EPA-65Q/2-75-026-Q
March 1975
Environmental Protection Technology Series
:::::::::
NS
j& **^
tf
:::::::::::::>::
::::::
,;.;.;.;
jiv»Vi
'I'X't'X'K'l'X'I'I'I'I'I'I'X'X'X'I'l'I'X'X'I'l'X'I'X'I'I't'I'X'I'l'I'I'^I'X'XO'XOX'X'X'I'X'X'^'X'I'I'I'X'X'X'I'I'l'I'I'X'I'I'I'X'I'I'i
£
55
\^'11^/
\ PRQlt0
ss
o
UJ
O
T
-------
EPA-650/2-75-026-0
TESTING OF A MOLECULAR SIEVE
USED TO CONTROL MERCURY EMISSION
FROM A CHLOR-ALKALI PLANT,
VOLUME I
by
John T. Chehaske and John R. Clinc
Engineering-Science, Inc.
7903 Westpark Drive
McLean, Virginia 22101
Contract No. 68-02-1406, Task 3
ROAP No. 21AFA-106
Program Element No. 1AB015
EPA Task Officer: E.J. Woolclridge
Control Systems Laboratory
National Environmental Research Center
Research Triangle Park, N. C. 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
WASHINGTON, D. C. 20460
March 1975
-------
EPA REVIEW NOTICE
This report has been reviewed by the National Environmental Research
Center - Research Triangle Park, Office of Research and Development,
EPA, and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environ-
mental Protection Agency, have been grouped into series. These broad
categories were established to facilitate further development and applica-
tion of environmental technology. Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields. These series arc:
1. ENVIRONMENTAL HEALTH EFFECTS RESEARCH
2. ENVIRONMENTAL PROTECTION TECHNOLOGY
3. ECOLOGICAL RESEARCH
4. ENVIRONMENTAL MONITORING
5. SUCIOECUNOM1C ENVIRONMENTAL STUDIES
6. SCIENTIFIC AND TECHNICAL ASSESSMENT REPORTS
9. MISCELLANEOUS
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to
develop and demonstrate instrumentation, equipment and methodology
to repair or prevent environmental degradation from point and non-
point sources of pollution. This work provides the new or improved
technology required for the control and treatment of pollution sources
to meet environmental quality standards.
This document is available to the public for sale through the National
Technical Information Service, Springfield, Virginia 22161.
-------
ABSTRACT
Performance testing for mercury emission control was conducted
by Engineering-Science, Inc. on the Union Carbide PuraSiv Hg*
unit currently controlling mercury emissions from the by-product
hydrogen stream of the mercury-cell chlor-alkali production facility
of Sobin Chlor-Alkali in Orrington, Maine.
Simultaneous samplings of adsorber inlet and outlet streams were
performed during five consecutive 24-hour adsorption cycles,
eight sampling runs per cycle, in accordance with a modified ver-
sion of EPA Reference Method 102. Inlet mercury concentrations
were stable at 6 to 10 milligrams per Normal cubic meter (mg/Nm ).
3
Outlet mercury concentrations of 0.5 and 3.5 mg/Nm were approached
at the beginning and end of each adsorption cycle, respectively.
This occurrence was due to operation of the unit at nearly 150% of
3
design capacity of 49.6 Nm /min.
If operated at design capacity, the estimate outlet mercury concen-
3
trations from the PuraSiv Hg unit would average 2.0 ug/Nm through-
out each adsorption cycle, representing a 99.9% collection efficiency.
This report was submitted in fulfillment of Contract Number 68-02-
1406, Task Number 3 by Engineering-Science, Inc. under the sponsor-
ship of the Environmental Protection Agency. Work was completed
as of January 31, 1975.
*Registered Trademark of the Union Carbide Corporation
iii
-------
CONTENTS
Page
Abstract ill
List of Figures v
List of Tables vi
Acknowledgements vii
Sections
I Introduction 1-1
II - Summary II-l
III Process Description and Operation III-l
IV Sampling and Analytical Procedures IV-1
V Discussion of Results V-l
IV
-------
FIGURES
No. Page
III-l Schematic Diagram of Mercury-Cell Process III-2
III-2 By-Product Hydrogen Stream Flow Diagram III-4
(Shown With A-l Adsorbing and A-2 Regen-
erating
IV-1 Location of Inlet Sampling Port and Detail IV-3
of Probe and Pitot Tube Assembly
IV-2 Location of Outlet Sampling IV-5
IV-3 Windowless Gas Cell Used in Analysis IV-12
V-l Sketch of Typical Adsorption Curve V-12
V-2 Sketch of Adsorption Curve Generated from V-13
Test Data - Cycle No. 5
V-3 Errors Due to Anisokinetic Conditions V-20
(Small Particles)
-------
TABLES
No.;. Page
III-l Chlorine Production Rates (Cycle No. 1) III-7
III-2 Chlorine Production Rates (Cycle No. 2) III-8
III-3 Chlorine Production Rates (Cycle No. 3} III-9
III-A Chlorine Production Rates (Cycle No. 4) III-10
III-5 Chlorine Production Rates (Cycle No. 5) III-ll
V-l PuraSiv Hg Mercury Control Perfor- V-2
mance Summary (Cycle No. 1)
V-2 PuraSiv Hg Mercury Control Perfor- V-3
mance Summary (Cycle No. 2)
V-3 PuraSiv Hg Mercury Control Perfor- v-4
mance Summary (Cycle No. 3)
V-4 PuraSiv Hg Mercury Control Perfor- v-5
mance Summary (Cycle No. 4)
V-5 PuraSiv Hg Mercury Control Perfor- v-6
mance Summary (Cycle No. 5)
V-6 Mercury Inlet Data Summary (Metric) v-15
V-7 Mercury Inlet Data Summary (English) v-17
VI
-------
ACKNOWLEDGEMENTS
The project management is sincerely grateful to the many persons
and organizations which expended their time and cooperation in
assisting with the completion of this test program.
Particular thanks are extended to Mr. Pete DeAngelis and his
staff at Sobin Chlor-Alkali for generously allowing the testing to
be conducted at the Orrington facility as well as for providing tech-
nical and physical support to the test team.
Mr. Carmen Yon and Mr. Tom Holcotnbe of Union Carbide were extremely
helpful by providing much information about the PuraSiv Hg process
and technical knowledge of the system operation.
vii
-------
SECTION I
INTRODUCTION
Engineering-Science, Inc. (ES) was contracted by the Environmental
Protection Agency, Office of Research and Development, Control
Systems Laboratory, for the purpose of testing a PuraSiv Kg
unit for mercury control at the Sobin Chlor-Alkali plant in Orring-
ton» Maine. This particular plant is a manufacturer of chlorine
and sodium hydroxide by the conventional mercury-cell process. The
specific assignment of this contract required the determination of
mercury emission measurements in order to permit a technical/econo-
mical evaluation of the performance of the proprietary Union Carbide
PuraSiv Hg Process currently controlling mercury emissions from the
plant's by-product hydrogen stream. This particular adsorption
unit had been in service for approximately two years at the time
of the testing.
The by-product hydrogen stream originating from the decomposer por-
tion of the mercury-cell process undergoes primary cooling and
demisting, then mild compression followed by secondary cooling and
demisting. At this point, the hydrogen stream is preheated and
passed through an adsorption bed for additional mercury removal.
The PuraSiv Hg Process is designed such that there is always one
adsorption bed in service while a second adsorber is undergoing
regeneration. The cycles for adsorption and regeneration are alter-
nated between the two beds on a 24-hour basis.
Sites for mercury sampling were located at the outlet line of the
adsorption bed and on the inlet line to the adsorber, downstream
of the final demister but prior to the preheater. A total of five
complete 24-hour adsorption cycles were tested during the period
of September 24-29, 1974. All sample runs consisted of a continuous
1-1
-------
2-hour sampling of inlet and outlet streams simultaneously. An
hour interval between successive sample runs was utilized for the
purpose of sample recovery and preparation of new sampling trains.
All field testing and subsequent laboratory analyses were performed
by personnel of ES and the subcontractor, Commonwealth Laboratory,
The report for this test program has been published in two volumes
with Volume I containing the body of the report and with Volume II
consisting of the appendices.
1-2
-------
SECTION II
SUMMARY
Source testing was conducted on the Union Carbide PuraSiv Hg unit
at the Sobin Chlor-Alkali plant in Orrington, Maine in order to
permit a technical/economical evaluation of the mercury removal
performance of the proprietary adsorption system. This parti-
cular unit is currently controlling mercury emissions from the by-
product hydrogen stream associated with the daily production of
200 tons of chlorine by the mercury-cell chlor-alkali process.
The specific objectives of this test program were:
a. to establish the levels, with times of occurrence, of the
effluent mercury emissions from the PuraSiv Hg unit;
b. to define the mercury collection efficiency of the molecular
sieves during a complete adsorption cycle, as well as at speci-
fic intervals within a given cycle; and
c. to verify that the adsorber mercury emissions do not exceed
60 ppbv, as claimed by Union Carbide, regardless of the inlet
mercury loadings to the adsorber.
This information was successfully obtained as detailed in Section
V of this report.
The by-product hydrogen stream exiting from the decomposer under-
goes conventional treatment steps of primary cooling, demisting,
centrifugal collection, secondary cooling and final demisting.
This stream is then passed through an adsorbent bed which
entraps additional mercury. There are two separate adsorbers which
alternate on a 24-hour basis between adsorption service and regen-
eration, such that one bed is always "on-line."
A portion of the "clean" hydrogen from the "on-line" adsorber is
heated and passed down through the "off-line" bed in order to strip
II-l
-------
the previously collected mercury from the adsorbent. This resultant
mercury-laden regeneration gas is returned to the inlet of the
primary cooler. Another portion of the "clean" hydrogen is recy-
cled directly back to the primary cooler inlet in order to satisfy
the suction demands of the positive-displacement blower located
downstream of the initial demisting stage. The remainder of the
hydrogen from the adsorber outlet is either used for boiler combus-
tion and/or vented to the atmosphere.
Mercury sampling runs were conducted simultaneously on the adsorber
inlet and outlet streams for a period of 2 hours in accordance with
a modified version of the sampling procedures detailed in EPA
*
Reference Method 102. A copy of the method is contained in
Appendix F. A total of five complete adsorption cycles were
tested during the test program.
