United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
Research Triangle Park NC 27711
Research and Development
EPA/600/S4-87/034 Jan. 1988
Project Summary
Benzene Continuous Emission
Monitoring Systems for
Gasoline Bulk Storage
Jon N. Bolstad
The full report summarizes a study
of continuous emission monitors for
measuring benzene emissions from
gasoline bulk storage terminals. The
work was performed for the Quality
Assurance Division, Source Branch, of
the U.S. Environmental Protection
Agency (EPA) by Pacific Environmental
Services under Contract No. 68-02-
3997, Task 33. The EPA is considering
regulating the emissions of benzene
from bulk storage of gasoline products
and is likely to require data regarding
benzene emissions. The evaluation of
continuous emission monitors for
benzene is a necessary part of the
overall effort.
This study was performed in three
phases: a literature review, a laboratory
evaluation, and a field evaluation. The
study objectives were to determine the
commercial availability of continuous
monitors for benzene and the reported
performance of these units, evaluate
their suitability for measuring benzene
in gasoline vapors, and determine their
applicability to the source category.
There are no benzene-specific con-
tinuous emission analyzers commer-
cially available. If a benzene-specific
system is required, and semicontinuous
(i.e., gas chromatographic analysis) is
not acceptable for regulatory purposes,
nondispersive ultraviolet analysis holds
promise for accomplishing this task.
The nondispersive ultraviolet technique
can be used to measure benzene in a
matrix of other hydrocarbons, and
nondispersive ultraviolet continuous
systems do exist, so combining the
laboratory experience with the contin-
uous monitor experience into a proto-
type benzene continuous monitor
appears to be a promising (and cost-
effective) means to obtain such an
instrument.
This Project Summary was devel-
oped by EPA's Environmental Monitor-
ing Systems 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).
Introduction
The U.S. Environmental Protection
Agency (EPA) is considering regulating
benzene emissions from the gasoline
bulk storage industry. A means of
measuring benzene emissions may be
required if the regulation contains
specific emission limits rather than
specific equipment or operating prac-
tices. This study was undertaken to
evaluate the current availability and
performance of continuous emission
monitors (CEMs) for benzene in the event
that continuous benzene monitoring was
deemed necessary in the regulation. This
investigation focused on the availability
and performance of lexisting commercial
instruments, but the^findings are
intended to serve as the basis for
developing a test method suitable for
Federal Register promulgation.
The study was performed in three
phases: a literature review, a laboratory
evaluation, and a field evaluation. The
objectives were to determine the com-
mercial availability of continuous mon-
itors for benzene and the reported
performance of these units, evaluate
their suitability for measuring benzene
-------
in gasoline vapors, and determine their
applicability to the source catgeory.
Experimental Procedure
Literature Review
The technical literature was reviewed
to determine potential analytical
procedures for measuring benzene and
to establish the expected range of total
hydrocarbon emissions and the benzene
fraction of the total emissions from the
gasoline bulk storage industry.
Manufacturers of chemical analysis
instrumentation were surveyed both by
direct contact and through the literature
to ascertain the types of analyzers being
produced; their suitability for use in an
explosion-prone environment; their
range and sensitivity; their utility
requirements; and other considerations
germane to the use of each instrument
as a bulk storage terminal continuous
emission monitor. Other instruments
were also tested as potential process
monitoring devices or to measure a
benzene surrogate.
Laboratory Testing
A test plan was designed, and selected
instruments were subjected to quality
assurance (QA) testing in the controlled
environment of the laboratory. Certified
calibration gases of benzene/air mix-
tures were used to define the repeata-
bility of the test instruments, as well as
detection limits, response times, and drift
characteristics. In addition to benzene/
air mixtures, a cylinder gas mixture
simulating gasoline vapor was used to
test the ability of these instruments to
detect benzene without interference
from the other compounds expected in
the terminal samples. The measurement
techniques considered during this study
included: gas chromatograph (GC) with
flame mnization detector (FID), GC with
photoionization detector (PID), GC with
argon mnization detector (AID), and a
nondispersive infrared (NDIR) analyzer.
A nondispersive ultraviolet (NDUV)
analyzer was also desired, but no com-
mercial unit was found.
Field Tests
Five lower explosive limit (LEL)
analyzers were field tested, by using a
heated-oven GC with a PID as the
reference analyzer. Two instruments
employed NDIR detectors, and the other
three used catalytic oxidation to generate
a measurement signal. The field tests
were conducted at a terminal in
Baltimore. This terminal uses a refrig-
ertion condenser to recover vapors from
a five-bay loading rack. Samples were
extracted from the exhaust stack through
a stainless steel sampling line to a
manifold in a remote sampling trailer.
The GC was equipped with a gas sam-
pling loop, whereas the LEL analyzers
were used in a flow-through or flow-over
configuration.
