cytogenetic study of 290 Yorkers exposed to
benzene.u Report Dow Chemical U.S.A. 1977.
Dobrokhotov, Y.B. (1972) "The mutagenic influence of benzene
and toluene under experimental conditions," Gig. Sanit;
37:36-39.
Fornl, A., and L. Moreo (1967) "Cytogenetic'studies in a
case of benzene leukaemia,'1 Eur. J our. Cancer, 3:251-255.
Fornl, A., and L. Moreo. 1969. "Chromosome studies in a case
of benzene-1nduced erythroleukaemia." Cancer, 3:251-255.
Fornl, A., et al. (1971a) "Chromosome studies in workers
exposed to benzene or toluene or both," Arch. Environ.
Health, 22:373-378.
Forni, A., et al. {1971b) "Chromosome changes and their
evolution in subjects with past exposure to benzene," Arch.
Env1ron. Health, 23:325-391.
Fredga»K«, Reitalu, J, and Berlin, M. Chromosome studies in
workers exposed to benzene. Report #771018 of the Institute
of Genetics, Univ. of Lund and Department of Env. Health,
University of Lund, Sweden.
Haberlandt, W., and B. Ments 1971 "Deviation on number and
structure of chromosomes in industrial workers exposed to
benzene," Zbl> Arbeitsmed, 21:338-341.
Hartwich, G., and G. Schwanitz (1972)
"Chromosomenuntersuchungen nach chronischer
Benzol-Exposition." Dtsch. Med. Wschr. 97:45-49.
Hartwich, G«, et al. (1969) "Chromsome anomalies in a case
of benzene leukemia." Ger. Med. Monthly, 14:449-450.
Khan, H., and M. H. Khan (1973) "Cytogenetic studies
following chronic exposure to benzene," Arch. Toxikol.,
31:39-49.
Kilian, D. J., and R. C. Daniel (1978) "A cytogenetic study
of workers exposed to benzene in the Texas Division of Dow
Chemical, U.S.A.," Feb. 27, 1978, Dow Chemical, Freeport,
Texas.
Kissling, M., and B. Speck (1971) "Chromosomal aberrations
1n experimental benzene intoxication," Helv. Med. Acta,
36:59-66.
-------
Koizumi, A., et al. (1974) "Cytokinetlc and cytogenetic
changes in cultured human leucocytes and HeLa cells induced
by benzene," Ind. Health (Japan), 12:23-29.
Ltapkalo, A.A. (1973) "Genetic activity of benzene and
toluene," Gig Tr. Prof. Zabol., 17:24-28.
Lyon, J.P. (1975) "Mutagenicity Studies with Benzene," Ph.D.
Thesis, University of California
Philip, P., and M.K. Jensen. (1970) "Benzene-induced
chromosome abnormalities in rat bone marrow cells," Acta.
Pathol. Microbiol. Scand. Sect. A., 78:489-490.
Picciano, D. (1978) Unpublished report.
Pollini, G., and R. Colorabi (1964) "Lymphocyte chromosome
damage in benzene blood dyscrasia," Med. Lav. 55:641-654.
Pollini, G., et al. (1964) "Relationship between chromosoma
alterations and severity of benzol blood dyscrasia," Med.
Lav. 55:735-751.2
Sellyei, M., and E. Kelemen (1971) Chromosome study in a
case if g.ranul ocyti c leukemia with "Pelgerisaticn" 7 years
after benzene pancytopenia~ Eur. J our. Cancer, 7:83-85.
Shahin, M.M. (1977) Unpublished results.. The University c
Alberta, Canada. Cited in Mutation Research 47:75-97
(1978).
Simmon, Y.F., et al. (1977) "Mutagenic activity of chemical
identified in drinking water." Second International Conf.
on Environmental Mutagens, Edinburgh, Scotland, July 1977.
Tough, I.M., and W.M. Court Brown (1965). "Chromosome
aberrations and exposure to ambient benzene, "Lancet, 1:634
Tough, I.M., et al. (1970) "Chromosome studies in workers
exposed to atmospheric benzene. The possible influence of
age," Eur. Jour. Cancer, 6:49-55.
Vigliani, E.C., and A. Forni. 1969. Benzene, chromosome
changes and leukemia," Jour. Occup. Med.» 11:148-149.
V1 gl iani, E.C., and A. Forni (1976) "Benzene and leukemia,1"
Environ. Res., 11:122-127. '
Vigliani, E.C., and G. Sanita (1964) "Benzene and leukemia*
N.E. Jour. Med., 271:872-876.
-------
Warren, E.W., VanVoorhee, R.F., and Sampson, A.F., "Post.
Hearing Brief of API, UPRA, and Individual Member Companies
Regarding the Proposed Revised Permanent Standards for
Occuopation Exposure to Benzene," OSHA Docket Mo. H-Q59
September 22, 1977.
Young-, R.J. National Institute of Occupational Safety and
Health, Personal Communication, October, 1977.
-39-
------- |