There were no great variations of mercury concentrations in the
adsorber inlet stream. Concentrations at this point ranged from
3 +
6 to 10 milligrams per Normal cubic meter (mg/Nm ) and did not
correlate with any changes in chlorine production rates. Outlet
mercury concentrations exhibited a definite cyclic pattern. Outlet
mercury concentrations approaching 0.5 mg/Nm were measured during
the initial hours of the cycle, after which the outlet stream was
virtually mercury-free until the 20th hour of the adsorption cycle.
After this initiation of "breakthrough," mercury concentrations
escalated during the last hours of the cycle to values as high as
3
3.5 mg/Nm .
This cyclic variation in outlet mercury concentration was due to
the fact that the adsorbers were being operated at approximately
"Reference Method for Determination of Particulate and Gaseous
Mercury Emissions from Stationary Sources (Hydrogen Streams),"
as published in the Federal Register, Vol. 38, No. 66, Part II,
April 6, 1973.
Normal conditions are defined as 760 millimeters of mercury and
21.1 degrees Centigrade.
II-2
-------
145% of the design flow conditions. An excessive loading of
mercury to the adsorber resulted in the adsorbent capacity
being spent earlier than designed with consequent "breakthrough"
after completion of only 80% to 85% of the adsorption cycle. At
this point, unheated regeneration gas being passed down through
the regenerated bed to prepare it for adsorption was actually
depositing mercury on the adsorbent near the top of the bed. As an
adsorption cycle was begun, mercury in the inlet stream was initially
"trapped" near the bottom of the bed, and the gas stream approached
the top of the bed as a clean gas. Encountering mercury at the
top of the bed resulted in the clean gas stripping the mercury
from the adsorbent at the top. Thus, outlet mercury concentrations,
on the order of 0.5 mg/Nm , were measured at the beginning of the
adsorption cycles.
A realistic evaluation of the PuraSiv Hg unit would exclude
the mercury discharged during the "wash out" in the first part of
the adsorption cycle as well as the mercury emissions occurring
after early "breakthrough." Outlet mercury concentrations, more
representative of the adsorber operation at design conditions, were
experienced after "wash out" and before "breakthrough" when mercury
concentrations were consistently below 10 micrograms per Normal cubic
meter (ug/Nm ). Examination of these results reveal a mean concen-
tration of mercury in the outlet equivalent to 2.0 ug/Nm with a
standard deviation of 0.35 yg/Nm . On a 95% confidence basis,
this corresponds to a mean mercury concentration of 2.0 + 0.73 pg/Nm ,
which is well below the guaranteed maximum outlet mercury concentra-
tion of 60 ppbv (500 ug/Nm ) as claimed by Union Carbide.
Isokinetic sampling conditions for the inlet samples were consis-
tently low due to a small error in the estimated gas moisture con-
tent. Actual sampling rates averaged 79% of the theoretical isokinetic
sampling rate. In many systems, sampling at less than the isokinetic
II-3
-------
rate results in calculated pollutant concentrations that are
higher than the actual concentrations. However, in systems
containing only gaseous and/or very small particulate pollutants
(such as mist or fume), no error is introduced by anisokJnetic
sampling. Because the measured mercury concentrations were below
the equilibrium saturation concentration and, furthermore, because
the mist elimination units located upstream of the sampling site are
purported to remove more than 99% of all particles larger than 3
microns, it is believed that the underisokinetic sampling did not
affect the accuracy of the measured mercury concentrations.
-------
SECTION III
PROCESS DESCRIPTION AND OPERATION
The Sobin plant began chlorine production by the mercury-cell
process in 1967 utilizing the De Nora 2AH5 type cell. The present
production consists of twenty individual cells that operate on a
continuous basis (24 hours/day, 7 days/week) in order to maintain
a daily production rate of 200 tons of chlorine.
As Figure III-l illustrates, the mercury-cell consists of two
distinct sections: the electrolyzer and the decomposer. Sodium
chloride brine is pumped into the inlet end-box and flows cocur-
rently between a liquid mercury cathode and a stationary anode.
Application of a high current density between the electrodes
causes an electrolytic reaction which produces chlorine gas at
the anode and a sodium amalgam at the cathode. The chlorine gas
is exhausted to further treatment such as cooling, drying, compres-
sion and liquefaction before it reaches the final product stage.
Spent brine from the cell is recycled to the purified brine prepara-
tion facilities and the sodium amalgam passes into the decomposer.
In the decomposer, the amalgam serves as an anode to a short-cir-
cuited graphite cathode. By continuously adding purified water
to the decomposer, a second reaction is initiated whereby sodium
hydroxide and by-product hydrogen gas are formed. Before reaching
the final product stage, the sodium hydroxide undergoes additional
treatment such as filtration and concentration. The mercury regen-
erated in the decomposer is returned to the inlet end-box of the
electrolyzer for reuse. The by-product hydrogen stream is subjec-
ted to a series of treatment steps for the purpose of removal of
entrained mercury prior to final disposition of the hydrogen. In
most instances approximately 93% of the total mercury emissions from
a mercury-cell chlor-alkali plant are from this by-product hydrogen
stream.
III-l
-------
FIGURE III-l, SCHEMATIC DIAGRAM OF MERCURY-CELL PROCESS
Treated
Brine
Inlet .
End-Box
Aqueous
Layer "
re
CT>
Electrolyzer
©-
Outlet End-Box
-Aqueous Layer
Water Collection
System
Decomposer
(Denuder)
C XI
o o
Oi i.
O D.
S_ i
o >,
>,co
Caustic Soda
Solution
III-2
-------
In addition to the by-product hydrogen stream, there are two other
sources of mercury emissions from the conventional mercury-cell
process. The end-box ventilation system which exhausts mercury
vapor from the end-boxes of each cell accounts for approximately
5% of the total potential mercury emission from the mercury-cell
process. In order to prevent excessive exposure of the cell-room
personnel to mercury, the cell-room must be adequately ventilated.
The amount of mercury exhausted from the cell-room ventilation air
is a function of the effectiveness of the cell-room maintenance
and housekeeping procedures, but this source of mercury emission
can amount to as much as 1 to 2% of the total potential mercury
emission from the process.
Figure III-2 illustrates the mercury removal techr.ioues employed
on the by-product hydrogen stream at the Sobin plant. First, the
mercury-laden hydrogen, saturated with water vapor, passes through
a primary cooler and a York demister. In the process, portions of
the water and mercury vapors, as well as some existing mercury mist,
are removed. The hydrogen stream is then compressed to approximately
four to five psig by a water-sealed oositive-displacement blower,
after which the stream encounters a centrifugal knock-out pot, a
water-cooled chiller and a Brink mist eliminator in succession.
All of these operations, with the exception of the primary cooler,
are housed in the one-room compressor building. Following the Brink
mist eliminator, the hydrogen stream is piped outside to the PuraSiv
Hg skid.
A steam heater is used to preheat the hydrogen stream, after ehich
the hydrogen enters the bottom of one of the two adsorption beds
which contain the proprietary adsorbent material. Although there
are two adsorption beds, only one bed is on-line at any given time
while simultaneously the second bed is being regenerated. These
adsorbers alternate on a 24-hour cycle between adsorption and
regeneration, and the cycle is switched daily at 5:00 p.m.
III-3
-------
FIGURE 111-2, BY-PRODUCT HYDROGEN STREAM FLOW DIAGRAM
(SHOWN WITH A-l ADSORBING AND A-2 REGENERATING)
Well
Well
« from
Decomposer
(40.5)
Primary
Cooler
York
Demister
Vent
Blower Centrifugal
K.O. Pot
Secondary
Cooler
Brink Mist
Eliminator
Recycle «2 (19.5)
To Boiler (40.5)-.
Regeneration
(9.1)
TI - Temperature Indicator
TIC - Temperature Indicator Controller
PI - Pressure Indicator ,
{ ) - Typical H2 Flow Rate in NnT/m1n Vent'
Outlet 6(60) c - Valve Closed
Sample Point '
C
Adsorber
A-l
(69.1)
Inlet
Sample Point
OPI
hOPi
0.1)
Orifice
Manometer
CO-
C
Ico-
Inlet H
\
V-1 Regeneration
Heater
(69.1)
C
-COi
Adsorber
A-2
Manometer
--:j6 Orifice
^Preheater
-------
As the hydrogen stream moves up through the adsorption bed,
mercury which escaped the previous cooling and demisting stages
is "trapped" by the adsorbent. This results in an adsorber
effluent hydrogen stream with a considerably reduced mercury
content. The treated hydrogen stream is then split and disposed of
in four ways: (1) regeneration gas, (2) recycle gas, (3) boiler
fuel, and (4) stack vent.
The spent adsorption bed is regenerated by passing a hot hydrogen
gas stream down through the bed. A portion of the treated hydro-
gen stream is passed through an electrical heater raising the gas
temperature to approximately 500°F, after which the heated gas
passes down through the bed removing the mercury from the adsorbent.
The regenerator exhaust is a heavily mercury-laden hydrogen stream
which is returned to the inlet of the primary cooler.
Because the water-sealed blower is of the positive-displacement
type, it demands a constant volumetric flow at the blower suction.
In order to satisfy the blower demand and hence maintain consis-
tent pressures throughout the system, a portion of the treated
hydrogen stream is recycled to the inlet of the primary cooler
as dictated by an automatic pressure controller.
Currently, the blower is oversized for the system in order to
accommodate an increase in by-product hydrogen associated with
the planned expansion of the chlorine production facilities. In
the interim, however, the existing hydrogen production rate com-
bined with the returned regeneration stream is not nearly adequate
to meet the blower requirements. Consequently, a large portion
(approximately 48%) of the by-product hydrogen stream is passed
through the adsorption process and then returned to the blower
suction upstream of the primary cooler. However, return of the
"clean" hydrogen at this point mixes this "clean" hydrogen with
the decomposer by-product hydrogen stream which is saturated
III-5
-------
with mercury vapor and contains entrained liquid mercury particles.
Because liquid mercury is present, thermodynamic considerations
dictate that some of this liquid mercury will evaporate into the
original clean portion of the combined gas stream in an effort to
achieve equilibrium. Consequently, although the total amount of
mercury entering the primary cooler is unchanged, the amount of
mercury present in the vapor state which enters the primary cooler
has been increased above the amount of mercury vapor coming only
from the decomposer, and hence has increased the mercury vapor
loading to the PuraSiv Hg unit.