Results and Discussion
Laboratory Testing
Five separate instrument config-
urations were evaluated during the
laboratory testing phase of this project.
The NDIR analyzer was the only unit with
the capability to operate in a continuous
analysis mode. The other instruments
were a mid-sized laboratory-type GC
equipped with a temperature-
programmable oven and FID and PID
detectors, a portable heated-oven GC
with an AID, and a portable ambient-
temperature GC with a PID.
GC/FID/Heated Oven—The response
was linear over the entire range of
sample concentrations of interest (1 to
100 ppm), though the calibration curves
show some day-to-day variation and
some deviation from linearity. Least-
squares linear regressions calculated for
the daily analytical results yielded
correlation coefficients ranging from
0.9441 to 0.998, averaging 0.98. With
a 100-pL sample size, the intercept of
the linear calibration lines averaged
-0.53 ppm, ranging from -0.33 to +1.04
ppm. The standard error of estimate
(predicted value) from the linear
regression calculations ranged from 0.24
ppm to 6.00 ppm, averaging 2.29 ppm.
The relative standard deviation (RSD) and
the absolute standard deviation of
replicate analyses were used as
measures of precision (repeatability).
Ninety-five percent of the values were
less than 50 percent of the RSD and the
arithmetic average RSD was 13.77
percent. FID analysis of a paraffinic-
aromatic-naphthenic hydrocarbon
mixture normally yields individual
component results very nearly
proportional to the weight fraction of the
components in the mixture, as the FID
acts as a "carbon counter" for
compounds consisting only of carbon and
hydrogen. The benzene concentration in
the gasoline vapor standard was 3.28
percent by weight. Thus, as expected, the
average benzene fraction of the chro-
matographed sample (16 analyses) was
3.23 percent, with a standard deviation
of 0.90 percent.
GC/PID/Heated Oven—The precisic
and accuracy of the PID performance w;
very similar to that of the FID. The linei
regression performed for the multipoii
calibrations shows strong linearity, wii
correlation coefficients ranging froi
0.946 to 0.999, averaging 0.984. Th
standard error of estimate for th
predicted values ranged from 0.828 1
7.38 and averaged 2.83 ppm. N
unexpected differences in performanc
were observed between PID lamps (
different ionization energies; th
expected lower relative response factc
for the lower-energy lamp (9.5 electron
volts (eV) vs. 10.2 eV) was observed. A
with the FID, the lower limit of reliabl
detection was about 2 ppm; use of th
9.2-eV lamp would increase th
detection limit to 4 ppm. Some portio
of the gasoline vapor is not detected b
the 10.2-eV PID (compounds will
ionization potentials greater than 10.1
eV), which thus yields a lower tota
response and higher calculated benzeni
fraction of the total compared to FIC
analysis. The average benzene fractior
of the total response was 4.96 percent
with a standard deviation of 4.14 percent
Although the average value is consisten
with expectations, the deviation i:
considerably higher than that obtainec
with the FID.
GC/AID/Heated Oven—The portable
heated-oven GC with AID obtained foi
testing was built for ambient (ppb-levei;
measurements, and the unit was
available for only a short period of time.
As a result, the relatively high
concentrations being tested prevented
obtaining any reliable quantification
data.
GC/PID/'Ambient Oven—The test
unit, a portable GC with PID and ambient
temperature oven, with appropriate
column packing and programming of the
flow control valves permitted an
adequate separation of benzene from
gasoline vapors, in contrast to
expectations. The unit was evaluated
both with syringe injections of benzene
standards and with samples pumped
from a Tedlar® bag. Plotted peak heights
at particular detector gain settings were
measured for comparison to the data
machine-reported in parts per million or
volt-seconds. Apparently, because the
internal volume of the analytical system
was large compared to the sample
volume, a measurable amount of the
previous sample remained in the system.
Succeeding analyses of identical
samples gave results closer to the true
value. The unit was not available long
-------
enough to perform rigorous tests on its
ability to quantify benzene in gasoline
vapor, although a few tests were
conducted to ensure that it could be done.
Two replicate analyses gave results of
206.8 and 212.5 ppm benzene, which
agree well with the GC/PID/heated-
oven results.
NDIR—The NDIR analyzer was able to
quantify the benzene-in-air standards
reasonably well, but, as expected, could
not disinguish benzene from the rest of
the components in the gasoline vapor
mixture.
Field Testing
LEL Analyzers—When compared to
each other, the LEL analyzers had a fairly
consistent response to gasoline vapors.
The response characteristics of the LEL
analyzers to specific hydrocarbon
compounds are not known. The
instruments' responses all followed the
same general trend, that is, an increase
in one meant an increase in the others.
When least-squares linear regressions
were calculated between the GC (the
independent variable) and each of the
LEL analyzers, the correlations were not
particularly good, with r2 values in the
0.4-to-0.8 range. Comparing the
individual LELanalyzers with the average
value of all four yielded much better
correlations, as shown in Table 1.
