Report No. 76-AKM-iI
                                                 Without Appendices
   CD
O
                                           Final:' Report
                                        Elk Roofing  Company
                                        Stephens, Arkansas
                 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
PARTICULATE AND GASEOUS EMISSIONS FROM THE
 ASPHALT ROOFING PROCESS BLOWING STILL AT
          THE ELK ROOFING PLANT
            STEPHENS, ARKANSAS
                    to
      ENVIRONMENTAL PROTECTION AGENCY
               May 7, 1977
                    by
        P.  R.  Webb, R.  E. Barrett,
      W. C. Baytos, and S. E.  Miller
          Contract No.  68-02-1409
                Task No.  31
                 BATTELLE
           Columbus Laboratories
              505 King Avenue
           Columbus,  Ohio  43201

-------
                            TABLE OF CONTENTS
INTRODUCTION 	      1

SUMMARY OF RESULTS	     11

     Particulate Emissions 	     11
     Gaseous Hydrocarbon Emissions 	     12
     SO  and NO  Emissions	     20
     CO Emissions	     22
     Gaseous Emissions - Analysis of Flask Samples 	     22
     POM Emissions	     22
     Aldehyde Emissions  	     25
     Process Samples Analyses  	     25
     Visible Emissions 	     25

PROCESS DESCRIPTION AND OPERATION  	     29

     Process Description 	     29
     Emission Control System 	     31
     Process Operation 	     33
     Process Instrumentation 	     36
     Process Conditions During Testing 	     39

LOCATION OF SAMPLING POINTS  	     42

SAMPLING AND ANALYTICAL PROCEDURES 	     45

     Particulate Sampling  	     45
     Particulate Analysis  	     52
     Hydrocarbon Measurements  	     55
     Carbon Monoxide Measurements  	     57
     Continuous S0?/N0  Measurements 	     58
     Gas Composition Flask Sampling and Analysis 	     60
     POM Sampling and Analysis	     60
     Aldehyde Sampling and Analysis  	     62
     Visible Emission Measurements 	     63
     Analysis of Process Sampling and Recovery Oil 	     65
                               APPENDIX A

FIELD DATA RELATED TO PARTICULATE SAMPLING	     A-l

                               APPENDIX B

COMPLETE PARTICULATE RESULTS WITH SAMPLE CALCULATIONS  	     B-l

-------
                             TABLE OF CONTENTS
                               (Continued)

                                                                      Page

                               APPENDIX C

GASEOUS HYDROCARBON FIELD DATA AND CALCULATED EMISSIONS AT EACH
  SAMPLING POINT	     C-l

                               APPENDIX D

CO, NO , SO  FIELD DATA AND CALCULATED EMISSIONS	     D-l

                               APPENDIX E

EPA DRAFT METHOD FOR DETERMINATION OF PARTICULATE AND TOTAL
  GASEOUS HYDROCARBON EMISSIONS FROM THE ASPHALT ROOFING
  INDUSTRY	     E-l

                               APPENDIX F

POM FIELD DATA AND CALCULATED EMISSIONS	     F-l

                               APPENDIX G

VISIBLE EMISSIONS  	     G-l

                               APPENDIX H

ALDEHYDE EMISSIONS - DATA AND CALCULATIONS 	     H-l

                               APPENDIX I

LABORATORY RESULTS 	     1-1

                               APPENDIX J

PROCESS DATA	     J-l

                               APPENDIX K

BATTELLE SAMPLING CREW TEST LOG	     K-l

                               APPENDIX L

PROJECT PARTICIPANTS 	     L-l

-------
                            LIST OF TABLES


                                                                   Page

Table 1.  Sampling Task Log	    4

Table 2.  Performance Summary of Emission Reduction System-
   Saturant Blows 	   14

Table 3.  Performance Summary of Emission Reduction System -
   Coating Blows  	   15

Table 4.  Inlet Results, Elk Roofing, Stephens, Ark. (English
   Units)	   16

Table 5.  Outlet Results Elk Roofing, Stephens, Ark. (English)   .   17

Table 6.  Inlet Results, Elk Roofing, Stephens, Ark. (Metric
   Units)	   18

Table 7.  Outlet Results Elk Roofing, Stephens, Ark. (Metric
   Units)	   19

Table 8.  S02 and NOX Readings by Continuous Monitoring
   Analysis	   21

Table 9.  Analysis of Flask Sample from Elk Roofing Products
   Plant (EPA Sample S76-006-034)	   23

Table 10.  Summary of POM Data for Elk Roofing Plant	   24

Table 11.  Aldehyde Results for Elk Roofing Plant	   26

Table 12.  Ultimate Analysis of Process Sample from Elk Roofing
   Plant	   27

Table 13.  Summary of Visible Emission Data Elk Roofing Co.,
   Stephens, Arkansas 	   28

Table 14.  Summary of Process Data  .	   34

Table 15.  Summary of Mass Collected in Front Half of Stack Gas
   Sampling Train, Elk Roofing, Stephens, Arkansas - August 19-26,
   1975	   56

-------
                           LIST OF FIGURES
                                                                   Page
Figure 1.  Block Diagram Showing Relative Locations of Process
   Components and Sample Points 	    3

Figure 2.  Still Layout Showing Pertinent Features and Process
   Monitoring Instrumentation Elk Roofing 	   30

Figure 3.  Schematic of Layout of Asphalt Blowing Stills,
   Incinerator, and Waste Heat Boiler at Elk Roofing Plant  ...   32

Figure 4.  Emission Control System Layout Showing Monitoring
   Instrumentation Elk Roofing  	   37

Figure 5.  Water Flowmeter Elk Roofing  	   38

Figure 6.  Afterburner Inlet Duct Configuration Showing
   Sampling Port and Sample Point Locations 	   43

Figure 7.  Afterburner Outlet Stack Configuration Showing
   Sampling Ports and Sample Point Locations  	   44

Figure 8.  Blowing Still Process Sample Train Configuration-
   Inlet  	   49

Figure 9.  Blowing Still Process Sample Train Configuration-
   Outlet 	   51

Figure 10.  Cold Trap Apparatus Used to Evaporate TCE From
   Probe Wash for Inlet Samples for Elk Roofing Plant	   53

Figure 11.  Sampling Train for Determining CO Emissions at the
   Afterburner Outlet 	   59

Figure 12.  POM Sample Train Configuration for Asphalt Blowing
   Still Process  	 .........   61

Figure 13.  Relative Position of Visible Emission Observers with
   Respect to the Outlet Stack and Adjacent Structures  , , .  ,  .   64

-------
                PARTICULATE AND GASEOUS EMISSIONS FROM THE
                 ASPHALT ROOFING PROCESS BLOWING STILL AT
                          THE ELK ROOFING PLANT
                            STEPHENS, ARKANSAS
                             I.  INTRODUCTION

          In accordance with Section III of the Clean Air Act of 1970,
the Environmental Protection Agency (EPA) is charged with the respon-
sibility of developing standards of performance for emissions from new
stationary sources (New Source Performance Standards) which may contri-
bute significantly to air pollution.  A standard of performance developed
under the Act for emissions of air pollutants must be based on emissions
data from the best emission reduction systems that have been adequately
demonstrated for the particular industry being studied, taking into
account economic considerations.
          Presently, EPA is assembling background information, including
emissions data, for the promulgation of New Source Performance Standards
for the asphalt roofing industry.  To obtain necessary emissions data,
EPA is having various organizations measure emissions from several
asphalt roofing plants.
          As one part of this effort, EPA engaged Battelle-Columbus to
measure stack emissions from the blowing still process at an asphalt
roofing plant of the Elk Roofing Products Company in Stephens, Arkansas.
The blowing stills and emissions control device, (afterburner) at this
plant are about two years old.   This plant was tested during August 20 to
26, 1975, to determine emissions being emitted into the atmosphere from
the asphalt blowing still process and to determine the efficiency of the
afterburner used to control these emissions.  Battelle-Columbus was
responsible for sampling and analyses of the particulate emissions and
gaseous emissions, and acquisition and analysis of process samples.
Simultaneously, Monsanto Research Corporation conducted opacity readings
and Midwest Research Institute staff obtained necessary process data.

-------
          The Elk Roofing Products plant at Stephens uses a blowing still
process to produce asphalt materials used for the production of roofing
products.  Emissions from the blowing stills are controlled by passing
the exhaust gas as through an afterburner unit.  Figure 1 is a flow diagram
of the plant showing relative locations of process components and sampling
points
          Emissions were measured simultaneously at the inlet and outlet
of the afterburner for two blowing processes; (1) saturant blows, and
(2) coating blows.
          During six runs, particulate emission samples were obtained
simultaneously at test points TP-1, TP-2 using an EPA "draft method for
determination of particulate and total gaseous hydrocarbon emissions
from the asphalt roofing industry".  Simultaneously, hydrocarbon emissions
were measured at these points with a flame-ionization detector (FID).  Inlet
and outlet gas composition was determined from integrated gas samples which
were collected in appropriate containers during the course of each run;
anlaysis of CO  and 0  was by Orsat; CO concentration being determined by
nondispersive infrared (NDIR).  Visible emissions testing of outlet stack
emissions was conducted by two certified observers during the particulate
sampling
          Evacuated flask samples were taken at the outlet of the after-
burner for analysis by gas chromatography.   Continuous measurements for
NO  and SO. concentration were made at the afterburner outlet during one
  X       £*
run by an electrochemical method.  Four runs were made to determine aldehyde
concentrations utilizing the Los Angeles wet chemistry method^  .  Polycyclic
organic material (POM) was measured at the inlet and outlet of the afterburner
utilizing a modified EPA Method 5 train in conjunction with a BCL developed
POM collection column.  Process samples, collected during the course of all
runs, included recovered oil, crude or flux asphalt, saturant, and coating
asphalt.  Trace metals of selected particulate emission samples were deter-
mined by optical emission spectroscopy.
          Table 1, the sampling test log, shows the actual times sampling
was conducted.

-------
                                    \
                                    /T
/TP2
                                   Heat
                                 exchanger
 Knock
  out
  box
                                Afterburner
  Knock
   out
   box

        Recovery
         Oil
                    TPI
FIGURE 1.   BLOCK DIAGRAM SHOWING  RELATIVE LOCATIONS
           OF PROCESS COMPONENTS  AND SAMPLE POINTS.

-------
                                       TABLE 1.   SAMPLING TASK LOG

Incinerator Particulate
THC
Vis. Em. CO
Date Time Temperature TP-1 TP-2 Tp-1 TP-2 TP-2 TP-2 Comments
8-19-75 12:
12:
15:
17:
17:
17:
17:
18:
18:
18:

8-20-75 13:
13:
13:
13:
13:
13:
14:
14:
15:
15:
15:
15:
15:
15:
15:
15:
16:
16:
17:
17:
17:
45 1520
47
13 1400
02
10 -I
12
30 -r
09 J




15 1540 -L
29

30 T
33
34 1550
35 -|
36
54
30
33
08 J










- SB
T
LCP
09 J- CP
10 1520
16 J
54 1520
55 i
56
59 i
04
11
28 1540
29

F



1
t p
.
30 -«•





















•SB




1, ]



SB









•*• Thunderstorms, Sampling
Terminated








• SB

.






-L SB
• SB






u .


