United States
Environmental Protection
Agency
National Risk Management
Research Laboratory
Cincinnati, OH 45268
Research and Development
EPA/600/SR-02/007
May 2003
Project Summary
Characterization of Mercury
Emissions at a Chlor-Alkali Plant
John S. Kinsey
Current estimates indicate that up to
160 short tons (146 Mg) of mercury
(Hg) are consumed by the chlor-alkali
industry each year. Very little quantita-
tive information is currently available,
however, on the actual Hg losses from
these facilities. The Hg cell building roof
vent is considered to be the most sig-
nificant potential emission point in chlor-
alkali plants, especially when the cells
are opened for maintenance. Because
of their potential importance, chlor-al-
kali plants have been identified as need-
ing more accurate measurements of
Hg emissions. To obtain a better under-
standing of the fate of Hg within their
manufacturing process, the Olin Corpo-
ration voluntarily agreed to cooperate
with the U.S. Environmental Protection
Agency (EPA) in a comprehensive study
of Hg emissions from their Augusta,
GA, facility, in collaboration with other
members of the Chlorine Institute rep-
resenting the active chlor-alkali plants
in the United States.
To investigate the Hg releases from
the Olin chlor-alkali facility, the EPA's
National Risk Management Research
Laboratory, Air Pollution Prevention and
Control Division (EPA-APPCD) in Re-
search Triangle Park, NC, organized a
special study involving multiple organi-
zations and personnel. However, only
the research conducted by EPA-APPCD
involving roof vent monitoring and air
flow studies conducted in the Olin cell
building is discussed in this report.
The overall objective of the cell build-
ing roof vent monitoring was to deter-
mine the total elemental mercury (Hg°)
mass flux under a range of typical win-
tertime meteorological conditions, in-
cluding both normal operation of the
cell building as well as routine mainte-
nance of Hg cells and decomposers.
Secondary objectives of the research
were to perform an air flow mass bal-
ance for the building and to compare
various Hg monitoring methods under
a variety of sampling conditions. Both
objectives were met during the Febru-
ary 2000 field sampling campaign,
which showed an average Hg°emission
rate of 0.36 g/min from the roof ventila-
tor as determined over the 9-day moni-
toring period.
This Project Summary was developed
by the National Risk Management Re-
search Laboratory's Air Pollution Pre-
vention and Control Division, Research
Triangle Park, NC, to announce key find-
ings of the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).
Introduction
Current estimates indicate that up to
160 short tons (146 Mg) of Hg are con-
sumed by the chlor-alkali industry each
year. Very little quantitative information is
currently available, however, on the ac-
tual Hg losses from these facilities. The
most significant potential emission point
in chlor-alkali plants is thought to be the
Hg cell building roof vent, especially when
the cells are opened for maintenance.
Because of their potential importance,
chlor-alkali plants have been identified
as needing more accurate measurements
of Hg emissions.
To better understand the fate of Hg
within their manufacturing process, the
Olin Corporation voluntarily agreed to
-------
cooperate with the U.S. EPA in a compre-
hensive study of the Hg emissions from
their Augusta, GA, facility. This effort was
conducted in collaboration with other
members of the Chlorine Institute repre-
senting active chlor-alkali plants in the
United States. Chlorine Institute members
have committed to reduce overall Hg° con-
sumption by 50% (from 1990-95 levels)
by the year 2005.
To investigate Hg releases from the
Olin chlor-alkali facility, EPA-APPCD in
Research Triangle Park, NC, organized a
special study involving multiple organiza-
tions and personnel, with each major as-
pect of the study addressed by a separate
principal investigator based on the indi-
vidual area of expertise. However, only
the research conducted by EPA-APPCD
involving roof vent monitoring and airflow
studies conducted in the Olin cell build-
ing is discussed in this report.
The overall objective of the field sam-
pling program was to determine the total
Hg release from the plant using parallel
sampling approaches under typical win-
tertime meteorological conditions, includ-
ing both normal operation of the cell
building as well as routine maintenance
of the electrolytic cells and decomposers.