Rather than simply vent all the remaining treated hydrogen to the
atmosphere, some hydrogen is sent to a boiler at the plant in order to
recover the heat value of the stream. When the quantity of available
hydrogen exceeds the boiler requirements, an automatic control valve
actuates, bleeding the excess hydrogen through a stack into the atmos-
phere.
Although plant personnel do not record hourly chlorine production
rates per se, process data is recorded which will permit the cal-
culation of chlorine production rates. The chlorine production
is directly proportional to the number of cells on-line, the vol-
tage (DC) impressed across the electrolytic cells, and the total
amperage consumed. These three parameters are recorded hourly in
the plant operating log. The plant production guideline is that
192 tons of chlorine per day are generated with 20 cells, 92 volts
and 280,000 amperes. Production rates at other operating condi-
tions can be calculated by utilizing a ratio. For example, when
operating at 88 volts, 280,000 amperes and 19 cells, the hourly
chlorine production rate is:
_ 19 cells x 88 volts x 280 x_10^ amps 192 tons Clp/day
C12 rate - 2Q cells x 92 volts x 280 x 10^ amps X 24 hours/day
= 7.2 tons Cl^/hour
Process data for the duration of the test period and calculated pro-
duction rates are contained in Tables III-l through III-5 on the
following pages.
III-6
-------
Table III-l. CHLORINE PRODUCTION RATES
Date: September 24-25, 1974
Test Cycle: No. 1
Time
1700
1800
1900
2000
2100
2200
2300
2AOO
0100
0200
0300
0400
0500
0600
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
No. of Cells
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
Total Amperes x 10~
282
282
282
282
282
282
283
283
283
283
282
282
282
282
282
282
282
282
282
282
282
280
280
281
Total Volts (DC)
93
93
93
93
92
92
92
92
92
92
92
92
92
92
92
92
92
92
92
92
92
92
93
93
Tons Cl /Hour
8.14
8.14
8.14
8.14
8.06
8.06
8.09
8.09
8.09
8.09
8.06
8.06
8.06
8.06
8.06
8.06
8.06
8.06
8.06
8.06
8.06
8.00
8.09
8.12
Total Cl? Production During Cycle
194 tons
Hean Cl_ Production Rate During Cycle = 8.08 tons/hr
Standard Deviation of C12 Production Rate
0.034 tons/hr
III-7
-------
Table III-2. CHLORINE PRODUCTION RATES
Date: September 25-26, 1974
Test Cycle: No. 2
Time
1700
1800
1900
2000
2100
2200
2300
2400
0100
0200
0300
0400
0500
0600
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
No. of Cells
20
20
20
20
20
20
20
20
20
20
20
20
20
19
19
19
18
18
18
18
19
19
19
20
Total Amperes x 10
282
282
281
282
282
282
282
282
282
282
282
282
283
281
281
280
280
280
280
280
280
280
280
280
Total Volts (DC)
93
93
93
93
93
93
93
93
93
93
93
93
93
87
87
87
84
84
84
84
87
87
87
93
Tons Cl«/Hour
8.14
8.14
8.12
8.14
8.14
8.14
8.14
8.14
8.14
8.14
8.14
8.14
8.17
7.21
7.21
7.19
6.57
6.57
6.57
6.57
7.19
7.19
7.19
8.09
Total Cl, Production During Cycle = 183 tons
Mean Cl_ Production Rate During Cycle » 7.64 tons/hr
Standard Deviation of Cl. Production Rate - 0.62 tons/hr
III-8
-------
Table III-3. CHLORINE PRODUCTION RATES
Date: September 26-27, 1974
Test Cycle: No. 3
Time
1700
1800
1900
2000
2100
2200
2300
2400
0100
0200
0300
0400
0500
0600
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
No. of Cells
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
19
19
19
20
20
20
20
Total Amperes x 10
280
280
Plant Down
180
281
282
282
282
282
282
282
282
282
282
282
282
282
280
280
280
200
280
282
281
Total Volts (DC)
93
93
79
93
93
93
93
93
93
93
93
93
93
93
93
93
88
88
88
85
93
94
93
Tons Cl /Hour
8.09
8.09
4.42
8.12
8.14
8.14
8.14
8.14
8.14
8.14
8.14
8.14
8.14
8.14
8.14
8.14
7.27
7.27
7.27
5.28
8.09
8.23
8.12
Total Cl- Production During Cycle - 178 tons (approximated due to plant
shutdown)
Mean Cl- Production Rate During Cycle =7.74 tons/hr
Standard Deviation of Cl» Production Rate - 0.94 tons/hr
III-9
-------
Table III-4. CHLORINE PRODUCTION RATES
Date: September 27-28, 1974
Test Cycle: NO. 4
Time
1700
1800
1900
2000
2100
2200
2300
2400
0100
0200
0300
0400
0500
0600
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
No. of Cells
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
Total Amperes x 10~
281
280
280
281
281
282
282
282
282
282
282
282
282
282
281
280
281
280
280
280
280
280
280
280
Total Volts (DC)
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
Tons Cl /Hour
8.12
8.09
8.09
8.12
8.12
8.14
8.14
8.14
8.14
8.14
8.14
8.14
8.14
8.14
8.12
8.09
8.12
8.09
8.09
8.09
8.09
8.09
8.09
8.09
Total Cl, Production During Cycle
195 tons
Mean Cl, Production Rate During Cycle 8.12 tons/hr
Standard Deviation of Cl^ Production Rate
0.023 tons/hr
111-10
-------
Table III-5. CHLORINE PRODUCTION RATES
Date: September 28-29, 1974
Test Cycle: No. 5
Time
1700
1800
1900
2000
2100
2200
2300
2400
0100
0200
0300
0400
0500
0600
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
No. of Cells
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
Total Amperes x 10
280
280
280
280
280
280
280
280
280
280
280
280
280
280
280
280
280
280
280
280
280
280
280
280
Total Volts (DC)
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
93
Tons Cl /Hour
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
Total Cl- Production During Cycle = 194 tons
Mean Cl- Production Rate During Cycle « 8.09 tons/hr
Standard Deviation of Cl- Production Rate = 0
III-ll
-------
It is imperative to realize that the actual mercury-cell process
parameters do not appreciably affect the mercury concentration or
the hydrogen flow rate to the PuraSiv Hg unit. Obviously,
the rate of chlorine production determines the amount of by-product
hydrogen production which in turn determines the total quantity of
mercury fed to the decomposer. However, between the decomposer
and the adsorbent are two coolers, a centrifugal separator, and two
demisters. These units are designed to remove liquid mercury
droplets and mist so that the hydrogen entering the adsorber contains
only vaporized mercury. This pretreatment system acts as a buffer
between the adsorption system and the mercury-cell process, resulting
3
in relatively constant (6 to 10 mg/Nra ) mercury concentrations
at the adsorber inlet.
The adsorber inlet flow rate is also essentially independent of cell-
room process conditions. By varying the amount of recycled hydro-
gen, the positive-displacement blower is able to maintain a nearly
constant inlet flow rate to the adsorbent. (During the testing pro-
gram, inlet flow variation was less than 10%.) Because both the
flow rate and the mercury concentration entering the adsorber are
not affected by cell-room operating conditions, extensive process
data acquisition was not deemed necessary to the successful execu-
tion of the testing program.
In addition, it was beyond the scope of this test program to ascer-
tain the influences of the PuraSiv Hg operating parameters.
The adsorption process variables were fixed and controlled through-
out the duration of the test period in accordance with the plant's
normal operating procedure. Because no deviations from normal
operating conditions were encountered, evaluation of the effect
of adsorption process variables on adsorbent performance was not
possible.
111-12
-------
SECTION IV
SAMPLING AND ANALYTICAL PROCEDURES
At the coordination meeting held at the plant site on the first
day of testing, representatives of Battelle-Columbus Laboratories
expressed their opinion that isokinetic sampling of the inlet was
necessary. Battelle was interested in the testing because the
results will be used by Battelle in the performance of an associ-
ated Control Systems Laboratory contract. The necessity to sample
isokinetically is dependent upon the presence of mercury particles.
Although equilibrium data predicts no particles, the attainment
of equilibrium within the system could neither be proved or dis-
proved to the satisfaction of those concerned.
Since isokinetic sampling of the inlet had not been planned, no
provisions had been made to install sample ports as specified by
Method 102, nor did any such sample port location exist that would
permit sampling of two complete duct traverses. To achieve such
capability would have required the fabrication and installation
of special piping. A complete plant shutdown, including a nitro-
gen purging of all hydrogen lines, would have been necessary for
installation of the piping. For obvious reasons, the plant was
not interested in this approach. Therefore, a compromise was
reached to do a partial sampling traverse with an existing probe
and pitot tube which were located between the Brink demister and
the preheater. The EPA Project Officer agreed to this modifica-
tion of the Work Plan as well as the necessary modifications to
Method 102 which are described in this section.
LOCATION OF SAMPLING PORTS
As stated, the inlet sampling site used for this test was an
existing sample port complete with a probe and pitot tube previously
IV-1
-------
installed for a Union Carbide test program. The port consisted
of a 4-inch flanged pipe nozzle installed in a straight run of
6-inch inlet pipe within the compressor room. There was a straight,
unobstructed run of pipe for a distance of 8 feet and 6 inches
upstream of the sample port, and, likewise, a straight run of pipe
for a distance of 8 feet and 4 inches downstream of the port. A
schematic diagram of the sample port location is contained in
Figure IV-1. A detailed sketch of the probe and pitot tube assem-
bly showing their relative positions and the location of the tra-
verse points is also provided in Figure IV-1. The pitot tube was
a stainless steel Dwyer "pocket" model with a 1/8-inch outside
diameter. Calibration of this pitot tube in the laboratory sub-
sequent to the test program yielded a pitot coefficient factor
of 0.99. The sample probe was a scaled-down version of the more
familiar "standard size" button-hook nozzle/probe assembly and
was fabricated from 0.12-inch I.D. stainles steel tubing. Traverse
points were located as specified by EPA Method 102. The inlet
sampling location easily exceeded the minimum requirements of
being eight stack diameters downstream and two stack diameters
upstream from any flow disturbance. Thus, a minimum total of four
traverse points, two points on a diameter, were required. Due to
the limited length of the probe and pitot tube, any sample points
on the diameter past the centerline could not be reached. There
was no alternate equipment available to be used to resolve this
situation. Thus, two points on the one-half diameter were chosen.
