vvEPA
United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park NC 27711
EMB Report 80-BYC-8
March 1981
Air
Benzene
Coke Oven By-Product
Plants
Emission Test Report
U.S. Steel
Fairless Hills,
Pennsylvania
-------
SET 1957 04 1280
BENZENE SAMPLING PROGRAM
AT COKE BY-PRODUCT RECOVERY PLANTS:
UNITED STATES STEEL CORPORATION,
FAIRLESS HILLS, PENNSYLVANIA
EPA Contract 68-02-2813
Work Assignment 48
ESED Project No. 74/4j
Prepared For:
Mr. Daniel Bivins
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Emission Measurement Branch, ESED, MD-13
Research Triangle Park, North Carolina 27711
March 1981
SCOTT ENVIRONMENTAL SERVICES.
A Division Of
SCOTT ENVIRONMENTAL TECHNOLOGY, INC.
Plumsteadville, Pennsylvania 18949
Scott Environmental Technology Inc.
-------
1957-04-1280 "Page TC--1
TABLE OF CONTENTS
Page No.
1.0 INTRODUCTION 1-1
2.0 SUMMARY OF RESULTS 2-1
3.0 DISCUSSION OF RESULTS 3-1
3.1 COOLING TOWER 3-1
3.2 WASH OIL DECANTER 3-2
3.3 TAR DEHYDRATORS 3-5
4.0 PROCESS DESCRIPTION " . 4-1
5.0 FIELD SAMPLING AND ANALYSIS METHODOLOGY 5-1
5.1 DETERMINATION OF BENZENE FROM
STATIONARY SOURCES: EPA METHOD 110
AND MODIFICATIONS 5-1
5.2 SAMPLE HANDLING 5-4
5.3 FIELD ANALYSIS 5-4
6.0 FIELD SAMPLING 6-1
6.1 COOLING TOWER 6-1
6.2 WASH OIL DECANTER 6-3
6.3 TAR DEHYDRATORS 6-3
7.0 LABORATORY SAMPLE ANALYSIS 7-1
7.1 SAMPLE PREPARATION 7-1
7.2 PURGE AND TRAP PROCEDURE FOR EXTRAC-
TION OF BENZENE FROM LIQUID PHASE
TO GASEOUS PHASE 7-2
7.3 GAS CHROMATOGRAPH 7-4
8.0 QUALITY CONTROL AND QUALITY ASSURANCE 8-1
8.1 FIELD ANALYSIS PROCEDURES 8-1
8.2 PROCEDURES FOR ANALYSIS OF PROCESS
LIQUIDS 8-2
Scott Environmental Technology Inc.
-------
SET 1957 04 1280 Page 1-1
1.0 INTRODUCTION
Scott Environmental Services, a division of Scott: Environmental
Technology, Inc. conducted a sampling program at United States Steel
Corporation in Fairless Hills, Pennsylvania to determine benzene emissions
from four sources in the coke by-products recovery plant. The work was
performed for the United States Environmental Protection Agency, Emission
Measurement Branch, under Contract Number 68-02-2813, Work Assignment 48.
The Fairless Works was the fourth of seven coke by-product plants visited to
collect data for a possible National Emission Standard for Hazardous: Air
Pollutants for benzene.
Sampling was conducted at U.S. Steel from September 8th through
llth, 1980. Air and liquid samples for benzene analysis were collected
from the cooling tower direct water final cooler, wash oil decanter , and
from the two tar dehydrators.
Scott Environmental Technology Inc.
-------
SET 1957 04 1280 Page 2-1
2.0 SUMMARY OF RESULTS
Benzene Emission Rate
Process Ib/hr kg/hr
Cooling tower-direct water final cooler 70.2 31.3
Wash oil decanter 0.87 0.39
Tar dehydrator-west tank 6.95 '3.15
Tar dehydrator-east tank 2.45 1.11
Scott Environmental Technology Inc.
-------
SET 1957 04 1280 . Page 3-1
3.0 DISCUSSION OF RESULTS
3.1 COOLING TOWER
Hot water from the direct water final cooler is pumped to a hot
well, where it is circulated over a 60-foot high atmospheric cooling tower.
The tower has a 17 foot diameter fan on top for pulling air countercurrent
to the falling water to effect the cooling. The tower also effectively
acts as a stripper for benzene contained in the hot well water.
Four EPA Method 110 runs were done on the cooling tower on
September 9th and llth, 1980. Table 3-1 presents the results of the tests.
