United States Office of
Environmental Protection Research and Development
Agency Washington, DC 20460
EPA-600/&-96-138
November 1996
* EPA Evaluation of Styrene
Emissions From a
Shower Stall/Bathtub
Manufacturing Facility
Prepared for Office of Air Quality Planning and Standards
Prepared by Air ar\d Energy Engineering Research Laboratory
-------
FOREWORD
The U.S. Environmental Protection Agency is charged by Congress with pro-
tecting the Nation's land, air, and water resources. Under a mandate of national
environmental laws, the Agency strives to formulate and implement actions lead-
ing to a compatible balance between human activities and the ability of natural
systems to support and nurture life. To meet this mandate, EPA's research
program is providing data and technical support for solving environmental pro-
blems today and building a science knowledge base necessary to manage our eco-
logical resources wisely, understand how pollutants affect our health, and pre-
vent or reduce environmental risks in the future.
The National Risk Management Research Laboratory is the Agency's center for
investigation of technological and management approaches for reducing risks
from threats to human health and the environment. The focus of the Laboratory's
research program is on methods for the prevention and control of pollution to air,
land, water, and subsurface resources; protection of water quality in public water
systems; remediation of contaminated sites and ground-water; and prevention and
control of indoor air pollution. The goal of this research effort is to catalyze
development and implementation of innovative, cost-effective environmental
technologies; develop scientific and engineering information needed by EPA to
support regulatory and policy decisions; and provide technical support and infbr-
mation transfer to ensure effective implementation of environmental regulations
and strategies.
This publication has been produced as part of the Laboratory's strategic long-
term research plan. It is published and made available by EPA's Office of Re-
search and Development to assist the user community and to link researchers
with their clients.
E. Timothy Oppelt, Director
National Risk Management Research Laboratory
EPA REVIEW NOTICE
This report has been peer and administratively reviewed by the U.S. Environmental
Protection Agency, and approved for publication. Mention of trade names or
commercial products does not constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Information
Service, Springfield, Virginia 22161.
-------
EPA-600/R-96-138
November 1996
EVALUATION OF STYRENE EMISSIONS FROM A
SHOWER STALL/BATHTUB MANUFACTURING FACILITY
Prepared by:
Larry Felix, Randy Merritt, and Ashley Williamson
Southern Research Institute
Environmental Sciences Research Department
P. O. Box 55305
Birmingham, AL 35255-5305
EPA Contract Number 68-D2-0062
Task No. 12, Phase 1
Task Officer Bobby E. Daniel
Air and Energy Engineering Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
Prepared for:
U.S. Environmental Protection Agency
Office of Research and Development
Washington, D.C. 20460
-------
ABSTRACT
Current EPA emission factors (AP-42) for styrene emissions from the production of polyester
resin reinforced plastic products represent a composite of spraying and post-spraying emissions from
shower stall/bathtub manufacturing plants that use compressed air-powered spray guns to apply
catalyzed styrene resins to prepared molds. Because each step of manufacture (gel coating, first-stage
spray lay-up, and second-stage spray lay-up) creates large surface areas from which volatile styrene
monomer can evaporate, non-spraying emissions can constitute a large fraction of the styrene emitted
to the atmosphere. Thus, it is of interest to quantify the level of non-spraying styrene emissions
characteristic of this industry.
In this study, emissions measurements were carried out at a representative facility (Eljer
Plumbingware in Wilson, NC) that manufactures polyester resin reinforced shower stalls and bathtubs
by spraying styrene-based resins onto molds in vented, open, spray booths. Styrene emissions were
characterized for the three stages of manufacture by measuring styrene concentrations at the vents of
spray booths used in each part of the process. In addition, styrene concentrations were measured at
each ventilation fan exhaust. Emission levels were determined using EPA Method 18 to obtain
integrated emissions samples and total hydrocarbon (THC) analyzers to measure continuous emissions
levels during the EPA Method 18 sampling.
Analysis of the EPA Reference Method data indicates that: (1) styrene monomer is the only
volatile organic compound released in this process; (2) overall, approximately 4% of all material
sprayed is lost to atmospheric emissions as styrene (approximately 19% of all styrene sprayed); and (3)
emissions vary for each phase of manufacture, with post-spraying emissions of styrene (from curing
molds) constituting a large part, approximately 29% of all emissions.
-------
TABLE OF CONTENTS
Abstract ii
List of Figures iv
List of Tables iv
Acknowledgments v
Metric to Nonmetric Conversions vi
1. Introduction 1
2. Project Description 4
Experimental Approach 4
Test Matrix 5
Eljer Plumbingware Facility 6
Sampling Locations 12
3. Sampling and Analytical Procedures 15
Plant Process Information 15
EPA Method 18 Sampling 17
Method 18 Sampling Equipment 18
Method 18 Sampling Procedures '. 18
Method 18 Styrene Analysis 19
Sampling with THC Analyzers 21
Volumetric Flow Rate Determination 24
4. Results 25
Recommendations for Future Sampling 37
5. References 39
Appendix A Quality Control Evaluation Report 40
Summary 40
Significant QA/QC Problems 41
Data Quality 48
Precision , 50
Bias 52
Completeness 53
Representativeness 53
Comparability 54
-------
LIST OF FIGURES
Figure Page
1. Layout of the Eljer Plumbingware Facility 8
2. Overall Arrangement for Roof Sampling Locations 13
3. Equipment Arrangement Used for Sampling with THC Analyzers 23
LIST OF TABLES
Table Page
1. Emission Factors for Uncontrolled Polyester Resin 2
2. Eljer Plumbingware Test Matrix 7
3. Plant Raw Material Consumption, June 15-17,1993 16
4. Styrene Concentration and Emission Rate Summary for Eljer Plumbingware (June 1993) 30
5. Styrene Emissions per Unit Area of Mold Sprayed, Eljer Plumbingware (June 1993) 31
6. Determination of Non-Spray Booth Styrene Emission Rate 32
7. Plant Raw Material and Styrene Usage, June 15-17,1993 32
8. Styrene Emissions for Each Part of the Manufacturing Process 33
9. Distribution of Styrene Emissions from Each Part of the Manufacturing Process, Including Styrene
Emissions not Captured by Spray Booths 37
A-1. Results of Analyses of EPA Performance Evaluation Audit Sample 45
A-2. Results of Analyses of Matheson Calibration Gas 45
A-3. GC-FID Calibrations Performed at Eljer Plumbingware 47
A-4. Data Quality Indicator Goals for Critical Measurements Estimated in QAPP 50
A-5. Data Quality Indicator Values for Method 18 and THC Measurements Made at
Eljer Plumbingware 51
IV
-------
ACKNOWLEDGMENTS
The authors would like to thank J. N. Eason, Plant Manager at Eljer Plumbingware in Wilson, NC,
for providing a site to carry out this evaluation and Rollie Nagel, Manager of Safety and Environmental
Health at Eljer, for his support and help during the test. The authors would also like to thank EPA
Task Officer Bobby Daniel of the EPA Air and Energy Engineering Research Laboratory for his help.
support, and coordination throughout this work.
-------
Metric to Nonmetric Conversions
Readers more familiar with nonmetric units may use the following factors to convert to that
system.
Metric Multiplier Ylplrlc
kPa
kPa
°C
I (1000 cm3)
m
m2
m3
kg
1000 kg (metric ton)
1450.38
4.0145
1.8T + 32
0.26417
3.2808
10.7637
35.3134
2.2026
1.1023
psig
in. H20
°F
gal.
ft
ft2
ft3
fb
ton (short)
VI
-------
SECTION 1
INTRODUCTION
Current EPA emission factors (Table 1) for styrene emissions from the production of polyester
resin reinforced plastic products represent a composite of spraying and post-spraying emissions (from
curing molds) from shower stall/bathtub manufacturing plants that use compressed air-powered spray
guns to apply catalyzed styrene resins to prepared molds.1 Because each step of manufacture creates
large surface areas from which volatile styrene monomer can evaporate, post-spraying emissions can
constitute a large fraction of the styrene emitted to the atmosphere. Thus, it is of interest to quantify
the level of spraying and post-spraying styrene emissions characteristic of this industry.
Shower stalls and bathtubs are among the many kinds of products fabricated from liquid
polyester resin that has been extended with various inorganic filler materials and reinforced with glass
fibers. These composite materials are often referred to collectively as fiberglass reinforced plastic
(FRP) or "fiberglass". Depending on the size, shape, and intended use, any one of several
manufacturing processes can be used for fabrication. For the manufacture of shower stalls and
bathtubs, the preferred technique is spray lay-up or sprayup. Regardless, all of these processes
involve the application of a liquid resin that is mixed with a catalyst to initiate polymerization. In
polymerization, a liquid unsaturated polyester is cross-linked with a vinyl-type monomer, usually
styrene, by the action of the catalyst. Common catalysts are organic peroxides, typically methyl ethyl
ketone peroxide (MEKP) or benzoyl peroxide. Resins may contain inhibitors, to avoid self-curing during
resin storage, and promoters, to allow polymerization to occur at lower temperatures."
In the production of fiberglass shower stalls and bathtubs, exhaust air from the spray booths
used for mold-coating and plant ventilation air outlets represent the major point sources of VOC
emissions. Thus, at a particular facility, the number of manufacturing steps that involve the spraying of
styrene-based resins, the amount of styrene sprayed in each step of manufacture, and the amount of
-------
Table 1. Emission Factors for Uncontrolled Polyester Resin1'4
PRODUCT FABRICATION PROCESSES (a)
(100 x mass of VOC emitted/mass of monomer input)
Process
Resin
NVS
Hand lay-up
Spray lay-up
Continuous lamination
Pultrusion(d)
Filament winding(e)
Marble casting
Closed molding(g)
5-
9-
4-
4-
5-
1 -
1 -
10
13
7
7
10
3
3
vs(b)
2
3
1
1
2
1
1
-7
-9
- 5
- 5
- 7
- 2
- 2
Emission Gel Coat
Factor
Rating*
C
B
B
D
D
B
D
NVS
26 -35
26 -35
(c)
(c)
(c)
(0
(c)
VS(b)
8-25
8-25
(C)
(C)
(c)
(0
(c)
Emission
Factor
Rating*
D
B
(a) Ranges represent the variability of processes and sensitivity of emissions to process parameters.
Single value factors should be selected with caution. NVS = nonvapor-suppressed resin. VS =
vapor-suppressed resin.
(b) Factors are 30-70% of those for nonvapor-suppressed resins.
(c) Gel coat is not normally used in this process.
(d) Resin factors for the continuous lamination process are assumed to apply.
(e) Resin factors for the hand lay-up process are assumed to apply.
(0 Factors unavailable. However, when cast parts are subsequently sprayed with gel coat, hand and
spray lay-up gel coat factors are assumed to apply.
(g) Resin factors for marble casting, a semiciosed process, are assumed to apply.
* Emission factors developed from the results of facility source tests (B Rating), laboratory tests (C
Rating), and through technology transfer estimations (D Rating).2
-------
styrene that is volatilized during the spraying and curing of molds determines the amount of styrene
emitted to the atmosphere.
This study was undertaken to quantify styrene emission factors at a shower stall/bathtub
manufacturing plant determined by the EPA to be a representative facility. Once styrene emissions
were measured, the emissions measurements and raw material usage data from the plant were used to
determine emission factors for each phase in the manufacturing process.
Testing was carried out at Eljer Plumbingware, in Wilson, North Carolina and was part of a
larger effort that also involved the evaluation of a pilot-scale liquid chemical scrubber for styrene
removal. Styrene emissions measurements were originally scheduled for the week of June 14, 1993
and the liquid chemical scrubber evaluation was originally scheduled for the following week. Because
the plant was operating on a four-day production week during this time (Monday through Thursday),
instead of the five-day production week that had been expected, emissions testing had to be extended
through Monday, June 21, 1993 to obtain a suitable set of emissions data. A full day of testing could
not be carried out on Monday because portions of that day had to be devoted to preparing for the
upcoming liquid chemical scrubber evaluation.
Section 2 contains a detailed description of the facility and sampling locations. Detailed
descriptions of the sampling methodology are presented in Section 3 and the results of this evaluation
along with a discussion of these results are presented in Section 4. The quality assurance and quality
control measures taken during this evaluation as well as the results of these measures are contained in
the Quality Control Evaluation Report in Appendix A.
-------
SECTION 2
PROJECT DESCRIPTION
EXPERIMENTAL APPROACH
Styrene is an integral part of the industrial process that produces fiberglass bath tubs and
shower stalls. In the first step of this manufacturing process, styrene monomer is mixed with polyester
resin and a pigment to create a 'gel coat" that is sprayed onto a previously prepared mold. Methyl
ethyl ketone peroxide (MEKP) catalyst is added to the mix externally as it exits the compressed air-
powered spray gun used to apply the mix to a mold. Molds are reusable and before each use the mold
is waxed and coated with a mold-release agent that also helps to provide a high gloss to the finished
product. In subsequent manufacturing steps, styrene and polyester resin are mixed with inert fillers and
sprayed onto the previously coated mold along with chopped fiberglass. Between each application the
fiberglass strands are compacted on the mold by manual rolling after which the coated mold is set
aside while the resin is allowed to cure. Because curing is an exothermic process, succeeding
manufacturing steps are usually not carried out until the coated mold has cooled. Fiberglass provides
structural support for the finished article, while the cross-linked styrene and polyester resin act as a
glue to hold the matrix together, and the inert fillers provide additional structural support and can also
provide fire retardant properties. The final stage of manufacture is to separate the finished fiberglass
product from the mold, and prepare it for shipment
The purpose of this project is to develop quantitative emission factors specific to a spray layup
polyester resin shower stall and bathtub manufacturing process. To develop these emission factors,
styrene emissions were quantified from every point of air exhaust to the atmosphere at a fiberglass
shower stall and bathtub manufacturing plant located in Wilson, North Carolina. Specifically, the
following information was required to develop quantitative emission factors for styrene:
• Determination of the emission rate of styrene emitted from process exhaust vents
during normal production.