Secondary objectives of the research were
to perform an air flow mass balance for the
building and to compare various Hg moni-
toring methods under a variety of sam-
pling conditions. Both objectives were met.
Experimental Procedures
A combination of measurement methods
was used for data collection at the Olin
chlor-alkali facility because parallel ap-
proaches reduce the overall uncertainty
of the estimates and provide useful con-
straints on measurement accuracy. The
methods used were roof vent monitoring,
tracer gas analyses, and manual velocity
measurements. Each approach is sum-
marized below.
Roof Vent Monitoring
For this study, long-path instruments
were used in lieu of extractive sampling
using a manifold system to allow measure-
ments to be made on a spatially integrated
basis and to eliminate problems with repre-
sentative sampling typical of point mea-
surements. The primary instrumentation
used in roof vent monitoring consisted of:
• Ultraviolet differential optical absorp-
tion spectrometer (UV-DOAS) for
measurement of gas-phase Hg° con-
centrations;
• Optical scintillometer (anemometer)
for determination of air velocity; and
• Fourier transform infrared (FTIR) spec-
trometer for measurement of sulfur
hexafluoride (SF6) tracer gas concen-
trations.
The long-path instruments were
mounted on wooden sampling platforms
erected at the east and west ends of the
cell building roof vent. Except for the opti-
cal anemometer, the signals from all in-
struments were directed by optical fiber
to computerized data acquisition systems
located in a trailer parked directly be-
neath the roof ventilator at the west end
of the cell building. For the optical an-
emometer, the microprocessor and asso-
ciated laptop computer used for data
acquisition were located on the sampling
platform itself. Each instrument was set
up and calibrated according to the oper-
ating manual and/or approved Quality
Assurance Project Plan for the study. A
Met One Model 062 temperature control-
ler and meteorological station and asso-
ciated laptop computer were also installed
on the west sampling platform to monitor
air temperature and relative humidity.
Tracer Gas Analyses
In addition to the roof ventilator, SF6
tracer gas concentrations were also de-
termined in various cell building open-
ings using manually operated Tedlar® bag
samplers and a closed-cell Nicolet Ma-
gna 760 FTIR spectrometer to analyze
the gas samples. Manual bag sampling
was accomplished by drawing sample air
into the Tedlar® bag over a nominal 24-hr
period to obtain levels of SF6 released
along the open areas on the basement
and cell room floors. Multiple sampling
locations were used to obtain a distribu-
tion of tracer concentrations at key loca-
tions in and around the cell building.
Manual Velocity Measurements
Manual anemometer measurements
were performed to evaluate air velocity in
the roof vent as an independent check on
the optical anemometer as well as to de-
termine the air velocity in various build-
ing openings to perform an overall flow
balance for the cell building. A hand-held
Davis Instruments TurboMeter® propeller
anemometer was eventually used for
manual velocity measurements. The pro-
peller anemometer was also compared
with a hand-held hot wire instrument during
selected measurement periods. Propeller
anemometer readings were obtained both
in the roof ventilator and in cell building
openings. For the vent measurements,
readings were made at selected locations
across the width of the ventilator throat
both at the same height as the optical
anemometer measurement path and also
~ 20 cm below the throat exit. For the
various building openings, anemometer
readings were obtained at the approxi-
mate geometric center of each opening.
Data Reduction and Analysis
In the roof vent monitoring, Hg° con-
centrations measured by the UV-DOAS
were plotted as a chronology, and sum-
mary statistics were calculated for each
24-hr period. In addition, a second data
set consisting of 1-min averages was
downloaded for emission rate calcula-
tions. A similar calculation procedure was
also used for analysis of the optical an-
emometer results, with an appropriate
temperature and pressure correction ap-
plied to the optical anemometer results
prior to combination with the DOAS data.
The roof vent FTIR measurements gen-
erated individual FTIR spectra (64 sepa-
rate scans every 5 min), and the individual
spectra were analyzed by post-process-
ing to determine the concentration of SF6
and other gases of interest. Analysis of
the roof vent FTIR data disclosed that the
FTIR detector was optically saturated due
to poor instrument setup in the field. Be-
cause of this detector saturation, the re-
sponse of the instrument was highly
non-linear, making quantitative interpre-
tation of the spectra impossible. Results
of manual tracer gas measurements us-
ing sample collection into a Tedlar® bag
with FTIR measurement are also reported.
