xvEPA
United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park NC 27711
EMB Report 78-OCM-6
May 19?6
Air
Benzene
Nitrobenzene
Manufacture
Emission Test Report
E. I. Dupont De Nemours
and Company
Beaumont, Texas
-------
EMISSION TEST OF A NITROBENZENE PLANT AT E. I. du PONT de NEMOURS
AND COMPANY, INC., BEAUMONT, TEXAS
by
George W. Scheil
William H. Maxwell
FINAL REPORT
EPA Contract No. 68-02-2814, Work Assignment No. 9
EPA Project No. 78-OCM-6
MRI Project No. 4468-L(9)
For
Emission Measurement Branch
Field Testing Section
Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Attn: Mr. J. E. McCarley, Jr.
-------
PREFACE
The work reported herein was conducted by Midwest Research Institute under
Environmental Protection Agency (EPA) Contract No. 68-02-2814, Work Assignment
No. 9.
The project was under the supervision of Mr. Doug Fiscus, Head, Field Pro-
grams Section, and Mr. William Maxwell, Program Manager. Mr. Maxwell served as
field team leader and was assisted in the field by Messrs. George Scheil and
Ron Jones, and Mr. Roy Neulicht of EPA, and in the lab by Messrs. George Scheil
and Dennis Wallace and Ms. Alice Shan.
Approved for:
MIDWEST RESEARCH INSTITUTE
L. J. Shannon, Executive Director
Environmental and Materials
Sciences Division
July 30, 1979
11
-------
CONTENTS
Preface ii
Figures • iv
Tables. v
1. Introduction 1
2. Summary and Discussion of Results. 2
3. Process Description and Operation. ........ . 19
4. Location of Sample Points 21
5. Sampling and Analytical Procedures 26
Appendices
A. Representative Sample GG Plots 31
B. Table of Retention Indices 35
G. Field Data 38
D. Report of Analyses of EPA Samples 70
E. Draft EPA Benzene Method 99
F. Computer Program Explanation and Listing ........... 114
G. Sample Calculation 119
iii
-------
FIGURES
Number Page
1 Block diagram of nitrobenzene production .. 20
2 General site diagram - du Pont/Beaumont 22
3 Vapor input stream, water scrubber - site No.l 23
4 Vapor output stream, water scrubber - site No. 2 24
5 Vapor outlet stream, waste acid tank - site No. 3 25
6 TGNMO sampling train 28
7 TGNMO sampling train leak check diagram 30
A-l Typical chromatogram, September 28, 1978, site No. 1 .... 32
A-2 Typical chromatogram, September 28, 1978, site No. 2 .... 33
A-3 Typical chromatogram, September 28, 1978, site No. 3 .... 34
IV
-------
TABLES
Number
1 Summary of Benzene Results of Integrated Bag Samples . . • . <
2 Summary of Nitrobenzene Results of Integrated Bag Samples. . . 4
3 Summary of Total Hydrocarbon (THC) Results of Integrated
Bag Samples* ...... 5
4 Summary of General Results, PPM as Benzene 7-9
5 Possible Peak Identifications 10
6 Scrubber Water Analyses* ................... 11
7 Sample Gondensate Organics Analysis* •• 12
8 Summary of NOX Results ..*...*.* 13
9 Summary of Stack Gas Data 14
10 Summary of TGNMO Test Results. 16
11 Comparison of Results for TGNMO and Integrated Bag Sampling. . 18
-------
SECTION 1
INTRODUCTION
This report presents the results of source testing performed during the
period September 25 to 29, 1978, by Midwest Research Institute (MRI) and the
U.S. Environmental Protection Agency (EPA) on the nitrobenzene facility of
E. I. du Pont de Nemours and Company, Inc., Beaumont, Texas. The portions of
the process sampled were the waste acid and organic receivers coming from the
process centrifuge. The waste acid receiver vapor stream was sampled before
being vented to the aniline incinerator. The waste organic receiver vapor
stream was sampled before and after the water scrubber. Samples were also ob-
tained of the liquid effluent of the water scrubber.
The vapor streams were analyzed for benzene, nitrobenzene, total hydro-
carbons (THC), and nitrogen oxides (NOX). Duct temperature, pressure, and
flow rate measurements were also made on the two vapor streams prior to the
control devices.
The results of these tests are to be evaluated by EPA as part of the de-
velopment of emission standards for this industry.
-------
SECTION 2
SUMMARY AND DISCUSSION OF RESULTS
The results of the analyses for benzene are given in Table 1. The benzene
peak was found to contain an impurity during the field analyses (the peak was
broader than normal). Further tests showed that benzene was present as a lead-
ing shoulder on a much larger peak. Later work showed that this larger peak is
probably cyclohexane. It is assumed that this leading shoulder is benzene. It
must be pointed out that later tests indicated that 1-methyl cyclopentene can-
not be resolved from benzene on the column used. However, the presence of this
compound is doubtful in this process. The location of the benzene peak was
proved by running mixtures of the sample gases with the benzene standard gases.
The benzene data presented in Table 1 are reliable within the limits of the
standard deviations shown. See Section 5 for a detailed description of the ana-
lytical technique used. The mass emission rates for Tables 1 to 3 are of limited
accuracy due to the inaccuracy of the measured flow rates.