-------
• Determination of the emission rates of any fugitive styrene emissions during storage,
transfer or mixing of the resin or gel coat by process material balance.
• Relation of emission rates of styrene to the amount of raw materials used and to the
number of units produced by collecting process data during tests of process exhaust
vents, measuring the time required to complete each unit, determining material
balances for the production process, and by computing emission factors for vents and
for fugitive emissions (weight of emissions per weight of material processed).
With the exceptions noted below, two methods were used to measure styrene emissions. A
heated Tedlar™ bag sampler was used to obtain an integrated sample of the contaminated air exiting a
representative point in each process exhaust vent (EPA Reference Method 18).sa Concurrently,
styrene emissions were measured on a continuous basis using a Total Hydrocarbon Analyzer (THC)
equipped with Flame lonization Detectors (FIDs). Sample times ranged from 40 to 45 minutes, typically
the time required to spray eight to ten molds. Sample times were dictated by the plant production rate
and the time available for sampling during a particular period of spraying.
Test Matrj^
Fifteen locations were sampled for styrene. Eleven of these locations were roof vents from
spray booths (seven booths were in use throughout the evaluation while the other four were used only
for building ventilation), while the four other locations were exhaust fans located on the side of the
building. Three exhaust fans were devoted to area ventilation for the building and one exhaust fan was
used to provide ventilation for the resin mixing room.
As noted in Section 1, because production was carried out only four days per week (Monday
through Thursday) instead of five days per week, one less day of testing was available than was
scheduled. The original test schedule called for one day of setup (Monday), four days of testing
(Tuesday through Friday), a weekend off, and the next week to be devoted to the evaluation of a pilot-
scale liquid chemical scrubber with Monday being a setup day. Unfortunately, it was not known that the
plant was on a four-day production schedule until Monday, June 14. On that day, after consulting with
the EPA Project Officer, it was decided that to obtain the maximum amount of reliable data during the
week of June 14, EPA Method 18 sampling would be focused on the seven active spray booths and
that styrene emissions from non-active booths and building exhaust fans (including the resin mixing
5
-------
room exhaust fan) would be measured with THC analyzers only. Because these measurements could
not all be completed during the week of June 14, THC sampling of exhaust fan emissions was carried
out on Monday of the following week (while preparations were underway to begin the evaluation of the
liquid chemical scrubber). No more time could be taken from the evaluation of the liquid chemical
scrubber because the following week was also to be a four-day work week with one less day of
sampling than was expected. With the permission of the EPA Project Officer, headspace samples from
styrene storage tanks were also not taken because the vats of styrene containing mix are open at their
tops, so it was decided that a sample would be taken from the air exiting the resin mix room. Table 2
shows the sampling and analytical test matrix that was used for this testing. Details of the sampling
and analytical procedures used for this evaluation are presented in Section 3.
ELJER PLUMBINGWARE FACILITY
The Eljer Plumbingware facility, diagrammatically shown in Figure 1, is located in Wilson, North
Carolina. In this figure the location of each process vent is indicated. There are a total of fifteen vents
where air is exhausted. Eleven of these vents are from spray booths, seven of which were used for
spraying during this test. Three large fans shown on Figure 1 are used to provide a continuous source
of outside ventilation air to the plant. During this testing, three booths were not used (marked as
"unused" in Figure 1) and the air exhaust on these booths was closed off (normally, all fifteen vents are
used throughout the day). Ventilation air is turned on approximately 15 minutes before the workday
starts and is shut off approximately fifteen minutes after the workday ends.
-------
Table 2. Eljer Plumbingware Test Matrix
Sample Location
Gel Coat Booth #1
Gel Coat Booth #2
Gel Coat Booth #3
Lay-Up Booth #1
Uy-Up Booth #2
Lay-Up Booth #3
Lay-Up Booth #4'
Lay-Up Booth #5'
Lay-Up Booth #6'
Back-Up Booth 01
Back-Up Booth #2'
Exhaust Fan #1
Exhaust Fan #2
Exhaust Fan #3
Exhaust Fan, Resin
Mixing Room
No. of Runs
1
4
5
4
4
e
2
3
3
1
4
5
1
3
3
1
3
3
2
1
2
1
2
1
2
4
5
2
1
Sample Type
Gas Velocity
Styrene
Sytrene/Gas Velocity
Gas Velocity
Styrene
Styrene/Gas Velocity
Gas Velocity
Styrene
Styrene/Gas Velocity
Gas Velocity
Styrene
Styrene/Gas Velocity
Gas Velocity
Styrene
Styrene/Gas Velocity
Gas Velocity
Sytrene
Styrene/Gas Velocity
Gas Velocity,
Styrene/Gas Velocity .
Gas Velocity
Styrene/Gas Velocity
Gas Velocity
Styrene/Gas Velocity
Gas Velocity
Styrene
Styrene/Gas Velocity
Gas Velocity
Styrene/Gas Velocity
Gas Velocity
Styrene
Gas Velocity
Styrene
Gas Velocity
Styrene
Gas Velocity
Styrene
Procedure
Method 1-2
Method 18 •
THC Analyzer
Method 1-2
Method 18
THC Analyzer
Method 1-2
Method 18
THC Analyzer
Method 1-2
Method 18
THC Analyzer
Method 1-2
Method 18
THC Analyzer
Method 1-2
Method 18
THC Analyzer
Method 1-2
THC Analyzer
Method 1-2
THC Analyzer
Method 1-2
THC Analyzer
Method 1-2
Method 18
THC Analyzer
Method 1-2
THC Analyzer
Velocity Meter
THC Analyzer
Velocity Meter
THC Analyzer
Velocity Mater
THC Analyzer
Velocity Meter
THC Analyzer
Sample
Duration
30 mln.
45 min.
45 min.
30 min.
45 min.
45 min.
30 min.
45 min.
45 min.
30 min.
45 min.
45 min.
30 min.
45 min.
45 min.
30 mln.
45 min.
45 min.
30 min.
33 min.
30 min.
12 min.
30 mln.
10 min.
30 min.
45 min.
45 min.
30 min.
10 min.
30 min.
18 min.
30 min.
33 min.
30 min.
31 min.
30 min.
14 min.
Analysis Method
Pitot
GC/FID
THC/FIO. Pitot
Pitot
GC/FID
THC/FIO, Pitot
Pitot
GC/FID
THC/FID. Pitot
Pitot
GC/FIO
THC/FID, Pilot
Pitot
GC/FID
THC/FID, Pitot
Pitot
GC/FID
THC/FID. Pitot
Pitot
THC/FID, Pitot
Pitot
THC/FID, Pitot
Pitot
THC/FID, Pitot
Pitot
GC/FID
THC/FID, Pitot
Volumetric
THC/FID
Volumetric
THC/FID
Volumetric
THC/FID
Volumetric
THC/FID
Volumetric
TCH/FID
No active spraying was cam'ed out in these booths during the test period. Used for ventilation
only.
-------
190ft-
EXHAUST
FAN
t
VAN
-.
MOBILE
LAB
TRAILER
RESIN
MIXING
ROOM
GELCOAT
STAGING
AREA
T
13.5 ft
_L
SERVICE
ACCESS
J.
F1BERGLASS
ROVING FEEDER
BACK-UP
BOOTHil
4
,
/
BACK-UP
BOOTHM
LAY-UP
BOOTH*8
WY-UP
BOOTH ti
LAY-UP
iOOIMM
LAY4fl>
BOOTH ta
MOLD SEPARATING
STATION
J
UY4JP
LAY-UP
SOOTH *1
GEL COAT PUMPS
UNUSED
BOOTH
UNUSED
BOOTH
UNUSED
BOOTH
RJ.TEBMAT
SUPPORT
PLANT
VENTILATION
AIR FANS
HOOF EXHAUST,
VENT STACK
PLANT OFFIC
SPACE
265ft
OELCOAT
BOOTH M
QELCOAT
BOOTH*!
*
QELCOAT
BOOTH *t
MOLD REPAIR SHOP
EXHAUST FANS
rr*
Figure 1. Layout of Eljer Rumbhgware Facility
-------
During this evaluation, the workday started at 0700. One-half hour breaks in production
occurred at 1000 hours (morning break) and 1200 hours (lunch break). The workday ended at 1400.
Thus, during a typical day of testing, all ventilation fans ran for 7.5 hours while three periods of mold
spraying took place: 0700 to 1000, 1030 to 1200, and 1230 to 1400, accounting for six hours of actual
spraying.
Each stage of manufacture except for mold separation or "pulling" is carried out in a spray booth.
At the Eljer facility the spray booths were not constructed in place but are prefabricated units
manufactured by Sinks, Inc. The available volume in each spray booth is comprised of a height of about
3.05 m (10 ft), a width of about 4.11 m (13.5 ft), and a depth of about 3.66 m (12 ft). The booths are
approximately 1 m deeper but 3.66 m back from the mouth of the booth an expanded metal grate is
mounted across the width and height of the booth on which a large sheet of air conditioning-type filter
material is mounted. The filter material is usually changed every other day. While these filters appear
quite dirty when they are removed, the results of velocity traverses made at the exit vent of a booth before
and after the filter material was replaced showed that air flow was essentially unaffected by a clean or
dirty filter.
Each spray booth is continuously vented with air from the interior of the plant that is pulled into
the booth entrance, through the filter mat, and a five-blade fan unit mounted approximately 2 m below the
roof of the building. Air pulled into the fan exits vertically through a 0.91 m (3 ft) diameter stack mounted
on the roof of the facility. Each exhaust fan has a nominal rated flow of 411 mVmin (14,500 acfm).
During testing, fan capacities of from 353.5 to 427.9 m3/min (dry, referenced to 20°C) were measured.
There are three distinct manufacturing steps that are required to produce a fiberglass shower
stall or bath tub at the Eljer facility. First, a prepared mold is mounted on a cart and wheeled into one of
the three gel coat spray booths located near the mold repair shop. In the spray booth, the mold and cart
are designed to slide onto the arm of a permanently mounted pedestal assembly that can be hydraulically
elevated above the floor of the spray booth. The mold and cart are also designed to rotate on the arm of
the pedestal so that all parts of the mold are accessible for spraying. This mounting system is duplicated
in every spray booth at the Eljer facility.
-------
Gel coat is a mixture of styrene monomer, polyester resin, and pigment (chromatographic
analysis, 32.2% styrene) and is purchased as a prepared mix in 55 gallon drums. During this test, the gel
coat mix contained no additive to suppress styrene vapor emissions. At the time of this test at least four
colors of pigment were observed: white, off white, pink, and blue. However, plant records only keep track
of white and colored gel coat usage.
About two to three minutes are required to coat a bath tub mold (approximately 2.5 m2) with gel
coat and about five minutes are required to coat a large shower stall mold (7-8 m2) with gel coat After
spraying is completed, the mold is oriented upright and the pedestal is lowered until the wheeled cart
mounted to the mold contacts the floor. The mold and its attached cart are then wheeled out of the booth
to await the next stage of manufacture. Between each stage of manufacture the coated mold is set aside
to cure and harden for about an hour. Curing generates heat, so there is a time interval between
sprayings to allow the coated mold to cool.
The second stage of manufacture is called the "first lay-up' or "initial laminating" step and occurs
in two parts. In this stage, the mold is conveyed to one of the first lay-up booths and, as with the first
step of manufacture, mounted on a pedestal and prepared for spraying. The mix sprayed in this stage is
composed of a powdered inert filler added to a mixture of styrene monomer and polyester resin to form a
slurry that contains approximately 50% solids (chromatographic analysis, 21.4% styrene). The lay-up mix
is prepared in the resin mix room shown in Figure 1 and is pumped to the point of delivery.
Two coats of this slurry are sprayed onto the mold and during the spraying operation, chopped
fiberglass roving (3 to 4 cm long) is also blown at about a 30° angle into the stream of spray as it exits
the spray nozzle. The spray mixes with the strands of chopped fiberglass and forms a entangled mat of
resin impregnated fiberglass on the surface of the mold. The inert filler and the chopped fiberglass help
provide structural support to the finished product Between sprayings, the mold is left in the booth while
from two to four workers quickly compact and flatten the matted surface of the mold with small, hand-held
rollers. After the second spraying, the mold is wheeled from the booth and rolled again. The total time
for both sprayings usually takes two to three minutes and rolling can take another one to two minutes.
However, because one person is used to operate the sprayer in the three lay-up booths, the time between
sprayings averages from seven to ten minutes while other molds are being sprayed in the other lay-up
10
-------
booths. As with the first stage of manufacture, this step is brief and requires only three to five minutes to
complete. When this step is completed the coated mold is once again set aside to cure.