The third run was voided because the concentration in the collected sample
was over twice as high as in the first two, indicating a contaminated
sampling bag was used. When this was verified by the bag log book entries,
the run was declared void and a fourth test was run. The results of the
third test are included in Table 3-1 for comparisons of temperature and
flow rate.
A 24-point sampling and velocity traverse was made across two
diameters of the 17 foot fan shroud to obtain an integrated sample. The
average result for the three good tests was 70.2 Ib/hr. All stack flowrates
were corrected to the average conditions at which the benzene concentrations?
were measured in the Tedlar bags; assumed to be saturated at 68°F and
29.92 in. HE. (2*5% moisture). Example calculations are shown in Appendix A.
The fourth test yielded higher mass emission rates than the first
two tests although the benzene concentration was similar, due to hieher
measured stack velocities. This could be a result of the high winds on
that day, which affected the vane anemometer used to measure velocity,
although it was also somewhat windy on the first day, as noted on the field
data sheets in Appendix A.
Scott Environmental Techndosy Inc
-------
SET 1957 04 1280 Page 3-2
The stack velocity data of all four tests show that one quadrant
of the fan shroud had consistently higher velocities than the others. This
could be a consequence of localized wind effects, since the wind was pre-
dominantly from that side of the tower, or could be the result of a varia-
tion in the internal packing structure of the tower which could cause a
higher air flow on one side.
Liquid samples were dipped from the hot well and cold well after
each of the first three test runs, with temperatures of approximately 95° and
78°F respectively. As shown in Table 3-1, the first hot well sample contained
much less benzene than the other two. Replicate analyses verified this re-
sult. This apparently indicates a great deal of variability in the feed
water benzene concentration, due to stratification or incomplete mixing in
the hot well.
3.2 WASH OIL DECANTER
The wash oil decanter separates water from the wash oil before it
enters the wash oil cooler by settling out the water due to density differences.
The decanter is vented to the atmosphere, and benzene contained in the hot
wash oil could be directly emitted from this process.
Triplicate Method 110 tests were run on the decanter. The data
shown in Table 3-2 indicate an average emission rate of 0.87 Ib/hr benzene.
The moisture content of the stream was high; the average for the three runs
was 60.5% moisture, as determined from the volume of water collected in the
water trap. (See sample calculations in Appendix A). Three moisture
determinations were run using silica gel tubes, dry gas meter and pump
Scott Environmental Technology Inc.
-------
©
TABLE 3-1
yn COOLING TOWER DATA SUMMARY
* Process: Cooling Tower - direct water final cooler Stack
g Plant: U. S. Steel, Fairless Hills, Pa. Stack
3
2. Stack
^ Run Sample Temp.
5* No. Date Period ( F)
8. 1 9/9/80 1120-1245 82
"1 2 9/9/80 1417-1530 85
3 9/9/80 1547-1700 87
4 9/11/80 1525-1650 80
Standard conditions: Saturated at
1 '
Liquid Sample Data
Sample Location
Cooling Tower - hot well
Cooling Tower - cold well
Barometric Stack
Pressure Velocity
(in.Hg) (ft/min.)
30.12 1350
30.12 1170
30.12 1230
29.99 1680
68°F, 29.92^in. Hg
Date
9/9/80
9/9/80
Flowrate
Stack
Conditions
(ACFM)
306,000
266,000
279,000
381,000
Time
1410
1535
1715
1410
1535
1715
Diameter:
Area: 227
Flowrate
Standard
Conditions
(SCFM)
297000
255,000
265,000
369,000
Sample
Temp. °F
93
96
97
78
78
79
H
17 ft. H
2 Oi
ft. v
Benzene ,L
Benzene Emission £J
Concentration Rate ° ;
(ppm) (Ib/hr.)
18.77 67.6
19.80 61.2
44.4* — ;
i
18.30 81.9 '
*Contaminated Ave. 70.2
Tedlar Bag
Benzene Concentration
(ppm by weight)
8.2
48.2
48.8 . ^
oj
OQ
i.o "•
i.o Y
0.8 w i
-------
Process: Wash Oil Decanter
TABLE 3-2
WASH OIL DECANTER DATA SUMMARY
Stack Diameter: 8"
CO'
w
VO
Plant: U. S. Steel, Fairless Hills, PA
Stack Barometric
Run Sample Temp. Pressure
No. Date Period (°F) (in. Hg)
1 9/10/80 1410-1440 193 29.92
2 9/10/80 1508-1538 193 29.92
3 9/10/80 1555-1625 193 29.92
Standard conditions: Saturated at 68°F, 29.
