EPA PROJECT  REPORT NUMBER 75-CBK-2
   CD
o
                                           CABOT  CORPORATION

                                        Waverly,  West  Virginia
                 UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
                             Office of Air and Waste Management
                         Office of Air Quality Planning and Standards
                               Emission Measurement Branch
                           Research Triangle Park. North Carolina

-------
              FINAL REPORT
                   on
SAMPLING AND ANALYSIS OF SOURCE EMISSION
    SAMPLES FROM A CARBON BLACK PLANT
                   to
     ENVIRONMENTAL PROTECTION AGENCY
                May 12, 1975
                   by


     S. E. Miller and R.  E.  Barrett
       EPA Contract No.  68-02-1409
            Task Order No.  9
                BATTELLE
          Columbus Laboratories
             505 King Avenue
          Columbus, Ohio  43201

-------
                            TABLE OF CONTENTS
                                                                      Page
INTRODUCTION	„	      1
     Summary and Conclusions 	      2
     Plant and Process Detail	      4
     Sampling Locations. . . . „	      4
          Total Stack Flow	      7
     Sampling and Analytical Procedures	      7
          Particulate and POM	      7
          Sample Recovery	      9
          Analysis for POM Species	     11
          Hydrogen Sulfide (H S) by Method 11	     14
          Sulfur Dioxide (SO-) by Method 6	     14
          Hydrocarbons and Reduced Sulfur Compounds	     17
          Carbon Monoxide (CO), Carbon Dioxide (CO ) and
            Oxygen (0^	     20
          Moisture	     20
          Visible Emissions	     22
          Infrared Telespectrometer Measurements 	     22
     Recommendations on Sampling Procedures	 .     22
          H S - EPA Method 11	     22
          SO  - EPA Method 6		     24

                                APPENDICES
APPENDIX A.  PROCESS DESCRIPTION
APPENDIX B.  INFRARED REMOTE SENSING MEASUREMENTS
APPENDIX C.  FIELD DATA FOR VISUAL DETERMINATIONS
             FIELD DATA FOR METHOD 5 POM RUNS
             FIELD DATA FOR "PACKED COLUMN" POM RUNS
             S02 AMD H S TITRATION DATA

-------
                              List of Tables
                                                                      Page
Table 1.  Summary of Emissions from a Carbon  Black Plant	       3
Table 2.  Total Stack Gas Flow	       8
Table 3.  Particulate Data Unit No. 3	      10
Table 4.  Summary of Hydrogen-Sulfide Data -  EPA Method  11  .  .  „  .      15
Table 5.  Summary of Sulfur Dioxide Data - EPA Method 6	      16
Table 6.  Summary of Results - Gas Chromatography,
            Mass Spectrometry	      18
Table 7.  Summary of Results - Gas Chromatography,
            Mass Spectrometry	„	      19
Table 8.  Summary of Orsat Data	      21
Table 9.  Summary of Visible Emissions	      23


                             List of Figures

Figure 1.  Flow Diagram, Cabot Carbon Black Plant	      5
Figure 2.  Stack Geometry and Sampling Port Locations	      6

-------
                     SAMPLING AND ANALYSIS OF SOURCE
                EMISSION SAMPLES FROM A CARBON-BLACK PLANT
                                   by
                     S. E. Miller and R. E. Barrett

                               INTRODUCTION

          This report summarizes the results of a sampling program recently
conducted at the Cabot Corporation "Ohio River" carbon-black plant in
Waverly, West Virginia.  The program was conducted by Battelle-Columbus
pursuant to the terms of Task Order No. 9, Contract No. 68-02-1409 with
the U.S. Environmental Protection Agency.
          The objective of this program was to obtain accurate and detailed
data on air pollutant emissions from the Cabot Corporation Waverly plant for
the Standard Support Document for the Carbon-black industry.
          The field-sampling portion of the program was conducted during the
week of December 9-13, 1974.  Battelle's Columbus Laboratories (BCL)  was
responsible for measuring gas compositions in the stack upstream of the
flares  (i.e., below the flares) for Polycyclic Organic Matter (POM),  H.S,
S09, CO, and selected hydrocarbons.  In addition, Battelle measured CO,., 0 ,
and moisture content of the gas and visible emissions.  Concurrent with the
BCL measurements, EPA used an infrared telespectrometer system to obtain
remotely emission spectra of flare species over the spectral range of 2 to
14 microns.  Operation of the plant and control systems were monitored by
PEDCO-Environmental of Cincinnati, Ohio, who are also under contract  to EPA
to develop information for the emission standard support document.
          Because the sampling described in this report was conducted up-
stream of the flare, the measured values of combustible species (CO,  hydro-
carbons, and POM) should not be considered as equivalent to emissions.
Actual emissions of CO and hydrocarbons would be less than the measured
values.  Actual emissions of POM are impossible to estimate based on the
current knowledge of POM reactions.
          The initial sampling and analysis schedule for this task called
for BCL to obtain 2 gas samples from each of the 3 process trains.  However,