The third, and final, spraying step is called the "second lay-up" or "back up" step and takes place
in one of the two second lay-up booths shown in the upper left comer of Figure 1 (Back-Up Booth #1 or
Back-Up Booth #2). During this evaluation, only back-up booth #1 was used. Like the lay-up step, this
operation is carried out in two parts with two spraying steps. In this step, a blend of powdered inert filler
(incorporating a fire retardant) is added to a mixture of styrene monomer and polyester resin to form a
slurry that is contains approximately 50% solids (chromatographic analysis, 20.9% styrene). As with the
lay-up mix, the back-up mix is prepared in the resin mix room shown in Figure 1 and is pumped to the
point of delivery.
The back-up mixture is also sprayed with chopped fiberglass fibers and forms the final two layers
of the product. As with the second stage of manufacture, the mold is first moved into the back-up booth
where a fresh layer of the back-up slurry/chopped fiberglass mix is sprayed onto the mold. The mold is
then moved out in front of the booth where precut chipboard and corrugated paper supports are pressed
and molded into the wet slurry/fiberglass layer on the sides and bottom of the mold. The mold is then
moved back into the booth for a final spraying that covers all of the chipboard and heavy corrugated
paper supports. After the mold emerges from the back-up booth for the second time it is manually rolled
and set aside to cure for the last time.
Two molds are usually worked on at a time. Thus, while the supporting layer of chip board and
corrugated paper is applied to one mold, another mold is being sprayed with its first layer of back-up mix.
When the second mold is moved back into the booth for its final spraying, the first mold has its supporting
layer of chip board and corrugated paper applied. This step is brief and requires only seven to ten
minutes to complete two molds.
The last phase of manufacture is "pulling" or separation of the mold from the completed shower
stall or bath tub. After the finished fiberglass piece is trimmed and inspected it is prepared for shipment.
During the time of the emissions testing, on a daily basis, this facility consumed approximately
500 kg gel coat mix and approximately 4000 kg of lay-up mix and back-up mix. On the basis of a four-
day work week, which was typical for that time, and a 50-week work year, yearly gel coat mix
11
-------
consumption would be approximately 100,000 kg, and yearly consumption of lay-up and back-up mix
would be approximately 800,000 kg.
SAMPLING LOCATIONS
Two types of sampling locations were encountered in this evaluation. The first type of sampling
location was located on the roof of the facility and consisted of a 0.91 m (3 ft) diameter stack that
extended approximately 1.5 m (5 ft) above the surface of the roof. Eleven such stacks are used to
convey vent air to the outside from the eleven spray booths with operating exhaust fans. Moveable split
circular doors are positioned approximately 0.3 m (1 ft) below the top of each stack. These split doors are
designed so that when the exhaust fan feeding the stack is turned on, the doors fold together, up and out
of the way, kept open by the outwardly moving air stream. When the exhaust fan feeding the stack is
turned off, the doors close to prevent the entrance of rain or vermin. The actual exhaust fan for each
stack is located approximately 2 m (6.6 ft) below roof level. The fans typically have five blades. Figure 2
shows the overall arrangement for one of these sampling locations.
As Figure 2 shows, because of the eductor-type shroud that is mounted near the surface of the
roof, and because of the split door covers, it was necessary to conduct all sampling from as close to the
top of the roof as was. possible. Below the eductor-type shroud and above the rain cap, two small 2.5 cm
(1 in.) holes were drilled in the side of each stack at a 90° angle to each other, just large enough to allow
the entry of an s-type pitot or Method 18 sampling probe. Because the vents were at essentially
atmospheric pressure, no significant amount of air entered or was lost through these small ports. The
ports were covered over at the end of the evaluation.
The second type of sampling location was at the exhaust of a three or four-bladed exhaust fan
on the side of the building shown in Figure 1, about 3 m (10 ft) above the ground. These fans were
protected from the elements by horizontal louvers on the discharge side of the fan that close when the fan
is turned off, but are kept open by air exiting the fan when it is turned on. Because the louvers tended to
oscillate from nearly fully open to approximately 75% open while they operated, the louvers had to be held
fully open to be able to measure the velocity of the air exiting any one of these fans. Even so, because
these fans operate with no stack to straighten or direct their exhaust, air flow is very uneven.
12
-------
Weather Flaps
Held Open by Air Flow
AIRFLOW
Sampling Port
ROOF LINE
-3ft.
Figure 2. Overall Arrangement for Roof SampOng Locations.
13
-------
Three exhaust fans were located on the northwest side of the building. Another fan of this type was
located on the southeast side of the building and was used to ventilate the resin mixing room. Air velocity
was measured with a direct reading velocity meter that was checked against the s-pitot used at the spray
booth exhausts for air flow measurement.
14
-------
SECTION 3
SAMPLING AND ANALYTICAL PROCEDURES
The sampling and analytical procedures used for this evaluation incorporated the most recent
revisions of the published EPA methods, where applicable. In this section, descriptions of each sampling
and analytical method that was used in this evaluation are presented.
In order to meet the test objectives in an expedient manner, a trailer housing a mobile laboratory
was set up on site. All gas chromatographic (GC) analysis of the integrated Tedlar bag samples for
styrene were performed there. This trailer was also equipped for sampling two separate emissions
sources with THC monitors.7 The trailer was supplied by DEECO, Inc. of Gary, NC who also provided
personnel to conduct the Method 18 sampling and GC-FID analysis. The mobile laboratory was located
on the northeastern side of the building, outside, next to the gel coat staging area. In addition to the
mobile laboratory, a second vehicle, a van, was equipped for THC sampling and was used in this
evaluation and for the evaluation of a liquid chemical scrubber during the week following this testing. The
van was located in front of the mobile laboratory, along the northeastern side of the building.
Electrical power was provided to the mobile laboratory from plant service (a 220 VAC line)
through cabling supplied by Southern Research. Electrical power was connected and disconnected by a
local electrical contractor familiar with the plant power system. Electrical power to the van for the THC
analyzers and ancillary electrical equipment was obtained from one leg of the 220 VAC line that fed the
mobile laboratory. Power for roof top sampling was carried by heavy-duty extension cords from a power
distribution center installed by the electrical contractor for the duration of the test.
PLANT PROCESS INFORMATION
Eljer Plumbingware provided plant raw material usage records for the three days during which
spray booth samples were taken. These records are reproduced in Table 3. These data gave raw
material usage in terms of kilograms of product used in each phase of manufacture and in
15
-------
Table 3. Plant Raw Material Consumption. June 15-17, 1993*
Item
Gel Coat Mix (32.2% Styrene*)
Area Sprayed
White Mix
Colored Mix
MEKP
Total
Lay-Up Mix (21.4% Styrene*)
Area Sprayed
Resin (incl. styrene)
Inert Filler
Pigment
Vapor Suppressant
MEKP
Fiberglass Roving
Total
Back-Up Mix (20.9% Styrene*)
Area Sprayed
Resin (incl. styrene)
Inert Filler
Pigment
Vapor Suppressant
MEKP
Fiberglass Roving
Total
Units
m2
kg
kg
kg
kg
m2
kg
kg
kg
kg
.kg
kg
kg
m2
kg
kg
kg
kg
kg
kg
kg
6/15/93
664.6
290.8
197.3
7.3
495.3
640.5
958.0
1272.3
1.8
5.0
27.7
306.2
2571.0
640.5
775.2
1166.6
1.4
4.1
29.0
163.3
2139.6
6/16/93
646.0
295.7
220.4
7.7
523.9
642.1
909.9
1208.8
1.4
4.5
26.3
306.2
2457.1
642.1
721.2
1085.4
1.4
3.6
27.2
161.9
2000.8
6/17/93
653.0
286.2
226.8
7.7
520.7
652.1
923.5
1226.5
1.4
4.5
26.8
349.3
2532.0
652.1
716.2
1077.7
1.4
3.6
26.8
165.6
1991.3
* Results for sample obtained on 6/22/93 at 1020 hours.
16
-------
kilograms of product used per square foot of mold sprayed. The total area of mold that was sprayed in
each phase of manufacture was also provided as part of the daily record.
To relate the emissions data taken at each active spray booth to raw material usage, records
were kept of the area of molds that were sprayed in each active booth while sampling was underway.
Because plant production records do not track where a mold was sprayed, portable video cameras were
used to record the spraying activity that took place in a particular spray booth while the air exhausted from
that booth was being sampled. Afterwards, by inspecting the video record, mold types could be identified
and, with the help of catalog data sheets provided by Eljer Plumbingware, the total mold area sprayed
could be determined for each period of sampling. In this way, styrene emissions per unit area of mold
sprayed could be determined to compare with styrene use per unit area of mold sprayed.
Unfortunately, one video camera failed partway through the test and two data sets from the same
spray booth were lost (gel coat booth #2). Data from two other spray booths were lost when one camera
was inadvertently pointed at the wrong spray booth (lay-up booth #3 instead of lay-up booth #2) and when
plant production was shifted away from one booth so that no molds were sprayed during the time that
booth was sampled (gel coat booth #1).
EPA METHOD 18 SAMPLING
Samples for on-site analysis of styrene content were collected using EPA Method 18, according to
Section 7.1.1 of the method: Integrated Bag Sampling and Analysis, Evacuated Container Sampling
Procedure.6 Time integrated samples were collected for a typical collection time of 45 minutes by drawing
spray booth exhaust air into new 15 liter Tedlar or Teflon bags at constant sampling rates. With one
exception, samples were collected in triplicate or quadruplicate at the locations described in Table 2.
Each sample was analyzed on site by gas chromatography utilizing a flame ionization detector (GC/FID).
Stack parameters of velocity and flow rate were determined by Methods 1-2.
17
-------
Method 18 Sampling Equipment
Styrene-containing samples of spray booth exhaust were collected in evacuated Tedlar and Teflon
bags. The sampling train consisted of a heated Teflon probe and sampling line, heated leakproof rigid
sampling container, and VOST-train sampling boxes with sampling pump and dry gas meters. Constant
sampling rates were used, typically 0.25 liter/minute. To prepare for sampling, Tedlar or Teflon bags were
blanked with dry air, evacuated, and sealed into the sampling containers.
All bags were heated to 11°C (20°F) above ambient to minimize styrene loss. This temperature
was chosen because rooftop temperatures averaged about 38°C (100°F) during most testing. The Teflon
probes and sampling lines were also maintained at 49°C (120°F). Thus, bags were evacuated, installed
in the sampling containers, and preheated to 49°C (120°F) before being taken to the sampling location.
Before, during, and after sampling, and until they were analyzed, the 15 liter Tedlar bags were maintained
at 49°C (120°F) within the sampling container.
Method18 Sampling Procedures
The order of testing was determined by spray booth availability and distance from the mobile
laboratory. Testing started with gel coat booths 1 through 3 which were the furthest away from the mobile
laboratory. Lay-Up booths 1 through 3 were sampled next, then Back-Up booth #1 and any other spray
booths that needed to be repeated.
To prepare for sampling, in the mobile laboratory, evacuated Tedfar or Teflon bags were placed in
the rigid, leakproof sampling container and the inlet of the bag was connected to one of two Swagelock*
fittings mounted in the top of the container. The lid was placed on the container and sealed. At this point
the heating jacket on the container was allowed to bring the container and its bag to 49°C (120°F). When
the container had stabilized at 498C (120°F), it was transported to the roof of the facility. On
the roof, each sample container was leak checked, and the heated probe and sample line were attached
to the sample container. The probe was placed at a point where the local velocity was approximately •
18
-------
equal to the average velocity measured for the stack, as close to the center of the stack as was possible,
and the bag and sample line were conditioned with stack gas by filling and evacuating the bag three times
in succession. Sampling began and pertinent times and sampling parameters were entered on the
appropriate run sheet. Sampling continued for 45 minutes with the exception of four runs that were cut 5
minutes short because spraying in a particular booth had temporarily ceased.
When sampling was completed, the heated sampling container and Tedlar or Teflon bag was
returned to the mobile laboratory for analysis. For analysis, contents of the heated bag were sampled
directly by the GC/FID in the mobile laboratory within one to two hours after sampling was completed.
The bag was then removed and discarded while a new bag was fitted into the rigid sample container and
allowed to heat.
Method 18 Styrene Analysis
Samples were analyzed by GC/FID in the on-site mobile laboratory. The GC/FID instrument
conditions were as follows:
Shimadzu GC-14A Gas Chromatograph with 0.5 ml sample loop.
• Column - 30 m long megabore column (0.53 mm I.D) 1 urn film thickness, Restek RTX-1 (Equivalent
to J&W DB-1)
Detector - Flame lonization
Temperature program - isothermal at 100°C
Integrator - Shimadzu CR-501.
To calibrate the GC-FID, the sample loop was flushed with a standard and then injected into the
GC-FID and the area count was measured by the integrator. This was repeated for each standard. A
19
-------
calibration curve was developed by performing a least squares fit using the concentrations and
corresponding area counts of the standards. A calibration curve was developed at the beginning of each
day of sampling.
Initially, it was planned to calibrate the GC with styrene standards prepared according to Section
6.2.2.2 of Method 18 (Liquid Injection Technique). However, at the time of this test, the equipment
required to generate such styrene standards on site was not available. Therefore, precision of the GC-
FID was established by repetitive sampling of styrene standards contained in commercially available
compressed gas cylinders. Immediately before this test, two bottles each of low and mid-range styrene
calibration standards (nominal 5 .and 50 ppmv of styrene in nitrogen) were ordered from Matheson Gas
Products. When received, the styrene concentrations were certified by vendor analysis to be 3 ppmv (for
both bottles of nominal 5 ppmv styrene gas) and 52 and 54 ppmv (for the two bottles of nominal 50 ppmv
styrene calibration gas). For a high styrene calibration standard, an unused bottle of nominal 200 ppmv
styrene calibration gas (in nitrogen) left over from a previous EPA evaluation carried out at the Eljer facility
in October 1992 was used. The concentration of styrene in this cylinder was certified by Matheson Gas
Products to be 195 ppmv when it was received in October 1992.