The nitrobenzene peaks had a significant tailing on the column used. To
obtain the most representative data, the areas were measured by cutting and
weighing the peaks for nitrobenzene and the benzene standards. These results
are given in Table 2. The relative constancy of the concentrations, except for
No. 2 on the September 28 test, indicates that this component may be limited
by vapor pressure in these samples. Extrapolation of data from Lange's Handbook
of Chemistry indicates that at 20°C the expected vapor concentration of nitro-
benzene is about 400 ppm. The results in Table 2 are reported as benzene. How-
ever, the response factor for nitrobenzene versus benzene was determined to be
1.0+ 10% by injection of known'amounts of liquid into bags filled with measured
volumes of nitrogen.
Table 3 shows the THC results for each sample obtained by a short empty
column direct into the flame. Both benzene and propane standards were used.
Measurements were by peak height.
Table 4 shows the concentrations of all the observed peaks in the chroma-
tograms expressed as benzene. These measurements were by peak area, determined
as peak height times width at one-half the height. Since some of the peaks were
distorted, other measurement techniques such as planimetry or cutting and weigh-
ing of the chromatograms would be expected to yield more accurate results; how-
ever, these techniques are time consuming. An example is the differences in the
results by the different methods for nitrobenzene (peak 20) as shown in the re-
sults of Tables 2 and 4.
2
-------
TABLE 1. SUMMARY OF BENZENE RESULTS OF INTEGRATED BAG SAMPLES^-
a/
Date
September 26
September 27
September 28
Concentration—
ppm as benzene
Site concentration, std. deviation Average
No. 1 - organic receiver 490 + 75 545
scrubber inlet 6<)0 + 135
No. 2 - organic receiver 590 j- 90 545
scrubber outlet 500 + 110
No. 3 - acid receiver 230 + 38 210
vapor stream 205 + 41
200 + 110
No. 1 650 + 69 595
680 + 39
460 + 86
No. 2 540 + 47 565
590 + 69
No. 3 175 + 46 150
130 + 46
No. 1 130 + 80 135
110 + 33
160 + 87
No. 2 190 + 110 205
285'+ 56
140 + 37
No. 3 95+50 80
65 + 45
Emission ratc£' gram Honzciic
as benzene emltted/inceagram
Ib/hr kg/hr aniline produced
0.0917 0.0416 2.65^/
0.0917 0.0416 2-65-
0.4625 0.2098 13. 31!'-1'
0.1001 0.0454 2.90SL/
0.0950 0.0431 2.75^'
0.3303 0.1498 9. Sol/
0.0227 0.0103 1.31-?-'
0.0345 0.0156 1 . 99£/
0.1762 0.0799 10. 20S.I
a/ Condensatc trap instaLied prior to Integrated bag. Sec Table 7 and Section 5 for discussion.
t>/ Peak believed to be benzene but could also be I-methyl eyelopentene. See Section 5 for description of
measuring method.
c/ All mass emission data calculated using the avernge flow rate (Table 9) for each test site.
^!/ Production rate Is 107,100 Mg/yr aniline based on data supplied by the company. See Section 1 of this
report.
.>/ Production r;»te Is 53._S50 Mg/yr aniline.
-------
TABLE 2. SUMMARY OF NITROBENZENE RESULTS OF INTEGRATED BAG SAMPLE3»b/
Date
September 26
September 27
September 28
No
No
No
No
No
No
No
No
No
Site
. 1 - organic receiver
scrubber Inlet
. 2 - organic receiver
scrubber outlet
. 3 - acid receiver
vapor stream
. I
. 2
. 3
. 1
. 2
. 3
Concentration—
ppiu as benzene
260
289
210
258
292
285
246
82
340
Emission rate grain Nitrobenzene emitted (as
as benzene benzene) /Mcgagram aniline pro-
0
0
0
0
0
0
0
0
0
Ib/hr
.0437
.0486
.4625
.0434
.0491
.6276
.0414
.0138
.7487
0
0
0
0
0
0
0
u
0
kg/lir
.0198
.0220
.2092
.0197
.0223
.2840
.0188
.0063
.3387
du
1.
I.
13.
1.
1.
18.
2.
0.
43.
ced
26
40
39
26
42
16
40
80
33
a_/ Nitrobenzene results are reported as benzene. Instrument response factor for nitrobenzene was determined
to be within 107. of the instrument response to benzene.
h/ Condensate trap installed prior to integrated bag. See Table 7 and Section 5 for discussion.
c/ Concentration determined by cutting and weighing the chromatograms for benzene standard and the nitro-
benzene peak and comparing.