The specified location of Point No. 1 was located less than 1 inch
from the stack wall; thus in accordance with the Method, traverse
Point No. 1 was located at 1.0 inch. Point No. 2 was located at
1.5 inches from the stack wall.
It should be noted that, due to the presence of the Swagelok fitting
on the probe assembly, the probe could be traversed from the wall
to the centerline of the pipe. However, the compression fitting on
IV-2
-------
FIGURE IV-1, LOCATION OF INLET SAMPLING PORT AND DETAIL
OF PROBE AND PITOT TUBE ASSEMBLY
n o D
Z*
LFrom Brink
Demister
To Preheater
r - -,- - , C'4" ., , \ .
."[>, B q «UT
; ©^; 6" SCH 40 Pipe J
'<. l^^^Inlet Sanple Port With
N, Probe and Pi tot Tube
See Large Detail
1/8" O.D.
Pitot Tube
Flow
Whitey Whitey
OKS2 4354
Valve Valve
Whitey
OKS2
Valve
Not to Scale
IV-3
-------
the pitot tube assembly required the pitot tube to remain sta-
tionary at 1/4-inch beyond the pipe centerline. As shown later
in this section, a velocity traverse was performed on the one-half
diameter that was accessible. However, this was done under the
presence of leaking hydrogen caused by removing the pitot tube
fitting from the flanged pipe; this condition would have been
intolerable for the duration of the test program.
Because the two adsorption beds are alternated in service between
adsorption and regeneration, the two units are connected via a
manifold in such a manner that the treated hydrogen stream exiting
from the on-line adsorber provides both the hydrogen used for regen-
eration of the second adsorber, as well as the hydrogen stream to
be used for combustion, recycle, or simply vented. It is this
latter hydrogen stream, referred to as the adsorber "outlet" which
was sampled during the test period. A schematic diagram of this
manifold arrangement and the location of the sampling site for
the "outlet" is shown in Figure IV-2.
The existing sample tap was a 1/2-inch pipe nipple installed in
the 6-inch "outlet" pipe approximately 3 feet downstream of the
manifold takeoff and 9 feet upstream of any flow disturbance.
From this sample tap, 1/4-inch O.D. stainless steel tubing traverses
4 feet horizontally to the side of adsorber A-l and then down the
side of the adsorber for another 9 feet, terminating at a valve.
The sample train was connected with a 4-foot length of Teflon
tubing to a Swagelok fitting at the valve, and all "outlet" samples
were collected through this connection to the existing sample tap.
SAMPLING PROCEDURES
The expected magnitude of the inlet flow rate within the snail
diameter pipe indicated the possibility of a high Reynolds Number
with the associated high degree of turbulence and flat velocity
IV-4
-------
FIGURE IV-2, LOCATION OF OUTLET SAMPLING
See Detail Below H- Outlet
Regeneration
A-l
9'
^rr\r\ 11 T r*i i
1 or VILW
4'
Samplina Train
Connected Here
SIDE VIEW
IV-5
-------
profile. A velocity traverse with the existing pitot tube con-
firmed these expectations, shown below.
Distance from Wall A£ Stack Temp., °C
1/2" 0.58 22
1" 0.59 22
1-1/2" 0.58 22
2" 0.58 22
2-1/2" 0.59 22
3" 0.59 22
As noted, due to the design of the existing pitot tube and probe,
only half of the diameter could be traversed. That half of the
diameter on the side away from the sampling port was inaccessible
because of the limited length of probe and pitot tube.
Because this flat velocity profile existed at a high Reynolds
Number, the inlet gas stream would be expected to be extensively
mixed, i.e., a homogeneous gas phase with no segregation of parti-
culate matter in any specific region of the duct. Therefore, the
composition of the stream in the inaccessible portion of the duct
should be identical to the gas stream on the one-half of the dia-
meter available for sampling. Hence, the collected samples are
believed to be a true representation of the composition of the
adsorber inlet stream.
Sampling of the adsorber inlet stream was conducted in accordance
with the test procedures as specified in the aforementioned EPA
Method 102, using the standard sampling train of that method as
described in Appendix F, with the following modifications:
a. The pitot tube and probe varied in design from those prescribed
by Method 102 and permitted a traverse of only one-half of the
duct diameter.
b. A section of Teflon tubing with a Swagelok fitting on one end
and a ground-glass socket on the other end was employed to
IV-6
-------
connect the existing probe to the inlet of .the first impinger.
A Whitey needle valve was also installed in this section of
Teflon tubing.
c. Since the stream to be sampled was under positive pressure,
the vacuum pump in the sampling train was by-passed and the
sample was simply pushed through the sampling train. The
needle valve as described above was used to regulate the samp-
ling flow rate.
Isokinetic sampling runs were conducted on the adsorber inlet stream
for a period of two hours. Readings were taken every five minutes
and the probe was alternated between the two traverse points every
10 minutes.
The adsorber outlet stream was sampled simultaneously over a two-
hour period in accordance with the specifications of EPA Method 102,
using the standard sampling train as described in Appendix F. The
following modifications were made to those procedures when sampling
the adsorber outlet stream:
a. No probe or pitot tube was used since the adsorber outlet stream
was completely gaseous. Consequently, no point-by-point traverse
was required.
b. A section of Teflon tubing, similar to the one used for the
inlet stream sampling, was employed to connect the sample tap
to the first impinger.
c. The vacuum pump was by-passed and the sample was allowed to
push itself through the sampling train. The sample rate
was regulated by adjusting the Whitey needle valve in the
section of Teflon tubing.
In the cases of both inlet and outlet streams, by adopting the
described modifications, there is no reason to expect less accuracy
in the results than that normally obtained by Method 102. In the
IV-7
-------
case of the inlet stream, insertion of the "standard size" probe
and S-type pitot tube into the 6-inch diameter sample duct would
have created flow disturbances which would have resulted in erro-
neous measurements. Utilization of the miniature probe and pitot
tube, as described, most probably prevented a possible source of
error. Because isokinetic sampling rates were not a critical
factor in determination of the mercury concentrations (as detailed
in Section V), the accuracy in measurement of the sampled gas
volume, corrected to normal conditions, represents the accuracy
of the sampling train. Because the test instruments used to deter-
mine the sample gas volume at standard conditions were either
calibrated prior to field testing and/or the accuracy of the instru-
ments were known, based upon the contractor's prior experience
with Method 102 and other related Methods, it is estimated that
sampling train results differed no more than +_ 5% from the actual
values.
During the 2-hour sampling periods of the adsorber outlet stream,
readings were also taken every five minutes. Sample flow rates
were maintained at approximately 1 cfm throughout the runs. The
field data sheets indicating what sampling parameters were moni-
tored for all sampling runs are contained in Appendix G.
At the completion of the 2-hour simultaneous samplings of the
inlet and outlet streams, the sample boxes were disconnected at
the outlet from the fourth impinger and at the Swagelok fitting
at the sample tap. Sample recovery was performed, as prescribed
in the aforementioned EPA Method 102, with one additional step.
The Teflon tubing from each train was also rinsed with 0.1M 1C1
and the washings were added to the recovered sample of the inlet
and outlet, respectively. This tubing was likewise cleaned with
a distilled water wash along with the impingers and other glass-
ware prior to preparation of a new sampling train. The field
cleanup data sheets for all sampling runs are contained in Appen-
dix H.
IV-8
-------
All sample recovery, glassware cleaning, and sample train prepara-
tion were performed in a covered van located at the rear of the
plant's office building parking lot. This location was situated
upwind of the prevailing wind direction from the plant proper,
approximately 150 meters from the nearest mercury emission source.
Checks of the mercury level inside the cleanup van were made with
Drager tubes having a lower limit of detectability of 0.005 ppm
by volume. These tests showed no detectable mercury.
There were no sampling equipment failures experienced during the
six days of testing. There was, however, one plant upset which
occurred on the evening of September 26 which involved a plant
shutdown. This event necessitated the curtailment of the simul-
taneous sampling run in progress at that time and also required
shortening the sampling period of the subsequent simultaneous
sampling run in order to remain in phase with the 24-hour cycles
of the adsorbers, which were still scheduled to be switched at
5:00 p.m. on the following day.
In addition to recording the standard sampling data as specified
by Method 102, certain process data relevant to the PuraSiv Hg
operation were monitored at 5-minute intervals as well. Those
process parameters of interest were the following:
a. gas flow rate to the on-line adsorber;
b. regeneration gas flow rate to the off-line adsorber;
c. adsorber inlet pressure after preheater;
d. adsorber outlet pressure;
e. gas temperature near the inlet sampling site, i.e., downstream
of the secondary cooler but upstream from the Brink demister;
f. gas temperature at the preheater outlet; and
g. adsorber inlet pressure before preheater.
IV- 9
-------
Recordings of this data were incorporated into the standard field
data sheets which are contained in Appendix G. The instruments
and gauges used to obtain these readings are described in Appendix
I. These instruments were all permanently installed in the various
process lines by the plant. For this reason, calibration of the
instruments could not be conducted by the test team. Because of
the positive pressure in the system and the lack of necessary taps
in the piping, the tests were conducted utilizing the existing
plant instrumentation.
ANALYTICAL PROCEDURES
Upon receipt of the recovered samples at the laboratory, the samples
were analyzed spectrophotometrically for mercury content as speci-
fied in EPA Method 102. The following modifications of Method 102
were used:
a. Flameless spectrophotometric analysis for mercury content is
extremely sensitive for mercury detection. In addition, the
specified mercury concentration range for the standard solu-
tions is quite high, since at that concentration range the
calibration curve obtained is extremely flat. Such a cali-
bration curve renders a poor resolution between two slight
concentration differences. Therefore, in order to improve
resolution, the mercury concentrations of the standard solu-
tions were lowered by dilution with 0.1M 1C1 and the samples
were analyzed at levels of 1 to 5 yg/1 of Hg. Not only did
this procedure improve the resolution, but also less scrubbing
time was required and system contamination and "memory" were
decreased.
b. The scrubber solution volume was increased to 200 ml in order
to improve repeatability.
c. The normal method of spectrophotometric analysis utilizes a
quartz glass cell through which the sample flows during
IV-10
-------
absorbency measurement. This frequently results in the "win-
dows," i.e., the two ends, of the quartz tube becoming contam-
inated with mercury and analytical errors occurring due to
this interference. Because the flameless atomic absorption
analytical technique is extremely sensitive for mercury detec-
tion, the need to eliminate any and all possible sources of
interference is obvious. In order to resolve this situation,
a special windowless quartz glass cell was employed for the
mercury analysis of this test program. As shown in Figure IV-3,
the mercury-laden zero grade air from the scrubber flask enters
at the center of the quartz tube and flows toward both open
ends. While the sample is flowing within the tube, the absor-
bency can be measured accurately without the possible inter-
ference of "dirty" cell windows. This modification also allows
easier intermittent cleaning of the quartz cell. In addition,
the exhaust hood was lowered closer to the cell in order to
improve suction around the cell.