Time integrated bag samples were analyzed in the same manner by flushing the sample loop,
injecting the sample, noting the retention time, and measuring the area count. Identification of styrene
was based on retention time established by the calibration standards. The area count was converted into
a concentration using the least squares calibration developed with the calibration standards. Each sample
was analyzed in duplicate. No interferents were anticipated due to the fact that no other chemicals except
MEKP are used in the manufacturing process. MEKP is used in very small quantities relative to the
amount of styrene-containing mix sprayed (1.1 to 1.5% by weight) and would not be detected by the
GC/FID or the THC analyzers. Also, during a previous test conducted at the Eljer facility, GC-FID analysis
of EPA Method 18 samples obtained using the adsorption tube procedure showed only the presence of
styrene in the air exhausted from gel-coat, lay-up, and back-up spray booths.8
20
-------
At the time of testing the EPA Project Officer decided against requiring a separate analysis of an
appropriate styrene audit material. However, after sampling was over, inconsistent styrene concentrations
from concurrent GC/FID and THC measurements suggested that it would be appropriate to request that
an audit sample be supplied for analysis. The audit cylinder was analyzed using the Absorption Tube
Procedure of Method 18 and the results of this analysis was found to be well within ± 10% of the verified
concentration. Unfortunately, this showed that there was substantial error in the factory analyses of the
standard gas mixtures of styrene used for field calibration, necessitating an overall correction of the
results. Subsequently, it was determined from THC analyses using a THC analyzer that had been
calibrated with the EPA styrene Audit Standard that the calibration gas standards obtained from Matheson
Gas Products actually contained 2.2, 39.4 (Matheson-certified 54 ppmv), and 170.8 ppmv of styrene,
respectively (see Appendix A). As is also shown in Appendix A, the Matheson-certified 52 ppmv styrene
cylinder was determined by EPA Method 18 analysis (adsorption tube procedure) to contain 39.1 ppmv of
styrene. The manner in which these errors in styrene concentration were detected and the effort that was
required to correct the data taken in the field are detailed in the Quality Control Evaluation Report in
Appendix A.
SAMPLING WITH THC ANALYZERS
JUM Instruments Model VE-7 total hydrocarbon (THC) analyzers were used to monitor styrene
emissions from active spray booths during time periods when Method 18 sampling was conducted at
these spray booths. In addition, the THC analyzers were used to monitor styrene emissions from
otherwise active spray booths during mid-morning breaks and lunch breaks and to monitor styrene
emissions from all other points of ventilation air exhaust.
Four JUM THC analyzers were used for this sampling. Two of the THC analyzers were
associated with the GC/FID system in the mobile laboratory. The other two THC analyzers were kept in
the van and were used to corroborate data taken with the THC analyzers in the mobile laboratory. Figure
3 shows how. these analyzers were configured. As is shown in this figure, two long sampling hoses were
21
-------
used to convey a sample of exhaust air from a process vent to one pair of THC analyzers. These 0.635
cm (0.25 in.) outside diameter sample lines were made of Teflon and were heated to approximately 66°C
(150°F) and were 22.9 m (75 ft) long. At the flow rates used for sampling, transit time through the lines
averaged between 43 and 51 seconds.
Data from two of the THC analyzers, those in the mobile laboratory, were recorded on a
dedicated chart recorder and logged on a PC-based data acquisition system (DAS) developed by DEECO.
Inc.7 Output from each THC analyzer was averaged over a period of one minute and the one minute
averages were recorded by the data logger. The chart recorder recorded instrumental output on a
continuous basis. A S-type pitot probe with a K-type thermocouple was positioned in the stack in close
proximity to the Method 18 sampling probe and the probe used for the THC analyzers. One minute
averages of the pressure drop across the S-type pitot probe and the temperature recorded by the K-type
thermocouple attached to the probe were also logged on datalogger. In this manner, if the flow rate of air
exiting a particular spray booth was observed to change or fluctuate significantly, the run could be
aborted. Overall, flow rates through a given spray booth exhaust were observed to vary insignificantly
over time and no runs had to be aborted because of unstable exhaust flow.
The two THC analyzers in the mobile laboratory were calibrated with 650 ppm ±1% propane in
the morning and in the afternoon and instrument zero was verified with hydrocarbon-free air (less than 0.1
ppm THC). to be able to develop comparisons with the other two THC analyzers that were calibrated
with styrene, mid-range styrene calibration gas (nominal 50 ppmv) was also sampled by these instruments
and the result was recorded. See the Quality Control Evaluation Report in Appendix A for details of this
comparison. Data from the two THC analyzers in the van were logged on a separate DAS. This PC-
based DAS recorded instrumental output once a second and logged the values on a floppy disk along
with local time. Outputs from these THC analyzers were also recorded on a two-channel chart recorder
which was annotated and became part of the experimental record.
The two analyzers in the van were calibrated before sampling started in the morning, and at the
\
mid-morning break with mid-range (nominal 50 ppmv) calibration gas and were also checked for linearity
22
-------
TO HEATED
SAMPLING LINE
TO EPA METHOD 18
(NIOSH METHOD 1501)
SAMPLING APPARATUS
(Used for Other Sampling)
OUTPUT TO DAS AND
CHART RECORDER
EXHAUST
TO OTHER THC ANALYZER
SPAN AND ZERO GAS INPUTS
t
J.U.M. VE-7
THC Analyzer
ZERO
AIR
3'
ppm
52'
ppm
195-
ppm
TO OTHER THC
ANALYZER
STYRENE CALIBRATION GAS
f Manufacturer* analyst, actual value*
worn lower. S«a Appendix A)
Figure 3. Equpment Arrangement Used for Sampl'ng with THC Analyzers
23
-------
with low (nominal 5 ppmv) and high-range (nominal 200 ppmv) styrene calibration standards. The
calibration was also checked at the end of the day. As part of each calibration check, instrument zero
was verified with hydrocarbon-free air (less than 0.1 ppm THC). A 10.7 ppm propane standard was also
used to determine an instrumental response for propane.
VOLUMETRIC FLOW RATE DETERMINATION
For spray booth exhausts, volumetric flow was determined according to EPA Method 2.6 A type
K thermocouple and S-type pitot probe were used to determine flue gas temperature, and velocity,
respectively. EPA Method 1 was used to determine the number of traverse points.6 Parameters
measured included pitot pressure drop, stack temperature, static, and barometric pressure. A computer
program was used to calculate the average velocity.
Stack exhaust gas was assumed to be ambient air. Throughout the test relative humidity
averaged approximately 60% (monitored on-site with a relative humidity sensor, Cole Parmer Model 3310-
40). At typical plant air temperatures (27°C or 80°F) this corresponds to 2% water vapor in the air, the
value that was used for dry gas corrections.
The other ventilation exhausts were three or four-bladed exhaust fans. Because of the very
uneven air flows and the absence of a flow straightening exhaust stack, it was essentially impossible to
determine the location of suitable traverse points or to make measurements with an s-type pitot probe.
Thus, with the permission of the EPA Project Officer, a direct reading air velocity meter (Air-Neotronics
Model 50-4) with a long enough averaging time to suppress the effects of buffeting from the fan exhaust
was used to determine an average velocity. The direct reading air velocity meter was checked against the
s-type pitot probe in one of the'spray booth exhaust vents.
24
-------
SECTION 4
RESULTS
Emission factors were calculated from emission rates determined from THC analyzer and
Method 18 sampling and knowledge of the amount of styrene used in each stage of manufacture at the
Eljer facility (from production records for each day of testing, shown in Table 3). Knowledge of the total of
area sprayed during Method 18 sampling (from inspection of videotapes made of spraying carried out in
the booth being sampled to identify the type and surface area of mold sprayed) provided the amount of
styrene expended per square meter of mold sprayed. Comparison of the styrene emission rate
determined from THC analyzer and Method 18 sampling at the exhaust of a given spray booth with the
amount of styrene sprayed yielded the emission rate for that part of the manufacturing process.
Thus, If M, is the mass of styrene sprayed per square foot of mold area for a given phase of the
spraying operation (supplied from plant production records and analysis of samples of mix from each
spraying operation) and A, is the surface area of the fth mold sprayed, the mass of styrene sprayed during
the spraying of that mold, M,, is equal to the product of M, and A;.
M, - M. • A, (1)
If during a period of time, Ts, when THC analyzer or Method 18 sampling was conducted, n molds were
sprayed, the total mass of styrene sprayed, MT was:
A, (2)
25
-------
Now, if CSB is the concentration of styrene measured at the exhaust stack of the active spray
booth under test (from the THC analyzer or Method 18 measurement), and if the average air flow rate
through the booth under test is QSB, then the rate at which gaseous styrene is emitted through the spray
booth exhaust vent (emission rate over the time of the Method 18 measurement) is:
sa
(3)
and the total mass of styrene emitted during the time Ts is:
Msa • ESB " Ts (4)
It should be noted that this calculation presumes that none of the styrene sprayed in a given
spray booth escaped through the front of the booth into the plant There are two reasons why this
assumption is warranted: First, all spraying is directed into the booth, i.e. in reviewing the videotape
record made to identify which molds were sprayed while sampling was carried out in a particular booth,
the operator of the spray gun was never observed to direct the spray gun toward the mouth of the booth.
Second, the net .air velocity into the booth (across the mouth of the booth) ranged from 0.47 to 0.57 m/s.
which should be great enough to capture alt styrene vapor generated in the booth.
Indeed, because styrene emissions associated with spraying do not escape from spray booths
into the plant, some styrene emissions from other parts of the manufacturing process must be captured by
air being swept into the spray booths. At the Eljer facility (and perhaps at other similar facilities), molds
that have been sprayed are frequently left near the mouths of spray booths where spraying is in progress.
Thus, styrene evolved from a curing mold can be captured by an nearby spray booth. While this is
probably not a common occurrence for gel coat booths at the Eljer facility (because of limited space in
front of these booths), it is an integral part of the manufacturing process for the latter two stages of
spraying. In fact, at any one time, it is common for as many fifteen molds to be in various stages of
manufacture in the general vicinity of the lay-up and back-up booths. Also, molds are generally left in a
26
-------
lay-up booth between sprayings where the surface of the mold is rolled flat. In AP-42 it is noted that
styrene emissions are increased by such manual rolling.1
Thus, the concentration of styrene measured at the exhaust of a given active spray booth, CSB, is
actually the sum of the concentration of styrene vapor emitted in the spray booth, Cs, and the
concentration of styrene vapor swept into the spray booth from the area adjacent to the spray booth. CP:
CSB - Cs * Cp (5)
and equation (3) should be written as:
ESB = (Cs «• CP) • QSB (6)
Then the emission factor, Fse, for the spray booth under test is the ratio of the amount of styrene
emitted, MSB, from equation (4) to the amount of styrene sprayed, MT, from equation (2):
= -7 (7)
Styrene emissions not captured by the active spray booth exhaust fans (styrene evolved from
recently sprayed molds and from open resin mixing tanks) were determined in a similar manner. During
periods of normal production styrene emissions were measured at every point where plant ventilation air
was exhausted to the atmosphere (this included four ventilation exhaust fans and four spray booths that
were not in use but whose exhaust fans were kept running throughout the day). Also, during the two daily
breaks in production (morning break and lunch break), styrene emissions were measured at normally
active spray booths where the exhaust fans were left running. Thus, for a given ventilation exhaust fan.
unused spray booth, or inactive spray booth, if Cy, is the concentration of styrene measured at the exhaust
point under test (from the THC analyzer), and if the average air flow rate for this location is Qvi, then the
rate at which gaseous styrene is emitted through a ventilation fan, an unused spray booth exhaust vent,
or an inactive spray booth exhaust vent (mass emission rate over the time of the THC measurement) is:
27
-------
EVi = Cvi • Qvi (8)
Because these emissions are not directly tied to a mold surface area or stage of the production
process, mass emissions per unit time were determined by equation (8). and extrapolated to a typical day of
production using the amount ol time spent spraying and the amount ol time spent on break. Thus, if Tp is the
total time spent spraying during the three periods of mold production and if TB is the total time spent on break
(morning break plus lunch break) then the total mass of styrene emitted from a given ventilation exhaust fan
or unused spray booth is:
Mvi = EVJ ' P"P + TB) (9)
and the total mass of styrene emitted from an inactive spray booth (during a break in production) is:
M.SBJ = EVJ ' TB (10)
Thus, the emission factor, Fv, associated with styrene emissions not captured by the spray booth
exhaust fans (from recently sprayed molds and from open resin mixing tanks) is the ratio of the total amount
of styrene vented through the four ventilation exhaust fans and four unused spray booths and the seven
inactive spray booths to the total amount of styrene sprayed during a given day of production, MST:
MST
The amount of styrene sprayed on a given day, MST, was calculated from plant production records
and chemical analysis (by chromatography) of samples of each mix sprayed (to determine styrene content).
Knowledge of MST and the average mold area sprayed per day (from the three spraying operations) makes it
possible to relate styrene emissions not captured by active spray booth exhaust fans to both the mass of
styrene that was sprayed as well as total mold area.
