United States
Environmental Protection
Agency
Research and Development
Kl''
Atmospheric Research and Exposure O \\ „
Assessment Laboratory "^-4
Research Triangle Park NC 27711 , *
EPA/600/S3-89/004 July 1989 '
Project Summary
Sampling and Analysis of
Butadiene at a Synthetic
Rubber Plant
J. D. Goodrich. W. G. DeWees, and R. R. Segall
Butadiene emission samples were
collected from the process vent
stream of a plant manufacturing
synthetic rubber from styrene and
butadiene. Samples were collected
by modification of the evacuated
container sampling procedure,
outlined In Section 7.1.1 of EPA
Method 18. On-site analysis of
samples was performed using a gas
chromatograph equipped with a
flame ionizatlon detector. The
precision of butadiene concen-
trations determined from simultan-
eous samples collected at a nominal
sampling rate of 0.050 L/mln, rather
than at the recommended sampling
rate of 0.5 L/min, was determined. In
addition, simultaneous samples were
collected at both 0.20 L/min and 0.050
L/min and analyzed to determine if
the mean values or precisions of the
measured concentrations were
influenced by the sampling rate.
Acceptable precision was observed
upon analysis of emission samples
collected at both sampling rates, and
the mean values and precisions of
butadiene levels determined were
statistically equal for simultaneous
samples collected using the two
sampling rates.
This Project Summary was devel-
oped by EPA's Atmospheric Research
and Exposure Assessment Laboratory,
Research Triangle Park, NC, to
announce key findings of the research
project that Is fully documented In a
separate report of the same title (see
Project Report ordering information at
back).
Procedure
The principles of sampling and analysis
followed the general procedures
described in EPA Method 18, with the
exception that emission samples were
collected at flow rates of 0.050 L/min and
0.20 L/min. rather than 0.5 L/min.
The precision of the sampling and
analysis techniques was determined
using quad-trains. During the field test,
gases were withdrawn from a duct
carrying process emissions in route to an
open flare, and all quad-trains drew
gases from a common manifold. A
schematic of the sampling apparatus is
included as Figure 1. Samples were
collected in quadruplicate in the 5 L
Tedlar bags. In addition, during two of the
twelve sampling runs, samples were
collected using dual quad-trains. One
train utilized 5 L bags and the other train
utilized 20 L bags. All bags were
analyzed on-site, as was an audit sample.
The results from analysis of the four bags
were averaged and a standard deviation,
variance, and 95% confidence interval for
each sampling run were calculated.
One small bag sample from nine of the
first ten sampling runs was returned to
the base laboratory for reanalysis by
Entropy. All bags collected during the
final two sampling runs, except one
sample from the twelfth run that leaked,
were also retained. A full complement of
four large and four small bags was
available as the eleventh sampling run
-------
XAD-2
Sorbent
Tube
To
Ambient
3/8" Teflon Tubing
(Exhaust) to Ambient
Accesses
for Large
Bag Samplers
When Used
Bulkhead
Quick
Connects
PVC
Bag
Samplers
M '
T
Bag
A
I
[ (
•
Bag
8
Critical
Orifices
Activated
Charcoal
Filter
Ambient
Figure 1. Sample collection apparatus.
involved sample collection using the dual
quad-train setup. These samples were
analyzed within one week of completion
of the field test. Following reanalysis at
Entropy's base laboratory, selected bag
samples were made available to EMSL-
RTP, Quality Assurance Division (QAD)
personnel for independent verification of
butadiene concentrations measured.
Analyzed cylinders containing approxi-
mately 10, 150, and 2000 ppmv of
gaseous butadiene in nitrogen were
purchased from a supplier of high purity
gases and used to calibrate the gas
chromatograph. The cylinders were
analyzed after final blending by the
supplier, who certified the butadiene
concentrations to ± 2%.
Instrument response was obtained from
a least squares fit of the calibration data.
An audit of these results was conducted
daily by assaying the concentration of a
blind audit sample provided under the
EPA's performance audit program. The
audit sample contained butadiene diluted
with nitrogen gas at a concentration in the
ppmv range.
The accuracy of the butadiene deter-
minations was examined by the on-site
analysis of a blind audit sample provided
by Research Triangle Institute, Research
Triangle Park, North Carolina under the
EPA's performance audit program, which
utilizes cylinder gases containing volatile
organics in the ppmv range. The audit
cylinders available were known to contain
butadiene in the 5-60 ppmv range. Fiel
and laboratory analyses of the cylindi
provided a quality assurance check of th
three-point calibration curves coverin
the approximate range 10 to 2000 ppmv.