-------
 during  the field-sampling period, Unit No. 1 was not in operation due to
 periodic maintenance and installation of new bags in the baghouse.  Con-
 sequently, the sampling schedule was changed to include 3 samples from each
 of Units No. 2 and No. 3.  The sampling operation was conducted as four
 separate runs with multiple samples obtained during some runs.  The run
 schedule was as follows:
     Run No. 1   Unit 3       December 10  (velocity profile obtained
                                December 9)
     Run No. 2   Unit 3       December 11
     Run No. 3   Unit 2       December 11
     Run No. 4   Unit 2       December 12
     Run No. 5   Unit 3       December 12  (POM run only)
 Multiple samples obtained during one run are identified by postscripts
 on the  run numbers  (e.g., 5A, 4B) .

                         Summary and Conclusions

          Table 1 summarizes results of the gas composition measurements
made by BCL at the Cabot carbon-black plant.
          Upstream of the flare (in the stack)  the exhaust gas from this
source is incompletely oxidized and, hence, at  the point at which sampling
was conducted the majority of the sulfur is present in reduced species
 (H S and CS^) rather than as S0_.  Based on the mass spectrometer data, if
all the reduced sulfur compounds were oxidized  to SCL in the flare, the
equivalent emission of SO  would be:
          Run No. 1      62 to 65 Kg SO /hr equivalent
          Run No. 2     100 to 103 Kg SO /hr equivalent
          Run No» 4     100 to 103 Kg SO /hr equivalent.
          Some disagreement is observed in the  H S data acquired by Method 11
and by mass spectrometer analysis.  As discussed elsewhere in this report,
the large concentrations of H S present in the  stack gas were beyond the
normal range of Method 11.  Also, interference  reactions with other sulfur
compounds present in the stack gas are suspected.  Hence, the mass spec-
trometer values were accepted as more accurate.

-------
TABLE 1.  SUMMARY OF GAS COMPOSITIONS MEASURED UPSTREAM OF
          THE FLARE AT A CARBON-BLACK PLANT


Gas Analysis,
volume percent
co2
°2
CO
Moisture, vol percent
1

3


--
—
--
43
Total Dry Gas Flow Rate
in Stack Nm /min 327
Total Gas Flow Rate
at Sampling Point C,
Nm3/min
Flow of Specific
Species, kg/hr
S02 (Method 6)
S02 (GC)
H S (Method 11)
H2S (GC)
CS2 (GC)
COS (GC)
H2 (GC)
CH (GC)
C2H2 (GC)
C2H4 (GC)
C2H6 (GC)
Total C and C,
hydrocarbons (JGC)
Total POM
(Packed Column)
Opacity, percent

38.3'

0.38
< 2.6
18.2
19.4
12.4
4.4
276
86.1
250
7.8
< 0.1
4.3

0.00030
—
Run Number
2 3
Unit Number
3 2


3.7
0.5
13.3
45 48
336 429

38.4

--
< 2.6
38.3 31.0
22.8
31.9
3.2
284
92.6
260
8.2
< 0.1
5.0

—
8.8 13.8
4 5(a)

2 3


3.1
0.8
13.3
45
378

38.9

4.83
<3.0
38.0
18.2
35.8
4.9
267
68.6
307
4.3
< 0.1
5.6

__
—
(a) This run made for POM only.

-------
          Measurement of carbon monoxide (CO) was attempted by non-dispersive
infrared (Method 10), but levels were far in excess of the limit of the avail-
able instrument (6000 ppm).  CO determinations were made by two alternate
methods, i.e., EPA Method 3 (Orsat) and gas chromatography.