-------
TABLE 3. SUMMARY OF TOTAL HYDROCARBON (THC) RESULTS OF INTEGRATED BAG
Cone en true ion
Date
September 26'
September 27
September 28
Site
No. 1 - organic receiver
scrubber inlet
No. 2 - organic receiver
scrubber outlet
No. 3 - acid receiver
vapor stream
No. 1
No. 2
No. 3
No. I
No. 2
No. 3
As benzene
(ppm)
7,520
7,970
3,140
8,100
6,440
3,880
6,210
6,210
3,360
As propane
(l>pm)
9,320
9,870
3,890
11,400
9,050
5,460
8,580
8,580
4,640
As
Ib/hr
1.26
1.34
6.92
1.36
1.08
8.54
1.05
1.04
7.40
Emission rate
benzene
kg/hr
0.57
0.61
3.14
0.62
0.49
3.87
0.48
0.47
3.36
As propane
Ib/hr
0.88
0.94
4.84
1.08
0.86
6.78
0.81
0.81
5.77
kg/hr
0.40
0.43
2.20
0.49
0.39
3.08
0.37
0.37
2.62
gram Total
Hydrocarbons emitted
(as benzene) /megayram
aniline produced
36.37
38.92
200.34
39.56
31.26
246.92
61.25
59.97
428.74
aj Measured by peak height.
b/ CondensiiLe trap installed prior uo integrated bag. See Table 7 and Section 5 for discussion.
-------
TABLE 4. SUMMARY OF GENERAL RESULTS,- PPM AS BENZENE
Peak No.
September 26
1
2
3
4
5
6
7
8
9
10
11, llaS/
12
13
14
15
16
17
18
19
20!/
Total
THC result
No. Ik/
4.6
6.4
\J 9 *T
34.6
40.6
9.7
2,110
199
251
1,480
1,830
465
615
-
132
271
108
45.8
_
232
7,830 (104%)
7,520
Site
No. 2£/
3.6
1.7
2.2
28.7
33.9
8.1
1,490
85.2
145
1,230
1,530
250
355
,
109
306
84.5
42.3
_
200
5,910 (74%)
7,970
No. 3£L/
3.6
0.2
0.5
4.7
8.5
2.4
547
167
53.1
430
703
126
189
-
53.9
116
52.4
33.3
_
141
2,630 (93%)
3,140
(from Table 3)
(continued)
6
-------
TABLE 4. SUMMARY OF GENERAL RESULTS,- PPM AS BENZENE (continued)
Peak No.
September 27
1
2
3
4
5
6
7
8
9
10
11, llaS/
12
13
14
15
16
17
18
19
20I/
Total
THC result
No. lb_/
4.2
1.2
0.6
26.3
27.0
5.7
1,140
48.0
130
1,100
1,450
276
133
-
127
233
107
47.6
0.9
149
5,010 (62%)
8,100
Site
No. 2£/
4.6
1.3
1.3
23.3
25.9
6.6
1,220
73.2
122
952
1,270
193
108
-
101
199
91.5
38.7
mm
142
4,570 (71%)
6,440
No. 3d./
1.8
-
-
3.7
5.2
1.9
412
22.4
43.1
407
632
105
150
-
52.6
138
49.6
31.8
149
2,060 (53%)
3,880
(from Table 3)
(continued)
7
-------
TABLE 4. SUMMARY OF GENERAL RESULTS,- PPM AS BENZENE (continued)
Peak No.
September 28
1
2
3
4
5
6
7
8
9
10
11, llaSJ
12
13
14
15
16
17
18
19
20f/
Total
THC result
[from Table 3)
No. Lk/
3.0
1.0
0.9
19.1
22.6
6.3
1,340
81.3
141
885
1,320
210
443
7.1
120
279
96.1
70.1
3.3
315
5,360 (86%)
6,210
Site
No. 2£/
2.7
0.8
0.8
18.5
20.0
5.8
1,170
73.9
122
867
1,220
195
345
-
Ill
248
110
55.2
70.2
4,640 (75%)
6,210
No. 3d./
2.1
.
-
3.5
4.5
1.1
397
22.9
40.9
354
574
105
192
-
57.5
148
68.8
45.3
178
2,190 (65%)
3,360
a/ Integrated bag sample analysis with condensate trap installed prior to the bag.
See Section 5 for discussion.
b/ Organic receiver scrubber inlet.
c/ Organic receiver scrubber outlet.
d/ Acid receiver vapor stream.
e/ Sum of benzene and superimposed peak.
f_/ Nitrobenzene.
8
-------
The two compounds of greatest interest to EPA were benzene and nitroben-
zene, and these results are shown in Tables 1 and 2, respectively, using the
more appropriate measurement techniques. The other compounds listed in Table 4
are provided as secondary information to this test program. Because of this
secondary interest only, these data were reported using the much quicker method
of data reduction. For comparison purposes, the nitrobenzene (peak 20) results
in Table 4 were calculated using the same data reduction technique as was em-
ployed for the other peaks shown in Table 4. It should also be understood that
all concentrations are reported as if each of these compounds was benzene; the
instrument response factor for each of these compounds was not determined. Some
typical chromatograms are shown in Appendix A. Some of the 60 hydrocarbons
searched for possible identication of the various peaks are listed in Table 5.
The retention indices for all of the compounds measured in the search are given
in Appendix B.
Table 6 shows the results of analyses of the three scrubber water samples
obtained during the test after extraction into carbon disulfide. The scrubber
water flow rate was 34 liters/min (9 gal/min).
Table 7 compares the benzene, nitrobenzene, and THC results obtained from
the integrated gas samples with the results of analyses of the composite sam-
ples obtained from the condensate traps in each of the sampling lines. The
equivalent vapor concentrations represent the concentrations resulting from
vaporizing the material in a volume of gas equal to the estimated total inte-
grated gas volumes. The main conclusion from this table is that the vapor stream
concentrations are similar on the inlet and outlet of the scrubber but the
entrained liquid is greatly reduced by the scrubber.