- F
•*- F
SB - Sampling Break
CP - Changed Ports
F - Finished Run

-------
TABLE 1.  (Continued)

Incinerator Particulate THC Vis.
Em. CO
Date Time Temperature TP-1 TP-2 TP-1 TP-2 TP-2 TP-2 Comments
8-21-75 10
10
11
13
13
13
13
13
13
13
15
15
15
15
15
15
17
17
17
17
:08 T
:20 1520
:48 J-SB
:12 1440
:13 -i
:14
:15
:25
: 31
:32
:13

•
_



• CP
:18 1520
:23 J
:25 -i
:30
:38
:38
:50 1500 J
•




• CP



• F
:52



»














• F
• F




• •



•*• SB


I,

• F
:54 •>• F

-------
TABLE 1.  (Continued)
Incinerator Particulate THC S02-Instr. N0v-Instr. Aldehyde
Date Time Temperature TP-1 TP-2 TP-1 TP-2 TP-1 TP-2 TP-1 TP-2 TP-1 TP-2 Comments
8-22-75 06:22 1400
06:25 -T
06:26
06:36
06:47
08:25 J




-CP
08:30 1520
08:32 T
08:36
08:41
10:09
10:21
10:57
11:02 -1







•-CP



• F
11:03 J



















IF
i-F
• F
- F
11:05 1500
12:00 T
12:30 1480
12:31 -LF
8-24-75 11:10 . T T
12:02 T
13:02
13:05
13:10
13:13
13:14
13:15
14:50
15:05






1

•LSB
j


i \

f \


T SO At TP-1
J-F ,f . ^ 6yEPA-6
15:26 J- CP r r f

-------
TABLE 1.  (Continued)
Incinerator Particulate THC S02~Instr. N0x-Instr. Aldehyde
Date Time Temperature TP-1 TP-2 TP-1 TP-2 TP-1 TP-2 TP-1 TP-2 TP-1 TP-2 Comments
8-24-75 15:30 L CP
15:38
15 : 39 T
15:58 1
16:03
16:16
16:30
16:34
17:19
17:24
17:39
17:50
17:53
17:55
17:57
17:58 J














•F





























•F
• F






• SB







IF IF



I, T

• F
• F
•F •'-F
• F
8-25-75 08:30 1480 1
08:33 -I
08:36
08:47
09:00
09:07
09:26
10:07 1510
10:08
10:09







- CP
10:11


T






-CP
10:15 J
r \
•*• SB

I
A SB


-S6
10:20






•SB
10:26 f

-------
TABLE 1.  (Continued)
Incinerator Particulate THC S02-Instr. N02~Instr. Aldehyde
Date Time Temperature TP-1 TP-2 TP-1 TP-2 TP-1 TP-2 TP-1 TP-2 TP-1 TP-2 Comments
8-25-75 10:
10:
10:
10:
11:
11:
11:
11:
12:
12:
12:
12:
12:
12:
15:
15:
15:
8-26-75 08:
08:
09:
09:
10:
10:
10:
13:
13:
13:
14:
14:
15:
15:
15:
i c; .
30
32 -i
34
40
25
30
35
47
05
08
10 1520
11














-F

- F














- F

- F
17 J
26 i
00
14 1510
25
49 1520 T
50
04
08
13 1500
14







•CP
15 -**CP
36 1520
37
43
31
34
02
11
17 1500
on

L







J- SB







-"-F
•F
• 1?
-•










• SB
r
L F





I T
F
-*• F



c



•- F













-------
                                            TABLE 1.   (Continued)
Date Time
8-27-75 12:07
12:08
12:09
12:11
12:21
12:30
12:37
12:48
14:08
14:17
14:25
14:30
14:37
14:50
14:55
15:14
15:30
16:35
16:37
Incinerator THC PPOM Aldehyde
Temperature TP-1 TP-2 TP-1 TP-2 TP-1 TP-2 Comments
1540
-₯•

•*• SB
1500
-"• SB
^•CP


1520

-r
IT
1530
1510 * F



•p

• CP


-•-F




-*• F
SB-Sampling Break, CP-Change Ports, F-Finish Run

-------
                                     10
          The following sections of this report cover the summary of
results, process description and operation,  location of sampling points,
and sampling and analytical procedures.   Detailed descriptions of procedures,
field and laboratory data,  and calculations  are presented in various
appendices, as noted.

-------
                                    11
                       II.  SUMMARY OF RESULTS

          The composition  (dry) of the inlet and outlet gases  (at the
afterburner were primarily nitrogen (78.2 to 87.6 percent) and oxygen
(10.8 to 20.2 percent), with about 0.2 to 5.6 percent C02.  Both  the
inlet and outlet gas had very high moisture contents - 27.2 to 44.7
percent at the inlet and 11.6 to 20.8 percent at the outlet.  This high
moisture results from the injection of steam over the heated asphalt
to reduce the fire hazard.
          Average gas flow (dry) and gas temperatures for the inlet and
outlet were similar for saturant and coating blows and were as follows:
                             Average Gas Flow,    Average Temperature,
                                         3
                             dscfm     Nm /min       F           C
          Saturant Blow,
            inlet            1867        52        390         198
            outlet           8878       250        390         198
          Coating Blow,
            inlet            1887        53        420         215
            outlet           9039       254        385         196
The increase in gas flow rate through the afterburner, and a related
decrease in moisture, related to the operation of the afterburner.  At
the afterburner,    fuel (natural gas) and air are mixed with the inlet
gas and the fuel is combusted.  The emission control aspect of the after-
burner is that the combustion process incinerates combustible pollutants
as well.
          Sampling for the saturant runs ranged from 73.5 to 112.1 per-
cent of isokinetic.  For the  coating  blow runs,  sampling  ranged from 77.5
to 98.7 percent of isokinetic.
                         Particulate Emissions

          Particulate emissions from each run and average values are given
in Table 2 and 3 for saturant and coating blows, respectively.  The com-
plete printout from all particulate runs is presented in Tables 4 through 7.

-------
                                     12
Saturant Blow
          Average inlet and outlet particulate concentrations for sat-
                             3                                "}
urant-blows were 27,974 mg/Nm  (12.18 gr/dscf) and 365.0 rag/Mm  (0.159
gr/dscf), respectively.  Average saturator blow emissions were:
                  Inlet      80.0 kg/hr     176.4 Ib/hr
                  Outlet      5.6 kg/hr      12.3 Ib/hr
Based on these data, the average efficiency of the afterburner (on
particulate) was 93 percent for saturant blows.
Coating Blow
          Average inlet and outlet particulate concentrations for coating
                       3                                3
blows were 34,363 mg/Nm  (14.60 gr/dscf)  and 221.4 mg/Nm  (0.096  gr/dscf);
respectively.  Average coating blow emissions were:
                  Inlet      98.6 kg/hr     217.4 Ib/hr
                  Outlet      3.3 kg/hr       7.2 Ib/hr
Based on these data the average efficiency of the afterburner (on
particulate) was 96.7 percent for coating blows.
                     Gaseous Hydrocarbon Emissions

          Continuous gaseous hydrocarbon monitoring was conducted using
an FID to analyze samples drawn from the particulate sampling train,
just downstream of the filter.
          Average values of hydrocarbon emissions for each run are
presented in Tables 2 and 3 for saturant and coating blows, respectively.

Saturant Blows
          For saturant blows, the average gaseous hydrocarbon concentra-
tions were 7675 and 29.7 ppm at the afterburner inlet and outlet, re-

-------
                                    13
spectively.  Gaseous hydrocarbon emissions for saturant blows were:

                  Inlet      16.0   kg/hr     35.3  Ib/hr
                  Outlet      0.295 kg/hr      0.6  Ib/hr

Based on these data, the afterburner was 98.2 percent efficient at
reducing ga.seous hydrocarbon emissions during saturant blows.
Coating Blows


          For the coating blows, the average gaseous hydrocarbon concen-

trations were 6506 and 64.7 ppm at the afterburner inlet and outlet,
respectively.  Gaseous hydrocarbon emissions for coating blows were:

                    Inlet      14.0   kg/hr     30.4  Ib/hr
                    Outlet      0.686 kg/hr      1.51 Ib/hr

Based on these data, the afterburner was 95.0 percent efficient at
reducing gaseous hydrocarbons during coating blows.

-------
.TABLE  2.   PERFORMANCE  SUMMARY  OF  EMISSION REDUCTION SYSTEM-SATURANT BLOWS
Run Number 2
Date 8-20-75
Stack Conditions
Sample Location AB Inlet AB Outlet
Sample Number B-3 B-4
Volumetric Flow Rate,
dscfm 1715 8673
Stack Temperature, F 393 399
Moisture, Volume Percent 44.7 18.8
Particulates — Probe, Upstream
Impingers, Prefilter, Filter,
gr/dscf 16.64 0.286
Ib/hr 226.9 22.6
Afterburner Efficiency, percent
This Run 90.0
Average, Three Runs
Gaseous Hydrocarbons
ppa as CH4 9152 78.8
gr/dscf 2.699 0.023
Ib/hr 39.43 1.72
Afterburner Efficiency, percent
This Run 95.6
Average, Three Runs
6 7 Average
8-25-76 8-26-75
AB Inlet AB Outlet AB Inlet AB Outlet AB Inlet AB Outlet
B-ll B-12 B-13 B-14 . 	 	
1884 8476 2001 9485 1867 8878
405 393 371 378 ' 390 390
41.0 20.8 27.2 11.6 37.6 17.1
11.34 0.087 8.561 0.104 12.18 0.159
174.7 6.3 127.5 8.1 176.4 12.3
96.4 93.6 93.0
93.3
8146 7.1 5726 3.3 7675 29.7
2.403 0.002 1.689 0.001 2.264 0.009
38.06 0.15 28.49 0.08 35.33 0.65
	 - " T
99.6 99.7 98.2
98.3

-------
TABLE 3.  PERFORMANCE SUMMARY OF EMISSION REDUCTION SYSTEM -  COATING  BLOWS
Run Number
Date
Stack Conditions
Sample Location
Sample Number
Volumetric Flow Rate, dscfin
Stack Temperature, F
Moisture, Volume Percent
Particulates — Probe, Upstream
Impingers, Prefilter, Filter,
gr/dscf
Ib/hr
Afterburner Efficiency, percent
This Run
Average, Three Runs
Gaseous Hydrocarbons
ppm as CH,
gr/dscf
Ib/hr
Afterburner Efficiency, percent
This Run
Avo^-ar»o TV»i-«o Utirtc
3
8-21-75
AB Inlet AB Outlet
B-5 B-6
1838 9045
419 391
33.6 15.6

15.06 0.123
210.8 9.0
95.7
6420 69.2
1.894 0.020
28.83 1.57
94.6
4
8-22-75
AB Inlet AB Outlet
B-7 B-8
1943 9549
430 382
33.9 15.0

15.42 0.066
229.3 5.4
97.6
96.7
7066 103.3
2.084 0.030
33.73 2.49
92.6
95.2
5
8-24-75
AB Inlet AB Outlet
B-9 B-10
1881 8524
411 382
34.9 16.5

14.41" 0.099
212.1 7.1
96.7
6100 21.7
1.799 0.006
28.75 0.46
98.4
Average
AB Inlet AB Outlet
	 	
1887 9039
420 385
34.1 15.7

14.60 0.096
217.4 . 7.2
96.7
6506 64.7
1.919 0.019
30.44 . 1-51
95.0

-------
I
TABLE 4. INLET RESULTS
INLET RESULTS, EL< ROOFIMG, STEPHENS, ARK.
; RUN 'JO.
3 TEST DATE
0
z
§ VOLUME OF SflS SAMPLFO, OSCF
PERCENT MOISTURE R* VOLUME
AVERAGE STOCK TEMPERATURE, F
STACK VOLJttTRIC FLOX RATE, OSCFM
STACK VOLUMETRIC FLOW RATE, ACFM
PERCENT IS3KINETIC
PERCENT EXCESS AIR
PERCENT OPACITY
FEED RATE, TONS/HR

Z
8/20
64.5
44.7
393
1715
500 <»
B5.3
88

ND
, ELK ROOFING, STEPHENS, ARK.