Manual Velocity Measurements
and Flow Balance Calculations
For the manual velocity measurements
in the roof vent, data from a field note-
book were plotted with respect to the physi-
cal boundaries of the ventilator throat and
averages calculated for each set of ob-
servations. These averages were then
compared to similar values obtained from
the optical anemometer data for the same
time period. In addition, the data points
obtained at both edges of the ventilator
were extrapolated by linear regression to
the point of zero velocity. These locations
were then used to determine the effective
flow area of the vent for the emission rate
calculations by trigonometric analysis.
For the cell building openings, manual
velocity data were multiplied by the cross-
sectional area of each opening and com-
bined with the total volumetric flow of the
electrically powered ventilation fans to
obtain the total air entering the cell build-
ing at ambient temperature and pressure.
Similar calculations were also performed
for the roof vent using the applicable op-
tical anemometer data for the same mea-
surement period and the effective flow
area as described earlier. A flow balance
for the entire building was then calcu-
lated using three techniques:
-------
• The total flow values obtained for the
building inlets and roof vent were
corrected to standard temperature
and pressure conditions, and the two
values were compared on a volu-
metric basis.
• The mass of air entering and leaving
the building was calculated, and a
similar comparison was made.
• A method developed by the Occi-
dental Chemical Corporation as part
of their direct mass balance model-
ing effort was also used.
Results
Continuous monitoring was conducted
at the roof vent for Hg° concentration and
air velocity from which 1-min average Hg°
emission rates were calculated. In addi-
tion, continuous monitoring was also at-
tempted for SF6 tracer gas as a separate
measure of the air flow rate from the vent.
Hg° Monitoring Results
Raw 30-second averages generated by
the UV-DOAS were reduced to produce
daily plots of Hg° monitoring results as
well as summary statistics calculated for
each day. The measured Hg° concentra-
tion varied over an order of magnitude
from ~ 73 to 7.3 mg/m3. The overall aver-
age for the study period was 24 mg/m3 of
Hg°. Similar plots and statistics were also
created from analysis of the 1-min optical
anemometer data. The air velocities mea-
sured by the optical anemometer varied
from 0.24 to 1.5 m/s with an overall aver-
age for the monitoring period of 0.93 m/s.
The 1-min average Hg° emission rates
calculated from the monitoring data were
plotted for the 9-day study period with
summary statistics calculated from these
data shown in Table 1. As indicated by
Table 1, the Hg° emission rate varied
over about 2 orders of magnitude from
0.08 to 1.22 g/min. An overall average for
the monitoring period of 0.36 g/min was
also determined.
Tracer Gas Results
Tracer gas results include both roof vent
monitoring conducted using the open-path
FTIR and manual bag sampling con-
ducted in various building openings.
• Roof Vent Monitoring: Results of the
FTIR measurements at the roof vent
were found to be unusable for deter-
mination of volumetric air flow due to
optical saturation of the detector.
However, three common greenhouse
gases (carbon monoxide, nitrous ox-
ide, and methane) were found in
measurable quantities in the roof vent
effluent; exact source(s) of these
gases could not be determined from
available data.
Table 1. Summary of Calculated Hg° Emission Rates
Hg° Emission Rate (g/min)
Date
2/17/00
2/18/00
2/19/00
2/20/00
2/21/00
2/22/00
2/23/00
2/24/00
2/25/00
Maximum
0.82
0.58
0.64
1.22
0.88
0.69
0.65
0.83
0.87
Minimum
0.24
0.15
0.10
0.13
0.12
0.12
0.08
0.24
0.33
Mean
0.38
0.33
0.26
0.35
0.27
0.31
0.30
0.46
0.58
Standard
Deviation
0.076
0.075
0.085
0.19
0.11
0.080
0.090
0.12
0.11
iMumuei ui
Observations
(n)'
747
1339
1364
1368
1311
1340
1300
1130
450
luicii ucniy
Emissions
(g/day)b
N/A
481
370
510
387
453
438
662
N/A
Mean
0.80
0.17
0.36
472
a Dimensionless.
b Sum of measured 1-min values adjusted to a standard day of 1440 min to account for missing data.