Liquid Sample Data
Sample Location
Wash Oil Decanter -
dipped from hatchway
Stack
Velocity
(ft/min.)
795
820
810
92 in. Hg.
Date
9/10/80
9/10/80
9/10/80
Flowrate
Stack
Conditions
(ACFM)
280
290
280
Time
1640
1640
1640
Stack Area:
Flowrate
Standard
Conditions
(SCFM)
96
67
110
Sample
Temp.
(°F)
205
205
205
0.35 Ft2
Benzene
Benzene Emission
Concentration Rate
(ppm) (Ib/hr.)
706 0.82
1052 0.85
694 0.95
Ave. 0.87
Benzene
Concentration
(ppm by weight)
42.2
40.3
50.3
r
o
r
t-*
CO-
o
pi
00
ID
Ave. 44.3
U)
-------
SET 1957-04-1280 Page 3-5
as described in Section 6.2, as an additional check to.insure the accuracy
of the Method 110 data. The average for these three runs was 59.1%
moisture, which confirmed the accuracy of our previous findings. These
calculations can also be found in Appendix A.
Triplicate liquid samples were dipped from the hatchway next to
the sampling vent immediately after the three tests were run. The liquid
temperature was 205 F and the average benzene concentration in the
samples was 44.3 ppm by weight.
3.3 TAR DEHYDRATORS
Two tanks in series heat the tar from the tar decanter to drive
off the entrained water. Benzene contained in the tar is also potentially
released with the water.
Method 110 was modified by bubbling the sample stream through
propylene carbonate to remove the naphthalene, as described in Section 6.3.
Significant amounts of benzene were found in the propylene carbonate
solution, and these amounts were added to the total benzene found in the
collected bag samples to determine mass emissions from this source.
As shown in Table 3-3, the mass emission rate from the west tank
(1st in series) ranged from 1 to 15 Ib/hr. with an average of 6.95 Ib/hr.
and the east tank emission rate varied from 1 to 4 Ib/hr. with an average
of 2.45 Ib/hr. There was great variability between velocity readings,
not only between runs but from reading to reading on each single run, as
can be seen from the field data sheets in Appendix B. Also the benzene
concentration varied from about 700 ppm to over 3000 ppm between the dif-
ferent runs. Generally the benzene concentration increased when stack
temperature increased.
Scott Environmental Technology Inc.
-------
©
TABLE 3-3
J? TAR DEHYDRATORS DATA SUMMARY
m Process: Tar dehydrators - east & west tanks Stack Diameter: 8"
3 Plant: U. S. Steel, ?airless Hills, PA Stack Area: 0.349
£ Flowrate Flowrate
_j Stack Barometric Stack
n. Run Sample Temp. Pressure
g No. Date Period (°F) (in.Hg)
^ WEST TANK (1st in series)
^ 1 9/11/80 1015-1045 174 29.99
2 9/11/80 1350-1420 182 29.99
3 9/11/80 1730-1800 166 29.99
EAST TANK (2nd in series)
1 9/11/80 1100-1130 174 . 29.99
2 9/11/80 1345-1415 142 29.99
3 9/11/80 1730-1800 178 29.99
Standard conditions: saturated at 68°F»
Liquid Sample Data
Sample Location
West Tank - inlet
West Tank - dipped from tank
East Tank - dipped from tank
Velocity
(ft/min.)
576
1820
990
310 -
460
670
29.92 in. Hg.
Date
9/11/80
9/11/80 .
9/11/80
Stack
Conditions
(ACFM)
200
640
350
110
160
.230
Time
1805
'
1820
1825
Standard
Conditions
(SCFM)
120
410
260
76 -
120
110
Sample
Temp. ( F)
196
207
219
2
ft.
Benzene
Concentration
(ppm)
740
3010
1520
2410
840
2990
Benzene Cone.
(ppm by weight)
285
295
251
629
599
634_
282
286
261
W
w
H
I-1
VO
-J
Benzene $>
Emission ^L
Rate g
(Ib/hr) ° !
1
1
1.04
15.01
4.79
Ave. 6.95
2.22
1.17
3.89
Ave. 2.45
i
DJ
00
n
Average V i
1
277
!