3 U
8/21 8/22
9C.o 9S.7
33.6 33.9
U19 !»30
1833 19t»3
U603 <»960
77.5 78. k
221 267
NA NA
ND ND

5
8/24
105.5
34.9
411
1881
4765
82.4
516
NA
ND
(ENGLISH UNITS)

6
8/25
79.8
41.3
405
1884
5239
90.5
111
NA
ND

7
8/26
66.7
27.2
371
2001
4322
73.5
113
NA -
ND
AVEKAflE
SATURATOR COATING
2,6,7 3,4,5
70.4 98.3
37.6 34.1
390 420 " "
1867 1887
4855 4776
83.1 79.4
104 335
NA NA 	
ND ND
*
PARTICULATES - PROBE, CYC, FILTER CATCH
MS
GR/OSCF -
GR/ACF -
L8/HR - (D
L8/TON FEED
59783.7
16.636
5.697
2?6.9
ND
83595.8 98814.5
15.058 15.1*16
6.039 • 6.033
210.3 229.3
ND ND
98706.2
14.405
5.680
212.1
ND
58771.5
11.333
4.073
174.7
ND
37089.1
a. 561
3.961
127.5
ND
55214.8 95372.2
12.178 14.60
4.577 5.907
176.4 217.4 .
ND ND

PARTICULftTF.S - TOTAL CATCH • • '
1
1C
SR/HSCF -
GR/ACF -
L3/HR - (1)
L3/TOI^ FEED
,
PERCENT IMPINGER CATCH
(1) Results are average of concentration method
•MM ^^Mtffei-{^MHB.n^

69783.7
16.636
5.697
326.9
ND

.0


88595.8 98811*. 5
15.053 15.1*16
6.008 6.033
21C.3 • 229.3
ND • ND

.0 .0
ND
- 	 . ,. N/l

98706.2
14.405
5.680
21-2.1
' . ND

• 0
•
58771.5
11.338
4.073
174.7
ND

.0

37089.1
8.561
3.961
127.5
ND
.
.0

55214.8 95372.2 	 	
- 12.178 14.60
4.577 5. SO;
176.4 217.4
ND * ND

0 0
- No data.
I^B-NottMh ic 	

-------
TABLE 5. OUTLET RESULTS
OUT! FT RESULTS fLK SCO FT'NT, , ^~FPUfM<,^ APK.
- CMf' VC 2
o TEST DATE • fa/20
oVOLUNE CF GAS SAMPLED, DSCF 105.4
PERCENT MOISTURE 3Y VOLUME 18.8
AVEf-AGE STACK TEMPERATURE, F 399
STACK VOLUMETRIC FLOW RATE", OSCFM 8673
STACK VOLUMETRIC FLOW RATE, ACFK 17357
PERCENT ISOKINETIC 112.1
PERCENT EXCESS AIR 354
PERCENT OPACITY ' ND
FEED RATE, TONS/HR ^

PARTICULATES - PRO 6E , CYC, FILTER CATCH



IC/LJ^J — I T ) • •} 1 C
LB/7 Of ' F^^O ^*

P£ I?T Tflll ATP"*; -— .- T flT A 1 f* fi T'"* H

MG 1954.9
G.R/CSCF - 0.266
GR/ACF - 0.143
L3/HR - (D 22.6
L3/TCN FEED ND

PERCENT I.-1PI.NGER CATCH .0
(1) Results are average of concentration method
end_..area ratio ~ethod.. 	
ELK ROOFING, STEPHENS


8/21 8/22
122.0 !«.«,. 7
15. & 15.0
391 382
9C45 9549
17213 17895
67.6 98.7
299 2563
ND ND
ND ND



•30 c. t 7- • •- OCH •:>
n r P.^ ..... .. n r ^^
or. ^ d
ND ND



982.7 624.5
0.123 C.C66
0 . C 65 0 . 0 35
9.0 5.4
ND ND

.0 .0
ND '
, ARK. (ENGLISH)

AVERAGE
c c. -T SATITRATOR COATTN1^
8/24 8/25 8/26 2,6,7 3,4,5
132.1 90.3 89.9 95.2 133.1
16.5 20.8 11.6 17-1 15.7
362 393 378 .390 385
8524 8476 9485 8878 9°39
1625. 1729C 17CCC 17216 17119
95.3 93.3 91.3 100.6 93.9
21U 165 193 237 1024
ND ND ND ND ND
ND ND ND ND ND


o n -L7 r G<;a o.oai 0.051
71 fi T SI 12"? 7^?
ND NT) NT> ND NT) 	



351.4 511.2 6C6.1 1024.1 81S.5
G.C'99 C.C87 0.104 0.159 0.096
0.052 0.343 0.058 ' 0.081 0.051
7.1 6.. 3 8.1 12.3 7.2
ND ND ND ND ND

.C .0 .0 .° 0;
= No data

-------
TABLE 6. INLET RESULTS, ELK ROOFING, STEPHENS, ARK. (METRIC UNITS) • ' -
INLET RESULTS, ELK -POOLING, STEPHENS, ARK." »,,r.04rr
1 RUN NO. ?
S TEST DATE ~ 8/20
o
z
§ VOLUME OF G4S SAMPLFO, NOM 1.83
b.
PERCENT MOISTURE 3Y VOLUME 1*1*. 7
AVERAGE STACK TEM^RaTURE , C 300
STACK VOL-METRIC FLOW RATE, NCMM t*8
STACK VOL'JIETRi: FLOrf RATE, CrtM 11*1
PERCENT ISDKINF.TIC 85.3
PERCENT EXCESS AIR 88
PERCENT OPACITY NA
FEED RATE, MTON^R nu
3
8/21
2.56
33.6
215
51
129
77.5
221
NA
ND
,,
8/22
2.78
33.9
221
51*
139
78.1*
367
NA
ND
567
8/3<» 8/25 3/36
3.98 2.35 1.88
34.9 41.0 27.2
210 207 188
53 53 56
134 147 121
82.4 90.5 73.5
516 111 113
NA NA NA
.ND ND ND
SATURATOR
" 2,6,7 '-
.2.06
37.6
198
52
136
83.1
104
NA
ND
COATING
3,4,5 - — '
2.93 " 	
34.1 	
215
53
134
79.4
335
NA
' ND "

PARTICULATES - PROBE, CY C, FILTEP CATCH
HG 69783.7
M3/NCM - 3<*212.6
MG/C-I - 13085.0
KG/HR - .(1) . 103.9
KG/MTON FEED - __NP
88595.8
3t»589.0
13800. 8
95.5
ND
9SR14.5
35411.3
13857.5
104.0
ND
98706.2 58771.5 37089.1
33088.1 26043.0 19665.4
13047.3 9356.4 9097.6
96.2 79.2 57.9
ND ND ND
55214.8
27973.7
10515.0
80.0
ND
95372.2
34362.8
13568.5
98.6
ND

PARTICULATES - TOTAL CATCH
••
KG , , 69783.7.
MS/NCM - 38312.6
MG/CM - 13085.0
KG/HR - UJ 102.9
KG/MTON FEED ' ND

PERCENT IMPINGER CAT^H .0
(1) Results art average of concentration method
^^^Ki-rc^^^B-i /^H^H,/4_ ._ _ _.

88595. 8
31.589.0
13800.8
95.6
•ND

.0


98814.5
T5411.3
13857.5
104.0
-ND

.0
ND
NA

98705.2 58771.5 37089.1
33088.1 26043.0 19665.4
13047.3 9356.4 9097.6
96.2 • 79.2 57.9
ND • ND ND

.0 .3 .0
= No data.
^^•\o t ^^•ica^^H 	
" 55214.8
27973.7
10515.0
SD.O
ND
0


95372.2
34362.8
13568.5
98.6
ND
0

oo

-------
                                    TABLE 7.   OUTLET RESULTS ELK ROOFING.  STEPHENS.  ARK.  (METRIC UNITS)
nuTifT P~"IMT<; t L" c-nnrj.Mr,., ^TrDMtf!Si Af-"
- rlltl HO,
o TEST DATE
gVOLUKE OF GAS SAMPLED, NCK
PERCENT MOISTURE BY VOLUME
AVERAGE STACK TEMPERATURE, c
STACK VOLUMETRIC FLOW RATE, NCKK
STACK VOLUMETRIC FLCW RATE, CUM
PERCENT ISGKINCTIC
PERCENT EXCESS AIR
PERCtNT OPACITY
FEED RATE, MTON/HR
"> 7 d r
6/20 8/21 8/22 8/2»f
2. 97 3.46 4. C8 3.73
18.3 15.6 15. C 1&. 5
203 199 194 194 .
244 25E 269 240
469 463 5G4 u58
112.1 87.6 58.7 95.3
354 299 2563 210
ND ND ND ND
ND ND ND ND

8/25
2.55
2;. 8 -
2CO
239
467
96.3
165
ND
ND
7
8/26
2.54
11.6
192
267
479
SI. 3
193
ND
ND
AVEF
SATI!RATOH_
2,6,7
2.81
17.1
198
250
485
100.6
2.37
ND
• ND
IAGS
	 COAT DIG 	
3,4,5
3.95
15.7
195
255
482
93.9
1024
ND
ND"

PARTICULATfcS - PRO BE , CYC , FILTE R CATCH
Mr
Mr. y f r p -

KG/HP - ' '
KC/MTrN FEED

PAPTICULATES - TOT/L CATCH

HG
MG/NCM -
MG/CM -
KG/HR - .-. o •>
984. C
2* 9
ND


511.2
2CJ.2
98.0
2.9
ND
. AT A 1

1334 0
3 7
ND


6C6.1
235.5
133.0
3.7
ND
1024 1
36S 0
186.2
5.6
ND


1024.1
365.0

5.6
ND
819 5
221 A ., , ,
11^ 6 . ,
3.3
ND 	


819.5
221.4
IZu. u
3.3
ND

PERCENT IlPINGER CATCH
.0 .0 .C .C
.0
.0
0
0
(1)   Results are average of concentration method
     and area ratio -.ethod
                                                                              ND  =  No data.

-------
                                     20
                          S02 and NOX Emissions

          SC>2 and NOX emission measurements were made during saturant
and coating blows at both the afterburner inlet and outlet.  Table 8
summarizes the S00 and NO  emission concentrations measured at these
                 2       x
points.
          SC>2 concentrations at the inlet and outlet ranged up to 920
and 350 ppm, respectively.   As the fueled natural gas afterburner would not
be expected to affect S02 present in the inlet gas stream, the reduction in
S0? concentration is attributed to dilution of the inlet gas with after-
burner combustion air.
          Due to the lack of a 862 scrubber upstream of the NOX analyzer
(as described in Section V)  the NOX emissions data are not considered
valid.  However, measured NOX concentration at the inlet. and outlet
ranged up to 1900 and 500 ppm, respectively.   As with S0_, the NO  reduc-
                                                        *~        X
tion across the afterburner is thought to be primarily due to dilution.
          Using the mean S0» concentrations and the outlet stack gas
flow for Run B-10,  the emission rates for SO- were calculated to be:
              Saturant blows - outlet     5.50 kg/hr     12.1 Ib/hr
              Coating blows - outlet      6.49 kf/hr     14.3 Ib/hr.
          Method 6 measurements for SO  were made at the afterburner
inlet for two saturant blow runs.  The S02 concentrations measured
during the runs were:
                         Run B-l     123 ppm
                         Run B-3     357 ppm.
The Method 6 S09 number is much lower than the mean values obtained
with the continuous analyzer.  The latter value generally is in agreement
with the continuous analyzer data. If it is assumed that the range of S0»
values omitted during the complete saturator blow is 0 - 730 ppm and that
the ratio of the average value to the maximum value is the same at the
inlet and outlet, then a mean S00 emission concentration of 294 ppm is pre-
                                ^
dieted for B-ll (inlet).