There are no known literature references or other written
documentation demonstrating or negating the validity of this
windowless cell procedure. The method was developed indepen-
dently by subcontractor personnel who possessed prior experience
in the analysis of mercury by flameless atomic absorption (FAA).
Prior to the analyses of the samples collected from this test
program, the windowless cell technique was demonstrated by using
an EPA reference sample obtained from the Analytical Quality Con-
trol Laboratory, National Environmental Research Center, U.S.
Environmental Protection Agency in Cincinnati, Ohio. The "known"
value of the mercury concentration in this reference sample was
2.40 yg/1. The measured value of the mercury concentration in
that sample when using the windowless cell method was 2.76 yg/1.
Personnel associated with the above Analytical Quality Control
Laboratory indicated that FAA results for mercury analysis may
vary by as much as + 50% from the "known" value when employing
IV-11
-------
FIGURE IV-3, V.'INDOWLESS GAS CELL USED IN ANALYSIS
Suction
Hood
hv
Gas Stream With
Mercury Vapor
Gas Flows
Tube From
Scrubber Flask
IV-12
-------
the conventional single-pass FAA procedure at low concen-
*
trations of mercury (<10 yg/1). The result obtained with
the windowless cell should provide some indication of the
reliability of that method under the analytical conditions
described for this test program.
d. Possible background readings attributable to entrained parti-
cles and/or water vapor were decreased by employing a Kjeldahl
bulb "condenser," submerged in an ice bath, between the scrubber
flask and the gas cell.
e. A recent collaborative test program has shown that the analyti-
cal procedure as contained in Method 102 may not provide re-
producible accurate results when analyzing samples obtained
**
from mercury-cell chlor-alkali plants. Test results, pub-
lished subsequent to the laboratory analyses of this project,
showed that accurate and reproducible mercury determinations
were obtained only when analyzing the samples at mercury con-
centration levels of 0.2 to 5.0 pg/1. Fortunately, this was
one of the modifications of Method 102 employed in this test
program. In the above-mentioned collaborative test program,
mercury analyses employing the mercury concentration range of
0.2 to 5.0 ug/1 demonstrated an accuracy of less than 1%
difference between actual and measured concentrations.
Therefore, because the analytical procedures of that program
are so similar to the actual practices used for this test,
the analytical results obtained from this test program are
estimated to be no more than + 2% of the actual mercury contents.
*Personal communications from analytical chemists at Analytical
Quality Control Laboratory, U.S. Environmental Protection Agency,
Cincinnati, Ohio. March 1975. Engineering-Science, Inc. McLean,
Virginia.
** "Evaluations of the EPA Method for the Determination of Mercury
Emissions from Stationary Sources," Mitchell, W. J. and Midgett,
M. R., Quality Assurance and Environmental Monitoring Laboratory,
U. S. Environmental Protection Agency, October, 1974.
IV-13
-------
A detailed description of the analytical procedures and the labora-
tory report are contained in Appendix F and Appendix J, respectively.
In order to obtain the required mercury concentration range of 1
to 5 yg/1, special dilution procedures were needed. All "prepared
samples" were made by adding 10 ml of ION NaOH to a 50 ml aliquot
of the sample returned from the field. In all analyses, the total
volume of solution charged to the aeration scrubber was 200 ml.
In making the scrubber solutions for the adsorber outlet
samples, the following quantities were used: (1) 10 ml ION NaOH;
(2) 20 ml standard Method 102 reducing agent; (3) 120 ml distilled
water; and (A) 50 ml of the "prepared sample." However, the scrub-
ber solutions for the adsorber inlet samples were made up
on an individual sample-by-sample basis, in order to achieve a Hg
concentration in the 1 to 5 yg/1 working range. In all cases, 10 ml
ION NaOH and 20 ml of the reducing agent were added. By employing
a microsyringe, only a few microliters of the inlet "prepared sample"
were then added and the volume was diluted to 200 ml total with
distilled water. If the resultant solution was found to have a Hg
concentration greater than 5 yg/1, the scrubber solution for that
particular sample was made up again using fewer microliters of
the "prepared sample" for that run.
IV-14
-------
SECTION V
DISCUSSION OF RESULTS
In order to determine the extent of mercury removal by the pro-
prietary PuraSiv Hg process, source sampling was conducted
on this system during a period of five consecutive 24-hour adsorp-
tion cycles. Throughout the duration of each of these cycles, a
series of eight distinct sampling tests were executed simultaneously
on the adsorber feed and discharge streams. The test data obtained
permitted calculation of the mercury concentrations and mercury
mass flow rates to and from the adsorption unit at specific periods
of the adsorption cycle. The results, as well as the mercury removal
efficiency of the PuraSiv Hg unit, are tabulated separately
for each of the five adsorption cycles tested as shown in Tables V-l
through V-5.
All mercury concentrations as determined by the sample concentration
method and all gas flow rates are calculated at standard conditions
(760 mm Hg, 21.1°C). Descriptions of the sample concentration method
of calculation and of the ratio-of-area method of calculation are
contained in Appendix A. Mercury mass flow rates for the inlet
stream were calculated using the appropriate mercury concentration
and gas flow rate as determined by the pitot tube measurements at
the inlet sampling site. The mercury mass flow rates for the out-
let stream were calculated in the same manner after the outlet flow
rate was determined by subtracting the regeneration gas flow rate
from the total inlet flow rate. Example calculations of mercury
concentration and mass flow rate for both streams are contained in
Appendix B.
It should be noted that the adsorber outlet referenced in Tables V-l
through V-5 is not the total flow discharging from the adsorber.
As detailed in Section III, the total "clean" hydrogen gas from
the adsorber is separated via manifold piping into the outlet
V-l
-------
Table V-l. PURASIV HG MERCURY CONTROL PERFORMANCE SUMMARY
September 24-25, 1974
Cycle No
Adsorber
Is 24-hours duration
A
Hour of
Adsorption Cycle
0 -
3.25 -
6.5 -
9.67 -
12.5 -
15.75 -
18.67 -
22 -
2
5.25
8.5
11.67
14.5
17.75
20.67
24
Hg Concentration, mg/Nm
Inlet Outlet
7.98
7.78
7.26
7.46
9.50
10.10
7.94
8.00
0.455
0.026
<0.001
0.002
< 0.001
0.009
0.795
3.56
Hg Flow
Inlet
33,300
33,200
31,600
31,800
43,500
42,100
32,800
32,400
Rate, mg/hr
Outlet
1,670
97
4
7
4
33
2,850
12,500
Hg Control
Efficiency,*
94.2
99.7
99.9
99.9
99.9
99.9
90.0
55.4
Chlorine Produc.
Rate, tons/hi
8.14
8.09
8.09
6.06
E.06
£.06
£.06
£.07
Average Hg Collection Efficiency During Cycle « 92.4%
Total Hg in Feed Stream During Cycle - 691 g
Total Maximum Hg Discharge During Cycle - 47 g
NOTEs Nm is defined as one normal cubic meter, i.e., one cubic meter at standard conditions of
760 mm Hg and 21.1eC.
Average Hg Collection Efficiency During Cycle has been calculated as the mean value of the
individual efficiencies for each sampling period. This value is not based on the total Hg
in and out of the sieve during an adsorption cycle.
-------
Table V-2. PURASIV HG MERCURY CONTROL PERFORMANCE SUMMARY
September 25-26, 1974
Cycle No.
Adsorber
. 2 : 24-hours duration
B
Hour of
Adsorption Cycle
0 -
3 -
5.75 -
9 -
12 -
15 -
18 -
21 -
2
5
7.75
11
14
17
20
23
Hg Concentration, mg/Nm
Inlet Outlet
8.09
7.91
7.01
6.97
6.96
7.52
8.22
5.54
0.476
0.486
0.003
0.002
0.002
0.003
0.219
3.20
Hg Flow Rate, mg/hr Hg Control
Inlet Outlet Efficiency, %
33,400
32,600
28,600
28,500
28,300
30,600
32,600
22,300
1,710
1,740
11
7
7
11
750
11,100
94.1
93.8
99.9
99.9
99.9
99.9
97.3
42.2
Chlorine Produc.
Rate, tons/hr
8.13
8.14
8.14
8.14
7.53
6.78
6.78
7.49
Average Hg Collection Efficiency » 90,9%
Total Hg in Feed Stream During Cycle - 712 g
Total Maximum Hg Discharge During Cycle » 51 g
NOTE: Nm is defined as one normal cubic meter, i.e., one cubic meter at standard conditions of
760 mm Hg and 21.1"C.
Average Hg Collection Efficiency During Cycle has been calculated as the mean value of the
individual efficiencies for each sampling period. This value is not based on the total Hg
in and out of the sieve during an adsorption cycle.
-------
f
Table V-3.
Cycle No. 3: 22.25-hours duration
Adsorber A
PURASIV HG MERCURY CONTROL PERFORMANCE SUMMARY
September 26-27, 1974
Hour of
Adsorption Cycle
0
2
4.25
7.25
10.25
13.25
16.25
19.25
- 1.75
- 3.25
- 6.25
- 9.25
- 12.25
- 15.25
- 18.25
- 21.25
Hg Concentration, mg/Nm
Inlet Outlet
8.73
6.94
6.86
5.86
6.68
6.31
7.28
7.87
0.456
0.016
0.003
0.001
0.001
0.002
0.002
0.067
Hg Flow Rate, mg/hr Hg Control
Inlet Outlet Efficiency, %
36,200
28,000
27,900
24,400
27,200
25,400
29,300
32,300
1,640
56
11
4
4
7
7
240
94.8
99.8
99.9
99.9
99.9
99.9
99.9
99.1
Chlorine Produc.