621
276
-------
SET 1957-04-1280 Page 3-7
Liquid samples were collected from each of the tanks through the
manway and from the inlet to the west tank in the funnel. As shown in
Table 3-3, the west tank inlet had an average benzene concentration of
280 ppm, while the west tank process liquid contained 620 ppm benzene.
It is possible that the funnel is not the only inlet to the west tank, or
that there is great variability in the composition of the inlet liquid
because of multiple sources.
Scott Environmental Technology Inc.
-------
SET 1957-04-1280 Page 4-1
4.0 PROCESS DESCRIPTION
The U.S. Steel facility at Fairless Hills, Pennsylvania, operates
two coke batteries, each with 87 Wilputte ovens. The unit operations at the
coke by-product plant are primary cooling, tar decanting, exhausting, tar
electrostatic precipitation, ammonia recovery, naphthalene recovery, final
cooling, and light oil recovery.
The gas leaving the ovens is collected in a collecting ma.in where
it is sprayed with flushing liquor to reduce the temperature to about 175°F.
When the gas and flushing liquor leave the battery area, the flushing
liquor separates from the gas and travels to the flushing liquor descanter.
The gas is transported from the collecting main through a crossover main,
and it proceeds to the primary coolers; gas enters the primary cooler at
about 175°F and leaves at 95°F. Water from the primary cooler goes to the
tar decanter, and the tar from the tar decanters and flushing liquor
decanter is pumped to the tar dehydrator. Following the primary coolers,
the gas stream is pressurized by steam turbine-driven exhausters and then
it enters the electrostatic precipitator. The flushing liquor decanter
separates the dirty liquor into flushing liquor, tear, and sludge. Flushing
liquor is returned to the batteries for reuse, and any excess is pumped to
the phenolized weak ammonia tanks. After passing through the electrostatic
precipitor, the gas is conveyed to the ammonia recovery area. Ammonia is
recovered by using a 5 to 6 percent sulfuric acid solution; the final pro-
duct of this operation is ammonium sulfate crystals.
The coke oven gas proceeds from the ammonia recovery area to a direct
water final cooler. The final cooler water flows from the cooler to a
separation basin where crude naphthalene is skimmed from the surface, dried
Scott Environmental Technolosylnc-
-------
SET 1957-04-1280 " " Page 4-2
to 3 to 4 percent moisture, and pumped to storage. The water is pumped
from the separation basin to the hot well of the final cooler cooling
tower.
After cooling, the coke oven gas enters the two wash oil (scrubbers
which are in series with the countercurrent flow of the wash oil and the
gas stream. The benzolized wash oil is pumped to the light: oil recovery
area where light oil is removed by steam stripping. The light oil flows to
the crude residue separation column to separate the primary light oil
(0° to 150° fraction). The primary light oil vapors are cooled in a
condenser, and the primary light oil goes to a light oil decanter, which;is
not vented to the atmosphere, for water removal. The primary light oil is
collected in a tank and pumped to storage before it is shipped. The second-
ary light oil (150° to 290° F fraction) is separated, recovered, and stored.
No further processing of the recovered oils is performed at the Fairless
Hills plant.
After being stripped of light oils, the debenzolized wash oil flows to
a wash oil decanter. Water is removed, and the wash oil is cooled in an
indirect cooler and is returned to the wash oil scrubbers.
The clean coke oven gas exiting the wash oil scrubbers is used for
fuel. Approximately 35 percent of this gas is used for battery underfire
while the remainder is piped to the mill for operations such as firing the
boilers and furnaces.
4.1 PROCESS OPERATING PARAMETERS
The emissions from the final cooler cooling tower, wash oil decanter,
and tar dehydrator were tested.
Scott Environmental Technology Inc.
-------
SET 1957-04-1280 Page 4-3
the following process data were collected during the benzene emission
testing.
A. Process parameters
The water flow rate in the final cooler cooling tower was
2,100 gallons/minute.
The flow rate of benzolized wash oil from the wash oil
decanter was 27,500 gallons/hour.
The residence time of tar in the tar dehydrator was 24 hours.
B. Production data
Item 9/9/80 9/10/80 9/11/80
Coal charged/day (tons) 3,770 3,776 4,027
Coke produced (tons) 2,760 2,713 2,811
Coke oven gas produced
(cubic feet) 40,678 x 103 40,918 x 103 41,672 x 103
Light oil produced
(gallons) 7,185 9,733 7,571
Naphthalene produced
(gallons) 490 450 475
Tar produced (gallons) 31,030 28,960 33,100
The coal mixture was 70 percent high volatile coal and 30 percent
low volatile coal.