-------
                          21
    TABLE 8.  SO- AND NO  READINGS BY CONTINUOUS
                                 / \
              MONITORING ANALYSISW
Inlet
<400 -
N.A.
S
Outlet
B-10(e)
0 - 350
141
Saturant Blow:
         Run
         Range, ppm
         Mean, ppm

Coating Blow:                   ,_.
         Run                B-ir ;           B-10
         Range, ppm      <400 - 920^      46 - 330
         Mean, ppm          N.A.(d)            166


                       NO (g)
Saturant Blow:
         Run              .    B-9             B-12
         Range, ppm         0 - 1600        245 - 500
         Mean, ppm            902              391

Coating Blow:
         Run                  B-9            B
         Range, ppm      60 - 1900          50 - 435
         Mean, ppm            814              260

(a)  S02 data are from EnviroMetrics analyzer; NOX
     data are from DynaScience analyzer.
(b)  Data taken during a portion of a coating blow
     representing last 10. minutes of saturant blow.

(c)  Calibration gas cylinders empty at end of run
     and thus, analyzer calibration could not be
     verified.
(d)  Mean values not available as complete blow was
     not sampled.

(e)  Data taken during saturant blow preceeding
     coating blow for which B-10 particulate samples
     were collected.

(f)  This coating blow did not appear normal as flow
     was stopped during the process.

(g)  No S02 scrubber was used ahead of the analyzer
     used to make the NOX measurements.  Thus, they
     may contain a contribution due to the S02> as
     well as NOX.

-------
                                      22
                                CO Emissions


           CO emissions  were measured using an NDIR analyzer at the after-

 burner outlet during one saturant blow arid one coating blow.   The emis-
 sions  of CO were:

                                     Saturant     Coating
                                       Blow       Blow

           CO concentration, ppm
             Minimium                   250        670
             Maximum                   1150       3950
             Weighted average           637       1539

           CO emissions,  kg/hr         13.0       32.3
                         Ib/hr         28.7       71.3


               Gaseous Emission -  Analysis  of  Flask Samples


           The flask sample analysis  of outlet gas (Table  9) agrees,  in

 general, with Orsat data and, thus,  this data is presumed to  be valid.
 Analysis of flask  sample shows emission of less than 5 ppm NO , 100 ppm
                                                              X
 S09, 3000 ppm CO,  and about 200 ppm  of gaseous hydrocarbons,  at the
 particular time at'.-which the sample  was drawn.  The FID response to this

 hydrocarbon concentration would have been  about 330 ppm,  CH,  equivalent.


                              POM  Emissions


           Table 10 presents POM emissions  results from one run made
 during a coating blow.   The POM concentration at the inlet was very high,
              3
 112,308 |j,g/Nm . With the large amount of  dilution air. the outlet, concen-
                             o
.tration was only 74.97  yg/nM ,  giving an afterburner efficiency of 99.93
 percent on destruction  of POM.  Only 7.7 percent of the total POM at the
 outlet was recognized as being carcinogenic compounds.
           Later runs to "calibrate"  the effects of using  a high column

 temperature (175°F) for POM sampling (as w.as  used for sampling at the Elk

-------
                                 23
TABLE 9.  ANALYSIS OF FLASK SAMPLE FROM ELK ROOFING PRODUCTS PLANT
          (EPA SAMPLE S76-006-034)

          (For Sample Collected from the Particulate Sampling
           Train Downstream of the Afterburner  1436-1441,
           August 21, 1975)
Component


°2
co2
N2
A
CO
H2
S°2
NO
X
COS
CH OH
C2H5OH
HCHO

CH COCH
CH4
C2H6
C3H8
C4H10
C5H12
C2H4
C3H6
C4H8
C5H10
C2H2
C6H6
Analysis
Volume
percent
12.7
3.71
82.1
1.03
0.30
0.08
0.01
O.0005
0.0002
O.005
<0.005
<0.005
ppm by
Volume
<0.5
80
0.5
1.0
<0.5
<0.5
80
3.0
1.0
<0.5
27
3

-------
                                TABLE 10.  SUMMARY OF  POM DATA  FOR ELK  ROOFING  PLANT
Raw Data (POM by GC-MS,
No Correction for Blanks, )ug.

NAS
Component Notation
Anthracene/Phenanthrene
Methyl anthracenes
Fluoranthene
Pyrene
Methyl Pyrene/FIucranthene
Benzo(c)phenanthrene ***
Chrysene/Benz(a)anthracene *
Methyl chrysenes *
Benzo fluoranthenes **
Benz(a)pyrene ***
Benz(e)pyrene
Perylene '•.
3-Msthylcholanthrene';. ****
Indeno(l,2,3,-cd)pyrene - *
Benzo (ghi) perylene
Dibenzo (a, h) anthracene ***
Dibenzo(c,g)carbazole ***
Dibenz(ai and ah)pyrenes ***
Coronene
TOTAL
Sample Volume, Nin
_dscf
Inlet
(EPA Sample
S75-006-112 & 113
BCL Sample 2-1)
41,152
112,128
1,920
2,316
44,096
ND
12,032
39,360
1,152
512
896
ND








2.28
76.4 •
Outlet
(EPA Sample
S 75-006-106
BCL Sample 15)
79.0
• 69.3
6.0
7.8
51.8
'1.0
4.0
3.5
4.7
, 
'ND










POM in Sample
(Corrected for
blank) , ug
Inlet
41,152
112,128
1,920
2,816
44,096
—
12,032
39,3'60
1,152
512
896
—










Outlet
78.5
69.3
5.9
7.7
51.8
1.0
4.0
3.5
4.7
("O
—










POM, Loading in
Gas St.ream,
ug/Nni3
Inlet Outlet
18,049
49,179
842:'
1,235
19,340
•
'5,227
17,263
505
225
393
—







112,308


25.5
22.5
1.92
2.50
16.8
0.32
1.30
1.14
1.53
£'">
—







74.97


Separation of Benz(a)pyrene and Benz(e)pyrene was conducted only on one inlet sample due to cost limitations.

-------
                                   25
Roofing suggested that the POM collection efficiency of the adsorbent
column is greatly reduced at elevated temperature.  Thus, the results;
presented in Table 10 must be taken as minimum POM emission values.
Actual POM emissions may have been higher by one or two orders of
magnitude.

                           Aldehyde Emissions

          Table 11 presents results of aldehyde emission measurements
made during coating blows.  Aldehyde emissions'were 3.50 and 5.17 kg/hr
at the afterburner inlet and outlet, respectively.  However, neglecting
one run in which no aldehyde reduction was observed through the after-
burner, the average aldehyde emissions at the outlet were 0.131 kg/hr,
giving an afterburner efficiency of 96.3 percent on aldehydes.  It must
be noted, however, that inlet and outlet sampling was conducted at random
periods and not simultaneously.

          Table 12 reports on the proximate analysis (VM, FC ash, mois-
ture) the ultimate analysis (C, H, S, N, and 0), the specific gravity,
and the heating value for five process samples collected during emission
sampling at the Elk Roofing Co. plant.
                             Visible Emissions

          Table  13 reports on visible emissions readings made at the Elk
Roofing Co. plant during various periods of stack  sampling.  All re-
ported opacity readings were zero opacity.

-------
                                            26
                    TABLE 11.  ALDEHYDE RESULTS FOR ELK ROOFING PLANT
Run Number
AL-l-IN
AL-2-IN
AL-3-IN
AL-4-OUT
AL-5-IN
AL-6-OUT
AL-7-IN
AL-8-OUT
AL-9-OUT



Date
8-22-75
8-22-75
8-24-75
8-24-75
8-24-75
8-24-75
8-25-75
8-25-75
8-27-75



Gas Sample Aldehyde Stack
Volume, Concentration, Flow Rate,
Clock Time Nm3(a) mgHCHO/Nm3 (b) Nm3/min
1009-1021 0.0269
1200-1231 0.0812
1558-1616 0.0538
1603-1616 0.0790
1719-1739 0.0591
1724-1757 0.0948
1130-1147 0.0487
1135-1205 0.0801
1237-1437 0.3085
Average Inlet
Average Outlet
(includes Run AL-8)
Average Outlet
(without Run AL-8)
Efficiency
(without Run AL-8)
713.8 54
1428.6 54
542.8 53
7.6 240
296.1 53
4.4 240
2218.8 56
1260.9 267
14.6 258
1040.6
321.9
8.9

Emission Rate
kg/hr
2.31
4.63
1.73
0.109
0.912
0.063
7.46
20.20
0.226
3.41
5.15
0.133
96.3
(a)   Corrected  to  standard  conditions, 20 C and 29.92 Hg.




(b)   Results  corrected  for  blank.

-------
                   TABLE 12.  ULTIMATE ANALYSES OF PROCESS  SAMPLES FROM ELK ROOFING PLANT
                                                                                         (a)
Sample
Flux
(Unblown
Asphalt)
EPA Sample No ... S75-006-136
Volatile matter (VM) , percent
Fixed carbon (FC) , percent
Ash, percent
Moisture, percent
Carbon, percent
Hydrogen, percent
Sulfur, percent
Nitrogen, percent
Oxygen, percent
Specific gravity
Heating value, Btu/lb
85.25
14.72
0.01
0.02
78.85
5.47
3.52
0.14
11.99
0.9598
18,652
Saturant
Asphalt
S75-006-137
83.87
16.07
0.04
0.02
78.96
5.48
3.28
1.54
10.68
1.0046
18,504
Coating Recovery Oil, Recovery Oil,
Asphalt From Afterburner From Afterburner Drain
S75-006-139 S75-006-004 S75-006-133
82.02
17.92
0.06
0.00
79.15
5.48
3.34
0.72
11.25
0.8541
18,346
93.02
5.82
0.34
0.82
77.09
5.36
2.61
0.54
13.24
0.8786
18,517
89.14
5.91
0.63
4.32
74.16
5.15
2.71
0.57
12.46
0.9810
17,608
                                                                                                                N3
(a)
    All data reported on "as received" basis.

-------
                             TABLE  13.      SUMMARY  0? VISIBLE EMISSIOii  D'ATA
                                           ELK  ROOFIM'J CO.,  STEPHEN:!,  ARKANSAS
                                                                          Mind
       21) HP.  Clock                Dlsi. .  lu      Direction        V/ti>'l      Velocity
          "Observer    .°..-'>irce (ft)    fi-on So:;:-'?    Mrectlon      Mf'H
                                                                                   :.'e?.th'-r    Background     Time-Opacity
5-19 12:17-17:02


5-14 12:1:7-17:02

<5-lQ 17:12-1«:29

°-l
-------
                                     29
                  III.  PROCESS DESCRIPTION AND OPERATION

          This roofing plant has two asphalt blowing (oxidizing) stills.
For controlling emissions from the stills, the plant uses an afterburner
equipped with a waste heat boiler.  The stills and the afterburner are
about one year old.  Details of the blowing process, design data on the
afterburner and operational data monitored during the emissions tests are
presented next.

                           Process, Description

          Asphalt is oxidized or air-blown to raise its softening point,
reduce penetration, and achieve other desired rheological and physical
properties.  At this plant, asphalt is blown in batches in vertical stills.
The flux asphalt is usually pumped into the still at 400°F to 460°F and
the oxidization process is started by forcing air into the asphalt through
a perforated pipe arrangement or sparger located near the bottom of the
still.  Initially, the purge valve is open (see Figure 2) and some of the
air bypasses the sparge line and enters the still above the asphalt.  This
purge valve closes gradually, over a period of 4 to 6 minutes.  When closed,
all the air passes through the sparger.  Emission tests were initiated
when the compressor was started, although plant operators do not consider
the blow to have started until the purge valve is completely closed.