Several other organic components i
addition to butadiene were present i
samples collected from the plant'
process vent. In particular, seven
butane/butene isomers were observe*
Because the chromatographic characte
istics of some of these are quite similar 1
butadiene, a chromatographic systei
capable of high resolving power wj
necessary. During the field test, samp
analysis involved a chromatograph
system utilizing two packed GC columr
connected in series as part of a bacl
-------
flush-to-vent configuration. The first
:olumn, a 10' x 1/8" stainless steel
column containing 20% SP-2100/0.1%
Carbowax 1500 on 100/120 mesh
Supelcoport, effected the separation of
butadiene and other permanent gases
from the higher boiling components
within approximately three minutes. In
that time these gases had passed into the
second column, a 2 m x 1/8" stainless
steel column containing 0.19% picric acid
on 80/100 mesh Carbopack C. The
sampling valve was returned to the
load/isolate position at this point to direct
a separate carrier flow through the
second column allowing for the
separation of butadiene from other
permanent gases present, while the
original carrier flow was simultaneously
reversed through the first column to
backflush higher boiling sample
components to a vent. The separation
was performed isothermally at 60°C, and
both carrier flows were approximately 20
mL/min in the forward direction.
Results and Discussion
The range and sensitivity were
determined for the Tedlar bag sampling
technique prior to conducting the field
test. An estimation of the limit of
detection and quantifiable limit was
carried out. Using a sample size of 1 mL,
these values were determined to be 0.43
ppmv and 2.0 ppmv, respectively.
The upper limit of detection, where the
GC column becomes saturated and/or the
detector response becomes non-linear,
was also estimated. The peak shapes
observed for prepared samples that
contained butadiene at levels of
approximately one percent indicated that
the column was saturated. The retention
time observed for these prepared
samples was also significantly shortened.
The shift in retention time was potentially
great enough to prevent the complete
separation of butadiene from other
components observed in samples
collected during the field tests.
The results from analysis of the four
bags composing each quad-train were
averaged and a standard deviation,
variance, and 95% confidence interval
were calculated. This information is
summarized in Table 1.
Two samples could be discarded as
statistical outliers at a confidence level of
95%. Two estimates for measures of the
precision of the method are given, one in
which the outliers are included and one in
which they are not.
The average coefficient of variation
observed for all the small bag samples
Table 1. Summary of Analytical Results and Precision of Data
Date
4/28
4/28
4/28
4/28
4/28
4/28
4/28
4/29
4/29
4/29
4/30
4/30
4/30
4/30
Run ID
51
52@
53
L54-
54
55
56@
61
62
63
77
L72
72
73
Mean'
322
24
30
444
459
171
546
426
29
114
149
14
13
3480
Std. Dev.
7.4
5.7
3.7
47
39
17
150
76
3.0
11
6.6
0.6
0.4
40
Coef. of
Var. (%)
2.3
24
7.8
11
8.5
9.8
27
18
10
10
4.5
5
3
1.2
Variance
55
32
61
2200
1500
279
21000
5700
9.0
120
44
0.3
0.2
1800
"Butadiene concentrations in ppmr
~ One bag sample collected during this run was found to leak; the results from
analysis of this bag are not included in any statistical evaluation of the data.
@The results summarized below are obtained for sample runs 52 and 56 when
statistical outliers are excluded.
Date
Run ID
Mean"
Std. Dev.
Coef. of
Var. (%;
Variance
4128
4/28
52
56
27
473
1.4
15
5.1
3.1
2
217
was approximately 11%, but dropped to
6.9% when the two statistical outliers
were eliminated from their respective
sample sets and the means and standard
deviations of those samples recalculated.
The average coefficient of variation
observed for the two sampling runs
utilizing the large Tedlar bags was
approximately 8%; the analysis of one
large bag sample collected as part of
these runs was discarded because the
bag did not pass a post-analysis leak
check.
During one dual quad-train sampling
run, the butadiene level determined using
samples collected in the small Tedlar
bags at an approximate flow rate of 0.050
L/min was 459 ppmv and the standard
deviation observed among the four
replicates was 39 ppmv. The level
determined using the large bag samples
collected at a flow rate of about 0.20
L/min was 444 ppmv and the observed
standard deviation was 47 ppmv. For the
other dual quad-train run, the level
measured using the small bags was 13
ppmv, with a standard deviation of 0.6
ppmv, while the respective data collected
using the large bags were 13 ppmv and
0.4 ppmv. A variance-ratio test, or F-test,
showed that standard deviations obtained
using the different bag sizes and flow
rates were not significantly different at
the 95% confidence level for either of the
two sampling runs. This fact, and the
similarity of the average coefficients of
variation calculated after analysis of
samples collected using both sampling
rates, suggests that the precisions for the
two sampling protocols are invariant.