28
-------
The overall results of the sampling to determine styrene emissions are detailed below and the
results of individual THC and EPA Method 18 runs are shown in Table 4. Styrene emissions per unit area of
mold sprayed are presented in Table 5. Table 6 shows how styrene emissions were determined for styrene
emissions not captured by the spray booth exhaust fans. Table 7 summarizes total mold surface area
sprayed, plant raw material usage, and styrene use for each stage of manufacture for each day of testing.
Table 8 combines the results shown in Tables 4 through 7 and presents average styrene emissions as a
function of the area sprayed, the total mass that was sprayed, and the total amount of styrene that was
sprayed. Finally, Table 9 shows relative styrene emissions for each stage of manufacture, including
emissions not captured by spray booth exhaust fans.
As Tables 4 and 5 show, styrene emissions based on measurements with the THC analyzers were
generally much greater than those determined by Method 18. The two techniques had been expected to
produce comparable results. Indeed, it was felt that if one technique was to be biased toward a lower
measurement, it would likely be the THC analyzers, because some styrene could be lost in the long heated
sampling lines used to convey the ventilation air samples to the analyzers. Note, however, that Method 18
employs a complex procedure to obtain a time-averaged sample of an aerosol which must be transported to
a GC-FID for analysis. In contrast, sampling with a THC analyzer is relatively straightforward and immediate.
Nevertheless, as Table 8 shows (within the uncertainty of the data) that styrene emissions as determined
from the two methods' data do overlap. It should be emphasized that because of the lack of multiple
measurements, no uncertainty could be determined for styrene emissions not captured by spray booth
exhaust fans (based only on THC analyzer data). Therefore, the uncertainties in Table 8 are minimum
values. The lack of multiple measurements for such emissions is especially unfortunate because styrene
emissions not captured by active spray booth exhaust fans are one of the largest sources of styrene
emissions.
Research conducted subsequent to the above analysis of test results has also shown differences
between EPA Method 18 measurements of styrene emissions and those using THC analyzers.910 However,
the reason for these differences remains the subject of research. It has been suggested that in the Method
29
-------
Table 4. Styrene Concentration and Emission Rate Summary for Eijer Plumbingware (June 1993)
Sampling
Location
Gel Coat #1
Gel Coat #2
Gel Coat #3
Lay-Up 11
Lay-Up#2
Lay-Up #3
8ack-Upi1
Date
6/15/93
6/17/13
6/15/93
6/17/93
6/16/93
6/16/93
6/17/93
6/16/93
6/17/93
6/16/93
6/16/93
6/17/93
Sampling
Time Span
0900-0945
1030-1115
1120-1205
1245-1330
0900-0945
1030-1115
1120-1205
1245-1330
0740-0825
0830X3915
0920-1005
0740-0825
0830-0915
0920-1005
1100-1145
1030-1110
1125-1205
1100-1145
1030-1110
1125-1205
1245-1330
1245-1330
0825-0910
0825-0910
Vent ,
Row Rate
(dsemm
374.3
374.3
374.3
374.3
373.6
373.6
373.6
373.6
362.2
362.2
362.2
358.8
358.8
358.8
358.8
427.9
427.9
427.9
404.2
404.2
370.1
370.1
378.4
378.4
— Styr
LabTHC
(ppm)
45.4
59.5
hi/A
8.9
60.5
67.3
N/A
71.6
60.9
35 3
49.4
51.4
49.5
50.9
42.0
51.6
51.4
44.0
29.7
31.3
100.7
105.1
68.3
74.3
;ne Concent
VanTHC
(ppm)
N/A
N/A
59.7
5.9
N/A
N/A
65.2
63.8
N/A
N/A
51.1
N/A
N/A
N/A
35.2
49.0
47.6
44.8
28.6
302
N/A
N/A
N/A
76.0
•ation —
Method 18
(ppm)
21.5
33.3
39.2
6.3
24.7
44.3
62.3
36.8
39.7
27.3
41,4
27.8
30.5
35.3
29.7
39.5
33,7
34.3
26,0
37.6
92.8
92.3
45,4
40.2
FOR VENTS OR BOOTHS NO STYRENE WAS SPRAYED
Exhaust 1
Exhaust 2
Exhaust 3
Resin Mix
Lay-Up #4
Lay-Up #5
Lay-Up #6
Back-Up #2
6/21/93
6/21/93
6/21/93
6/21/93
6/21/93
6/21/93
6/21/93
6/21/93
0812-0830
0840-0913
0916-0940
1326-1340
1154-1229
1302-1315
1250-1300
1316-1324
363.5
163.1
331.4
187.9
396.9
379.1
353.5
367.3
8.0
9.7
9.6
72
8.7
102
13.9
8.2
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A"
N/A
N/A
N/A
N/A
N/A
— Styr
LabTHC
(kg/in)
4.413
5.788
N/A
0,864
5.869
6.527
N/A
6,945
5.732
3.319
4.647
4.791
4.618
4.743
3,911.
5.736
5.711
4.892
3.116
3285
9,680
10.100
6.711
7.305
0.758
0.412
0.823
0.352
0.894
1.005
1.273
0.779
ene Emissio
VanTHC
(kg/hr)
N/A
N/A
5.806
0.575
N/A
N/A
6,331
6.188
N/A
N/A
4.813
N/A
N/A
N/A
3.282
5.447
5.288
4.979
3,003
3.175
N/A
N/A
N/A
7.472
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
i Rate —
Method 18
(kg/hr)
2.093
3243
3.810
0.616
2.399
4.298
6.046
3.575
3.735
2.569
3,986
2.590
2.835
3.289
2.772
4.392
3.751
3.811
2,727
3.946
8.925
8.873
4.462
3.949 i
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Total (All Vents) = 6.295
AVERAGE EMISSIONS PER BOOTH DURING MORNING AND LUNCH BREAKS (NO SPRAYIt*
Gel Coat #1
Get Coat #2
Gel Coat #2
Back-Up 11
Lay-Up #1
6/15/93
6/15/93
6/22/93
6/17/93
6/16/93
1003-1024
1003-1024
1005-1031
1000-1030
1207-1224
374.3
373.6
373.6
370.1
358.8
6.9
9.9
N/A
52
7.0
N/A
N/A
7.5
N/A
N/A
JG ACTIVITY)
N/A
N/A
N/A
N/A
N/A
0.674
0.961
0.732
0,498
0.653
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Average (per booth) = 7.0 Average (per booth) = 0.668
-------
Table 5. Styrene Emissions per Unit Area of Mold Sprayed, Eljer Plumbingware (June 1993)
Sampling
Location
Gel Coat #1
Gel Coat #2
Gel Coat #3
Date
6/15/93
6/17/93
6/15/93
6/17/93
6/16/93
Sampling
Time Span
0900-0945
1030-1115
1120-1205
1245-1330
0900-0945
1030-1115
1120-1205
1245-1330
0740-0825
0830-0915
0920-1005
Sample
Time
(min)
45
45
45
45
45
45
45
45
45
45
45
Mold
Area
(m2)
44.50
43.66
36.14
0.00
38.55
N/A"
N/A'
54.63
34.18
22.30
23.97
Vent
Row Rate
(dscmm)
374.3
374.3
374.3
374.3
373.6
373.6
373.6
373.6
362.2
362.2
3622
— Slyrei
THC
(Q/m*)
74.4
99.4
120.5
N/A
114.2
N/A
N/A
90.2
125.7
111.6
148.0
Gel Coat Booth Average = 11 0.5
Lay-Up #1
Lay-Up #2
Lay-Up #3
6/16/93
6/17/93
6/16/93
6/17/93
6/16/93
0740-0825
0830-0915
0920-1005
1100-1145
1030-1110
1125-1205
1100-1145
1030-1110
1125-1205
Baek-Up #1
6/16/93
6/17/93
1245-1330
1245-1330
0825-0910
0825-0910
Relative J
45
45
45
4S
40
40
45
40
40
Standard Oe\
28.43
30.10
27.22
34.19
22.99
19.51
N/A"
40.13
28.24
nation (%) a
358.8
358.8
358.8
358.8
427.9
427.9
427.9
404.2
404.2
20.5
126.4
115.1
130.7
78.9
162.1
187.9
N/A
50.8
76.3
Lay-Up Booth Average = 1 1 6.0
Relative J
45
45
45
45
Standard Da\
103.58
103.58
80.27
80.27
rtaton (%) =
370.1
370.1
378.4
378.4
39.7
70.1
73.1
62.7
69.0
Back-Up Booth Average = 68.7
Relative Standard Deviation (%) = 6.4
ie Emitted —
Method 18
{q/m'5
35.3
55.7
79.1
N/A
46.7
N/A
N/A
49. 1f
81.9
86.4
121.9
69.5
40.6
68.3
70.6
90.6
60.8t
127.3
128.2
N/A
45.3
93.1
85.5
35.3
64.6
64.3
41.7
36.9
51.9
28.2
Video camera failure.
" Video Camera pointed at wrong booth, no data,
t Value may be in error due to shift in GC calibration during analysis.
-------
Table 6. Determination of Non-Spray Booth Styrene Emission Rate
Activity
All Vents During Spraying' =
All Vents During Breaks" =
Usually Active Spray Booths During Breaks1 =
All Vents, All Day, (THC measurements only) =
Amount
37.773
9.443
7.012
54.227
Unit
kg/day
kg/day
kg/day
kg/day
* Active spray booths excluded, (6 h of ventilation/day) x (6.295 kg/h from non-spraying vents).
** Active spray booths excluded, (1.5 h of breaks/day) x (6.295 kg/h from non-spraying vents).
f Booths normally used for spraying, (1.5 h of breaks/day) x (0.668 kg/h per booth during
breaks) x (7 booths).
Table 7. Plant Raw Material and Styrene Usage, June 15-17,1993
Date
6/15/93
6/16/93
6/17/93
Total
Mass of Material Used'
Gel Coat
(kg)
488.1
516.2
513.0
1517.3
Lay Up
(kg)
2237.1
2124.6
2155.9
6517.6
Back Up
(kg)
1947.3
1811.6
1798.9
5557.8
— Amount Sprayed —
Total"
(kg)
4672.4
4452.4
4467.9
13592.7
Styrene"
(kg)
1042.9
999.5
1002.5
3044.9
— Styrene Emitted —
THC
(kg)
246.0
244.2
246.9
737.1
Method 18
(kg)
188.4
187.4
189.2
565.0
Date
6/15/93
6/16/93
6/17/93
Total
Area Sprayed
Gel Coat1
(m2)
664.6
646.0
653.0
1963.6
Lay Up
(m2)
640.5
642.0
652.1
1934.5
Back Up
(m2)
640.5
642.0
652.1
1934.5
Areal Density
Gel Coat
(kg/m2)
0.7344
0.7990
0.7856
0.7727
Lay Up
(kg/m2)
3.4930
3.3092
3.3063
3.3691
Back Up
(kg/m2)
3.0404
2.8217
2.7588
2.8729
* Not counting MEKP. See Table 3 for details of raw material use.
" Gel Coat = 32.2% styrene, Lay-Up Mix = 21.4% styrene, Back-Up Mix = 20.9% styrene.
f Includes molds rejected due to imperfections.
32
-------
Table 8. Styrene Emissions for Each Part of the Manufacturing Process
(a) Styrene Emissions per Unit Area of Mold Sprayed
Emissions from:
Gel Coat Booths
Lay Up Booths
Back Up Booths
Non-Spray Booth Emissions
All Emissions
THC Analyzer
Styrene
(Q/m°)
110.5
116.0
68.7
83.7
378.9
Pop. Std.
Dev.
(Q/mJ)
21.2
43.1
3.8
N/A
48.1'
EPA Method 18
Styrene
(9/m2)
69.5
85.5
51.9
83.7
290.6
Pop. Std.
Dev.
(g/m!)
26.4
28.3
12.6
N/A
40.7'
(b) Percent of Total Mass Used in Each Stage of Manufacture that was Emitted as Styrene
Emissions from:
Gel Coat Booths
Lay Up Booths
Back Up Booths
Non-Spray Booth Emissions
All Emissions
THC Analyzer •
Styrene
14.3
3.4
2.4
1.2
5.4
Rel. Std. Dev.
2.7
1.3
1.5
N/A
0.7'
EPA Method 18
Styrene
9.0
2.5
1.8
1.2
4.2
Rel. Std. Dev.
3.4
0.8
0.4
N/A
0.6'
(c) Percent of Styrene Used in each Stage of Manufacture that was Emitted
Emissions from:
Gel Coat Booths
Lay Up Booths
Back Up Booths
Non-Spray Booth Emissions
All Emissions
THC Analyzer
Styrene
44.4
16.1
11.4
5.3
24.2
Rel. Std. Dev.
8.5
6.0
0.6
N/A
3.3'
EPA Method 18
Styrene
27.9
11.9
8.6
5.3
18.6
Rel. Std. Dev.
10.6
3.9
2.1
N/A
2.8'
Minimum estimate. Assumes each process independent with no contribution from the non-spraying emissions
component.
33
-------
18 procedure, styrene can polymerize before analysis and may also have a very low vapor pressure at
stack or instrument conditions. Both of these conditions would result in a lower measurement for styrene.
Although it is not specifically noted in AP-42,1 it is reasonable to assume that styrene emission
factors cited in this standard for polyester resin plastics products fabrication include emissions not captured
by active spray booths. Thus, in order to compare the results obtained in this study with those cited in AP-
42, it is necessary to apportion styrene emissions from non-spray booth emissions to those parts of the
process associated with spraying operations: gel coating, lay-up and back-up. The following approach was
followed to apportion non-spray booth emissions to each of the spraying operations.