          The asphalt temperature increases as blowing progresses, due to the
exothermic nature of the oxidization process.  (No external heat is added.)
This increase in temperature, in turn, increases the rate of oxidization
of the asphalt.  Water is sprayed into the surface of the hot asphalt as
necessary to keep its temperature safely below the flash point.  The blowing
duration varies with the type of asphalt used, the blowing temperature and
the required asphalt characteristics (penetration, softening point and
consistency).  At this plant, the same flux asphalt is used for saturating and

-------
                                       30
                                    FIGURE 2

               STILL LAYOUT SHOWING PERTINENT FEATURES AND PROCESS
                     MONITORING INSTRUMENTATION ELK ROOFING
                  — Oil Recovery Cyclone
                                                                   Bypass valve
Water
                        Still
Sparger
Air
Blower

-------
                                     31
coating asphalt.  Only the blowing duration is changed, from 1-1/2 hours

for saturant to 4-1/2 hours for coating.  Asphalt samples are checked

periodically during the blow and when the desired characteristics are
achieved, oxidization is terminated by shutting off the air supply.

                         Emission Control System


          Figure 3 is a schematic of the layout of the blowing stills,  the
afterburner and the waste heat boiler at the Elk Roofing Plant.   Even
though emissions from both stills are ducted to the afterburner,  blowing

was only done in one still at a time during sampling.   The capacity of  the
blowing stills, the physical dimensions and design information on the
afterburner are summarized below.

          Type of stills:  vertical
          No. of stills:  two
          Working capacity:  9,600 gal. each
          Physical  dimensions:  Height: 30 ft (asphalt level  does not exceed
                                               20 ft)
          Pressure  drop of knock out boxes:  4" H20
          Type of air blower:  rotary lobe blower (100 hp)
          Identification:  Roots ID No. 847-205-110, Model 812 RAG-J
          Manufacturer:  Dresser Industries, Inc.
                         Roots Blower and Vacuum Pump  Division
                         Connersville, Indiana

          Control  Device:  HIRT incinerator equipped with waste heat boiler

          Manufacturer:  HIRT Combustion Engineers, Montebello, California
          Model No.:   H1H13MX
          Design operating temp:  1700°F (actual^ 1600°F)
          Maximum temp.: "l800°F
          Retention time:  0.5 sec at 1400°F
          Exhaust fan:   24,000 acfm at 500°F (75 hp)
          Natural  gas burner:  16,500 scfh at 5 psig

          waste Heat Boiler

          Model No.:   WHB 400
          Manufacturer:  Scotch Marine Boiler, Abilene, Texas
          Rating:   29,360 Btu/hr
          Produces:  1,500 Ib steam/hr
          Design water  pressure:  150 psig
          No.  of tubes:  371

-------
Air
Process



i ,









1

I
I

Knock
Cham

n H
1
1
1



1 Asphalt
Blowing
Still
1
1
L_i::l.





1
Out
Ders




j

S







Fumes













j



Asphalt
Blowing
Still





•


:$?-

I
1







r




Stc

1
Combustion Air



vy ^
Y

A
/\
A,r / \
1
j. r
T 1



I
Natural
Gas




Waste /CX
_^. Heot ^^

Boiler Fan
t
Process
Steam

T.^^ Primary Secondary
Damper Damper
for for
LMiunon LMiunon
Air Air
                                                                                                      LO
                                                                                                      NJ
        Figure 3 - Schematic of Layout of Asphalt Blowing Stills, Incinerator, and
                     Waste Heat Boiler at Elk Roofing Plant

-------
                                     33
          Design Cons tra in ts
          Flow:  1,600 to 3,200 scfm
          Quench steam:  0 to 2,800 scfm
          Fume temperature:  400°F                   g
          Hydrocarbon present in fumes:  0 to 12 X 10  Btu/hr
          Oxygen:  0 to 21 percent
          Primary dilution air:  1,500 to 6,000 scfm
          Secondary dilution air:  0 to 1,500 scfm
          Combustion air:  riate unknown
                            Process Operation

          Process parameters monitored durinq sampling were:
          1.  Blowing duration (from start-up to shut-down of air blower)
          2.  Air flow through still (cfm)
          3.  Temperature of asphalt entering/leaving stills
          4.  Temperature of asphalt in still as a function of time during
blowing
          5.  Amount of asphalt in still at start/end of blow cycle
          6.  Cooling water used during blowing as a function of time
          7.  Specification on asphalt - before and after blowing
          8.  Quantity of oil recovered from incinerator
          9.  Afterburner operating temperature
         10.  Waste heat boiler outlet temperature
         11.  Pressure and quantity of steam generated by waste heat boiler
as a function of time
         12.  Temperature of fumes emitted from blowing still prior to
dilution with air
         13.  Natural gas consumption by afterburner plus storage and preheater
tanks as a function of time
         14.  Asphalt transfers into (and out of) still not being blown

          The blowing duration was generally 1-1/2 hours for blowing saturant
and 4-1/2 hours for blowing coating.  The actual times during sampling
varied somewhat and are recorded in Table 14

-------
                                           TABLE 14.  SUMMARY OF PROCESS DATA
Amount of
asphalt Average Temp during
charged airflow blowing Softening point
Typp nf hlnwinn into still Duration of through still (°F) of product
August
12:45
August
5:13 p
August
8:20 a
August
1:34 p
August
3:54 p
August
10:20
August
1:12 p
August
6:22 a
August
11:10
August
19,
a.m.
19,
.ni. -
20,
,m. -
20,
.m. -
20,
.m. -
21,
a.m.
21,
.m. -
22.
.in. -
22,
a.m.
24,
1976
-5:03 p.m.
1976
6:15 p.m.
1976
12:53 a.m.
1976
3:10 p.m.
1976
5:28 p.m.
1976
-11:47 a.m.
1976
!.':50 p.m.
1976
11:05 a.m.
1976
-12:30 p.m.
1976
Cycle " (gal.) blowing cycle (cfm) min max
Coating 9,600 4 hr 18 min 1,530 450 510
Saturant 9,600 1 hr 2 min ' 1,550 460 510
(sampling abandoned due to storm)
Coating 9.600 4 hr 33 min 1,530 440 510
Saturant 9,600 1 hr 36 min 1,550 435 500
Saturant 9,600 1 hr 34 min 1,530 450 520
Saturant 9,600 1 hit 27 min 1,550 425 515
Coating 9,600 4 hr 38 rain 1,530 440 510
Coating 9,600 4 hr 43 min 1,530 420 512
Saturant 9,600 1 hr 20 min 1,550 430 490
Coating 9,600 5 hr 1 min 1,530 421 508
(°F)
245
-
240
141
146
143
251
248
-
_
12:54 a.m.-5:55  p.m.

-------
                                                  TABLE 14.  (Continued)


Amount of
asphalt
charged
Type of blowing into still

August 25, 1976
8:30 a.m. -10:07 a.m.
August 25, 1976
10:29 a.m. -3:14 p.m.
August 26, 1976
8:49 a.m. -10:13 a.m.
August 26, 1976
1:36 p.m. -3:17 p.m.
August 27, 1976
cycle
Saturant

Coating

Saturant

Saturant

Coating
(Ral.)
9,600

9,600

9,600

9,600

9,600
Averaoe Temp during
airflow blowing Softening point
Duration of
blowing cycle
1 hr 37 min

4 hr 45 min

1 hr 24 min

1 hr 41 min

4 hr 28 min
through still (°F)
(cfm) min max
1,550 455 508

1,530 468 518

1,530 430 515

1,550 410 490

1,530 465 518
of product
(°F)
141

259

133

127

243
12:07 a.m.-4:35 p.m.

-------
                                     36
          Tables 1 through 15 of Appendix J show the values recorded
during each run for the process parameters delineated above and Table 14
summarizes some of these data.

                        Process Instrumental on

          Process parameters were monitored at 15-minute intervals during
saturant blows and every 30 minutes during coating blows.  The instrumentation
used was located as illustrated in Figures 2 and 4 and, where appropriate,
are also briefly described below:
          1.  Airflow through the still was calculated from the pressure
change across the compressor, using information supplied by the compressor
manufacturer.
          2.  The operating temperature of the afterburner was measured
with a temperature detector and displayed on the afterburner control panel.
          3.  The amount of asphalt in the still is estimated using a
crude float and weight technique.  The float rides on the surface of the
asphalt and the marker (weight), which is connected to the float by cable,
hangs down the side of the tank.  As the asphalt level rises, the marker
drops.  The tank is assumed to be full when the marker reaches a mark painted
on the side of the tank.
          4.  Cooling water requirements were measured directly from a float
gage made by Schutte and Koerting.  This device consists of a tapered glass
cylinder and a float (see Figure 5).  Increases in the water flowrates
raise the float in the vertically mounted glass cylinder, and vice-versa.
Flowrate is measured by comparing float height to flow rate calibration
marks on the glass cylinder.
          5.  Asphalt samples were obtained from a tap on the side of the
still near the bottom before the starting of blowing and after it was
completed.  Tests were run to determine softening point and penetration.
          6.  The temperature of the fumes at the incinerator inlet,
before dilution, was displayed on a gage located in the insulated fume dust
located just upstream of a velocity section (and the dilution air inlets).

-------
'Exhaust
 Gas
                                     FIGURE 4

                         EMISSION CONTROL SYSTEM LAYOUT SHOWING
                         MONITORING  INSTRUMENTATION ELK ROOFING
             /"
                                                                 Control
                                                                 Panel-
               Waste  Heat
            Recovery  Boiler
'  Pi 1uted

i  Fumes
                                                 Incinerator

                                                 Combustion air
                                                                                       Fumes
                                                                             ,  W  Velocity
                                                                             1   Ip Section
                                                                              /\
-*- Air

^-Primary dilution
 air damper
                         \Secondary  dilution
                          air  damper

-------
                38
          FIGURE 5

WATER FLOWMETER  ELK  ROOFING
                      -Water Flow
                         Tapered
                         Glass Cylinder
                      H-Float
                         Calibrations
                         Water Flow

-------
                                     39
          7.  The quantity of oil recovered from the afterburner during a
complete saturant blow cycle was about 1/3 gal.  This was deemed negligible
compared to a charge of 9,600 gal. of asphalt in the still, so this
parameter was not monitored in subsequent runs.  The quantity of steam
generated by the waste heat boiler was read from a Foxboro recorder.
The multiplier for the readings was obtained from the Foxboro Company.
          8.  Natural gas consumption was charted on a Foxboro circular
recorder.  There was no way to accurately determine incinerator usage,
since the flux tank and (after August 23, 1975) the preheater tanks were
on the same gas line.  It was assumed, however, that changes in gas consumption
could be attributed to the heating value of the blowing fumes.  Using this
approach, the percent reduction is in the range of 10 to 22 percent.
During one run, the gas supply to the flux tank heater was turned off for
a few minutes and gas requirements dropped about 2,000 scf/hr.  The plan
to repeat this test after August 23, 1975, when the preheaters were in
use and the flux tank heater was operating at maximum capacity, was aborted
because natural gas consumption rates were fluctuating widely for no
discernible reason on the date selected.
          9.  It was not possible to determine the accuracy of process
parameter values recorded from information provided by plant personnel.
Instrument calibrations and maintenance are usually performed by plant
personnel and we could not obtain records of these dates.