                        Plant and Process Details

          The Cabot Ohio River plant contains 3 process units where carbon
black is produced from residium tar.  A process-train flow diagram is
shown schematically in Figure 1.  Each process train is equipped with a
separate baghouse for collecting the carbon-black product and atmospheric
flares  for disposal of off gases.  The plant has a vertically oriented
design with the reactor effluent going to the baghouses at the top of
the structure and the carbon black passing to successively lower levels
for each processing step.  A portion of the flue gas (about 13 to 15
percent) from one train is returned as auxiliary fuel for the direct-
fired dryer ovens.
          During the field-sampling period, process operations at the
Cabot Ohio River plant were maintained at normal operating levels on
Units 2 and 3 while Unit 1 was out of service.  No major upset or change
in operating condition was encountered during the sampling periods.
During  the last day of sampling, the flares on Unit 2 were turned off
and could not be re-ignited because of high winds.  This should have no
significant effect on the BCL stack sampling results since the sampling
port for this unit was located 10 feet below the flares.  Further details
on this process and an operations log have been provided by PedCo and
appear in Appendix A.

                           Sampling Locations

         Stack geometry and sample-port locations are shown in Figure 2.
Integrated gas samples were extracted from the main stacks approximately
10 feet below the flares (Point A).  Particulate measurements for POM
(polycyclic organic matter) determinations were made at a lower level take-

-------
                                                                                                   ROSE
                                                                                                • Sampling
                                                                            Flare
Air
Preheater

r -0
'
!
l
Residuum Y


Feed A ^\
•* 	 Air

Baghouse
'
> '\/\/\/
1 ^V' 1
Tar ' T 1
„ af.-r Natural Water
Heate Gas ,Quench Carbon Black

Gaseous
— .L. . ,. — A
1 Sampling

^^
1
t
1

': Stack
Fuel — >
Air — >

^v
\
- POM;
Sampling
f
Direct Fired
Dryer Ovens'
NOTE:  The plant has three of these trains.
                                  FIGURE 1.  FLOW DIAGRAM, CABOT CARBON BLACK PLANT

-------
          	 B-Rose  Sampling
                     Flare
                  A- Gaseous  Sampling
                    (I"  pipe  coupling)
               Existing Platform
                30" i.d.  Stack
                            C-PNA Sampling
                        (3"' pipe coupling)
                                               90*
                                                         I
                                                               30
12"i.d. Duct
                                                     To Dryer
FIGURE 2,  STACK GEOMETRY AND SAMPLING PORT LOCATIONS

-------
off from the main stack  (Point C) where a portion of the flue gas is fired
as auxiliary fuel to heat the drying ovens.  POM measurements were made on
Unit No. 3 only.

Total Stack Flow

          Gas velocity and stack temperature traverses were made in one
direction only using Methods 1 and 2 procedures described in the Federal
Register of December 23, 1971.  A two direction traverse could not be made
due to physical constraints of the sample platform.  Total stack-gas flow
was calculated by computing gas velocities at each traverse point and then
determining the average gas velocity and flow rate.  Table 2 gives the total
stack-gas flow for each traverse.
          Because of the number of samples to be collected, velocity profiles
could not be measured simultaneously with collection of each sample.  Hence,
velocity profiles were determined prior to or after other sampling as can be
noted from the sampling times shown in Tables 1 through 7.

                    Sampling and Analytical Procedures

Particulate and POM

          Three particulate runs, to obtain samples for POM analysis, were
made on Unit No. 3.  Each run was conducted in parallel employing (1) EPA
Method 5 (with a back-up "packed column" between the third and fourth
impingers), and (2) a "packed column" sampling train.  The sample probe for
each train was located approximately 5 inches inside the duct, where velocity
profiles showed the velocity head pressure to be fairly uniform at 1.1 in.
HO.  A 2-hour sampling period was employed for each run.  Due to excessive
pressure drop across the packed column isokinetic sampling was not maintained.
It is believed that the divergence from isokinetic sampling was of little
consequence since the particle size was quite small.  Particulate samples
were not used in establishing mass emission rates because the sampling location
(Point C, Figure 2, or equivalent to upstream of the flare) was not suitable
for obtaining particulate mass emission.

-------
TABLE 2.  TOTAL STACK GAS FLOW
Run
No.
1
2
3
4
Unit
No.
3
3
2
2
Total Stack Gas Flow,
Date
12/9/74
12/11/74
12/11/74
12/12/74
Time
1600
0930
1300
1130
dscm/min
327
336
429
378
dscf/min
11,549
11,879
15,164
13,360

-------
          Mass loading particulate data, obtained from the filters used in
the POM runs, are presented in Table 3.  The two filters from Run No.  1 were
inadvertently destroyed in the recovery process for POM analysis before
weights could be obtained.  Since all 3 POM runs were made in a similar
manner (except for sampling rate), it is suggested that the mass particulate
loading obtained on the filters from Run Noc.  2 and 5 are representative of
particulate loading during Run No. 1, after the mass loadings obtained in
Runs No. 2 and 5 are adjusted so that they conform to the sample volumes
obtained in Run No. 1.
          The POM analysis program provided for only one run (Run No.  1) to
be analyzed.  Because the POM data were not as significant as expected, it
was mutually agreed that POM analysis of Run Nos. 2 and 3 would not be con-
ducted at this time.  However, these samples are being retained in POM
storage until the final report is prepared for distribution.