Rounded to three significant figures.
Table 2. Results of Air Flow Balance Calculations for the Olin Cell Building3
Date
2/24/00
2/25/00
Volume Balance
(% Closure)
82
100
Mass Balance
(% Closure)
82
99
OxyChem DMB
Results'"
(% Closure)
79
100
Mass Balance
% Difference
2.9
-0.9
a Rounded to two significant figures. D
b Occidental Chemical Corporation direct mass balance (DMB) method as provided by Michael Shaffer. D
• Manual Bag Sampling: The average
concentration of SF6 for the low release
days was just above the detection
limit for the instrumentation. The aver-
age concentration for the high release
days was below the method detec-
tion limit. Although the concentrations
of SF6 measured on February 20,
2000, were less than 5 times the
method detection limit, the concentra-
tions detected were significantly
higher, on average, than any other
sampling day, suggesting minimal Hg
transport during this sampling period.
Air Flow Study Results
Air flow was determined for the cell
building using manual velocity measure-
ments, with associated air flow balance
calculations.
• Roof Vent Monitoring: Air velocity
measurements, performed manually
using a propeller anemometer as well
as an optical anemometer in the
same time period, showed that the
average air velocities determined by
the two methods were within + 10%,
quite acceptable considering the dif-
ferences in measurement technique.
Based upon these measurements, the
measurement path of the optical an-
emometer was considered to be lo-
cated at a point representative of the
average velocity and thus appropri-
ate for use in the emission rate cal-
culations.
Air Flow Balance: The results of the
cell building flow balance calcula-
tions are shown in Table 2 for the
three methods used. Unusually good
closure was obtained in each of the
three flow balance calculations per-
formed, and the three methods also
correlate well with each other. The
high degree of closure of these flow
balances lends further credibility to
the air velocity measurements made
by the optical anemometer in the roof
-------
ventilator to adequately characterize
the air flow from the cell building.
Discussion
No specific pattern could be discerned
from the daily plots of the Hg° emission
rate determined from the roof vent
monitoring conducted in this study. An
attempt was made to correlate various
episodic events where the emission rate
rose for a period of time and then dropped
back to some nominal level with either
process operation or maintenance events
using plant records, but this analysis failed
to find any useful association. Although
the concentration of Hg° was found to
be relatively homogeneous across the
lateral dimension of the roof vent, such
was found not to be the case along
the longitudinal dimension. The differ-
ences observed constitute yet another ar-
gument supporting spatially integrated
readings in lieu of point sampling with a
manifold system. Another observation
made during the study involved the im-
pact of the high electromagnetic field on
instrument operation. For future studies
of this type, optical modems and cables
should be used for the optical anemom-
eter to allow logging of the data at a
remote location to reduce the amount of
lost data and make troubleshooting much
easier for the operator.
Conclusions and
Recommendations
Conclusions drawn from the use of the
equipment, methods, and data analysis
procedures to determine the total Hg° re-
lease and volumetric air flow from the
Olin chlor-alkali cell building are pre-
sented below.
• Hg° concentrations measured by the
UV-DOAS varied over an order of
magnitude from -73 to 7.3 mg/m3. The
overall average for the 9-day study
period was 24 mg Hg°/m3.
• Hg° emission rates measured in the roof
ventilator varied from 0.08 to 1.2 g/min.
An overall average for the monitoring
period of 0.36 g/min was calculated.
• A comparison between the concen-
tration of Hg° measured by the UV-
DOAS and similar measurements
conducted using a hand-held instru-
ment across the width of the roof vent
showed that the Hg° concentrations
were relatively consistent across the
vent and compare reasonably well to
the average concentration obtained
with the UV-DOAS.