                   Process Conditions During Testing

          There are a number of conditions affecting the test program at
this plant which are worthy of note, as detailed below.
          a.  The same bulk asphalt is used for saturating and coating
asphalt.   Only the blowing duration is changed, from 1-1/2 hours for
saturant to 4-1/2 hours for coating.   It was therefore decided that a
saturant run could be "synthesized" by merely sampling the first 1-1/2
hours of a 4-1/2 hour coating blow.

-------
                                     40
          b.  The bulk asphalt source was changed midway through our
tests, on August 23, 1975.  The procurement specifications are the
same for both asphalts, but plant personnel assured us that the asphalts
have different characteristics.  They claim that the softening point (SP)
of the asphalt normally used tends to drop with time, whereas the replace-
ment asphalt maintains its SP.  One saturant run and two coating runs
were conducted with the regular asphalt and two saturant and one coating
run were made with the replacement asphalt.  No significant differences
were noticeable during the tests, however, in either the blowing times
or the final softening points  -  which were usually in the 127 to 146°F
range for the saturants and in the 240 to 259°F range for coatings.
          c.  The settings of the dilution air dampers for the afterburner
were changed on August 23, 1975, when the secondary dilution air damper
was found to be sticking open.  This change was reflected in the test data
for subsequent tests by higher CO levels at the exhaust and an increase
in gas flow through the still (airflow at the A/B exhaust remained relatively
constant, indicating that dilution air had reduced.
          d.  The asphalt normally used is delivered by pipeline at about
450°F.  The replacement asphalt, however, arrived in tanker trucks at
around 300°F.  As a result, the temperature of the bulk asphalt storage
tank could not be maintained at the normal 450 to 475°F.   An old blowing
still  was pressed into service on August 26, 1975 as a preheater.   The
asphalt for one saturant blow (1/2 run), however, was received at the still
at only 410°F, about 50°F below normal.  No water was added during this blow
and the asphalt temperature still never rose above 490°F (it is normally
kept down to 500°F by the addition of cooling water).  Data collected during
this run may therefore not be representative (gas HC, which was monitored
continuously, was lower than on previous runs).
          e.  A circular recorder was used to monitor the natural  gas
consumption of the afterburner.  This is useful as an indicator of the
amount of heating value provided by the fumes.  Also included on the same
line,  however, are the flux tank heater and some old blowing stills which
are now used as preheaters.  The normal usage of the flux heater (the only
heater in use the first week) was estimated by turning it off for a short
time and determining the differential usage.  Unfortunately, when the flux

-------
                                     41
source was changed, the delivery temperature dropped dramatically as a
result, the flux tank heater was cranked up to maximum capacity and one
of the old blowing stills was started up and operated at capacity.  It
was planned to have both of these units cut off for a few minutes on
August 27 to determine their rate of consumption, but the readings were
so erratic on that date that this was not done on the grounds that no
meaningful differential usage could be determined.
          f.  Test samples also include emissions from the asphalt in
the still not being blown.  For that reason, the asphalt level in this
still, and transfers in and out, were also recorded.
          g.  Inlet fume concentration was, of necessity, measured down-
stream of the knock out boxes, which collect some of the still emissions
and return them to the process.  Thus, the measured inlet emissions were
already partly controlled, but the extent of this control could not be
determined.
          h.  At 5 minutes into run # 2 of August 19, 1975, the incinerator
temperature rapidly rose about 300°F (to almost 1700°F) and steam flow
increased about 37 percent.  At the same time, natural gas consumption
decreased almost 16 percent and the dilution air dampers opened wide.  Blowing
was suspended momentarily by cutting off the air supply.  Conditions returned
to normal after a few minutes.  This response, which was noted on all blows,
although to a much lesser degree, can be tied into the closing of the bypass
valve on the still air supply.*  The magnitude of this occurrence was
attributed by plant personnel to the use of fresh asphalt (it was received
that morning).  They stated that flux asphalt is more volatile when first
received, but is normally stored for several days prior to use.  They also
noted that visible emissions; from the uncontrolled flux tank are more evident
when the flux asphalt is first received.
* Asphalt blowing is initiated by starting the compressor at this plant,
however, little air passes through the asphalt at first.   Then, as the
bypass valve gradually closes (over a 6 to 6-1/2 minute period), airflow
through the asphalt increases.

-------
                                       42
                       IV.  LOCATION OF SAMPLING POINTS

          Sampling port and point locations were determined as outlined in
the EPA Federal Register. December 23, 1971, Method 1.  Accordingly, the
blowing still inlet and outlet stack geometries, as presented in Figures
3 and 4, indicate the sample port locations relative to the nearest upstream
and downstream flow disturbances, for test points TP-1, TP-2, respectively.
The number of upstream stack diameters from any flow disturbance are pre-
sented as follows:

                      Upstream Sample Port Configuration
                                         Upstream          Number of
         Test Point       Location       Diameters       Sample Points
             1             Inlet           9.75                8
             2             Outlet          5                  32

          As indicated,  the desired minimum number of 2 downstream stack
diameters was met for both stacks.   Detailed sample point locations for
each port are presented  in Figures  6 and 7.

-------
                             43
 I I  I I
10
t~
4)
aj


T>
m

OS
t
 0)
 
-------
                                              44
    o
    Flow
    -36"
s
1

"oc
c


/,• •



"<
c
f
1

l/>
\-
a
di
• I
! 5
c
tf






<
i
<
«
i
I .!
- i
C
u

Percent Distance from Percent Distance from
of Outside of Sam- of Outside of Sam-
Point Diameter pie Port, in. Point Diameter pie Port, in.
1 1.6 5 3/4 9 62.5 27
2 A. 9 6 1/4 10 71.7 30 3/8
3 8.5 7 1/2 11 78.0 32 5/8
4 12.5 9 12 83.1 34 1/2
, 5 16.9 10 1/2 13 87.5 36
' 6 22.0 12 1/2 14 91.5 37 1/2
7 28.3 14 5/8 15 95.1 38 3/4
8 37.5 18 16 98.4 39 1/4



.. i.P." 	 	 • -.». /
X^TV / N
/ 12- \
/ l0*' \
	 -Q
A C. 8.
Port \ ••! /
n \ '5 /
X A* S
- x^ ?:3 ^/
= — i "r^"^
5
5 ~<£
B Port
TP-2
                       FIGURE 7.  AFTERBURNER OUTLET STACK CONFIGURATION
                                  SHOWING SAMPLING PORTS AND  SAMPLE  POINT
                                  LOCATIONS

-------
                                     45
                 V.  SAMPLING AND ANALYTICAL PROCEDURES
                          Particulate Sampling
Background on Particulate Sampling at
Asphalt Roofing Plants
          Ortanic pollutants are generated in the manufacture of asphalt

roofing products.  These compounds may be divided into 2 categories—parti-

culates (oil droplets) and gaseous hydrocarbons (organics in the vapor
state at the filtration temperature).


          Method development tests were conducted by EPA at two asphalt

roofing plants to evaluate the proposed sampling trains to collect these

pollutants.  These studies resulted in the use of a modified Method 5

system to isokinetically sample the gas stream.  The modifications to

Method 5 include:


          1.  Change of filtration temperature from 120 C to 50 C.

              The physical state of the organic matter is a function
              of temperature.  Therefore, if was necessary to select
              a filtration temperature to provide a basis for evalua-
              ting the different control systems and the emissions
              from the different plants.  The 50 C upper limit was
              selected to be consistent with the optimum operating
              temperature of 40 C for the collection systems, i.e.,
              filtration and electrostatic precipitation.

          2.  Use of a precollector filter to reduce the oil droplet
              loading on the primary filter.

              This change was necessary to prevent oil from seeping
              through the glass fiber filter mat during periods of
              high oil droplet concentrations.  A procedure to avoid
              having to quantitatively remove the oil from the
              precollector was added to the method.  This involves
              weighing the precollector system before and after
              sampling to obtain the mass collected by difference.

          3.  Extraction of a small portion (1.0 £/min.) of the sample
              gas after filtration and use of continuous flame
              ionization detector (FID) analyzer to measure the
              gaseous hydrocarbon (HC) contents of this portion.

-------
                           46
    The sample gas to the FID analyzer is transported
    through heated lines to prevent condensation.  The
    FID analyzer monitors the level of total gaseous
    hydrocarbons by ionizing the hydrocarbons in a
    hydrogen rich flame and measuring the current pro-
    duced.  This type measurement system has been used
    by EPA as a reference method at other emission
    sources and is accurate over a wide-range of
    concentrations.

    To develop background information on the chemical
    composition of the emitted organics, specific hydro-
    carbons were identified using techniques which
    included infrared analysis, gas chromatography (GC),
    and GC mass spectroscopy (GC-MS).   The oil droplet
    fractions were semi-quantitatively analyzed and
    found to contain almost every category of hydro-
    carbons, e.g., ringed, straight chained, partially
    oxidized, etc.  However, the gaseous portion which
    was defined using quantitative GC-MS analyses
    showed that the hydrocarbons were primarily 1-
    through 3-carbon chains for noncombustion control
    devices, and methane for after-burners.   Therefore,
    the FID response to the gaseous hydrocarbons was
    determined to be satisfactory.  These data are
    described in the individual test reports and in
    the test reports concerning the method development.

4.  Change in cleanup reagent from acetone to 1,1,1-
    trichloroethane.

    Sample cleanup and recovery procedures were also
    developed and tested during the method development
    program.  Various solvents were used, e.g., acetone,
    chloroform, hexane, 1,1,1-trichloroethane, diethyl
    ether, methylene chloride, and trichloroethylene.
    The chlorinated hydrocarbons proved to be most
    effective solvents.  Chloroform and methylene
    chloride were rejected as unsafe due to the toxic
    chemical exposure criteria established by OSHA.
    The solvent, 1,1,1-trichloroethane (TCE) was
    decided upon because it was most effective in dis-
    solving the baked-on oil and tars and, due to its
    lower vapor pressure, was potentially less toxic.

5.  Change in analysis procedure to minimize sample loss
    through evaporation.

    In the laboratory the cleanup reagent presented
    some problems.  The low vapor pressure of TCE caused
    an increase in the time necessary to evaporate the
    samples at ambient temperature to a final weight.
    Experiments were conducted to quantify the loss of
    light hydrocarbons by condensing the vapors from the

-------
                                     47
              evaporation process and analyzing them by gas chromato-
              graphy.  Results showed that the hydrocarbon loss for
              outlet sample fractions was minimal.
              A continuous weight loss was recorded for the samples
              over a period of several weeks after removal from the
              condenser.  The weight loss was most significant for
              inlet samples.  The outlet samples also continued to
              lose weight, but to a lesser degree.  Consequently,
              the criterion of "constant weight" was defined as "a
              less than 10 percent weight change between two sequen-
              tial weighings 24 hours apart."  Most samples weighed
              in this manner reached a constant weight between the
              24 to 48 hour weighings.

          6.  Collection and analysis procedure for condensed water.
              In cases where moisture contents of the stack gases
              were above 10 percent, condensation in the filtration
              section of the sample train occurrs.  These conditions
              happen when sampling from blowing stills.  Thus
              impingers, or some other collection device, must be
              provided upstream of the filter to prevent excessive
              carryover of water and/or oil onto the filter.  In
              this analyses, the oil is extracted from the water
              phase using a separatory funnel and TCE.  The remain-
              ing water fraction is evaporated at 100 C, desiccated,
              and weighed.

          Previous investigators used test methods which differed from the
EPA approach.  These methods, e.g., LAAPCD and conventional Method 5

including impinger analysis, measured both filterable and condensible
hydrocarbons as particulate,.  The gaseous hydrocarbons were measured by
flame ionization analysis; the sample gas, however, was taken directly
from the stack.  The gases were neither filtered nor cooled to 50 C.  In

some cases the data gave similar emission rates.  In other cases, large

differences occurred.  Since EPA did not conduct comparative tests, it

cannot be determined if these differences were due to process operating
conditions or to differences in the test methods.
          The blowing still facility is a batch process.  It must be

tested over a complete cycle which may be several hours.  Emissions, flow-
rates, and temperatures are a function of time.  Careful attention is

required to ensure that the cycle being tested is indicative of the normal

process operation.