Sample Recovery

          The sample for POM analysis was recovered from the EPA Method 5
train as required by this method.  To assure maximum recovery of POM,  an
additional methylene chloride rinse of the acetone rinsed impingers was made.
In order to facilitate evaluation of the effectiveness of each portion of
the Method 5 sampling train, each portion was  separately analyzed as detailed
below.  The collected samples were recovered from the Battelle adsorbent
sampling columns by continuous extraction with pentane for 24 hours.

Sample Number                           Sample
  1-1               Methylene chloride soxhlet extract (24 hours) of Method 5
                    filter together with acetone probe rinse.
  1-2               Methylene chloride extract of the aqueous impinger contents
                    and subsequent water rinse, together with the acetone
                    rinse of the empty impingers.
  1-3               Further rinse of the empty impingers with methylene chloride,
  1-4               Pentane extract of adsorbent sampling column back-up to
                    the Method 5 sampling train.
  1-5               Pentane extract of the parallel adsorbent sampling column,
                    together with the methylene chloride Soxhlet extract
                    (24 hours) of its upstream filter, and the acetone rinse
                    of the sampling probe.

-------
                            10
           TABLE  3.   PARTICUIATE DATA  UNIT  NO.  3

Run
No.
1A
IB
2A
2B
5A
5B
Mass(1)
Loadings ,
gr.
ND
ND
0.2487
0.1761
0.2410
0.1685
Sample
Volume
DSCF
66.79
36.63
51.94
35.26
53.29
37.54
Stack Gas
Volume
DSCFM
1353
--
1356
--
1376

Stack
Moisture,
Vol. Percent
44.6
--
44.4
--
43.4

(1)   Filter only

(2)   Runs  labeled "A"  are  using Method  5  sampling  train,
     Runs  labeled "B"  are  using the  packed  column  sampling  train

-------
                                  11
An internal standard of 9-methyl anthracene was added to each of the above


extracts, which were then reduced to a volume of about 2 mis by a rotary


evaporator.  A Rosen liquid chromatographic separation was then carried out


on each of the extracts.   The eluting fraction containing all POM species


between anthracene and dibenzpyrenes, together with the eight NAS three and


four star carcinogens, was collected and reduced to approximately 200 ul


using a Kuderna-Danish evaporator.



Analysis for POM Species



        The above five samples were each quantitatively divided into two portions,


one portion was quantitatively analyzed by gas chromatography-mass spectrometry


(GC-MS) for the POM species present (except benz(a)pyrene).   The other fraction


was  quantitatively analyzed  by.thin layer chromatography and  spectrofluorimetry.


GC-MS analysis was carried out using a Finnigan 1015 quadrupole mass  spectrometer


with a chemical ionization source.   The reagent carrier gas used was  methane.


Gas  chromatographic separation was  carried out using a  12 foot 37» Dexil 300


column, programmed from  170  C to 350 C at 4 C min  . Data  acquisition was


carried out using a System Industries 150 Data Acquisition  System with facilities


 for  CRT display and print-out with  XY plotter.  The POM species present in each


 sample were detected by  means of multiple ion overlay computer routines,


which utilize prior knowledge of individual mass spectra and  also make use of


 the  methane ionization adduct ions  at M+29 and M+41.  Isomer assignments were

                         4
 made by means of relative GO retention times whenever possible.


           Following the  identification of the POM species present, quantifi-


 cation was carried out with  reference to the internal standard using a

-------
                                   12
computer routine which permits integration of the appropriate mass spectral

    *-•
ions present in any computer reconstructed GC peak.  We have previously


demonstrated that this new integration routine has a reproducibility and


accuracy of better than 7  percent over a very wide range of mass spectrometer con-


ditions for the major species of interest in this program.  Over  the narrower


range of conditions used in this work, the accuracy is rather better.