Table 8 presents the results of the NOX analyses. The value of these re-
sults is questionable as all of the sample lines contained liquid, possibly
nitric acid. Twenty- to 200-fold dilutions were required for the analyses.
Table 9 presents a summary of the stack gas data used in the emission cal-
culations for the other tables. The flow rate measurements have limited accuracy
due to the liquid entrainment and the low AP pitot readings. Therefore only an
average for all 3 days is presented.
All field data (excluding TGNMO) may be found in Appendix C. Sample calcu-
lations are given in Appendix G.
During this test program emission testing was conducted for Volatile Organic
Carbon (VOC) using a test method under development by the EPA. The test work
was conducted by EPA personnel. The purpose of the tests was two-fold. First,
hands-on experience with the method during actual field use was desired so that
problem areas could be identified. Secondly, comparative data obtained with
the TGNMO and integrated bag/THC sampling methods were desired. Since results
from the two sampling procedures would be compared, simultaneous samples were
conducted from the same sampling location.
-------
TABLE 5. POSSIBLE PEAK IDENTIFICATIONS
Peak No. Retention index Possible component
1
2
3
4
5
6
7
8
9
10
11- .
lla-
12
13
14
15
16
17
18
19
20
100-200
300
400
480
500
525
550
580
600
620
650
655
670
680
690
700
715
725
735
-
™
Air, methane, ethylene, acetylene, ethane
Propylene, propane
Isobutane, 1-butene, 2-methylpropene,
n-butane
2-Methylbutane
n-Pentane, 1-pentene, 2-methyl-l-butene
2 ,2 -Dimethylbutane
4-Methyl-l-pentene, cyclopentane, 2 -methyl-
pentane, 2,3-dimethylbutane, 4-^methyl-2-
pentene (trans)
3-Methylpentane
n-Hexane, 2-ethyl-l-butene
Methyl cyclopentane, 3-methyl-2-pentene
(trans), 2-hexene (cis)
Benzene, 1-methyl cyclopentene
Cyclohexane
2 ,3-Dimethylpentane
Cyclohexene, 3-methylhexane
2 ,2 ,4-Trimethylpentane
1-Heptene, 3-heptene
Methyl cyclohexane, 2-heptene (cis)
2 ,4-Dimethylhexane, 2 ,5-dimethylhexane
2 ,3,4-Trimethylpentane
Unknown
Nitrobenzene
a/ Peak 11 is the smaller lead shoulder on Peak lla.
10
-------
TABLE 6. SCRUBBER WATER ANALYSES
Sample
1
2
3
Date taken
September 27
September 28
September 28
Benzene
jxg/ml kg/hrS/
8.7 0.0177
1.1 0.0022
1.5 0.0031
Nitrobenzene
jug/ml
1,400
360
475
as benzene
kg/hr§/
2.85
0.74
0.97
Other organics
/xg/ml as benzene
110
90
85
kg/hr£/
0.224
0.184
0.173
a/ Water flow rate of 34 ^/min (9 gal/min) used to calculate mass emission rate.
-------
TABLE 7. SAMPLE CONDENSATE ORGANICS ANALYSIS
Benzene:
Total present in condensate (ing)-'
Condensate equivalent vapor con-
centration (ppm)3jJi'
Integrated gas analysis average
Total benzene present in stream
(ppm)aji/
Nitrobenzene:
Total present in condensate (nig)S.'
Condensate equivalent vapor con-
centration (ppm)aik'
Integrated gas analysis average
Total nitrobenzene present in
stream (ppm)2jJi'
Total hydrocarbons:
Total present in condensate (rig-
as benzene)
Condensate equivalent vapor con-
centration (ppm)JiP/
Integrated gas analysis average
(ppm) -
Total hydrocarbons present in
stream (ppm)3.ii?/
Site No. 1
scrubber inlet
10
16
425
441
11,300
11,680
255
11,835
11,300
11,600
7,300
18,900
Site No. 2
scrubber outlet
0.18
0.3
440
440
25
29
220
249
46
69
6,900
6,970
Site No. 3
acid receiver
0.037
0.1
145
145
375
480
280
760
377
482
3,500
3,980
ji/ Condensate suspected to be primarily entrained liquid.
Sample was not collected Isokinetlcal ly.
b/ Calculated ns benzene.
-------
TABLE 8. SUMMARY OF NOX RESULTS
NOX emissions^/
Date
September 26
September 27
September 28
Site Run No.