Hate, tons/hr
8.09
8.13
8.14
8.14
8.14
7.85
6.61
8.15
Average Hg Collection Efficiency - 99.2*
Total Hg in Feed Stream During Cycle 628 g
Total Maximum Hg Discharge During Cycle - 10 g
NOTE: Nm is defined as one normal cubic meter, i.e., one cubic meter at standard conditions
760 mm Hg and 21.1°C.
Average Hg Collection Efficiency During Cycle has been calculated as the mean value of the
individual efficiencies for each sampling period. This value is not based on the total Hg
in and out of the sieve during an adsorption cycle.
-------
<
Table V-4. PURASIV HG MERCURY CONTROL PERFORMANCE SUMMARY
September 27-28, 1974
Cycle No
Adsorber
. 4: 24-hour s duration
B
Hour of
Adsorption Cycle
0 -
3 -
6 -
9 -
12 -
15 -
18 -
21 -
2
5
8
11
14
17
20
23
3
Hg Concentration, mg/Nm
Inlet Outlet
7.83
6.03
7.24
7.65
6.52
6.88
7.77
8.86
0.039
0.002
0.001
0.001
0.001
0.001
0.030
0.996
Hg Flow Rate, mg/hr Hg Control
Inlet Outlet Efficiency, %
32,300
24,900
31,100
34,000
28,400
29,300
33,400
37,500
140
7
4
4
4
4
110
3,670
99.5
99.9
99.9
99.9
99.9
99.9
99.6
88.8
Chlorine Produc.
Rate, tons/hr
8,10
8.13
8.14
9.14
8.13
8.10
8.09
8.09
Average Hg Collection Efficiency « 98.4%
Total Hg in Feed Stream During Cycle = 653 g
Total Maximum Hg Discharge During Cycle - 18 g
NOTE: Nm is defined as one normal cubic meter, i.e., one cubic meter at standard conditions
760 mm Hg and 21.1°C.
Average Hg Collection Efficiency During Cycle has been calculated as the mean value of the
individual efficiencies for each sampling period. This value is not based on the total Hg
in and out of the sieve during an adsorption cycle.
-------
<
Table V-5. PURASIV HG MERCURY CONTROL PERFORMANCE SUMMARY
September 28-29, 1974
Cycle No. 5: 24-hours duration
Adsorber A
Hour of
Adsorption Cycle
0 -
3 -
6 -
9 -
12 -
15 -
18 -
22 -
2
5
8
11
14
17
20
24
3
Hg Concentration, rcg/Nm
Inlet Outlet
10.3
8.74
10.5
9.75
9.57
10.2
8.77
9.49
0.310
0.002
0.002
0.001
0.001
0.001
0.091
3,53
Hg Flow Rate, mg/hr
Inlet Outlet
43,100
36,400
43,600
40,000
39,300
41,000
35,400
37,700
1,120
7
7
4
4
3
310
12,000
Hg Control
Efficiency, %
97.0
99.9
99.9
99.9
99.9
99.9
99.0
62.8
Chlorine Produc.
Rate, tons/hr
8.09
8.09
8.09
8.09
8.09
8.09
8.09
8.09
Average Hg Collection Efficiency * 94.8%
Total Hg in Feed Stream During Cycle = 946 g
Total Maximum Hg Discharge During Cycle - 40 g
NOTE: Nm is defined as one normal cubic meter, i.e., one cubic meter at standard conditions
760 mm Hg and 21.1'C.
Average Hg Collection Efficiency During Cycle has been calculated as the mean value of the
individual efficiencies for each sampling period. This value is not based on the total Hg
in and out of the sieve during an adsorption cycle.
-------
stream and the regeneration stream. The regeneration hydrogen
stream is an integral part of the PuraSiv Hg process and is
eventually recycled to the inlet of the process. The outlet stream
is subsequently divided into recycle (but not regeneration) gas,
boiler combustion gas and/or vent gas to the atmosphere. Although
actual plant operating conditions currently provide for recycling,
approximately 28% of the total outlet hydrogen, the mercury emis-
sions for a given cycle are calculated on the basis of maximum
possible emission from the adsorption unit, i.e., no recycle hydro-
gen requirements, no hydrogen combustion, but instead venting of
all hydrogen (except that required for regeneration) directly to
the atmosphere. This procedure permits a more realistic evalua-
tion of the PuraSiv Hg performance in lieu of a determination
of the actual mercury emission from the plant.
Obviously, the outlet stream and regeneration stream have identical
mercury concentrations upon initial discharge from the adsorber
as one stream. This permits calculation of the mercury removal
efficiency on the basis of adsorber inlet and outlet mercury con-
centrations. Collection efficiency could just as well be calculated
by a mercury mass balance on the adsorption unit, being certain to
include the mercury emitted from the adsorber but recycled through
the regeneration gas stream.
Measured inlet mercury concentrations ranged consistently from 6
to 10 mg/Nm with the highest concentration measured being 10.5
3
mg/Nm . As shown in Appendix C, at measured stack conditions of
1,108 mm Hg and 21°C, the equilibrium saturation concentration of
3
mercury is calculated to be 14.4 mg/m . This saturation concentra-
tion when corrected to dry standard conditions is equivalent to 10.9
mg/Nm .
The results tabulated in Tables V-l through V-5 reveal a definite
cyclic pattern of the mercury content in the adsorber outlet stream.
V-7
-------
During the initial 2 to 4 hours of the adsorption cycles, mercury
concentrations approaching 0.5 mg/Nm were measured, after which
3
outlet mercury concentrations at the level of 1 to 2 ug/Nm were
consistently determined. Finally, in the late hours of the adsorp-
tion cycle (approximately the 20th hour), "breakthrough" occurs as
the adsorbent becomes mercury-laden causing sharp rises in the
outlet mercury concentration. Approximately 85% of the mercury
discharged from the adsorber during the total cycle occurred during
the final four hours of the cycle.
These higher outlet mercury concentrations at the beginning and
end of each cycle occur because the plant is currently operating
the adsorber system at approximately 145% of design capacity.
This additional flow to the adsorbent is due to plant modifications
of the blower which increased its capacity in anticipation of
future plant expansion. Although plant production has not yet been
increased, the quantity of mercury vapor charged to the adsorbent
is directly proportional to the inlet hydrogen flow rate, as
explained in Section III. Therefore, the total weight of mercury
vapor charged to the adsorbent is roughly equivalent to the
amount expected from a 280 tons Cl»/day mercury-cell process.
Because the system was not designed to remove this amount of
mercury over a 24-hour adsorption cycle, the adsorbent mercury reten-
tion capacity is realized prior to the end of the cycle.
Toward the end of a regeneration cycle, the regeneration hydrogen
gas is no longer heated but is still passed down through the adsorbent
bed to cool the adsorbent for the upcoming adsorption cycle. If
"breakthrough" has occurred in the adsorption cycle, this results
in deposition of mercury from the "clean" regeneration gas upon
the adsorbent of the recently regenerated adsorption bed. Once the
regenerated bed enters adsorption service, mercury from the inlet
stream is entrapped near the bottom portion of the bed and then
V-8
-------
near the top of the bed this "cleaned" hydrogen gas encounters
the mercury-laden adsorbent caused by previously cooling the bed
with "dirty" hydrogen due to early "breakthrough." The mercury
is subsequently stripped from the adsorbent during the initial hours
of the adsorption cycle. This explains the appearance of mercury
in the outlet stream during the start of the adsorption cycle.
It is doubtful if "breakthrough" would occur when operating the
adsorption process at design flow rates and, consequently, mercury
in the adsorber outlet during cycle start would also not occur.
Due to an accidental plant shutdown on the evening of September 26,
Adsorber A was in adsorption service for a total duration of 22.5
hours. This fact explains the relatively low mercury content (0.067
3 3
mg/Nm compared to the usual 3.5 mg/Nm ) in the adsorber outlet at
the end of Cycle No. 3. Although "breakthrough" had commenced,
the rapid escalation in the mercury discharge during cycle end
was avoidable when Adsorber A was switched to regeneration on its
routine time schedule. In addition, this occurrence meant that
Adsorber B was not cooled with "dirty" regeneration gas and thus
low (0.039 mg/Nm ) mercury concentrations in the outlet stream
were measured at the start of Cycle No. 4.
If this particular PuraSiv Hg process were to be operated at
or near the actual design capacity, early "breakthrough" and the
subsequent rapid rise in mercury emissions near the end of a cycle
would be avoided. Furthermore, the "wash out" of mercury during
the beginning of an adsorption cycle would not occur since the
previously regenerated bed would not have been cooled with "dirty"
gas from the adsorber outlet. Therefore, a realistic evaluation
of the actual performance capability of this process should exclude
the mercury emissions occurring at these two periods in the cycle.
A true representation of the outlet mercury concentrations attain-
able with this system is reflected after initial "wash out" but
V-9
-------
prior to excessive "breakthrough" when the mercury concentration
3
is consistently less than 10 yg/Nm . Examination of these results
reveals a mean mercury concentration in the adsorber outlet of
2.0 yg/Nm with a standard deviation of the mean equal to 0.35
yg/Nm . These values result in a 95% confidence limit of the mean
of 2.0 + 0.73 yg/Nm3.
Should the PuraSiv Hg unit be operated at 100% of design capa-
3 3
city (49.6 Nm /min inlet, 9.1 Nm /min regeneration, no recycle),
it is estimated that the outlet mercury concentration would average
3
2.0 yg/Nm throughout the duration of the 24-hour adsorption cycle
at an overall collection efficiency of 99.9%. This would result
in a total 24-hour mercury discharge of 0.12 g/day.
Any determination of emission factors for the PuraSiv Hg unit
would be misleading as well as impractical since the mercury
charged to the adsorber is only indirectly related to the actual
rate of chlorine production. In fact, examination of Tables V-l
through V-5 will show that at times mercury charged to the adsorber
would increase when chlorine production decreased and vice versa.
The amount of mercury that enters the adsorber is actually a func-
tion of the production rate, the recycle rate, and the operation
of the equipment between the decomposer and the adsorption
unit, i.e., the primary cooler, the York demister, the centrifugal
knock-out pot, the secondary cooler and the Brink demister. Granted,
a "slug" of mercury through this system would most likely result
in a higher instantaneous mercury rate to the adsorber, but with
the consistent efficiency of the intermediate control devices,
the adsorber should be expected to experience more or less the
same inlet mercury loadings regardless of the normally small varia-
tions in chlorine production rates.