Scott Environmental Technology Inc.
-------
SET .1957-04-1280 . Page 5-1
5.0 FIELD SAMPLING AND ANALYSIS METHODOLOGY
5.1 DETERMINATION OF BENZENE FROM STATIONARY SOURCES:
EPA METHOD 110 AND MODIFICATIONS
EPA Method 110 consists of drawing a time-integrated stack gas
sample through a probe into a Tedlar* sample bag, which is enclosed in a
leak-free drum, by use of a pump hooked to the drum outlet which slowly
evacuates the drum, causing the bag to fill. A copy of the method is
included in'Appendix D.
The method was modified by Scott because as it stands the
method doesn't account for moisture in the sample stream, and is only
designed to measure benzene concentration, not mass emission rate. The
following modifications were made to all tests done using Method 110:
1. To obtain mass emission rates, velocity and temperature
readings were taken at the top of the stack at 5 minute intervals during
the 30-minute sampling runs. This information was used to calculate flow-
rate, which was used in conjunction with the benzene concentration to
yield the mass emission rate. Velocity readings were made using a vane
anemometer with direct electronic readout.
2. A personnel sampling pump was substituted for the pump,
needle valve, and flowmeter of the method. The personnel pumps have
built-in flowmeters and rate adjustment screws and have the further
advantage of being intrinsically safe, as required in many areas of
the coke plant.
* Mention of trade names or specific products does not constitute endorsement
by the U.S. Environmental Protection Agency.
Scott Environmental Technology Inc.
-------
SET 1957-04-1280
Page 5-2
3. Swagelok fittings were used in place of quick-connects.
4. Rather than discarding Teflon sample lines after each set
of samples, they were washed with propylene carbonate and/or acetone and
flushed with nitrogen before reuse.
5. An orifice and magnehelic gauge were inserted in the sampling
line before the Tedlar bag to indicate that air flow was reaching the
bag.
6. A water knockout trap was inserted between the probe and
magnehelic gauge to collect any condensate in the sample line.
7. The following cleanup procedures were followed:
If any condensate was collected in the trap or sample line, it
was measured and saved for analysis. The probe, line and trap were then
washed with propylene carbonate, which was also saved for analysis. Any
benzene found in these washes and water catches was added to the total found
in the sample bag to determine mass emission rates.
Bag volumes were measured whenever water was collected in the
trap by emptying the bag through a dry gas meter after the sample was
analyzed. The volume of water collected in the trap was then converted
to an equivalent air volume and was added to the volume in the bag to
determine the percent moisture in the sample stream.
After the probe, line and trap washes were completed, the lines
were washed with acetone to remove the propylene carbonate film and flushed
with nitrogen to dry.
Figure 5-1 shows the modified Method 110 setup.
Scott Environmental Technology Inc.
-------
SET 1957-04-1280
Page 5-3
FIGURE 5-1
r/\/v/<
Inc.
MODIFIED METHOD I/O
SAMPLING TRAIN
-------
SET 1957-04-1280 Page 5-4
5.2. SAMPLE HANDLING .
After being collected the gas samples were immediately
transported to the gas chromatograph and analyzed. The elapsed time
between sample collection and analysis never exceeded one hour.
To verify that there was no sample degradation in samples of this type
some of the samples were retained for 24 hours and reanalyzed. The
loss of benzene and isobutane observed was typically less than 3%..
/
5.3 FIELD ANALYSIS
All gas samples collected were analyzed using a Shimadzu
GC Mini 1 gas chromatograph equipped with dual flame ionization
detectors, dual electrometers, heated sample loop and a backflush
system. Figure 5-2 shows a schematic of the backflush apparatus.
The backflush system is composed of a ten port sequence reversal valve
and two columns, a scrubber column for retaining high molecular weight
compounds and an analytical column. When the system is in the inject
mode the scrubber column and the analytical column are connected in
series allowing sample components to move from the precolumn to the
analytical column. In the backflush mode the columns are disconnected
from each other and become two separate systems each with its own
carrier gas source. This arrangement allows the separation and measure-
ment of low molecular weight compounds while the scrubber column is
being backflushed of heavier sample components. Backflush times for
different mixtures of sample components must be predetermined to insure
that the compound(s) of interest are transferred to the analytical
column before backflushing is started.