No. 1 1
Organic receiver
scrubber inlet
No. 2
Organic receiver
scrubber outlet
No. 3
Acid receiver
vapor stream
No. 1 2
No. 2
No. 3
No. 1 3
No. 2
Ho. 3
ppui
35,700
19,200
4,120
3,450
10,800
10,000
25,200
9,670
3,530
3,280
8,500
8,330
27,400
6,390
. 3,450
3,450
11,000
5,550
Ib/dscf
0.00425
0.00228
0.00049
0.00041
0.00128
0.00119
0.00300
0.00115
0.00042
0.00039
0.00101
0.00099
0.00326
0.00076
0.00041
0.00041
0.00131
0.00066
ing/dscm
68,100
36,500
7,850
6,570
20,500
19,100
48,100
18,400
6,730
6,250
16,200
15,900
52,200
12,200
6,570
6,570
21 ,000
10,600
Ib/hr
3.53
1.89
0.41
0.34
13.92
12.94
2.50
0.95
0.35
0.32
10.98
10.76
2.71
0.63
0.34
0.34
14.24
7.18
kg/hr
1.60
0.86
0.18
0.15
6.31
5.88
1.13
0.43
0.16
0.15
4.99
4.89
1.23
0.29
0.15
0.15
6.47
3.26
<>/ All sample lines contained moisture, possibly nitric acid, so all values may be questionable.
-------
TABLE 9. SUMMARY OF STACK GAS DATA
Stack Stack gas
Stack
temp* velocity head Molecular pressure, absolute
Site
No. 1
Organic receiver
scrubber inlet
No. 2
Organic receiver
scrubber outlet
No. 3
Acid receiver
vapor stream
°F in 1^0 weight, stack
(°C) (inm 1120) gas
UQi/ 0.002 28.60
(43) (0.051)
80JL/ 28.60
(27)
120S/ 0.33 28.00
(49) (8.41)
in Hg
(uuu Hg)
30.13£/
(765)
30. 13-'
(765)
30. l^7
(765)
Mole fraction
dry gas
0.914^
0.966-
0.899-'
Stack Clow rate
dsct/min
(dscm/rain)
13.84H/
(0.3918)
13.84"7
(0.3918)
181.21-
(5.131)
a/ Average of three runs.
J>/ All streams saturated.
c/ IJ readings were very low; flow ratts have poor accuracy.
-------
A "t" - connector was placed in the integrated bag sampling line (after the
condensate trap) and the TGNMO sampling train was connected to this "t." In
normal practice the condensate trap used in the sampling line during these
tests woi/ld not be used for TGNMO sampling; the TGNMO sampling probe would be
placed directly into the duct or would be attached (as close to the duct as
possible) to any permanent probe tip located in the duct. However, since a
primary purpose of these tests was to compare the results of the TGNMO analy-
sis directly with the results of the integrated bag total hydrocarbon (THC)
analyses, it was necessary that the two systems sample the same conditioned
gas stream.
In general, the TGNMO sampling method consists of pulling a sample from
the duct through an organic condensate trap (-78°C) into an evacuated tank.
The organic contents of the condensate trap and the non-methane organic frac-
tion of the gas sampling tank are analyzed by oxidation to carbon dioxide (CO )
with subsequent measurement of the CO by a non-dispersive .infrared (NDIR)
analyzer; results are expressed as parts per million carbon (ppmc).
Table 10 summarizes the results obtained with the TGNMO method. The test
location, run number, sample volume, sample time, volatile organics, condens-
able organics, and total gaseous non-methane organic' concentrations (ppmc) are
presented. The volatile organics are those non-methane organics measured in
the tank fraction of the sampling train, while the condensable organics are
those organics collected in the condensate trap fraction. The total gaseous non-
methane concentration is simply the sum of the two sample fractions. The com-
plete laboratory report is included in Appendix D.
During the sampling, a problem was encountered with sample flow blockage.
A problem with flow blockage was experienced during runs 1 and 2 at the or-
ganic vent scrubber outlet and with run 1 at the waste receiver vent. The prob-
lem with flow blockage at the waste receiver vent was probably due to packing
the condensate trap too deeply in dry ice; particular care was taken to main-
tain the proper dry ice level during runs 2 and 3 and no problems were encoun-
tered. In an attempt to maintain a constant flow during this test run (Waste
Receiver-Run 1), the sample flow control valve was completely opened near the
end of the test; this resulted in a situation of initially no sample flow due
to blockage followed by an extremely rapid sample flow (half the tank volume
in less than 10 min). Apparently, this situation did significantly affect the
results for this test run; an abnormally high result was obtained for the tank
portion of the sample train (see Table 10). The problem with flow blockage at
the organic vent scrubber outlet was probably caused by an excessive amount
of moisture in the effluent stream. Both the first and second test runs at this
sample location were short due to the blockage problem. The results (see Table
10) for these two test runs (especially run 2) are lower than for the third
run, during which no blockage occurred. In order to prevent flow blockage dur-
ing run 3, a condensate trap maintained at a temperature of 5°C was added to
15
-------
TABLE 10. SUMMARY OF TGNMO TEST RESULTS
Sample
location
Organic
venc
inlet
Organic
venc
ouclec
Waste
acid re-
ceiver
Sun No.
1
2
3
1
2
3
IS/
2
3
Dace
9/26
9/27
9/28
9/26
9/27
9/28
9/26
9/27
9/28
Sample
volume
(cc)
3,457
4,082
4,543
1,171
1,079
4,339
4,564
3,506
4,312
Samole
time
(min)
35
100
105
30
25
105
45
38
105
Volatile
Volatile
1,390
1,470
1,980
130
530
1,520
20,970
1,830
1,060
orsanic carbon, oom Ci
Condensable
36,300
46,450
21,860
18,020
5,270
30,627k/
22,190
14,980
16,620
Monntethane
38,190
47,920
23,340£/
13,150
5,800
32,147
43,160
16,810
17,680
a/ Comparison of Pose Tesc Field and Laboratory vacuum readings indicates this tank leaked;
therefore, reported results are lower than true value; result calculated on post-case
vacuum reading yields 27,390 ppm.
b/ 3,457 ppm for first condensate trap (5°C); 27,170 for second condensace trap C-78°C).
c/ Sample rate changed significantly during run; may have affected results.