          Benz(a)pyrene was separately quantified by the well-proven method


involving spectrofluorimetric determination of the protonated species, following


separation of the sample by thin layer chromatography  (TLC).  These experiments


were conducted in dim amber lighting, to minimize sample decomposition.   Samples


were spotted in triplicate onto activated alumina TLC  plates, and developed


using a 5 percent solution of diethyl ether in hexane.   (The TLC  plates were


previously pre-cleaned by 48 hours of Soxhlet extraction with an  azeotropic


mixture of chloroform and acetone; this Battelle developed  procedure  consistently


reduces the plate "background" to less than 2 ng benz(a)pyrene equivalent.)


          The developed samples were cut from the TLC plates with the aid  of


benz(a)pyrene markers on the sides of the plates, and the alumina was eluted


with peroxide-free diethyl ether.  This solvent was removed with a stream of


nitrogen, the sample was immediately dissolved in 1.0 ml of concentrated


sulfuric acid and transferred to a quartz cuvette for spectrofluorimetric


quantisation in a Turner Model 210, using excitation of 4700 ± 25 A and emission


of  5400 ± 100 A.  The spectrofluorimeter was calibrated by means of approximately


twenty determinations of 10 to 200 ng of pure benz(a)pyrene using an  identical


procedure to that employed for the unknown samples.  The results of the POM


quantification by GC-MS and spectrofluorimetry are tabulated below:

-------
                                  13
                                          POM In Sample, ne
POM Species




Anthracene, phenanthrene




Fluoranthene




Pyrene




Benz(a)anthracene, chrysene




Benz(a)pyrene






Total POM (ng)
1-1
1750
9600
7700
-
63
1-2
8000
16200
54200
1900
69
1-3
2360
4600
2080
1400
24
1-4
10500
5650
21500
400
58
1-5
85800
10100
38700
1400
122
19,100   80,400   10,500    38,100   136,000
          Apart from the small quantity of benz(a)pyrene, none other of the




eight NAS three and four star carcinogens were detected.   The absence of




significant quantities of the POM species larger than four condensed benzene




rings suggests very rapid quenching of the combustion effluents, which may be




an inherent characteristic of carbon black manufacture.




          The EPA Method 5 sampling train (110,000 ng) compared reasonably well




with the Battelle adsorbent sampling column (136,000), but the non-quantitative




nature of Method 5 is underlined by finding 38,100 ng of POM in the adsorbent




sampling column used to back-up Method 5.

-------
                                   14
Hydrogen Sulfide (H S) by Method 11

          Four H2S runs were made on each unit (Unit Nos. 2 and 3),
employing EPA Method 11, as outlined in the Federal Register of March 8,
1974.  The results of the 8 runs are tabulated in Table 4.  During the
initial H2S run, a yellow precipitate (CdS) was visually observed in the
impingers.  This is an indication that the capacity of the 3 absorbing
impingers has been exceeded.  To compensate for the relatively high con-
centration of H2S in the stack gas, subsequent H S runs were made at a
lower sampling rate.  The analysis titration for H S was conducted at the
sampling location in order to prevent loss of iodine from the sample.
Table C-1 in the Appendix lists data and calculations associated with the
H S determinations by Method 11.
          In addition to EPA Method 11, H S and other sulfur compounds of
interest (CS , COS, and SO ) were analyzed by mass spectrometry (MS) at BCL.  The
analytical results for the MS runs are included in Tables 6 and 7.  As was
pointed out earlier, some disagreement is observed in the H?S data acquired
by the two methods„  Visual observations at the sampling site gave some
indication that the large concentrations of H S present in the stack gas
were beyond the normal range for Method 11; hence, the values obtained by
MS analysis are deemed to be more accurate.  However, if the Method 11
sampling train is overloaded with H S, it would be expected to result in
emission values that are too low.  Conversely, the data in Table 1 show that
the Method 11 train generally produced higher H2S emission values than
obtained by mass spectrometry.  Possibly the iodine used in the Method 11
procedure is reacted with CS^ so that the Method 11 procedure is not unique
for H2S.

Sulfur Dioxide (S02> by Method 6
          Three SO. runs were made on each unit (Unit Nos. 2 and 3),
employing EPA Method 6 as outlined in the Federal Register of December 23,
1971.  One sample from each unit (Run Nos. 2 and 3) was voided due to a
misaligned sampling probe on the first impinger which resulted in a leak
in the sampling system.  The results of the S02 runs are tabulated in
Table 5.  Sample analysis was conducted at the sampling location.
Table C-2 in the Appendix lists data and calculations associated with
the S00 determination by Method 6.