1. Obtain an average emission factor for each of the spraying operations by averaging the Method
18 and THC measurements for each of the three spraying operations (gel coating, lay-up and
back-up)
2. Apportion emissions not directly associated with spraying to each of the three spraying
operations as follows:
2.1 Assume that vents that are in the close proximity to a particular spraying operation capture
the emissions (from curing molds) generated by that operation. Some non-spraying
emissions are surely captured by active spray booths. However, for the purposes of
apportioning emissions to obtain a comparison with AP-42, we will assume that all
emissions captured by an active spray booth are emissions directly related to that phase of
manufacture.
2.2 Thus, because the area of the plant used for gel coating is more or less off by itself, we
assume that the three exhaust fans on the side of the plant adjacent to the gel coat spray
booths capture curing emissions from molds previously sprayed in one of the three gel coat
booths. Likewise, we will assume that lay-up booths 4, 5, and 6, (that were not used for
spraying) capture styrene emissions from molds that were sprayed in lay-up booths 1, 2,
34
-------
and 3, and we will assume that back-up booth 2 (that was not used for spraying) captures
styrene emissions from molds that were sprayed in back up booth 1.
2.3 Similarly, we will assume that during breaks in spraying (morning break, lunch) any
emissions captured by a particular spray booth came from molds that were sprayed in the
phase of manufacture normally associated with that spray booth. This includes styrene
emissions captured during the 15 minutes before work commenced, the 30 minute
midmoming break, the 30 minute lunch break, and the 15 minutes that the vents are
operated at the end of the day after spraying has ended.
2.4 Assume that resin mixing room emissions that are exhausted to the atmosphere can be
apportioned to emissions from the lay-up and back-up operations because these resins are
only used in those phases of manufacture; gel coat is supplied in pre-mixed 55 gallon
drums. We will apportion the resin mixing room emissions according to the mass of mix
sprayed in each of the two operations.
• Using these assumptions, the following emission factors were calculated:
Gel Coat 47.5% of the styrene sprayed in that phase of manufacture
Lay-Up 20.0% of the styrene sprayed in that phase of manufacture
Back-Up 12.1 % of the styrene sprayed in that phase of manufacture
These data suggest that spray booth emissions are higher than those cited in AP-42 for gel coating
and spray lay-up (see Table 1, reproduced from AP-42, for styrene emission factors).1 AP-42 cites a value
of from 26 to 35% of styrene monomer being emitted for gel coat that contains no vapor suppressing
additives (as was the case at Eljer). The results calculated according to the above procedure to apportion
non-spraying emissions to each part of the manufacturing process that incorporates spraying suggest that
nearly 48% of the styrene in the gel coat mix is lost to the atmosphere.
35
-------
AP-42 makes no distinction between styrene emissions from lay-up booths or from back-up booths.
and indicates that, with vapor suppressing additives in the mix, from 3 to 9% of the styrene sprayed in this
operation is emitted. If vapor-suppressing additives are not added to the mix, emissions rise to from 9 to
13% of the styrene sprayed. At Eljer, vapor suppressants are added to the lay-up and back-up mix.
However, the levels of styrene emissions measured there suggest that the emissions levels are probably
higher than what AP-42 cites as typical for non-vapor suppressed emissions, particularly for the lay-up phase
of manufacture. Thus, results calculated according to the above apportioning procedure show that styrene
emissions to the atmosphere averaged 20% of the styrene sprayed in the lay-up booths and 12% of the
styrene sprayed in the single back-up booth.
These generally higher than expected emission levels may be, at least in part, due to the nature of
the process. At the Eljer facility (which is probably like other such manufacturing plants), molds that have
been sprayed are frequently left near the mouths of spray booths where spraying is in progress. Hence,
styrene evolved from a curing mold can be captured by an adjacent spray booth. While this practice is not
common in the gel coat booths (because of limited space in front of the booths at the Eljer facility); this
practice is an integral part of the manufacturing process for the latter two stages of spraying, in fact, at any
one time, it is common for as many fifteen molds to be in various stages of manufacture in the general
vicinity of the lay-up and back-up booths. Also, molds are generally left in a lay-up booth between sprayings
where the surface of the mold is rolled flat. In AP-42 it is noted that styrene emissions are increased by such
manual rolling.1
Finally, AP-42 provides no separate estimate of styrene emissions not captured by spray booths.
While such emissions are certainly a function of ventilation system design and the specific equipment at a
given facility, at Eljer it was found that 6% of all the styrene sprayed exits the facility through openings other
than spray booth exhausts. As noted in Table 9, this corresponds to rrom 22 to 29% (depending on the
measurement method) of all styrene emitted to the atmosphere; thus, styrene emissions not captured by
spray booths represent a source of styrene emissions as great as (or possibly greater than) styrene
emissions associated with any one of the spraying operations.
36
-------
Table 9. Distribution of Styrene Emissions from Each Part of the Manufacturing
Process, Including Styrene Emissions not Captured by Spray Booths
Date
6/15/93
6/16/93
6/17/93
Average
From THC Analyzer Measurements
Gel
Coat
Booths
29.9
29.2
29.2
29.4
Lay
Up
Booths
30.2
30.5
30.6
30.5
Back Up
Booths
17.9
18.1
18.2
18.0
Non-
Spraying
22.0-
22.2
22.0
22.1
All
Source
s
100.0
. 100.0
100.0
100.0
From Method 18 Measurements
Gel Coat
Booths
24.5
24.0
24.0
24.2
Lay
Up
Booth
s
29.1
29.3
29.5
29.3
Back Up
Booths
17.6
17.8
17.9
17.7
Non-
Spraying
28.8
28.9
28.6
28.8
All
Source
s
100.0
100.0
100.0
100.0
RECOMMENDATIONS FOR FUTURE SAMPLING
Three methods are available to measure hydrocarbon emission rates for an emission source such
as Eljer Plumbingware. Two of the three methods were used here: EPA Method 18 (using Tedlar bags to
obtain time-averaged samples) and THC analyzers (direct sampling through heated hoses). The third
method that is appropriate is a variation of Method 18 that uses a charcoal-filled adsorption tube instead of a
Tedlar bag to obtain a time integrated aerosol sample. With this method, styrene (or another VOC) that was
adsorbed on the activated charcoal during sampling is desorbed in a known volume of carbon disulfide
(usually in the laboratory). Then, a known volume of the carbon disulfide and desorbed hydrocarbon mixture
is injected directly into a GC, usually equipped with an FID.
The advantage of the first two methods is that results are obtained on site and while the Method 18
procedure using Tedlar bags is complex, it has been used successfully for some time and is well
documented.5 THC analyzers have also been used for a number of years and usually employ an FID for
detection. With these devices, time averaging is accomplished by averaging recorded instrument response
over the period of interest. Important caveats for THC analysis are that first, if more than one hydrocarbon is
37
-------
present, some way must be found to determine the relative concentration of each component (presuming the
split remains constant throughout the sampling period) and second, the instrument should be calibrated with
the compound that is being measured (at a concentration near that which is expected, if possible). An on
site GC-FID could be used for speciation (from direct sampling) or samples obtained with adsorption tubes
(taken concurrently with other measurements) could be used to provide compound speciation.
The problems encountered in this investigation were exacerbated by inaccurate styrene
concentration determinations on the part of Matheson Gas Products for the calibration gas standards.
However, the general disagreement between the THC analyzers and the EPA Method 18 Tedlar bag results
cannot be due to inaccurate calibration gases because both instruments were calibrated with the same
calibration gas standards. Thus, though absolute response in the field was in error, this problem was
eliminated when the data were corrected according to the procedures documented in Appendix A. This
disagreement does suggest that it would have been desirable to use. a third method to ascertain which of the
two other methods were in error. Some EPA Method 18 adsorption tubes should have been run during the
test (concurrent with other sampling) to provide a third, separate determination of styrene concentration. In
the future, such a procedure should be followed. Thus, in a future emissions test, two techniques (perhaps
Method 18 with Tedlar bags and THC analyzers) should be used to measure hydrocarbon concentration
and, at a minimum, some Method 18 adsorption tubes should be taken concurrently with other samples. In
the case of a disagreement, results from the analysis of the adsorption tubes could be used to determine
which technique provides the best results. In the case where time is available, Method 18 samples with
adsorption tubes could be taken on the first day of sampling and be analyzed immediately so that the results
from the adsorption tube analysis could be compared with results from the other two methods.
Two final recommendations are in order. First, all calibration standards should be independently
checked before a field evaluation commences. Second, it is also recommended that prior to any emissions
testing, an EPA audit standard should be obtained and evaluated to validate laboratory standard operating
procedures.
38
-------
SECTION 5
REFERENCES
1. Office of Air Quality Planning and Standards, Compilation of Air Pollution Emission Factors, Volume 1:
Stationary and Point Sources, AP-42, Fourth Edition (GPO 055-000-00251-7). U.S.Environmental
Protection Agency, Research Triangle Park. NC. September 1985 [supplemented October 1986 GPO
055-000-00265-7) and September 1988 (GPO-055-000-00278-9)], Section 4.12, Polyester Resin
Plastics Fabrication, Table 4.12-1.
2. Rogozen, M. B., Control Techniques for Organic Gas Emissions from Fiber glass Impregnation and
Fabrication Processes, ARB/R-82/165, California Air Resources Board, Sacramento, CA, (NTIS
PB82-251109), June 1982.
3. Modern Plastics Encyclopedia, 1986-1987,63(1 OA), McGraw-Hill, Inc., New York, NY, October 1986.
4. LaFlam, G. A., Emission Factor Documentation for AP-42 Section 4.12: Polyester Resin Plastics
Product Fabrication, Pacific Environmental Services, Inc., Durham, NC, November 1987.
5. Von Lehmden, D.J., DeWees, W.G., and Nelson, C.. Quality Assurance Handbook for Air Pollution
Measurement Systems: Volume III. Stationary Sources Specific Methods, Section 3.16. U.S.
Environmental Protection Agency, Research Triangle Park, NC. EPA/600/4-77/027b, NTIS PB80-
112303, May 1979.
6. Code of Federal Regulations 40 (Parts 53 to 60), Revised as of July 1,1991, Office of the Federal
Register, National Archives and Records Administration, U. S. Government Printing Office,
Washington. DC 20402.
7. Steinsberger, S.C.. Buynitzky. W.D., DeWees, W.G., Knoll, J.E., Midgett, M. R., Hartman, M.. U.S. EPA
Evaluation of the Hazardous Organic Mass Emission Rate (HOMER) System for Measurement of
Hazardous Organic NESHAPS Emissions. Presented at the 86th Annual Meeting of the Air and Waste
Management Association, Denver, CO, June 13-18,1993.
8. Felix, L, Merritt, R., Williamson, A.. Evaluation of the Polyad® FB Air Purification and Solvent Recovery
Process for Styrene Removal, U. S. Environmental Protection Agency, Research Triangle Park, NC.
EPA-600/R-93-212, NTIS PB94-130317, November 1993.
9. Personal Communication from Bahner, M.A., Research Triangle Institute, Research Triangle Park. NC,
to J.W. Jones, U.S. Environmental Protection Agency, Research Triangle Park, NC, April, 1996.
10. Schweitzer, J., Pultrusion Industry Council Phase II Emission Study Report, Society of the Plastics
Industry (SPI) Composites Institute, New York, NY, February, 1996.
39
-------
APPENDIX A
QUALITY CONTROL EVALUATION REPORT
SUMMARY
A Quality Assurance Project Plan (QAPP) was written and approved for this project. No field audits
were planned or performed. However, as stated in the QAPP, vendor-certified calibration gases (3, 52,54,
and 195 ppmv of styrene in nitrogen and zero air with less than 0.1 ppm THC content) served as field
performance audit samples for EPA Method 18 and THC sampling. Unfortunately, as documented below,
the concentrations of the styrene calibration gases were incorrectly determined by the vendor, Matheson
Gas Products. Actual concentrations were determined to be 2.2, 39.1, 39.4, and 170.8 ppmv of styrene,
respectively. EPA personnel were on site to oversee diagnostic measurements. In the field, QC was
addressed by adherence to standard sampling protocols either as specified for EPA Method 18 or by
• *
following a standard operating procedure (modified as needed for this particular sampling task) with the THC
analyzers as specified in the THC instruction manual.
In the Southern Research Analytical Chemistry facilities, QC is addressed by strict adherence to
standard operating procedures (SOP) previously defined and implemented. While random audits can occur
while the field samples from any project are being analyzed, and audits are regularly performed by the QA
officer at this facility, no audit was planned or performed as part of this project.
Overall, the data quality indicator (DQI) goals were not achieved. This is partially due to reduced
plant availability because the work week at the plant was one day less than was expected when the project
was initiated. Mainly, the DQI goals were not achieved because significant problems were encountered with
40
-------
the calibration gases purchased for this test. These difficulties are discussed in detail in the following section
on QA/QC problems.