16
-------
the sampling train just preceding the condensate trap packed in dry ice. This
appears to have alleviated the blockage problem since the third test run was
conducted without any abnormal occurrences. The condensable organic valve re-
ported in Table 10 for run 3 is the sum of the organic analysis for both traps,
The other anomaly which occurred during these tests was a leak in one of
the sample tanks. Upon arrival at the test site, it was noted that two of the
evacuated tanks had leaked during shipping. The tank fittings were tightened
and both tanks were reevacuated; one tank still had a small leak. Examination
of the post-test vacuum readings taken on this tank in the field and in the
laboratory indicate the tank leaked during shipping. The organic concentra-
tions value reported by the laboratory is based on the larger sample volume
(i.e., actual sample volume plus leakage volume). If the concentration values
are recalculated using the sample volume measured in the field, the measured
concentrations are 2,317 ppm volatiles, 25,572 ppm condensables, and 27,889
ppm total non-methane organics.
For comparison purposes, Table 11 presents the results obtained by both
the TGNMO method and the integrated bag sample method. The total hydrocarbon
(THC) values for the integrated bag samples were obtained by directly analyz-
ing the bag sample with a flame ionization detector (FID) calibrated with
propane. The results obtained with the FID were multiplied by three (3) to
convert the data to a basis of ppm C. ; this value is reported in Table 11.
17
-------
TABLE 11. COMPARISON OF RESULTS FOR TGNMO AND INTEGRATED BAG SAMPLING
Volatile organic carbon—' Total hydrocarbon, £/
Sample location
Organic vent
inlet
Organic vent
outlet
Waste acid
receiver
Run
1
2
3
1
2
3
1
2
3
ppm Ci
38,190
47,920
27,890
18,150
5,800
32,147
43,160
16,810
17,680
a.— u
a
+ 27
+ 28
+ 8
-63
-308
+ 20
+ 73
+ 3
+ 21
ppm C
27,
34,
25,
29,
27,
25,
11,
16,
13,
1
960
200
740
610
150
740
670
380
920
a/ TGNMO method.
b/ Integrated bag method, flame ionization detection with propane calibration
X3.
18
-------
SECTION 3
PROCESS DESCRIPTION AND OPERATION
The nitrobenzene-aniline production facility at Beaumont was built in
1972 and is a single train of equipment with a nameplate capacity of 145,000
metric tons/year of aniline (all nitrobenzene produced is used to make ani-
line).
The starting materials, benzene and nitric acid, are reacted in the liq-
uid phase under carefully controlled conditions in the presence of concentrated
H2S04» The resulting crude nitrobenzene and spent acid phases are then sepa-
rated. The spent acid is extracted with benzene to recover nitrobenzene and
then concentrated in a direct contact t^SO^ concentrator. Concentrated acid
is recycled to the nitration reactor. The benzene extractant also is directed
to the nitration reactor. Crude nitrobenzene from the separation step is washed
with recycled water and neutralized. The aqueous waste from this step is sent
to a wastewater treatment facility. Benzene and water are stripped from the
neutralized nitrobenzene stream in a vacuum distillation column. The benzene
is recycled to the nitration step, and the purified nitrobenzene is forwarded
to intermediate storage to await use in the aniline process. Figure 1 is a sim-
plified flow diagram for the process.
During the sampling runs, the plant was operating stably at a constant
rate as determined by monitoring feed rates to the nitration reactors. These
rates were as follows:
a/
Annualized aniline equivalent-
Date (Mg/yr)
9/26/78 107,100
9/27/78 107,100
9/28/78 53,550
a/ Assuming the plant operates 6,833 hr/yr.
19
-------
Recycle Benzene
N)
O
A Vent
.@-L
Vents
1 U
Liquid -Liquid
Separation
and Extraction
Water and Vent
Base
Crude
NB
Nitrobenzene
Wash and
Neutralization
Vent
Vent
I 1
to WWT
Waste Acid
Storage
Sulfuric
Acid
Concentration
Acid Purge
Nitrobenzene
Storage
Figure 1. Block diagram of nitrobenzene production.
-------
SECTION 4
LOCATION OF SAMPLE POINTS
Figure 2 presents a general site diagram of the sampling location. All
of the sample streams were carried to ground level from the respective ducts
or sources via stainless steel sample lines.
Figure 3 shows a diagram of sampling site No. 1, the waste organic vapor
input stream to the water scrubber. An existing 1.77 cm (0.5 in.) stainless
steel sampling line was used. As shown in Figure 2, this line tapped the va-
por line approximately 4.3 m (14 ft) above ground level. The "S"-shaped pitot
tube was located in the center of the 10.2 cm (4 in.) stainless steel duct ap-
proximately 0.6 m (2 ft) above the waste organic receiver. A mercury thermom-
eter was taped to the exterior of the duct at approximately the same location
as the pitot tube.