-------
                    TABLE 4.  SUMMARY OF HYDROGEN-SULFIDE DATA - EPA METHOD 11
Run No.
1A
IB
2A
2B
3A
3B(a>
4A(a)
4B(a)
Unit
No.
3
3
3
3
2
2
2
2
Date
12/10/74
12/10/74
12/11/74
12/11/74
12/11/74
12/11/74
12/12/74
12/12/74
Time
1135-1150
1245-1300
1040-1050
1120-1130
1640-1655
1748-1803
0930-0940
1000-1010
Sample
Volume ,
Nm3
0.0165
0.0105
0.0051
0.0036
0.0056
0.0080
0.0049
0.0044
Moisture,
vol percent
43
47
45
45
48
47
46
49
H2S,
rag
12.0
11.8
9.9
6.7
8.8
6.7
8.4
7.2
H9S
•£ . o
mg/Nm
727
1123
1941
1861
1571
838
1714
1636
Emission
kg/hr
14.3
22.0
39.1
37.5
40.4
21.6
38.9
37.1
Ib/hr
31.5
48.5
86.2
82 . 7
89.1
47.6
85.8
81.8
(a)   Stack flare out.

-------
                       TABLE 5.   SUMMARY OF SULFUR DIOXIDE DATA - EPA METHOD 6
Run No.
1A
IB
2
3
4A(b)
4B(b)
Unit
No.
3
3
3
2
2
2
Date
12/10/74
12/10/74
12/11/74
12/11/74
12/12/74
12/12/74
Time
1135-1235
1620-1720
1040-1140
1645-1745
0936-1041
1300-1400
Sample so so
Volume, Moisture, ™2' 2'
Nm3 vol percent mg mg/Nn
0.
0.
0.
0.
0.
0.
131
128
125
128
120
135
41
41
2.6
2.1
43
43
0.81 6.2
4.15 32.4
0 Run
0 Run
21.89 182.4
32.83 243.2
_ Emission
i~" kg/hr
0.12
0.64
voided 
-------
                                    17

Hydrocarbons and Reduced Sulfur Compounds

          Three samples were obtained from each unit for combined gas
chromatography (GC) and mass spectrometry (MS) analysis.  The grab samples
were taken in 3-liter volumetric glass bulbs and analyzed at BCL.  The GC
unit included a Packard, Series 800 with a flame ionization detector, and
an Aerograph, Model 202 with a thermal conductivity detector.  The MS used
was a Consolidated Electrodynamics Corporation, Model 21-620.
          The GC and MS are independent analyses; however, GC analyses are
often used to compliment MS analyses when interference problems arise.  For
example, ethylene, ethane, carbon monoxide, and nitrogen all have the same
nominal mass to charge ratio of 28 on the mass spectrometer.  The GC provides
accurate analysis for the above compounds which then may be used to make
proper adjustments in the MS analyses.  The MS provides a broad range of
compounds not easily obtained by GC.
          The hydrocarbons were analyzed using the GC instrument equipped
with the flame ionization detector and a Porapak Q* column.  The sulfur
compounds, oxygen, nitrogen, hydrogen and argon are calculated from the mass
spectra after proper adjustment for some hydrocarbons which might interfere.
          Standard gas mixtures were used for calibration of the flame
ionization detector.  One standard mixture contains the saturated straight
chain hydrocarbons and a second mixture contains the unsaturated straight
chain hydrocarbons.  Research grade gases are used to standardize the mass
spectrometer data.
          The analytical results for the GC-MS runs are compiled in Tables
6 and 7 for Units 3  and 2, respectively.  Light hydrocarbon compounds
(through C.) are reported.  The sulfur compounds of interest and additional
components derived from the GC runs and MS runs using a mass-to-charge ratio
through 100, are also reported in Tables 6 and 7.
*  Trade name