Scott Environmental Technology Inc
-------
a
tt >
o I
Ul ill
§
K)
A >
CARRIER GAS A
A
fU
-------
SET 1957-04-1280 < Page 5-6
Samples for chromatographic analysis were drawn into a 20
cc glass syringe then introduced to the sample loop inlet. The
samples once in the sample loop were allowed to come to atmospheric
pressure by waiting 15 seconds prior to injection. The following
chromatographic conditions were maintained:
Column Temperature (isothermal) - 100 C
Injector and Detector Temperature - 200 C
o
5 ml Sample Loop, Temperature - 50 C
Carrier Gas Flow Rate - 32 cc/min.
Hydrogen Flow Rate - 40 cc/min.
Air Flow Rate - 240 cc/min.
Analysis Time - 5 min.
Detector - Flame loniziation
The columns used for field analysis were:
A - Scrubber Column
10% FFAP on Supelcoport 80/100
1/8" x 1 m Stainless Steel
B - Analytical Column
20% SP-2100, 0.1% Carbowax 1500
100/120 Supelcoport
1/8" x 10' Stainless Steel.
Scott Environmental Technology Inc
-------
SET 1957-04-1280 Page 6-1.
6.0 FIELD SAMPLING
6.1 COOLING TOWER
The cooling tower stands about 60 feet high and has a 17-foot
diameter fan on top surrounded by a 15-foot high shroud. Self-contained
breathing apparatus was required by the plant for anyone working a.t the
level of the top of the shroud.
Sampling was conducted in accordance with EPA Method 110,
modified as described in Section 5.1, and utilizing a 24-point sampling
and velocity traverse to collect an integrated sample and an accurate velocity
profile. The sampling time at each traverse point was two minutes.
The third run was voided because the concentration of benzene
in the collected sample was over twice as high as in the first two, indicating
a contaminated sampling bag. When checking the bag history, it was found
that the bag had been previously used at a very high concentration source,
and cleaning the bag following our standard procedure would have left
considerable residual benzene. The bag had not been checked for background
benzene prior to use due to an oversight when packing to leave the previous
plant. The run was therefore declared void and a fourth test was run.
Liquid samples were extracted from the hot and cold wells after
each of the first three test runs using an aluminum can on a rope. Amber
glass bottles were then filled from the can and the samples were returned to
Scott's laboratory for analysis.
Scott Environmental Technolosylnc
-------
1957-04-1280
Page 6-2
•if
TOWER
COOLING TOWER
FAN AT DECK. L£\/E"L -\
FAN MOTOR
WOOC DECK ^60'HIGH
Inc.
. FIGURE 6-1 COOLING TOWER FAN SHROUD
-------
SET 1957-04-1280 Page 6-3
6.2 WASH OIL DECANTER
Three Method 110 tests were conducted on the wash oil decanter.
The moisture content of the stack gas was high, and the water trap catch
volume was measured to use in calculating the percent moisture. As a check
on the accuracy of the moisture content determined from the water catch,
three separate moisture determination tests were run by drawing a source
sample through tared silica gel tubes connected to a calibrated dry gas
meter using a personnel sampling pump. The average results of the two
moisture determination methods were 60.5% for the water catch and 59.1%
for the silica gel, indicating that the method using the water catch volume
is accurate and can be used with assurance for other process calculations.
The wash oil in the decanter is very hot (205 F) and the hatchways
on the tank (see Figure 6-2) are normally open in summer to release heat.
During sampling, the hatchways were closed so emissions were coming only
from the vent stack.
Liquid samples were dipped from the hatchway on the inlet end
of the tank, next to the sampling vent. The liquid level was within 6"
of the top of the decanter. Samples were dipped with an aluminum sampling
container and allowed to cool slightly before transferring to amber glass
bottles.
6.3 TAR DEHYDRATORS
Triplicate Method 110 tests were run simultaneously on the east
and west tar dehydrators shown in Figure g-3- In addition to the method
modifications described in Section 5.1, some additional changes were made
to deal with the problem of naphthalene constantly plugging the probe and
Scott Environmental Technology Inc.
-------
SET 1957-04-1280
Page 6-4
WALK
HATCHES
WASH''OIL
DECANTER
HATCHES
o
TEST POINT
OUTLET END
WALK
INLET END
inc.
FIGURE 6-2 WASH OIL DECANTER
-------
SET 1957-04-1280
Page 6-5
WALK
WEST'TANK
1ST IN SERIES
FUNNEL INL&T
TEST
PT. O
EAST TANK
2ND I-N SERIES
O
WALK
MANWAYS
Inc.