Sampling site No. 2, the vapor output stream from the water scrubber, is
shown in Figures 3 and 4. A 0.6 cm (0.25 in.) stainless steel line was run from
the top of the scrubber to the ground for sampling purposes. No flow or temper-
ature measurements of the air flow were made at this point. Effluent water sam-
ples were taken from the valve in the seal leg, and water flow rates were taken
from plant instrumentation.
Figure 5 presents a diagram of sampling site No. 3, the vapor stream off
the waste acid receiver. A 0.6 cm (.0.25. in.) stainless steel line was again
run from the sample tap to ground level for sampling. The "S"-shaped pitot was
located in the center of the 5 cm (2 in.) stainless steel duct approximately
1.2 m (4 ft) above the waste acid receiver. A mercury thermometer was used as
for the waste organic line for temperature measurements.
21
-------
Sample
Site #3
To Incinerator
r= Water Spray In
ll
Vapor
Packing
Liquid
Fraction
Waste
Acid
Receiver
Figure 2. General site diagram - Du Pont/Beaumont.
22
-------
1/4"
Stainless
Steel.
Sampling
Line
20"
Sample
Site #2
1/2" Stainless
Steel Sampling
Line
Sample
Site #1
4" Stainless
Steel
Organic
Vapors
Organic
Recovery
Tank
Organic
Solvent
Out
Figure 3. Vapor input stream, water scrubber - site No. 1.
23
-------
-Vapor Exhaust
1/4" Stainless Steel-
Sample Line
Liquid.
Level
Sea I Leg
Spray Water In
Teflon Sheet Cap
(Not Sealed)
Ceramic Ball
Packing
10"O.D.
t
Organic
Vapors
In
-Effluent Water Sample Point
Figure 4. Vapor output stream, water scrubber - site No. 2,
24
-------
1/4" Stainless.
Steel Sample
Line
Sample
Site #3
Acid Out
1" Flanged Gate Valve
Pitot Tube Location
1/2" Flanged Gate Valve
4"
Waste Acid
Receiver
Figure 5. Vapor outlet stream, waste acid tank - site No. 3.
25
-------
SECTION 5
SAMPLING AND ANALYTICAL PROCEDURES
The hydrocarbon samples were obtained according to the September 27,
1977, Environmental Protection Agency (EPA) draft benzene method (Appendix F).
Seventy liter aluminized Mylar bags were used with sample times of 1.5 to 2 hr.
The sample box and bag were both heated to approximately 49°C (120°F) using
copper steam lines and insulation. The sample lines were also insulated and
steam heated. Only a short segment of the line just prior to the sample bag was
left unheated. A midget glass impinger was fitted into the line ahead of the
bag to serve as a condensate knockout trap. This was done to remove as much liq-
uid as possible from the sample stream before the bag. Between the bag and the
condensate trap, a "tee" was placed in the line to allow for simultaneous test-
ing by EPA using a different sampling system. Both the sample box and bag were
leak-checked prior to each of the three runs. The boxes were transported to the
field lab immediately upon completion of sampling. They were heated in the lab
using an electric drum heater until all GC analyses had been completed.
Fyrite analysis of the two vapor streams was done for oxygen and carbon
dioxide content.
NOX samples were obtained from the three sample lines according to EPA
Reference Method 7 (Federal Register. Vol. 42, No. 160, August 18, 1977). Two
samples per run were obtained following the completion of the hydrocarbon runs.
The sample lines were not heated during this testing.
The samples were recovered in the field lab, transferred to shipping bot-
tles, and trucked to MRI for analysis according to Method 7.
Duct velocity measurements were obtained from locations one and three.
No significant (0.002 in. ^0) flow was ever detected from the waste organic
vapor stream (site No. !)• Difficulties were encountered at the waste acid vapor
stream (site No. 3) because of the vapor saturation conditions existing in the
duct. The pitot lines continually filled with liquid, even when knockout traps
were employed. Finally, on the final day, a reverse nitrogen purge was used to
clear the lines so that several spot readings could be obtained. Temperature
readings from both ducts were obtained by taping and insulating a mercury ther-
mometer to the duct.
26
-------
All of the organic analyses were performed on a Varian Model 2440 gas
chromatograph with flame ionization detector, column bypass valve, and a gas
sampling valve with matched, heated 2-cc sample loops. The carbon disulfide
extracts were injected by syringe into the injection port. The column used dur-
ing the field analyses was a 2 m x 1/8 in. OD stainless steel column packed
with 3% SP-2100 on 100/120 mesh Supelcoport. A 2 m x 1/8 in. OD nickel column
packed with 20% SP-2100/0.1% Carbowax 1500 on 100/120 mesh Supelcoport was used
for the compound identification and liquid extract analyses. This mixed phase
column gives similar results to the original column but with much less peak
tailing and better resolution. Both columns were programmed from 0 to 40° at
4°/mm and 40 to 160° at 10°/mm (200° limit on the first column).