• Comparison of roof vent monitoring
data obtained by the UV-DOAS and
point measurements made using a
Tekran Model 2537A automated Hg
analyzer at the entrance to the vent
exhibited a relatively high degree of
scatter with only about 63% of the
variance explained by linear regres-
sion. The data do, however, show
comparable trends in concentration
of Hg° with time. Scatter in the data is
potentially due to a combination of
factors including differences in analy-
sis method, non-representative sam-
pling, and sampling line losses.
• The SF6 tracer gas results obtained
using the long-path FTIR in the roof
vent were found to be unusable for
the purpose of determining volumet-
ric air flow due to optical saturation
of the detector. Results of the 24-hr,
time-integrated bag sampling showed
SF6 tracer gas concentrations either
at or below the instrumental detec-
tion limit except for one sampling
period on February 20, 2000.
• The average roof vent air velocity mea-
sured by a hand-held anemometer
as compared to that obtained by the
optical anemometer showed that the
two methods agreed within + 10%.
• Very good closure (79 to 100%) was
obtained for each of the three air
flow balance calculations performed
for the cell building. The three meth-
ods also correlate well with each
other, and the high degree of closure
of these flow balances lends further
credibility to the air velocity measure-
ments made by the optical anemom-
eter in the roof ventilator.
• No specific pattern could be discerned
from daily plots of Hg° emission rates.
Various episodic events were observed
during the study where the emission
rate rose for a period of time, then
dropped back to some nominal level
which could not be correlated to ei-
ther process operation or mainte-
nance events using plant records.
• Although the concentration of Hg°
was found to be relatively homoge-
neous across the lateral dimension
of the roof vent, concentrations of Hg°
were not consistent along the length
of the ventilator.
On the basis of the results obtained for
this study, the following recommendations
are applicable:
• This study was conducted at one
chlor-alkali plant, in a time window of
approximately 2 weeks. For more
thorough characterizations of opera-
tions in this industry, extended moni-
toring at a single location and/or
monitoring at more plants is recom-
mended to better characterize main-
tenance events and other operational
transients.
• Roof vent instrumentation may be a
useful tool for process monitoring in
some facilities to identify problems in
the operation of the cells that may
require corrective action. The long-
term suitability of these instruments
must be established, however, by
additional on-site evaluations.
• The high electromagnetic field at the
facility has an adverse effect upon
instrument operation. For future stud-
ies of this type, optical modems and
cables should be used to allow log-
ging of data at a remote location to
reduce data loss and make trouble-
shooting much easier for the operator.
• The variation in Hg° concentrations
along the length of the ventilator vs.
the homogeneous values observed
for Hg° across the lateral dimension
argue strongly for the use of spatially
integrated measurements rather than
point sampling with a manifold sys-
tem.
• Roof vent tracer gas data from this
study were not usable. Since the use
of a tracer is well accepted for deter-
mining flow rates, the possibility of
use of tracer gas for future flow mea-
surement studies should not be aban-
doned. Greater care is needed,
however, to verify proper instrument
setup and operation.
• Different tracer gases such as CF4
could be used using UV-DOAS, mak-
ing concurrent sampling and analy-
sis of Hg and tracer gas highly
desirable. Additional research is also
recommended to determine the best
way to diffuse the tracer gas into the
cell room.
• Additional measurements of non-el-
emental (oxidized) forms of Hg should
also be conducted to determine their
overall environmental significance.
-------
John S. Kinsey is also the EPA Project Officer (see below).
The complete report, entitled "Characterization of Mercury Emissions at a Chlor-
Alkali Plant, Volumes I and II," will be available at http://www.epa.gov/ORD/
NRMRL/Pubs/600R02/007 or as Order No. PB2003-100241; Cost: $98.00,
subject to change from:
National Technical Information Serviceu
5285 Port Royal Roadu
Springfield, VA 22161-0001U
Telephone: (703) 605-6000U
(800) 553-6847 (U.S. only)
The EPA Project Officer can be contacted at:
Air Pollution Prevention and Control Division
National Risk Management Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711-0001
United States
Environmental Protection Agency
CenterforEnvironmental Research Information
Cincinnati, OH 45268
PRESORTED STANDARD
POSTAGES FEES PAID
EPA
PERMIT No. G-35
Official Business
Penalty for Private Use
$300
EPA/600/SR-02/007
------- |