FIGURE 6-3 TAR DEHYDRATORS - EAST & WEST TANKS
-------
SET 1957-04-1280 Page 6-6
sample line. The sample stream was bubbled through propylene carbonate to
knock out the naphthalene, using a large diameter glass elbow as a probe.
After the naphthalene was scrubbed out, the sample stream passed through
Teflon tubing and on into the sampling drum as usual. Figure 6-4 shows the
propylene carbonate sampling train. The glass probe was connected directly
to two impingers each containing 100 ml of propylene carbonate and a third
empty impinger. The impingers were contained in a bucket which wa.s hooked
onto the top of the stack.
Cleanup consisted of saving the impinger catches and wasihes in
addition to the sample line and water trap washes for analysis of benzene.
Scott Environmental Technolosy Inc
-------
SET 1957 04 1280 Page 7-1
7.0 LABORATORY SAMPLE ANALYSIS
Two types of liquid samples were collected: process liquids, and
sample line and water trap catches and washes. All liquid samples were
stored in amber glass bottles and returned to Scott's Plumsteadville laboratory
for analysis. "
7.1 SAMPLE PREPARATION
Depending upon the complexity of the sample, one of the following
sample preparation procedures was followed prior to the "purge and trap"
procedure and analysis.
Samples Containing Immiscible Liquid Phases
Using a clinical centrifuge (International Equipment Company,
Massachusetts) immiscible liquid phases were separated and each phase was
analyzed separately for benzene.
Samples Containing Solid and Immiscible Liquid Phases
Samples containing solids of higher density than the liquid phase
were separated by centrifuge or by simple decantation of the liquid. The
different phases in the liquid fraction were then further separated by
centrifuging. Solid and liquid phases were analyzed separately.
•Samples Containing Finely Crystalline Solid Suspension
In analyzing these samples the stoppered sample jars were shaken
for at ],east half an hour for homogenizing the solution. The uniform
distribution of suspended fine crystalline solid particles was tested by
determining the percentage of dry solid in several aliquots of the homoge-
nized mixture. A weighed amount of the mixture was analyzed for benzene.
Scott Environmental Technotosy Inc
-------
SET 1957 04 1280 . , „ Page 7-2
Sampling System Washings
All washings were clear solutions having only one liquid phase.
The total weight of the liquid phase was determined using a balance correct
to ±0.1 g. The total weight of each washing wzs more than 25 grams, so an
error of 0.1 g in weighing the mass will contribute an error of only 0.4%
to the final analytical data. A weighed aliquot of the washing was analyzed
for benzene by following the "purge and trap" and analysis procedures out-
lined in the following sections, and using this analysis data the weight
of benzene present in the total mass of washing was calculated.
7.2 PURGE AND TRAP PROCEDURE FOR EXTRACTION OF BENZENE FROM LIQUID PHASE
TO GASEOUS PHASE
An accurately weighed quantity of the sample to be analyzed was
diluted with 20-25 ml of propylene carbonate in a specially designed glass
purging apparatus which was kept immersed in a thermostatted water bath
maintained at 78°C. Benzene free nitrogen gas was bubbled through the
propylene carbonate solution in the purging apparatus at the rate oJ:
0.2 - 0.3 liters/minute, and collected in leak free Tedlar bags. Under
these experimental conditions, 1 1/2 - 2 hours were sufficient to purge
off all the benzene from the. liquid phase to the gaseous phase. The total
volume of nitrogen gas used to purge the sample was accurately measured
by a calibrated dry gas meter. A diagram of the purge and trap set-up is
shown in Figure 7-1.
Propylene carbonate was found to be an ideal diluting solvent
for the extraction of benzene from all types of liquid samples containing
viscous tar, pitch, light and heavy oil and insoluble particulates. It
was chosen for its high boiling point, low density, and good solvating
capacity.
Scott Environmental Technology Inc.