A graphical presentation of the mercury discharged from the adsorber
outlet during the different hours of the adsorption cycle, i.e.,
mg Hg/hr versus cycle hour, would clearly show the cyclic pattern
V-10
-------
of the mercury emission rate during the adsorption cycle. However,
the magnitude of the range of mercury discharge rates (<10 to
>10,000 mg/hr) prevents the construction of any meaningful curve
with sufficient detail. Figure V-l represents an oversimplified
version of how such a graph would appear.
Naturally, the nature of the test program did not produce instan-
taneous mercury mass flow rates required to graphically construct
a curve such as Figure V-l. Again, an oversimplification of how
the test results would appear, if presented graphically, is shown
in Figure V-2. The graph would consist of 8 separate values of
the mercury flow rate, each existing for a period of two hours.
Between successive mercury flow rates is a void data space repre-
senting the hour interval in which no sampling was conducted while
sample recovery and sample train preparation were performed. Since
the production process was quite steady throughout and the opera-
ting parameters of the PuraSiv Hg unit were fixed, any changes
in mercury content of the outlet appear to be unidirectional, i.e.,
a decreasing mercury content declined until it leveled off, after
which a rising mercury content at breakthrough continued to esca-
late rapidly. Therefore, for a first approximation, it would seem
reasonable that a mercury flow rate at some time when no sampling
was being done can be estimated by linear interpolation. Using
this assumption for linear interpolation, addition of the dotted
lines in Figure V-2, now completes the description of an adsorp-
tion "curve" generated from the actual test results. The area
under this "curve" represents the total amount of mercury discharged
over a given cycle and this is easily determined through graphical
integration. A sample graphical integration procedure is located
in Appendix D. Likewise, this same manner of curve construction
should be applicable to the mercury flow rates in the inlet stream.
The same type of graphical integration was employed to calculate
the total mercury charged to an adsorber during a given cycle.
V-ll
-------
FIGURE V-l, SKETCH OF TYPICAL ADSORPTION CURVE
OJ
i-
d)
JD
S-
o
I/)
-o
-------
FIGURE V-2, SKETCH OF ADSORPTION CURVE GENERATED FROM
TEST DATA - CYCLE NO, 5
S_
.c
cr.
i.
01
.p
6
TD
IO
0)
«
o;
o
d
O
S-
o>
H3.UUU
£ 40,000
* t
35,000
12,000
11,000
10,000
9,000
(J * W\J
tJ 7,000
o
6,000
5,000
4,000
3,000
2,000
1,000
(
i i i i i
« *
-
. -
- 4
I
-
~
'
-
__
.
- . ~
-
-
_
_ . _
*w
3 4 8 12 16 20 24
Adsorption Cycle Time, Hrs.
V-13
-------
The results for each stream during each of the five cycles is
indicated at the bottom of Tables V-l through V-5, respectively.
Additional summarized results for each sample run conducted on
the adsorber inlet stream are presented in Tables V-6 and V-7.
Further detailed results and data for each of these runs are
contained on individual computer print-out sheets located in
Appendix E.
Upon inspection of Tables V-6 and V-7, the initial impulse is
to reject all inlet sampling runs on the grounds of unacceptable
anisokinetic conditions, i.e., the isokinetic values for all tests
range from 76% to 81%. A closer evaluation of the actual test
conditions indicates that the isokinetic sampling rate of the
inlet stream is not a prime influence on the results obtained.
The consistently low isokinetic values were the result of three
problems, all of which were related to the fact that the sampled
gas stream was hydrogen rather than air. Accurate estimation
of the gas stream moisture content is much more important when
the gas is hydrogen than it is when the gas is air. Small errors
in moisture content estimation result in much larger errors in
the estimated gas stream molecular weight. To establish the
nomograph "C" factor, an estimated stack gas moisture content of
3% by volume was obtained from a psychrometric chart. Because
the stack static pressure was about 250 mm Hg higher than standard
pressure, the value obtained from the psychrometric chart was too
high. The actual saturation moisture content at stack conditions
was approximately 2%. In an air system, an insignificant sampling
rate error would result from using a nomograph set for 3% mois-
ture in a gas stream containing only 2%. In the hydrogen system,
however, use of the 3% moisture content estimation resulted in
sampling rates that ranged from 3% to 4% lower than would have
been obtained if the nomograph had been set up for 2% moisture.
V-1A
-------
Table V-6. MERCURY INLET DATA SUMMARY (METRIC)
SCA-1
SCA-3
SCA-5
SCA-7
SCA-9
SCA-11
SCA-13
SCA-1 5
SCA-1 7
SCA-19
SCA-21
SCA-23
CA-25
SCA-27
SCA-29
SCA-31
SCA-33
SCA-35
SCA-37
SCA-39
DRY GAS
VOLUME
NH3
2.512
2.483
2.595
2.CT3
2.642
2.623
2.473
2.560
2.494
2.528
2.5i9
2.578
2.610
2.585
2.532
2.485
2.492
1 .538
1.655
2.514
MlISTUKf
%
1.64
1.62
1.55
1.37
1.53
1.54
1.49
1.72
1.74
1.63
I.t6
1.36
1.85
1.90
1.74
2. 19
1.73
1.16
1.75
1.81
TTTAL
CATCH
MG
22.36
19.85
20.23
19.60
19.75
24.97
24.97
20.37
20. DO
20.50
20.13
18.11
18.24
18.02
19.09
20.48
13.82
13.45
11.50
17.27
PLOW
RATF
NV3/HIN
70.16
69.61
71.05
72.51
71.00
70.. 89
69.67
68.79
67.50
68.86
68.61
67.93
68.10
67.85
67.81
65.95
67.26
69.12
67.38
67.05
STACK
Tr-yp
C
22.2
22.2
22.2
22.2
22.2
22.2
21. 7
21.7
21.7
21. 7
21.7
21.7.
21.7
21. 7
22.2
21. 1
22.2
21.7
21.3
21.7
150-
KIN-TIC
*
76. 4
76. 1
77.0
79.3
79.4
79.0
75.7
79.4
78.8
78.4
79.0
81. 3
31.8
81.3
79.7
30.4
7°. 1
76.5
78.6
79.1
C1NC
MG/NM3
8.884
7.978
7.780
7.260
7.459
9.5JO
10.076
7.939
8.003
8.092
7.911
7.010
6.973
6.957
7.525
8.225
5.535
8.727
6.935
6.856
':f.
-------
Table V-6 (continued). MERCURY INLET DATA SUMMARY (METRIC)
SCA-43
SCA-45
SCA-47
SCA-49
SCA-51
SCA-53
SCA-55
SCA-57
SCA-59
SCA-61
SCA-63
SCA-65
SCA-67
SCA-69
SCA-71
SCA-73
SCA-75
SCA-77
SCA-79
SCA-81
DRY GAS
VOLUME
NM3
2.580
2.476
2.431
2.514
2.533
2.527
2.581
2.660
2.642
2.628
2.617
2.574
2.580
2.554
2.557
2.587
2.494
2.510
2.446
2.420
MOISTURE
%
1.71
1.94
1.88
1.72
1.59
1.70
1.20
1.02
1.51
1.89
1.74
1.59
1.67
1.66
1,77
2.01
1.64
2.00
1.64
1.89
TOTAL
CATCH
M6
17.27
15.65
13.11
19.83
19.88
15.28
18.73
-20.40
17.27
18.11
20.37
22.85
26.71
22.36
26.90
25.26
23,93
25.65
21,50
23,01
FLOW
KATE
NM3/MIN
67.97
67.04
67.11
68.40
68.64
68,74
71.68
74,05
72.57
71.05
71.68
7C.58
6S.49
69.43
69.17
68.36
68.50
67.07
67.29
66.26
STACK
TEMP *
C
21.7
21.1
21.1
21.1
21.7
21.7
21.1
21.1
21.1
21.1
21.6
21.7
21.7
21.7
21.7
21.7
21.9
21.7
21.7
21.7
ISO-
CINET 1C
?
81.0
73.8
78.9
78.4
73.7
78.5
76.8
76.6
77.7
78.9
77.9
77.8
79.2
78.5
78.9
80.8
77.7
79.9
77.6
77.9
CONC
KG/NM3
6.681
6.3C8
7.283
7.872
7.832
6.033
7.241
7.652
6.522
6.876
7.768
8.858
10.331
8.736
10.498
9.750
9.573
10.193
8.772
9.490
EMISSION
U/UAY
653.32
606.90
703.70
775.28
773.99
597.14
747.28
816.00
631.48
703.39
801.68
900.17
1033.61
873.34
1045.57
959.67
944.22
934.31
849.94
905.36
-------
Table V-7. MERCURY INLET DATA SUMMARY (ENGLISH)
I
M
^J
CODE
SCA-1
SCA-3
SCA-5
SCA-7
SCA-9
SCA-H
SCA-1 3
SCA-1 5
SCA-1 7
SCA-19
SCA-21
SCA-23
CA-25
SCA-27
SCA-29
SCA-31
SCA-33
SCA-35
SCA-3 7
SCA-3 9
SCA-41
DRY GAS
VOLUME
SCF
88.701
87.686
91.633
55.113
93.290
92.630
87.33J
90.418
88.064
89.283
89. '66
91.047
52.186
91.281
89.405
87.749
87.993
54.310
58.438
88.768
93.215
MJISTUP;
«
1.64
1.62
1.55
1.37
1.53
1.54
1.49
1.72
1.74
1.63
1.66
1.36
1.85
1.93
1. 74
2.19
1.73
1.16
1.75
1.81
1.88
TOTAL
CiTCh
MG
22.36
19.85
20.23
19.60
19.75.
24.97
24.97
20.37
20.00
20.50
20.13
IP. 11
18.24
18.02
19.09
20.48
13.82
13.45
11.50
17.27
15.51
FLCW I
PATK
SCFM
2478.
2458.
2509.
2561.
2508.
2503.
2460.
2429.
2384.
2432.
2423.
2399.
2405.
2396.
2395.
2329.
2375.
2441.
2379.
2396.
2449.
STACK
T - M P 1
f
72.
72.
72.
72.
72.
72.
71.
71.
71.
71.
71.
71.
71.
71.
72.
70.
72.
71.
70.
71.
71.