-------
            TABLE 6.  SUMMARY OF RESULTS - GAS CHROMATOGRAPHY, MASS SPECTROMETRY
Run No. 1
Date - 12/10/74
Time - 11:30
Emission
Sample
H2S
CS2
COS
so2
C02
CO
02
N2
H2
A
CH4
C2H2
C2H4
C2H6
C3H6
C3H8
C4H10
C4H1Q
ppm
700
200
90
< 50
38400
118000
4300
644000
168000
7400
6600
11800
340
< 5
6
29
(iso) 16
(n) 51
Kg/hr
19.4
12.4
4.4
< 2.6
1376
2691
112.0
14692
275.9
241
86.1
250
7.8
< 0.1
0.2
1.0
0.7
2.4
Ib/hr
42.7
27.3
9.6
< 5.7
3034
5933
246.9
32391
608
531
190
551.5
17.2
< 0.2
0.5
2.2
1.6
5.3
ppm
800
500
60
< 50
38600
104000
4700
660000
165000
7400
6700
11800
358
< 5
13
33
18
53
Run No. 2A
Date - 12/11/74
Time - 9:15
Emission
Kg/hr
22.8
31.9
2.9
< 2.6
1423
2440
126.0
15489
278
248
89.9
258
8.4
< 0.1
0.5
1.2
0.9
2.6
Ib/hr
50.4
70.3
6.6
< 5.7
3137
5379
277.7
34148
614
546
198
568
18.6
< 0.2
1.0
2.7
1.9
5.8
ppm
800
500
70
< 50
39000
115000
2000
645000
171000
7300
7100
12000
343
< 5
7
29
18
53
Run No. 2B
Date - 12/11/74
Time - 9:20
Emission
Kg/hr
22.8
31.9
3.5
< 2.6
1438
2698
53.6
15135
288.8
244.1
95.3
261.6
8.0
< 0.1
0.2
1.0
0.9
2.6
Ib/hr
50.4
70.3
7.7
< 5.7
3169
5948
118.2
33366
637
538
210.2
576.7
17.8
< 0.2
0.5
2.3
1.9
5.8
                                                                                                        00
Note: Standard conditions = 70 F, 29.92 in. Hg.

-------
            TABLE 7.  SUMMARY OF RESULTS - GAS CHROMATOGRAPHY, MASS SPECTROMETRY
Run No. 4A
Date - 12/12/74
Time - 1415
Emission
Sample
H2S
cs2
COS
S02
C02
CO
°2
N2
H2
A
CH4
C2H2
C2H4
C2H6
C3H6
C3Hg
C4H10
C4H10
ppm
600
400
80
< 50
40600
117000
9300
668000
139000
7600
4400
12300
162
< 5
9
35
(iso) 14
(n) 43
Kg/hr
19.3
28.7
4.5
< 3.0
1684
3087
280
17631
264
286
66.5
302
4.3
< 0.1
0.4
1.4
0.7
2.3
Ib/hr
42.4
63.. 3
10.0
< 6.6
3711
6806
618
38870
582
631
146.5
665
9.3
< 0.2
0.7
3.2
1.7
5.2
ppm
400
600
100
< 50
41400
117000
8500
664000
143000
7700
4600
12700
160
< 5
11
36
11
64
Run No. 4B
Date - 12/12/74
Time - 1420
Emission
Kg/hr
12.9
43.0
5.7
< 3.0
1717
3087
255.9
17530
272.1
289.4
69.6
311.7
4.2
< 0.1
0.4
1.5
0.6
3.6
Ib/hr
28.3
94.8
12.5
< 6.6
3784
6806
564.6
38646
599
639
153.3
686
9.2
< 0.2
0.9
3.4
1.3
7.8
ppm
700
500
80
< 50
40500
123000
8600
662000
140000
7700
4600
12600
161
< 5
8
30
14
63
Run No. 4C
Date - 12/12/74
Time - 1425
Emission
Kg/hr
22.4
35.8
4.5
< 3.0
1679
3246
256.9
17475
266
287.4
69.6
308.7
4.3
< 0.1
0.3
1.2
0.7
3.5
Ib/hr
49.6
79.1
10.0
< 6.6
3702
7156
572
38524
586
639
153.3
681
9.3
< 0.2
0.7
2.7
1.7
7.6
Note: Standard conditions = 70 F, 29.92 in. Hg.

-------
                                   20
Carbon Monoxide  (CO), Carbon Dioxide  (CO ) and Oxygen  (0 )

          Measurement of CO was attempted by non-dispersive infrared technique,
but  levels were  far in excess of the  limit of the available instrument
 (6000 ppm).  This problem was anticipated and CO determinations were made by
two  alternate methods, i.e., EPA Method 3 (Orsat) and  gas chromatography of
grab samples.  Since carbon monoxide  and nitrogen have the same nominal mass,
a  GC unit equipped with a thermal conductivity detector and a molecular
sieve column was used to analyze the  grab samples for  carbon monoxide.
          Five (Orsat) determinations of C09,  0 ,  and CO were made, using
EPA Method 3 procedures--two from Unit No.  3  and  three from Unit No.  2.
The determinations made on Unit No.  3 were from duplicate samples, one
analyzed by the Cabot Corporation and the other by BCL.  A summary of the
Orsat data is presented in Table 8.