With respect to plant availability, we were able to perform Method 18 sampling for only three of the
four days that were originally planned. Unfortunately, because the following week was scheduled for a
second phase of this Work Assignment, the testing of a liquid chemical scrubber, only a minimal extension of
this testing could be accommodated. Within these constraints the Project Officer concluded that in order to
obtain the needed data within the time available Method 18 sampling would focus on the operational spray
booths (a total of seven) while styrene emissions not captured by active spray booths (spray booths used as
ventilation air exhausts and separate ventilation air fan exhausts) would be measured with the THC
analyzers. Thus, from June 15 through 17, Method 18 testing was conducted at the spray booth exits and
the THC analyzers were used as continuous monitors to measure styrene emissions at the same locations
and times used for Method 18 sampling. On Monday, June 21, while preparations were underway to begin
Phase 2 of this Work Assignment, the THC analyzers were used to quantify styrene emissions not captured
by active spray booths that had not been determined during the previous week of testing.
SIGNIFICANT QA/QC PROBLEMS
One significant QA/QC problem was encountered. After sampling for both phases of the Work
Assignment had been completed, samples of the nominal 3 and 52 ppmv styrene cylinder gas standards
taken in the field on June 24 with Method 18 (Section 7.4, Absorption Tube Procedure, equivalent to NIOSH
Method 1501) were analyzed to verify sample recovery for the second phase of this Work Assignment.
These samples were analyzed on July 7. Styrene concentrations of 2.3 and 35.8 ppmv were determined
corresponding to vendor-certified values of 3 and 52 ppmv. Such large discrepancies between the styrene
concentrations certified by Matheson Gas Products and the styrene concentrations measured with the
adsorption tubes suggested that the vendor-certified concentrations of these calibration gases were in error
or that some part of the laboratory analysis performed by Southern was incorrect. Therefore, a two-pronged
41
-------
investigation followed that focused on the possibility that the styrene calibration gases were in error, that
Southern made incorrect determinations of the styrene content in the calibration gases, or a combination of
the two possibilities occurred.
With respect to the calibration gases, two bottles each of the low styrene concentration (nominal 5
ppmv styrene in nitrogen) and intermediate styrene concentration (nominal 50 ppmv styrene in nitrogen)
calibration gases were ordered from Matheson Gas Products for this test on May 13,1993 and were
received in early June. As indicated above, Matheson Gas Products certified that the styrene content in the
two bottles of low concentration gas were actually 3 ppmv while the styrene concentration in one of the
intermediate calibration standards was 52 ppmv and the other intermediate concentration standard was 54
ppmv. Two cylinders of the high calibration standard (nominal 200 ppmv styrene in nitrogen) were ordered
on September 30,1992 for an earlier EPA-sponsored test at the Eljer facility. These gases were received in
mid-October, 1992. One cylinder of this gas was not used during that test and was taken on this test for use
as a high styrene concentration calibration gas. Matheson certified that the styrene content was 195 ppmv
for this cylinder. Matheson Gas Products was contacted and a representative indicated that as far as their
records indicated, the cylinders were properly prepared and passivated and that stable styrene
concentrations were determined in their laboratory (and were recorded on the calibration tags supplied with
each cylinder) when the gases were shipped to Southern.
With respect to Southern's laboratory procedures, while conversations were being held with
Matheson Gas Products, two other samples of the 3 and 52 ppmv styrene calibration gases were taken on
July 13 and analyzed to check the procedures followed during the earlier analyses. In addition, different
high-purity liquid laboratory standards for styrene (from two different suppliers, Aldrich and Chem Service)
were used to prepare independent calibration standards that were checked against one another on the same
GC FID used for both sets of analyses. Approximately six calibration standards (of different concentrations
below, centered about, and above those measured from the earlier analyses of the adsorption tubes) were
prepared by adding a known quantity of each high-purity liquid styrene standard to a known quantity of high-
purity carbon disulfide. Known microliter volumes of these liquid mixtures were then injected into the GC-FID
42
-------
used for the adsorption tube analyses and the peak areas were recorded and averaged. No statistically
different result was determined for the two liquid styrene standards and the analyses of these two adsorption
tube samples were consistent with the earlier results. To make a definitive assessment of the actual styrene
content of the Matheson-certified 3 and 52 ppmv styrene calibration gases, on July 29 and 30. four
adsorption tube samples each were taken from each calibration gas cylinder (using Method 18, the
adsorption tube procedure). Two adsorption tube samples of each styrene calibration standard were taken
inside (at an average laboratory temperature of 22°C) and two adsorption tube samples were taken outside
with the calibration gas bottles in the direct sun (an an average temperature of 38°C). The reason samples
were taken at laboratory conditions and at conditions that mimicked ambient field temperatures experienced
at the Eljer facility was to determine if styrene gas was condensed within the sampling apparatus at room
temperature - a possible explanation for the apparent low recovery based on Matheson's certified values.
The adsorption tubes (from the same lot used at Eljer SKC, Inc. catalog # 226-01, coconut charcoal, Lot
120) were analyzed by removing the charcoal from the tubes and desorbing the styrene into high-purity
carbon disulfide. As part of the analytical procedure, the desorption efficiency of styrene from this lot of
coconut charcoal is separately determined each time a sample or set of samples is analyzed. The
desorption efficiency was determined to be 90.25%, equal to the value that has been determined in the past.
The results of these analyses, carried out during the first week of August, was that no difference could be
detected between samples obtained inside or outside the laboratory and that the Matheson-certified 3 ppmv
styrene gas was 2.69 ppmv with an RSD of 3.55% while the Matheson-certified 52 ppmv styrene gas was
39.1 ppmv with an RSD of 0.55%. No error was found in the analytical procedures followed in these
analyses, in the preparation of the two sets of calibration standards, or in the behavior or operation of the
GC-FID used for these analyses.
Next, a performance evaluation audit standard was requested from EPA to determine with certainty
if the error was due to our analytical procedures, the cylinder was sent to Southern on September 17 and
the results of Southern's triplicate analysis (using the Method 18, Adsorption Tube Procedure) of the styrene
content in the cylinder was reported to the EPA on September 21. After it was determined that Southern's
43
-------
analysis was within 96.6% of the actual styrene concentration of 58.6 ppmv (with an accuracy of ± 2.2%), it
was concluded that the concentrations reported on Matheson's analysis of all the gas samples provided for
this test were in error. The results of the tests of the EPA performance evaluation audit sample are shown in
Table A-1. Table A-2 shows the results of tests performed to determine the actual styrene content of these
gases. No other corrective actions were required or taken during the collection of samples and data or
during subsequent analysis of samples collected during testing.
The values reported in Table A-2 were obtained by two separate methods. First, as part of the
investigation discussed above, EPA Method 18, Adsorption Tube Procedure (equivalent to NIOSH Method
1501), was used to make triplicate determinations of the styrene content of each of the nominal 3, 52, 54,
and 195 ppm styrene calibration gases. All of these determinations were completed by September 14.
Second, on September 29, a JUM VE-7 THC analyzer (one of the THC analyzers used in the sampling van)
was allowed to stabilize for 24 hours on filtered ambient laboratory air and was then spanned with 10.7 ppm
± 1% propane (unfortunately, other propane standards were not available when these measurements were
performed) and zeroed with a THC-free zero air standard (s, 0.1 ppm of hydrocarbon compounds). The THC
analyzer was then used to sample the 58.6 ppm EPA audit standard, as well as the nominal 3, 54, and 195
ppm styrene calibration gases (at this time the cylinder containing the 52 ppm calibration gas had been
exhausted). Styrene content was determined based on the response of each of the calibration gases to the
value measured for the EPA audit standard. Zero and span checks performed at the beginning, middle and
at the end of the THC measurements confirmed instrumental stability.
These results required that, at best, all of the data be scaled to reflect the true concentrations of
styrene present in the gas cylinders obtained from Matheson Gas Products that were used for field
calibrations. At worst, the data could be completely compromised because the styrene within the cylinders
supplied by Matheson could have been slowly polymerizing since the cylinders were prepared and the
styrene concentrations measured after the test would not represent styrene concentrations
44
-------
Table A-1. Results of Analyses of EPA Performance Evaluation Audit Sample*
Sample"
1.
2.
3.
Average ± RSD
Measured (ppm)
55.6
57.0
57.1
56.6 ±1.5%
Actual (ppm)
58.6
58.6
58.6
58.6 ± 2.2%
Rel % Difference
-5.1
-2.7
-2.6
-3.4
* Cylinder CLM 008308. Specified as containing styrene at a concentration under 100
ppm with the balance gas being nitrogen. Content later quoted by EPA to be 58.6 ppm
± 2.2% RSD.
" Analysis by EPA Method 18, Absorption Tube Procedure, with GC/FID. Aside from the
diluent (CS2), styrene was the only material detected.
Table A-2. Results of Analyses of Matheson Calibration Gas
Matheson Analysis'
(ppm)
3
52
54
195
Method 18 Analysis"
(ppm)
2.69
39.1
37.8
176.8
RSD (%)
3.55
0.55
2.07
4.06
THC Analysis1
(ppm)
2.16
N/An
39.45
170.8
RSD (%)
0.25
-.
0.12
0.18
Comparability
(% Diff.)
21.9
..
-4.3
3.5
* As indicated on gas cylinder, ppm styrene in nitrogen.
** Absorption Tube Procedure using charcoal tubes.
1 THC calibrated with 10.7 ppm propane in nitrogen. Response referred to styrene by analysis
of EPA performance evaluation audit sample (58.6 ppm styrene measured 151.06 ± 0.24
ppm with propane-based calibration).
n Cylinder exhausted before THC measurements could be made.
present in the cylinders at the time of the test. The latter eventuality was explored with Matheson in the initial
conversations that were directed toward determining the source of the disagreement. As indicated above.
Matheson Gas Products asserted that the cylinders were properly prepared and passivated. While
Matheson was unable to explain why the concentrations were so far from those determined by their original
in-house analysis, they did maintain that if the temperature indicating strips on the sides of the cylinders had
45
-------
not changed color (indicating exposure to temperatures that could degrade the sample), styrene
concentrations within the cylinders should have remained stable through the time period of the test and our
subsequent determination of the actual styrene concentrations within the cylinders. Because none of the
temperature indicating strips on the sides of the cylinders had changed color (indicating the temperature of
the cylinder had reached or exceeded 125°F), we proceeded to correct the data assuming that styrene
concentrations in the calibration cylinders measured after the test were representative of styrene
concentrations present during testing.
Both THC data and EPA Method 18 data required correction. Correction of the THC data was
relatively straightforward. Lab THC data was taken on THC analyzers calibrated with propane (850 ppm, ±
1%). During the test, the nominal 54 ppm styrene calibration gas (actually 39.45 ppm by later THC analysis)
was observed to produce an equivalent propane response of 115 ppm on THC#1 and 116 ppm on THC #2.
Thus, these data were scaled by the ratio of 0.3430 (or 39.45/115) for THC #1 and 0.3401 (or 39.45/116) for
THC #2. Both THC analyzers in the sampling van were calibrated with the same nominal 52 ppm styrene
calibration gas (actually 39.1 ppm by Method 18 analysis). Because this cylinder was emptied before the
THC measurements reported in Table A-2 could be made, results obtained with these analyzers were scaled
by a ratio of 0.7519 (or 39.1/52).
The EPA Method 18 data required more extensive correction. The concentration of a VOC as
measured by a gas chromatograph is determined by first measuring the instrumental response (peak area)
to a set of calibration gases and then relating the peak area measured for each calibration gas to the known
concentration of that gas. Because a FID produces a linear response to concentration, a linear fit is usually
made to the peak area (x) versus concentration calibration (y) data.
In the field, after measuring the peak area response to each calibration gas (nominal 3, 54, and 195
ppm styrene), linear fits were made to relate peak area to styrene content for subsequent determinations.
Thus, to correct the GC data obtained on this test, each linear (calibration) fit obtained in the field was
recalculated based on what the actual styrene concentration of each calibration gas was later determined to
be and the styrene concentrations obtained during the time that calibration was used were corrected with the
new linear fits. For these corrections, the THC analyzer-determined values for the calibration gases were
46
-------
used because during those measurements repeated, direct comparisons were made with the EPA audit
standard. Table A-3 presents the calibration data determined for each day of testing (on site).
Table A-3. GC-FID Calibrations Performed at Eljer Plumbingware*
Nominal Styrene
Concentration
(ppm)
3
54
195
Actual Styrene
Concentration
(ppm)
2.16
39.45
170.8
GC-FID
Peak Area
6/15/93
(0935-1151)
4994
118910
343248
GC-FID
Peak Area
6/16/93
(0819-0902)
3903
100744
364290
GC-FID
Peak Area
6/17/93
(1052-1145)
6986
273351"
968576
GC-FID
Peak Area
6/21/93
(before 0800)
9428
201447
780310
Performed by DEECO, Inc..
Reanalysis at 1346 gave peak area of 176150
During the test, full instrumental calibrations were performed early each morning and afterwards,
occasional checks were performed with the nominal 54 ppm calibration standard to assure that instrument
response had not drifted. Because of the limited time available for sampling (due to the plant operating
schedule) only one full calibration was performed each day.
These corrections are not completely satisfactory. In particular, the peak area measured for the
195 ppm styrene calibration gas (actually 170.8 ppm by subsequent THC analysis), appears to be low as
compared to the other two calibration gases. It could be that this relatively high concentration of styrene did
change between the time the test was performed and the time that the concentration of this calibration gas
was measured in the laboratory. Another likely possibility is that the Teflon sampling line that was used to
convey this gas to the GC was not kept warm enough (the lab trailer was air conditioned and maintained at
approximately 75° F) and some of the styrene condensed in the line before it could reach the inlet of the GC.