-------
NITROGEN
CYLINDER,
P>
00
FIGURE 7-1 PURGE AND TRAP METHOD EQUIPMENT SET-UP
-------
SET 1957 04 1280 -..- Page 7-4
t
7.3 GAS CHROMATOGRAPH . "
A Perkin-Elmer 900 gas .chromatograph was used for the analysis
of the purge bags. A 10 ft. by 1/8 inch stainless steel column packed with
20% SP-2100/0.1% Carbowax 1500 on 80/120 mesh Supelcoport was used for the
analysis. This column gave complete resolution of the benzene peak from
other components present in the purge bags. The 'peak height* method was
utilized to calculate the concentration of benzene in the purge bags
analyzed. The Perkin-Elmer 900 used for analysis was not equipped with
a backflushing unit. Gas chromatograph conditions were as follows:
GC column temperature: 70°C isothermal
Detector temperature: 190°C
5 ml loop at a temperature of 120°C
Carrier gas flow rate: 30 cc/rain He
Hydrogen flow rate: 45 cc/min
Oxygen flow rate: 400 cc/min
*
Detector: Flame lonization Detector (FID)
In addition to benzene, the purge bags contained other volatile
hydrocarbons present in the liquid samples such as toluene and naphthalene.
Because this chromatograph was not equipped with a backflush, it was
necessary to elute all heavy organics from the column by heating the column
to 150°C after every two injections for one hour with the carrier gas on.
After cooling the column to 70°C the absence of any organic in the column
which might overlap the benzene peak in the next analysis was checked. When
the column was found to be satisfactorily clean, the next analysis was
continued under the conditions previously described.
Scott Environmental Technology Inc.
-------
SET 1957 04 1280 : Page 8-1
8.0 QUALITY CONTROL AND QUALITY ASSURANCE
The following sections will address quality control and quality
assurance procedures for the-, field analysis of benzene in air samples and
the laboratory analysis of process liquids and BaP samples.
8.1 FIELD ANALYSIS PROCEDURES
All samples were analyzed in duplicate and as a rule peak heights
were reproduced to within 5%. For some very high concentration samples
(percent range) it was necessary to make dilutions for analysis. When this
was done a fresh dilution was prepared for each injection and peak heights
were reproduced to within 10%. To verify that the system was retaining no
benzene, frequent injections of the standard and nitrogen were made. In all
cases the result was satisfactory.
The Tedlar bags that were reused for sampling were flushed three
times with nitrogen and allowed to sit overnight after being filled to
approximately three quarters of their capacity. They were analyzed for
benzene content the following day. The background concentrations of the
bags were recorded and varied from 0 to 10 ppm benzene. Care was taken to
use sample bags whose background concentration was very low compared to the
expected concentration of the source.
The accuracy and linearity of the gas chromatographic techniques
*
used in this program were tested through the use of EPA Audit Samples. Two
standards, a 122.5 ppm and 6.11 ppm benzene were used to analyze the audit
cylinders.
Scott Environmental Techndosy Inc
-------
SET 1957 04 1280 ; Page 8r2
8.2 PROCEDURES FOR ANALYSIS OF PROCESS LIQUIDS .
Scott's benzene standards, checked against EPA Audit Standards,
were used as reference standards throughout this program. The accuracy and
linearity of the gas chromatographic technique for benzene analysis was
tested through the use of EPA Audit Standards which were available to Scott.
Gas chromatographic analysis of the samples and standard were performed
under identical conditions to assure the accuracy of the analytical data
generated.
Each batch of propylene carbonate which was used as the diluting
solvent in the purge and trap technique was analyzed for benzene content by
subjecting 25 ml of propylene carbonate to the purge and trap procedure
followed by gas chromatographic analysis of the trapped gas under identical
conditions as described in Section 5.2. All batches of analytical grade
propylene carbonate were found to be free from benzene.
Every day before the analysis of samples the purging apparatus and
trapping bags were tested for absence of benzene. Whenever the whole system
was found to be free from benzene to the lowest detectable limit of the
instrument, the samples were analyzed using the purging apparatus and the
trapping gas sampling bags.
Generally an accurately weighed mass of each sample was-subjected
to purge and trap procedure only once and the trapped gas sample was repeat-
edly analyzed by GC until the analytical data of consecutive GC analyses varied
by ±0.5% or less.
Scott Environment^'Tedrnclosy tnc
-------
SET 1957 041280 Pa8e 8~3
For randomly selected samples, Che whole analytical procedure was
repeated with a different weighed mass of the source sample to check the
validity and accuracy of the analytical methodology. The analytical data
for different runs were found not to vary by more than 5%.
By purging the sample with nitrogen under the experimental con-
ditions as utilized by Scott, the recovery of benzene from the sample was
quantitative and this has been verified .by analyzing a standard benzene
solution in propylene carbonate containing tar and pitch.
Scott Environmental Technolosylnc
------- |