Moisture

          Moisture content of the gas stream was  determined by EPA
Method 4 procedures (measuring condensed water) during runs for deter-
mination of H?S and S0?.   Moisture results are reported in Tables 4
and 5.

-------
                                   21
                   TABLE 8.  SUMMARY OF ORSAT DATA
Gas Analysis.
Run No.
2A
2B<«>
4A
4B
4C
Unit No.
3
3
2
2
2
Date
12/11/74
12/11/74
12/12/74
12/12/74
12/12/74
Time
3:00 p.m.
3:00 p.m.
8:50 a.m.
11:40 a.m.
2:30 p.m.
C02
3.65
2.20
2.91
3.10
3.30
02
0.51
0.30
0.90
0.50
0.90
percent
CO
13.30
12.80
12.45
13.80
13.50
(a)   Analysis by Cabot.

-------
                                   22
Visible Emissions

          Visual determination of the opacity of emissions from Unit Nos. 3
and 2 were obtained by two qualified observers.  Two determinations were
made on each unit by both observers.  A summary of the visible emission data
are compiled in Table 9.  Observation records obtained by the observers are
appended to this report.

Infrared Telespectrometer Measurements

          The EPA infrared telespectrometer system, which bears the achronym
ROSE (Remote Optical Sensing of Emissions), consists of a conventional quarter-
meter grating monochromator, HgrGe and PbS detectors, associated electronic
amplification and recording systems, and a specially designed telescopic
optical system.  This system allows the infrared emission spectra of warm
(> 400 K) gaseous pollutants to be obtained from remote locations over the
near infrared spectral regions of good atmospheric transmittance.  For the
flare measurements the ROSE system, which is housed in a mobile laboratory,
viewed the flare at a slant range of 65 meters at an elevation angle of 30°.
Data were obtained both with the flare on and off and also at various heights
above the stack exit.  The results of these measurements are described in
Appendix B.  The ROSE system is described in detail in EPA Report R2-72-052.

                   Recommendations on Sampling Procedures

          The following comments are based upon observations made during
the stack-emission-sampling program at the Cabot Corporation carbon-black
plant in Waverly, West Virginia.

H S - EPA Method 11

          It appears that EPA Method 11 is not completely applicable (at
least without modification) in determining H_S emissions from stationary
sources where high concentrations of H?S  and moisture are prevalent.
The method is based upon procedures developed by the petroleum industry

-------
                  23
TABLE 9. SUMMARY OF VISIBLE EMISSIONS
Opacity.
Run
No.
2A
2B
3A
3B
Stack
No.
3
3
2
2
Minimun
Date
12/11/74
12/11/74
12/11/74
12/11/74

12:
15:
11:
14:
Time
48-13:
20-16:
02-12:
09-15:

48
20
02
09
Ob. A
5
5
5
5
Ob.B
5
5
5
5
percent
Maximum
Ob. A
15
15
40
30
Ob.B
15
15
40
40


Average
Ob. A
9.6
7.4
15.9
9.7
Ob.B
9.9
8.3
18.6
11.0

-------
                                   24
and natural gas producers-API Method 772-54.  It was designed for use in
gas streams where relatively low concentrations of HLS and moisture are
encountered.  The possibility of interference of other reduced sulfur
compounds should be investigated.
          There were indications that the high concentration of HLS
(727-1941 mg) encountered at the Waverly plant exceeded the capacity of the
three absorbing impingers in the sampling train.  This was visually observed
by the yellow color (CdS) in the impingers, which even occurred at sampling
rates considerably slower than those recommended in the Method.
          There are several practical modifications which can be made to
Method 11 where relatively high concentrations of H?S (> 100 ppm) are
anticipated.  One modification, requiring very little readjustment of
the procedure, would be to increase the strength of the H S absorbant
solution [Cd(OH) ] and the iodine-thiosulfate titrating reagent.
          When high moisture content (40 to 50 percent) is encountered in
the stack gas, as was found at the Cabot carbon-black plant in Waverly,
an additional empty impinger in the number two position (immediately behind
the 3 percent H_02 impinger) would  help prevent splashover of condensed
moisture and acid mist from the first impinger into the H S absorbing
impingers.
S02 - EPA Method 6
         The aforementioned moisture problem may also seriously effect SCL
determinations employing EPA Method 6.  An empty impinger located in the
number two position of this train would also help prevent splashover of
acid mist and sulfur trioxide.

-------