Analyzer D was not operating properly,
so the low value is not surprising. The
high correlations for the catalytic
oxidation units (Analyzers A,B,C) are not
unexpected considering the detection
method.
Benzene/Total Hydrocarbons—Virtu-
ally no benzene (<1 ppm) was found in
the exhaust gas samples, but significant
amounts were present in the inlet to the
condenser. Several tests of the inlet gas
to the refrigeration system showed from
0.64 to 5.86 percent of the total chro-
matographable (PID detection) organic
compounds was benzene. To analyze the
inlet gas with the LEL units, the inlet
stream was diluted with ambient air until
the LEL readings were on scale. No
attempt was made to quantify the actual
dilution rate; it was estimated to be about
20:1 at the lowest ratio and 100:1 at the
highest ratio. The sampled gas concen-
tration of bepzene for the inlet samples
(diluted) ranged from 0.2 ppm to 13.1
ppm, and the undiluted concentration in
the inlet gas was estimated to be from
12 to 590 ppm. No relationship between
benzene and total hydrocarbons as
measured by the LEL analyzers was
observed. However, the same general
GC-LEL relationship discussed above
held with the inlet samples as with the
outlet samples.
Conclusions
Currently, no commercial analyzer is
available that can provide a continuous
measurement of the benzene
concentration in the exhaust gas from
a bulk gasoline storage terminal. Process
GCs designed for field use are available
and could be modified to serve as
semicontinuous analyzers with
acceptable precision and accuracy, but
they are usually expensive. If continuous
analysis is required, inexpensive "total"
hydrocarbon analyzers could be modified
to measure the expected range of
hydrocarbon concentrations. The results
would not quantitatively relate to
benzene emissions unless there were
data to correlate total hydrocarbons with
benzene. As alternatives to process GCs,
several different configurations of
laboratory GCs could accomplish the
requisite benzene separation and
quantification. Either FIDs or PIDs could
be used, and isothermal operation of
packed columns would be acceptable. A
Table 1. LEL Analyzer Regression Summaries
Analyzer
A
B
C
D
Slope
(Standard error)
1.323
(0.0298)
1.321
(0.0257)
0.980
(O.0390)
0.377
(0.0411)
Intercept
-1.545
-4.354
4.625
1.273
Correlation
Coefficient
0.976
0.982
0.929
0.636
heated column is not necessary to
achieve the benzene separation from the
other components of gasoline vapor, but
might be necessary to achieve analytical
cycle times short enough to make the GC
useful as a semicontinuous instrument.
Only one methodology seems to hold
much promise for near-term
development as a continuous benzene
monitor—NDUV. Commercial NDUV
instruments are used for other analytes,
so many of the operating problems and
issues have been resolved. The unknown
is whether a unit could be built at a
reasonable cost combining the needed
sensitivity and interference-rejection
capabilities to measure ppm-level
benzene in a gas stream containing
percent-level hydrocarbons, including
other aromatic compounds. Although
procedures still in their infancy, such as
tunable atomic line molecular
spectroscopy, hold promise, sufficient
data do not exist to assess the probable
performance of such instruments.
Recommendations
The NDUV technique for benzene
monitoring holds sufficient promise to
warrant further investigation. Some
manufacturers have expressed an inter-
est in pursuing this development. This
could permit relatively inexpensive (to
EPA) research. A prototype instrument
could be built and subjected to laboratory
and short-term field tests similar to those
described in the full report to establish
the performance specification data for
the applicable regulations. A surrogate
monitor does not seem practical, because
a great deal of data would be required
to develop the correlations between the
surrogate (total hydrocarbons) and ben-
zene emissions necessary for regulatory
purposes. Both the benzene and total
hydrocarbon emission rates are affected
by product composition and by the
terminal and control device operating
conditions. The limited field tests indicate
that developing the required correlations
would be difficult, at best.
The use of a dedicated GC also
warrants investigation if one analysis
ever 3 to 5 min is considered adequate.
A specially configured instrument would
need to be tested to establish
performance characteristics, because
data extrapolated from this study would
not reflect long-term unattended
operation.
-------
Jon Bolstad is with Pacific Environmental Services, Reston. VA 22090; the
EPA authors Robert G. Fuerst and Thomas J. Logan (a/so the EPA Project
Officer, see below) are with the Environmental Monitoring Systems
Labaoratory, Research Triangle Park, NC 27711.
The complete report, entitled "Benzene Continuous Emission Monitoring
Systems for Gasoline Bulk Storage," (Order No. PB 88-125 679/AS; Cost:
$14.95; 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:
Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
U.8.CFnC!ALM
A • iK GOsT*
<,-\"tNhur v tJ.i.HUiin
! rEB-5'83 ^VATF I
\ / -'SE S300J ~
Official Business
Penalty for Private Use S300
EPA/600/S4-87/034
PS
------- |