-------
                                     48
Particulate Sampling, Inlet

          Particulate sampling at the inlet was conducted using a sampling
train similar to that described by EPA in their "Draft Method for Determina-
tion of Particulate and Total Gaseous Hydrocarbon Emissions from the Asphalt
Roofing Industry".  The sampling train configuration is shown in Figure 8.
Because of the exceptionally high water and oil content present in gas
from the blowing still process, the sampling train included 4 impingers
in a water bath and a prefilter upstream of the filter.  The upstream
impingers were provided to condense and collect the water and oil which
otherwise would have plugged the filter.  At the beginning of a run, 100
ml of distilled water was placed in the first two impingers, the third
and fourth impingers were left empty.  The prefilter consisted of a glass
tube packed with a pre-weighed amount of glass wool.  The prefilter
removed most of the oil and water carry-over from the impingers.  Because
much of the water was removed from the gas sample upstream of the filtef,
only two impingers were used between the filter and vacuum pump.  These
two impingers were identical in use to the last two impingers of the
standard Method 5 sampling train.

          Additional features of the sampling train include (1) a tee
downstream of  the filter for withdrawing a portion of the gas sample for
continuous hydrocarbon analysis by FID, (2) maintaining the temperature
of sample gas at 32 to 38 C (90 to 100 F) through the prefilter and filter
(while downstream impingers, as usual, were in an ice bath), and (3)
washing the sampling train components with TCE followed by an acetone
rinse.

          Since the FID was connected in parallel with the particulate
sampling train, the total flow through the sample nozzle was the sum of
the two sample flows.   Therefore, to maintain isolinetic flow conditions
at the sample nozzle at about 0.75 scfm, the flow through the particulate
train downstream of the tee had to be reduced by an amount equal to the
flow through the FID.   This was accomplished by treating the FID flow as
a pseudo moisture.

-------
                     Glass wool prefilter
               \
       "~  Heat  "
       adjusted to
      maintain  IOOF
         at filter
                                              Conventional  filter

                                                      ^FID
Impinger
water bath



-*




\
V.

I
>
1
s
f
I
1
t
/!<
N^

— IOOF —
1



-
*~ Impinger*1
   ice bath
FIGURE 8.  BLOWING STILL PROCESS SAMPLE TRAIN  CONFIG JfvAT ION-INLET

-------
                                     50
          Using a nominal particulate sampling train flowrate of 0.75 cfm
and an FID flowrate of 0.083 cfm, the percent moistrue adjustment would
be 0.083/0.75 or 11 percent.  The nomograph was then adjusted by an
amount equal to the sum of the actual stack moisture plus the calculated
FID pseudo moisture.

          At the end of each run, the FID sample volume was calculated by
multiplying the FID run time by the FID flowrate.  The FID volume was then
corrected for moisture.  The total dry sample volume was then calculated
as the sum of the gas volumes passed through the particulate sampling
train and the FID.

Particulate Sampling, Outlet

          The particulate sampling train used for sampling at the outlet
is shown in Figure 9.  It is identical to that used at the inlet (Figure 8),
except that only two impingers were used upstream of the filter.  The blowing
still outlet gasses, after passing through an afterburner, did not contain
appreciable water or oil and, therefore, only two impingers were required to
protect the filter from overloading or becoming wet.  At the beginning of a
test the first impinger contained 100 ml of distilled water, the second
impinger was empty.

          The FID flowrate for outlet sampling was nominally 240 cm^/min.,
and when comparing this flow to the flowrate for the particulate sampling
train of 0.75 cfm, the FID flowrate was considered insignificant (about
1 percent) and no correction for the flow was made to the particulate
sampling train flowrate.  Appendix A contains field data sheets related to
the inlet and outlet particulate sampling, Orsat results, sampling weight
data, and sample identification.

-------
                      Glass wool prefilter\
       *~  Heat   •*
       adjusted to
       maintain IOOF
         at filter
Impinger
water bath
                                               Conventional filter

                                                        rFID
*~ Impingei*"
   ice bath
FIGURE 9.   BLOWING STILL PROCESS SAMPLE TRAIN CONFIGURATION-OUTLET

-------
                                     52
                          Particulate Analyses

TCE Wash of Nozzle and Probe

          The TCE wash of the nozzle and probe was dried by evaporating
the TCE in a distillation-condensation apparatus as shown in Figure 10.  The
particulate sample was considered dry when no TCE was observed to be moving
from the distillation flask to the condensating flask.  Infrared analyses
were made of the dried samples to determine if any TCE remained.  Results
of the infrared analyses showed that all dried samples contained less than
5 percent TCE, except one which contained about 10 percent TCE.  The samples
were weighted and the weights were used in particulate emissions calculations.
(Because of the possible volatile nature of these sample portions, the
samples were not desiccated.)

Impinger Catch and Wash

          Bulk Oil.  For the inlet samples, a large quantity of oily material
was collected in the impingers, upstream of the filter.   The oil was physically
extracted from the water using a separatory funnel and was then weighed.  For
the outlet samples there was no large quantity of oil present in the impingers
and, thus, no comparable sample existed.

          TCE Extraction.  Because the water portion of the upstream impinger
catch was actually an emulsion, a TCE extraction was made to remove organic
material.   The extract was dried by evaporating the TCE in ambient air.   The
sample was then desiccated for at least 24 hours and weighed.*

          The TCE extraction did not remove all organics from the water.
The residue from the water portions and a precipitate (not soluble in TCE),
appeared to be mostly organic.  The following sections describe further
treatment of the water and precipitate portions.
*  Weighing involved taking the weight after successive 24-hour periods of
   desiccation and using the mass value obtained by averaging the first two
   values that did not differ by more than 10 percent.  In essentially every
   case the values that were averaged were the 24 and 48-hour readings.

-------
                         53
                 Cylinder Air
                  330 cc/min.
     Evaporating Time
         6 days
                                                 Vent
                                      Dry Ice,
                                      -110 F
                                     ICE  Recovery
                                       99 +  %
   Distillation Flash
Condensation Flash
FIGURE 10.   COLD TRAP APPARATUS USED TO EVALPORATE TCE
            FROM PROBE WASH FOR INLET SAMPLES FOR ELK
            ROOFING PLANT

-------
                                     54
          Precipitate.  Following the ICE extraction, it was noted that a
water insoluble and TCE insoluble precipitate remained in the water portion.
This precipitate, which appeared to be very fine particles, was removed
from the water portion using a separatory funnel.  The precipitate was air
dried, desiccated for at least 24 hours, and weighed.*  Visible water was
evident in three of the samples after desiccation.  These three samples
(where an oil film appeared to be preventing water loss) were dried by
evaporating the water in an oven at 95 C (203 F).  The samples were then
weighed.*

          Water.  The water sample consisted of the water from the upstream
impingers that had been extracted with TCE.   Water and other material remain-
ing on the walls of the separating funnel after this extraction were washed
off with acetone and combined with the water sample.

          After removing the precipitate, as described above, the water was
evaporated at 95 C (203 F) until the residue appeared dry; the residue was
desiccated for at least 24 hours and weighed.*

          TCE Wash of Impingers.  The TCE wash was evaporated from the
impinger samples by air drying, desiccating for at least 24 hours, and
weighing.*

          Acetone Wash of Impingers.  The acetone was evaporated from the
impinger acetone wash samples by drying in air, desiccating for at least
24 hours, and weighing.*

Prefilters

          Because of the different nature of the prefilter catches for the
inlet and outlet runs, different procedures were used for obtaining weights
on these samples.
*  See Footnote on previous page.

-------
                                     55
          Inlet Run Prefliter.  The prefilters from the inlet runs contained
significant quantities of oil.  Thus, the prefilters were washed (while still
in the prefilter bulk) with approximately 210 ml of TCE (3 x 70 ml).  The
oil-laden TCE was filtered to remove pieces of fiberglas which were washed
from the prefilter.  The TCE was evaporated by air drying until the sample
appeared to be dry, and the residue was weighed.*

          Outlet Run Prefilter.  The prefilters from the outlet runs did not
appear oily, but did appear to contain a dry particulate material.  The
sample weights for the prefilters were obtained by removing them from the
prefilter bulb and weighing.* The sample weight was calculated as the dif-
ference between the prefilter gross and tare weights.
Filter
          The filters were desiccated for at least 24 hours, removed from
the desiccator in a constant temperature, constant humidity room, and
weighed.  Sample weight was taken as the difference between the filter
gross and tared weights.

Remarks

          Table 15 summarizes the particulate data obtained by the above
analyses and used in the particulate emissions calculations.


                        Hydrocarbon Measurements

          Hydrocarbon measurements were made by continuous monitoring, by
FID, of gas sampled at a point in the particulate sampling train between
the filter and the first downstream impinger, as shown in Figures 8  and  9.   Two  FID
instruments were used, the Beckman** Model 109 at the outlet and 402 at the
inlet.  Operation of the two instruments is basically the same, the primary
*  See footnote, Page 52.
** Mention of trade names or manufacturers is not intended to constitute
   endorsement by EPA or its contractors.

-------
                                           56
   TABLE 15.  SUMMARY OF MASS COLLECTED IN FRONT HALF OF STACK GAS SAMPLING TRAIN,
              ELK ROOFING, STEPHENS, ARKANSAS - AUGUST 19-26, 1975
RUN NUMBER
SAMPLE NUMBER
LOCATION
TCE rinse of
probe + nozzle
Oil catch
TCE extract
Precipitate
Water
TCE rinse of
implngers
Acetone rinse
of Impingers
Prefilter
Filter
Total
2
B-3
Inlet
2346.6*
48500.0
5387.1
188.4
811.3
9283.2
39.2
2930.8
297.6
69784.2
6
B-ll
Inlet
3941.0*
36900.0
2747.6
195.2
1268.3
9368.9
248.5
3675.5
426.5
58771.5
7
B-13
Inlet
7947.1*
19300.0
•811.7
208.0
545.6
6691.5
24.3
1463.9
97.0
37089.1
2
B-4
outlet
69.5*
0
6.4
11.5
892.1
844.5
7.1
118.5
5.3
1954.9
6
B-12
outlet
32.1*
0
0
8.6
270.8
190.3
3.7
0
5.7
511.2
7
B-14
outlet
83.0*
0
0
3.5
78.2
433.9
2.3
0.4
4.8
606.1

RUN NUMBER
SAMPLE NUMBER
LOCATION
TCE rinse of
probe + nozzle
Oil catch
TCE extract
Precipitate
Water
TCE rinse of
impingers
Acetone rinse
of impingers
Prefilter
Filter
Total
3
B-5
Inlet
16376.1
62200.0
982.5
CS
1647.2
4500.0
229.1
1961.5
699.4
88595.8
4
B-7
Inlet
22905.5*
60200.0
2833.9
274.6
1390.8
6349.9
85.4
3991.2
783.2
98814.5
5
B-9
Inlet
24376.3*
55600.0
2528.8
414.7
2147.4
8696.9
163.3
3831.4
958.1
98706.9
3
B-6
outlet
178.9
0
11.2
CS
279.5
496.0
6.4
5.6
5.1
982.7
4
B-8
outlet
21.4*
0
1.6
4.7
101.8
470.5
7.2
10.0
7.3
624.5
5
B-10
outlet
128.5
0
0.4
4.1
130.4
562.8
20.8
0
4.4
851.4
Note:  Run No. 1 (B-l and B-2 voided).
       CS - This portion included in another sample portion.
       All values reported are in milligrams (mg).
       * Blank correction for combined solvents varies from sum of individual solvent
         blank values.