Before the test this concentration of styrene had been observed to condense in Teflon sampling lines under
similar laboratory conditions. Because of this discrepancy, and because few of the styrene concentrations
measured were greater than 100 ppm (corrected), it was decided to eliminate the 195 ppm (actual 170.8
ppm) point from the set of calibration data from which new fits were determined to correct the original data.
47
-------
One other significant problem was encountered in correcting the GC data. On June 17, at the
morning calibration, GC response to styrene was observed to have increased dramatically as compared to
the response measured during earlier calibrations. Afterwards, instrumental response was monitored closery
while two samples (one from a Tedlar bag and another from a Teflon bag) were repeatedly analyzed.
Approximately one hour after that calibration, the GC was observed to have stabilized at a lower response
level and a single point calibration was performed with the 54 ppm (39.45 ppm actual) so that samples that
had been accumulating could be analyzed. Thus, we were forced to rely on a single point calibration for
Method 18 data acquired on June 17.
DATA QUALITY
The following procedures were used to determine how well data DQI goals were met:
• Precision is expressed as percent coefficient of variation:
%CV = 100x(S,/XavB)
where S, is the standard deviation of x number of data values from the data set and Xayg is the mean or
average of the x number of data values from the data set.
48
-------
Bigs is expressed as a difference or percent difference between measured and known values:
Bias = (X-T)
%RPD = 100
where T is the true value (reference standard) and X is the mean sample concentration. %RPD is the
relative percent difference.
* Completeness is expressed as a percent between successful analyses and total attempts:
Completeness = 100 S/A .
where S is the number of successful analyses and A is the total number of attempts.
• Comparability is expressed as a percent difference (%Diff) between the results for two methods:
%Diff = 100 (R,-
where R, is the result for one method and R2 is the result for the second method.
Table A-4 shows the DQI goals that were estimated for critical measurements in the QAPP, and
Table A-5 presents the DQI values for measurements carried out with EPA Reference Method 18 and the
THC analyzers in the sampling van and the mobile lab. Below, the precision, accuracy, and completeness of
the data that were obtained in this project are reviewed.
49
-------
Precision
Precision, as reported in Table A-5 (a), meets the DQI goals shown in Table A-4 for Method 18 with
the exception of the 170.8 ppm calibration point which was expected. Mainly, this is because field
calibrations were compromised by incorrect concentrations for the calibration gases. Using the procedure
described above, the values reported in Table A-4 (a) were obtained by generating new fits to the original
calibration data (excluding the 170,8 ppm point), as taken for each day of testing and then determining the
difference between the correct styrene concentration (from THC remeasurement) and styrene
concentrations determined from best linear fit to the calibration data.
Table A-4. Data Quality Indicator Goals for Critical Measurements Estimated in QAPP
Method and
Reference
EPA
Method 18,
Section 7.1
and
Section 7.3
Total
Hydrocarbon
Analyzer with
FID1.
Measurement
Parameter
VOC
Speciation and
Concentration
Hydrocarbon
compounds
in air.
Experimental
Condition
1. Spray booth.
Vent and air
exhausts.
2. Headspace
measurements
from styrene
storage tanks."
1. Spray booth
Vent and air
exhausts.
2. Headspace
measurements
from styrene
storage tanks,"
Expected
Precision
(RSD,%)
5.0"
±10fT
Expected
Accuracy
(% Bias)
10'
*5n
Completeness
(%)
90
90
Expected precision and bias for GC-FID analysis of samples obtained using EPA Method 18. Precision
and bias will be determined for the GC-FID used (Shimadzu GC-14A),
Headspace measurements eliminated from testing by Project Officer.
T J.U.M. Model VE-7 THC Analyzer with FID.
n Estimated values.
50
-------
Table A-5. Data Quality Indicator Values for Method 13 and THC Measurements Made at Eljer Plumbingware
(a) Method 18 Measurements In Mobile Lab
Gal Gas, % Bias
6/15/93
6/16/93
6/17/93
6/21/93
Average
Precision (% CV)
2.16 ppm Styrene
Cai Gas
(Method 18 Value)
2.23
2.23
2.19
2.20
2.21
0.9
Bias
%
3.2
3.2
1.4
1.9
2.4
39.4 ppm Styrene
Cal Gas
(Method 18 Value)
39.4
39.4
39.4
39.4
39.4
0.0
Bias
%
0
0
0
0
0
170.8 ppm Styrene
Cal Gas
(Method 18 Value)
112.6
140.5
136.5
152.6
135.3
12.1
Bias
%
-34
-18
-20
•11
-21
(b) THC Measurements Made in Sampling Van, THC'#1 and THC #2
THC#1
Cal Gas/% Bias
6/15/93
6/16/93
6/17/93
Average
Precision (% CV)
THC #2
Cal Gas/% Bias
6/15/93
6/16/93
6/17/93
Average
Precision (% CV)
2. 16 ppm Styrene
Cal Gas
(THC Value)
3.03
2.69
2.51
2.7
9.6
2. 16 ppm Styrene
Cal Gas
(THC Value)
2.82
2.60
2,58
2.67
5.0
Bias
%
40
25
16
27
Bias
%
31
20
19
24
170.8 ppm Styrene
Cal Gas
(THC Value)
178.2
—
_
170.8 ppm Styrene
Cal Gas
{THC Value)
174.9
-
-
Bias
%
4
-
Bias
%
. Si
2
'
-
(c)THC Measurements Made in Mobile Lab. THC #1 and THC #2
Date
6/15/93
6/16/63
6/17/93
6/21/93
Average
Precision (% CV)
THC#1
850 ppm Propane
Cal Gas
848
856
851
850
859
850
852
0.5
Bias
%
-0.2
. 0.7
0.1
0.0
1.1
0.00
0.3
THC #2
850 ppm Propane
Cal Gas
860
856
860
856
856
848
848
840
853
0.8
Bias
%
1.2
0.7
1.2
0.7
0.7
-0.2
-0.2
-1.2
0.4
51
-------
As with the Method 18 measurements, for the THC analyzers in the sampling van precision was
determined by repetitive sampling of calibration gases. Because of suspected problems with line losses the
170.8 ppm calibration gas was only sampled once. Precision was determined by repetitive sampling of the
2.16 ppm styrene calibration gas. The results of these determinations, shown in Table A-5(b), are within the
DQI goals shown in Table A-5.
With respect to the determination of precision for the THC analyzers in the mobile laboratory, these
analyzers were calibrated with 850 ppm propane. Precision was determined by repetitive sampling of the
850 ppm propane calibration gas. The results of these determinations, shown in Table A-5(c), are well within
the DQI goals shown in Table A-4. The relative response of each THC to styrene was determined on June -
i.
17. After both analyzers had been calibrated with the propane standard, nominal 54 ppm (actual 39.45 ppm)
styrene calibration gas was fed to both THC analyzers. THC #1 read 115 ppm and THC #2 indicated 116
ppm.
Bias
For Method 18 measurements, bias was much higher than the DQI goaf of 10% shown in Table A-
3. The 2.16 ppm data may be off because the styrene concentration is very low and a small difference in the
value as determined by the THC can result in a large percentage bias (a 5 ppm concentration of styrene in
nitrogen was originally ordered). In retrospect, it would have been better to use a higher concentration, in the
range of 15 ppm. The 170.8 ppm data are also low, by about the same percentage. However, in this case,
line losses could be partly at fault because some condensation of styrene within a Teflon sample line had
been observed in the past with this gas. In the air conditioned mobile lab, an unheated 2 m Teflon sample
line was used.
For THC analyzer measurements in the sample van, bias was determined for each measurement
of the 2.16 ppm span gases and is reported in Table A-5(b). At this concentration, average bias values for
both analyzers were much greater than the target DQI of ± 5%. Partly, this is because a low value of styrene
was sampled while the THC was calibrated with (what was thought to be) 52 ppm calibration gas. In
retrospect, it would have been better to use a higher concentration, either lower than the calibration standard
(in the range of 15 ppm) or higher than the calibration standard (in the range of 100 ppm). The single
52
-------
sample of 170.8 ppm styrene calibration gas did show a much tower bias for both THC analyzers, below the
expected DQI of 5%.
For THC analyzer measurements in the mobile lab, bias was determined for each measurement of
the 850 ppm propane span gas and is reported in Table A-5{c). As with the values measured for precision,
bias was very low. averaging less than 0.5% for both THC analyzers,
For Method 18 samples, completeness was 100%. Some data were not usable because the molds
that were sprayed during sampling runs when the video cameras malfunctioned could not be used to
determine emissions per area of mold sprayed, but this was not related to the sampling effort.
For THC analyzer measurements made in the sampling van, completeness was 50% (for active
spray booths). This is much lower than the OQI goal of 90%, but, data from these THC analyzers was only
intended to back up data taken by the other set of THC analyzers in the mobile laboratory. For the THC
analyzers in the mobile laboratory, a completeness of 91% was achieved (for active spray booths),
essentially equal to the OQI goal of 90%. Together, both THC analyzers were able to obtain corroborative
data for every Method 18 sample. With respect to THC samples taken to characterize styrene emissions not
captured by active spray booths, 100% of the data needed to quantify non-spraying emissions was obtained
by the THC analyzers in the mobile laboratory. It was not intended to utilize the THC analyzers in the
sampling van as part of this effort although these instruments were able to contribute some of the emissions
data measured during work day breaks at the active spray booths.
The physical layout of sampling locations at the Sjer facility dictated much of the sampling strategy
and sampling methodology practiced during this evaluation to obtain representative samples. The use of
heated sampling lines avoided condensation of styrene and the use of heated Tediar bags also prevented
condensation of styrene while the samples were being taken and subsequently analyzed. Location of the
sampling line inlets near the center of exit vents assured that samples were representative. Following the
53
-------
sample methodology recommended in EPA Method 18 also assured that representative samples were
obtained.
Comparability
The sampling plan for this project made provision for simultaneous sampling using two
measurement methods that should allow comparison of the results when suitably averaged over the same
sampling period. In general, for THC analyzer-based measurements, styrene emissions from the lay-up
booths and from the gel coat booths were much greater than those determined from Method 18
measurements. With the exception of the June 17 data, both methods measured essentially the same
emissions from the single back-up booth. However, these differences do not suggest that the
measurements are not comparable. Because standard deviations are relatively large for the lay-up and gel
coat booths, the data do overlap.
54
-------
TECHNICAL REPORT DATA
(Please read liuouctiant on the reverse before completing
1. REPORT NO.
EPA-600/R-96-138
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Evaluation of Styrene Emissions from a Shower Stall/
Bathtub Manufacturing Facility
5. REPORT DATE
November 1996
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Larry Felix, Randy Merritt,
and Ashley Williamson
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME ANO ADDRESS
Southern Research Institute
P.O. Box 55305
Birmingham. Alabama 35355-5305
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-D2-0062, Task 12/1
12. SPONSORING AGENCY NAME ANO ADDRESS
EPA, Office of Research and Development
Air Pollution Prevention and Control Division*
Research Triangle Park, NC 27711
I. TYPE OF REPORT ANO PERIOD O
Task Final; 11/93-4/94
D COVERED
14. SPONSORING AGENCY CODE
EPA/600/13
is. SUPPLEMENTARY NOTESAPPCD prOject officer Bobby E. Daniel is no longer with the
Agency. For details, contact Julian W. Jones, Mail Drop 61, 919/541-2489. (*) Prior
to April 1, 1995, the Division was the Air and Energy Engineering Research Lab.
16. ABSTRACT
The report gives results of emissions measurements, carried out at a rep-
resentative facility (Eljer Plumbingware in Wilson, NC) that manufactures polyesten
resin-reinforced shower stalls and bathtubs by spraying styrene-based resins onto
molds in vented, open, spray booths. Styrene emissions were characterized for the
three stages of manufacture by measuring styrene concentrations at the vents of
spray booths used in each part of the process. In addition, styrene concentrations
were measured at each ventilation fan exhaust. Emission levels were determined
using EPA Method 18 to obtain integrated emissions samples and total hydrocarbon
(THC) analyzers to measure continuous emissions levels during the EPA Method 18
sampling. Analysis of the EPA Reference Method data indicates that: (1) styrene
monomer is the only volatile organic compound released in the process; (2) overall,
approximately 4% of all material sprayed is lost to atmospheric emissions as sty-
rene (approximately 19% of all styrene sprayed); and (3) emissions vary for each
phase of manufacture, with post-spraying emissions of styrene (from curing molds)
constituting a large part, approximately 29% of all emissions. (NOTE: Current EPA
emission factors—AP-42—for styrene emissions from the production of polyester-
resin-reinforced plastics represent a composite of spraying/post- spraying emissions
17.
KEY WORDS ANO DOCUMENT ANALYSIS
DESCRIPTORS
b.lOENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Pollution
Styrene
Styrene Resins
Emission
Manufacturing
Hydrocarbons
Pollution Control
Stationary Sources
Shower Stalls
Bathtubs
13B
07C
1U.11J
14G
05C
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS {This Report)
Unclassified
21. NO. OF PAGES
61 .
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
55
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
National Risk Management Research Laboratory
Technology Transfer and Support Division
Cincinnati, Ohio 45268
OFFICIAL BUSINESS
PENALTY FOR PRIVATE USE. S3OO
AN EQUAL OPPORTUNITY EMPLOYER
// Your address is incorrect, please change on the above label
tear off; and return to the above address.
If you do not desire to continue receiving these technical
reports. CHECK HERE D; tear off label, and return it to the
above address.
Publication No. EPA-GOO/R-96-138
------- |