Likewise, when the means of butadiene
levels determined using the small and
large Tedlar bags were compared for the
two sampling runs using a Student's t
test and the null hypothesis, they were
found to be equivalent at the 95%
confidence level.
Selected samples were returned from
the field for reanalysis. Because the
reanalysis results were in overall
agreement with the results obtained
during the original on-site analyses, the
two sets of measurements were used to
estimate within- and between-laboratory
precision. The average coefficient of
variation for the Entropy's field and
laboratory analyses of samples collected
in small bags was 9.5%; the average
coefficient of variation was 13% when the
precisions of both small and large bags
are included. The average coefficient of
variation for the laboratory analysis of
-------
selected bag samples by Entropy and
EMSL, QAD was 4.9%.
Several analyses of the audit cylinder
were conducted while analyzing samples
collected during the field test. The
average value obtained for the
concentration of the audit cylinder was
approximately 6% below the certified
concentration.
Although the retention of butadiene by
Tedlar bags was not rigorously evaluated,
experiments were conducted to
demonstrate that bags containing field
samples could be reused after they had
been reblanked. Bags were emptied of
sample by complete evacuation and
flushed twice with compressed air before
being filled to half their capacities. After
allowing the bags to stand overnight,
analysis of the contained gas showed
them to be essentially free of butadiene.
Conclusions and
Recommendations
Four principal conclusions were drawn
from the data collected and from the
observations made during the field test.
The sampling and analytical protocol
used was fully validated, including the
use of the lowered sampling rate utilized
for collection of emission samples in
small Tedlar bags. The physical and
chemical composition of the process
effluent encountered in this investigation
leads to two important conclusions/
recommendations concerning sampling
methodology and to two important
conclusions/ recommendations concern-
ing sample analysis.
1. Butadiene levels determined in
samples collected in 5 L Tedlar bags
at an approximate flow rate of 0.050
L/min and in samples collected in
larger, 20 L bags at an approximate
flow rate of 0.20 L/min were were not
significantly different at the 95%
confidence level. Because a sampling
rate of 0.5 L/min yields a more
convenient sample size than that
obtained using the sampling rate
recommended by EPA Method 18
and the two differing sampling rates
utilized for collection of bag samples
during this study did not affect the
values or precisions of the
determinations made, the lowered
sampling rate is recommended for
collection of bag samples.
2. The presence of aspirated organic
droplets in the process vent stream
made the precise quantitation of
gaseous butadiene levels very
difficult. The precision among quad-
train samples was effectively
improved by using of filtering media
designed to eliminate such liquid
while passing butadiene gas. The
average coefficient of variation
observed for the 0.050 mL/min
sampling rate was approximately 7%.
The elimination of similar liquids from
other process effluents is highly
desirable during collection of bag
samples. Further investigation of
filtering media designed to
accomplish this while passing the
sample component of interest is
recommended.
3. The presence of several butan
butene isomers in the sample strea
necessitated the use of
chromatography column capable
separating butadiene from the:
potential interferants. A 2 m x 1/1
stainless steel column containir
0.19% picric acid on 80/100 mes
Carbopack C was utilized during th
investigation. This column, or anotlr
column similarly able to resolv
butadiene from these potential
interfering hydrocarbons
4. The presence of high boiling samp
components in the collected emissic
samples made use of a chromati
graphic system utilizing a two columi
backflush-to-vent configuration high
desirable. The forward column, a 10'
1/8" stainless steel column containir
20% SP-2100/0.1% Carbowax 15C
on 100/120 mesh Supelcopor
effected the separation of butadier
and other permanent gases froi
higher boiling sample component
before completion of the separatic
on the picric acid column. Sampl
analysis was complete in under 1
minutes, far more rapidly than i
possible with a conventional singl
column system. The use of a simil;
system is recommended to speed tr
analysis of butadiene samples whei
ever significant quantities of high<
boiling organics are present.
-------
L D. Goodrich, W. G. DeWees, and R. R. Sega// are with Entropy
Environmentalists, Inc., Research Triangle Park, NC 27711.
Jimmy C. Pau is the EPA Project Officer (see below).
The complete report, entitled ".Sampling and Analysis of Butadiene at a Synthetic
Rubber Plant," (Order No. PB 89-151 534/AS; Cost: $15.95. subject to change)
will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield. VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S3-89/004
u
0000329 PS
60604
------- |