IS'.'-
-------
Table V-7 (continued). MERCURY INLET DATA SUMMARY (ENGLISH)
H1
CD
CODE
SCA-43
SCA-45
SCA-47
SCA-49
SCA-51
SCA-53
SCA-55
SCA-57
SCA-59
SCA-61
SCA-63
SCA-65
SCA-67
SCA-69
SCA-71
SCA-73
SCA-75
SCA-77
SCA-79
SCA-81
DRY GAS
VOLUME
SCF
91.100
87.429
87.63C
88.774
89.452
89.255
91.154
93.929
93.310
92.816
92.411
90.909
91.113
90.198
90,299
91.370
88.090
88.636
86.369
85.447
MOISTURE
%
1.71
1.94
. 1.88
1.72
1.59
1.70
1.20
1.02
1.51
1.89
1.74
1.59
1.67
1.66
1.77
2.01
1.64
2.00
1.64
1.89
TOTAL
CATCh
MG
17.27
15.65
18.11
19.83
19.88
15.28
18.73
20.40
17.27
18.11
20.37
22.85
26.71
22.36
26.90
25.28
23.93
25.65
21. 5C
23.01
FLOW STACK ISO-
RATE TEMP KINETIC
SCFM F *
24C1.
2368.
237C.
2416.
2424.
2428.
2531.
2615.
2563.
2509.
2531.
2493.
2454.
2452.
2443.
2414.
2419.
236S.
2376.
2340.
71.
70.
70.
70.
71.
71.
70.
70.
70.
70.
71.
71.
71.
71.
71.
71.
71.
71.
71.
71.
81.0
78.8
78.9
78.4
78.7
78.5
76.8
76.6
77.7
78.9
77.9
77.8
79.2
78.5
78.9
80.8
77.7
79.9
77.6
77.9
CONC EMISSION
GR/SCF LS/OAY
O.OC3
0.003
0.003
0.003
0.003
0.003
0-003
0.004
0.003
0.003
0.003
0.004
O.OC5
0.004
0.005
0.004
0.004
0.004
0.004
0.004
1.44
1.34
1.55
1.71
1.71
1.32
1.65
1.80
1.-50
1.55
1.77
1.98
2.28
1.93
2,31
2.12
2.08
2.17
1.87
2.00
-------
The low molecular weight of hydrogen may result in the orifice
meter being operated in a nonlinear region. To determine the
effect this may have had on the isokinetics, a check of the
orifice meter calibration was made using the field sampling data.
At the orifice meter pressure differentials encountered during
field testing, the actual sampling rates as measured by the dry
gas meter ranged from 1% to 3% lower than predicted from pretest
calibration.
The combined effect of the inaccurate moisture estimate and the
shift in the orifice meter curve account for approximately one-third
of the total isokinetic error. Discussions with EPA personnel
provided the answer to the balance of the error. The "C" factor
equation and the modified procedure for using the operating
nomograph specified by Method 102 are not exact, but are approx-
imations. Under the particular conditions of these tests, a
sampling rate error ranging from 15% to 18% was introduced by
using these procedures. It is understood that the present
procedure is under study and a revision will be forthcoming.
The total system operated with very few fluctuations because the
PuraSiv Hg operating parameters were set by the plant and automatically
controlled throughout the test period. In view of these facts,
in addition to the highly turbulent flow at the inlet sampling
location, the existence of equilibrium conditions at that point
of the process would appear probable. It has previously been
shown that all inlet mercury concentrations measured were below
the equilibrium saturation concentration of mercury for the test
conditions. Hence, it is doubtful that any particulate mercury
did indeed exist in the inlet gas stream. This being the case,
the use of isokinetic sampling is not a necessary test requirement.
Proportional sampling is acceptable for such conditions.
However, while the existence of particulate mercury in the inlet
gas stream was unlikely, the case was re-examined based on the
V-19
-------
FIGURE V-3, ERRORS DUE TO ANISOKINETIC CONDITIONS
(SMALL PARTICLES)
(O
o
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.4
0.6
pmr (Smal1}
c
0.8
1.0
1.2
1.4
Ratio of Actual Sampling Rate to
Theoretical Isokinetic Sampling Rate
1.6
V-20
-------
assumption that particulate mercury dic[ exist at the inlet samp-
ling location. This occurrence would be expected to be in the
form of a fine liquid mist, i.e., very small particles (<10 microns)
Brink mist eliminators provide essentially 100% collection effi-
ciency on all particles larger than 3 microns in diameter. In
addition, most Brink demisters are designed to remove 94% to 99.7%
of all particles smaller than 3 microns in diameter. For this
reason, it can be assumed that any mercury particles present
would have to be smaller than 3 microns. Figure V-3 graphically
illustrates the effect of anisokinetic sampling of streams
*
containing only small particles. Regardless of the isokinetic
sampling ratio, when the pollutant mass rate is determined by
the sample concentration method (pmr ), the calculated mass rate
and true mass rate will be identical. On the other hand, when
determining the pollutant mass rate by the ratio-of-area method
of calculation (pmr ), the calculated values under the anisoki-
3
netic conditions experienced in this test program would be approx-
imately 20% less than the actual concentrations. Therefore,
since the sample concentration method of calculation was used
exclusively in this test project, even if mercury mist did exist
in the inlet stream, the underisokinetic sampling rates would not
influence the final measured mercury content. For additional
explanation of the two methods of calculation, see Appendix A.
Examination of Tables V-l through V-5 shows that the reported
3
inlet mercury concentrations (6 to 10 mg/Nm ) were consistently
below the estimated inlet equilibrium mercury saturation concen-
3
tration corrected to standard conditions (10.9 mg/Nm ). Although
the presence of liquid mercury particles at the inlet sampling
"Method of Interpreting Stack Sampling Data," Smith, W.S., et al,
National Air Pollution Control Association, Public Health Service,
U.S. Department of Health, Education and Welfare.
V-21
-------
site was not expected, similarly, a condition of under-saturation
of mercury vapor at that point was also unexpected and is ques-
tionable. There is no apparent reason to believe that the esti-
mated accuracies of the sampling train and of the analytical
procedure are unrealistic. However, there does exist the definite
possibility that the temperature and/or pressure recorded for the
inlet sampling site were inaccurate and, thereby, caused a bias
in the estimated inlet mercury saturation concentration.
The most likely explanation for this apparent discrepancy is that
the temperature at the inlet sampling site was probably not the
temperature as recorded from the existing thermometer which was
installed near the outlet of the secondary cooler. Should the
temperature at the inlet sampling site have been 18°C instead of
21°C, as read from that thermometer, measured mercury concentra-
tions would coincide with an inlet stream saturated with mercury
vapor. A change of 3 C in the inlet temperature would not alter
the test results for mercury concentrations as reported at dry
standard conditions. On the other hand, this same temperature
change reduces the calculated equilibrium mercury concentration
by 20%. Averaging the inlet mercury concentrations of Cycle Nos.
1, 2, 4, and 5 (Cycle No. 3 excluded because of the plant shut-
2
down) yields an average measured concentration of 8.4 mg/Nm .
The inlet equilibrium saturation concentration, corrected to
dry standard conditions, corresponding to stack conditions of
18 C and 5 psig, is 8.4 mg/Nm . Therefore, it is suggested that
the reported estimated inlet equilibrium concentration may have
been biased by a process instrument whose accuracy and calibra-
tion are unknown.
V-22
-------
TECHNICAL REPORT
(Please read luiLnjctiuns on the reverse
DATA
before complctingl
1. REPORT NO,
EPA-650/2-75-026-a, -b
3. RECIPIENT'S ACCESSIOI*NO.
I. TITLE AND SUBTITLE
Testing of a Molecular Sieve Used To Control
Mercury Emission from a Chlor-Alkali Plant,
Volume I and Volume II (Appendices)
5. REPORT DATE
March 1975
6. PERFORMING ORGANIZATION CODE
'. AUTHORIS]
8. PERFORMING ORGANIZATION REPORT NO,
John T. Chehaske and John R. Cline
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Air Pollution Control Department
Engineering-Science, Inc.
7903 Westpark Drive, McLean, VA 22101
10. PROGRAM ELEMENT NO.
1AB015; ROAP 21AFA-106
11. CONTRACT/GRANT NO.
68-02-1406, Task 3
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
NERC-RTP, Control Systems Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final Task: 9/74-1/75
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT.
The report gives results of performance testing for mercury emission
control of a Union Carbide PuraSiv Hg adsorbent unit currently used to control
mercury emissions from the by-product hydrogen stream of the mercury-cell
chlor-alkali production facility of Sobin Chlor-Alkali in Orrington, Maine. Adsorber
inlet and outlet streams were sampled simultaneously during five consecutive 24-hour
adsorption cycles, eight sampling runs per cycle, in accordance with a modified
version of the EPA Reference Method 102. Inlet mercury concentrations were stable
at 6 to 10 mg/normal cu m. Outlet mercury concentrations of 0. 5 and 3. 5 mg/normal
cu m were approached at the beginning and end of each adsorption cycle, respectively.
This occurrence was due to the PuraSiv unit's operation at nearly 150% of design capa-
city of 49. 6 normal cu m/min. If operated at design capacity, the estimated outlet
mercury concentrations from the PuraSiv Hg unit would average 2.0 micrograms/
normal cu m throughout each adsorption cycle, representing a 99.9% collection
efficiency.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Air Pollution
Absorbers (Materials)
Mercury (Metal)
Performance Tests
Air Pollution Control
Stationary Sources
Molecular Sieves
Chlor-Alkali Plants
PuraSiv Hg Unit
13B
11G
07B
14B
8. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report}
Unclassified
70 SECURITY CLASS (This page/
Unclassified
21. NO. OF PAGES
Volume T - 62
Veil limp IT -
22. PRICE
EPA Form 2220-1 (9-73)
V-23
-------
ENVIRONMENTAL PROTECTION AGENCY
Technical Publications Branch
Offic* of Administration
Triangle Park. N.C. 27711
POSTAGE AND FEES PAID
ENVIRONMENTAL PROTECTION AGENCY
EPA - 335
OFFICIAL BUSINESS
AN EQUAL OPPORTUNITY EMPLOYER
Return this sheet if you do NOT wish to receive this material I I.
or if change of address is needed f I. (Indicate change, including
ZIP code.)
PUBLICATION EPA-650/2-75-026-Q
------- |