To measure the benzene concentrations, computer calculations were neces-
sary. Each peak was broken into 3-sec wide segments and the height at each of
these segments was measured. These readings were then entered into a BASIC
computer program (see Appendix F) which superimposed each sample on the stan-
dard peaks and determined the maximum residual for the minor component (ben-
zene). A series of two unknown simultaneous equations were then solved about
the points of maximum residual and the peak maximum. The true value of the
minor component was the set of solutions having the smallest square of the re-
sulting deviations assuming both peaks have a peak shape identical to the
standard peak. The benzene concentrations were then calculated by the ratio of
the calculated peak heights to the standard peak height. The standard devia-
tions reported are for the set of solutions obtained for that day's standard
peaks (three to six standards). The data resulting from these calculations are
reliable within the limits of the standard deviations.
The THC results were obtained with the column bypassed with a short cap-
pillary tube and comparing the response against 824 ppm benzene and 2,010 ppm
propane standards. The integrated gas samples were compared to the 824 ppm ben-
zene standard.
The condensate samples (composites of all 3 days) and the scrubber water
samples were extracted twice with carbon disulfide, and 2 |il samples of the di-
sulfide extract were injected into the gas chromatograph. Liquid standards were
used for these samples containing 100 and 1,000 /j,/^ of benzene and nitro-
benzene. Twenty milliliter portions of the scrubber samples and the condensate
composite at the scrubber outlet were extracted with two, 5_ml portions of C$2 •
The other composites were extracted entirely with two, 5-ml portions of CS2«
Figure 6 is a schematic of the TGNMO sampling train. The TGNMO sampling
method is based on the Total Combustion Analysis test procedures established
by the Southern California Air Pollution Control District.^ During these
!_/ "Total Combustion Analysis: A Test Method for Measuring Organic Carbon,'
Salo, Albert £•; Oaks, William L.; MacPhee, Robert D.; Air Pollution
Control District - County of Los Angeles, August 1974.
27
-------
Connector
Probe
N>
00
Condensate Trap
Dry Ice
Rotameter
\
Flow Control
Connector
On/Off Valve
Vacuum/Pressure
Gauge
Evacuated
Sample Tank
Figure 6. TGNMO sampling train.
-------
tests the sample was drawn from the sample line through a chilled (-78°C)
stainless steel condensate trap into a 6-liter evacuated aluminum gas collec-
tion tank. A rotameter and fine adjust flow control valve were installed be-
tween the condensate trap and the evacuated tank to monitor and control the
sampling rate. During testing a constant sampling rate of 50 cc/min was main-
tained. The condensate trap was partially submerged in crushed dry ice during
the test and was kept packed in dry ice until analysis. During the last sample
run at the organic receiver scrubber outlet, an additional condensate trap was
added immediately preceding the regular trap and was kept in a water bath main-
tained at a temperature of 5°G. This trap was added to the sampling train to
prevent sample gas flow blockage caused by freezing of the gas stream moisture
in the inlet to the dry ice condensate trap. This problem occurred only at the
organic vent scrubber outlet where an excessive amount of moisture was present
in the effluent stream.
All sample tanks were evacuated in the laboratory prior to shipment to
the field site. The sampling trains were assembled in the field and each train
was leak-checked prior to testing. The pretest leak check was conducted by at-
taching a vacuum pump and manometer (see Figure 7) to the probe tip and then
evacuating the sampling train (with the exception of the evacuated tank) to a
vacuum of at least 625 mm Hg» After the train was evacuated, the leak check
valve 2 was closed and the train was left under vacuum for a 5-min period. Any
change in vacuum as measured on the manometer was recorded as the leak rate.
A leak rate of less than 0,1 in. mercury for a 5-min period was considered ac-
ceptable. The evacuated sample tank assembly was leak-checked prior to assem-
bling the train by simply assuring that the vacuum gauge on the tank did not
indicate any change in vacuum in a 1/2-hr period. A post-test leak check was
conducted on all sampling trains; in order to perform the post test leak check,
the leak check apparatus (with valve 2 closed) was connected to the probe tip.
The flow control valve to the tank was then opened until the sampling train
reached a constant vacuum; the sample tank flow control valve was closed, and
the vacuum measured on the manometer was recorded and monitored for a 5-min
period. Any change in the vacuum was recorded as the post-test leak rate.
The analytical work was performed by Truesdail Laboratories, Inc. Total
gaseous non-methane organics (TGNMO) were determined by combining the analyt-
ical results obtained from independent analyses of the condensate trap and the
evacuated tank sample fractions. The organic contents of the condensate trap
were oxidized to carbon dioxide (002) which was quantitatively collected and
then measured by a non-dispersive infrared (NDIR) analyzer. A fraction of the
sample collected in the evacuated tank was injected into a gas chromatograph
in order to achieve separation of the 'non-methane organics from carbon monox-
ide, carbon dioxide, and methane. Once separated, the four fractions were
oxidized to carbon dioxide and separately measured with the NDIR analyzer.
The volume of sample collected was calculated from vacuum and pressure read-
ings of the sample tank taken before and after sampling. The measured CO^ con-
centrations and the sample volume were used to calculate the volatile organic
carbon concentration of the source effluent as parts per million carbon (ppm
29
-------
Control
Valve
Connector
Control
Valve
2
Vacuum
Line
Mercury
Manometer
Bypass
Valve
O/acuum
'ump
Figure 7. TGNMO sampling train leak check diagram.
30
------- |