-------
                                     57
difference being that the Model 402 incorporates a selectable elevated-
temperature sampling line and analyzer oven.  Because of the high moisture
content of the gas, a  heated sampling line also was used on the Model 109
analyzer at the outlet.  The heated sampling lines on both analyzers were
maintained at 96 to 99 C (205 to 210 F) during sampling.  An ice water
trap was incorporated between the heated sampling lines and the instruments
to prevent moisture from reaching the analyzers.  The estimated temperature
of the gas leaving the ice traps was 60 F.

          Calibration of the analyzers was done on-site (prior to each run)
using a known concentration of methane (Ctfy), and hydrocarbon-free zero air.
The hydrocarbon-free zero air was also used as a carrier gas for the FID.
The calibration gases are "Certified Standard Gases" obtained from Matheson
Gas Products*, who supply a certificate of analysis for each gas.

          Tables 2 and 3 contain results of the gaseous hydrocarbon measurements.
Complete gaseous hydrocarbon results and calculated emissions at each
sampling point are given in Appendix C.  The molecular weight of methane
(16) was used when converting gaseous hydrocarbon data from volume to mass.


                      Carbon Monoxide Measurements

          Continuous carbon monoxide (CO) measurements were made at the
afterburner outlet (TP2) on two runs, Run Nos. B-4 and B-6.  Due to unstable
sampling conditions on the scaffolding (wind, rain, vibrations and voltage
fluctuations) the sensitive CO instrument was moved to the air conditioned
van and the remaining CO runs were monitored via integrated bag samples.
The instrument used for monitoring CO was a Beckman* Model 315B nondispersive
infrared analyzer.  The analyzer was calibrated before and after each run
with zero nitrogen and certified standard CO span gas of 1050 ppm.
*  Mention of trade names or manufacturers is not intended to constitute
   endorsement by EPA or its contractors.

-------
                                     58
          A schematic drawing of the sampling train used to monitor CO
emissions is shown in Figure 11.  The primary function of the impingers in
the train is the removal of water and C02, which are infrared-absorbing
and cause serious interference effects if not minimized.

          To calculate CO emissions, the minimum, maximum, and average CO
concentrations were determined for each sampling point  (where valid data
were obtained).  The CO was sampled from a single point (centroid at TP 2).
The CO concentration was assumed uniform throughout the duct and, thus, the
particulate sampling train velocity data were used to calculate CO emission
rates.  The mean time and CO concentrations data reported with the CO
calculations are based on the particulate sampling traverse point data.
The CO recorder charts and calculated emissions at each sampling port are
given in Appendix D.  The two continuous CO runs appear on the same charts
with the FID measurements, and are labeled Run Nos. B-4 and B-6.  To convert
the CO meter and/or recorder chart readings to ppm concentrations, a calibra-
tion curve (Appendix D) must be used.
                     Continuous S02/NOX Measurements

          An attempt was made to measure the concentration of S02 and NOX,
both on the inlet and outlet sides of the afterburner.  Two types of
monitoring instruments were used:  (1)  EnvironMetrics*,  Model No. NS 200
S02/NOX analyzer, and (2) DynaScience* Nitrogen Oxide analyzer.  Both
instruments use a replaceable dry electrochemical-type sensing cell and are
designed more for laboratory use, rather than for field use.

          Various problems, unrecognized at the time of sampling, made the
S02 data questionable and the NOX data invalid.  Some of  the problems
encountered were (1) negative meter response when NOX was measured on the
EnviroMetrics analyzer (possibly due to an unknown inference), (2) the
*  Mention of trade names or manufacturers is not intended to constitute
   endorsement by EPA or its contractors.

-------
                                       59
     Stainless Steel

         '/2V d.  *  f'
-
 Duct
                                       ml
 To
NDI R

 via.
                                                       j
            FIGURE 11.  SAMPLING TRAIN FOR DETERMINING CO  EMISSIONS

                        AT THE AFTERBURNER OUTLET

-------
                                     60
stack gas was not scrubbed for SC>2 removal (S02 is an interfering gas for NOX)
when NOX was measured on the DynaScience analyzer, and (3) running out of S02
calibrating gas at the end of Run B-ll (the value was inadvertently left
open and the cylinder contents were discharged), thus, there was no final
calibration.

          For information purposes, Table 8 summarizes the results obtained
with the instruments.  Explanatory footnotes are included.  The only valid
data obtained are the SC>2 concentration measured on the outlet gas during
the beginning of a coating blow (which should be equivalent to a saturant
blow).  The NOX data are presented for information purposes only:  this is
not valid data.

          For reporting purposes, the recorder strip charts and the workup
data sheets of the S02/NOX readings are included in Appendix D.


               Gas Compos'ition Flask Sampling and Analysis

          Three evacuated flasks were obtained from the FID sample line
during Run B-6 while sampling outlet emissions at TP-2.  One of the flask
samples, S75-006-034, was analyzed at Battelle using gas chromatography
techniques.  Table 9 reports the analysis of this flask.   Laboratory data
reported for this analysis are given in Appendix I.


                        POM Sampling and Analysis

          POM sampling was conducted using a modified EPA Method 5 train at
locations TP-1 and TP-2 (inlet and outlet) on August 27,  1975.  Figure 12
depicts the modified train component configuration used for sampling POM at
the inlet; for sampling at the outlet the heated impingers were omitted.
The modification consisted of adding an adsorbent sampling column between
the filter and the impingers.  Details of the design and operation of the
POM sampling train are given in Appendix F.

-------
Flue gas
  flow
                  Probe
  THIS PORTION OMITTED FOR
     SAMPLING AT OUTLET

      |              I
      |     Heated   |
Filter i   Impingers  i      . .   .    t
  A.   I     *x      I     Adsorbent
•Light Shield

        Impingers
                                                         i     Pump  Thermostated
                                                         I             Daeorwn ir
                                           4_ Y3ri_ac	J
                                                                      Reservoir
                                                            175 F
                                                     Ice Bath
                                                                                                              To  Flow
                                                                                                              Control
                                                                                                              Device
                                  FIGURE 12. POM SAMPLE TRAIN CONFIGURATION FOR
                                             ASPHALT BLOWING STILL PROCESS

-------
                                     62
           Because of  the unusually high water and oil  content  of  the  blowing
 still process,  it was necessary  to maintain a sufficiently high temperature
 throughout the  train  to keep  the water in  the vapor phase until after passing
 through  the adsorbent sampler  (to avoid plugging).  The probe  and  filter was
 kept at  near stack  temperature (about 350  F).  The temperature in  the first
 two impingers (inlet only) and adsorbent sampler was maintained at 175 F to
 keep the water  in the gas phase; the estimated dewpoint was 165 F.  Analysis
 was by gas chromatography-mass spectrometry (GC-MS).

          An extensive discussion of details of POM sample recovery and
 analysis are included in Appendix F.  Field data sheets and calculated
 emissions  for POM inlet and outlet runs are included in Appendix F.   Table 10
 summarizes POM  emissions results.

          The sampling methodology used for this work is state-of-the-art.
 However, it is  still undergoing validation studies and its accuracy and
 reliability for all sampling conditions is not proven.  The adsorbent
 sampler, the crucial component of the POM sampling train, has been validated
 for a limited number of in-house laboratory samples.

                     Aldehyde Sampling and Analysis

          The inlet and outlet stack gases of the afterburner control device
were sampled to determine aldehyde concentration.  It was assumed that the
 level of concentration would be relatively low and that a large sample
volume would be necessary to enable detection by titration.   Therefore,  to
 facilitate the sampling, the Method 5 sample Train used for this sampling
was set up without the filter in place so that large volumes of gases could
be sampled in a relatively short period of time.

          The referenced procedure used for the sampling and analysis of
 aldehydes was Los Angeles Method
(1)  Los Angeles Procedure for Sampling and Analysis for Aldehydes,
     Los Angeles Air Pollution Control District Publication Number
     APCD 5-46.

-------
                                     63
          One hundred cc's of one (1) percent sodium bisulfite was placed
in each of the first two impingers of the sampling train.  The third impinger
was empty and the fourth impinger contained 200 grams of indicating Drierite.

          The inlet and outlet stack gases were sampled simultaneously to
                                               3
obtain sample volumes of approximately 15-20 ft .  The reagents were treated
according to the Los Angeles Method   , stored in glass containers and
returned to BCL for analysis.  The results of these measurements are pre-
sented in Table 11.

          The analytical report card (which contains additional information
regarding the analyses) and the field data forms are presented in Appendix H.

                      Visible Emission Measurements

          Monsanto Research Corporation, as a contractor to the Environmental
Protection Agency under the "Field Sampling of Atmospheric Emissions" program,
was requested to provide visible emission data from the Elk Roofing Co. plant.

          At the Elk Roofing Co. plant, the effluent from the afterburner
and heat exchanger is ducted to the atmosphere through a 30-foot stack.
Visible emission data were recorded on the effluent from this stack, for a
period of about 14-1/2 hours.

          The visible emissions were determined as prescribed by EPA
Method 9 as given in the Federal Register, Vol. 39, No. 219, November 12,
1974.  Opacity readings were made from three locations within the plant site
so that the angle with the line of observation and the sun would be no more
than 70 degrees.  The physical layout of the plant showing the location of
the stills, afterburner, and heat exchanger as well as the location of
observation points and direction of observation are shown in Figure 13.  A
summary of the time of observation, observation point used, pertinent
distances, weather conditions and opacity is presented in Table 13.

-------
                        64
   Observers #1 and #2 positions were between
   Points A and B, 50-80 feet from the stack.
   For exact locations, see Table 13.
FIGURE 13.  RELATIVE POSITION OF VISIBLE EMISSION
            OBSERVERS WITH RESPECT TO THE OUTLET
            STACK AND ADJACENT STRUCTURES

-------
                                     65
During the entire observation period, the opacity was recoreded at 0 per-
cent, that is, no visible emission could be detected.

          The complete visible emission report, as presented by Monsanto
to EPA (which includes the field data sheets), is presented as Appendix G.

              Analysis of Process Sampling and Recovery Oil

          Process samples were collected of the following:  flux (bulk
unblown asphalt), saturant asphalt (after blowing), coating asphalt (after
blowing), recovery oil (from afterburner), and recovery oil (from after-
burner drain).

          Flux (unblown asphalt—EPA Sample No. S75-006-136), saturant and
coating asphalt (EPA Sample Nos. S75-006-137 and S75-006-129, respectively),
and recovery oils (EPA Sample Nos. S75-006-004 and S75-006-133) were
analyzed by ASTM procedures for proximate and ultimate analyses.  That
analysis was done by Drawin Nevey of Nalin Laboratories of Columbus, Ohio,
on Battelle Purchase Order No. L-6056.  Values are reported for volatile
matter, fixed carbon, moisture, ash, C, H, 0, S, and N.  They applicable
ASTM procedures used for these analyses were:
                     Proximate Analysis           D3172
                       Volatime Matter (VM)       D3175
                       Moisture (M)               D3173
                       Ash (A)                    D3174
                       Fixed Carbon               By difference
                                                  (100-VM-M-A)
                     Ultimate Analysis            D3176
                       Carbon (C) and             D3178
                         Hydrogen (H)
                       Sulfur (S)                 D3177
                       Nitrogen (N)               D3179
                       Oxygen (0)                 By difference
                                                  (100-C-H-N-S-M-A).
          Table 7 summarizes process sanalysis results.  Copies of the
laboratory reports of these analyses are included in Appendix I.

-------